From 61a0c9a6ffa003b3829afaa88117f2d152f76df1 Mon Sep 17 00:00:00 2001 From: Alistair Leslie-Hughes Date: Wed, 17 May 2023 08:35:40 +1000 Subject: [PATCH] Updated vkd3d to f0a16d84ce939716defe41268340757da3900055. --- libs/vkd3d/include/list.h | 270 ++ libs/vkd3d/include/private/list.h | 270 ++ libs/vkd3d/include/private/rbtree.h | 378 ++ libs/vkd3d/include/private/vkd3d_common.h | 27 +- libs/vkd3d/include/private/vkd3d_test.h | 432 +++ libs/vkd3d/include/vkd3d_d3d9types.h | 237 ++ libs/vkd3d/include/vkd3d_d3dcompiler.h | 92 + libs/vkd3d/include/vkd3d_d3dcompiler_types.h | 45 + libs/vkd3d/include/vkd3d_shader.h | 163 +- libs/vkd3d/include/vkd3d_utils.h | 123 + libs/vkd3d/include/vkd3d_windows.h | 289 ++ libs/vkd3d/libs/vkd3d-common/blob.c | 2 + libs/vkd3d/libs/vkd3d-shader/d3d_asm.c | 139 +- libs/vkd3d/libs/vkd3d-shader/d3dbc.c | 142 +- libs/vkd3d/libs/vkd3d-shader/dxbc.c | 162 +- libs/vkd3d/libs/vkd3d-shader/dxil.c | 3269 +++++++++++++++-- libs/vkd3d/libs/vkd3d-shader/hlsl.c | 353 +- libs/vkd3d/libs/vkd3d-shader/hlsl.h | 111 +- libs/vkd3d/libs/vkd3d-shader/hlsl.l | 56 +- libs/vkd3d/libs/vkd3d-shader/hlsl.y | 698 +++- libs/vkd3d/libs/vkd3d-shader/hlsl_codegen.c | 952 +++-- .../libs/vkd3d-shader/hlsl_constant_ops.c | 417 ++- libs/vkd3d/libs/vkd3d-shader/ir.c | 648 +++- libs/vkd3d/libs/vkd3d-shader/preproc.h | 2 +- libs/vkd3d/libs/vkd3d-shader/spirv.c | 923 +++-- libs/vkd3d/libs/vkd3d-shader/tpf.c | 1715 +++++---- .../libs/vkd3d-shader/vkd3d_shader_main.c | 296 +- .../libs/vkd3d-shader/vkd3d_shader_private.h | 230 +- libs/vkd3d/libs/vkd3d/command.c | 352 +- libs/vkd3d/libs/vkd3d/device.c | 437 ++- libs/vkd3d/libs/vkd3d/resource.c | 121 +- libs/vkd3d/libs/vkd3d/state.c | 409 ++- libs/vkd3d/libs/vkd3d/vkd3d_main.c | 4 +- libs/vkd3d/libs/vkd3d/vkd3d_private.h | 67 +- libs/vkd3d/libs/vkd3d/vkd3d_shaders.h | 568 +-- 35 files changed, 11546 insertions(+), 2853 deletions(-) create mode 100644 libs/vkd3d/include/list.h create mode 100644 libs/vkd3d/include/private/list.h create mode 100644 libs/vkd3d/include/private/rbtree.h create mode 100644 libs/vkd3d/include/private/vkd3d_test.h create mode 100644 libs/vkd3d/include/vkd3d_d3d9types.h create mode 100644 libs/vkd3d/include/vkd3d_d3dcompiler.h create mode 100644 libs/vkd3d/include/vkd3d_d3dcompiler_types.h create mode 100644 libs/vkd3d/include/vkd3d_utils.h create mode 100644 libs/vkd3d/include/vkd3d_windows.h diff --git a/libs/vkd3d/include/list.h b/libs/vkd3d/include/list.h new file mode 100644 index 00000000000..2e1d95f3fd4 --- /dev/null +++ b/libs/vkd3d/include/list.h @@ -0,0 +1,270 @@ +/* + * Linked lists support + * + * Copyright (C) 2002 Alexandre Julliard + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA + */ + +#ifndef __WINE_SERVER_LIST_H +#define __WINE_SERVER_LIST_H + +#include + +struct list +{ + struct list *next; + struct list *prev; +}; + +/* Define a list like so: + * + * struct gadget + * { + * struct list entry; <-- doesn't have to be the first item in the struct + * int a, b; + * }; + * + * static struct list global_gadgets = LIST_INIT( global_gadgets ); + * + * or + * + * struct some_global_thing + * { + * struct list gadgets; + * }; + * + * list_init( &some_global_thing->gadgets ); + * + * Manipulate it like this: + * + * list_add_head( &global_gadgets, &new_gadget->entry ); + * list_remove( &new_gadget->entry ); + * list_add_after( &some_random_gadget->entry, &new_gadget->entry ); + * + * And to iterate over it: + * + * struct gadget *gadget; + * LIST_FOR_EACH_ENTRY( gadget, &global_gadgets, struct gadget, entry ) + * { + * ... + * } + * + */ + +/* add an element after the specified one */ +static inline void list_add_after( struct list *elem, struct list *to_add ) +{ + to_add->next = elem->next; + to_add->prev = elem; + elem->next->prev = to_add; + elem->next = to_add; +} + +/* add an element before the specified one */ +static inline void list_add_before( struct list *elem, struct list *to_add ) +{ + to_add->next = elem; + to_add->prev = elem->prev; + elem->prev->next = to_add; + elem->prev = to_add; +} + +/* add element at the head of the list */ +static inline void list_add_head( struct list *list, struct list *elem ) +{ + list_add_after( list, elem ); +} + +/* add element at the tail of the list */ +static inline void list_add_tail( struct list *list, struct list *elem ) +{ + list_add_before( list, elem ); +} + +/* remove an element from its list */ +static inline void list_remove( struct list *elem ) +{ + elem->next->prev = elem->prev; + elem->prev->next = elem->next; +} + +/* get the next element */ +static inline struct list *list_next( const struct list *list, const struct list *elem ) +{ + struct list *ret = elem->next; + if (elem->next == list) ret = NULL; + return ret; +} + +/* get the previous element */ +static inline struct list *list_prev( const struct list *list, const struct list *elem ) +{ + struct list *ret = elem->prev; + if (elem->prev == list) ret = NULL; + return ret; +} + +/* get the first element */ +static inline struct list *list_head( const struct list *list ) +{ + return list_next( list, list ); +} + +/* get the last element */ +static inline struct list *list_tail( const struct list *list ) +{ + return list_prev( list, list ); +} + +/* check if a list is empty */ +static inline int list_empty( const struct list *list ) +{ + return list->next == list; +} + +/* initialize a list */ +static inline void list_init( struct list *list ) +{ + list->next = list->prev = list; +} + +/* count the elements of a list */ +static inline unsigned int list_count( const struct list *list ) +{ + unsigned count = 0; + const struct list *ptr; + for (ptr = list->next; ptr != list; ptr = ptr->next) count++; + return count; +} + +/* move all elements from src to before the specified element */ +static inline void list_move_before( struct list *dst, struct list *src ) +{ + if (list_empty(src)) return; + + dst->prev->next = src->next; + src->next->prev = dst->prev; + dst->prev = src->prev; + src->prev->next = dst; + list_init(src); +} + +/* move all elements from src to after the specified element */ +static inline void list_move_after( struct list *dst, struct list *src ) +{ + if (list_empty(src)) return; + + dst->next->prev = src->prev; + src->prev->next = dst->next; + dst->next = src->next; + src->next->prev = dst; + list_init(src); +} + +/* move all elements from src to the head of dst */ +static inline void list_move_head( struct list *dst, struct list *src ) +{ + list_move_after( dst, src ); +} + +/* move all elements from src to the tail of dst */ +static inline void list_move_tail( struct list *dst, struct list *src ) +{ + list_move_before( dst, src ); +} + +/* move the slice of elements from begin to end inclusive to the head of dst */ +static inline void list_move_slice_head( struct list *dst, struct list *begin, struct list *end ) +{ + struct list *dst_next = dst->next; + begin->prev->next = end->next; + end->next->prev = begin->prev; + dst->next = begin; + dst_next->prev = end; + begin->prev = dst; + end->next = dst_next; +} + +/* move the slice of elements from begin to end inclusive to the tail of dst */ +static inline void list_move_slice_tail( struct list *dst, struct list *begin, struct list *end ) +{ + struct list *dst_prev = dst->prev; + begin->prev->next = end->next; + end->next->prev = begin->prev; + dst_prev->next = begin; + dst->prev = end; + begin->prev = dst_prev; + end->next = dst; +} + +/* iterate through the list */ +#define LIST_FOR_EACH(cursor,list) \ + for ((cursor) = (list)->next; (cursor) != (list); (cursor) = (cursor)->next) + +/* iterate through the list, with safety against removal */ +#define LIST_FOR_EACH_SAFE(cursor, cursor2, list) \ + for ((cursor) = (list)->next, (cursor2) = (cursor)->next; \ + (cursor) != (list); \ + (cursor) = (cursor2), (cursor2) = (cursor)->next) + +/* iterate through the list using a list entry */ +#define LIST_FOR_EACH_ENTRY(elem, list, type, field) \ + for ((elem) = LIST_ENTRY((list)->next, type, field); \ + &(elem)->field != (list); \ + (elem) = LIST_ENTRY((elem)->field.next, type, field)) + +/* iterate through the list using a list entry, with safety against removal */ +#define LIST_FOR_EACH_ENTRY_SAFE(cursor, cursor2, list, type, field) \ + for ((cursor) = LIST_ENTRY((list)->next, type, field), \ + (cursor2) = LIST_ENTRY((cursor)->field.next, type, field); \ + &(cursor)->field != (list); \ + (cursor) = (cursor2), \ + (cursor2) = LIST_ENTRY((cursor)->field.next, type, field)) + +/* iterate through the list in reverse order */ +#define LIST_FOR_EACH_REV(cursor,list) \ + for ((cursor) = (list)->prev; (cursor) != (list); (cursor) = (cursor)->prev) + +/* iterate through the list in reverse order, with safety against removal */ +#define LIST_FOR_EACH_SAFE_REV(cursor, cursor2, list) \ + for ((cursor) = (list)->prev, (cursor2) = (cursor)->prev; \ + (cursor) != (list); \ + (cursor) = (cursor2), (cursor2) = (cursor)->prev) + +/* iterate through the list in reverse order using a list entry */ +#define LIST_FOR_EACH_ENTRY_REV(elem, list, type, field) \ + for ((elem) = LIST_ENTRY((list)->prev, type, field); \ + &(elem)->field != (list); \ + (elem) = LIST_ENTRY((elem)->field.prev, type, field)) + +/* iterate through the list in reverse order using a list entry, with safety against removal */ +#define LIST_FOR_EACH_ENTRY_SAFE_REV(cursor, cursor2, list, type, field) \ + for ((cursor) = LIST_ENTRY((list)->prev, type, field), \ + (cursor2) = LIST_ENTRY((cursor)->field.prev, type, field); \ + &(cursor)->field != (list); \ + (cursor) = (cursor2), \ + (cursor2) = LIST_ENTRY((cursor)->field.prev, type, field)) + +/* macros for statically initialized lists */ +#undef LIST_INIT +#define LIST_INIT(list) { &(list), &(list) } + +/* get pointer to object containing list element */ +#undef LIST_ENTRY +#define LIST_ENTRY(elem, type, field) \ + ((type *)((char *)(elem) - offsetof(type, field))) + +#endif /* __WINE_SERVER_LIST_H */ diff --git a/libs/vkd3d/include/private/list.h b/libs/vkd3d/include/private/list.h new file mode 100644 index 00000000000..2e1d95f3fd4 --- /dev/null +++ b/libs/vkd3d/include/private/list.h @@ -0,0 +1,270 @@ +/* + * Linked lists support + * + * Copyright (C) 2002 Alexandre Julliard + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA + */ + +#ifndef __WINE_SERVER_LIST_H +#define __WINE_SERVER_LIST_H + +#include + +struct list +{ + struct list *next; + struct list *prev; +}; + +/* Define a list like so: + * + * struct gadget + * { + * struct list entry; <-- doesn't have to be the first item in the struct + * int a, b; + * }; + * + * static struct list global_gadgets = LIST_INIT( global_gadgets ); + * + * or + * + * struct some_global_thing + * { + * struct list gadgets; + * }; + * + * list_init( &some_global_thing->gadgets ); + * + * Manipulate it like this: + * + * list_add_head( &global_gadgets, &new_gadget->entry ); + * list_remove( &new_gadget->entry ); + * list_add_after( &some_random_gadget->entry, &new_gadget->entry ); + * + * And to iterate over it: + * + * struct gadget *gadget; + * LIST_FOR_EACH_ENTRY( gadget, &global_gadgets, struct gadget, entry ) + * { + * ... + * } + * + */ + +/* add an element after the specified one */ +static inline void list_add_after( struct list *elem, struct list *to_add ) +{ + to_add->next = elem->next; + to_add->prev = elem; + elem->next->prev = to_add; + elem->next = to_add; +} + +/* add an element before the specified one */ +static inline void list_add_before( struct list *elem, struct list *to_add ) +{ + to_add->next = elem; + to_add->prev = elem->prev; + elem->prev->next = to_add; + elem->prev = to_add; +} + +/* add element at the head of the list */ +static inline void list_add_head( struct list *list, struct list *elem ) +{ + list_add_after( list, elem ); +} + +/* add element at the tail of the list */ +static inline void list_add_tail( struct list *list, struct list *elem ) +{ + list_add_before( list, elem ); +} + +/* remove an element from its list */ +static inline void list_remove( struct list *elem ) +{ + elem->next->prev = elem->prev; + elem->prev->next = elem->next; +} + +/* get the next element */ +static inline struct list *list_next( const struct list *list, const struct list *elem ) +{ + struct list *ret = elem->next; + if (elem->next == list) ret = NULL; + return ret; +} + +/* get the previous element */ +static inline struct list *list_prev( const struct list *list, const struct list *elem ) +{ + struct list *ret = elem->prev; + if (elem->prev == list) ret = NULL; + return ret; +} + +/* get the first element */ +static inline struct list *list_head( const struct list *list ) +{ + return list_next( list, list ); +} + +/* get the last element */ +static inline struct list *list_tail( const struct list *list ) +{ + return list_prev( list, list ); +} + +/* check if a list is empty */ +static inline int list_empty( const struct list *list ) +{ + return list->next == list; +} + +/* initialize a list */ +static inline void list_init( struct list *list ) +{ + list->next = list->prev = list; +} + +/* count the elements of a list */ +static inline unsigned int list_count( const struct list *list ) +{ + unsigned count = 0; + const struct list *ptr; + for (ptr = list->next; ptr != list; ptr = ptr->next) count++; + return count; +} + +/* move all elements from src to before the specified element */ +static inline void list_move_before( struct list *dst, struct list *src ) +{ + if (list_empty(src)) return; + + dst->prev->next = src->next; + src->next->prev = dst->prev; + dst->prev = src->prev; + src->prev->next = dst; + list_init(src); +} + +/* move all elements from src to after the specified element */ +static inline void list_move_after( struct list *dst, struct list *src ) +{ + if (list_empty(src)) return; + + dst->next->prev = src->prev; + src->prev->next = dst->next; + dst->next = src->next; + src->next->prev = dst; + list_init(src); +} + +/* move all elements from src to the head of dst */ +static inline void list_move_head( struct list *dst, struct list *src ) +{ + list_move_after( dst, src ); +} + +/* move all elements from src to the tail of dst */ +static inline void list_move_tail( struct list *dst, struct list *src ) +{ + list_move_before( dst, src ); +} + +/* move the slice of elements from begin to end inclusive to the head of dst */ +static inline void list_move_slice_head( struct list *dst, struct list *begin, struct list *end ) +{ + struct list *dst_next = dst->next; + begin->prev->next = end->next; + end->next->prev = begin->prev; + dst->next = begin; + dst_next->prev = end; + begin->prev = dst; + end->next = dst_next; +} + +/* move the slice of elements from begin to end inclusive to the tail of dst */ +static inline void list_move_slice_tail( struct list *dst, struct list *begin, struct list *end ) +{ + struct list *dst_prev = dst->prev; + begin->prev->next = end->next; + end->next->prev = begin->prev; + dst_prev->next = begin; + dst->prev = end; + begin->prev = dst_prev; + end->next = dst; +} + +/* iterate through the list */ +#define LIST_FOR_EACH(cursor,list) \ + for ((cursor) = (list)->next; (cursor) != (list); (cursor) = (cursor)->next) + +/* iterate through the list, with safety against removal */ +#define LIST_FOR_EACH_SAFE(cursor, cursor2, list) \ + for ((cursor) = (list)->next, (cursor2) = (cursor)->next; \ + (cursor) != (list); \ + (cursor) = (cursor2), (cursor2) = (cursor)->next) + +/* iterate through the list using a list entry */ +#define LIST_FOR_EACH_ENTRY(elem, list, type, field) \ + for ((elem) = LIST_ENTRY((list)->next, type, field); \ + &(elem)->field != (list); \ + (elem) = LIST_ENTRY((elem)->field.next, type, field)) + +/* iterate through the list using a list entry, with safety against removal */ +#define LIST_FOR_EACH_ENTRY_SAFE(cursor, cursor2, list, type, field) \ + for ((cursor) = LIST_ENTRY((list)->next, type, field), \ + (cursor2) = LIST_ENTRY((cursor)->field.next, type, field); \ + &(cursor)->field != (list); \ + (cursor) = (cursor2), \ + (cursor2) = LIST_ENTRY((cursor)->field.next, type, field)) + +/* iterate through the list in reverse order */ +#define LIST_FOR_EACH_REV(cursor,list) \ + for ((cursor) = (list)->prev; (cursor) != (list); (cursor) = (cursor)->prev) + +/* iterate through the list in reverse order, with safety against removal */ +#define LIST_FOR_EACH_SAFE_REV(cursor, cursor2, list) \ + for ((cursor) = (list)->prev, (cursor2) = (cursor)->prev; \ + (cursor) != (list); \ + (cursor) = (cursor2), (cursor2) = (cursor)->prev) + +/* iterate through the list in reverse order using a list entry */ +#define LIST_FOR_EACH_ENTRY_REV(elem, list, type, field) \ + for ((elem) = LIST_ENTRY((list)->prev, type, field); \ + &(elem)->field != (list); \ + (elem) = LIST_ENTRY((elem)->field.prev, type, field)) + +/* iterate through the list in reverse order using a list entry, with safety against removal */ +#define LIST_FOR_EACH_ENTRY_SAFE_REV(cursor, cursor2, list, type, field) \ + for ((cursor) = LIST_ENTRY((list)->prev, type, field), \ + (cursor2) = LIST_ENTRY((cursor)->field.prev, type, field); \ + &(cursor)->field != (list); \ + (cursor) = (cursor2), \ + (cursor2) = LIST_ENTRY((cursor)->field.prev, type, field)) + +/* macros for statically initialized lists */ +#undef LIST_INIT +#define LIST_INIT(list) { &(list), &(list) } + +/* get pointer to object containing list element */ +#undef LIST_ENTRY +#define LIST_ENTRY(elem, type, field) \ + ((type *)((char *)(elem) - offsetof(type, field))) + +#endif /* __WINE_SERVER_LIST_H */ diff --git a/libs/vkd3d/include/private/rbtree.h b/libs/vkd3d/include/private/rbtree.h new file mode 100644 index 00000000000..b5d38bca54c --- /dev/null +++ b/libs/vkd3d/include/private/rbtree.h @@ -0,0 +1,378 @@ +/* + * Red-black search tree support + * + * Copyright 2009 Henri Verbeet + * Copyright 2009 Andrew Riedi + * Copyright 2016 Jacek Caban for CodeWeavers + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA + */ + +#ifndef __WINE_WINE_RBTREE_H +#define __WINE_WINE_RBTREE_H + +#define RB_ENTRY_VALUE(element, type, field) \ + ((type *)((char *)(element) - offsetof(type, field))) + +struct rb_entry +{ + struct rb_entry *parent; + struct rb_entry *left; + struct rb_entry *right; + unsigned int flags; +}; + +typedef int (*rb_compare_func)(const void *key, const struct rb_entry *entry); + +struct rb_tree +{ + rb_compare_func compare; + struct rb_entry *root; +}; + +typedef void (rb_traverse_func)(struct rb_entry *entry, void *context); + +#define RB_FLAG_RED 0x1 + +static inline int rb_is_red(struct rb_entry *entry) +{ + return entry && (entry->flags & RB_FLAG_RED); +} + +static inline void rb_rotate_left(struct rb_tree *tree, struct rb_entry *e) +{ + struct rb_entry *right = e->right; + + if (!e->parent) + tree->root = right; + else if (e->parent->left == e) + e->parent->left = right; + else + e->parent->right = right; + + e->right = right->left; + if (e->right) e->right->parent = e; + right->left = e; + right->parent = e->parent; + e->parent = right; +} + +static inline void rb_rotate_right(struct rb_tree *tree, struct rb_entry *e) +{ + struct rb_entry *left = e->left; + + if (!e->parent) + tree->root = left; + else if (e->parent->left == e) + e->parent->left = left; + else + e->parent->right = left; + + e->left = left->right; + if (e->left) e->left->parent = e; + left->right = e; + left->parent = e->parent; + e->parent = left; +} + +static inline void rb_flip_color(struct rb_entry *entry) +{ + entry->flags ^= RB_FLAG_RED; + entry->left->flags ^= RB_FLAG_RED; + entry->right->flags ^= RB_FLAG_RED; +} + +static inline struct rb_entry *rb_head(struct rb_entry *iter) +{ + if (!iter) return NULL; + while (iter->left) iter = iter->left; + return iter; +} + +static inline struct rb_entry *rb_next(struct rb_entry *iter) +{ + if (iter->right) return rb_head(iter->right); + while (iter->parent && iter->parent->right == iter) iter = iter->parent; + return iter->parent; +} + +static inline struct rb_entry *rb_postorder_head(struct rb_entry *iter) +{ + if (!iter) return NULL; + + for (;;) { + while (iter->left) iter = iter->left; + if (!iter->right) return iter; + iter = iter->right; + } +} + +static inline struct rb_entry *rb_postorder_next(struct rb_entry *iter) +{ + if (!iter->parent) return NULL; + if (iter == iter->parent->right || !iter->parent->right) return iter->parent; + return rb_postorder_head(iter->parent->right); +} + +/* iterate through the tree */ +#define RB_FOR_EACH(cursor, tree) \ + for ((cursor) = rb_head((tree)->root); (cursor); (cursor) = rb_next(cursor)) + +/* iterate through the tree using a tree entry */ +#define RB_FOR_EACH_ENTRY(elem, tree, type, field) \ + for ((elem) = RB_ENTRY_VALUE(rb_head((tree)->root), type, field); \ + (elem) != RB_ENTRY_VALUE(0, type, field); \ + (elem) = RB_ENTRY_VALUE(rb_next(&elem->field), type, field)) + +/* iterate through the tree using using postorder, making it safe to free the entry */ +#define RB_FOR_EACH_DESTRUCTOR(cursor, cursor2, tree) \ + for ((cursor) = rb_postorder_head((tree)->root); \ + (cursor) && (((cursor2) = rb_postorder_next(cursor)) || 1); \ + (cursor) = (cursor2)) + +/* iterate through the tree using a tree entry and postorder, making it safe to free the entry */ +#define RB_FOR_EACH_ENTRY_DESTRUCTOR(elem, elem2, tree, type, field) \ + for ((elem) = RB_ENTRY_VALUE(rb_postorder_head((tree)->root), type, field); \ + (elem) != WINE_RB_ENTRY_VALUE(0, type, field) \ + && (((elem2) = RB_ENTRY_VALUE(rb_postorder_next(&(elem)->field), type, field)) || 1); \ + (elem) = (elem2)) + + +static inline void rb_postorder(struct rb_tree *tree, rb_traverse_func *callback, void *context) +{ + struct rb_entry *iter, *next; + RB_FOR_EACH_DESTRUCTOR(iter, next, tree) callback(iter, context); +} + +static inline void rb_init(struct rb_tree *tree, rb_compare_func compare) +{ + tree->compare = compare; + tree->root = NULL; +} + +static inline void rb_for_each_entry(struct rb_tree *tree, rb_traverse_func *callback, void *context) +{ + struct rb_entry *iter; + RB_FOR_EACH(iter, tree) callback(iter, context); +} + +static inline void rb_clear(struct rb_tree *tree, rb_traverse_func *callback, void *context) +{ + /* Note that we use postorder here because the callback will likely free the entry. */ + if (callback) rb_postorder(tree, callback, context); + tree->root = NULL; +} + +static inline void rb_destroy(struct rb_tree *tree, rb_traverse_func *callback, void *context) +{ + rb_clear(tree, callback, context); +} + +static inline struct rb_entry *rb_get(const struct rb_tree *tree, const void *key) +{ + struct rb_entry *entry = tree->root; + while (entry) + { + int c = tree->compare(key, entry); + if (!c) return entry; + entry = c < 0 ? entry->left : entry->right; + } + return NULL; +} + +static inline int rb_put(struct rb_tree *tree, const void *key, struct rb_entry *entry) +{ + struct rb_entry **iter = &tree->root, *parent = tree->root; + + while (*iter) + { + int c; + + parent = *iter; + c = tree->compare(key, parent); + if (!c) return -1; + else if (c < 0) iter = &parent->left; + else iter = &parent->right; + } + + entry->flags = RB_FLAG_RED; + entry->parent = parent; + entry->left = NULL; + entry->right = NULL; + *iter = entry; + + while (rb_is_red(entry->parent)) + { + if (entry->parent == entry->parent->parent->left) + { + if (rb_is_red(entry->parent->parent->right)) + { + rb_flip_color(entry->parent->parent); + entry = entry->parent->parent; + } + else + { + if (entry == entry->parent->right) + { + entry = entry->parent; + rb_rotate_left(tree, entry); + } + entry->parent->flags &= ~RB_FLAG_RED; + entry->parent->parent->flags |= RB_FLAG_RED; + rb_rotate_right(tree, entry->parent->parent); + } + } + else + { + if (rb_is_red(entry->parent->parent->left)) + { + rb_flip_color(entry->parent->parent); + entry = entry->parent->parent; + } + else + { + if (entry == entry->parent->left) + { + entry = entry->parent; + rb_rotate_right(tree, entry); + } + entry->parent->flags &= ~RB_FLAG_RED; + entry->parent->parent->flags |= RB_FLAG_RED; + rb_rotate_left(tree, entry->parent->parent); + } + } + } + + tree->root->flags &= ~RB_FLAG_RED; + + return 0; +} + +static inline void rb_remove(struct rb_tree *tree, struct rb_entry *entry) +{ + struct rb_entry *iter, *child, *parent, *w; + int need_fixup; + + if (entry->right && entry->left) + for(iter = entry->right; iter->left; iter = iter->left); + else + iter = entry; + + child = iter->left ? iter->left : iter->right; + + if (!iter->parent) + tree->root = child; + else if (iter == iter->parent->left) + iter->parent->left = child; + else + iter->parent->right = child; + + if (child) child->parent = iter->parent; + parent = iter->parent; + + need_fixup = !rb_is_red(iter); + + if (entry != iter) + { + *iter = *entry; + if (!iter->parent) + tree->root = iter; + else if (entry == iter->parent->left) + iter->parent->left = iter; + else + iter->parent->right = iter; + + if (iter->right) iter->right->parent = iter; + if (iter->left) iter->left->parent = iter; + if (parent == entry) parent = iter; + } + + if (need_fixup) + { + while (parent && !rb_is_red(child)) + { + if (child == parent->left) + { + w = parent->right; + if (rb_is_red(w)) + { + w->flags &= ~RB_FLAG_RED; + parent->flags |= RB_FLAG_RED; + rb_rotate_left(tree, parent); + w = parent->right; + } + if (rb_is_red(w->left) || rb_is_red(w->right)) + { + if (!rb_is_red(w->right)) + { + w->left->flags &= ~RB_FLAG_RED; + w->flags |= RB_FLAG_RED; + rb_rotate_right(tree, w); + w = parent->right; + } + w->flags = (w->flags & ~RB_FLAG_RED) | (parent->flags & RB_FLAG_RED); + parent->flags &= ~RB_FLAG_RED; + if (w->right) + w->right->flags &= ~RB_FLAG_RED; + rb_rotate_left(tree, parent); + child = NULL; + break; + } + } + else + { + w = parent->left; + if (rb_is_red(w)) + { + w->flags &= ~RB_FLAG_RED; + parent->flags |= RB_FLAG_RED; + rb_rotate_right(tree, parent); + w = parent->left; + } + if (rb_is_red(w->left) || rb_is_red(w->right)) + { + if (!rb_is_red(w->left)) + { + w->right->flags &= ~RB_FLAG_RED; + w->flags |= RB_FLAG_RED; + rb_rotate_left(tree, w); + w = parent->left; + } + w->flags = (w->flags & ~RB_FLAG_RED) | (parent->flags & RB_FLAG_RED); + parent->flags &= ~RB_FLAG_RED; + if (w->left) + w->left->flags &= ~RB_FLAG_RED; + rb_rotate_right(tree, parent); + child = NULL; + break; + } + } + w->flags |= RB_FLAG_RED; + child = parent; + parent = child->parent; + } + if (child) child->flags &= ~RB_FLAG_RED; + } + + if (tree->root) tree->root->flags &= ~RB_FLAG_RED; +} + +static inline void rb_remove_key(struct rb_tree *tree, const void *key) +{ + struct rb_entry *entry = rb_get(tree, key); + if (entry) rb_remove(tree, entry); +} + +#endif /* __WINE_WINE_RBTREE_H */ diff --git a/libs/vkd3d/include/private/vkd3d_common.h b/libs/vkd3d/include/private/vkd3d_common.h index 6b5cdf8cf2c..e9bff2fbba3 100644 --- a/libs/vkd3d/include/private/vkd3d_common.h +++ b/libs/vkd3d/include/private/vkd3d_common.h @@ -49,17 +49,38 @@ ((uint32_t)(ch0) | ((uint32_t)(ch1) << 8) \ | ((uint32_t)(ch2) << 16) | ((uint32_t)(ch3) << 24)) +#define TAG_AON9 VKD3D_MAKE_TAG('A', 'o', 'n', '9') +#define TAG_DXBC VKD3D_MAKE_TAG('D', 'X', 'B', 'C') +#define TAG_DXIL VKD3D_MAKE_TAG('D', 'X', 'I', 'L') +#define TAG_ISG1 VKD3D_MAKE_TAG('I', 'S', 'G', '1') +#define TAG_ISGN VKD3D_MAKE_TAG('I', 'S', 'G', 'N') +#define TAG_OSG1 VKD3D_MAKE_TAG('O', 'S', 'G', '1') +#define TAG_OSG5 VKD3D_MAKE_TAG('O', 'S', 'G', '5') +#define TAG_OSGN VKD3D_MAKE_TAG('O', 'S', 'G', 'N') +#define TAG_PCSG VKD3D_MAKE_TAG('P', 'C', 'S', 'G') +#define TAG_PSG1 VKD3D_MAKE_TAG('P', 'S', 'G', '1') +#define TAG_RD11 VKD3D_MAKE_TAG('R', 'D', '1', '1') +#define TAG_RDEF VKD3D_MAKE_TAG('R', 'D', 'E', 'F') +#define TAG_RTS0 VKD3D_MAKE_TAG('R', 'T', 'S', '0') +#define TAG_SDBG VKD3D_MAKE_TAG('S', 'D', 'B', 'G') +#define TAG_SHDR VKD3D_MAKE_TAG('S', 'H', 'D', 'R') +#define TAG_SHEX VKD3D_MAKE_TAG('S', 'H', 'E', 'X') +#define TAG_STAT VKD3D_MAKE_TAG('S', 'T', 'A', 'T') +#define TAG_TEXT VKD3D_MAKE_TAG('T', 'E', 'X', 'T') +#define TAG_XNAP VKD3D_MAKE_TAG('X', 'N', 'A', 'P') +#define TAG_XNAS VKD3D_MAKE_TAG('X', 'N', 'A', 'S') + static inline size_t align(size_t addr, size_t alignment) { return (addr + (alignment - 1)) & ~(alignment - 1); } -#ifdef __GNUC__ +#if defined(__GNUC__) || defined(__clang__) # define VKD3D_NORETURN __attribute__((noreturn)) # ifdef __MINGW_PRINTF_FORMAT # define VKD3D_PRINTF_FUNC(fmt, args) __attribute__((format(__MINGW_PRINTF_FORMAT, fmt, args))) # else -# define VKD3D_PRINTF_FUNC(fmt, args) /* __attribute__((format(printf, fmt, args))) */ +# define VKD3D_PRINTF_FUNC(fmt, args) __attribute__((format(printf, fmt, args))) # endif # define VKD3D_UNUSED __attribute__((unused)) # define VKD3D_UNREACHABLE __builtin_unreachable() @@ -86,7 +107,7 @@ static inline unsigned int vkd3d_popcount(unsigned int v) { #ifdef _MSC_VER return __popcnt(v); -#elif defined(__MINGW32__) +#elif defined(HAVE_BUILTIN_POPCOUNT) return __builtin_popcount(v); #else v -= (v >> 1) & 0x55555555; diff --git a/libs/vkd3d/include/private/vkd3d_test.h b/libs/vkd3d/include/private/vkd3d_test.h new file mode 100644 index 00000000000..081443c4fa6 --- /dev/null +++ b/libs/vkd3d/include/private/vkd3d_test.h @@ -0,0 +1,432 @@ +/* + * Copyright 2016 Józef Kucia for CodeWeavers + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA + */ + +#ifndef __VKD3D_TEST_H +#define __VKD3D_TEST_H + +#include "vkd3d_common.h" +#include +#include +#include +#include +#include +#include +#include + +extern const char *vkd3d_test_name; +extern const char *vkd3d_test_platform; + +static void vkd3d_test_start_todo(bool is_todo); +static int vkd3d_test_loop_todo(void); +static void vkd3d_test_end_todo(void); + +#define START_TEST(name) \ + const char *vkd3d_test_name = #name; \ + static void vkd3d_test_main(int argc, char **argv) + +/* + * Use assert_that() for conditions that should always be true. + * todo_if() and bug_if() do not influence assert_that(). + */ +#define assert_that assert_that_(__LINE__) + +#define ok ok_(__LINE__) + +#define skip skip_(__LINE__) + +#define trace trace_(__LINE__) + +#define assert_that_(line) \ + do { \ + unsigned int vkd3d_line = line; \ + VKD3D_TEST_ASSERT_THAT + +#define VKD3D_TEST_ASSERT_THAT(...) \ + vkd3d_test_assert_that(vkd3d_line, __VA_ARGS__); } while (0) + +#define ok_(line) \ + do { \ + unsigned int vkd3d_line = line; \ + VKD3D_TEST_OK + +#define VKD3D_TEST_OK(...) \ + vkd3d_test_ok(vkd3d_line, __VA_ARGS__); } while (0) + +#define todo_(line) \ + do { \ + unsigned int vkd3d_line = line; \ + VKD3D_TEST_TODO + +#define VKD3D_TEST_TODO(...) \ + vkd3d_test_todo(vkd3d_line, __VA_ARGS__); } while (0) + +#define skip_(line) \ + do { \ + unsigned int vkd3d_line = line; \ + VKD3D_TEST_SKIP + +#define VKD3D_TEST_SKIP(...) \ + vkd3d_test_skip(vkd3d_line, __VA_ARGS__); } while (0) + +#define trace_(line) \ + do { \ + unsigned int vkd3d_line = line; \ + VKD3D_TEST_TRACE + +#define VKD3D_TEST_TRACE(...) \ + vkd3d_test_trace(vkd3d_line, __VA_ARGS__); } while (0) + +#define todo_if(is_todo) \ + for (vkd3d_test_start_todo(is_todo); vkd3d_test_loop_todo(); vkd3d_test_end_todo()) + +#define bug_if(is_bug) \ + for (vkd3d_test_start_bug(is_bug); vkd3d_test_loop_bug(); vkd3d_test_end_bug()) + +#define todo todo_if(true) + +struct vkd3d_test_state +{ + LONG success_count; + LONG failure_count; + LONG skip_count; + LONG todo_count; + LONG todo_success_count; + LONG bug_count; + + unsigned int debug_level; + + unsigned int todo_level; + bool todo_do_loop; + + unsigned int bug_level; + bool bug_do_loop; + bool bug_enabled; + + const char *test_name_filter; + char context[8][128]; + unsigned int context_count; +}; +extern struct vkd3d_test_state vkd3d_test_state; + +static bool +vkd3d_test_platform_is_windows(void) +{ + return !strcmp(vkd3d_test_platform, "windows"); +} + +static inline bool +broken(bool condition) +{ + return condition && vkd3d_test_platform_is_windows(); +} + +static void vkd3d_test_printf(unsigned int line, const char *msg) +{ + unsigned int i; + + printf("%s:%u: ", vkd3d_test_name, line); + for (i = 0; i < vkd3d_test_state.context_count; ++i) + printf("%s: ", vkd3d_test_state.context[i]); + printf("%s", msg); +} + +static void +vkd3d_test_check_assert_that(unsigned int line, bool result, const char *fmt, va_list args) +{ + if (result) + { + InterlockedIncrement(&vkd3d_test_state.success_count); + if (vkd3d_test_state.debug_level > 1) + vkd3d_test_printf(line, "Test succeeded.\n"); + } + else + { + InterlockedIncrement(&vkd3d_test_state.failure_count); + vkd3d_test_printf(line, "Test failed: "); + vprintf(fmt, args); + } +} + +static void VKD3D_PRINTF_FUNC(3, 4) VKD3D_UNUSED +vkd3d_test_assert_that(unsigned int line, bool result, const char *fmt, ...) +{ + va_list args; + + va_start(args, fmt); + vkd3d_test_check_assert_that(line, result, fmt, args); + va_end(args); +} + +static void +vkd3d_test_check_ok(unsigned int line, bool result, const char *fmt, va_list args) +{ + bool is_todo = vkd3d_test_state.todo_level && !vkd3d_test_platform_is_windows(); + bool is_bug = vkd3d_test_state.bug_level && !vkd3d_test_platform_is_windows(); + + if (is_bug && vkd3d_test_state.bug_enabled) + { + InterlockedIncrement(&vkd3d_test_state.bug_count); + if (is_todo) + result = !result; + if (result) + vkd3d_test_printf(line, "Fixed bug: "); + else + vkd3d_test_printf(line, "Bug: "); + vprintf(fmt, args); + } + else if (is_todo) + { + if (result) + { + InterlockedIncrement(&vkd3d_test_state.todo_success_count); + vkd3d_test_printf(line, "Todo succeeded: "); + } + else + { + InterlockedIncrement(&vkd3d_test_state.todo_count); + vkd3d_test_printf(line, "Todo: "); + } + vprintf(fmt, args); + } + else + { + vkd3d_test_check_assert_that(line, result, fmt, args); + } +} + +static void VKD3D_PRINTF_FUNC(3, 4) VKD3D_UNUSED +vkd3d_test_ok(unsigned int line, bool result, const char *fmt, ...) +{ + va_list args; + + va_start(args, fmt); + vkd3d_test_check_ok(line, result, fmt, args); + va_end(args); +} + +static void VKD3D_PRINTF_FUNC(2, 3) VKD3D_UNUSED +vkd3d_test_skip(unsigned int line, const char *fmt, ...) +{ + va_list args; + va_start(args, fmt); + vkd3d_test_printf(line, "Test skipped: "); + vprintf(fmt, args); + va_end(args); + InterlockedIncrement(&vkd3d_test_state.skip_count); +} + +static void VKD3D_PRINTF_FUNC(2, 3) VKD3D_UNUSED +vkd3d_test_trace(unsigned int line, const char *fmt, ...) +{ + va_list args; + va_start(args, fmt); + vkd3d_test_printf(line, ""); + vprintf(fmt, args); + va_end(args); +} + +static void VKD3D_PRINTF_FUNC(1, 2) VKD3D_UNUSED +vkd3d_test_debug(const char *fmt, ...) +{ + char buffer[512]; + va_list args; + int size; + + size = snprintf(buffer, sizeof(buffer), "%s: ", vkd3d_test_name); + if (0 < size && size < sizeof(buffer)) + { + va_start(args, fmt); + vsnprintf(buffer + size, sizeof(buffer) - size, fmt, args); + va_end(args); + } + buffer[sizeof(buffer) - 1] = '\0'; + +#ifdef _WIN32 + OutputDebugStringA(buffer); +#endif + + if (vkd3d_test_state.debug_level > 0) + printf("%s\n", buffer); +} + +#ifndef VKD3D_TEST_NO_DEFS +const char *vkd3d_test_platform = "other"; +struct vkd3d_test_state vkd3d_test_state; + +static void vkd3d_test_main(int argc, char **argv); + +int main(int argc, char **argv) +{ + const char *test_filter = getenv("VKD3D_TEST_FILTER"); + const char *debug_level = getenv("VKD3D_TEST_DEBUG"); + char *test_platform = getenv("VKD3D_TEST_PLATFORM"); + const char *bug = getenv("VKD3D_TEST_BUG"); + + memset(&vkd3d_test_state, 0, sizeof(vkd3d_test_state)); + vkd3d_test_state.debug_level = debug_level ? atoi(debug_level) : 0; + vkd3d_test_state.bug_enabled = bug ? atoi(bug) : true; + vkd3d_test_state.test_name_filter = test_filter; + + if (test_platform) + { + test_platform = strdup(test_platform); + vkd3d_test_platform = test_platform; + } + + if (vkd3d_test_state.debug_level > 1) + printf("Test platform: '%s'.\n", vkd3d_test_platform); + + vkd3d_test_main(argc, argv); + + printf("%s: %lu tests executed (%lu failures, %lu skipped, %lu todo, %lu bugs).\n", + vkd3d_test_name, + (unsigned long)(vkd3d_test_state.success_count + + vkd3d_test_state.failure_count + vkd3d_test_state.todo_count + + vkd3d_test_state.todo_success_count), + (unsigned long)(vkd3d_test_state.failure_count + + vkd3d_test_state.todo_success_count), + (unsigned long)vkd3d_test_state.skip_count, + (unsigned long)vkd3d_test_state.todo_count, + (unsigned long)vkd3d_test_state.bug_count); + + if (test_platform) + free(test_platform); + + return vkd3d_test_state.failure_count || vkd3d_test_state.todo_success_count; +} + +#ifdef _WIN32 +static char *vkd3d_test_strdupWtoA(WCHAR *str) +{ + char *out; + int len; + + if (!(len = WideCharToMultiByte(CP_ACP, 0, str, -1, NULL, 0, NULL, NULL))) + return NULL; + if (!(out = malloc(len))) + return NULL; + WideCharToMultiByte(CP_ACP, 0, str, -1, out, len, NULL, NULL); + + return out; +} + +static bool running_under_wine(void) +{ + HMODULE module = GetModuleHandleA("ntdll.dll"); + return module && GetProcAddress(module, "wine_server_call"); +} + +int wmain(int argc, WCHAR **wargv) +{ + char **argv; + int i, ret; + + argv = malloc(argc * sizeof(*argv)); + assert(argv); + for (i = 0; i < argc; ++i) + { + if (!(argv[i] = vkd3d_test_strdupWtoA(wargv[i]))) + break; + } + assert(i == argc); + + vkd3d_test_platform = running_under_wine() ? "wine" : "windows"; + + ret = main(argc, argv); + + for (i = 0; i < argc; ++i) + free(argv[i]); + free(argv); + + return ret; +} +#endif /* _WIN32 */ +#endif /* VKD3D_TEST_NO_DEFS */ + +typedef void (*vkd3d_test_pfn)(void); + +static inline void vkd3d_run_test(const char *name, vkd3d_test_pfn test_pfn) +{ + if (vkd3d_test_state.test_name_filter && !strstr(name, vkd3d_test_state.test_name_filter)) + return; + + vkd3d_test_debug("%s", name); + test_pfn(); +} + +static inline void vkd3d_test_start_todo(bool is_todo) +{ + vkd3d_test_state.todo_level = (vkd3d_test_state.todo_level << 1) | is_todo; + vkd3d_test_state.todo_do_loop = true; +} + +static inline int vkd3d_test_loop_todo(void) +{ + bool do_loop = vkd3d_test_state.todo_do_loop; + vkd3d_test_state.todo_do_loop = false; + return do_loop; +} + +static inline void vkd3d_test_end_todo(void) +{ + vkd3d_test_state.todo_level >>= 1; +} + +static inline void vkd3d_test_start_bug(bool is_bug) +{ + vkd3d_test_state.bug_level = (vkd3d_test_state.bug_level << 1) | is_bug; + vkd3d_test_state.bug_do_loop = true; +} + +static inline int vkd3d_test_loop_bug(void) +{ + bool do_loop = vkd3d_test_state.bug_do_loop; + vkd3d_test_state.bug_do_loop = false; + return do_loop; +} + +static inline void vkd3d_test_end_bug(void) +{ + vkd3d_test_state.bug_level >>= 1; +} + +static inline void vkd3d_test_push_context(const char *fmt, ...) +{ + va_list args; + + if (vkd3d_test_state.context_count < ARRAY_SIZE(vkd3d_test_state.context)) + { + va_start(args, fmt); + vsnprintf(vkd3d_test_state.context[vkd3d_test_state.context_count], + sizeof(vkd3d_test_state.context), fmt, args); + va_end(args); + vkd3d_test_state.context[vkd3d_test_state.context_count][sizeof(vkd3d_test_state.context[0]) - 1] = '\0'; + } + ++vkd3d_test_state.context_count; +} + +static inline void vkd3d_test_pop_context(void) +{ + if (vkd3d_test_state.context_count) + --vkd3d_test_state.context_count; +} + +#define run_test(test_pfn) \ + vkd3d_run_test(#test_pfn, test_pfn) + +#endif /* __VKD3D_TEST_H */ diff --git a/libs/vkd3d/include/vkd3d_d3d9types.h b/libs/vkd3d/include/vkd3d_d3d9types.h new file mode 100644 index 00000000000..75d0461409d --- /dev/null +++ b/libs/vkd3d/include/vkd3d_d3d9types.h @@ -0,0 +1,237 @@ +/* + * Copyright 2002-2003 Jason Edmeades + * Copyright 2002-2003 Raphael Junqueira + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA + */ + +#ifndef __VKD3D_D3D9TYPES_H +#define __VKD3D_D3D9TYPES_H +#ifndef _d3d9TYPES_H_ + +#ifndef MAKEFOURCC +#define MAKEFOURCC(ch0, ch1, ch2, ch3) \ + ((DWORD)(BYTE)(ch0) | ((DWORD)(BYTE)(ch1) << 8) | \ + ((DWORD)(BYTE)(ch2) << 16) | ((DWORD)(BYTE)(ch3) << 24 )) +#endif + +#define D3DSI_INSTLENGTH_SHIFT 24 + +#define D3DSP_DCL_USAGE_SHIFT 0 +#define D3DSP_DCL_USAGEINDEX_SHIFT 16 +#define D3DSP_DSTMOD_SHIFT 20 + +#define D3DSP_SRCMOD_SHIFT 24 + +#define D3DSP_REGTYPE_SHIFT 28 +#define D3DSP_REGTYPE_SHIFT2 8 +#define D3DSP_REGTYPE_MASK (0x7 << D3DSP_REGTYPE_SHIFT) +#define D3DSP_REGTYPE_MASK2 0x00001800 + +#define D3DSP_WRITEMASK_0 0x00010000 +#define D3DSP_WRITEMASK_1 0x00020000 +#define D3DSP_WRITEMASK_2 0x00040000 +#define D3DSP_WRITEMASK_3 0x00080000 +#define D3DSP_WRITEMASK_ALL 0x000f0000 + +#define D3DPS_VERSION(major, minor) (0xffff0000 | ((major) << 8) | (minor)) +#define D3DVS_VERSION(major, minor) (0xfffe0000 | ((major) << 8) | (minor)) + +typedef enum _D3DDECLUSAGE +{ + D3DDECLUSAGE_POSITION = 0x0, + D3DDECLUSAGE_BLENDWEIGHT = 0x1, + D3DDECLUSAGE_BLENDINDICES = 0x2, + D3DDECLUSAGE_NORMAL = 0x3, + D3DDECLUSAGE_PSIZE = 0x4, + D3DDECLUSAGE_TEXCOORD = 0x5, + D3DDECLUSAGE_TANGENT = 0x6, + D3DDECLUSAGE_BINORMAL = 0x7, + D3DDECLUSAGE_TESSFACTOR = 0x8, + D3DDECLUSAGE_POSITIONT = 0x9, + D3DDECLUSAGE_COLOR = 0xa, + D3DDECLUSAGE_FOG = 0xb, + D3DDECLUSAGE_DEPTH = 0xc, + D3DDECLUSAGE_SAMPLE = 0xd, +} D3DDECLUSAGE; + +typedef enum _D3DSHADER_INSTRUCTION_OPCODE_TYPE +{ + D3DSIO_NOP = 0x00, + D3DSIO_MOV = 0x01, + D3DSIO_ADD = 0x02, + D3DSIO_SUB = 0x03, + D3DSIO_MAD = 0x04, + D3DSIO_MUL = 0x05, + D3DSIO_RCP = 0x06, + D3DSIO_RSQ = 0x07, + D3DSIO_DP3 = 0x08, + D3DSIO_DP4 = 0x09, + D3DSIO_MIN = 0x0a, + D3DSIO_MAX = 0x0b, + D3DSIO_SLT = 0x0c, + D3DSIO_SGE = 0x0d, + D3DSIO_EXP = 0x0e, + D3DSIO_LOG = 0x0f, + D3DSIO_LIT = 0x10, + D3DSIO_DST = 0x11, + D3DSIO_LRP = 0x12, + D3DSIO_FRC = 0x13, + D3DSIO_M4x4 = 0x14, + D3DSIO_M4x3 = 0x15, + D3DSIO_M3x4 = 0x16, + D3DSIO_M3x3 = 0x17, + D3DSIO_M3x2 = 0x18, + D3DSIO_CALL = 0x19, + D3DSIO_CALLNZ = 0x1a, + D3DSIO_LOOP = 0x1b, + D3DSIO_RET = 0x1c, + D3DSIO_ENDLOOP = 0x1d, + D3DSIO_LABEL = 0x1e, + D3DSIO_DCL = 0x1f, + D3DSIO_POW = 0x20, + D3DSIO_CRS = 0x21, + D3DSIO_SGN = 0x22, + D3DSIO_ABS = 0x23, + D3DSIO_NRM = 0x24, + D3DSIO_SINCOS = 0x25, + D3DSIO_REP = 0x26, + D3DSIO_ENDREP = 0x27, + D3DSIO_IF = 0x28, + D3DSIO_IFC = 0x29, + D3DSIO_ELSE = 0x2a, + D3DSIO_ENDIF = 0x2b, + D3DSIO_BREAK = 0x2c, + D3DSIO_BREAKC = 0x2d, + D3DSIO_MOVA = 0x2e, + D3DSIO_DEFB = 0x2f, + D3DSIO_DEFI = 0x30, + + D3DSIO_TEXCOORD = 0x40, + D3DSIO_TEXKILL = 0x41, + D3DSIO_TEX = 0x42, + D3DSIO_TEXBEM = 0x43, + D3DSIO_TEXBEML = 0x44, + D3DSIO_TEXREG2AR = 0x45, + D3DSIO_TEXREG2GB = 0x46, + D3DSIO_TEXM3x2PAD = 0x47, + D3DSIO_TEXM3x2TEX = 0x48, + D3DSIO_TEXM3x3PAD = 0x49, + D3DSIO_TEXM3x3TEX = 0x4a, + D3DSIO_TEXM3x3DIFF = 0x4b, + D3DSIO_TEXM3x3SPEC = 0x4c, + D3DSIO_TEXM3x3VSPEC = 0x4d, + D3DSIO_EXPP = 0x4e, + D3DSIO_LOGP = 0x4f, + D3DSIO_CND = 0x50, + D3DSIO_DEF = 0x51, + D3DSIO_TEXREG2RGB = 0x52, + D3DSIO_TEXDP3TEX = 0x53, + D3DSIO_TEXM3x2DEPTH = 0x54, + D3DSIO_TEXDP3 = 0x55, + D3DSIO_TEXM3x3 = 0x56, + D3DSIO_TEXDEPTH = 0x57, + D3DSIO_CMP = 0x58, + D3DSIO_BEM = 0x59, + D3DSIO_DP2ADD = 0x5a, + D3DSIO_DSX = 0x5b, + D3DSIO_DSY = 0x5c, + D3DSIO_TEXLDD = 0x5d, + D3DSIO_SETP = 0x5e, + D3DSIO_TEXLDL = 0x5f, + D3DSIO_BREAKP = 0x60, + + D3DSIO_PHASE = 0xfffd, + D3DSIO_COMMENT = 0xfffe, + D3DSIO_END = 0xffff, + + D3DSIO_FORCE_DWORD = 0x7fffffff, +} D3DSHADER_INSTRUCTION_OPCODE_TYPE; + +typedef enum _D3DSHADER_PARAM_DSTMOD_TYPE +{ + D3DSPDM_NONE = 0 << D3DSP_DSTMOD_SHIFT, + D3DSPDM_SATURATE = 1 << D3DSP_DSTMOD_SHIFT, + D3DSPDM_PARTIALPRECISION = 2 << D3DSP_DSTMOD_SHIFT, + D3DSPDM_MSAMPCENTROID = 4 << D3DSP_DSTMOD_SHIFT, + + D3DSPDM_FORCE_DWORD = 0x7fffffff, +} D3DSHADER_PARAM_DSTMOD_TYPE; + +typedef enum _D3DSHADER_PARAM_REGISTER_TYPE +{ + D3DSPR_TEMP = 0x00, + D3DSPR_INPUT = 0x01, + D3DSPR_CONST = 0x02, + D3DSPR_ADDR = 0x03, + D3DSPR_TEXTURE = 0x03, + D3DSPR_RASTOUT = 0x04, + D3DSPR_ATTROUT = 0x05, + D3DSPR_TEXCRDOUT = 0x06, + D3DSPR_OUTPUT = 0x06, + D3DSPR_CONSTINT = 0x07, + D3DSPR_COLOROUT = 0x08, + D3DSPR_DEPTHOUT = 0x09, + D3DSPR_SAMPLER = 0x0a, + D3DSPR_CONST2 = 0x0b, + D3DSPR_CONST3 = 0x0c, + D3DSPR_CONST4 = 0x0d, + D3DSPR_CONSTBOOL = 0x0e, + D3DSPR_LOOP = 0x0f, + D3DSPR_TEMPFLOAT16 = 0x10, + D3DSPR_MISCTYPE = 0x11, + D3DSPR_LABEL = 0x12, + D3DSPR_PREDICATE = 0x13, + + D3DSPR_FORCE_DWORD = 0x7fffffff, +} D3DSHADER_PARAM_REGISTER_TYPE; + +typedef enum _D3DSHADER_PARAM_SRCMOD_TYPE +{ + D3DSPSM_NONE = 0x0 << D3DSP_SRCMOD_SHIFT, + D3DSPSM_NEG = 0x1 << D3DSP_SRCMOD_SHIFT, + D3DSPSM_BIAS = 0x2 << D3DSP_SRCMOD_SHIFT, + D3DSPSM_BIASNEG = 0x3 << D3DSP_SRCMOD_SHIFT, + D3DSPSM_SIGN = 0x4 << D3DSP_SRCMOD_SHIFT, + D3DSPSM_SIGNNEG = 0x5 << D3DSP_SRCMOD_SHIFT, + D3DSPSM_COMP = 0x6 << D3DSP_SRCMOD_SHIFT, + D3DSPSM_X2 = 0x7 << D3DSP_SRCMOD_SHIFT, + D3DSPSM_X2NEG = 0x8 << D3DSP_SRCMOD_SHIFT, + D3DSPSM_DZ = 0x9 << D3DSP_SRCMOD_SHIFT, + D3DSPSM_DW = 0xa << D3DSP_SRCMOD_SHIFT, + D3DSPSM_ABS = 0xb << D3DSP_SRCMOD_SHIFT, + D3DSPSM_ABSNEG = 0xc << D3DSP_SRCMOD_SHIFT, + D3DSPSM_NOT = 0xd << D3DSP_SRCMOD_SHIFT, + + D3DSPSM_FORCE_DWORD = 0x7fffffff, +} D3DSHADER_PARAM_SRCMOD_TYPE; + +typedef enum _D3DSHADER_MISCTYPE_OFFSETS +{ + D3DSMO_POSITION = 0x0, + D3DSMO_FACE = 0x1, +} D3DSHADER_MISCTYPE_OFFSETS; + +typedef enum _D3DVS_RASTOUT_OFFSETS +{ + D3DSRO_POSITION = 0x0, + D3DSRO_FOG = 0x1, + D3DSRO_POINT_SIZE = 0x2, + + D3DSRO_FORCE_DWORD = 0x7fffffff, +} D3DVS_RASTOUT_OFFSETS; + +#endif /* _d3d9TYPES_H_ */ +#endif /* __VKD3D_D3D9TYPES_H */ diff --git a/libs/vkd3d/include/vkd3d_d3dcompiler.h b/libs/vkd3d/include/vkd3d_d3dcompiler.h new file mode 100644 index 00000000000..c78cf25b955 --- /dev/null +++ b/libs/vkd3d/include/vkd3d_d3dcompiler.h @@ -0,0 +1,92 @@ +/* + * Copyright 2010 Matteo Bruni for CodeWeavers + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA + */ + +#ifndef __VKD3D_D3DCOMPILER_H +#define __VKD3D_D3DCOMPILER_H +#ifndef __D3DCOMPILER_H__ + +#include + +#define D3DCOMPILE_DEBUG 0x00000001 +#define D3DCOMPILE_SKIP_VALIDATION 0x00000002 +#define D3DCOMPILE_SKIP_OPTIMIZATION 0x00000004 +#define D3DCOMPILE_PACK_MATRIX_ROW_MAJOR 0x00000008 +#define D3DCOMPILE_PACK_MATRIX_COLUMN_MAJOR 0x00000010 +#define D3DCOMPILE_PARTIAL_PRECISION 0x00000020 +#define D3DCOMPILE_FORCE_VS_SOFTWARE_NO_OPT 0x00000040 +#define D3DCOMPILE_FORCE_PS_SOFTWARE_NO_OPT 0x00000080 +#define D3DCOMPILE_NO_PRESHADER 0x00000100 +#define D3DCOMPILE_AVOID_FLOW_CONTROL 0x00000200 +#define D3DCOMPILE_PREFER_FLOW_CONTROL 0x00000400 +#define D3DCOMPILE_ENABLE_STRICTNESS 0x00000800 +#define D3DCOMPILE_ENABLE_BACKWARDS_COMPATIBILITY 0x00001000 +#define D3DCOMPILE_IEEE_STRICTNESS 0x00002000 +#define D3DCOMPILE_OPTIMIZATION_LEVEL0 0x00004000 +#define D3DCOMPILE_OPTIMIZATION_LEVEL1 0x00000000 +#define D3DCOMPILE_OPTIMIZATION_LEVEL2 0x0000c000 +#define D3DCOMPILE_OPTIMIZATION_LEVEL3 0x00008000 +#define D3DCOMPILE_RESERVED16 0x00010000 +#define D3DCOMPILE_RESERVED17 0x00020000 +#define D3DCOMPILE_WARNINGS_ARE_ERRORS 0x00040000 +#define D3DCOMPILE_RESOURCES_MAY_ALIAS 0x00080000 +#define D3DCOMPILE_ENABLE_UNBOUNDED_DESCRIPTOR_TABLES 0x00100000 +#define D3DCOMPILE_ALL_RESOURCES_BOUND 0x00200000 +#define D3DCOMPILE_DEBUG_NAME_FOR_SOURCE 0x00400000 +#define D3DCOMPILE_DEBUG_NAME_FOR_BINARY 0x00800000 + +#define D3DCOMPILE_EFFECT_CHILD_EFFECT 0x00000001 +#define D3DCOMPILE_EFFECT_ALLOW_SLOW_OPS 0x00000002 + +#define D3DCOMPILE_FLAGS2_FORCE_ROOT_SIGNATURE_LATEST 0x00000000 +#define D3DCOMPILE_FLAGS2_FORCE_ROOT_SIGNATURE_1_0 0x00000010 +#define D3DCOMPILE_FLAGS2_FORCE_ROOT_SIGNATURE_1_1 0x00000020 + +#define D3DCOMPILE_SECDATA_MERGE_UAV_SLOTS 0x00000001 +#define D3DCOMPILE_SECDATA_PRESERVE_TEMPLATE_SLOTS 0x00000002 +#define D3DCOMPILE_SECDATA_REQUIRE_TEMPLATE_MATCH 0x00000004 + +typedef enum D3DCOMPILER_STRIP_FLAGS +{ + D3DCOMPILER_STRIP_REFLECTION_DATA = 0x00000001, + D3DCOMPILER_STRIP_DEBUG_INFO = 0x00000002, + D3DCOMPILER_STRIP_TEST_BLOBS = 0x00000004, + D3DCOMPILER_STRIP_PRIVATE_DATA = 0x00000008, + D3DCOMPILER_STRIP_ROOT_SIGNATURE = 0x00000010, + D3DCOMPILER_STRIP_FORCE_DWORD = 0x7fffffff, +} D3DCOMPILER_STRIP_FLAGS; + +HRESULT WINAPI D3DCompile(const void *data, SIZE_T data_size, const char *filename, + const D3D_SHADER_MACRO *macros, ID3DInclude *include, const char *entrypoint, + const char *profile, UINT flags, UINT effect_flags, ID3DBlob **shader, ID3DBlob **error_messages); +HRESULT WINAPI D3DCompile2(const void *data, SIZE_T data_size, const char *filename, + const D3D_SHADER_MACRO *macros, ID3DInclude *include, const char *entrypoint, + const char *profile, UINT flags, UINT effect_flags, UINT secondary_flags, + const void *secondary_data, SIZE_T secondary_data_size, ID3DBlob **shader, + ID3DBlob **error_messages); +HRESULT WINAPI D3DCreateBlob(SIZE_T size, ID3DBlob **blob); +HRESULT WINAPI D3DGetBlobPart(const void *data, SIZE_T data_size, D3D_BLOB_PART part, UINT flags, ID3DBlob **blob); +HRESULT WINAPI D3DGetDebugInfo(const void *data, SIZE_T data_size, ID3DBlob **blob); +HRESULT WINAPI D3DGetInputAndOutputSignatureBlob(const void *data, SIZE_T data_size, ID3DBlob **blob); +HRESULT WINAPI D3DGetInputSignatureBlob(const void *data, SIZE_T data_size, ID3DBlob **blob); +HRESULT WINAPI D3DGetOutputSignatureBlob(const void *data, SIZE_T data_size, ID3DBlob **blob); +HRESULT WINAPI D3DPreprocess(const void *data, SIZE_T size, const char *filename, const D3D_SHADER_MACRO *macros, + ID3DInclude *include, ID3DBlob **shader, ID3DBlob **error_messages); +HRESULT WINAPI D3DStripShader(const void *data, SIZE_T data_size, UINT flags, ID3DBlob **blob); + +#endif /* __D3DCOMPILER_H__ */ +#endif /* __VKD3D_D3DCOMPILER_H */ diff --git a/libs/vkd3d/include/vkd3d_d3dcompiler_types.h b/libs/vkd3d/include/vkd3d_d3dcompiler_types.h new file mode 100644 index 00000000000..b3a47cdd912 --- /dev/null +++ b/libs/vkd3d/include/vkd3d_d3dcompiler_types.h @@ -0,0 +1,45 @@ +/* + * Copyright 2010 Matteo Bruni for CodeWeavers + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA + */ + +#ifndef __VKD3D_D3DCOMPILER_TYPES_H +#define __VKD3D_D3DCOMPILER_TYPES_H +#ifndef __D3DCOMPILER_H__ + +typedef enum D3D_BLOB_PART +{ + D3D_BLOB_INPUT_SIGNATURE_BLOB, + D3D_BLOB_OUTPUT_SIGNATURE_BLOB, + D3D_BLOB_INPUT_AND_OUTPUT_SIGNATURE_BLOB, + D3D_BLOB_PATCH_CONSTANT_SIGNATURE_BLOB, + D3D_BLOB_ALL_SIGNATURE_BLOB, + D3D_BLOB_DEBUG_INFO, + D3D_BLOB_LEGACY_SHADER, + D3D_BLOB_XNA_PREPASS_SHADER, + D3D_BLOB_XNA_SHADER, + D3D_BLOB_PDB, + D3D_BLOB_PRIVATE_DATA, + D3D_BLOB_ROOT_SIGNATURE, + D3D_BLOB_DEBUG_NAME, + D3D_BLOB_TEST_ALTERNATE_SHADER = 0x8000, + D3D_BLOB_TEST_COMPILE_DETAILS, + D3D_BLOB_TEST_COMPILE_PERF, + D3D_BLOB_TEST_COMPILE_REPORT +} D3D_BLOB_PART; + +#endif /* __D3DCOMPILER_H__ */ +#endif /* __VKD3D_D3DCOMPILER_TYPES_H */ diff --git a/libs/vkd3d/include/vkd3d_shader.h b/libs/vkd3d/include/vkd3d_shader.h index 01356ce3931..66a7de9b571 100644 --- a/libs/vkd3d/include/vkd3d_shader.h +++ b/libs/vkd3d/include/vkd3d_shader.h @@ -96,6 +96,11 @@ enum vkd3d_shader_structure_type * \since 1.9 */ VKD3D_SHADER_STRUCTURE_TYPE_VARYING_MAP_INFO, + /** + * The structure is a vkd3d_shader_scan_combined_resource_sampler_info structure. + * \since 1.10 + */ + VKD3D_SHADER_STRUCTURE_TYPE_SCAN_COMBINED_RESOURCE_SAMPLER_INFO, VKD3D_FORCE_32_BIT_ENUM(VKD3D_SHADER_STRUCTURE_TYPE), }; @@ -154,6 +159,40 @@ enum vkd3d_shader_compile_option_pack_matrix_order VKD3D_FORCE_32_BIT_ENUM(VKD3D_SHADER_COMPILE_OPTION_PACK_MATRIX_ORDER), }; +/** Individual options to enable various backward compatibility features. \since 1.10 */ +enum vkd3d_shader_compile_option_backward_compatibility +{ + /** + * Causes compiler to convert SM1-3 semantics to corresponding System Value semantics, + * when compiling HLSL sources for SM4+ targets. + * + * This option does the following conversions: + * + * - POSITION to SV_Position for vertex shader outputs, pixel shader inputs, + * and geometry shader inputs and outputs; + * - COLORN to SV_TargetN for pixel shader outputs; + * - DEPTH to SV_Depth for pixel shader outputs. + */ + VKD3D_SHADER_COMPILE_OPTION_BACKCOMPAT_MAP_SEMANTIC_NAMES = 0x00000001, + + VKD3D_FORCE_32_BIT_ENUM(VKD3D_SHADER_COMPILE_OPTION_BACKWARD_COMPATIBILITY), +}; + +/** + * Determines the origin of fragment coordinates. + * + * \since 1.10 + */ +enum vkd3d_shader_compile_option_fragment_coordinate_origin +{ + /** Fragment coordinates originate from the upper-left. This is the + * default; it's also the only value supported by Vulkan environments. */ + VKD3D_SHADER_COMPILE_OPTION_FRAGMENT_COORDINATE_ORIGIN_UPPER_LEFT = 0x00000000, + /** Fragment coordinates originate from the lower-left. This matches the + * traditional behaviour of OpenGL environments. */ + VKD3D_SHADER_COMPILE_OPTION_FRAGMENT_COORDINATE_ORIGIN_LOWER_LEFT = 0x00000001, +}; + enum vkd3d_shader_compile_option_name { /** @@ -193,6 +232,24 @@ enum vkd3d_shader_compile_option_name * \since 1.9 */ VKD3D_SHADER_COMPILE_OPTION_PACK_MATRIX_ORDER = 0x00000007, + /** + * This option is used to enable various backward compatibility features. + * + * \a value is a mask of values from enum vkd3d_shader_compile_option_backward_compatibility. + * + * \since 1.10 + */ + VKD3D_SHADER_COMPILE_OPTION_BACKWARD_COMPATIBILITY = 0x00000008, + /** + * This option specifies the origin of fragment coordinates for SPIR-V + * targets. + * + * \a value is a member of enum + * vkd3d_shader_compile_option_fragment_coordinate_origin. + * + * \since 1.10 + */ + VKD3D_SHADER_COMPILE_OPTION_FRAGMENT_COORDINATE_ORIGIN = 0x00000009, VKD3D_FORCE_32_BIT_ENUM(VKD3D_SHADER_COMPILE_OPTION_NAME), }; @@ -1383,19 +1440,33 @@ struct vkd3d_shader_descriptor_info * * This structure extends vkd3d_shader_compile_info. * - * When scanning a legacy Direct3D shader, vkd3d-shader enumerates each - * constant register set used by the shader as a single constant buffer - * descriptor, as follows: - * - The \ref vkd3d_shader_descriptor_info.type field is set to - * VKD3D_SHADER_DESCRIPTOR_TYPE_CBV. - * - The \ref vkd3d_shader_descriptor_info.register_space field is set to zero. - * - The \ref vkd3d_shader_descriptor_info.register_index field is set to a - * member of enum vkd3d_shader_d3dbc_constant_register denoting which set - * is used. - * - The \ref vkd3d_shader_descriptor_info.count field is set to one. - * - * In summary, there may be up to three such descriptors, one for each register - * set used by the shader: float, integer, and boolean. + * When scanning a legacy Direct3D shader, vkd3d-shader enumerates descriptors + * as follows: + * + * - Each constant register set used by the shader is scanned as a single + * constant buffer descriptor. + * There may therefore be up to three such descriptors, one for each register + * set used by the shader: float, integer, and boolean. + * The fields are set as follows: + * * The \ref vkd3d_shader_descriptor_info.type field is set to + * VKD3D_SHADER_DESCRIPTOR_TYPE_CBV. + * * The \ref vkd3d_shader_descriptor_info.register_space field is set to zero. + * * The \ref vkd3d_shader_descriptor_info.register_index field is set to a + * member of enum vkd3d_shader_d3dbc_constant_register denoting which set + * is used. + * * The \ref vkd3d_shader_descriptor_info.count field is set to one. + * - Each sampler used by the shader is scanned as two separate descriptors, + * one representing the texture, and one representing the sampler state. + * If desired, these may be mapped back into a single combined sampler using + * struct vkd3d_shader_combined_resource_sampler. + * The fields are set as follows: + * * The \ref vkd3d_shader_descriptor_info.type field is set to + * VKD3D_SHADER_DESCRIPTOR_TYPE_SRV and VKD3D_SHADER_DESCRIPTOR_TYPE_SAMPLER + * respectively. + * * The \ref vkd3d_shader_descriptor_info.register_space field is set to zero. + * * The \ref vkd3d_shader_descriptor_info.register_index field is set to the + * binding index of the original sampler, for both descriptors. + * * The \ref vkd3d_shader_descriptor_info.count field is set to one. */ struct vkd3d_shader_scan_descriptor_info { @@ -1412,6 +1483,53 @@ struct vkd3d_shader_scan_descriptor_info unsigned int descriptor_count; }; +/** + * This structure describes a single resource-sampler pair. It is returned as + * part of struct vkd3d_shader_scan_combined_resource_sampler_info. + * + * \since 1.10 + */ +struct vkd3d_shader_combined_resource_sampler_info +{ + unsigned int resource_space; + unsigned int resource_index; + unsigned int sampler_space; + unsigned int sampler_index; +}; + +/** + * A chained structure describing the resource-sampler pairs used by a shader. + * + * This structure extends vkd3d_shader_compile_info. + * + * The information returned in this structure can be used to populate the + * \ref vkd3d_shader_interface_info.combined_samplers field. This is + * particularly useful when targeting environments without separate binding + * points for samplers and resources, like OpenGL. + * + * No resource-sampler pairs are returned for dynamic accesses to + * resource/sampler descriptor arrays, as can occur in Direct3D shader model + * 5.1 shaders. + * + * Members of this structure are allocated by vkd3d-shader and should be freed + * with vkd3d_shader_free_scan_combined_resource_sampler_info() when no longer + * needed. + * + * \since 1.10 + */ +struct vkd3d_shader_scan_combined_resource_sampler_info +{ + /** Must be set to VKD3D_SHADER_STRUCTURE_TYPE_SCAN_COMBINED_RESOURCE_SAMPLER_INFO. */ + enum vkd3d_shader_structure_type type; + /** Optional pointer to a structure containing further parameters. */ + const void *next; + + /** Pointer to an array of resource-sampler pairs. */ + struct vkd3d_shader_combined_resource_sampler_info *combined_samplers; + /** The number of resource-sampler pairs in \ref combined_samplers. */ + unsigned int combined_sampler_count; +}; + /** * Data type of a shader varying, returned as part of struct * vkd3d_shader_signature_element. @@ -2313,6 +2431,21 @@ VKD3D_SHADER_API void vkd3d_shader_build_varying_map(const struct vkd3d_shader_s const struct vkd3d_shader_signature *input_signature, unsigned int *count, struct vkd3d_shader_varying_map *varyings); +/** + * Free members of struct vkd3d_shader_scan_combined_resource_sampler_info + * allocated by vkd3d_shader_scan(). + * + * This function may free members of + * vkd3d_shader_scan_combined_resource_sampler_info, but does not free the + * structure itself. + * + * \param info Combined resource-sampler information to free. + * + * \since 1.10 + */ +VKD3D_SHADER_API void vkd3d_shader_free_scan_combined_resource_sampler_info( + struct vkd3d_shader_scan_combined_resource_sampler_info *info); + #endif /* VKD3D_SHADER_NO_PROTOTYPES */ /** Type of vkd3d_shader_get_version(). */ @@ -2385,6 +2518,10 @@ typedef void (*PFN_vkd3d_shader_build_varying_map)(const struct vkd3d_shader_sig /** Type of vkd3d_shader_free_scan_signature_info(). \since 1.9 */ typedef void (*PFN_vkd3d_shader_free_scan_signature_info)(struct vkd3d_shader_scan_signature_info *info); +/** Type of vkd3d_shader_free_scan_combined_resource_sampler_info(). \since 1.10 */ +typedef void (*PFN_vkd3d_shader_free_scan_combined_resource_sampler_info)( + struct vkd3d_shader_scan_combined_resource_sampler_info *info); + #ifdef __cplusplus } #endif /* __cplusplus */ diff --git a/libs/vkd3d/include/vkd3d_utils.h b/libs/vkd3d/include/vkd3d_utils.h new file mode 100644 index 00000000000..686ddf386e7 --- /dev/null +++ b/libs/vkd3d/include/vkd3d_utils.h @@ -0,0 +1,123 @@ +/* + * Copyright 2016 Józef Kucia for CodeWeavers + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA + */ + +#ifndef __VKD3D_UTILS_H +#define __VKD3D_UTILS_H + +#include +#include + +#ifndef VKD3D_UTILS_API_VERSION +#define VKD3D_UTILS_API_VERSION VKD3D_API_VERSION_1_0 +#endif + +#ifdef __cplusplus +extern "C" { +#endif /* __cplusplus */ + +/** + * \file vkd3d_utils.h + * + * This file contains definitions for the vkd3d-utils library. + * + * The vkd3d-utils library is a collections of routines to ease the + * porting of a Direct3D 12 application to vkd3d. + * + * \since 1.0 + */ + +#define VKD3D_WAIT_OBJECT_0 (0) +#define VKD3D_WAIT_TIMEOUT (1) +#define VKD3D_WAIT_FAILED (~0u) +#define VKD3D_INFINITE (~0u) + +#ifdef LIBVKD3D_UTILS_SOURCE +# define VKD3D_UTILS_API VKD3D_EXPORT +#else +# define VKD3D_UTILS_API VKD3D_IMPORT +#endif + +/* 1.0 */ +VKD3D_UTILS_API HANDLE vkd3d_create_event(void); +VKD3D_UTILS_API HRESULT vkd3d_signal_event(HANDLE event); +VKD3D_UTILS_API unsigned int vkd3d_wait_event(HANDLE event, unsigned int milliseconds); +VKD3D_UTILS_API void vkd3d_destroy_event(HANDLE event); + +#define D3D12CreateDevice(a, b, c, d) D3D12CreateDeviceVKD3D(a, b, c, d, VKD3D_UTILS_API_VERSION) +VKD3D_UTILS_API HRESULT WINAPI D3D12CreateRootSignatureDeserializer( + const void *data, SIZE_T data_size, REFIID iid, void **deserializer); +VKD3D_UTILS_API HRESULT WINAPI D3D12GetDebugInterface(REFIID iid, void **debug); +VKD3D_UTILS_API HRESULT WINAPI D3D12SerializeRootSignature(const D3D12_ROOT_SIGNATURE_DESC *desc, + D3D_ROOT_SIGNATURE_VERSION version, ID3DBlob **blob, ID3DBlob **error_blob); + +/* 1.2 */ +VKD3D_UTILS_API HRESULT WINAPI D3D12CreateDeviceVKD3D(IUnknown *adapter, D3D_FEATURE_LEVEL feature_level, + REFIID iid, void **device, enum vkd3d_api_version api_version); +VKD3D_UTILS_API HRESULT WINAPI D3D12CreateVersionedRootSignatureDeserializer(const void *data, + SIZE_T data_size, REFIID iid, void **deserializer); +VKD3D_UTILS_API HRESULT WINAPI D3D12SerializeVersionedRootSignature(const D3D12_VERSIONED_ROOT_SIGNATURE_DESC *desc, + ID3DBlob **blob, ID3DBlob **error_blob); + +/* 1.3 */ +VKD3D_UTILS_API HRESULT WINAPI D3DCompile(const void *data, SIZE_T data_size, const char *filename, + const D3D_SHADER_MACRO *defines, ID3DInclude *include, const char *entrypoint, + const char *target, UINT flags, UINT effect_flags, ID3DBlob **shader, ID3DBlob **error_messages); +VKD3D_UTILS_API HRESULT WINAPI D3DCompile2(const void *data, SIZE_T data_size, const char *filename, + const D3D_SHADER_MACRO *defines, ID3DInclude *include, const char *entrypoint, + const char *target, UINT flags, UINT effect_flags, UINT secondary_flags, + const void *secondary_data, SIZE_T secondary_data_size, ID3DBlob **shader, + ID3DBlob **error_messages); +VKD3D_UTILS_API HRESULT WINAPI D3DCreateBlob(SIZE_T data_size, ID3DBlob **blob); +VKD3D_UTILS_API HRESULT WINAPI D3DPreprocess(const void *data, SIZE_T size, const char *filename, + const D3D_SHADER_MACRO *defines, ID3DInclude *include, + ID3DBlob **shader, ID3DBlob **error_messages); + +/** + * Set a callback to be called when vkd3d-utils outputs debug logging. + * + * If NULL, or if this function has not been called, libvkd3d-utils will print + * all enabled log output to stderr. + * + * Calling this function will also set the log callback for libvkd3d and + * libvkd3d-shader. + * + * \param callback Callback function to set. + * + * \since 1.4 + */ +VKD3D_UTILS_API void vkd3d_utils_set_log_callback(PFN_vkd3d_log callback); + +/** \since 1.10 */ +VKD3D_UTILS_API HRESULT WINAPI D3DGetBlobPart(const void *data, + SIZE_T data_size, D3D_BLOB_PART part, UINT flags, ID3DBlob **blob); +/** \since 1.10 */ +VKD3D_UTILS_API HRESULT WINAPI D3DGetDebugInfo(const void *data, SIZE_T data_size, ID3DBlob **blob); +/** \since 1.10 */ +VKD3D_UTILS_API HRESULT WINAPI D3DGetInputAndOutputSignatureBlob(const void *data, SIZE_T data_size, ID3DBlob **blob); +/** \since 1.10 */ +VKD3D_UTILS_API HRESULT WINAPI D3DGetInputSignatureBlob(const void *data, SIZE_T data_size, ID3DBlob **blob); +/** \since 1.10 */ +VKD3D_UTILS_API HRESULT WINAPI D3DGetOutputSignatureBlob(const void *data, SIZE_T data_size, ID3DBlob **blob); +/** \since 1.10 */ +VKD3D_UTILS_API HRESULT WINAPI D3DStripShader(const void *data, SIZE_T data_size, UINT flags, ID3DBlob **blob); + +#ifdef __cplusplus +} +#endif /* __cplusplus */ + +#endif /* __VKD3D_UTILS_H */ diff --git a/libs/vkd3d/include/vkd3d_windows.h b/libs/vkd3d/include/vkd3d_windows.h new file mode 100644 index 00000000000..7b0e972d828 --- /dev/null +++ b/libs/vkd3d/include/vkd3d_windows.h @@ -0,0 +1,289 @@ +/* + * Copyright 2016 Józef Kucia for CodeWeavers + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA + */ + +#ifndef __VKD3D_WINDOWS_H +#define __VKD3D_WINDOWS_H +#ifndef _INC_WINDOWS + +/* Nameless unions */ +#ifndef __C89_NAMELESS +# ifdef NONAMELESSUNION +# define __C89_NAMELESS +# define __C89_NAMELESSUNIONNAME u +# else +# define __C89_NAMELESS +# define __C89_NAMELESSUNIONNAME +# endif /* NONAMELESSUNION */ +#endif /* __C89_NAMELESS */ + +#if !defined(_WIN32) || defined(__WIDL__) + +# if !defined(__WIDL__) +# if !defined(VKD3D_WIN32_WCHAR) +# include +# endif +# include +# endif + +# ifdef __GNUC__ +# define DECLSPEC_ALIGN(x) __attribute__((aligned(x))) +# endif + +/* HRESULT */ +typedef int HRESULT; +# define SUCCEEDED(hr) ((HRESULT)(hr) >= 0) +# define FAILED(hr) ((HRESULT)(hr) < 0) + +# define _HRESULT_TYPEDEF_(x) ((HRESULT)x) + +# define S_OK _HRESULT_TYPEDEF_(0) +# define S_FALSE _HRESULT_TYPEDEF_(1) + +# define E_NOTIMPL _HRESULT_TYPEDEF_(0x80004001) +# define E_NOINTERFACE _HRESULT_TYPEDEF_(0x80004002) +# define E_POINTER _HRESULT_TYPEDEF_(0x80004003) +# define E_ABORT _HRESULT_TYPEDEF_(0x80004004) +# define E_FAIL _HRESULT_TYPEDEF_(0x80004005) +# define E_OUTOFMEMORY _HRESULT_TYPEDEF_(0x8007000E) +# define E_INVALIDARG _HRESULT_TYPEDEF_(0x80070057) + +# define DXGI_ERROR_NOT_FOUND _HRESULT_TYPEDEF_(0x887a0002) +# define DXGI_ERROR_MORE_DATA _HRESULT_TYPEDEF_(0x887a0003) +# define DXGI_ERROR_UNSUPPORTED _HRESULT_TYPEDEF_(0x887a0004) + +# define D3DERR_INVALIDCALL _HRESULT_TYPEDEF_(0x8876086c) + +/* Basic types */ +typedef unsigned char BYTE; +typedef unsigned int DWORD; +typedef int INT; +typedef unsigned int UINT; +typedef int LONG; +typedef unsigned int ULONG; +typedef float FLOAT; +typedef LONG BOOL; + +/* Assuming LP64 model */ +typedef char INT8; +typedef unsigned char UINT8; +typedef short INT16; +typedef unsigned short UINT16; +typedef int INT32; +typedef unsigned int UINT32; +# if defined(__WIDL__) +typedef __int64 INT64; +typedef unsigned __int64 UINT64; +# else +typedef int64_t DECLSPEC_ALIGN(8) INT64; +typedef uint64_t DECLSPEC_ALIGN(8) UINT64; +# endif +typedef INT64 LONG64; +typedef long LONG_PTR; +typedef unsigned long ULONG_PTR; + +typedef ULONG_PTR SIZE_T; + +# ifdef VKD3D_WIN32_WCHAR +typedef unsigned short WCHAR; +# else +typedef wchar_t WCHAR; +# endif /* VKD3D_WIN32_WCHAR */ +typedef void *HANDLE; + +/* GUID */ +# ifdef __WIDL__ +typedef struct +{ + unsigned long Data1; + unsigned short Data2; + unsigned short Data3; + unsigned char Data4[8]; +} GUID; +# else +typedef struct _GUID +{ + unsigned int Data1; + unsigned short Data2; + unsigned short Data3; + unsigned char Data4[8]; +} GUID; +# endif + +typedef GUID IID; +typedef GUID CLSID; +typedef GUID UUID; + +# ifdef INITGUID +# ifndef __cplusplus +# define DEFINE_GUID(name, l, w1, w2, b1, b2, b3, b4, b5, b6, b7, b8) \ + const GUID name DECLSPEC_HIDDEN; \ + const GUID name = \ + { l, w1, w2, { b1, b2, b3, b4, b5, b6, b7, b8 }} +# else +# define DEFINE_GUID(name, l, w1, w2, b1, b2, b3, b4, b5, b6, b7, b8) \ + EXTERN_C const GUID name DECLSPEC_HIDDEN; \ + EXTERN_C const GUID name = \ + { l, w1, w2, { b1, b2, b3, b4, b5, b6, b7, b8 }} +# endif +# else +# define DEFINE_GUID(name, l, w1, w2, b1, b2, b3, b4, b5, b6, b7, b8) \ + EXTERN_C const GUID name DECLSPEC_HIDDEN; +# endif /* INITGUID */ + +/* __uuidof emulation */ +#if defined(__cplusplus) && !defined(_MSC_VER) + +extern "C++" +{ + template const GUID &__vkd3d_uuidof(); +} + +# define __CRT_UUID_DECL(type, l, w1, w2, b1, b2, b3, b4, b5, b6, b7, b8) \ + extern "C++" \ + { \ + template<> inline const GUID &__vkd3d_uuidof() \ + { \ + static const IID __uuid_inst = {l, w1, w2, {b1, b2, b3, b4, b5, b6, b7, b8}}; \ + return __uuid_inst; \ + } \ + template<> inline const GUID &__vkd3d_uuidof() \ + { \ + return __vkd3d_uuidof(); \ + } \ + } + +# define __uuidof(type) __vkd3d_uuidof() +#else +# define __CRT_UUID_DECL(type, l, w1, w2, b1, b2, b3, b4, b5, b6, b7, b8) +#endif /* defined(__cplusplus) && !defined(_MSC_VER) */ + +typedef struct SECURITY_ATTRIBUTES SECURITY_ATTRIBUTES; +#endif /* !defined(_WIN32) || defined(__WIDL__) */ + + +#ifndef _WIN32 +# include +# include +# include + +# define COM_NO_WINDOWS_H + +# define FORCEINLINE inline + +# define CONTAINING_RECORD(address, type, field) \ + ((type *)((char *)(address) - offsetof(type, field))) + +# ifdef __x86_64__ +# define __stdcall __attribute__((ms_abi)) +# else +# if (__GNUC__ > 4) || ((__GNUC__ == 4) && (__GNUC_MINOR__ >= 2)) || defined(__APPLE__) +# define __stdcall __attribute__((__stdcall__)) __attribute__((__force_align_arg_pointer__)) +# else +# define __stdcall __attribute__((__stdcall__)) +# endif +# endif + +# define WINAPI __stdcall +# define STDMETHODCALLTYPE __stdcall + +# ifdef __GNUC__ +# define DECLSPEC_SELECTANY __attribute__((weak)) +# endif + +/* Macros for COM interfaces */ +# define interface struct +# define BEGIN_INTERFACE +# define END_INTERFACE +# define MIDL_INTERFACE(x) struct + +# ifdef __cplusplus +# define EXTERN_C extern "C" +# else +# define EXTERN_C extern +# endif + +# define CONST_VTBL const + +# define TRUE 1 +# define FALSE 0 + +# if defined(__cplusplus) && !defined(CINTERFACE) +# define REFIID const IID & +# define REFGUID const GUID & +# define REFCLSID const CLSID & +# else +# define REFIID const IID * const +# define REFGUID const GUID * const +# define REFCLSID const CLSID * const +# endif + +#if defined(__cplusplus) && !defined(CINTERFACE) +# define IsEqualGUID(guid1, guid2) (!memcmp(&(guid1), &(guid2), sizeof(GUID))) +#else +# define IsEqualGUID(guid1, guid2) (!memcmp(guid1, guid2, sizeof(GUID))) +#endif + +#elif !defined(__WIDL__) + +# include + +#endif /* _WIN32 */ + + +/* Define DECLSPEC_HIDDEN */ +#ifndef DECLSPEC_HIDDEN +# if defined(__MINGW32__) +# define DECLSPEC_HIDDEN +# elif defined(__GNUC__) +# define DECLSPEC_HIDDEN __attribute__((visibility("hidden"))) +# else +# define DECLSPEC_HIDDEN +# endif +#endif /* DECLSPEC_HIDDEN */ + +/* Define min() & max() macros */ +#ifndef NOMINMAX +# ifndef min +# define min(a, b) (((a) <= (b)) ? (a) : (b)) +# endif + +# ifndef max +# define max(a, b) (((a) >= (b)) ? (a) : (b)) +# endif +#endif /* NOMINMAX */ + +#ifndef DEFINE_ENUM_FLAG_OPERATORS +#ifdef __cplusplus +# define DEFINE_ENUM_FLAG_OPERATORS(type) \ +extern "C++" \ +{ \ + inline type operator &(type x, type y) { return (type)((int)x & (int)y); } \ + inline type operator &=(type &x, type y) { return (type &)((int &)x &= (int)y); } \ + inline type operator ~(type x) { return (type)~(int)x; } \ + inline type operator |(type x, type y) { return (type)((int)x | (int)y); } \ + inline type operator |=(type &x, type y) { return (type &)((int &)x |= (int)y); } \ + inline type operator ^(type x, type y) { return (type)((int)x ^ (int)y); } \ + inline type operator ^=(type &x, type y) { return (type &)((int &)x ^= (int)y); } \ +} +#else +# define DEFINE_ENUM_FLAG_OPERATORS(type) +#endif +#endif /* DEFINE_ENUM_FLAG_OPERATORS */ + +#endif /* _INC_WINDOWS */ +#endif /* __VKD3D_WINDOWS_H */ diff --git a/libs/vkd3d/libs/vkd3d-common/blob.c b/libs/vkd3d/libs/vkd3d-common/blob.c index 30205088b1b..fa2812619ac 100644 --- a/libs/vkd3d/libs/vkd3d-common/blob.c +++ b/libs/vkd3d/libs/vkd3d-common/blob.c @@ -17,6 +17,8 @@ */ #define COBJMACROS + +#define CONST_VTABLE #include "vkd3d.h" #include "vkd3d_blob.h" #include "vkd3d_debug.h" diff --git a/libs/vkd3d/libs/vkd3d-shader/d3d_asm.c b/libs/vkd3d/libs/vkd3d-shader/d3d_asm.c index f0c386f1b3a..82d1d71d9d3 100644 --- a/libs/vkd3d/libs/vkd3d-shader/d3d_asm.c +++ b/libs/vkd3d/libs/vkd3d-shader/d3d_asm.c @@ -106,9 +106,9 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_DEFAULT ] = "default", [VKD3DSIH_DEFB ] = "defb", [VKD3DSIH_DEFI ] = "defi", - [VKD3DSIH_DEQ ] = "deq", + [VKD3DSIH_DEQO ] = "deq", [VKD3DSIH_DFMA ] = "dfma", - [VKD3DSIH_DGE ] = "dge", + [VKD3DSIH_DGEO ] = "dge", [VKD3DSIH_DISCARD ] = "discard", [VKD3DSIH_DIV ] = "div", [VKD3DSIH_DLT ] = "dlt", @@ -140,7 +140,8 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_ENDLOOP ] = "endloop", [VKD3DSIH_ENDREP ] = "endrep", [VKD3DSIH_ENDSWITCH ] = "endswitch", - [VKD3DSIH_EQ ] = "eq", + [VKD3DSIH_EQO ] = "eq", + [VKD3DSIH_EQU ] = "eq_unord", [VKD3DSIH_EVAL_CENTROID ] = "eval_centroid", [VKD3DSIH_EVAL_SAMPLE_INDEX ] = "eval_sample_index", [VKD3DSIH_EXP ] = "exp", @@ -152,6 +153,7 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_FIRSTBIT_LO ] = "firstbit_lo", [VKD3DSIH_FIRSTBIT_SHI ] = "firstbit_shi", [VKD3DSIH_FRC ] = "frc", + [VKD3DSIH_FREM ] = "frem", [VKD3DSIH_FTOD ] = "ftod", [VKD3DSIH_FTOI ] = "ftoi", [VKD3DSIH_FTOU ] = "ftou", @@ -163,13 +165,15 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_GATHER4_PO_C_S ] = "gather4_po_c_s", [VKD3DSIH_GATHER4_PO_S ] = "gather4_po_s", [VKD3DSIH_GATHER4_S ] = "gather4_s", - [VKD3DSIH_GE ] = "ge", + [VKD3DSIH_GEO ] = "ge", + [VKD3DSIH_GEU ] = "ge_unord", [VKD3DSIH_HS_CONTROL_POINT_PHASE ] = "hs_control_point_phase", [VKD3DSIH_HS_DECLS ] = "hs_decls", [VKD3DSIH_HS_FORK_PHASE ] = "hs_fork_phase", [VKD3DSIH_HS_JOIN_PHASE ] = "hs_join_phase", [VKD3DSIH_IADD ] = "iadd", [VKD3DSIH_IBFE ] = "ibfe", + [VKD3DSIH_IDIV ] = "idiv", [VKD3DSIH_IEQ ] = "ieq", [VKD3DSIH_IF ] = "if", [VKD3DSIH_IFC ] = "ifc", @@ -197,6 +201,7 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_ISHR ] = "ishr", [VKD3DSIH_ITOD ] = "itod", [VKD3DSIH_ITOF ] = "itof", + [VKD3DSIH_ITOI ] = "itoi", [VKD3DSIH_LABEL ] = "label", [VKD3DSIH_LD ] = "ld", [VKD3DSIH_LD2DMS ] = "ld2dms", @@ -214,7 +219,8 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_LOGP ] = "logp", [VKD3DSIH_LOOP ] = "loop", [VKD3DSIH_LRP ] = "lrp", - [VKD3DSIH_LT ] = "lt", + [VKD3DSIH_LTO ] = "lt", + [VKD3DSIH_LTU ] = "lt_unord", [VKD3DSIH_M3x2 ] = "m3x2", [VKD3DSIH_M3x3 ] = "m3x3", [VKD3DSIH_M3x4 ] = "m3x4", @@ -228,7 +234,8 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_MOVC ] = "movc", [VKD3DSIH_MSAD ] = "msad", [VKD3DSIH_MUL ] = "mul", - [VKD3DSIH_NE ] = "ne", + [VKD3DSIH_NEO ] = "ne_ord", + [VKD3DSIH_NEU ] = "ne", [VKD3DSIH_NOP ] = "nop", [VKD3DSIH_NOT ] = "not", [VKD3DSIH_NRM ] = "nrm", @@ -304,6 +311,7 @@ static const char * const shader_opcode_names[] = [VKD3DSIH_USHR ] = "ushr", [VKD3DSIH_UTOD ] = "utod", [VKD3DSIH_UTOF ] = "utof", + [VKD3DSIH_UTOU ] = "utou", [VKD3DSIH_XOR ] = "xor", }; @@ -358,11 +366,6 @@ struct vkd3d_d3d_asm_compiler struct vkd3d_d3d_asm_colours colours; }; -static int shader_ver_ge(const struct vkd3d_shader_version *v, int major, int minor) -{ - return v->major > major || (v->major == major && v->minor >= minor); -} - static int VKD3D_PRINTF_FUNC(2, 3) shader_addline(struct vkd3d_string_buffer *buffer, const char *format, ...) { va_list args; @@ -391,13 +394,14 @@ static unsigned int shader_get_float_offset(enum vkd3d_shader_register_type regi } } -static void shader_dump_global_flags(struct vkd3d_d3d_asm_compiler *compiler, uint32_t global_flags) +static void shader_dump_global_flags(struct vkd3d_d3d_asm_compiler *compiler, + enum vkd3d_shader_global_flags global_flags) { unsigned int i; static const struct { - unsigned int flag; + enum vkd3d_shader_global_flags flag; const char *name; } global_flag_info[] = @@ -423,7 +427,7 @@ static void shader_dump_global_flags(struct vkd3d_d3d_asm_compiler *compiler, ui } if (global_flags) - vkd3d_string_buffer_printf(&compiler->buffer, "unknown_flags(%#x)", global_flags); + vkd3d_string_buffer_printf(&compiler->buffer, "unknown_flags(%#"PRIx64")", (uint64_t)global_flags); } static void shader_dump_sync_flags(struct vkd3d_d3d_asm_compiler *compiler, uint32_t sync_flags) @@ -477,6 +481,11 @@ static void shader_dump_uav_flags(struct vkd3d_d3d_asm_compiler *compiler, uint3 vkd3d_string_buffer_printf(&compiler->buffer, "_opc"); uav_flags &= ~VKD3DSUF_ORDER_PRESERVING_COUNTER; } + if (uav_flags & VKD3DSUF_RASTERISER_ORDERED_VIEW) + { + vkd3d_string_buffer_printf(&compiler->buffer, "_rov"); + uav_flags &= ~VKD3DSUF_RASTERISER_ORDERED_VIEW; + } if (uav_flags) vkd3d_string_buffer_printf(&compiler->buffer, "_unknown_flags(%#x)", uav_flags); @@ -637,7 +646,7 @@ static void shader_dump_decl_usage(struct vkd3d_d3d_asm_compiler *compiler, { struct vkd3d_string_buffer *buffer = &compiler->buffer; - if (semantic->resource.reg.reg.type == VKD3DSPR_SAMPLER) + if (semantic->resource.reg.reg.type == VKD3DSPR_COMBINED_SAMPLER) { switch (semantic->resource_type) { @@ -678,7 +687,7 @@ static void shader_dump_decl_usage(struct vkd3d_d3d_asm_compiler *compiler, else { /* Pixel shaders 3.0 don't have usage semantics. */ - if (!shader_ver_ge(&compiler->shader_version, 3, 0) + if (!vkd3d_shader_ver_ge(&compiler->shader_version, 3, 0) && compiler->shader_version.type == VKD3D_SHADER_TYPE_PIXEL) return; else @@ -880,9 +889,7 @@ static void shader_dump_register(struct vkd3d_d3d_asm_compiler *compiler, const break; case VKD3DSPR_COLOROUT: - shader_addline(buffer, "o"); - if (!shader_ver_ge(&compiler->shader_version, 4, 0)) - shader_addline(buffer, "C"); + shader_addline(buffer, "oC"); break; case VKD3DSPR_DEPTHOUT: @@ -904,7 +911,7 @@ static void shader_dump_register(struct vkd3d_d3d_asm_compiler *compiler, const case VKD3DSPR_TEXCRDOUT: /* Vertex shaders >= 3.0 use general purpose output registers * (VKD3DSPR_OUTPUT), which can include an address token. */ - if (shader_ver_ge(&compiler->shader_version, 3, 0)) + if (vkd3d_shader_ver_ge(&compiler->shader_version, 3, 0)) shader_addline(buffer, "o"); else shader_addline(buffer, "oT"); @@ -926,6 +933,7 @@ static void shader_dump_register(struct vkd3d_d3d_asm_compiler *compiler, const shader_addline(buffer, "aL"); break; + case VKD3DSPR_COMBINED_SAMPLER: case VKD3DSPR_SAMPLER: shader_addline(buffer, "s"); is_descriptor = true; @@ -1066,6 +1074,14 @@ static void shader_dump_register(struct vkd3d_d3d_asm_compiler *compiler, const shader_addline(buffer, "oStencilRef"); break; + case VKD3DSPR_UNDEF: + shader_addline(buffer, "undef"); + break; + + case VKD3DSPR_SSA: + shader_addline(buffer, "sr"); + break; + default: shader_addline(buffer, "", reg->type); break; @@ -1074,9 +1090,9 @@ static void shader_dump_register(struct vkd3d_d3d_asm_compiler *compiler, const if (reg->type == VKD3DSPR_IMMCONST) { shader_addline(buffer, "%s(", compiler->colours.reset); - switch (reg->immconst_type) + switch (reg->dimension) { - case VKD3D_IMMCONST_SCALAR: + case VSIR_DIMENSION_SCALAR: switch (reg->data_type) { case VKD3D_DATA_FLOAT: @@ -1096,7 +1112,7 @@ static void shader_dump_register(struct vkd3d_d3d_asm_compiler *compiler, const } break; - case VKD3D_IMMCONST_VEC4: + case VSIR_DIMENSION_VEC4: switch (reg->data_type) { case VKD3D_DATA_FLOAT: @@ -1126,7 +1142,7 @@ static void shader_dump_register(struct vkd3d_d3d_asm_compiler *compiler, const break; default: - shader_addline(buffer, "", reg->immconst_type); + shader_addline(buffer, "", reg->dimension); break; } shader_addline(buffer, ")"); @@ -1134,13 +1150,13 @@ static void shader_dump_register(struct vkd3d_d3d_asm_compiler *compiler, const else if (reg->type == VKD3DSPR_IMMCONST64) { shader_addline(buffer, "%s(", compiler->colours.reset); - /* A double2 vector is treated as a float4 vector in enum vkd3d_immconst_type. */ - if (reg->immconst_type == VKD3D_IMMCONST_SCALAR || reg->immconst_type == VKD3D_IMMCONST_VEC4) + /* A double2 vector is treated as a float4 vector in enum vsir_dimension. */ + if (reg->dimension == VSIR_DIMENSION_SCALAR || reg->dimension == VSIR_DIMENSION_VEC4) { if (reg->data_type == VKD3D_DATA_DOUBLE) { shader_print_double_literal(compiler, "", reg->u.immconst_double[0], ""); - if (reg->immconst_type == VKD3D_IMMCONST_VEC4) + if (reg->dimension == VSIR_DIMENSION_VEC4) shader_print_double_literal(compiler, ", ", reg->u.immconst_double[1], ""); } else @@ -1150,17 +1166,18 @@ static void shader_dump_register(struct vkd3d_d3d_asm_compiler *compiler, const } else { - shader_addline(buffer, "", reg->immconst_type); + shader_addline(buffer, "", reg->dimension); } shader_addline(buffer, ")"); } else if (reg->type != VKD3DSPR_RASTOUT && reg->type != VKD3DSPR_MISCTYPE - && reg->type != VKD3DSPR_NULL) + && reg->type != VKD3DSPR_NULL + && reg->type != VKD3DSPR_DEPTHOUT) { if (offset != ~0u) { - bool is_sm_5_1 = shader_ver_ge(&compiler->shader_version, 5, 1); + bool is_sm_5_1 = vkd3d_shader_ver_ge(&compiler->shader_version, 5, 1); if (reg->idx[0].rel_addr || reg->type == VKD3DSPR_IMMCONSTBUFFER || reg->type == VKD3DSPR_INCONTROLPOINT || (reg->type == VKD3DSPR_INPUT @@ -1181,7 +1198,7 @@ static void shader_dump_register(struct vkd3d_d3d_asm_compiler *compiler, const { shader_print_subscript_range(compiler, reg->idx[1].offset, reg->idx[2].offset); } - else + else if (reg->type != VKD3DSPR_SSA) { /* For descriptors in sm < 5.1 we move the reg->idx values up one slot * to normalise with 5.1. @@ -1256,7 +1273,7 @@ static void shader_dump_dst_param(struct vkd3d_d3d_asm_compiler *compiler, shader_dump_register(compiler, ¶m->reg, is_declaration); - if (write_mask) + if (write_mask && param->reg.dimension == VSIR_DIMENSION_VEC4) { static const char write_mask_chars[] = "xyzw"; @@ -1322,7 +1339,7 @@ static void shader_dump_src_param(struct vkd3d_d3d_asm_compiler *compiler, } if (param->reg.type != VKD3DSPR_IMMCONST && param->reg.type != VKD3DSPR_IMMCONST64 - && param->reg.type != VKD3DSPR_SAMPLER) + && param->reg.dimension == VSIR_DIMENSION_VEC4) { unsigned int swizzle_x = vkd3d_swizzle_get_component(swizzle, 0); unsigned int swizzle_y = vkd3d_swizzle_get_component(swizzle, 1); @@ -1374,7 +1391,7 @@ static void shader_dump_ins_modifiers(struct vkd3d_d3d_asm_compiler *compiler, if (mmask & VKD3DSPDM_PARTIALPRECISION) shader_addline(buffer, "_pp"); if (mmask & VKD3DSPDM_MSAMPCENTROID) shader_addline(buffer, "_centroid"); - mmask &= ~(VKD3DSPDM_SATURATE | VKD3DSPDM_PARTIALPRECISION | VKD3DSPDM_MSAMPCENTROID); + mmask &= ~VKD3DSPDM_MASK; if (mmask) FIXME("Unrecognised modifier %#x.\n", mmask); } @@ -1556,7 +1573,7 @@ static void shader_dump_instruction_flags(struct vkd3d_d3d_asm_compiler *compile break; case VKD3DSIH_TEX: - if (shader_ver_ge(&compiler->shader_version, 2, 0) && (ins->flags & VKD3DSI_TEXLD_PROJECT)) + if (vkd3d_shader_ver_ge(&compiler->shader_version, 2, 0) && (ins->flags & VKD3DSI_TEXLD_PROJECT)) shader_addline(buffer, "p"); break; @@ -1568,7 +1585,7 @@ static void shader_dump_instruction_flags(struct vkd3d_d3d_asm_compiler *compile static void shader_dump_register_space(struct vkd3d_d3d_asm_compiler *compiler, unsigned int register_space) { - if (shader_ver_ge(&compiler->shader_version, 5, 1)) + if (vkd3d_shader_ver_ge(&compiler->shader_version, 5, 1)) shader_print_uint_literal(compiler, ", space=", register_space, ""); } @@ -1578,6 +1595,36 @@ static void shader_print_opcode(struct vkd3d_d3d_asm_compiler *compiler, enum vk shader_opcode_names[opcode], compiler->colours.reset); } +static void shader_dump_icb(struct vkd3d_d3d_asm_compiler *compiler, + const struct vkd3d_shader_immediate_constant_buffer *icb) +{ + struct vkd3d_string_buffer *buffer = &compiler->buffer; + unsigned int i, j; + + vkd3d_string_buffer_printf(buffer, " {\n"); + if (icb->component_count == 1) + { + for (i = 0; i < icb->element_count; ) + { + for (j = 0; i < icb->element_count && j < 4; ++i, ++j) + shader_print_hex_literal(compiler, !j ? " " : ", ", icb->data[i], ""); + vkd3d_string_buffer_printf(buffer, "\n"); + } + } + else + { + assert(icb->component_count == VKD3D_VEC4_SIZE); + for (i = 0; i < icb->element_count; ++i) + { + shader_print_hex_literal(compiler, " {", icb->data[4 * i + 0], ""); + shader_print_hex_literal(compiler, ", ", icb->data[4 * i + 1], ""); + shader_print_hex_literal(compiler, ", ", icb->data[4 * i + 2], ""); + shader_print_hex_literal(compiler, ", ", icb->data[4 * i + 3], "},\n"); + } + } + shader_addline(buffer, "}"); +} + static void shader_dump_instruction(struct vkd3d_d3d_asm_compiler *compiler, const struct vkd3d_shader_instruction *ins) { @@ -1612,8 +1659,10 @@ static void shader_dump_instruction(struct vkd3d_d3d_asm_compiler *compiler, case VKD3DSIH_DCL_CONSTANT_BUFFER: vkd3d_string_buffer_printf(buffer, " "); shader_dump_register(compiler, &ins->declaration.cb.src.reg, true); - if (shader_ver_ge(&compiler->shader_version, 5, 1)) + if (vkd3d_shader_ver_ge(&compiler->shader_version, 6, 0)) shader_print_subscript(compiler, ins->declaration.cb.size, NULL); + else if (vkd3d_shader_ver_ge(&compiler->shader_version, 5, 1)) + shader_print_subscript(compiler, ins->declaration.cb.size / VKD3D_VEC4_SIZE / sizeof(float), NULL); shader_addline(buffer, ", %s", ins->flags & VKD3DSI_INDEXED_DYNAMIC ? "dynamicIndexed" : "immediateIndexed"); shader_dump_register_space(compiler, ins->declaration.cb.range.space); @@ -1629,7 +1678,7 @@ static void shader_dump_instruction(struct vkd3d_d3d_asm_compiler *compiler, case VKD3DSIH_DCL_GLOBAL_FLAGS: vkd3d_string_buffer_printf(buffer, " "); - shader_dump_global_flags(compiler, ins->flags); + shader_dump_global_flags(compiler, ins->declaration.global_flags); break; case VKD3DSIH_DCL_HS_MAX_TESSFACTOR: @@ -1637,15 +1686,7 @@ static void shader_dump_instruction(struct vkd3d_d3d_asm_compiler *compiler, break; case VKD3DSIH_DCL_IMMEDIATE_CONSTANT_BUFFER: - vkd3d_string_buffer_printf(buffer, " {\n"); - for (i = 0; i < ins->declaration.icb->vec4_count; ++i) - { - shader_print_hex_literal(compiler, " {", ins->declaration.icb->data[4 * i + 0], ""); - shader_print_hex_literal(compiler, ", ", ins->declaration.icb->data[4 * i + 1], ""); - shader_print_hex_literal(compiler, ", ", ins->declaration.icb->data[4 * i + 2], ""); - shader_print_hex_literal(compiler, ", ", ins->declaration.icb->data[4 * i + 3], "},\n"); - } - shader_addline(buffer, "}"); + shader_dump_icb(compiler, ins->declaration.icb); break; case VKD3DSIH_DCL_INDEX_RANGE: @@ -1659,6 +1700,10 @@ static void shader_dump_instruction(struct vkd3d_d3d_asm_compiler *compiler, ins->declaration.indexable_temp.register_idx, compiler->colours.reset); shader_print_subscript(compiler, ins->declaration.indexable_temp.register_size, NULL); shader_print_uint_literal(compiler, ", ", ins->declaration.indexable_temp.component_count, ""); + if (ins->declaration.indexable_temp.alignment) + shader_print_uint_literal(compiler, ", align ", ins->declaration.indexable_temp.alignment, ""); + if (ins->declaration.indexable_temp.initialiser) + shader_dump_icb(compiler, ins->declaration.indexable_temp.initialiser); break; case VKD3DSIH_DCL_INPUT_PS: diff --git a/libs/vkd3d/libs/vkd3d-shader/d3dbc.c b/libs/vkd3d/libs/vkd3d-shader/d3dbc.c index 1fd5ab2446d..3d139416b61 100644 --- a/libs/vkd3d/libs/vkd3d-shader/d3dbc.c +++ b/libs/vkd3d/libs/vkd3d-shader/d3dbc.c @@ -261,7 +261,7 @@ static const struct vkd3d_sm1_opcode_info vs_opcode_table[] = {VKD3D_SM1_OP_M3x3, 1, 2, VKD3DSIH_M3x3}, {VKD3D_SM1_OP_M3x2, 1, 2, VKD3DSIH_M3x2}, /* Declarations */ - {VKD3D_SM1_OP_DCL, 0, 2, VKD3DSIH_DCL}, + {VKD3D_SM1_OP_DCL, 0, 0, VKD3DSIH_DCL}, /* Constant definitions */ {VKD3D_SM1_OP_DEF, 1, 1, VKD3DSIH_DEF}, {VKD3D_SM1_OP_DEFB, 1, 1, VKD3DSIH_DEFB}, @@ -328,7 +328,7 @@ static const struct vkd3d_sm1_opcode_info ps_opcode_table[] = {VKD3D_SM1_OP_M3x3, 1, 2, VKD3DSIH_M3x3}, {VKD3D_SM1_OP_M3x2, 1, 2, VKD3DSIH_M3x2}, /* Declarations */ - {VKD3D_SM1_OP_DCL, 0, 2, VKD3DSIH_DCL}, + {VKD3D_SM1_OP_DCL, 0, 0, VKD3DSIH_DCL}, /* Constant definitions */ {VKD3D_SM1_OP_DEF, 1, 1, VKD3DSIH_DEF}, {VKD3D_SM1_OP_DEFB, 1, 1, VKD3DSIH_DEFB}, @@ -402,16 +402,6 @@ static uint32_t read_u32(const uint32_t **ptr) return *(*ptr)++; } -static bool shader_ver_ge(const struct vkd3d_shader_version *v, unsigned int major, unsigned int minor) -{ - return v->major > major || (v->major == major && v->minor >= minor); -} - -static bool shader_ver_le(const struct vkd3d_shader_version *v, unsigned int major, unsigned int minor) -{ - return v->major < major || (v->major == major && v->minor <= minor); -} - static bool has_relative_address(uint32_t param) { enum vkd3d_sm1_address_mode_type address_mode; @@ -434,8 +424,8 @@ static const struct vkd3d_sm1_opcode_info *shader_sm1_get_opcode_info( return NULL; if (opcode == info->sm1_opcode - && shader_ver_ge(&sm1->p.shader_version, info->min_version.major, info->min_version.minor) - && (shader_ver_le(&sm1->p.shader_version, info->max_version.major, info->max_version.minor) + && vkd3d_shader_ver_ge(&sm1->p.shader_version, info->min_version.major, info->min_version.minor) + && (vkd3d_shader_ver_le(&sm1->p.shader_version, info->max_version.major, info->max_version.minor) || !info->max_version.major)) return info; } @@ -457,18 +447,20 @@ static uint32_t swizzle_from_sm1(uint32_t swizzle) static void shader_sm1_parse_src_param(uint32_t param, const struct vkd3d_shader_src_param *rel_addr, struct vkd3d_shader_src_param *src) { - src->reg.type = ((param & VKD3D_SM1_REGISTER_TYPE_MASK) >> VKD3D_SM1_REGISTER_TYPE_SHIFT) + enum vkd3d_shader_register_type reg_type = ((param & VKD3D_SM1_REGISTER_TYPE_MASK) >> VKD3D_SM1_REGISTER_TYPE_SHIFT) | ((param & VKD3D_SM1_REGISTER_TYPE_MASK2) >> VKD3D_SM1_REGISTER_TYPE_SHIFT2); + + vsir_register_init(&src->reg, reg_type, VKD3D_DATA_FLOAT, 1); src->reg.precision = VKD3D_SHADER_REGISTER_PRECISION_DEFAULT; src->reg.non_uniform = false; - src->reg.data_type = VKD3D_DATA_FLOAT; src->reg.idx[0].offset = param & VKD3D_SM1_REGISTER_NUMBER_MASK; src->reg.idx[0].rel_addr = rel_addr; - src->reg.idx[1].offset = ~0u; - src->reg.idx[1].rel_addr = NULL; - src->reg.idx[2].offset = ~0u; - src->reg.idx[2].rel_addr = NULL; - src->reg.idx_count = 1; + if (src->reg.type == VKD3DSPR_SAMPLER) + src->reg.dimension = VSIR_DIMENSION_NONE; + else if (src->reg.type == VKD3DSPR_DEPTHOUT) + src->reg.dimension = VSIR_DIMENSION_SCALAR; + else + src->reg.dimension = VSIR_DIMENSION_VEC4; src->swizzle = swizzle_from_sm1((param & VKD3D_SM1_SWIZZLE_MASK) >> VKD3D_SM1_SWIZZLE_SHIFT); src->modifiers = (param & VKD3D_SM1_SRC_MODIFIER_MASK) >> VKD3D_SM1_SRC_MODIFIER_SHIFT; } @@ -476,18 +468,20 @@ static void shader_sm1_parse_src_param(uint32_t param, const struct vkd3d_shader static void shader_sm1_parse_dst_param(uint32_t param, const struct vkd3d_shader_src_param *rel_addr, struct vkd3d_shader_dst_param *dst) { - dst->reg.type = ((param & VKD3D_SM1_REGISTER_TYPE_MASK) >> VKD3D_SM1_REGISTER_TYPE_SHIFT) + enum vkd3d_shader_register_type reg_type = ((param & VKD3D_SM1_REGISTER_TYPE_MASK) >> VKD3D_SM1_REGISTER_TYPE_SHIFT) | ((param & VKD3D_SM1_REGISTER_TYPE_MASK2) >> VKD3D_SM1_REGISTER_TYPE_SHIFT2); + + vsir_register_init(&dst->reg, reg_type, VKD3D_DATA_FLOAT, 1); dst->reg.precision = VKD3D_SHADER_REGISTER_PRECISION_DEFAULT; dst->reg.non_uniform = false; - dst->reg.data_type = VKD3D_DATA_FLOAT; dst->reg.idx[0].offset = param & VKD3D_SM1_REGISTER_NUMBER_MASK; dst->reg.idx[0].rel_addr = rel_addr; - dst->reg.idx[1].offset = ~0u; - dst->reg.idx[1].rel_addr = NULL; - dst->reg.idx[2].offset = ~0u; - dst->reg.idx[2].rel_addr = NULL; - dst->reg.idx_count = 1; + if (dst->reg.type == VKD3DSPR_SAMPLER) + dst->reg.dimension = VSIR_DIMENSION_NONE; + else if (dst->reg.type == VKD3DSPR_DEPTHOUT) + dst->reg.dimension = VSIR_DIMENSION_SCALAR; + else + dst->reg.dimension = VSIR_DIMENSION_VEC4; dst->write_mask = (param & VKD3D_SM1_WRITEMASK_MASK) >> VKD3D_SM1_WRITEMASK_SHIFT; dst->modifiers = (param & VKD3D_SM1_DST_MODIFIER_MASK) >> VKD3D_SM1_DST_MODIFIER_SHIFT; dst->shift = (param & VKD3D_SM1_DSTSHIFT_MASK) >> VKD3D_SM1_DSTSHIFT_SHIFT; @@ -551,9 +545,9 @@ static bool add_signature_element(struct vkd3d_shader_sm1_parser *sm1, bool outp return false; element = &signature->elements[signature->element_count++]; + memset(element, 0, sizeof(*element)); element->semantic_name = name; element->semantic_index = index; - element->stream_index = 0; element->sysval_semantic = sysval; element->component_type = VKD3D_SHADER_COMPONENT_FLOAT; element->register_index = register_index; @@ -561,7 +555,8 @@ static bool add_signature_element(struct vkd3d_shader_sm1_parser *sm1, bool outp element->register_count = 1; element->mask = mask; element->used_mask = is_dcl ? 0 : mask; - element->min_precision = VKD3D_SHADER_MINIMUM_PRECISION_NONE; + if (sm1->p.shader_version.type == VKD3D_SHADER_TYPE_PIXEL && !output) + element->interpolation_mode = VKD3DSIM_LINEAR; return true; } @@ -853,6 +848,22 @@ static void shader_sm1_skip_opcode(const struct vkd3d_shader_sm1_parser *sm1, co return; } + /* DCL instructions do not have sources or destinations, but they + * read two tokens to a semantic. See + * shader_sm1_read_semantic(). */ + if (opcode_info->vkd3d_opcode == VKD3DSIH_DCL) + { + *ptr += 2; + } + /* Somewhat similarly, DEF and DEFI have a single source, but need to read + * four tokens for that source. See shader_sm1_read_immconst(). + * Technically shader model 1 doesn't have integer registers or DEFI; we + * handle it here anyway because it's easy. */ + else if (opcode_info->vkd3d_opcode == VKD3DSIH_DEF || opcode_info->vkd3d_opcode == VKD3DSIH_DEFI) + { + *ptr += 3; + } + *ptr += (opcode_info->dst_count + opcode_info->src_count); } @@ -951,9 +962,9 @@ static void shader_sm1_read_semantic(struct vkd3d_shader_sm1_parser *sm1, } static void shader_sm1_read_immconst(struct vkd3d_shader_sm1_parser *sm1, const uint32_t **ptr, - struct vkd3d_shader_src_param *src_param, enum vkd3d_immconst_type type, enum vkd3d_data_type data_type) + struct vkd3d_shader_src_param *src_param, enum vsir_dimension dimension, enum vkd3d_data_type data_type) { - unsigned int count = type == VKD3D_IMMCONST_VEC4 ? 4 : 1; + unsigned int count = dimension == VSIR_DIMENSION_VEC4 ? 4 : 1; if (*ptr >= sm1->end || sm1->end - *ptr < count) { @@ -975,7 +986,7 @@ static void shader_sm1_read_immconst(struct vkd3d_shader_sm1_parser *sm1, const src_param->reg.idx[2].offset = ~0u; src_param->reg.idx[2].rel_addr = NULL; src_param->reg.idx_count = 0; - src_param->reg.immconst_type = type; + src_param->reg.dimension = dimension; memcpy(src_param->reg.u.immconst_uint, *ptr, count * sizeof(uint32_t)); src_param->swizzle = VKD3D_SHADER_NO_SWIZZLE; src_param->modifiers = 0; @@ -1090,7 +1101,7 @@ static void shader_sm1_read_instruction(struct vkd3d_shader_sm1_parser *sm1, str goto fail; } - ins->handler_idx = opcode_info->vkd3d_opcode; + vsir_instruction_init(ins, &sm1->p.location, opcode_info->vkd3d_opcode); ins->flags = (opcode_token & VKD3D_SM1_INSTRUCTION_FLAGS_MASK) >> VKD3D_SM1_INSTRUCTION_FLAGS_SHIFT; ins->coissue = opcode_token & VKD3D_SM1_COISSUE; ins->raw = false; @@ -1132,19 +1143,19 @@ static void shader_sm1_read_instruction(struct vkd3d_shader_sm1_parser *sm1, str else if (ins->handler_idx == VKD3DSIH_DEF) { shader_sm1_read_dst_param(sm1, &p, dst_param); - shader_sm1_read_immconst(sm1, &p, &src_params[0], VKD3D_IMMCONST_VEC4, VKD3D_DATA_FLOAT); + shader_sm1_read_immconst(sm1, &p, &src_params[0], VSIR_DIMENSION_VEC4, VKD3D_DATA_FLOAT); shader_sm1_scan_register(sm1, &dst_param->reg, dst_param->write_mask, true); } else if (ins->handler_idx == VKD3DSIH_DEFB) { shader_sm1_read_dst_param(sm1, &p, dst_param); - shader_sm1_read_immconst(sm1, &p, &src_params[0], VKD3D_IMMCONST_SCALAR, VKD3D_DATA_UINT); + shader_sm1_read_immconst(sm1, &p, &src_params[0], VSIR_DIMENSION_SCALAR, VKD3D_DATA_UINT); shader_sm1_scan_register(sm1, &dst_param->reg, dst_param->write_mask, true); } else if (ins->handler_idx == VKD3DSIH_DEFI) { shader_sm1_read_dst_param(sm1, &p, dst_param); - shader_sm1_read_immconst(sm1, &p, &src_params[0], VKD3D_IMMCONST_VEC4, VKD3D_DATA_INT); + shader_sm1_read_immconst(sm1, &p, &src_params[0], VSIR_DIMENSION_VEC4, VKD3D_DATA_INT); shader_sm1_scan_register(sm1, &dst_param->reg, dst_param->write_mask, true); } else @@ -1250,7 +1261,7 @@ static enum vkd3d_result shader_sm1_init(struct vkd3d_shader_sm1_parser *sm1, return VKD3D_ERROR_INVALID_SHADER; } - if (!shader_ver_le(&version, 3, 0)) + if (!vkd3d_shader_ver_le(&version, 3, 0)) { vkd3d_shader_error(message_context, &location, VKD3D_SHADER_ERROR_D3DBC_INVALID_VERSION_TOKEN, "Invalid shader version %u.%u (token 0x%08x).", version.major, version.minor, code[0]); @@ -1333,12 +1344,22 @@ int vkd3d_shader_sm1_parser_create(const struct vkd3d_shader_compile_info *compi ++instructions->count; } - *parser = &sm1->p; - for (i = 0; i < ARRAY_SIZE(sm1->p.shader_desc.flat_constant_count); ++i) sm1->p.shader_desc.flat_constant_count[i].external = get_external_constant_count(sm1, i); - return sm1->p.failed ? VKD3D_ERROR_INVALID_SHADER : VKD3D_OK; + if (!sm1->p.failed) + vsir_validate(&sm1->p); + + if (sm1->p.failed) + { + WARN("Failed to parse shader.\n"); + shader_sm1_destroy(&sm1->p); + return VKD3D_ERROR_INVALID_SHADER; + } + + *parser = &sm1->p; + + return VKD3D_OK; } bool hlsl_sm1_register_from_semantic(struct hlsl_ctx *ctx, const struct hlsl_semantic *semantic, @@ -1811,6 +1832,37 @@ static void write_sm1_dp2add(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer write_sm1_instruction(ctx, buffer, &instr); } +static void write_sm1_ternary_op(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer, + D3DSHADER_INSTRUCTION_OPCODE_TYPE opcode, const struct hlsl_reg *dst, const struct hlsl_reg *src1, + const struct hlsl_reg *src2, const struct hlsl_reg *src3) +{ + struct sm1_instruction instr = + { + .opcode = opcode, + + .dst.type = D3DSPR_TEMP, + .dst.writemask = dst->writemask, + .dst.reg = dst->id, + .has_dst = 1, + + .srcs[0].type = D3DSPR_TEMP, + .srcs[0].swizzle = hlsl_swizzle_from_writemask(src1->writemask), + .srcs[0].reg = src1->id, + .srcs[1].type = D3DSPR_TEMP, + .srcs[1].swizzle = hlsl_swizzle_from_writemask(src2->writemask), + .srcs[1].reg = src2->id, + .srcs[2].type = D3DSPR_TEMP, + .srcs[2].swizzle = hlsl_swizzle_from_writemask(src3->writemask), + .srcs[2].reg = src3->id, + .src_count = 3, + }; + + sm1_map_src_swizzle(&instr.srcs[0], instr.dst.writemask); + sm1_map_src_swizzle(&instr.srcs[1], instr.dst.writemask); + sm1_map_src_swizzle(&instr.srcs[2], instr.dst.writemask); + write_sm1_instruction(ctx, buffer, &instr); +} + static void write_sm1_binary_op(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer, D3DSHADER_INSTRUCTION_OPCODE_TYPE opcode, const struct hlsl_reg *dst, const struct hlsl_reg *src1, const struct hlsl_reg *src2) @@ -2177,6 +2229,10 @@ static void write_sm1_expr(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *b } break; + case HLSL_OP3_CMP: + write_sm1_ternary_op(ctx, buffer, D3DSIO_CMP, &instr->reg, &arg1->reg, &arg2->reg, &arg3->reg); + break; + case HLSL_OP3_DP2ADD: write_sm1_dp2add(ctx, buffer, &instr->reg, &arg1->reg, &arg2->reg, &arg3->reg); break; @@ -2199,7 +2255,7 @@ static void write_sm1_jump(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *b struct sm1_instruction instr = { - .opcode = VKD3D_SM1_OP_TEXKILL, + .opcode = D3DSIO_TEXKILL, .dst.type = D3DSPR_TEMP, .dst.reg = reg->id, @@ -2289,6 +2345,8 @@ static void write_sm1_resource_load(struct hlsl_ctx *ctx, struct vkd3d_bytecode_ .src_count = 2, }; + if (load->load_type == HLSL_RESOURCE_SAMPLE_PROJ) + sm1_instr.opcode |= VKD3DSI_TEXLD_PROJECT << VKD3D_SM1_INSTRUCTION_FLAGS_SHIFT; assert(instr->reg.allocated); diff --git a/libs/vkd3d/libs/vkd3d-shader/dxbc.c b/libs/vkd3d/libs/vkd3d-shader/dxbc.c index 1cb00688c76..7834c1e1615 100644 --- a/libs/vkd3d/libs/vkd3d-shader/dxbc.c +++ b/libs/vkd3d/libs/vkd3d-shader/dxbc.c @@ -94,16 +94,24 @@ static bool require_space(size_t offset, size_t count, size_t size, size_t data_ return !count || (data_size - offset) / count >= size; } -static void read_dword(const char **ptr, uint32_t *d) +static uint32_t read_u32(const char **ptr) { - memcpy(d, *ptr, sizeof(*d)); - *ptr += sizeof(*d); + unsigned int ret; + memcpy(&ret, *ptr, sizeof(ret)); + *ptr += sizeof(ret); + return ret; } -static void read_float(const char **ptr, float *f) +static float read_float(const char **ptr) { + union + { + uint32_t i; + float f; + } u; STATIC_ASSERT(sizeof(float) == sizeof(uint32_t)); - read_dword(ptr, (uint32_t *)f); + u.i = read_u32(ptr); + return u.f; } static void skip_dword_unknown(const char **ptr, unsigned int count) @@ -117,7 +125,7 @@ static void skip_dword_unknown(const char **ptr, unsigned int count) WARN("Skipping %u unknown DWORDs:\n", count); for (i = 0; i < count; ++i) { - read_dword(ptr, &d); + d = read_u32(ptr); WARN("\t0x%08x\n", d); } } @@ -164,7 +172,7 @@ static int parse_dxbc(const struct vkd3d_shader_code *dxbc, struct vkd3d_shader_ return VKD3D_ERROR_INVALID_ARGUMENT; } - read_dword(&ptr, &tag); + tag = read_u32(&ptr); TRACE("tag: %#x.\n", tag); if (tag != TAG_DXBC) @@ -174,10 +182,10 @@ static int parse_dxbc(const struct vkd3d_shader_code *dxbc, struct vkd3d_shader_ return VKD3D_ERROR_INVALID_ARGUMENT; } - read_dword(&ptr, &checksum[0]); - read_dword(&ptr, &checksum[1]); - read_dword(&ptr, &checksum[2]); - read_dword(&ptr, &checksum[3]); + checksum[0] = read_u32(&ptr); + checksum[1] = read_u32(&ptr); + checksum[2] = read_u32(&ptr); + checksum[3] = read_u32(&ptr); vkd3d_compute_dxbc_checksum(data, data_size, calculated_checksum); if (memcmp(checksum, calculated_checksum, sizeof(checksum))) { @@ -191,7 +199,7 @@ static int parse_dxbc(const struct vkd3d_shader_code *dxbc, struct vkd3d_shader_ return VKD3D_ERROR_INVALID_ARGUMENT; } - read_dword(&ptr, &version); + version = read_u32(&ptr); TRACE("version: %#x.\n", version); if (version != 0x00000001) { @@ -201,10 +209,10 @@ static int parse_dxbc(const struct vkd3d_shader_code *dxbc, struct vkd3d_shader_ return VKD3D_ERROR_INVALID_ARGUMENT; } - read_dword(&ptr, &total_size); + total_size = read_u32(&ptr); TRACE("total size: %#x\n", total_size); - read_dword(&ptr, &chunk_count); + chunk_count = read_u32(&ptr); TRACE("chunk count: %#x\n", chunk_count); if (!(sections = vkd3d_calloc(chunk_count, sizeof(*sections)))) @@ -219,7 +227,7 @@ static int parse_dxbc(const struct vkd3d_shader_code *dxbc, struct vkd3d_shader_ const char *chunk_ptr; uint32_t chunk_offset; - read_dword(&ptr, &chunk_offset); + chunk_offset = read_u32(&ptr); TRACE("chunk %u at offset %#x\n", i, chunk_offset); if (chunk_offset >= data_size || !require_space(chunk_offset, 2, sizeof(DWORD), data_size)) @@ -233,8 +241,8 @@ static int parse_dxbc(const struct vkd3d_shader_code *dxbc, struct vkd3d_shader_ chunk_ptr = data + chunk_offset; - read_dword(&chunk_ptr, &chunk_tag); - read_dword(&chunk_ptr, &chunk_size); + chunk_tag = read_u32(&chunk_ptr); + chunk_size = read_u32(&chunk_ptr); if (!require_space(chunk_ptr - data, 1, chunk_size, data_size)) { @@ -359,10 +367,10 @@ static int shader_parse_signature(const struct vkd3d_shader_dxbc_section_desc *s return VKD3D_ERROR_INVALID_ARGUMENT; } - read_dword(&ptr, &count); + count = read_u32(&ptr); TRACE("%u elements.\n", count); - read_dword(&ptr, &header_size); + header_size = read_u32(&ptr); i = header_size / sizeof(uint32_t); if (align(header_size, sizeof(uint32_t)) != header_size || i < 2 || !require_space(2, i - 2, sizeof(uint32_t), section->data.size)) @@ -396,24 +404,24 @@ static int shader_parse_signature(const struct vkd3d_shader_dxbc_section_desc *s e[i].sort_index = i; if (has_stream_index) - read_dword(&ptr, &e[i].stream_index); + e[i].stream_index = read_u32(&ptr); else e[i].stream_index = 0; - read_dword(&ptr, &name_offset); + name_offset = read_u32(&ptr); if (!(e[i].semantic_name = shader_get_string(data, section->data.size, name_offset))) { WARN("Invalid name offset %#x (data size %#zx).\n", name_offset, section->data.size); vkd3d_free(e); return VKD3D_ERROR_INVALID_ARGUMENT; } - read_dword(&ptr, &e[i].semantic_index); - read_dword(&ptr, &e[i].sysval_semantic); - read_dword(&ptr, &e[i].component_type); - read_dword(&ptr, &e[i].register_index); + e[i].semantic_index = read_u32(&ptr); + e[i].sysval_semantic = read_u32(&ptr); + e[i].component_type = read_u32(&ptr); + e[i].register_index = read_u32(&ptr); e[i].target_location = e[i].register_index; e[i].register_count = 1; - read_dword(&ptr, &mask); + mask = read_u32(&ptr); e[i].mask = mask & 0xff; e[i].used_mask = (mask >> 8) & 0xff; switch (section->tag) @@ -431,10 +439,12 @@ static int shader_parse_signature(const struct vkd3d_shader_dxbc_section_desc *s } if (has_min_precision) - read_dword(&ptr, &e[i].min_precision); + e[i].min_precision = read_u32(&ptr); else e[i].min_precision = VKD3D_SHADER_MINIMUM_PRECISION_NONE; + e[i].interpolation_mode = VKD3DSIM_NONE; + TRACE("Stream: %u, semantic: %s, semantic idx: %u, sysval_semantic %#x, " "type %u, register idx: %u, use_mask %#x, input_mask %#x, precision %u.\n", e[i].stream_index, debugstr_a(e[i].semantic_name), e[i].semantic_index, e[i].sysval_semantic, @@ -452,7 +462,7 @@ static int isgn_handler(const struct vkd3d_shader_dxbc_section_desc *section, { struct shader_signature *is = ctx; - if (section->tag != TAG_ISGN) + if (section->tag != TAG_ISGN && section->tag != TAG_ISG1) return VKD3D_OK; if (is->elements) @@ -595,11 +605,11 @@ static int shader_parse_descriptor_ranges(struct root_signature_parser_context * for (i = 0; i < count; ++i) { - read_dword(&ptr, &ranges[i].range_type); - read_dword(&ptr, &ranges[i].descriptor_count); - read_dword(&ptr, &ranges[i].base_shader_register); - read_dword(&ptr, &ranges[i].register_space); - read_dword(&ptr, &ranges[i].descriptor_table_offset); + ranges[i].range_type = read_u32(&ptr); + ranges[i].descriptor_count = read_u32(&ptr); + ranges[i].base_shader_register = read_u32(&ptr); + ranges[i].register_space = read_u32(&ptr); + ranges[i].descriptor_table_offset = read_u32(&ptr); TRACE("Type %#x, descriptor count %u, base shader register %u, " "register space %u, offset %u.\n", @@ -638,12 +648,12 @@ static int shader_parse_descriptor_ranges1(struct root_signature_parser_context for (i = 0; i < count; ++i) { - read_dword(&ptr, &ranges[i].range_type); - read_dword(&ptr, &ranges[i].descriptor_count); - read_dword(&ptr, &ranges[i].base_shader_register); - read_dword(&ptr, &ranges[i].register_space); - read_dword(&ptr, &ranges[i].flags); - read_dword(&ptr, &ranges[i].descriptor_table_offset); + ranges[i].range_type = read_u32(&ptr); + ranges[i].descriptor_count = read_u32(&ptr); + ranges[i].base_shader_register = read_u32(&ptr); + ranges[i].register_space = read_u32(&ptr); + ranges[i].flags = read_u32(&ptr); + ranges[i].descriptor_table_offset = read_u32(&ptr); TRACE("Type %#x, descriptor count %u, base shader register %u, " "register space %u, flags %#x, offset %u.\n", @@ -671,8 +681,8 @@ static int shader_parse_descriptor_table(struct root_signature_parser_context *c } ptr = &context->data[offset]; - read_dword(&ptr, &count); - read_dword(&ptr, &offset); + count = read_u32(&ptr); + offset = read_u32(&ptr); TRACE("Descriptor range count %u.\n", count); @@ -698,8 +708,8 @@ static int shader_parse_descriptor_table1(struct root_signature_parser_context * } ptr = &context->data[offset]; - read_dword(&ptr, &count); - read_dword(&ptr, &offset); + count = read_u32(&ptr); + offset = read_u32(&ptr); TRACE("Descriptor range count %u.\n", count); @@ -723,9 +733,9 @@ static int shader_parse_root_constants(struct root_signature_parser_context *con } ptr = &context->data[offset]; - read_dword(&ptr, &constants->shader_register); - read_dword(&ptr, &constants->register_space); - read_dword(&ptr, &constants->value_count); + constants->shader_register = read_u32(&ptr); + constants->register_space = read_u32(&ptr); + constants->value_count = read_u32(&ptr); TRACE("Shader register %u, register space %u, 32-bit value count %u.\n", constants->shader_register, constants->register_space, constants->value_count); @@ -745,8 +755,8 @@ static int shader_parse_root_descriptor(struct root_signature_parser_context *co } ptr = &context->data[offset]; - read_dword(&ptr, &descriptor->shader_register); - read_dword(&ptr, &descriptor->register_space); + descriptor->shader_register = read_u32(&ptr); + descriptor->register_space = read_u32(&ptr); TRACE("Shader register %u, register space %u.\n", descriptor->shader_register, descriptor->register_space); @@ -777,9 +787,9 @@ static int shader_parse_root_descriptor1(struct root_signature_parser_context *c } ptr = &context->data[offset]; - read_dword(&ptr, &descriptor->shader_register); - read_dword(&ptr, &descriptor->register_space); - read_dword(&ptr, &descriptor->flags); + descriptor->shader_register = read_u32(&ptr); + descriptor->register_space = read_u32(&ptr); + descriptor->flags = read_u32(&ptr); TRACE("Shader register %u, register space %u, flags %#x.\n", descriptor->shader_register, descriptor->register_space, descriptor->flags); @@ -805,9 +815,9 @@ static int shader_parse_root_parameters(struct root_signature_parser_context *co for (i = 0; i < count; ++i) { - read_dword(&ptr, ¶meters[i].parameter_type); - read_dword(&ptr, ¶meters[i].shader_visibility); - read_dword(&ptr, &offset); + parameters[i].parameter_type = read_u32(&ptr); + parameters[i].shader_visibility = read_u32(&ptr); + offset = read_u32(&ptr); TRACE("Type %#x, shader visibility %#x.\n", parameters[i].parameter_type, parameters[i].shader_visibility); @@ -853,9 +863,9 @@ static int shader_parse_root_parameters1(struct root_signature_parser_context *c for (i = 0; i < count; ++i) { - read_dword(&ptr, ¶meters[i].parameter_type); - read_dword(&ptr, ¶meters[i].shader_visibility); - read_dword(&ptr, &offset); + parameters[i].parameter_type = read_u32(&ptr); + parameters[i].shader_visibility = read_u32(&ptr); + offset = read_u32(&ptr); TRACE("Type %#x, shader visibility %#x.\n", parameters[i].parameter_type, parameters[i].shader_visibility); @@ -900,19 +910,19 @@ static int shader_parse_static_samplers(struct root_signature_parser_context *co for (i = 0; i < count; ++i) { - read_dword(&ptr, &sampler_descs[i].filter); - read_dword(&ptr, &sampler_descs[i].address_u); - read_dword(&ptr, &sampler_descs[i].address_v); - read_dword(&ptr, &sampler_descs[i].address_w); - read_float(&ptr, &sampler_descs[i].mip_lod_bias); - read_dword(&ptr, &sampler_descs[i].max_anisotropy); - read_dword(&ptr, &sampler_descs[i].comparison_func); - read_dword(&ptr, &sampler_descs[i].border_colour); - read_float(&ptr, &sampler_descs[i].min_lod); - read_float(&ptr, &sampler_descs[i].max_lod); - read_dword(&ptr, &sampler_descs[i].shader_register); - read_dword(&ptr, &sampler_descs[i].register_space); - read_dword(&ptr, &sampler_descs[i].shader_visibility); + sampler_descs[i].filter = read_u32(&ptr); + sampler_descs[i].address_u = read_u32(&ptr); + sampler_descs[i].address_v = read_u32(&ptr); + sampler_descs[i].address_w = read_u32(&ptr); + sampler_descs[i].mip_lod_bias = read_float(&ptr); + sampler_descs[i].max_anisotropy = read_u32(&ptr); + sampler_descs[i].comparison_func = read_u32(&ptr); + sampler_descs[i].border_colour = read_u32(&ptr); + sampler_descs[i].min_lod = read_float(&ptr); + sampler_descs[i].max_lod = read_float(&ptr); + sampler_descs[i].shader_register = read_u32(&ptr); + sampler_descs[i].register_space = read_u32(&ptr); + sampler_descs[i].shader_visibility = read_u32(&ptr); } return VKD3D_OK; @@ -936,7 +946,7 @@ static int shader_parse_root_signature(const struct vkd3d_shader_code *data, return VKD3D_ERROR_INVALID_ARGUMENT; } - read_dword(&ptr, &version); + version = read_u32(&ptr); TRACE("Version %#x.\n", version); if (version != VKD3D_SHADER_ROOT_SIGNATURE_VERSION_1_0 && version != VKD3D_SHADER_ROOT_SIGNATURE_VERSION_1_1) { @@ -945,8 +955,8 @@ static int shader_parse_root_signature(const struct vkd3d_shader_code *data, } desc->version = version; - read_dword(&ptr, &count); - read_dword(&ptr, &offset); + count = read_u32(&ptr); + offset = read_u32(&ptr); TRACE("Parameter count %u, offset %u.\n", count, offset); if (desc->version == VKD3D_SHADER_ROOT_SIGNATURE_VERSION_1_0) @@ -980,8 +990,8 @@ static int shader_parse_root_signature(const struct vkd3d_shader_code *data, } } - read_dword(&ptr, &count); - read_dword(&ptr, &offset); + count = read_u32(&ptr); + offset = read_u32(&ptr); TRACE("Static sampler count %u, offset %u.\n", count, offset); v_1_0->static_sampler_count = count; @@ -995,7 +1005,7 @@ static int shader_parse_root_signature(const struct vkd3d_shader_code *data, return ret; } - read_dword(&ptr, &v_1_0->flags); + v_1_0->flags = read_u32(&ptr); TRACE("Flags %#x.\n", v_1_0->flags); return VKD3D_OK; diff --git a/libs/vkd3d/libs/vkd3d-shader/dxil.c b/libs/vkd3d/libs/vkd3d-shader/dxil.c index b78c78d34a7..1709212fa99 100644 --- a/libs/vkd3d/libs/vkd3d-shader/dxil.c +++ b/libs/vkd3d/libs/vkd3d-shader/dxil.c @@ -20,9 +20,18 @@ #define VKD3D_SM6_VERSION_MAJOR(version) (((version) >> 4) & 0xf) #define VKD3D_SM6_VERSION_MINOR(version) (((version) >> 0) & 0xf) +/* Two seems to be the maximum but leave some extra room. */ +#define VKD3D_SM6_MAX_METADATA_TABLES 4 #define BITCODE_MAGIC VKD3D_MAKE_TAG('B', 'C', 0xc0, 0xde) #define DXIL_OP_MAX_OPERANDS 17 +static const uint64_t MAX_ALIGNMENT_EXPONENT = 29; +static const uint64_t GLOBALVAR_FLAG_IS_CONSTANT = 1; +static const uint64_t GLOBALVAR_FLAG_EXPLICIT_TYPE = 2; +static const unsigned int GLOBALVAR_ADDRESS_SPACE_SHIFT = 2; +static const unsigned int SHADER_DESCRIPTOR_TYPE_COUNT = 4; + +static const unsigned int dx_max_thread_group_size[3] = {1024, 1024, 64}; enum bitcode_block_id { @@ -114,6 +123,19 @@ enum bitcode_function_code FUNC_CODE_INST_CMPXCHG = 46, }; +enum bitcode_metadata_code +{ + METADATA_STRING = 1, + METADATA_VALUE = 2, + METADATA_NODE = 3, + METADATA_NAME = 4, + METADATA_DISTINCT_NODE = 5, + METADATA_KIND = 6, + METADATA_LOCATION = 7, + METADATA_NAMED_NODE = 10, + METADATA_ATTACHMENT = 11, +}; + enum bitcode_type_code { TYPE_CODE_NUMENTRY = 1, @@ -139,9 +161,190 @@ enum bitcode_value_symtab_code VST_CODE_BBENTRY = 2, }; +enum bitcode_linkage +{ + LINKAGE_EXTERNAL = 0, + LINKAGE_APPENDING = 2, + LINKAGE_INTERNAL = 3, +}; + +enum dxil_component_type +{ + COMPONENT_TYPE_INVALID = 0, + COMPONENT_TYPE_I1 = 1, + COMPONENT_TYPE_I16 = 2, + COMPONENT_TYPE_U16 = 3, + COMPONENT_TYPE_I32 = 4, + COMPONENT_TYPE_U32 = 5, + COMPONENT_TYPE_I64 = 6, + COMPONENT_TYPE_U64 = 7, + COMPONENT_TYPE_F16 = 8, + COMPONENT_TYPE_F32 = 9, + COMPONENT_TYPE_F64 = 10, + COMPONENT_TYPE_SNORMF16 = 11, + COMPONENT_TYPE_UNORMF16 = 12, + COMPONENT_TYPE_SNORMF32 = 13, + COMPONENT_TYPE_UNORMF32 = 14, + COMPONENT_TYPE_SNORMF64 = 15, + COMPONENT_TYPE_UNORMF64 = 16, + COMPONENT_TYPE_PACKEDS8X32 = 17, + COMPONENT_TYPE_PACKEDU8X32 = 18, +}; + +enum dxil_semantic_kind +{ + SEMANTIC_KIND_ARBITRARY = 0, + SEMANTIC_KIND_VERTEXID = 1, + SEMANTIC_KIND_INSTANCEID = 2, + SEMANTIC_KIND_POSITION = 3, + SEMANTIC_KIND_RTARRAYINDEX = 4, + SEMANTIC_KIND_VIEWPORTARRAYINDEX = 5, + SEMANTIC_KIND_CLIPDISTANCE = 6, + SEMANTIC_KIND_CULLDISTANCE = 7, + SEMANTIC_KIND_OUTPUTCONTROLPOINTID = 8, + SEMANTIC_KIND_DOMAINLOCATION = 9, + SEMANTIC_KIND_PRIMITIVEID = 10, + SEMANTIC_KIND_GSINSTANCEID = 11, + SEMANTIC_KIND_SAMPLEINDEX = 12, + SEMANTIC_KIND_ISFRONTFACE = 13, + SEMANTIC_KIND_COVERAGE = 14, + SEMANTIC_KIND_INNERCOVERAGE = 15, + SEMANTIC_KIND_TARGET = 16, + SEMANTIC_KIND_DEPTH = 17, + SEMANTIC_KIND_DEPTHLESSEQUAL = 18, + SEMANTIC_KIND_DEPTHGREATEREQUAL = 19, + SEMANTIC_KIND_STENCILREF = 20, + SEMANTIC_KIND_DISPATCHTHREADID = 21, + SEMANTIC_KIND_GROUPID = 22, + SEMANTIC_KIND_GROUPINDEX = 23, + SEMANTIC_KIND_GROUPTHREADID = 24, + SEMANTIC_KIND_TESSFACTOR = 25, + SEMANTIC_KIND_INSIDETESSFACTOR = 26, + SEMANTIC_KIND_VIEWID = 27, + SEMANTIC_KIND_BARYCENTRICS = 28, + SEMANTIC_KIND_SHADINGRATE = 29, + SEMANTIC_KIND_CULLPRIMITIVE = 30, + SEMANTIC_KIND_COUNT = 31, + SEMANTIC_KIND_INVALID = SEMANTIC_KIND_COUNT, +}; + +enum dxil_element_additional_tag +{ + ADDITIONAL_TAG_STREAM_INDEX = 0, + ADDITIONAL_TAG_GLOBAL_SYMBOL = 1, /* not used */ + ADDITIONAL_TAG_RELADDR_MASK = 2, + ADDITIONAL_TAG_USED_MASK = 3, +}; + +enum dxil_shader_properties_tag +{ + SHADER_PROPERTIES_FLAGS = 0, + SHADER_PROPERTIES_GEOMETRY = 1, + SHADER_PROPERTIES_DOMAIN = 2, + SHADER_PROPERTIES_HULL = 3, + SHADER_PROPERTIES_COMPUTE = 4, + SHADER_PROPERTIES_AUTO_BINDING_SPACE = 5, + SHADER_PROPERTIES_RAY_PAYLOAD_SIZE = 6, + SHADER_PROPERTIES_RAY_ATTRIB_SIZE = 7, + SHADER_PROPERTIES_SHADER_KIND = 8, + SHADER_PROPERTIES_MESH = 9, + SHADER_PROPERTIES_AMPLIFICATION = 10, + SHADER_PROPERTIES_WAVE_SIZE = 11, + SHADER_PROPERTIES_ENTRY_ROOT_SIG = 12, +}; + +enum dxil_binop_code +{ + BINOP_ADD = 0, + BINOP_SUB = 1, + BINOP_MUL = 2, + BINOP_UDIV = 3, + BINOP_SDIV = 4, + BINOP_UREM = 5, + BINOP_SREM = 6, + BINOP_SHL = 7, + BINOP_LSHR = 8, + BINOP_ASHR = 9, + BINOP_AND = 10, + BINOP_OR = 11, + BINOP_XOR = 12 +}; + +enum dxil_fast_fp_flags +{ + FP_ALLOW_UNSAFE_ALGEBRA = 0x1, + FP_NO_NAN = 0x2, + FP_NO_INF = 0x4, + FP_NO_SIGNED_ZEROS = 0x8, + FP_ALLOW_RECIPROCAL = 0x10, +}; + +enum dxil_overflowing_binop_flags +{ + /* Operation is known to never overflow. */ + OB_NO_UNSIGNED_WRAP = 0x1, + OB_NO_SIGNED_WRAP = 0x2, +}; + +enum dxil_possibly_exact_binop_flags +{ + /* "A udiv or sdiv instruction, which can be marked as "exact", indicating that no bits are destroyed." */ + PEB_EXACT = 0x1, +}; + enum dx_intrinsic_opcode { + DX_LOAD_INPUT = 4, DX_STORE_OUTPUT = 5, + DX_CREATE_HANDLE = 57, + DX_CBUFFER_LOAD_LEGACY = 59, +}; + +enum dxil_cast_code +{ + CAST_TRUNC = 0, + CAST_ZEXT = 1, + CAST_SEXT = 2, + CAST_FPTOUI = 3, + CAST_FPTOSI = 4, + CAST_UITOFP = 5, + CAST_SITOFP = 6, + CAST_FPTRUNC = 7, + CAST_FPEXT = 8, + CAST_PTRTOINT = 9, + CAST_INTTOPTR = 10, + CAST_BITCAST = 11, + CAST_ADDRSPACECAST = 12, +}; + +enum dxil_predicate +{ + FCMP_FALSE = 0, + FCMP_OEQ = 1, + FCMP_OGT = 2, + FCMP_OGE = 3, + FCMP_OLT = 4, + FCMP_OLE = 5, + FCMP_ONE = 6, + FCMP_ORD = 7, + FCMP_UNO = 8, + FCMP_UEQ = 9, + FCMP_UGT = 10, + FCMP_UGE = 11, + FCMP_ULT = 12, + FCMP_ULE = 13, + FCMP_UNE = 14, + FCMP_TRUE = 15, + ICMP_EQ = 32, + ICMP_NE = 33, + ICMP_UGT = 34, + ICMP_UGE = 35, + ICMP_ULT = 36, + ICMP_ULE = 37, + ICMP_SGT = 38, + ICMP_SGE = 39, + ICMP_SLT = 40, + ICMP_SLE = 41, }; struct sm6_pointer_info @@ -201,6 +404,8 @@ enum sm6_value_type { VALUE_TYPE_FUNCTION, VALUE_TYPE_REG, + VALUE_TYPE_ICB, + VALUE_TYPE_HANDLE, }; struct sm6_function_data @@ -210,6 +415,12 @@ struct sm6_function_data unsigned int attribs_id; }; +struct sm6_handle_data +{ + const struct sm6_descriptor_info *d; + struct vkd3d_shader_register reg; +}; + struct sm6_value { const struct sm6_type *type; @@ -219,6 +430,8 @@ struct sm6_value { struct sm6_function_data function; struct vkd3d_shader_register reg; + const struct vkd3d_shader_immediate_constant_buffer *icb; + struct sm6_handle_data handle; } u; }; @@ -277,6 +490,59 @@ struct dxil_block size_t record_count; }; +enum sm6_metadata_type +{ + VKD3D_METADATA_KIND, + VKD3D_METADATA_NODE, + VKD3D_METADATA_STRING, + VKD3D_METADATA_VALUE, +}; + +struct sm6_metadata_node +{ + bool is_distinct; + unsigned int operand_count; + struct sm6_metadata_value *operands[]; +}; + +struct sm6_metadata_kind +{ + uint64_t id; + char *name; +}; + +struct sm6_metadata_value +{ + enum sm6_metadata_type type; + const struct sm6_type *value_type; + union + { + char *string_value; + const struct sm6_value *value; + struct sm6_metadata_node *node; + struct sm6_metadata_kind kind; + } u; +}; + +struct sm6_metadata_table +{ + struct sm6_metadata_value *values; + unsigned int count; +}; + +struct sm6_named_metadata +{ + char *name; + struct sm6_metadata_value value; +}; + +struct sm6_descriptor_info +{ + enum vkd3d_shader_descriptor_type type; + unsigned int id; + struct vkd3d_shader_register_range range; +}; + struct sm6_parser { const uint32_t *ptr, *start, *end; @@ -291,19 +557,36 @@ struct sm6_parser struct sm6_type *types; size_t type_count; + struct sm6_type *bool_type; + struct sm6_type *metadata_type; + struct sm6_type *handle_type; struct sm6_symbol *global_symbols; size_t global_symbol_count; + const char *entry_point; + struct vkd3d_shader_dst_param *output_params; + struct vkd3d_shader_dst_param *input_params; struct sm6_function *functions; size_t function_count; + struct sm6_metadata_table metadata_tables[VKD3D_SM6_MAX_METADATA_TABLES]; + struct sm6_named_metadata *named_metadata; + unsigned int named_metadata_count; + + struct sm6_descriptor_info *descriptors; + size_t descriptor_capacity; + size_t descriptor_count; + + unsigned int indexable_temp_count; + struct sm6_value *values; size_t value_count; size_t value_capacity; size_t cur_max_value; + unsigned int ssa_next_id; struct vkd3d_shader_parser p; }; @@ -970,14 +1253,18 @@ static const struct dxil_block *sm6_parser_get_level_one_block(const struct sm6_ return found; } -static char *dxil_record_to_string(const struct dxil_record *record, unsigned int offset) +static char *dxil_record_to_string(const struct dxil_record *record, unsigned int offset, struct sm6_parser *sm6) { unsigned int i; char *str; assert(offset <= record->operand_count); if (!(str = vkd3d_calloc(record->operand_count - offset + 1, 1))) + { + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, + "Out of memory allocating a string of length %u.", record->operand_count - offset); return NULL; + } for (i = offset; i < record->operand_count; ++i) str[i - offset] = record->operands[i]; @@ -1147,6 +1434,8 @@ static enum vkd3d_result sm6_parser_type_table_init(struct sm6_parser *sm6) switch ((width = record->operands[0])) { case 1: + sm6->bool_type = type; + break; case 8: case 16: case 32: @@ -1166,6 +1455,7 @@ static enum vkd3d_result sm6_parser_type_table_init(struct sm6_parser *sm6) case TYPE_CODE_METADATA: type->class = TYPE_CLASS_METADATA; + sm6->metadata_type = type; break; case TYPE_CODE_NUMENTRY: @@ -1188,7 +1478,7 @@ static enum vkd3d_result sm6_parser_type_table_init(struct sm6_parser *sm6) case TYPE_CODE_STRUCT_ANON: case TYPE_CODE_STRUCT_NAMED: - if (!dxil_record_validate_operand_min_count(record, 2, sm6)) + if (!dxil_record_validate_operand_min_count(record, 1, sm6)) return VKD3D_ERROR_INVALID_SHADER; if (record->code == TYPE_CODE_STRUCT_NAMED && !struct_name) { @@ -1227,12 +1517,15 @@ static enum vkd3d_result sm6_parser_type_table_init(struct sm6_parser *sm6) break; } + if (!ascii_strcasecmp(struct_name, "dx.types.Handle")) + sm6->handle_type = type; + type->u.struc->name = struct_name; struct_name = NULL; break; case TYPE_CODE_STRUCT_NAME: - if (!(struct_name = dxil_record_to_string(record, 0))) + if (!(struct_name = dxil_record_to_string(record, 0, sm6))) { ERR("Failed to allocate struct name.\n"); return VKD3D_ERROR_OUT_OF_MEMORY; @@ -1274,6 +1567,16 @@ static inline bool sm6_type_is_integer(const struct sm6_type *type) return type->class == TYPE_CLASS_INTEGER; } +static bool sm6_type_is_bool_i16_i32_i64(const struct sm6_type *type) +{ + return type->class == TYPE_CLASS_INTEGER && (type->u.width == 1 || type->u.width >= 16); +} + +static bool sm6_type_is_bool(const struct sm6_type *type) +{ + return type->class == TYPE_CLASS_INTEGER && type->u.width == 1; +} + static inline bool sm6_type_is_i8(const struct sm6_type *type) { return type->class == TYPE_CLASS_INTEGER && type->u.width == 8; @@ -1289,6 +1592,11 @@ static inline bool sm6_type_is_floating_point(const struct sm6_type *type) return type->class == TYPE_CLASS_FLOAT; } +static bool sm6_type_is_scalar(const struct sm6_type *type) +{ + return type->class == TYPE_CLASS_INTEGER || type->class == TYPE_CLASS_FLOAT || type->class == TYPE_CLASS_POINTER; +} + static inline bool sm6_type_is_numeric(const struct sm6_type *type) { return type->class == TYPE_CLASS_INTEGER || type->class == TYPE_CLASS_FLOAT; @@ -1299,6 +1607,11 @@ static inline bool sm6_type_is_pointer(const struct sm6_type *type) return type->class == TYPE_CLASS_POINTER; } +static bool sm6_type_is_aggregate(const struct sm6_type *type) +{ + return type->class == TYPE_CLASS_STRUCT || type->class == TYPE_CLASS_VECTOR || type->class == TYPE_CLASS_ARRAY; +} + static bool sm6_type_is_numeric_aggregate(const struct sm6_type *type) { unsigned int i; @@ -1321,6 +1634,11 @@ static bool sm6_type_is_numeric_aggregate(const struct sm6_type *type) } } +static bool sm6_type_is_array(const struct sm6_type *type) +{ + return type->class == TYPE_CLASS_ARRAY; +} + static inline bool sm6_type_is_struct(const struct sm6_type *type) { return type->class == TYPE_CLASS_STRUCT; @@ -1364,6 +1682,27 @@ static const struct sm6_type *sm6_type_get_pointer_to_type(const struct sm6_type return NULL; } +/* Call for aggregate types only. */ +static const struct sm6_type *sm6_type_get_element_type_at_index(const struct sm6_type *type, uint64_t elem_idx) +{ + switch (type->class) + { + case TYPE_CLASS_ARRAY: + case TYPE_CLASS_VECTOR: + if (elem_idx >= type->u.array.count) + return NULL; + return type->u.array.elem_type; + + case TYPE_CLASS_STRUCT: + if (elem_idx >= type->u.struc->elem_count) + return NULL; + return type->u.struc->elem_types[elem_idx]; + + default: + vkd3d_unreachable(); + } +} + /* Never returns null for elem_idx 0. */ static const struct sm6_type *sm6_type_get_scalar_type(const struct sm6_type *type, unsigned int elem_idx) { @@ -1388,6 +1727,11 @@ static const struct sm6_type *sm6_type_get_scalar_type(const struct sm6_type *ty } } +static unsigned int sm6_type_max_vector_size(const struct sm6_type *type) +{ + return min((VKD3D_VEC4_SIZE * sizeof(uint32_t) * CHAR_BIT) / type->u.width, VKD3D_VEC4_SIZE); +} + static const struct sm6_type *sm6_parser_get_type(struct sm6_parser *sm6, uint64_t type_id) { if (type_id >= sm6->type_count) @@ -1451,7 +1795,7 @@ static enum vkd3d_result sm6_parser_symtab_init(struct sm6_parser *sm6) symbol = &sm6->global_symbols[sm6->global_symbol_count]; symbol->id = record->operands[0]; - if (!(symbol->name = dxil_record_to_string(record, 1))) + if (!(symbol->name = dxil_record_to_string(record, 1, sm6))) { ERR("Failed to allocate symbol name.\n"); return VKD3D_ERROR_OUT_OF_MEMORY; @@ -1492,10 +1836,10 @@ static const char *sm6_parser_get_global_symbol_name(const struct sm6_parser *sm static unsigned int register_get_uint_value(const struct vkd3d_shader_register *reg) { - if (!register_is_constant(reg) || !data_type_is_integer(reg->data_type)) + if (!register_is_constant(reg) || (!data_type_is_integer(reg->data_type) && !data_type_is_bool(reg->data_type))) return UINT_MAX; - if (reg->immconst_type == VKD3D_IMMCONST_VEC4) + if (reg->dimension == VSIR_DIMENSION_VEC4) WARN("Returning vec4.x.\n"); if (reg->type == VKD3DSPR_IMMCONST64) @@ -1508,6 +1852,17 @@ static unsigned int register_get_uint_value(const struct vkd3d_shader_register * return reg->u.immconst_uint[0]; } +static uint64_t register_get_uint64_value(const struct vkd3d_shader_register *reg) +{ + if (!register_is_constant(reg) || !data_type_is_integer(reg->data_type)) + return UINT64_MAX; + + if (reg->dimension == VSIR_DIMENSION_VEC4) + WARN("Returning vec4.x.\n"); + + return (reg->type == VKD3DSPR_IMMCONST64) ? reg->u.immconst_uint64[0] : reg->u.immconst_uint[0]; +} + static inline bool sm6_value_is_function_dcl(const struct sm6_value *value) { return value->value_type == VALUE_TYPE_FUNCTION; @@ -1530,6 +1885,11 @@ static inline bool sm6_value_is_register(const struct sm6_value *value) return value->value_type == VALUE_TYPE_REG; } +static bool sm6_value_is_handle(const struct sm6_value *value) +{ + return value->value_type == VALUE_TYPE_HANDLE; +} + static inline bool sm6_value_is_constant(const struct sm6_value *value) { return sm6_value_is_register(value) && register_is_constant(&value->u.reg); @@ -1540,6 +1900,11 @@ static inline bool sm6_value_is_undef(const struct sm6_value *value) return sm6_value_is_register(value) && value->u.reg.type == VKD3DSPR_UNDEF; } +static bool sm6_value_is_icb(const struct sm6_value *value) +{ + return value->value_type == VALUE_TYPE_ICB; +} + static inline unsigned int sm6_value_get_constant_uint(const struct sm6_value *value) { if (!sm6_value_is_constant(value)) @@ -1547,6 +1912,11 @@ static inline unsigned int sm6_value_get_constant_uint(const struct sm6_value *v return register_get_uint_value(&value->u.reg); } +static unsigned int sm6_parser_alloc_ssa_id(struct sm6_parser *sm6) +{ + return sm6->ssa_next_id++; +} + static struct vkd3d_shader_src_param *instruction_src_params_alloc(struct vkd3d_shader_instruction *ins, unsigned int count, struct sm6_parser *sm6) { @@ -1555,7 +1925,7 @@ static struct vkd3d_shader_src_param *instruction_src_params_alloc(struct vkd3d_ { ERR("Failed to allocate src params.\n"); vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, - "Out of memory allocating instruction src paramaters."); + "Out of memory allocating instruction src parameters."); return NULL; } ins->src = params; @@ -1571,7 +1941,7 @@ static struct vkd3d_shader_dst_param *instruction_dst_params_alloc(struct vkd3d_ { ERR("Failed to allocate dst params.\n"); vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, - "Out of memory allocating instruction dst paramaters."); + "Out of memory allocating instruction dst parameters."); return NULL; } ins->dst = params; @@ -1579,16 +1949,27 @@ static struct vkd3d_shader_dst_param *instruction_dst_params_alloc(struct vkd3d_ return params; } +static void register_init_with_id(struct vkd3d_shader_register *reg, + enum vkd3d_shader_register_type reg_type, enum vkd3d_data_type data_type, unsigned int id) +{ + vsir_register_init(reg, reg_type, data_type, 1); + reg->idx[0].offset = id; +} + static enum vkd3d_data_type vkd3d_data_type_from_sm6_type(const struct sm6_type *type) { if (type->class == TYPE_CLASS_INTEGER) { switch (type->u.width) { + case 1: + return VKD3D_DATA_BOOL; case 8: return VKD3D_DATA_UINT8; case 32: return VKD3D_DATA_UINT; + case 64: + return VKD3D_DATA_UINT64; default: FIXME("Unhandled width %u.\n", type->u.width); return VKD3D_DATA_UINT; @@ -1612,6 +1993,31 @@ static enum vkd3d_data_type vkd3d_data_type_from_sm6_type(const struct sm6_type return VKD3D_DATA_UINT; } +static void register_init_ssa_vector(struct vkd3d_shader_register *reg, const struct sm6_type *type, + unsigned int component_count, struct sm6_parser *sm6) +{ + enum vkd3d_data_type data_type; + unsigned int id; + + id = sm6_parser_alloc_ssa_id(sm6); + data_type = vkd3d_data_type_from_sm6_type(sm6_type_get_scalar_type(type, 0)); + register_init_with_id(reg, VKD3DSPR_SSA, data_type, id); + reg->dimension = component_count > 1 ? VSIR_DIMENSION_VEC4 : VSIR_DIMENSION_SCALAR; +} + +static void register_init_ssa_scalar(struct vkd3d_shader_register *reg, const struct sm6_type *type, + struct sm6_parser *sm6) +{ + register_init_ssa_vector(reg, type, 1, sm6); +} + +static void dst_param_init(struct vkd3d_shader_dst_param *param) +{ + param->write_mask = VKD3DSP_WRITEMASK_0; + param->modifiers = 0; + param->shift = 0; +} + static inline void dst_param_init_scalar(struct vkd3d_shader_dst_param *param, unsigned int component_idx) { param->write_mask = 1u << component_idx; @@ -1619,40 +2025,88 @@ static inline void dst_param_init_scalar(struct vkd3d_shader_dst_param *param, u param->shift = 0; } +static void dst_param_init_vector(struct vkd3d_shader_dst_param *param, unsigned int component_count) +{ + param->write_mask = (1u << component_count) - 1; + param->modifiers = 0; + param->shift = 0; +} + +static void dst_param_init_ssa_scalar(struct vkd3d_shader_dst_param *param, const struct sm6_type *type, + struct sm6_parser *sm6) +{ + dst_param_init(param); + register_init_ssa_scalar(¶m->reg, type, sm6); +} + static inline void src_param_init(struct vkd3d_shader_src_param *param) { param->swizzle = VKD3D_SHADER_SWIZZLE(X, X, X, X); param->modifiers = VKD3DSPSM_NONE; } +static void src_param_init_scalar(struct vkd3d_shader_src_param *param, unsigned int component_idx) +{ + param->swizzle = vkd3d_shader_create_swizzle(component_idx, component_idx, component_idx, component_idx); + param->modifiers = VKD3DSPSM_NONE; +} + static void src_param_init_from_value(struct vkd3d_shader_src_param *param, const struct sm6_value *src) { src_param_init(param); param->reg = src->u.reg; } -static void register_address_init(struct vkd3d_shader_register *reg, const struct sm6_value *address, - unsigned int idx, struct sm6_parser *sm6) +static void src_param_init_vector_from_reg(struct vkd3d_shader_src_param *param, + const struct vkd3d_shader_register *reg) +{ + param->swizzle = VKD3D_SHADER_NO_SWIZZLE; + param->modifiers = VKD3DSPSM_NONE; + param->reg = *reg; +} + +static void register_index_address_init(struct vkd3d_shader_register_index *idx, const struct sm6_value *address, + struct sm6_parser *sm6) { - assert(idx < ARRAY_SIZE(reg->idx)); if (sm6_value_is_constant(address)) { - reg->idx[idx].offset = sm6_value_get_constant_uint(address); + idx->offset = sm6_value_get_constant_uint(address); } else if (sm6_value_is_undef(address)) { - reg->idx[idx].offset = 0; + idx->offset = 0; } else { struct vkd3d_shader_src_param *rel_addr = shader_parser_get_src_params(&sm6->p, 1); if (rel_addr) src_param_init_from_value(rel_addr, address); - reg->idx[idx].offset = 0; - reg->idx[idx].rel_addr = rel_addr; + idx->offset = 0; + idx->rel_addr = rel_addr; } } +static void instruction_dst_param_init_ssa_scalar(struct vkd3d_shader_instruction *ins, struct sm6_parser *sm6) +{ + struct vkd3d_shader_dst_param *param = instruction_dst_params_alloc(ins, 1, sm6); + struct sm6_value *dst = sm6_parser_get_current_value(sm6); + + dst_param_init_ssa_scalar(param, dst->type, sm6); + param->write_mask = VKD3DSP_WRITEMASK_0; + dst->u.reg = param->reg; +} + +static void instruction_dst_param_init_ssa_vector(struct vkd3d_shader_instruction *ins, + unsigned int component_count, struct sm6_parser *sm6) +{ + struct vkd3d_shader_dst_param *param = instruction_dst_params_alloc(ins, 1, sm6); + struct sm6_value *dst = sm6_parser_get_current_value(sm6); + + dst_param_init_vector(param, component_count); + register_init_ssa_vector(¶m->reg, sm6_type_get_scalar_type(dst->type, 0), component_count, sm6); + dst->u.reg = param->reg; +} + /* Recurse through the block tree while maintaining a current value count. The current * count is the sum of the global count plus all declarations within the current function. * Store into value_capacity the highest count seen. */ @@ -1699,7 +2153,7 @@ static size_t sm6_parser_get_value_index(struct sm6_parser *sm6, uint64_t idx) WARN("Ignoring upper 32 bits of relative index.\n"); i = (uint32_t)sm6->value_count - (uint32_t)idx; - /* This may underflow to produce a forward reference, but it must not exceeed the final value count. */ + /* This may underflow to produce a forward reference, but it must not exceed the final value count. */ if (i >= sm6->cur_max_value) { WARN("Invalid value index %"PRIx64" at %zu.\n", idx, sm6->value_count); @@ -1717,6 +2171,78 @@ static size_t sm6_parser_get_value_index(struct sm6_parser *sm6, uint64_t idx) return i; } +static bool sm6_value_validate_is_register(const struct sm6_value *value, struct sm6_parser *sm6) +{ + if (!sm6_value_is_register(value)) + { + WARN("Operand of type %u is not a register.\n", value->value_type); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "A register operand passed to a DXIL instruction is not a register."); + return false; + } + return true; +} + +static bool sm6_value_validate_is_handle(const struct sm6_value *value, struct sm6_parser *sm6) +{ + if (!sm6_value_is_handle(value)) + { + WARN("Handle parameter of type %u is not a handle.\n", value->value_type); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_RESOURCE_HANDLE, + "A handle parameter passed to a DX intrinsic function is not a handle."); + return false; + } + return true; +} + +static bool sm6_value_validate_is_pointer(const struct sm6_value *value, struct sm6_parser *sm6) +{ + if (!sm6_type_is_pointer(value->type)) + { + WARN("Operand result type class %u is not a pointer.\n", value->type->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "A pointer operand passed to a DXIL instruction is not a pointer."); + return false; + } + return true; +} + +static bool sm6_value_validate_is_numeric(const struct sm6_value *value, struct sm6_parser *sm6) +{ + if (!sm6_type_is_numeric(value->type)) + { + WARN("Operand result type class %u is not numeric.\n", value->type->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "A numeric operand passed to a DXIL instruction is not numeric."); + return false; + } + return true; +} + +static bool sm6_value_validate_is_bool(const struct sm6_value *value, struct sm6_parser *sm6) +{ + const struct sm6_type *type = value->type; + if (!sm6_type_is_bool(type)) + { + WARN("Operand of type class %u is not bool.\n", type->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "A bool operand of type class %u passed to a DXIL instruction is not a bool.", type->class); + return false; + } + return true; +} + +static const struct sm6_value *sm6_parser_get_value_safe(struct sm6_parser *sm6, unsigned int idx) +{ + if (idx < sm6->value_count) + return &sm6->values[idx]; + + WARN("Invalid value index %u.\n", idx); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Invalid value index %u.", idx); + return NULL; +} + static size_t sm6_parser_get_value_idx_by_ref(struct sm6_parser *sm6, const struct dxil_record *record, const struct sm6_type *fwd_type, unsigned int *rec_idx) { @@ -1860,12 +2386,80 @@ static inline double bitcast_uint64_to_double(uint64_t value) return u.double_value; } +static enum vkd3d_result value_allocate_constant_array(struct sm6_value *dst, const struct sm6_type *type, + const uint64_t *operands, struct sm6_parser *sm6) +{ + struct vkd3d_shader_immediate_constant_buffer *icb; + const struct sm6_type *elem_type; + unsigned int i, size, count; + + elem_type = type->u.array.elem_type; + /* Multidimensional arrays are emitted in flattened form. */ + if (elem_type->class != TYPE_CLASS_INTEGER && elem_type->class != TYPE_CLASS_FLOAT) + { + FIXME("Unhandled element type %u for data array.\n", elem_type->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "The element data type for an immediate constant buffer is not scalar integer or floating point."); + return VKD3D_ERROR_INVALID_SHADER; + } + + /* Arrays of bool are not used in DXIL. dxc will emit an array of int32 instead if necessary. */ + if (!(size = elem_type->u.width / CHAR_BIT)) + { + WARN("Invalid data type width %u.\n", elem_type->u.width); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "An immediate constant buffer is declared with boolean elements."); + return VKD3D_ERROR_INVALID_SHADER; + } + size = max(size, sizeof(icb->data[0])); + count = type->u.array.count * size / sizeof(icb->data[0]); + + if (!(icb = vkd3d_malloc(offsetof(struct vkd3d_shader_immediate_constant_buffer, data[count])))) + { + ERR("Failed to allocate buffer, count %u.\n", count); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, + "Out of memory allocating an immediate constant buffer of count %u.", count); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + if (!shader_instruction_array_add_icb(&sm6->p.instructions, icb)) + { + ERR("Failed to store icb object.\n"); + vkd3d_free(icb); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, + "Out of memory storing an immediate constant buffer object."); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + + dst->value_type = VALUE_TYPE_ICB; + dst->u.icb = icb; + + icb->data_type = vkd3d_data_type_from_sm6_type(elem_type); + icb->element_count = type->u.array.count; + icb->component_count = 1; + + count = type->u.array.count; + if (size > sizeof(icb->data[0])) + { + uint64_t *data = (uint64_t *)icb->data; + for (i = 0; i < count; ++i) + data[i] = operands[i]; + } + else + { + for (i = 0; i < count; ++i) + icb->data[i] = operands[i]; + } + + return VKD3D_OK; +} + static enum vkd3d_result sm6_parser_constants_init(struct sm6_parser *sm6, const struct dxil_block *block) { enum vkd3d_shader_register_type reg_type = VKD3DSPR_INVALID; const struct sm6_type *type, *elem_type; enum vkd3d_data_type reg_data_type; const struct dxil_record *record; + enum vkd3d_result ret; struct sm6_value *dst; size_t i, value_idx; uint64_t value; @@ -1911,14 +2505,19 @@ static enum vkd3d_result sm6_parser_constants_init(struct sm6_parser *sm6, const dst = sm6_parser_get_current_value(sm6); dst->type = type; dst->value_type = VALUE_TYPE_REG; - dst->u.reg.type = reg_type; - dst->u.reg.immconst_type = VKD3D_IMMCONST_SCALAR; - dst->u.reg.data_type = reg_data_type; + vsir_register_init(&dst->u.reg, reg_type, reg_data_type, 0); switch (record->code) { case CST_CODE_NULL: - /* Register constant data is already zero-filled. */ + if (sm6_type_is_array(type)) + { + FIXME("Constant null arrays are not supported.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Constant null arrays are not supported."); + return VKD3D_ERROR_INVALID_SHADER; + } + /* For non-aggregates, register constant data is already zero-filled. */ break; case CST_CODE_INTEGER: @@ -1961,7 +2560,19 @@ static enum vkd3d_result sm6_parser_constants_init(struct sm6_parser *sm6, const break; case CST_CODE_DATA: - WARN("Unhandled constant array.\n"); + if (!sm6_type_is_array(type)) + { + WARN("Invalid type %u for data constant idx %zu.\n", type->class, value_idx); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "The type of a constant array is not an array type."); + return VKD3D_ERROR_INVALID_SHADER; + } + if (!dxil_record_validate_operand_count(record, type->u.array.count, type->u.array.count, sm6)) + return VKD3D_ERROR_INVALID_SHADER; + + if ((ret = value_allocate_constant_array(dst, type, record->operands, sm6)) < 0) + return ret; + break; case CST_CODE_UNDEF: @@ -1983,7 +2594,18 @@ static enum vkd3d_result sm6_parser_constants_init(struct sm6_parser *sm6, const return VKD3D_OK; } -static struct vkd3d_shader_instruction *sm6_parser_require_space(struct sm6_parser *sm6, size_t extra) +static bool bitcode_parse_alignment(uint64_t encoded_alignment, unsigned int *alignment) +{ + if (encoded_alignment > MAX_ALIGNMENT_EXPONENT + 1) + { + *alignment = 0; + return false; + } + *alignment = (1u << encoded_alignment) >> 1; + return true; +} + +static struct vkd3d_shader_instruction *sm6_parser_require_space(struct sm6_parser *sm6, size_t extra) { if (!shader_instruction_array_reserve(&sm6->p.instructions, sm6->p.instructions.count + extra)) { @@ -1999,15 +2621,194 @@ static struct vkd3d_shader_instruction *sm6_parser_add_instruction(struct sm6_pa { struct vkd3d_shader_instruction *ins = sm6_parser_require_space(sm6, 1); assert(ins); - shader_instruction_init(ins, handler_idx); + vsir_instruction_init(ins, &sm6->p.location, handler_idx); ++sm6->p.instructions.count; return ins; } +static void sm6_parser_declare_indexable_temp(struct sm6_parser *sm6, const struct sm6_type *elem_type, + unsigned int count, unsigned int alignment, unsigned int init, struct sm6_value *dst) +{ + enum vkd3d_data_type data_type = vkd3d_data_type_from_sm6_type(elem_type); + struct vkd3d_shader_instruction *ins; + + ins = sm6_parser_add_instruction(sm6, VKD3DSIH_DCL_INDEXABLE_TEMP); + ins->declaration.indexable_temp.register_idx = sm6->indexable_temp_count++; + ins->declaration.indexable_temp.register_size = count; + ins->declaration.indexable_temp.alignment = alignment; + ins->declaration.indexable_temp.data_type = data_type; + ins->declaration.indexable_temp.component_count = 1; + /* The initialiser value index will be resolved later so forward references can be handled. */ + ins->declaration.indexable_temp.initialiser = (void *)(uintptr_t)init; + + register_init_with_id(&dst->u.reg, VKD3DSPR_IDXTEMP, data_type, ins->declaration.indexable_temp.register_idx); +} + +static bool sm6_parser_declare_global(struct sm6_parser *sm6, const struct dxil_record *record) +{ + const struct sm6_type *type, *scalar_type; + unsigned int alignment, count; + uint64_t address_space, init; + struct sm6_value *dst; + bool is_constant; + + if (!dxil_record_validate_operand_min_count(record, 6, sm6)) + return false; + + if (!(type = sm6_parser_get_type(sm6, record->operands[0]))) + return false; + if (sm6_type_is_array(type)) + { + if (!sm6_type_is_scalar(type->u.array.elem_type)) + { + FIXME("Unsupported nested type class %u.\n", type->u.array.elem_type->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Global array variables with nested type class %u are not supported.", + type->u.array.elem_type->class); + return false; + } + count = type->u.array.count; + scalar_type = type->u.array.elem_type; + } + else if (sm6_type_is_scalar(type)) + { + count = 1; + scalar_type = type; + } + else + { + FIXME("Unsupported type class %u.\n", type->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Global variables of type class %u are not supported.", type->class); + return false; + } + + is_constant = record->operands[1] & GLOBALVAR_FLAG_IS_CONSTANT; + + if (record->operands[1] & GLOBALVAR_FLAG_EXPLICIT_TYPE) + { + address_space = record->operands[1] >> GLOBALVAR_ADDRESS_SPACE_SHIFT; + + if (!(type = sm6_type_get_pointer_to_type(type, address_space, sm6))) + { + WARN("Failed to get pointer type for type class %u, address space %"PRIu64".\n", + type->class, address_space); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_MODULE, + "Module does not define a pointer type for a global variable."); + return false; + } + } + else + { + if (!sm6_type_is_pointer(type)) + { + WARN("Type is not a pointer.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "The type of a global variable is not a pointer."); + return false; + } + address_space = type->u.pointer.addr_space; + } + + if ((init = record->operands[2])) + { + if (init - 1 >= sm6->value_capacity) + { + WARN("Invalid value index %"PRIu64" for initialiser.", init - 1); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Global variable initialiser value index %"PRIu64" is invalid.", init - 1); + return false; + } + } + + /* LINKAGE_EXTERNAL is common but not relevant here. */ + if (record->operands[3] != LINKAGE_EXTERNAL && record->operands[3] != LINKAGE_INTERNAL) + WARN("Ignoring linkage %"PRIu64".\n", record->operands[3]); + + if (!bitcode_parse_alignment(record->operands[4], &alignment)) + WARN("Invalid alignment %"PRIu64".\n", record->operands[4]); + + if (record->operands[5]) + WARN("Ignoring section code %"PRIu64".\n", record->operands[5]); + + if (!sm6_parser_get_global_symbol_name(sm6, sm6->value_count)) + WARN("Missing symbol name for global variable at index %zu.\n", sm6->value_count); + /* TODO: store global symbol names in struct vkd3d_shader_desc? */ + + if (record->operand_count > 6 && record->operands[6]) + WARN("Ignoring visibility %"PRIu64".\n", record->operands[6]); + if (record->operand_count > 7 && record->operands[7]) + WARN("Ignoring thread local mode %"PRIu64".\n", record->operands[7]); + /* record->operands[8] contains unnamed_addr, a flag indicating the address + * is not important, only the content is. This info is not relevant. */ + if (record->operand_count > 9 && record->operands[9]) + WARN("Ignoring external_init %"PRIu64".\n", record->operands[9]); + if (record->operand_count > 10 && record->operands[10]) + WARN("Ignoring dll storage class %"PRIu64".\n", record->operands[10]); + if (record->operand_count > 11 && record->operands[11]) + WARN("Ignoring comdat %"PRIu64".\n", record->operands[11]); + + dst = sm6_parser_get_current_value(sm6); + dst->type = type; + dst->value_type = VALUE_TYPE_REG; + + if (is_constant && !init) + { + WARN("Constant array has no initialiser.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "A constant global variable has no initialiser."); + return false; + } + + if (address_space == ADDRESS_SPACE_DEFAULT) + { + sm6_parser_declare_indexable_temp(sm6, scalar_type, count, alignment, init, dst); + } + else if (address_space == ADDRESS_SPACE_GROUPSHARED) + { + FIXME("Unsupported TGSM.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "TGSM global variables are not supported."); + return false; + } + else + { + FIXME("Unhandled address space %"PRIu64".\n", address_space); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Global variables with address space %"PRIu64" are not supported.", address_space); + return false; + } + + ++sm6->value_count; + return true; +} + +static const struct vkd3d_shader_immediate_constant_buffer *resolve_forward_initialiser( + size_t index, struct sm6_parser *sm6) +{ + const struct sm6_value *value; + + assert(index); + --index; + if (!(value = sm6_parser_get_value_safe(sm6, index)) || !sm6_value_is_icb(value)) + { + WARN("Invalid initialiser index %zu.\n", index); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Global variable initialiser value index %zu is invalid.", index); + return NULL; + } + else + { + return value->u.icb; + } +} + static enum vkd3d_result sm6_parser_globals_init(struct sm6_parser *sm6) { const struct dxil_block *block = &sm6->root_block; + struct vkd3d_shader_instruction *ins; const struct dxil_record *record; + enum vkd3d_result ret; uint64_t version; size_t i; @@ -2030,7 +2831,8 @@ static enum vkd3d_result sm6_parser_globals_init(struct sm6_parser *sm6) break; case MODULE_CODE_GLOBALVAR: - FIXME("Global variables are not implemented yet.\n"); + if (!sm6_parser_declare_global(sm6, record)) + return VKD3D_ERROR_INVALID_SHADER; break; case MODULE_CODE_VERSION: @@ -2050,6 +2852,24 @@ static enum vkd3d_result sm6_parser_globals_init(struct sm6_parser *sm6) } } + for (i = 0; i < block->child_block_count; ++i) + { + if (block->child_blocks[i]->id == CONSTANTS_BLOCK + && (ret = sm6_parser_constants_init(sm6, block->child_blocks[i])) < 0) + return ret; + } + + /* Resolve initialiser forward references. */ + for (i = 0; i < sm6->p.instructions.count; ++i) + { + ins = &sm6->p.instructions.elements[i]; + if (ins->handler_idx == VKD3DSIH_DCL_INDEXABLE_TEMP && ins->declaration.indexable_temp.initialiser) + { + ins->declaration.indexable_temp.initialiser = resolve_forward_initialiser( + (uintptr_t)ins->declaration.indexable_temp.initialiser, sm6); + } + } + return VKD3D_OK; } @@ -2063,7 +2883,7 @@ static void dst_param_io_init(struct vkd3d_shader_dst_param *param, param->shift = 0; /* DXIL types do not have signedness. Load signed elements as unsigned. */ component_type = e->component_type == VKD3D_SHADER_COMPONENT_INT ? VKD3D_SHADER_COMPONENT_UINT : e->component_type; - shader_register_init(¶m->reg, reg_type, vkd3d_data_type_from_component_type(component_type), 0); + vsir_register_init(¶m->reg, reg_type, vkd3d_data_type_from_component_type(component_type), 0); } static void sm6_parser_init_signature(struct sm6_parser *sm6, const struct shader_signature *s, @@ -2071,7 +2891,7 @@ static void sm6_parser_init_signature(struct sm6_parser *sm6, const struct shade { struct vkd3d_shader_dst_param *param; const struct signature_element *e; - unsigned int i; + unsigned int i, count; for (i = 0; i < s->element_count; ++i) { @@ -2079,8 +2899,11 @@ static void sm6_parser_init_signature(struct sm6_parser *sm6, const struct shade param = ¶ms[i]; dst_param_io_init(param, e, reg_type); - param->reg.idx[0].offset = i; - param->reg.idx_count = 1; + count = 0; + if (e->register_count > 1) + param->reg.idx[count++].offset = 0; + param->reg.idx[count++].offset = i; + param->reg.idx_count = count; } } @@ -2112,15 +2935,25 @@ static void sm6_parser_emit_signature(struct sm6_parser *sm6, const struct shade param = &ins->declaration.dst; } + ins->flags = e->interpolation_mode; *param = params[i]; + + if (e->register_count > 1) + { + param->reg.idx[0].rel_addr = NULL; + param->reg.idx[0].offset = e->register_count; + } } } static void sm6_parser_init_output_signature(struct sm6_parser *sm6, const struct shader_signature *output_signature) { - sm6_parser_init_signature(sm6, output_signature, - (sm6->p.shader_version.type == VKD3D_SHADER_TYPE_PIXEL) ? VKD3DSPR_COLOROUT : VKD3DSPR_OUTPUT, - sm6->output_params); + sm6_parser_init_signature(sm6, output_signature, VKD3DSPR_OUTPUT, sm6->output_params); +} + +static void sm6_parser_init_input_signature(struct sm6_parser *sm6, const struct shader_signature *input_signature) +{ + sm6_parser_init_signature(sm6, input_signature, VKD3DSPR_INPUT, sm6->input_params); } static void sm6_parser_emit_output_signature(struct sm6_parser *sm6, const struct shader_signature *output_signature) @@ -2128,6 +2961,14 @@ static void sm6_parser_emit_output_signature(struct sm6_parser *sm6, const struc sm6_parser_emit_signature(sm6, output_signature, VKD3DSIH_DCL_OUTPUT, VKD3DSIH_DCL_OUTPUT_SIV, sm6->output_params); } +static void sm6_parser_emit_input_signature(struct sm6_parser *sm6, const struct shader_signature *input_signature) +{ + sm6_parser_emit_signature(sm6, input_signature, + (sm6->p.shader_version.type == VKD3D_SHADER_TYPE_PIXEL) ? VKD3DSIH_DCL_INPUT_PS : VKD3DSIH_DCL_INPUT, + (sm6->p.shader_version.type == VKD3D_SHADER_TYPE_PIXEL) ? VKD3DSIH_DCL_INPUT_PS_SIV : VKD3DSIH_DCL_INPUT_SIV, + sm6->input_params); +} + static const struct sm6_value *sm6_parser_next_function_definition(struct sm6_parser *sm6) { size_t i, count = sm6->function_count; @@ -2150,6 +2991,303 @@ static struct sm6_block *sm6_block_create() return block; } +static enum vkd3d_shader_opcode map_binary_op(uint64_t code, const struct sm6_type *type_a, + const struct sm6_type *type_b, struct sm6_parser *sm6) +{ + bool is_int = sm6_type_is_bool_i16_i32_i64(type_a); + bool is_bool = sm6_type_is_bool(type_a); + enum vkd3d_shader_opcode op; + bool is_valid; + + if (!is_int && !sm6_type_is_floating_point(type_a)) + { + WARN("Argument type %u is not bool, int16/32/64 or floating point.\n", type_a->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "An argument to a binary operation is not bool, int16/32/64 or floating point."); + return VKD3DSIH_INVALID; + } + if (type_a != type_b) + { + WARN("Type mismatch, type %u width %u vs type %u width %u.\n", type_a->class, + type_a->u.width, type_b->class, type_b->u.width); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_TYPE_MISMATCH, + "Type mismatch in binary operation arguments."); + } + + switch (code) + { + case BINOP_ADD: + case BINOP_SUB: + /* NEG is applied later for subtraction. */ + op = is_int ? VKD3DSIH_IADD : VKD3DSIH_ADD; + is_valid = !is_bool; + break; + case BINOP_AND: + op = VKD3DSIH_AND; + is_valid = is_int; + break; + case BINOP_ASHR: + op = VKD3DSIH_ISHR; + is_valid = is_int && !is_bool; + break; + case BINOP_LSHR: + op = VKD3DSIH_USHR; + is_valid = is_int && !is_bool; + break; + case BINOP_MUL: + op = is_int ? VKD3DSIH_UMUL : VKD3DSIH_MUL; + is_valid = !is_bool; + break; + case BINOP_OR: + op = VKD3DSIH_OR; + is_valid = is_int; + break; + case BINOP_SDIV: + op = is_int ? VKD3DSIH_IDIV : VKD3DSIH_DIV; + is_valid = !is_bool; + break; + case BINOP_SREM: + op = is_int ? VKD3DSIH_IDIV : VKD3DSIH_FREM; + is_valid = !is_bool; + break; + case BINOP_SHL: + op = VKD3DSIH_ISHL; + is_valid = is_int && !is_bool; + break; + case BINOP_UDIV: + case BINOP_UREM: + op = VKD3DSIH_UDIV; + is_valid = is_int && !is_bool; + break; + case BINOP_XOR: + op = VKD3DSIH_XOR; + is_valid = is_int; + break; + default: + FIXME("Unhandled binary op %#"PRIx64".\n", code); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Binary operation %#"PRIx64" is unhandled.", code); + return VKD3DSIH_INVALID; + } + + if (!is_valid) + { + WARN("Invalid operation %u for type %u, width %u.\n", op, type_a->class, type_a->u.width); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_INVALID_OPERATION, + "Binary operation %u is invalid on type class %u, width %u.", op, type_a->class, type_a->u.width); + } + + return op; +} + +static void sm6_parser_emit_binop(struct sm6_parser *sm6, const struct dxil_record *record, + struct vkd3d_shader_instruction *ins, struct sm6_value *dst) +{ + struct vkd3d_shader_src_param *src_params; + enum vkd3d_shader_opcode handler_idx; + const struct sm6_value *a, *b; + uint64_t code, flags; + bool silence_warning; + unsigned int i = 0; + + a = sm6_parser_get_value_by_ref(sm6, record, NULL, &i); + b = sm6_parser_get_value_by_ref(sm6, record, a->type, &i); + if (!a || !b) + return; + + if (!dxil_record_validate_operand_count(record, i + 1, i + 2, sm6)) + return; + + code = record->operands[i++]; + if ((handler_idx = map_binary_op(code, a->type, b->type, sm6)) == VKD3DSIH_INVALID) + return; + + vsir_instruction_init(ins, &sm6->p.location, handler_idx); + + flags = (record->operand_count > i) ? record->operands[i] : 0; + silence_warning = false; + + switch (handler_idx) + { + case VKD3DSIH_ADD: + case VKD3DSIH_MUL: + case VKD3DSIH_DIV: + case VKD3DSIH_FREM: + if (!(flags & FP_ALLOW_UNSAFE_ALGEBRA)) + ins->flags |= VKD3DSI_PRECISE_X; + flags &= ~FP_ALLOW_UNSAFE_ALGEBRA; + /* SPIR-V FPFastMathMode is only available in the Kernel executon model. */ + silence_warning = !(flags & ~(FP_NO_NAN | FP_NO_INF | FP_NO_SIGNED_ZEROS | FP_ALLOW_RECIPROCAL)); + break; + case VKD3DSIH_IADD: + case VKD3DSIH_UMUL: + case VKD3DSIH_ISHL: + silence_warning = !(flags & ~(OB_NO_UNSIGNED_WRAP | OB_NO_SIGNED_WRAP)); + break; + case VKD3DSIH_ISHR: + case VKD3DSIH_USHR: + case VKD3DSIH_IDIV: + case VKD3DSIH_UDIV: + silence_warning = !(flags & ~PEB_EXACT); + break; + default: + break; + } + /* The above flags are very common and cause warning spam. */ + if (flags && silence_warning) + { + TRACE("Ignoring flags %#"PRIx64".\n", flags); + } + else if (flags) + { + WARN("Ignoring flags %#"PRIx64".\n", flags); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_IGNORING_OPERANDS, + "Ignoring flags %#"PRIx64" for a binary operation.", flags); + } + + src_params = instruction_src_params_alloc(ins, 2, sm6); + src_param_init_from_value(&src_params[0], a); + src_param_init_from_value(&src_params[1], b); + if (code == BINOP_SUB) + src_params[1].modifiers = VKD3DSPSM_NEG; + + dst->type = a->type; + + if (handler_idx == VKD3DSIH_UMUL || handler_idx == VKD3DSIH_UDIV || handler_idx == VKD3DSIH_IDIV) + { + struct vkd3d_shader_dst_param *dst_params = instruction_dst_params_alloc(ins, 2, sm6); + unsigned int index = code != BINOP_UDIV && code != BINOP_SDIV; + + dst_param_init(&dst_params[0]); + dst_param_init(&dst_params[1]); + register_init_ssa_scalar(&dst_params[index].reg, a->type, sm6); + vsir_register_init(&dst_params[index ^ 1].reg, VKD3DSPR_NULL, VKD3D_DATA_UNUSED, 0); + dst->u.reg = dst_params[index].reg; + } + else + { + instruction_dst_param_init_ssa_scalar(ins, sm6); + } +} + +static void sm6_parser_emit_dx_cbuffer_load(struct sm6_parser *sm6, struct sm6_block *code_block, + enum dx_intrinsic_opcode op, const struct sm6_value **operands, struct vkd3d_shader_instruction *ins) +{ + struct sm6_value *dst = sm6_parser_get_current_value(sm6); + struct vkd3d_shader_src_param *src_param; + const struct sm6_value *buffer; + const struct sm6_type *type; + + buffer = operands[0]; + if (!sm6_value_validate_is_handle(buffer, sm6)) + return; + + vsir_instruction_init(ins, &sm6->p.location, VKD3DSIH_MOV); + + src_param = instruction_src_params_alloc(ins, 1, sm6); + src_param_init_vector_from_reg(src_param, &buffer->u.handle.reg); + register_index_address_init(&src_param->reg.idx[2], operands[1], sm6); + assert(src_param->reg.idx_count == 3); + + type = sm6_type_get_scalar_type(dst->type, 0); + assert(type); + src_param->reg.data_type = vkd3d_data_type_from_sm6_type(type); + + instruction_dst_param_init_ssa_vector(ins, sm6_type_max_vector_size(type), sm6); +} + +static const struct sm6_descriptor_info *sm6_parser_get_descriptor(struct sm6_parser *sm6, + enum vkd3d_shader_descriptor_type type, unsigned int id, const struct sm6_value *address) +{ + const struct sm6_descriptor_info *d; + unsigned int register_index; + size_t i; + + for (i = 0; i < sm6->descriptor_count; ++i) + { + d = &sm6->descriptors[i]; + + if (d->type != type || d->id != id) + continue; + + if (!sm6_value_is_constant(address)) + return d; + + register_index = sm6_value_get_constant_uint(address); + if (register_index >= d->range.first && register_index <= d->range.last) + return d; + } + + return NULL; +} + +static void sm6_parser_emit_dx_create_handle(struct sm6_parser *sm6, struct sm6_block *code_block, + enum dx_intrinsic_opcode op, const struct sm6_value **operands, struct vkd3d_shader_instruction *ins) +{ + enum vkd3d_shader_descriptor_type type; + const struct sm6_descriptor_info *d; + struct vkd3d_shader_register *reg; + struct sm6_value *dst; + unsigned int id; + + type = sm6_value_get_constant_uint(operands[0]); + id = sm6_value_get_constant_uint(operands[1]); + if (!(d = sm6_parser_get_descriptor(sm6, type, id, operands[2]))) + { + WARN("Failed to find resource type %#x, id %#x.\n", type, id); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Descriptor for resource type %#x, id %#x was not found.", type, id); + return; + } + + dst = sm6_parser_get_current_value(sm6); + dst->value_type = VALUE_TYPE_HANDLE; + dst->u.handle.d = d; + + reg = &dst->u.handle.reg; + /* Set idx_count to 3 for use with load instructions. + * TODO: set register type from resource type when other types are supported. */ + vsir_register_init(reg, VKD3DSPR_CONSTBUFFER, VKD3D_DATA_FLOAT, 3); + reg->idx[0].offset = id; + register_index_address_init(®->idx[1], operands[2], sm6); + reg->non_uniform = !!sm6_value_get_constant_uint(operands[3]); + + /* NOP is used to flag no instruction emitted. */ + ins->handler_idx = VKD3DSIH_NOP; +} + +static void sm6_parser_emit_dx_load_input(struct sm6_parser *sm6, struct sm6_block *code_block, + enum dx_intrinsic_opcode op, const struct sm6_value **operands, struct vkd3d_shader_instruction *ins) +{ + struct vkd3d_shader_src_param *src_param; + const struct shader_signature *signature; + unsigned int row_index, column_index; + const struct signature_element *e; + + row_index = sm6_value_get_constant_uint(operands[0]); + column_index = sm6_value_get_constant_uint(operands[2]); + + vsir_instruction_init(ins, &sm6->p.location, VKD3DSIH_MOV); + + signature = &sm6->p.shader_desc.input_signature; + if (row_index >= signature->element_count) + { + WARN("Invalid row index %u.\n", row_index); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Invalid input row index %u.", row_index); + return; + } + e = &signature->elements[row_index]; + + src_param = instruction_src_params_alloc(ins, 1, sm6); + src_param->reg = sm6->input_params[row_index].reg; + src_param_init_scalar(src_param, column_index); + if (e->register_count > 1) + register_index_address_init(&src_param->reg.idx[0], operands[1], sm6); + + instruction_dst_param_init_ssa_scalar(ins, sm6); +} + static void sm6_parser_emit_dx_store_output(struct sm6_parser *sm6, struct sm6_block *code_block, enum dx_intrinsic_opcode op, const struct sm6_value **operands, struct vkd3d_shader_instruction *ins) { @@ -2190,14 +3328,14 @@ static void sm6_parser_emit_dx_store_output(struct sm6_parser *sm6, struct sm6_b return; } - shader_instruction_init(ins, VKD3DSIH_MOV); + vsir_instruction_init(ins, &sm6->p.location, VKD3DSIH_MOV); if (!(dst_param = instruction_dst_params_alloc(ins, 1, sm6))) return; dst_param_init_scalar(dst_param, column_index); dst_param->reg = sm6->output_params[row_index].reg; if (e->register_count > 1) - register_address_init(&dst_param->reg, operands[1], 0, sm6); + register_index_address_init(&dst_param->reg.idx[0], operands[1], sm6); if ((src_param = instruction_src_params_alloc(ins, 1, sm6))) src_param_init_from_value(src_param, value); @@ -2213,17 +3351,29 @@ struct sm6_dx_opcode_info /* 8 -> int8 + b -> constant int1 + c -> constant int8/16/32 i -> int32 + H -> handle v -> void o -> overloaded */ static const struct sm6_dx_opcode_info sm6_dx_op_table[] = { + [DX_CBUFFER_LOAD_LEGACY ] = {'o', "Hi", sm6_parser_emit_dx_cbuffer_load}, + [DX_CREATE_HANDLE ] = {'H', "ccib", sm6_parser_emit_dx_create_handle}, + [DX_LOAD_INPUT ] = {'o', "ii8i", sm6_parser_emit_dx_load_input}, [DX_STORE_OUTPUT ] = {'v', "ii8o", sm6_parser_emit_dx_store_output}, }; -static bool sm6_parser_validate_operand_type(struct sm6_parser *sm6, const struct sm6_type *type, char info_type) +static bool sm6_parser_validate_operand_type(struct sm6_parser *sm6, const struct sm6_value *value, char info_type, + bool is_return) { + const struct sm6_type *type = value->type; + + if (info_type != 'H' && !sm6_value_is_register(value)) + return false; + switch (info_type) { case 0: @@ -2231,8 +3381,15 @@ static bool sm6_parser_validate_operand_type(struct sm6_parser *sm6, const struc return false; case '8': return sm6_type_is_i8(type); + case 'b': + return sm6_value_is_constant(value) && sm6_type_is_bool(type); + case 'c': + return sm6_value_is_constant(value) && sm6_type_is_integer(type) && type->u.width >= 8 + && type->u.width <= 32; case 'i': return sm6_type_is_i32(type); + case 'H': + return (is_return || sm6_value_is_handle(value)) && type == sm6->handle_type; case 'v': return !type; case 'o': @@ -2252,7 +3409,7 @@ static bool sm6_parser_validate_dx_op(struct sm6_parser *sm6, enum dx_intrinsic_ info = &sm6_dx_op_table[op]; - if (!sm6_parser_validate_operand_type(sm6, dst->type, info->ret_type)) + if (!sm6_parser_validate_operand_type(sm6, dst, info->ret_type, true)) { WARN("Failed to validate return type for dx intrinsic id %u, '%s'.\n", op, name); /* Return type validation failure is not so critical. We only need to set @@ -2262,7 +3419,7 @@ static bool sm6_parser_validate_dx_op(struct sm6_parser *sm6, enum dx_intrinsic_ for (i = 0; i < operand_count; ++i) { const struct sm6_value *value = operands[i]; - if (!sm6_value_is_register(value) || !sm6_parser_validate_operand_type(sm6, value->type, info->operand_info[i])) + if (!sm6_parser_validate_operand_type(sm6, value, info->operand_info[i], false)) { WARN("Failed to validate operand %u for dx intrinsic id %u, '%s'.\n", i + 1, op, name); vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, @@ -2292,7 +3449,7 @@ static void sm6_parser_emit_unhandled(struct sm6_parser *sm6, struct vkd3d_shade return; type = sm6_type_get_scalar_type(dst->type, 0); - shader_register_init(&dst->u.reg, VKD3DSPR_UNDEF, vkd3d_data_type_from_sm6_type(type), 0); + vsir_register_init(&dst->u.reg, VKD3DSPR_UNDEF, vkd3d_data_type_from_sm6_type(type), 0); /* dst->is_undefined is not set here because it flags only explicitly undefined values. */ } @@ -2303,7 +3460,7 @@ static void sm6_parser_decode_dx_op(struct sm6_parser *sm6, struct sm6_block *co if (op >= ARRAY_SIZE(sm6_dx_op_table) || !sm6_dx_op_table[op].operand_info) { FIXME("Unhandled dx intrinsic function id %u, '%s'.\n", op, name); - vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_UNHANDLED_INTRINSIC, + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_UNHANDLED_INTRINSIC, "Call to intrinsic function %s is unhandled.", name); sm6_parser_emit_unhandled(sm6, ins, dst); return; @@ -2372,238 +3529,1860 @@ static void sm6_parser_emit_call(struct sm6_parser *sm6, const struct dxil_recor "Ignoring %u function call operands beyond the parameter list.", j); } - if (!fn_value->u.function.is_prototype) + if (!fn_value->u.function.is_prototype) + { + FIXME("Unhandled call to local function.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Call to a local function is unsupported."); + return; + } + if (!sm6_value_is_dx_intrinsic_dcl(fn_value)) + WARN("External function is not a dx intrinsic.\n"); + + if (!operand_count) + { + WARN("Missing dx intrinsic function id.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND_COUNT, + "The id for a dx intrinsic function is missing."); + return; + } + + op_value = operands[0]; + if (!sm6_value_is_constant(op_value) || !sm6_type_is_integer(op_value->type)) + { + WARN("dx intrinsic function id is not a constant int.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Expected a constant integer dx intrinsic function id."); + return; + } + sm6_parser_decode_dx_op(sm6, code_block, register_get_uint_value(&op_value->u.reg), + fn_value->u.function.name, &operands[1], operand_count - 1, ins, dst); +} + +static enum vkd3d_shader_opcode sm6_map_cast_op(uint64_t code, const struct sm6_type *from, + const struct sm6_type *to, struct sm6_parser *sm6) +{ + enum vkd3d_shader_opcode op = VKD3DSIH_INVALID; + bool from_int, to_int, from_fp, to_fp; + bool is_valid = false; + + from_int = sm6_type_is_integer(from); + to_int = sm6_type_is_integer(to); + from_fp = sm6_type_is_floating_point(from); + to_fp = sm6_type_is_floating_point(to); + + /* NOTE: DXIL currently doesn't use vectors here. */ + if ((!from_int && !from_fp) || (!to_int && !to_fp)) + { + FIXME("Unhandled cast of type class %u to type class %u.\n", from->class, to->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Cast of type class %u to type class %u is not implemented.", from->class, to->class); + return VKD3DSIH_INVALID; + } + if (to->u.width == 8 || from->u.width == 8) + { + FIXME("Unhandled 8-bit value.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Cast to/from an 8-bit type is not implemented."); + return VKD3DSIH_INVALID; + } + + /* DXC emits minimum precision types as 16-bit. These must be emitted + * as 32-bit in VSIR, so all width extensions to 32 bits are no-ops. */ + switch (code) + { + case CAST_TRUNC: + /* nop or min precision. TODO: native 16-bit */ + if (to->u.width == from->u.width || (to->u.width == 16 && from->u.width == 32)) + op = VKD3DSIH_NOP; + else + op = VKD3DSIH_UTOU; + is_valid = from_int && to_int && to->u.width <= from->u.width; + break; + case CAST_ZEXT: + case CAST_SEXT: + /* nop or min precision. TODO: native 16-bit */ + if (to->u.width == from->u.width || (to->u.width == 32 && from->u.width == 16)) + { + op = VKD3DSIH_NOP; + is_valid = from_int && to_int; + } + else if (to->u.width > from->u.width) + { + op = (code == CAST_ZEXT) ? VKD3DSIH_UTOU : VKD3DSIH_ITOI; + assert(from->u.width == 1 || to->u.width == 64); + is_valid = from_int && to_int; + } + break; + case CAST_FPTOUI: + op = VKD3DSIH_FTOU; + is_valid = from_fp && to_int && to->u.width > 1; + break; + case CAST_FPTOSI: + op = VKD3DSIH_FTOI; + is_valid = from_fp && to_int && to->u.width > 1; + break; + case CAST_UITOFP: + op = VKD3DSIH_UTOF; + is_valid = from_int && to_fp; + break; + case CAST_SITOFP: + op = VKD3DSIH_ITOF; + is_valid = from_int && to_fp; + break; + case CAST_FPTRUNC: + /* TODO: native 16-bit */ + op = (from->u.width == 64) ? VKD3DSIH_DTOF : VKD3DSIH_NOP; + is_valid = from_fp && to_fp; + break; + case CAST_FPEXT: + /* TODO: native 16-bit */ + op = (to->u.width == 64) ? VKD3DSIH_FTOD : VKD3DSIH_NOP; + is_valid = from_fp && to_fp; + break; + case CAST_BITCAST: + op = VKD3DSIH_MOV; + is_valid = to->u.width == from->u.width; + break; + default: + FIXME("Unhandled cast op %"PRIu64".\n", code); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Cast operation %"PRIu64" is unhandled.\n", code); + return VKD3DSIH_INVALID; + } + + if (!is_valid) + { + FIXME("Invalid types %u and/or %u for op %"PRIu64".\n", from->class, to->class, code); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Cast operation %"PRIu64" from type class %u, width %u to type class %u, width %u is invalid.\n", + code, from->class, from->u.width, to->class, to->u.width); + return VKD3DSIH_INVALID; + } + + return op; +} + +static void sm6_parser_emit_cast(struct sm6_parser *sm6, const struct dxil_record *record, + struct vkd3d_shader_instruction *ins, struct sm6_value *dst) +{ + struct vkd3d_shader_src_param *src_param; + enum vkd3d_shader_opcode handler_idx; + const struct sm6_value *value; + const struct sm6_type *type; + unsigned int i = 0; + + if (!(value = sm6_parser_get_value_by_ref(sm6, record, NULL, &i))) + return; + + if (!dxil_record_validate_operand_count(record, i + 2, i + 2, sm6)) + return; + + if (!(type = sm6_parser_get_type(sm6, record->operands[i++]))) + return; + + dst->type = type; + + if (sm6_type_is_pointer(type)) + { + *dst = *value; + dst->type = type; + ins->handler_idx = VKD3DSIH_NOP; + return; + } + + if ((handler_idx = sm6_map_cast_op(record->operands[i], value->type, type, sm6)) == VKD3DSIH_INVALID) + return; + + vsir_instruction_init(ins, &sm6->p.location, handler_idx); + + if (handler_idx == VKD3DSIH_NOP) + { + dst->u.reg = value->u.reg; + return; + } + + src_param = instruction_src_params_alloc(ins, 1, sm6); + src_param_init_from_value(src_param, value); + + instruction_dst_param_init_ssa_scalar(ins, sm6); + + /* bitcast */ + if (handler_idx == VKD3DSIH_MOV) + src_param->reg.data_type = dst->u.reg.data_type; +} + +struct sm6_cmp_info +{ + enum vkd3d_shader_opcode handler_idx; + bool src_swap; +}; + +static const struct sm6_cmp_info *sm6_map_cmp2_op(uint64_t code) +{ + static const struct sm6_cmp_info cmp_op_table[] = + { + [FCMP_FALSE] = {VKD3DSIH_INVALID}, + [FCMP_OEQ] = {VKD3DSIH_EQO}, + [FCMP_OGT] = {VKD3DSIH_LTO, true}, + [FCMP_OGE] = {VKD3DSIH_GEO}, + [FCMP_OLT] = {VKD3DSIH_LTO}, + [FCMP_OLE] = {VKD3DSIH_GEO, true}, + [FCMP_ONE] = {VKD3DSIH_NEO}, + [FCMP_ORD] = {VKD3DSIH_INVALID}, + [FCMP_UNO] = {VKD3DSIH_INVALID}, + [FCMP_UEQ] = {VKD3DSIH_EQU}, + [FCMP_UGT] = {VKD3DSIH_LTU, true}, + [FCMP_UGE] = {VKD3DSIH_GEU}, + [FCMP_ULT] = {VKD3DSIH_LTU}, + [FCMP_ULE] = {VKD3DSIH_GEU, true}, + [FCMP_UNE] = {VKD3DSIH_NEU}, + [FCMP_TRUE] = {VKD3DSIH_INVALID}, + + [ICMP_EQ] = {VKD3DSIH_IEQ}, + [ICMP_NE] = {VKD3DSIH_INE}, + [ICMP_UGT] = {VKD3DSIH_ULT, true}, + [ICMP_UGE] = {VKD3DSIH_UGE}, + [ICMP_ULT] = {VKD3DSIH_ULT}, + [ICMP_ULE] = {VKD3DSIH_UGE, true}, + [ICMP_SGT] = {VKD3DSIH_ILT, true}, + [ICMP_SGE] = {VKD3DSIH_IGE}, + [ICMP_SLT] = {VKD3DSIH_ILT}, + [ICMP_SLE] = {VKD3DSIH_IGE, true}, + }; + + return (code < ARRAY_SIZE(cmp_op_table)) ? &cmp_op_table[code] : NULL; +} + +static void sm6_parser_emit_cmp2(struct sm6_parser *sm6, const struct dxil_record *record, + struct vkd3d_shader_instruction *ins, struct sm6_value *dst) +{ + struct vkd3d_shader_src_param *src_params; + const struct sm6_type *type_a, *type_b; + bool is_int, is_fp, silence_warning; + const struct sm6_cmp_info *cmp; + const struct sm6_value *a, *b; + uint64_t code, flags; + unsigned int i = 0; + + if (!(dst->type = sm6->bool_type)) + { + WARN("Bool type not found.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_MODULE, + "Module does not define a boolean type for comparison results."); + return; + } + + a = sm6_parser_get_value_by_ref(sm6, record, NULL, &i); + b = sm6_parser_get_value_by_ref(sm6, record, a->type, &i); + if (!a || !b) + return; + + if (!dxil_record_validate_operand_count(record, i + 1, i + 2, sm6)) + return; + + type_a = a->type; + type_b = b->type; + is_int = sm6_type_is_bool_i16_i32_i64(type_a); + is_fp = sm6_type_is_floating_point(type_a); + + code = record->operands[i++]; + + if ((!is_int && !is_fp) || is_int != (code >= ICMP_EQ)) + { + FIXME("Invalid operation %"PRIu64" on type class %u.\n", code, type_a->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Comparison operation %"PRIu64" on type class %u is invalid.", code, type_a->class); + return; + } + + if (type_a != type_b) + { + WARN("Type mismatch, type %u width %u vs type %u width %u.\n", type_a->class, + type_a->u.width, type_b->class, type_b->u.width); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_TYPE_MISMATCH, + "Type mismatch in comparison operation arguments."); + } + + if (!(cmp = sm6_map_cmp2_op(code)) || !cmp->handler_idx || cmp->handler_idx == VKD3DSIH_INVALID) + { + FIXME("Unhandled operation %"PRIu64".\n", code); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Comparison operation %"PRIu64" is unhandled.", code); + return; + } + + vsir_instruction_init(ins, &sm6->p.location, cmp->handler_idx); + + flags = (record->operand_count > i) ? record->operands[i] : 0; + silence_warning = false; + + if (is_fp) + { + if (!(flags & FP_ALLOW_UNSAFE_ALGEBRA)) + ins->flags |= VKD3DSI_PRECISE_X; + flags &= ~FP_ALLOW_UNSAFE_ALGEBRA; + /* SPIR-V FPFastMathMode is only available in the Kernel execution model. */ + silence_warning = !(flags & ~(FP_NO_NAN | FP_NO_INF | FP_NO_SIGNED_ZEROS | FP_ALLOW_RECIPROCAL)); + } + if (flags && silence_warning) + { + TRACE("Ignoring fast FP modifier %#"PRIx64".\n", flags); + } + else if (flags) + { + WARN("Ignoring flags %#"PRIx64".\n", flags); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_IGNORING_OPERANDS, + "Ignoring flags %#"PRIx64" for a comparison operation.", flags); + } + + src_params = instruction_src_params_alloc(ins, 2, sm6); + src_param_init_from_value(&src_params[0 ^ cmp->src_swap], a); + src_param_init_from_value(&src_params[1 ^ cmp->src_swap], b); + + instruction_dst_param_init_ssa_scalar(ins, sm6); +} + +static void sm6_parser_emit_extractval(struct sm6_parser *sm6, const struct dxil_record *record, + struct vkd3d_shader_instruction *ins, struct sm6_value *dst) +{ + struct vkd3d_shader_src_param *src_param; + const struct sm6_type *type; + const struct sm6_value *src; + unsigned int i = 0; + uint64_t elem_idx; + + if (!(src = sm6_parser_get_value_by_ref(sm6, record, NULL, &i))) + return; + + if (!dxil_record_validate_operand_min_count(record, i + 1, sm6)) + return; + + if (record->operand_count > i + 1) + { + FIXME("Unhandled multiple indices.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Multiple extractval indices are not supported."); + return; + } + + type = src->type; + if (!sm6_type_is_aggregate(type)) + { + WARN("Invalid extraction from non-aggregate.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Source type of an extractval instruction is not an aggregate."); + return; + } + + elem_idx = record->operands[i]; + if (!(type = sm6_type_get_element_type_at_index(type, elem_idx))) + { + WARN("Invalid element index %"PRIu64".\n", elem_idx); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Element index %"PRIu64" for an extractval instruction is out of bounds.", elem_idx); + return; + } + if (!sm6_type_is_scalar(type)) + { + FIXME("Nested extraction is not supported.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Extraction from nested aggregates is not supported."); + return; + } + dst->type = type; + + vsir_instruction_init(ins, &sm6->p.location, VKD3DSIH_MOV); + + src_param = instruction_src_params_alloc(ins, 1, sm6); + src_param_init_from_value(src_param, src); + src_param->swizzle = vkd3d_shader_create_swizzle(elem_idx, elem_idx, elem_idx, elem_idx); + + instruction_dst_param_init_ssa_scalar(ins, sm6); +} + +static void sm6_parser_emit_gep(struct sm6_parser *sm6, const struct dxil_record *record, + struct vkd3d_shader_instruction *ins, struct sm6_value *dst) +{ + const struct sm6_type *type, *pointee_type; + unsigned int elem_idx, operand_idx = 2; + enum bitcode_address_space addr_space; + const struct sm6_value *elem_value; + struct vkd3d_shader_register *reg; + const struct sm6_value *src; + bool is_in_bounds; + + if (!dxil_record_validate_operand_min_count(record, 5, sm6) + || !(type = sm6_parser_get_type(sm6, record->operands[1])) + || !(src = sm6_parser_get_value_by_ref(sm6, record, NULL, &operand_idx)) + || !sm6_value_validate_is_register(src, sm6) + || !sm6_value_validate_is_pointer(src, sm6) + || !dxil_record_validate_operand_min_count(record, operand_idx + 2, sm6)) + { + return; + } + + if (src->u.reg.idx_count > 1) + { + WARN("Unsupported stacked GEP.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "A GEP instruction on the result of a previous GEP is unsupported."); + return; + } + + is_in_bounds = record->operands[0]; + + if ((pointee_type = src->type->u.pointer.type) != type) + { + WARN("Type mismatch, type %u width %u vs type %u width %u.\n", type->class, + type->u.width, pointee_type->class, pointee_type->u.width); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_TYPE_MISMATCH, + "Type mismatch in GEP operation arguments."); + } + addr_space = src->type->u.pointer.addr_space; + + if (!(elem_value = sm6_parser_get_value_by_ref(sm6, record, NULL, &operand_idx))) + return; + + /* The first index is always zero, to form a simple pointer dereference. */ + if (sm6_value_get_constant_uint(elem_value)) + { + WARN("Expected constant zero.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "The pointer dereference index for a GEP instruction is not constant zero."); + return; + } + + if (!sm6_type_is_array(pointee_type)) + { + WARN("Invalid GEP on type class %u.\n", pointee_type->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Source type for index 1 of a GEP instruction is not an array."); + return; + } + + if (!(elem_value = sm6_parser_get_value_by_ref(sm6, record, NULL, &operand_idx))) + return; + + /* If indexing is dynamic, just get the type at offset zero. */ + elem_idx = sm6_value_is_constant(elem_value) ? sm6_value_get_constant_uint(elem_value) : 0; + type = sm6_type_get_element_type_at_index(pointee_type, elem_idx); + if (!type) + { + WARN("Invalid element index %u.\n", elem_idx); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Element index %u for a GEP instruction is out of bounds.", elem_idx); + return; + } + + if (operand_idx < record->operand_count) + { + FIXME("Multiple element indices are not implemented.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Multi-dimensional addressing in GEP instructions is not supported."); + return; + } + + if (!(dst->type = sm6_type_get_pointer_to_type(type, addr_space, sm6))) + { + WARN("Failed to get pointer type for type %u.\n", type->class); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_MODULE, + "Module does not define a pointer type for a GEP instruction."); + return; + } + + reg = &dst->u.reg; + *reg = src->u.reg; + reg->idx[1].offset = 0; + register_index_address_init(®->idx[1], elem_value, sm6); + reg->idx[1].is_in_bounds = is_in_bounds; + reg->idx_count = 2; + + ins->handler_idx = VKD3DSIH_NOP; +} + +static void sm6_parser_emit_load(struct sm6_parser *sm6, const struct dxil_record *record, + struct vkd3d_shader_instruction *ins, struct sm6_value *dst) +{ + const struct sm6_type *elem_type = NULL, *pointee_type; + struct vkd3d_shader_src_param *src_param; + unsigned int alignment, i = 0; + const struct sm6_value *ptr; + uint64_t alignment_code; + + if (!(ptr = sm6_parser_get_value_by_ref(sm6, record, NULL, &i))) + return; + if (!sm6_value_validate_is_register(ptr, sm6) + || !sm6_value_validate_is_pointer(ptr, sm6) + || !dxil_record_validate_operand_count(record, i + 2, i + 3, sm6)) + return; + + if (record->operand_count > i + 2 && !(elem_type = sm6_parser_get_type(sm6, record->operands[i++]))) + return; + + if (!elem_type) + { + elem_type = ptr->type->u.pointer.type; + } + else if (elem_type != (pointee_type = ptr->type->u.pointer.type)) + { + WARN("Type mismatch.\n"); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_TYPE_MISMATCH, + "Type mismatch in pointer load arguments."); + } + + dst->type = elem_type; + + if (!sm6_value_validate_is_numeric(dst, sm6)) + return; + + alignment_code = record->operands[i++]; + if (!bitcode_parse_alignment(alignment_code, &alignment)) + WARN("Invalid alignment %"PRIu64".\n", alignment_code); + + if (record->operands[i]) + WARN("Ignoring volatile modifier.\n"); + + vsir_instruction_init(ins, &sm6->p.location, VKD3DSIH_MOV); + + src_param = instruction_src_params_alloc(ins, 1, sm6); + src_param_init_from_value(&src_param[0], ptr); + src_param->reg.alignment = alignment; + + instruction_dst_param_init_ssa_scalar(ins, sm6); +} + +static void sm6_parser_emit_ret(struct sm6_parser *sm6, const struct dxil_record *record, + struct sm6_block *code_block, struct vkd3d_shader_instruction *ins) +{ + if (!dxil_record_validate_operand_count(record, 0, 1, sm6)) + return; + + if (record->operand_count) + FIXME("Non-void return is not implemented.\n"); + + ins->handler_idx = VKD3DSIH_NOP; +} + +static void sm6_parser_emit_vselect(struct sm6_parser *sm6, const struct dxil_record *record, + struct vkd3d_shader_instruction *ins, struct sm6_value *dst) +{ + struct vkd3d_shader_src_param *src_params; + const struct sm6_value *src[3]; + unsigned int i = 0; + + if (!(src[1] = sm6_parser_get_value_by_ref(sm6, record, NULL, &i)) + || !(src[2] = sm6_parser_get_value_by_ref(sm6, record, src[1]->type, &i)) + || !(src[0] = sm6_parser_get_value_by_ref(sm6, record, NULL, &i))) + { + return; + } + dxil_record_validate_operand_max_count(record, i, sm6); + + for (i = 0; i < 3; ++i) + { + if (!sm6_value_validate_is_register(src[i], sm6)) + return; + } + + dst->type = src[1]->type; + + if (!sm6_value_validate_is_bool(src[0], sm6)) + return; + + vsir_instruction_init(ins, &sm6->p.location, VKD3DSIH_MOVC); + + src_params = instruction_src_params_alloc(ins, 3, sm6); + for (i = 0; i < 3; ++i) + src_param_init_from_value(&src_params[i], src[i]); + + instruction_dst_param_init_ssa_scalar(ins, sm6); +} + +static bool sm6_metadata_value_is_node(const struct sm6_metadata_value *m) +{ + return m && m->type == VKD3D_METADATA_NODE; +} + +static bool sm6_metadata_value_is_value(const struct sm6_metadata_value *m) +{ + return m && m->type == VKD3D_METADATA_VALUE; +} + +static bool sm6_metadata_value_is_string(const struct sm6_metadata_value *m) +{ + return m && m->type == VKD3D_METADATA_STRING; +} + +static bool sm6_metadata_get_uint_value(const struct sm6_parser *sm6, + const struct sm6_metadata_value *m, unsigned int *u) +{ + const struct sm6_value *value; + + if (!m || m->type != VKD3D_METADATA_VALUE) + return false; + + value = m->u.value; + if (!sm6_value_is_constant(value)) + return false; + if (!sm6_type_is_integer(value->type)) + return false; + + *u = register_get_uint_value(&value->u.reg); + + return true; +} + +static bool sm6_metadata_get_uint64_value(const struct sm6_parser *sm6, + const struct sm6_metadata_value *m, uint64_t *u) +{ + const struct sm6_value *value; + + if (!m || m->type != VKD3D_METADATA_VALUE) + return false; + + value = m->u.value; + if (!sm6_value_is_constant(value)) + return false; + if (!sm6_type_is_integer(value->type)) + return false; + + *u = register_get_uint64_value(&value->u.reg); + + return true; +} + +static enum vkd3d_result sm6_parser_function_init(struct sm6_parser *sm6, const struct dxil_block *block, + struct sm6_function *function) +{ + struct vkd3d_shader_instruction *ins; + size_t i, block_idx, block_count; + const struct dxil_record *record; + bool ret_found, is_terminator; + struct sm6_block *code_block; + struct sm6_value *dst; + + if (sm6->function_count) + { + FIXME("Multiple functions are not supported yet.\n"); + return VKD3D_ERROR_INVALID_SHADER; + } + if (!(function->declaration = sm6_parser_next_function_definition(sm6))) + { + WARN("Failed to find definition to match function body.\n"); + return VKD3D_ERROR_INVALID_SHADER; + } + + if (block->record_count < 2) + { + /* It should contain at least a block count and a RET instruction. */ + WARN("Invalid function block record count %zu.\n", block->record_count); + return VKD3D_ERROR_INVALID_SHADER; + } + if (block->records[0]->code != FUNC_CODE_DECLAREBLOCKS || !block->records[0]->operand_count + || block->records[0]->operands[0] > UINT_MAX) + { + WARN("Block count declaration not found or invalid.\n"); + return VKD3D_ERROR_INVALID_SHADER; + } + + if (!(block_count = block->records[0]->operands[0])) + { + WARN("Function contains no blocks.\n"); + return VKD3D_ERROR_INVALID_SHADER; + } + if (block_count > 1) + { + FIXME("Branched shaders are not supported yet.\n"); + return VKD3D_ERROR_INVALID_SHADER; + } + + if (!(function->blocks[0] = sm6_block_create())) + { + ERR("Failed to allocate code block.\n"); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + function->block_count = block_count; + code_block = function->blocks[0]; + + sm6->cur_max_value = function->value_count; + + for (i = 1, block_idx = 0, ret_found = false; i < block->record_count; ++i) + { + sm6->p.location.column = i; + + if (!code_block) + { + WARN("Invalid block count %zu.\n", function->block_count); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, + "Invalid block count %zu.", function->block_count); + return VKD3D_ERROR_INVALID_SHADER; + } + + /* block->record_count - 1 is the instruction count, but some instructions + * can emit >1 IR instruction, so extra may be used. */ + if (!vkd3d_array_reserve((void **)&code_block->instructions, &code_block->instruction_capacity, + max(code_block->instruction_count + 1, block->record_count), sizeof(*code_block->instructions))) + { + ERR("Failed to allocate instructions.\n"); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + + ins = &code_block->instructions[code_block->instruction_count]; + ins->handler_idx = VKD3DSIH_INVALID; + + dst = sm6_parser_get_current_value(sm6); + dst->type = NULL; + dst->value_type = VALUE_TYPE_REG; + is_terminator = false; + + record = block->records[i]; + switch (record->code) + { + case FUNC_CODE_INST_BINOP: + sm6_parser_emit_binop(sm6, record, ins, dst); + break; + case FUNC_CODE_INST_CALL: + sm6_parser_emit_call(sm6, record, code_block, ins, dst); + break; + case FUNC_CODE_INST_CAST: + sm6_parser_emit_cast(sm6, record, ins, dst); + break; + case FUNC_CODE_INST_CMP2: + sm6_parser_emit_cmp2(sm6, record, ins, dst); + break; + case FUNC_CODE_INST_EXTRACTVAL: + sm6_parser_emit_extractval(sm6, record, ins, dst); + break; + case FUNC_CODE_INST_GEP: + sm6_parser_emit_gep(sm6, record, ins, dst); + break; + case FUNC_CODE_INST_LOAD: + sm6_parser_emit_load(sm6, record, ins, dst); + break; + case FUNC_CODE_INST_RET: + sm6_parser_emit_ret(sm6, record, code_block, ins); + is_terminator = true; + ret_found = true; + break; + case FUNC_CODE_INST_VSELECT: + sm6_parser_emit_vselect(sm6, record, ins, dst); + break; + default: + FIXME("Unhandled dxil instruction %u.\n", record->code); + return VKD3D_ERROR_INVALID_SHADER; + } + + if (sm6->p.failed) + return VKD3D_ERROR; + assert(ins->handler_idx != VKD3DSIH_INVALID); + + if (is_terminator) + { + ++block_idx; + code_block = (block_idx < function->block_count) ? function->blocks[block_idx] : NULL; + } + if (code_block) + code_block->instruction_count += ins->handler_idx != VKD3DSIH_NOP; + else + assert(ins->handler_idx == VKD3DSIH_NOP); + + sm6->value_count += !!dst->type; + } + + if (!ret_found) + { + WARN("Function contains no RET instruction.\n"); + return VKD3D_ERROR_INVALID_SHADER; + } + + return VKD3D_OK; +} + +static bool sm6_block_emit_instructions(struct sm6_block *block, struct sm6_parser *sm6) +{ + struct vkd3d_shader_instruction *ins = sm6_parser_require_space(sm6, block->instruction_count + 1); + + if (!ins) + return false; + + memcpy(ins, block->instructions, block->instruction_count * sizeof(*block->instructions)); + sm6->p.instructions.count += block->instruction_count; + + sm6_parser_add_instruction(sm6, VKD3DSIH_RET); + + return true; +} + +static enum vkd3d_result sm6_parser_module_init(struct sm6_parser *sm6, const struct dxil_block *block, + unsigned int level) +{ + size_t i, old_value_count = sm6->value_count; + struct sm6_function *function; + enum vkd3d_result ret; + + for (i = 0; i < block->child_block_count; ++i) + { + if ((ret = sm6_parser_module_init(sm6, block->child_blocks[i], level + 1)) < 0) + return ret; + } + + sm6->p.location.line = block->id; + sm6->p.location.column = 0; + + switch (block->id) + { + case CONSTANTS_BLOCK: + /* Level 1 (global) constants are already done in sm6_parser_globals_init(). */ + if (level < 2) + break; + function = &sm6->functions[sm6->function_count]; + sm6->cur_max_value = function->value_count; + return sm6_parser_constants_init(sm6, block); + + case FUNCTION_BLOCK: + function = &sm6->functions[sm6->function_count]; + if ((ret = sm6_parser_function_init(sm6, block, function)) < 0) + return ret; + /* The value index returns to its previous value after handling a function. It's usually nonzero + * at the start because of global constants/variables/function declarations. Function constants + * occur in a child block, so value_count is already saved before they are emitted. */ + memset(&sm6->values[old_value_count], 0, (sm6->value_count - old_value_count) * sizeof(*sm6->values)); + sm6->value_count = old_value_count; + break; + + case BLOCKINFO_BLOCK: + case MODULE_BLOCK: + case PARAMATTR_BLOCK: + case PARAMATTR_GROUP_BLOCK: + case VALUE_SYMTAB_BLOCK: + case METADATA_BLOCK: + case METADATA_ATTACHMENT_BLOCK: + case TYPE_BLOCK: + break; + + default: + FIXME("Unhandled block id %u.\n", block->id); + break; + } + + return VKD3D_OK; +} + +static bool sm6_parser_allocate_named_metadata(struct sm6_parser *sm6) +{ + struct dxil_block *block; + unsigned int i, j, count; + + for (i = 0, count = 0; i < sm6->root_block.child_block_count; ++i) + { + block = sm6->root_block.child_blocks[i]; + if (block->id != METADATA_BLOCK) + continue; + for (j = 0; j < block->record_count; ++j) + count += block->records[j]->code == METADATA_NAMED_NODE; + } + + if (!count) + return true; + + return !!(sm6->named_metadata = vkd3d_calloc(count, sizeof(*sm6->named_metadata))); +} + +static enum vkd3d_result metadata_value_create_node(struct sm6_metadata_value *m, struct sm6_metadata_table *table, + unsigned int dst_idx, unsigned int end_count, const struct dxil_record *record, struct sm6_parser *sm6) +{ + struct sm6_metadata_node *node; + unsigned int i, offset; + + m->type = VKD3D_METADATA_NODE; + if (!(m->value_type = sm6->metadata_type)) + { + WARN("Metadata type not found.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_METADATA, + "The type for metadata values was not found."); + return VKD3D_ERROR_INVALID_SHADER; + } + if (!(node = vkd3d_malloc(offsetof(struct sm6_metadata_node, operands[record->operand_count])))) + { + ERR("Failed to allocate metadata node with %u operands.\n", record->operand_count); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, + "Out of memory allocating a metadata node with %u operands.", record->operand_count); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + m->u.node = node; + + node->is_distinct = record->code == METADATA_DISTINCT_NODE; + + offset = record->code != METADATA_NAMED_NODE; + + for (i = 0; i < record->operand_count; ++i) + { + uint64_t ref; + + ref = record->operands[i] - offset; + if (record->operands[i] >= offset && ref >= end_count) + { + WARN("Invalid metadata index %"PRIu64".\n", ref); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_METADATA, + "Metadata index %"PRIu64" is invalid.", ref); + vkd3d_free(node); + return VKD3D_ERROR_INVALID_SHADER; + } + + if (!node->is_distinct && ref == dst_idx) + { + WARN("Metadata self-reference at index %u.\n", dst_idx); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_METADATA, + "Metadata index %u is self-referencing.", dst_idx); + vkd3d_free(node); + return VKD3D_ERROR_INVALID_SHADER; + } + + node->operands[i] = (record->operands[i] >= offset) ? &table->values[ref] : NULL; + if (record->code == METADATA_NAMED_NODE && !sm6_metadata_value_is_node(node->operands[i])) + { + WARN("Named node operand is not a node.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_METADATA, + "The operand of a metadata named node is not a node."); + vkd3d_free(node); + return VKD3D_ERROR_INVALID_SHADER; + } + } + + node->operand_count = record->operand_count; + + return VKD3D_OK; +} + +static enum vkd3d_result sm6_parser_metadata_init(struct sm6_parser *sm6, const struct dxil_block *block, + struct sm6_metadata_table *table) +{ + unsigned int i, count, table_idx, value_idx; + struct sm6_metadata_value *values, *m; + const struct dxil_record *record; + const struct sm6_value *value; + enum vkd3d_result ret; + char *name; + + for (i = 0, count = 0; i < block->record_count; ++i) + count += block->records[i]->code != METADATA_NAME; + + if (!(values = vkd3d_calloc(count, sizeof(*values)))) + { + ERR("Failed to allocate metadata tables.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, + "Out of memory allocating metadata tables."); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + table->values = values; + + for (i = 0, name = NULL; i < block->record_count; ++i) + { + record = block->records[i]; + + table_idx = table->count; + m = &values[table_idx]; + + switch (record->code) + { + case METADATA_NAMED_NODE: + if (!name) + { + WARN("Named node has no name.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_METADATA, + "A metadata named node has no name."); + return VKD3D_ERROR_INVALID_SHADER; + } + + /* When DXC emits metadata value array reference indices it assumes named nodes + * are not included in the array. Store named nodes separately. */ + m = &sm6->named_metadata[sm6->named_metadata_count].value; + sm6->named_metadata[sm6->named_metadata_count].name = name; + name = NULL; + + if ((ret = metadata_value_create_node(m, table, UINT_MAX, count, record, sm6)) < 0) + return ret; + ++sm6->named_metadata_count; + /* Skip incrementing the table count. */ + continue; + + case METADATA_DISTINCT_NODE: + case METADATA_NODE: + if ((ret = metadata_value_create_node(m, table, table_idx, count, record, sm6)) < 0) + return ret; + break; + + case METADATA_KIND: + if (!dxil_record_validate_operand_min_count(record, 2, sm6)) + return VKD3D_ERROR_INVALID_SHADER; + + m->type = VKD3D_METADATA_KIND; + m->u.kind.id = record->operands[0]; + if (!(m->u.kind.name = dxil_record_to_string(record, 1, sm6))) + { + ERR("Failed to allocate name of a kind.\n"); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + break; + + case METADATA_NAME: + /* Check the next record to avoid freeing 'name' in all exit paths. */ + if (i + 1 == block->record_count || block->records[i + 1]->code != METADATA_NAMED_NODE) + { + WARN("Name is not followed by a named node.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_METADATA, + "A metadata node name is not followed by a named node."); + return VKD3D_ERROR_INVALID_SHADER; + } + /* LLVM allows an empty string here. */ + if (!(name = dxil_record_to_string(record, 0, sm6))) + { + ERR("Failed to allocate name.\n"); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + continue; + + case METADATA_STRING: + /* LLVM allows an empty string here. */ + m->type = VKD3D_METADATA_STRING; + if (!(m->u.string_value = dxil_record_to_string(record, 0, sm6))) + { + ERR("Failed to allocate string.\n"); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + break; + + case METADATA_VALUE: + if (!dxil_record_validate_operand_count(record, 2, 2, sm6)) + return VKD3D_ERROR_INVALID_SHADER; + + m->type = VKD3D_METADATA_VALUE; + if (!(m->value_type = sm6_parser_get_type(sm6, record->operands[0]))) + return VKD3D_ERROR_INVALID_SHADER; + + if (record->operands[1] > UINT_MAX) + WARN("Truncating value index %"PRIu64".\n", record->operands[1]); + value_idx = record->operands[1]; + if (!(value = sm6_parser_get_value_safe(sm6, value_idx))) + return VKD3D_ERROR_INVALID_SHADER; + + if (!sm6_value_is_constant(value) && !sm6_value_is_undef(value) && !sm6_value_is_icb(value) + && !sm6_value_is_function_dcl(value)) + { + WARN("Value at index %u is not a constant or a function declaration.\n", value_idx); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_METADATA, + "Metadata value at index %u is not a constant or a function declaration.", value_idx); + return VKD3D_ERROR_INVALID_SHADER; + } + m->u.value = value; + + if (value->type != m->value_type) + { + WARN("Type mismatch.\n"); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_TYPE_MISMATCH, + "The type of a metadata value does not match its referenced value at index %u.", value_idx); + } + + break; + + default: + FIXME("Unhandled metadata type %u.\n", record->code); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_METADATA, + "Metadata type %u is unhandled.", record->code); + return VKD3D_ERROR_INVALID_SHADER; + } + ++table->count; + } + + return VKD3D_OK; +} + +static enum vkd3d_shader_component_type vkd3d_component_type_from_dxil_component_type(enum dxil_component_type type) +{ + switch (type) + { + case COMPONENT_TYPE_I1: + return VKD3D_SHADER_COMPONENT_BOOL; + case COMPONENT_TYPE_I16: + case COMPONENT_TYPE_I32: + return VKD3D_SHADER_COMPONENT_INT; + case COMPONENT_TYPE_U16: + case COMPONENT_TYPE_U32: + return VKD3D_SHADER_COMPONENT_UINT; + case COMPONENT_TYPE_F16: + case COMPONENT_TYPE_F32: + case COMPONENT_TYPE_SNORMF32: + case COMPONENT_TYPE_UNORMF32: + return VKD3D_SHADER_COMPONENT_FLOAT; + case COMPONENT_TYPE_F64: + case COMPONENT_TYPE_SNORMF64: + case COMPONENT_TYPE_UNORMF64: + return VKD3D_SHADER_COMPONENT_DOUBLE; + default: + FIXME("Unhandled component type %u.\n", type); + return VKD3D_SHADER_COMPONENT_UINT; + } +} + +static enum vkd3d_shader_minimum_precision minimum_precision_from_dxil_component_type(enum dxil_component_type type) +{ + switch (type) + { + case COMPONENT_TYPE_F16: + return VKD3D_SHADER_MINIMUM_PRECISION_FLOAT_16; + case COMPONENT_TYPE_I16: + return VKD3D_SHADER_MINIMUM_PRECISION_INT_16; + case COMPONENT_TYPE_U16: + return VKD3D_SHADER_MINIMUM_PRECISION_UINT_16; + default: + return VKD3D_SHADER_MINIMUM_PRECISION_NONE; + } +} + +static const enum vkd3d_shader_sysval_semantic sysval_semantic_table[] = +{ + [SEMANTIC_KIND_ARBITRARY] = VKD3D_SHADER_SV_NONE, + [SEMANTIC_KIND_POSITION] = VKD3D_SHADER_SV_POSITION, + [SEMANTIC_KIND_TARGET] = VKD3D_SHADER_SV_NONE, +}; + +static enum vkd3d_shader_sysval_semantic sysval_semantic_from_dxil_semantic_kind(enum dxil_semantic_kind kind) +{ + if (kind < ARRAY_SIZE(sysval_semantic_table)) + { + return sysval_semantic_table[kind]; + } + else + { + return VKD3D_SHADER_SV_NONE; + } +} + +static const struct sm6_metadata_value *sm6_parser_find_named_metadata(struct sm6_parser *sm6, const char *name) +{ + const struct sm6_metadata_node *node; + unsigned int i; + + for (i = 0; i < sm6->named_metadata_count; ++i) + { + if (strcmp(sm6->named_metadata[i].name, name)) + continue; + + node = sm6->named_metadata[i].value.u.node; + if (!node->operand_count) + return NULL; + if (node->operand_count > 1) + { + FIXME("Ignoring %u extra operands for %s.\n", node->operand_count - 1, name); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_IGNORING_OPERANDS, + "Ignoring %u extra operands for metadata node %s.", node->operand_count - 1, name); + } + return node->operands[0]; + } + + return NULL; +} + +static bool sm6_parser_resources_load_register_range(struct sm6_parser *sm6, + const struct sm6_metadata_node *node, struct vkd3d_shader_register_range *range) +{ + unsigned int size; + + if (!sm6_metadata_value_is_value(node->operands[1])) + { + WARN("Resource data type is not a value.\n"); + return false; + } + if (!sm6_type_is_pointer(node->operands[1]->value_type)) + { + WARN("Resource type is not a pointer.\n"); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_TYPE_MISMATCH, + "Resource metadata value type is not a pointer."); + } + + if (!sm6_metadata_get_uint_value(sm6, node->operands[3], &range->space)) + { + WARN("Failed to load register space.\n"); + return false; + } + if (!sm6_metadata_get_uint_value(sm6, node->operands[4], &range->first)) + { + WARN("Failed to load register first.\n"); + return false; + } + if (!sm6_metadata_get_uint_value(sm6, node->operands[5], &size)) + { + WARN("Failed to load register range size.\n"); + return false; + } + if (!size || (size != UINT_MAX && !vkd3d_bound_range(range->first, size, UINT_MAX))) + { + WARN("Invalid register range, first %u, size %u.\n", range->first, size); + return false; + } + range->last = (size == UINT_MAX) ? UINT_MAX : range->first + size - 1; + + return true; +} + +static enum vkd3d_result sm6_parser_resources_load_cbv(struct sm6_parser *sm6, + const struct sm6_metadata_node *node, struct sm6_descriptor_info *d, struct vkd3d_shader_instruction *ins) +{ + struct vkd3d_shader_register *reg; + unsigned int buffer_size; + + if (node->operand_count < 7) + { + WARN("Invalid operand count %u.\n", node->operand_count); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND_COUNT, + "Invalid operand count %u for a CBV descriptor.", node->operand_count); + return VKD3D_ERROR_INVALID_SHADER; + } + if (node->operand_count > 7 && node->operands[7]) + { + WARN("Ignoring %u extra operands.\n", node->operand_count - 7); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_IGNORING_OPERANDS, + "Ignoring %u extra operands for a CBV descriptor.", node->operand_count - 7); + } + + if (!sm6_metadata_get_uint_value(sm6, node->operands[6], &buffer_size)) + { + WARN("Failed to load buffer size.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_RESOURCES, + "Constant buffer size metadata value is not an integer."); + return VKD3D_ERROR_INVALID_SHADER; + } + + vsir_instruction_init(ins, &sm6->p.location, VKD3DSIH_DCL_CONSTANT_BUFFER); + ins->resource_type = VKD3D_SHADER_RESOURCE_BUFFER; + ins->declaration.cb.size = buffer_size; + ins->declaration.cb.src.swizzle = VKD3D_SHADER_NO_SWIZZLE; + ins->declaration.cb.src.modifiers = VKD3DSPSM_NONE; + + reg = &ins->declaration.cb.src.reg; + vsir_register_init(reg, VKD3DSPR_CONSTBUFFER, VKD3D_DATA_FLOAT, 3); + reg->idx[0].offset = d->id; + reg->idx[1].offset = d->range.first; + reg->idx[2].offset = d->range.last; + + ins->declaration.cb.range = d->range; + + return VKD3D_OK; +} + +static enum vkd3d_result sm6_parser_descriptor_type_init(struct sm6_parser *sm6, + enum vkd3d_shader_descriptor_type type, const struct sm6_metadata_node *descriptor_node) +{ + struct vkd3d_shader_instruction *ins; + const struct sm6_metadata_node *node; + const struct sm6_metadata_value *m; + struct sm6_descriptor_info *d; + enum vkd3d_result ret; + unsigned int i; + + for (i = 0; i < descriptor_node->operand_count; ++i) + { + m = descriptor_node->operands[i]; + if (!sm6_metadata_value_is_node(m)) + { + WARN("Resource descriptor is not a node.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_RESOURCES, + "Resource descriptor is not a metadata node."); + return VKD3D_ERROR_INVALID_SHADER; + } + + node = m->u.node; + if (node->operand_count < 6) + { + WARN("Invalid operand count %u.\n", node->operand_count); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND_COUNT, + "Invalid operand count %u for a descriptor.", node->operand_count); + return VKD3D_ERROR_INVALID_SHADER; + } + + if (!vkd3d_array_reserve((void **)&sm6->descriptors, &sm6->descriptor_capacity, + sm6->descriptor_count + 1, sizeof(*sm6->descriptors))) + { + ERR("Failed to allocate descriptor array.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, + "Out of memory allocating the descriptor array."); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + d = &sm6->descriptors[sm6->descriptor_count]; + d->type = type; + + if (!sm6_metadata_get_uint_value(sm6, node->operands[0], &d->id)) + { + WARN("Failed to load resource id.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_RESOURCES, + "Resource id metadata value is not an integer."); + return VKD3D_ERROR_INVALID_SHADER; + } + + if (!sm6_parser_resources_load_register_range(sm6, node, &d->range)) + { + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_RESOURCES, + "Resource register range is invalid."); + return VKD3D_ERROR_INVALID_SHADER; + } + + if (!(ins = sm6_parser_require_space(sm6, 1))) + { + ERR("Failed to allocate instruction.\n"); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + + switch (type) + { + case VKD3D_SHADER_DESCRIPTOR_TYPE_CBV: + if ((ret = sm6_parser_resources_load_cbv(sm6, node, d, ins)) < 0) + return ret; + break; + default: + FIXME("Unsupported descriptor type %u.\n", type); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_RESOURCES, + "Resource descriptor type %u is unsupported.", type); + return VKD3D_ERROR_INVALID_SHADER; + } + + ++sm6->descriptor_count; + ++sm6->p.instructions.count; + } + + return VKD3D_OK; +} + +static enum vkd3d_result sm6_parser_resources_init(struct sm6_parser *sm6) +{ + const struct sm6_metadata_value *m = sm6_parser_find_named_metadata(sm6, "dx.resources"); + enum vkd3d_shader_descriptor_type type; + const struct sm6_metadata_node *node; + enum vkd3d_result ret; + + if (!m) + return VKD3D_OK; + + node = m->u.node; + if (node->operand_count != SHADER_DESCRIPTOR_TYPE_COUNT) + { + WARN("Unexpected descriptor type count %u.\n", node->operand_count); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_RESOURCES, + "Descriptor type count %u is invalid.", node->operand_count); + return VKD3D_ERROR_INVALID_SHADER; + } + + for (type = 0; type < SHADER_DESCRIPTOR_TYPE_COUNT; ++type) + { + if (!(m = node->operands[type])) + continue; + + if (!sm6_metadata_value_is_node(m)) + { + WARN("Resource list is not a node.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_RESOURCES, + "Resource list is not a metadata node."); + return VKD3D_ERROR_INVALID_SHADER; + } + + if ((ret = sm6_parser_descriptor_type_init(sm6, type, m->u.node)) < 0) + return ret; + } + + return VKD3D_OK; +} + +static void signature_element_read_additional_element_values(struct signature_element *e, + const struct sm6_metadata_node *node, struct sm6_parser *sm6) +{ + unsigned int i, operand_count, value; + enum dxil_element_additional_tag tag; + + if (node->operand_count < 11 || !node->operands[10]) + return; + + if (!sm6_metadata_value_is_node(node->operands[10])) + { + WARN("Additional values list is not a node.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Signature element additional values list is not a metadata node."); + return; + } + + node = node->operands[10]->u.node; + if (node->operand_count & 1) + { + WARN("Operand count is not even.\n"); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_IGNORING_OPERANDS, + "Operand count for signature element additional tag/value pairs is not even."); + } + operand_count = node->operand_count & ~1u; + + for (i = 0; i < operand_count; i += 2) + { + if (!sm6_metadata_get_uint_value(sm6, node->operands[i], &tag) + || !sm6_metadata_get_uint_value(sm6, node->operands[i + 1], &value)) + { + WARN("Failed to extract tag/value pair.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Signature element tag/value pair at index %u is not an integer pair.", i); + continue; + } + + switch (tag) + { + case ADDITIONAL_TAG_STREAM_INDEX: + e->stream_index = value; + break; + case ADDITIONAL_TAG_RELADDR_MASK: + /* A mask of components accessed via relative addressing. Seems to replace TPF 'dcl_index_range'. */ + if (value > VKD3DSP_WRITEMASK_ALL) + { + WARN("Invalid relative addressed mask %#x.\n", value); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_INVALID_MASK, + "Mask %#x of relative-addressed components is invalid.", value); + } + break; + case ADDITIONAL_TAG_USED_MASK: + if (value > VKD3DSP_WRITEMASK_ALL) + { + WARN("Invalid used mask %#x.\n", value); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_INVALID_MASK, + "Mask %#x of used components is invalid.", value); + value &= VKD3DSP_WRITEMASK_ALL; + } + e->used_mask = value; + break; + default: + FIXME("Unhandled tag %u, value %u.\n", tag, value); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_IGNORING_OPERANDS, + "Tag %#x for signature element additional value %#x is unhandled.", tag, value); + break; + } + } +} + +static enum vkd3d_result sm6_parser_read_signature(struct sm6_parser *sm6, const struct sm6_metadata_value *m, + struct shader_signature *s) +{ + unsigned int i, j, column_count, operand_count, index; + const struct sm6_metadata_node *node, *element_node; + struct signature_element *elements, *e; + unsigned int values[10]; + + if (!m) + return VKD3D_OK; + + if (!sm6_metadata_value_is_node(m)) + { + WARN("Signature element list is not a node.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Signature element list is not a metadata node."); + return VKD3D_ERROR_INVALID_SHADER; + } + + node = m->u.node; + operand_count = node->operand_count; + + if (!(elements = vkd3d_calloc(operand_count, sizeof(*elements)))) + { + ERR("Failed to allocate %u signature elements.\n", operand_count); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, + "Out of memory allocating %u signature elements.", operand_count); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + + for (i = 0; i < operand_count; ++i) + { + m = node->operands[i]; + + if (!sm6_metadata_value_is_node(m)) + { + WARN("Signature element is not a node.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Signature element is not a metadata node."); + return VKD3D_ERROR_INVALID_SHADER; + } + + element_node = m->u.node; + if (element_node->operand_count < 10) + { + WARN("Invalid operand count %u.\n", element_node->operand_count); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Invalid signature element operand count %u.", element_node->operand_count); + return VKD3D_ERROR_INVALID_SHADER; + } + if (element_node->operand_count > 11) + { + WARN("Ignoring %u extra operands.\n", element_node->operand_count - 11); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_IGNORING_OPERANDS, + "Ignoring %u extra operands for a signature element.", element_node->operand_count - 11); + } + + for (j = 0; j < 10; ++j) + { + /* 1 is the semantic name, 4 is semantic index metadata. */ + if (j == 1 || j == 4) + continue; + if (!sm6_metadata_get_uint_value(sm6, element_node->operands[j], &values[j])) + { + WARN("Failed to load uint value at index %u.\n", j); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Signature element value at index %u is not an integer.", j); + return VKD3D_ERROR_INVALID_SHADER; + } + } + + e = &elements[i]; + + if (values[0] != i) + { + FIXME("Unsupported element id %u not equal to its index %u.\n", values[0], i); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "A non-sequential and non-zero-based element id is not supported."); + return VKD3D_ERROR_INVALID_SHADER; + } + + if (!sm6_metadata_value_is_string(element_node->operands[1])) + { + WARN("Element name is not a string.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Signature element name is not a metadata string."); + return VKD3D_ERROR_INVALID_SHADER; + } + e->semantic_name = element_node->operands[1]->u.string_value; + + e->component_type = vkd3d_component_type_from_dxil_component_type(values[2]); + e->min_precision = minimum_precision_from_dxil_component_type(values[2]); + + j = values[3]; + e->sysval_semantic = sysval_semantic_from_dxil_semantic_kind(j); + if (j != SEMANTIC_KIND_ARBITRARY && j != SEMANTIC_KIND_TARGET && e->sysval_semantic == VKD3D_SHADER_SV_NONE) + { + WARN("Unhandled semantic kind %u.\n", j); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "DXIL semantic kind %u is unhandled.", j); + return VKD3D_ERROR_INVALID_SHADER; + } + + if ((e->interpolation_mode = values[5]) >= VKD3DSIM_COUNT) + { + WARN("Unhandled interpolation mode %u.\n", e->interpolation_mode); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Interpolation mode %u is unhandled.", e->interpolation_mode); + return VKD3D_ERROR_INVALID_SHADER; + } + + e->register_count = values[6]; + column_count = values[7]; + e->register_index = values[8]; + e->target_location = e->register_index; + if (e->register_index > MAX_REG_OUTPUT || e->register_count > MAX_REG_OUTPUT - e->register_index) + { + WARN("Invalid row start %u with row count %u.\n", e->register_index, e->register_count); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "A signature element starting row of %u with count %u is invalid.", + e->register_index, e->register_count); + return VKD3D_ERROR_INVALID_SHADER; + } + index = values[9]; + if (index >= VKD3D_VEC4_SIZE || column_count > VKD3D_VEC4_SIZE - index) + { + WARN("Invalid column start %u with count %u.\n", index, column_count); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "A signature element starting column %u with count %u is invalid.", index, column_count); + return VKD3D_ERROR_INVALID_SHADER; + } + + e->mask = vkd3d_write_mask_from_component_count(column_count); + e->used_mask = e->mask; + e->mask <<= index; + + signature_element_read_additional_element_values(e, element_node, sm6); + e->used_mask <<= index; + + m = element_node->operands[4]; + if (!sm6_metadata_value_is_node(m)) + { + WARN("Semantic index list is not a node.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Signature element semantic index list is not a metadata node."); + return VKD3D_ERROR_INVALID_SHADER; + } + + element_node = m->u.node; + for (j = 0; j < element_node->operand_count; ++j) + { + if (!sm6_metadata_get_uint_value(sm6, element_node->operands[j], &index)) + { + WARN("Failed to get semantic index for row %u.\n", j); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Signature element semantic index for row %u is not an integer.", j); + } + else if (!j) + { + e->semantic_index = index; + } + else if (index != e->semantic_index + j) + { + WARN("Semantic index %u for row %u is not of an incrementing sequence.\n", index, j); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Signature element semantic index %u for row %u is not of an incrementing sequence.", index, j); + } + } + } + + vkd3d_free(s->elements); + s->elements = elements; + s->element_count = operand_count; + + return VKD3D_OK; +} + +static enum vkd3d_result sm6_parser_signatures_init(struct sm6_parser *sm6, const struct sm6_metadata_value *m) +{ + enum vkd3d_result ret; + + if (!sm6_metadata_value_is_node(m)) { - FIXME("Unhandled call to local function.\n"); - vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, - "Call to a local function is unsupported."); - return; + WARN("Signature table is not a node.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE, + "Signature table is not a metadata node."); + return VKD3D_ERROR_INVALID_SHADER; } - if (!sm6_value_is_dx_intrinsic_dcl(fn_value)) - WARN("External function is not a dx intrinsic.\n"); - if (!operand_count) + if (m->u.node->operand_count && (ret = sm6_parser_read_signature(sm6, m->u.node->operands[0], + &sm6->p.shader_desc.input_signature)) < 0) { - WARN("Missing dx intrinsic function id.\n"); - vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND_COUNT, - "The id for a dx intrinsic function is missing."); - return; + return ret; } - - op_value = operands[0]; - if (!sm6_value_is_constant(op_value) || !sm6_type_is_integer(op_value->type)) + if (m->u.node->operand_count > 1 && (ret = sm6_parser_read_signature(sm6, m->u.node->operands[1], + &sm6->p.shader_desc.output_signature)) < 0) { - WARN("dx intrinsic function id is not a constant int.\n"); - vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, - "Expected a constant integer dx intrinsic function id."); - return; + return ret; } - sm6_parser_decode_dx_op(sm6, code_block, register_get_uint_value(&op_value->u.reg), - fn_value->u.function.name, &operands[1], operand_count - 1, ins, dst); + /* TODO: patch constant signature in operand 2. */ + + sm6_parser_init_input_signature(sm6, &sm6->p.shader_desc.input_signature); + sm6_parser_init_output_signature(sm6, &sm6->p.shader_desc.output_signature); + + return VKD3D_OK; } -static void sm6_parser_emit_ret(struct sm6_parser *sm6, const struct dxil_record *record, - struct sm6_block *code_block, struct vkd3d_shader_instruction *ins) +static void sm6_parser_emit_global_flags(struct sm6_parser *sm6, const struct sm6_metadata_value *m) { - if (!dxil_record_validate_operand_count(record, 0, 1, sm6)) - return; - - if (record->operand_count) - FIXME("Non-void return is not implemented.\n"); + enum vkd3d_shader_global_flags global_flags, mask, rotated_flags; + struct vkd3d_shader_instruction *ins; - ins->handler_idx = VKD3DSIH_NOP; + if (!sm6_metadata_get_uint64_value(sm6, m, (uint64_t*)&global_flags)) + { + WARN("Failed to load global flags.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_PROPERTIES, + "Global flags metadata value is not an integer."); + return; + } + /* Rotate SKIP_OPTIMIZATION from bit 0 to bit 4 to match vkd3d_shader_global_flags. */ + mask = (VKD3DSGF_SKIP_OPTIMIZATION << 1) - 1; + rotated_flags = global_flags & mask; + rotated_flags = (rotated_flags >> 1) | ((rotated_flags & 1) << 4); + global_flags = (global_flags & ~mask) | rotated_flags; + + ins = sm6_parser_add_instruction(sm6, VKD3DSIH_DCL_GLOBAL_FLAGS); + ins->declaration.global_flags = global_flags; } -static enum vkd3d_result sm6_parser_function_init(struct sm6_parser *sm6, const struct dxil_block *block, - struct sm6_function *function) +static enum vkd3d_result sm6_parser_emit_thread_group(struct sm6_parser *sm6, const struct sm6_metadata_value *m) { + const struct sm6_metadata_node *node; struct vkd3d_shader_instruction *ins; - const struct dxil_record *record; - bool ret_found, is_terminator; - struct sm6_block *code_block; - struct sm6_value *dst; - size_t i, block_idx; + unsigned int group_sizes[3]; + unsigned int i; - if (sm6->function_count) + if (sm6->p.shader_version.type != VKD3D_SHADER_TYPE_COMPUTE) { - FIXME("Multiple functions are not supported yet.\n"); + WARN("Shader of type %#x has thread group dimensions.\n", sm6->p.shader_version.type); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_PROPERTIES, + "Shader has thread group dimensions but is not a compute shader."); return VKD3D_ERROR_INVALID_SHADER; } - if (!(function->declaration = sm6_parser_next_function_definition(sm6))) + + if (!m || !sm6_metadata_value_is_node(m)) { - WARN("Failed to find definition to match function body.\n"); + WARN("Thread group dimension value is not a node.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_PROPERTIES, + "Thread group dimension metadata value is not a node."); return VKD3D_ERROR_INVALID_SHADER; } - if (block->record_count < 2) + node = m->u.node; + if (node->operand_count != 3) { - /* It should contain at least a block count and a RET instruction. */ - WARN("Invalid function block record count %zu.\n", block->record_count); + WARN("Invalid operand count %u.\n", node->operand_count); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND_COUNT, + "Thread group dimension operand count %u is invalid.", node->operand_count); return VKD3D_ERROR_INVALID_SHADER; } - if (block->records[0]->code != FUNC_CODE_DECLAREBLOCKS || !block->records[0]->operand_count - || block->records[0]->operands[0] > UINT_MAX) + + for (i = 0; i < 3; ++i) { - WARN("Block count declaration not found or invalid.\n"); - return VKD3D_ERROR_INVALID_SHADER; + if (!sm6_metadata_get_uint_value(sm6, node->operands[i], &group_sizes[i])) + { + WARN("Thread group dimension is not an integer value.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_PROPERTIES, + "Thread group dimension metadata value is not an integer."); + return VKD3D_ERROR_INVALID_SHADER; + } + if (!group_sizes[i] || group_sizes[i] > dx_max_thread_group_size[i]) + { + char dim = "XYZ"[i]; + WARN("Invalid thread group %c dimension %u.\n", dim, group_sizes[i]); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_PROPERTIES, + "Thread group %c dimension %u is invalid.", dim, group_sizes[i]); + return VKD3D_ERROR_INVALID_SHADER; + } } - if (!(function->block_count = block->records[0]->operands[0])) + ins = sm6_parser_add_instruction(sm6, VKD3DSIH_DCL_THREAD_GROUP); + ins->declaration.thread_group_size.x = group_sizes[0]; + ins->declaration.thread_group_size.y = group_sizes[1]; + ins->declaration.thread_group_size.z = group_sizes[2]; + + return VKD3D_OK; +} + +static enum vkd3d_result sm6_parser_entry_point_init(struct sm6_parser *sm6) +{ + const struct sm6_metadata_value *m = sm6_parser_find_named_metadata(sm6, "dx.entryPoints"); + const struct sm6_metadata_node *node, *entry_node = m ? m->u.node : NULL; + enum dxil_shader_properties_tag tag; + unsigned int i, operand_count; + const struct sm6_value *value; + enum vkd3d_result ret; + + if (!entry_node || entry_node->operand_count < 2 || !(m = entry_node->operands[0])) { - WARN("Function contains no blocks.\n"); + WARN("No entry point definition found.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_ENTRY_POINT, + "No entry point definition found in the metadata."); return VKD3D_ERROR_INVALID_SHADER; } - if (function->block_count > 1) + + if (m->type != VKD3D_METADATA_VALUE) { - FIXME("Branched shaders are not supported yet.\n"); + WARN("Entry point definition is not a value.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_ENTRY_POINT, + "Entry point definition is not a metadata value."); return VKD3D_ERROR_INVALID_SHADER; } - if (!(function->blocks[0] = sm6_block_create())) + value = m->u.value; + if (!sm6_value_is_function_dcl(value)) { - ERR("Failed to allocate code block.\n"); - return VKD3D_ERROR_OUT_OF_MEMORY; + WARN("Entry point value is not a function definition.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_ENTRY_POINT, + "Entry point metadata value does not contain a function definition."); + return VKD3D_ERROR_INVALID_SHADER; } - code_block = function->blocks[0]; - sm6->cur_max_value = function->value_count; + sm6->entry_point = value->u.function.name; + if (!sm6_metadata_value_is_string(entry_node->operands[1]) + || ascii_strcasecmp(sm6->entry_point, entry_node->operands[1]->u.string_value)) + { + WARN("Entry point function name %s mismatch.\n", sm6->entry_point); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_ENTRY_POINT_MISMATCH, + "Entry point function name %s does not match the name in metadata.", sm6->entry_point); + } - for (i = 1, block_idx = 0, ret_found = false; i < block->record_count; ++i) + if (entry_node->operand_count >= 3 && (m = entry_node->operands[2]) + && (ret = sm6_parser_signatures_init(sm6, m)) < 0) { - sm6->p.location.column = i; + return ret; + } - if (!code_block) + if (entry_node->operand_count >= 5 && (m = entry_node->operands[4])) + { + if (!sm6_metadata_value_is_node(m)) { - WARN("Invalid block count %zu.\n", function->block_count); - vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND, - "Invalid block count %zu.", function->block_count); + WARN("Shader properties list is not a node.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_PROPERTIES, + "Shader properties tag/value list is not a metadata node."); return VKD3D_ERROR_INVALID_SHADER; } - /* block->record_count - 1 is the instruction count, but some instructions - * can emit >1 IR instruction, so extra may be used. */ - if (!vkd3d_array_reserve((void **)&code_block->instructions, &code_block->instruction_capacity, - max(code_block->instruction_count + 1, block->record_count), sizeof(*code_block->instructions))) + node = m->u.node; + if (node->operand_count & 1) { - ERR("Failed to allocate instructions.\n"); - return VKD3D_ERROR_OUT_OF_MEMORY; + WARN("Operand count is not even.\n"); + vkd3d_shader_parser_warning(&sm6->p, VKD3D_SHADER_WARNING_DXIL_IGNORING_OPERANDS, + "Operand count for shader properties tag/value pairs is not even."); } + operand_count = node->operand_count & ~1u; - ins = &code_block->instructions[code_block->instruction_count]; - ins->handler_idx = VKD3DSIH_INVALID; - - dst = sm6_parser_get_current_value(sm6); - dst->type = NULL; - dst->value_type = VALUE_TYPE_REG; - is_terminator = false; - - record = block->records[i]; - switch (record->code) + for (i = 0; i < operand_count; i += 2) { - case FUNC_CODE_INST_CALL: - sm6_parser_emit_call(sm6, record, code_block, ins, dst); - break; - case FUNC_CODE_INST_RET: - sm6_parser_emit_ret(sm6, record, code_block, ins); - is_terminator = true; - ret_found = true; - break; - default: - FIXME("Unhandled dxil instruction %u.\n", record->code); + if (!sm6_metadata_get_uint_value(sm6, node->operands[i], &tag)) + { + WARN("Tag is not an integer value.\n"); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_PROPERTIES, + "Shader properties tag at index %u is not an integer.", i); return VKD3D_ERROR_INVALID_SHADER; - } - - if (sm6->p.failed) - return VKD3D_ERROR; - assert(ins->handler_idx != VKD3DSIH_INVALID); + } - if (is_terminator) - { - ++block_idx; - code_block = (block_idx < function->block_count) ? function->blocks[block_idx] : NULL; + switch (tag) + { + case SHADER_PROPERTIES_FLAGS: + sm6_parser_emit_global_flags(sm6, node->operands[i + 1]); + break; + case SHADER_PROPERTIES_COMPUTE: + if ((ret = sm6_parser_emit_thread_group(sm6, node->operands[i + 1])) < 0) + return ret; + break; + default: + FIXME("Unhandled tag %#x.\n", tag); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_PROPERTIES, + "Shader properties tag %#x is unhandled.", tag); + break; + } } - if (code_block) - code_block->instruction_count += ins->handler_idx != VKD3DSIH_NOP; - else - assert(ins->handler_idx == VKD3DSIH_NOP); - - sm6->value_count += !!dst->type; - } - - if (!ret_found) - { - WARN("Function contains no RET instruction.\n"); - return VKD3D_ERROR_INVALID_SHADER; } return VKD3D_OK; } -static bool sm6_block_emit_instructions(struct sm6_block *block, struct sm6_parser *sm6) +static void sm6_metadata_value_destroy(struct sm6_metadata_value *m) { - struct vkd3d_shader_instruction *ins = sm6_parser_require_space(sm6, block->instruction_count + 1); - - if (!ins) - return false; - - memcpy(ins, block->instructions, block->instruction_count * sizeof(*block->instructions)); - sm6->p.instructions.count += block->instruction_count; - - sm6_parser_add_instruction(sm6, VKD3DSIH_RET); - - return true; + switch (m->type) + { + case VKD3D_METADATA_NODE: + vkd3d_free(m->u.node); + break; + case VKD3D_METADATA_KIND: + vkd3d_free(m->u.kind.name); + break; + case VKD3D_METADATA_STRING: + vkd3d_free(m->u.string_value); + break; + default: + break; + } } -static enum vkd3d_result sm6_parser_module_init(struct sm6_parser *sm6, const struct dxil_block *block, - unsigned int level) +static void sm6_parser_metadata_cleanup(struct sm6_parser *sm6) { - size_t i, old_value_count = sm6->value_count; - struct sm6_function *function; - enum vkd3d_result ret; + unsigned int i, j; - for (i = 0; i < block->child_block_count; ++i) + for (i = 0; i < ARRAY_SIZE(sm6->metadata_tables); ++i) { - if ((ret = sm6_parser_module_init(sm6, block->child_blocks[i], level + 1)) < 0) - return ret; + for (j = 0; j < sm6->metadata_tables[i].count; ++j) + sm6_metadata_value_destroy(&sm6->metadata_tables[i].values[j]); + vkd3d_free(sm6->metadata_tables[i].values); } - - sm6->p.location.line = block->id; - sm6->p.location.column = 0; - - switch (block->id) + for (i = 0; i < sm6->named_metadata_count; ++i) { - case CONSTANTS_BLOCK: - function = &sm6->functions[sm6->function_count]; - sm6->cur_max_value = function->value_count; - return sm6_parser_constants_init(sm6, block); - - case FUNCTION_BLOCK: - function = &sm6->functions[sm6->function_count]; - if ((ret = sm6_parser_function_init(sm6, block, function)) < 0) - return ret; - /* The value index returns to its previous value after handling a function. It's usually nonzero - * at the start because of global constants/variables/function declarations. Function constants - * occur in a child block, so value_count is already saved before they are emitted. */ - memset(&sm6->values[old_value_count], 0, (sm6->value_count - old_value_count) * sizeof(*sm6->values)); - sm6->value_count = old_value_count; - break; - - case BLOCKINFO_BLOCK: - case MODULE_BLOCK: - case PARAMATTR_BLOCK: - case PARAMATTR_GROUP_BLOCK: - case VALUE_SYMTAB_BLOCK: - case METADATA_BLOCK: - case METADATA_ATTACHMENT_BLOCK: - case TYPE_BLOCK: - break; - - default: - FIXME("Unhandled block id %u.\n", block->id); - break; + sm6_metadata_value_destroy(&sm6->named_metadata[i].value); + vkd3d_free(sm6->named_metadata[i].name); } - - return VKD3D_OK; + vkd3d_free(sm6->named_metadata); } static void sm6_type_table_cleanup(struct sm6_type *types, size_t count) @@ -2669,6 +5448,8 @@ static void sm6_parser_destroy(struct vkd3d_shader_parser *parser) sm6_type_table_cleanup(sm6->types, sm6->type_count); sm6_symtab_cleanup(sm6->global_symbols, sm6->global_symbol_count); sm6_functions_cleanup(sm6->functions, sm6->function_count); + sm6_parser_metadata_cleanup(sm6); + vkd3d_free(sm6->descriptors); vkd3d_free(sm6->values); free_shader_desc(&parser->shader_desc); vkd3d_free(sm6); @@ -2679,10 +5460,20 @@ static const struct vkd3d_shader_parser_ops sm6_parser_ops = .parser_destroy = sm6_parser_destroy, }; +static struct sm6_function *sm6_parser_get_function(const struct sm6_parser *sm6, const char *name) +{ + size_t i; + for (i = 0; i < sm6->function_count; ++i) + if (!ascii_strcasecmp(sm6->functions[i].declaration->u.function.name, name)) + return &sm6->functions[i]; + return NULL; +} + static enum vkd3d_result sm6_parser_init(struct sm6_parser *sm6, const uint32_t *byte_code, size_t byte_code_size, const char *source_name, struct vkd3d_shader_message_context *message_context) { const struct shader_signature *output_signature = &sm6->p.shader_desc.output_signature; + const struct shader_signature *input_signature = &sm6->p.shader_desc.input_signature; const struct vkd3d_shader_location location = {.source_name = source_name}; uint32_t version_token, dxil_version, token_count, magic; unsigned int chunk_offset, chunk_size; @@ -2690,8 +5481,9 @@ static enum vkd3d_result sm6_parser_init(struct sm6_parser *sm6, const uint32_t enum bitcode_block_abbreviation abbr; struct vkd3d_shader_version version; struct dxil_block *block; + struct sm6_function *fn; enum vkd3d_result ret; - unsigned int i; + unsigned int i, j; count = byte_code_size / sizeof(*byte_code); if (count < 6) @@ -2838,11 +5630,12 @@ static enum vkd3d_result sm6_parser_init(struct sm6_parser *sm6, const uint32_t return ret; } - if (!(sm6->output_params = shader_parser_get_dst_params(&sm6->p, output_signature->element_count))) + if (!(sm6->output_params = shader_parser_get_dst_params(&sm6->p, output_signature->element_count)) + || !(sm6->input_params = shader_parser_get_dst_params(&sm6->p, input_signature->element_count))) { - ERR("Failed to allocate output parameters.\n"); + ERR("Failed to allocate input/output parameters.\n"); vkd3d_shader_error(message_context, &location, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, - "Out of memory allocating output parameters."); + "Out of memory allocating input/output parameters."); return VKD3D_ERROR_OUT_OF_MEMORY; } @@ -2869,6 +5662,7 @@ static enum vkd3d_result sm6_parser_init(struct sm6_parser *sm6, const uint32_t "Out of memory allocating DXIL value array."); return VKD3D_ERROR_OUT_OF_MEMORY; } + sm6->ssa_next_id = 1; if ((ret = sm6_parser_globals_init(sm6)) < 0) { @@ -2876,7 +5670,35 @@ static enum vkd3d_result sm6_parser_init(struct sm6_parser *sm6, const uint32_t return ret; } - sm6_parser_init_output_signature(sm6, output_signature); + if (!sm6_parser_allocate_named_metadata(sm6)) + { + ERR("Failed to allocate named metadata array.\n"); + return VKD3D_ERROR_OUT_OF_MEMORY; + } + + for (i = 0, j = 0; i < sm6->root_block.child_block_count; ++i) + { + block = sm6->root_block.child_blocks[i]; + if (block->id != METADATA_BLOCK) + continue; + + if (j == ARRAY_SIZE(sm6->metadata_tables)) + { + FIXME("Too many metadata tables.\n"); + vkd3d_shader_error(message_context, &location, VKD3D_SHADER_ERROR_DXIL_INVALID_METADATA, + "A metadata table count greater than %zu is unsupported.", ARRAY_SIZE(sm6->metadata_tables)); + return VKD3D_ERROR_INVALID_SHADER; + } + + if ((ret = sm6_parser_metadata_init(sm6, block, &sm6->metadata_tables[j++])) < 0) + return ret; + } + + if ((ret = sm6_parser_entry_point_init(sm6)) < 0) + return ret; + + if ((ret = sm6_parser_resources_init(sm6)) < 0) + return ret; if ((ret = sm6_parser_module_init(sm6, &sm6->root_block, 0)) < 0) { @@ -2889,22 +5711,31 @@ static enum vkd3d_result sm6_parser_init(struct sm6_parser *sm6, const uint32_t return ret; } - if (!sm6_parser_require_space(sm6, output_signature->element_count)) + if (!sm6_parser_require_space(sm6, output_signature->element_count + input_signature->element_count)) { vkd3d_shader_error(message_context, &location, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, "Out of memory emitting shader signature declarations."); return VKD3D_ERROR_OUT_OF_MEMORY; } sm6_parser_emit_output_signature(sm6, output_signature); + sm6_parser_emit_input_signature(sm6, input_signature); - for (i = 0; i < sm6->function_count; ++i) + sm6->p.shader_desc.ssa_count = sm6->ssa_next_id; + + if (!(fn = sm6_parser_get_function(sm6, sm6->entry_point))) { - if (!sm6_block_emit_instructions(sm6->functions[i].blocks[0], sm6)) - { - vkd3d_shader_error(message_context, &location, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, - "Out of memory emitting shader instructions."); - return VKD3D_ERROR_OUT_OF_MEMORY; - } + WARN("Failed to find entry point %s.\n", sm6->entry_point); + vkd3d_shader_parser_error(&sm6->p, VKD3D_SHADER_ERROR_DXIL_INVALID_ENTRY_POINT, + "The definition of the entry point function '%s' was not found.", sm6->entry_point); + return VKD3D_ERROR_INVALID_SHADER; + } + + assert(sm6->function_count == 1); + if (!sm6_block_emit_instructions(fn->blocks[0], sm6)) + { + vkd3d_shader_error(message_context, &location, VKD3D_SHADER_ERROR_DXIL_OUT_OF_MEMORY, + "Out of memory emitting shader instructions."); + return VKD3D_ERROR_OUT_OF_MEMORY; } dxil_block_destroy(&sm6->root_block); @@ -2955,6 +5786,12 @@ int vkd3d_shader_sm6_parser_create(const struct vkd3d_shader_compile_info *compi compile_info->source_name, message_context); vkd3d_free(byte_code); + if (!sm6->p.failed && ret >= 0) + vsir_validate(&sm6->p); + + if (sm6->p.failed && ret >= 0) + ret = VKD3D_ERROR_INVALID_SHADER; + if (ret < 0) { WARN("Failed to initialise shader parser.\n"); diff --git a/libs/vkd3d/libs/vkd3d-shader/hlsl.c b/libs/vkd3d/libs/vkd3d-shader/hlsl.c index 5fe9047bf25..501bff8cfb8 100644 --- a/libs/vkd3d/libs/vkd3d-shader/hlsl.c +++ b/libs/vkd3d/libs/vkd3d-shader/hlsl.c @@ -216,13 +216,15 @@ bool hlsl_type_is_resource(const struct hlsl_type *type) return false; } -enum hlsl_regset hlsl_type_get_regset(const struct hlsl_type *type) +/* Only intended to be used for derefs (after copies have been lowered to components or vectors) or + * resources, since for both their data types span across a single regset. */ +static enum hlsl_regset type_get_regset(const struct hlsl_type *type) { - if (type->class <= HLSL_CLASS_LAST_NUMERIC) + if (hlsl_is_numeric_type(type)) return HLSL_REGSET_NUMERIC; if (type->class == HLSL_CLASS_ARRAY) - return hlsl_type_get_regset(type->e.array.type); + return type_get_regset(type->e.array.type); if (type->class == HLSL_CLASS_OBJECT) { @@ -245,6 +247,11 @@ enum hlsl_regset hlsl_type_get_regset(const struct hlsl_type *type) vkd3d_unreachable(); } +enum hlsl_regset hlsl_deref_get_regset(struct hlsl_ctx *ctx, const struct hlsl_deref *deref) +{ + return type_get_regset(hlsl_deref_get_type(ctx, deref)); +} + unsigned int hlsl_type_get_sm4_offset(const struct hlsl_type *type, unsigned int offset) { /* Align to the next vec4 boundary if: @@ -324,7 +331,7 @@ static void hlsl_type_calculate_reg_size(struct hlsl_ctx *ctx, struct hlsl_type { if (hlsl_type_is_resource(type)) { - enum hlsl_regset regset = hlsl_type_get_regset(type); + enum hlsl_regset regset = type_get_regset(type); type->reg_size[regset] = 1; } @@ -452,11 +459,11 @@ struct hlsl_type *hlsl_type_get_component_type(struct hlsl_ctx *ctx, struct hlsl } unsigned int hlsl_type_get_component_offset(struct hlsl_ctx *ctx, struct hlsl_type *type, - enum hlsl_regset regset, unsigned int index) + unsigned int index, enum hlsl_regset *regset) { + unsigned int offset[HLSL_REGSET_LAST + 1] = {0}; struct hlsl_type *next_type; - unsigned int offset = 0; - unsigned int idx; + unsigned int idx, r; while (!type_is_single_component(type)) { @@ -468,19 +475,22 @@ unsigned int hlsl_type_get_component_offset(struct hlsl_ctx *ctx, struct hlsl_ty case HLSL_CLASS_SCALAR: case HLSL_CLASS_VECTOR: case HLSL_CLASS_MATRIX: - if (regset == HLSL_REGSET_NUMERIC) - offset += idx; + offset[HLSL_REGSET_NUMERIC] += idx; break; case HLSL_CLASS_STRUCT: - offset += type->e.record.fields[idx].reg_offset[regset]; + for (r = 0; r <= HLSL_REGSET_LAST; ++r) + offset[r] += type->e.record.fields[idx].reg_offset[r]; break; case HLSL_CLASS_ARRAY: - if (regset == HLSL_REGSET_NUMERIC) - offset += idx * align(type->e.array.type->reg_size[regset], 4); - else - offset += idx * type->e.array.type->reg_size[regset]; + for (r = 0; r <= HLSL_REGSET_LAST; ++r) + { + if (r == HLSL_REGSET_NUMERIC) + offset[r] += idx * align(type->e.array.type->reg_size[r], 4); + else + offset[r] += idx * type->e.array.type->reg_size[r]; + } break; case HLSL_CLASS_OBJECT: @@ -493,7 +503,8 @@ unsigned int hlsl_type_get_component_offset(struct hlsl_ctx *ctx, struct hlsl_ty type = next_type; } - return offset; + *regset = type_get_regset(type); + return offset[*regset]; } static bool init_deref(struct hlsl_ctx *ctx, struct hlsl_deref *deref, struct hlsl_ir_var *var, @@ -501,7 +512,9 @@ static bool init_deref(struct hlsl_ctx *ctx, struct hlsl_deref *deref, struct hl { deref->var = var; deref->path_len = path_len; - deref->offset.node = NULL; + deref->rel_offset.node = NULL; + deref->const_offset = 0; + deref->data_type = NULL; if (path_len == 0) { @@ -528,7 +541,8 @@ bool hlsl_init_deref_from_index_chain(struct hlsl_ctx *ctx, struct hlsl_deref *d deref->path = NULL; deref->path_len = 0; - deref->offset.node = NULL; + deref->rel_offset.node = NULL; + deref->const_offset = 0; assert(chain); if (chain->type == HLSL_IR_INDEX) @@ -591,7 +605,7 @@ struct hlsl_type *hlsl_deref_get_type(struct hlsl_ctx *ctx, const struct hlsl_de assert(deref); - if (deref->offset.node) + if (hlsl_deref_is_lowered(deref)) return deref->data_type; type = deref->var->data_type; @@ -745,7 +759,7 @@ struct hlsl_type *hlsl_new_uav_type(struct hlsl_ctx *ctx, enum hlsl_sampler_dim { struct hlsl_type *type; - if (!(type = vkd3d_calloc(1, sizeof(*type)))) + if (!(type = hlsl_alloc(ctx, sizeof(*type)))) return NULL; type->class = HLSL_CLASS_OBJECT; type->base_type = HLSL_TYPE_UAV; @@ -811,17 +825,15 @@ struct hlsl_ir_function *hlsl_get_function(struct hlsl_ctx *ctx, const char *nam return NULL; } -struct hlsl_ir_function_decl *hlsl_get_func_decl(struct hlsl_ctx *ctx, const char *name) +struct hlsl_ir_function_decl *hlsl_get_first_func_decl(struct hlsl_ctx *ctx, const char *name) { - struct hlsl_ir_function_decl *decl; struct hlsl_ir_function *func; struct rb_entry *entry; if ((entry = rb_get(&ctx->functions, name))) { func = RB_ENTRY_VALUE(entry, struct hlsl_ir_function, entry); - RB_FOR_EACH_ENTRY(decl, &func->overloads, struct hlsl_ir_function_decl, entry) - return decl; + return LIST_ENTRY(list_head(&func->overloads), struct hlsl_ir_function_decl, entry); } return NULL; @@ -1102,7 +1114,7 @@ bool hlsl_copy_deref(struct hlsl_ctx *ctx, struct hlsl_deref *deref, const struc if (!other) return true; - assert(!other->offset.node); + assert(!hlsl_deref_is_lowered(other)); if (!init_deref(ctx, deref, other->var, other->path_len)) return false; @@ -1124,7 +1136,8 @@ void hlsl_cleanup_deref(struct hlsl_deref *deref) deref->path = NULL; deref->path_len = 0; - hlsl_src_remove(&deref->offset); + hlsl_src_remove(&deref->rel_offset); + deref->const_offset = 0; } /* Initializes a simple variable dereference, so that it can be passed to load/store functions. */ @@ -1159,7 +1172,7 @@ struct hlsl_ir_node *hlsl_new_store_index(struct hlsl_ctx *ctx, const struct hls unsigned int i; assert(lhs); - assert(!lhs->offset.node); + assert(!hlsl_deref_is_lowered(lhs)); if (!(store = hlsl_alloc(ctx, sizeof(*store)))) return NULL; @@ -1325,6 +1338,40 @@ struct hlsl_ir_node *hlsl_new_if(struct hlsl_ctx *ctx, struct hlsl_ir_node *cond return &iff->node; } +struct hlsl_ir_switch_case *hlsl_new_switch_case(struct hlsl_ctx *ctx, unsigned int value, + bool is_default, struct hlsl_block *body, const struct vkd3d_shader_location *loc) +{ + struct hlsl_ir_switch_case *c; + + if (!(c = hlsl_alloc(ctx, sizeof(*c)))) + return NULL; + + c->value = value; + c->is_default = is_default; + hlsl_block_init(&c->body); + if (body) + hlsl_block_add_block(&c->body, body); + c->loc = *loc; + + return c; +} + +struct hlsl_ir_node *hlsl_new_switch(struct hlsl_ctx *ctx, struct hlsl_ir_node *selector, + struct list *cases, const struct vkd3d_shader_location *loc) +{ + struct hlsl_ir_switch *s; + + if (!(s = hlsl_alloc(ctx, sizeof(*s)))) + return NULL; + init_node(&s->node, HLSL_IR_SWITCH, NULL, loc); + hlsl_src_from_node(&s->selector, selector); + list_init(&s->cases); + if (cases) + list_move_head(&s->cases, cases); + + return &s->node; +} + struct hlsl_ir_load *hlsl_new_load_index(struct hlsl_ctx *ctx, const struct hlsl_deref *deref, struct hlsl_ir_node *idx, const struct vkd3d_shader_location *loc) { @@ -1332,7 +1379,7 @@ struct hlsl_ir_load *hlsl_new_load_index(struct hlsl_ctx *ctx, const struct hlsl struct hlsl_type *type; unsigned int i; - assert(!deref->offset.node); + assert(!hlsl_deref_is_lowered(deref)); type = hlsl_deref_get_type(ctx, deref); if (idx) @@ -1568,7 +1615,7 @@ static bool clone_block(struct hlsl_ctx *ctx, struct hlsl_block *dst_block, if (!list_empty(&src->uses)) { - if (!vkd3d_array_reserve((void **)&map->instrs, &map->capacity, map->count + 1, sizeof(*map->instrs))) + if (!hlsl_array_reserve(ctx, (void **)&map->instrs, &map->capacity, map->count + 1, sizeof(*map->instrs))) { hlsl_block_cleanup(dst_block); return false; @@ -1605,7 +1652,7 @@ static bool clone_deref(struct hlsl_ctx *ctx, struct clone_instr_map *map, { unsigned int i; - assert(!src->offset.node); + assert(!hlsl_deref_is_lowered(src)); if (!init_deref(ctx, dst, src->var, src->path_len)) return false; @@ -1787,6 +1834,58 @@ static struct hlsl_ir_node *clone_index(struct hlsl_ctx *ctx, struct clone_instr return dst; } +void hlsl_free_ir_switch_case(struct hlsl_ir_switch_case *c) +{ + hlsl_block_cleanup(&c->body); + list_remove(&c->entry); + vkd3d_free(c); +} + +void hlsl_cleanup_ir_switch_cases(struct list *cases) +{ + struct hlsl_ir_switch_case *c, *next; + + LIST_FOR_EACH_ENTRY_SAFE(c, next, cases, struct hlsl_ir_switch_case, entry) + { + hlsl_free_ir_switch_case(c); + } +} + +static struct hlsl_ir_node *clone_switch(struct hlsl_ctx *ctx, + struct clone_instr_map *map, struct hlsl_ir_switch *s) +{ + struct hlsl_ir_switch_case *c, *d; + struct hlsl_ir_node *ret; + struct hlsl_block body; + struct list cases; + + list_init(&cases); + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + if (!(clone_block(ctx, &body, &c->body, map))) + { + hlsl_cleanup_ir_switch_cases(&cases); + return NULL; + } + + d = hlsl_new_switch_case(ctx, c->value, c->is_default, &body, &c->loc); + hlsl_block_cleanup(&body); + if (!d) + { + hlsl_cleanup_ir_switch_cases(&cases); + return NULL; + } + + list_add_tail(&cases, &d->entry); + } + + ret = hlsl_new_switch(ctx, map_instr(map, s->selector.node), &cases, &s->node.loc); + hlsl_cleanup_ir_switch_cases(&cases); + + return ret; +} + static struct hlsl_ir_node *clone_instr(struct hlsl_ctx *ctx, struct clone_instr_map *map, const struct hlsl_ir_node *instr) { @@ -1825,6 +1924,9 @@ static struct hlsl_ir_node *clone_instr(struct hlsl_ctx *ctx, case HLSL_IR_STORE: return clone_store(ctx, map, hlsl_ir_store(instr)); + case HLSL_IR_SWITCH: + return clone_switch(ctx, map, hlsl_ir_switch(instr)); + case HLSL_IR_SWIZZLE: return clone_swizzle(ctx, map, hlsl_ir_swizzle(instr)); } @@ -1946,76 +2048,38 @@ void hlsl_pop_scope(struct hlsl_ctx *ctx) ctx->cur_scope = prev_scope; } -static int compare_param_hlsl_types(const struct hlsl_type *t1, const struct hlsl_type *t2) +static bool func_decl_matches(const struct hlsl_ir_function_decl *decl, + const struct hlsl_func_parameters *parameters) { - int r; - - if ((r = vkd3d_u32_compare(t1->class, t2->class))) - { - if (!((t1->class == HLSL_CLASS_SCALAR && t2->class == HLSL_CLASS_VECTOR) - || (t1->class == HLSL_CLASS_VECTOR && t2->class == HLSL_CLASS_SCALAR))) - return r; - } - if ((r = vkd3d_u32_compare(t1->base_type, t2->base_type))) - return r; - if (t1->base_type == HLSL_TYPE_SAMPLER || t1->base_type == HLSL_TYPE_TEXTURE) - { - if ((r = vkd3d_u32_compare(t1->sampler_dim, t2->sampler_dim))) - return r; - if (t1->base_type == HLSL_TYPE_TEXTURE && t1->sampler_dim != HLSL_SAMPLER_DIM_GENERIC - && (r = compare_param_hlsl_types(t1->e.resource_format, t2->e.resource_format))) - return r; - } - if ((r = vkd3d_u32_compare(t1->dimx, t2->dimx))) - return r; - if ((r = vkd3d_u32_compare(t1->dimy, t2->dimy))) - return r; - if (t1->class == HLSL_CLASS_STRUCT) - { - size_t i; - - if (t1->e.record.field_count != t2->e.record.field_count) - return t1->e.record.field_count - t2->e.record.field_count; - - for (i = 0; i < t1->e.record.field_count; ++i) - { - const struct hlsl_struct_field *field1 = &t1->e.record.fields[i]; - const struct hlsl_struct_field *field2 = &t2->e.record.fields[i]; + size_t i; - if ((r = compare_param_hlsl_types(field1->type, field2->type))) - return r; + if (parameters->count != decl->parameters.count) + return false; - if ((r = strcmp(field1->name, field2->name))) - return r; - } - return 0; - } - if (t1->class == HLSL_CLASS_ARRAY) + for (i = 0; i < parameters->count; ++i) { - if ((r = vkd3d_u32_compare(t1->e.array.elements_count, t2->e.array.elements_count))) - return r; - return compare_param_hlsl_types(t1->e.array.type, t2->e.array.type); + if (!hlsl_types_are_equal(parameters->vars[i]->data_type, decl->parameters.vars[i]->data_type)) + return false; } - - return 0; + return true; } -static int compare_function_decl_rb(const void *key, const struct rb_entry *entry) +struct hlsl_ir_function_decl *hlsl_get_func_decl(struct hlsl_ctx *ctx, const char *name, + const struct hlsl_func_parameters *parameters) { - const struct hlsl_ir_function_decl *decl = RB_ENTRY_VALUE(entry, const struct hlsl_ir_function_decl, entry); - const struct hlsl_func_parameters *parameters = key; - size_t i; - int r; + struct hlsl_ir_function_decl *decl; + struct hlsl_ir_function *func; - if ((r = vkd3d_u32_compare(parameters->count, decl->parameters.count))) - return r; + if (!(func = hlsl_get_function(ctx, name))) + return NULL; - for (i = 0; i < parameters->count; ++i) + LIST_FOR_EACH_ENTRY(decl, &func->overloads, struct hlsl_ir_function_decl, entry) { - if ((r = compare_param_hlsl_types(parameters->vars[i]->data_type, decl->parameters.vars[i]->data_type))) - return r; + if (func_decl_matches(decl, parameters)) + return decl; } - return 0; + + return NULL; } struct vkd3d_string_buffer *hlsl_type_to_string(struct hlsl_ctx *ctx, const struct hlsl_type *type) @@ -2193,6 +2257,10 @@ struct vkd3d_string_buffer *hlsl_modifiers_to_string(struct hlsl_ctx *ctx, unsig vkd3d_string_buffer_printf(string, "extern "); if (modifiers & HLSL_STORAGE_NOINTERPOLATION) vkd3d_string_buffer_printf(string, "nointerpolation "); + if (modifiers & HLSL_STORAGE_CENTROID) + vkd3d_string_buffer_printf(string, "centroid "); + if (modifiers & HLSL_STORAGE_NOPERSPECTIVE) + vkd3d_string_buffer_printf(string, "noperspective "); if (modifiers & HLSL_MODIFIER_PRECISE) vkd3d_string_buffer_printf(string, "precise "); if (modifiers & HLSL_STORAGE_SHARED) @@ -2239,6 +2307,7 @@ const char *hlsl_node_type_to_string(enum hlsl_ir_node_type type) [HLSL_IR_RESOURCE_LOAD ] = "HLSL_IR_RESOURCE_LOAD", [HLSL_IR_RESOURCE_STORE] = "HLSL_IR_RESOURCE_STORE", [HLSL_IR_STORE ] = "HLSL_IR_STORE", + [HLSL_IR_SWITCH ] = "HLSL_IR_SWITCH", [HLSL_IR_SWIZZLE ] = "HLSL_IR_SWIZZLE", }; @@ -2305,21 +2374,34 @@ static void dump_deref(struct vkd3d_string_buffer *buffer, const struct hlsl_der if (deref->var) { vkd3d_string_buffer_printf(buffer, "%s", deref->var->name); - if (deref->path_len) + if (!hlsl_deref_is_lowered(deref)) { - vkd3d_string_buffer_printf(buffer, "["); - for (i = 0; i < deref->path_len; ++i) + if (deref->path_len) { vkd3d_string_buffer_printf(buffer, "["); - dump_src(buffer, &deref->path[i]); + for (i = 0; i < deref->path_len; ++i) + { + vkd3d_string_buffer_printf(buffer, "["); + dump_src(buffer, &deref->path[i]); + vkd3d_string_buffer_printf(buffer, "]"); + } vkd3d_string_buffer_printf(buffer, "]"); } - vkd3d_string_buffer_printf(buffer, "]"); } - else if (deref->offset.node) + else { + bool show_rel, show_const; + + show_rel = deref->rel_offset.node; + show_const = deref->const_offset != 0 || !show_rel; + vkd3d_string_buffer_printf(buffer, "["); - dump_src(buffer, &deref->offset); + if (show_rel) + dump_src(buffer, &deref->rel_offset); + if (show_rel && show_const) + vkd3d_string_buffer_printf(buffer, " + "); + if (show_const) + vkd3d_string_buffer_printf(buffer, "%uc", deref->const_offset); vkd3d_string_buffer_printf(buffer, "]"); } } @@ -2440,6 +2522,7 @@ const char *debug_hlsl_expr_op(enum hlsl_ir_expr_op op) [HLSL_OP1_ABS] = "abs", [HLSL_OP1_BIT_NOT] = "~", [HLSL_OP1_CAST] = "cast", + [HLSL_OP1_CEIL] = "ceil", [HLSL_OP1_COS] = "cos", [HLSL_OP1_COS_REDUCED] = "cos_reduced", [HLSL_OP1_DSX] = "dsx", @@ -2449,6 +2532,7 @@ const char *debug_hlsl_expr_op(enum hlsl_ir_expr_op op) [HLSL_OP1_DSY_COARSE] = "dsy_coarse", [HLSL_OP1_DSY_FINE] = "dsy_fine", [HLSL_OP1_EXP2] = "exp2", + [HLSL_OP1_FLOOR] = "floor", [HLSL_OP1_FRACT] = "fract", [HLSL_OP1_LOG2] = "log2", [HLSL_OP1_LOGIC_NOT] = "!", @@ -2485,6 +2569,7 @@ const char *debug_hlsl_expr_op(enum hlsl_ir_expr_op op) [HLSL_OP2_NEQUAL] = "!=", [HLSL_OP2_RSHIFT] = ">>", + [HLSL_OP3_CMP] = "cmp", [HLSL_OP3_DP2ADD] = "dp2add", [HLSL_OP3_MOVC] = "movc", [HLSL_OP3_TERNARY] = "ternary", @@ -2540,6 +2625,10 @@ static void dump_ir_jump(struct vkd3d_string_buffer *buffer, const struct hlsl_i case HLSL_IR_JUMP_RETURN: vkd3d_string_buffer_printf(buffer, "return"); break; + + case HLSL_IR_JUMP_UNRESOLVED_CONTINUE: + vkd3d_string_buffer_printf(buffer, "unresolved_continue"); + break; } } @@ -2659,6 +2748,32 @@ static void dump_ir_index(struct vkd3d_string_buffer *buffer, const struct hlsl_ vkd3d_string_buffer_printf(buffer, "]"); } +static void dump_ir_switch(struct hlsl_ctx *ctx, struct vkd3d_string_buffer *buffer, const struct hlsl_ir_switch *s) +{ + struct hlsl_ir_switch_case *c; + + vkd3d_string_buffer_printf(buffer, "switch ("); + dump_src(buffer, &s->selector); + vkd3d_string_buffer_printf(buffer, ") {\n"); + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + if (c->is_default) + { + vkd3d_string_buffer_printf(buffer, " %10s default: {\n", ""); + } + else + { + vkd3d_string_buffer_printf(buffer, " %10s case %u : {\n", "", c->value); + } + + dump_block(ctx, buffer, &c->body); + vkd3d_string_buffer_printf(buffer, " %10s }\n", ""); + } + + vkd3d_string_buffer_printf(buffer, " %10s }", ""); +} + static void dump_instr(struct hlsl_ctx *ctx, struct vkd3d_string_buffer *buffer, const struct hlsl_ir_node *instr) { if (instr->index) @@ -2714,6 +2829,10 @@ static void dump_instr(struct hlsl_ctx *ctx, struct vkd3d_string_buffer *buffer, dump_ir_store(buffer, hlsl_ir_store(instr)); break; + case HLSL_IR_SWITCH: + dump_ir_switch(ctx, buffer, hlsl_ir_switch(instr)); + break; + case HLSL_IR_SWIZZLE: dump_ir_swizzle(buffer, hlsl_ir_swizzle(instr)); break; @@ -2855,7 +2974,7 @@ static void free_ir_resource_load(struct hlsl_ir_resource_load *load) static void free_ir_resource_store(struct hlsl_ir_resource_store *store) { - hlsl_src_remove(&store->resource.offset); + hlsl_src_remove(&store->resource.rel_offset); hlsl_src_remove(&store->coords); hlsl_src_remove(&store->value); vkd3d_free(store); @@ -2874,6 +2993,14 @@ static void free_ir_swizzle(struct hlsl_ir_swizzle *swizzle) vkd3d_free(swizzle); } +static void free_ir_switch(struct hlsl_ir_switch *s) +{ + hlsl_src_remove(&s->selector); + hlsl_cleanup_ir_switch_cases(&s->cases); + + vkd3d_free(s); +} + static void free_ir_index(struct hlsl_ir_index *index) { hlsl_src_remove(&index->val); @@ -2934,6 +3061,10 @@ void hlsl_free_instr(struct hlsl_ir_node *node) case HLSL_IR_SWIZZLE: free_ir_swizzle(hlsl_ir_swizzle(node)); break; + + case HLSL_IR_SWITCH: + free_ir_switch(hlsl_ir_switch(node)); + break; } } @@ -2967,14 +3098,12 @@ static void free_function_decl(struct hlsl_ir_function_decl *decl) vkd3d_free(decl); } -static void free_function_decl_rb(struct rb_entry *entry, void *context) -{ - free_function_decl(RB_ENTRY_VALUE(entry, struct hlsl_ir_function_decl, entry)); -} - static void free_function(struct hlsl_ir_function *func) { - rb_destroy(&func->overloads, free_function_decl_rb, NULL); + struct hlsl_ir_function_decl *decl, *next; + + LIST_FOR_EACH_ENTRY_SAFE(decl, next, &func->overloads, struct hlsl_ir_function_decl, entry) + free_function_decl(decl); vkd3d_free((void *)func->name); vkd3d_free(func); } @@ -3004,17 +3133,15 @@ void hlsl_add_function(struct hlsl_ctx *ctx, char *name, struct hlsl_ir_function { func = RB_ENTRY_VALUE(func_entry, struct hlsl_ir_function, entry); decl->func = func; - - if (rb_put(&func->overloads, &decl->parameters, &decl->entry) == -1) - ERR("Failed to insert function overload.\n"); + list_add_tail(&func->overloads, &decl->entry); vkd3d_free(name); return; } func = hlsl_alloc(ctx, sizeof(*func)); func->name = name; - rb_init(&func->overloads, compare_function_decl_rb); + list_init(&func->overloads); decl->func = func; - rb_put(&func->overloads, &decl->parameters, &decl->entry); + list_add_tail(&func->overloads, &decl->entry); rb_put(&ctx->functions, func->name, &func->entry); } @@ -3092,7 +3219,7 @@ unsigned int hlsl_combine_swizzles(unsigned int first, unsigned int second, unsi return ret; } -static const struct hlsl_profile_info *get_target_info(const char *target) +const struct hlsl_profile_info *hlsl_get_target_info(const char *target) { unsigned int i; @@ -3395,6 +3522,10 @@ static bool hlsl_ctx_init(struct hlsl_ctx *ctx, const struct vkd3d_shader_compil else if (option->value == VKD3D_SHADER_COMPILE_OPTION_PACK_MATRIX_COLUMN_MAJOR) ctx->matrix_majority = HLSL_MODIFIER_COLUMN_MAJOR; } + else if (option->name == VKD3D_SHADER_COMPILE_OPTION_BACKWARD_COMPATIBILITY) + { + ctx->semantic_compat_mapping = option->value & VKD3D_SHADER_COMPILE_OPTION_BACKCOMPAT_MAP_SEMANTIC_NAMES; + } } return true; @@ -3455,14 +3586,12 @@ int hlsl_compile_shader(const struct vkd3d_shader_code *hlsl, const struct vkd3d } entry_point = hlsl_source_info->entry_point ? hlsl_source_info->entry_point : "main"; - if (!(profile = get_target_info(hlsl_source_info->profile))) + if (!(profile = hlsl_get_target_info(hlsl_source_info->profile))) { FIXME("Unknown compilation target %s.\n", debugstr_a(hlsl_source_info->profile)); return VKD3D_ERROR_NOT_IMPLEMENTED; } - vkd3d_shader_dump_shader(compile_info->source_type, profile->type, &compile_info->source); - if (compile_info->target_type == VKD3D_SHADER_TARGET_D3D_BYTECODE && profile->major_version > 3) { vkd3d_shader_error(message_context, NULL, VKD3D_SHADER_ERROR_HLSL_INCOMPATIBLE_PROFILE, @@ -3501,7 +3630,7 @@ int hlsl_compile_shader(const struct vkd3d_shader_code *hlsl, const struct vkd3d if ((func = hlsl_get_function(&ctx, entry_point))) { - RB_FOR_EACH_ENTRY(decl, &func->overloads, struct hlsl_ir_function_decl, entry) + LIST_FOR_EACH_ENTRY(decl, &func->overloads, struct hlsl_ir_function_decl, entry) { if (!decl->has_body) continue; @@ -3567,7 +3696,7 @@ struct hlsl_ir_function_decl *hlsl_compile_internal_function(struct hlsl_ctx *ct hlsl_release_string_buffer(ctx, internal_name); return NULL; } - func = hlsl_get_func_decl(ctx, internal_name->buffer); + func = hlsl_get_first_func_decl(ctx, internal_name->buffer); hlsl_release_string_buffer(ctx, internal_name); return func; } diff --git a/libs/vkd3d/libs/vkd3d-shader/hlsl.h b/libs/vkd3d/libs/vkd3d-shader/hlsl.h index 743a746f2bf..a0065572539 100644 --- a/libs/vkd3d/libs/vkd3d-shader/hlsl.h +++ b/libs/vkd3d/libs/vkd3d-shader/hlsl.h @@ -21,13 +21,10 @@ #define __VKD3D_SHADER_HLSL_H #include "vkd3d_shader_private.h" -#include "wine/rbtree.h" +#include "rbtree.h" #include "d3dcommon.h" #include "d3dx9shader.h" -enum vkd3d_sm4_register_type; -enum vkd3d_sm4_swizzle_type; - /* The general IR structure is inspired by Mesa GLSL hir, even though the code * ends up being quite different in practice. Anyway, here comes the relevant * licensing information. @@ -284,6 +281,7 @@ enum hlsl_ir_node_type HLSL_IR_RESOURCE_STORE, HLSL_IR_STORE, HLSL_IR_SWIZZLE, + HLSL_IR_SWITCH, }; /* Common data for every type of IR instruction node. */ @@ -357,11 +355,16 @@ struct hlsl_attribute #define HLSL_STORAGE_IN 0x00000800 #define HLSL_STORAGE_OUT 0x00001000 #define HLSL_MODIFIER_INLINE 0x00002000 +#define HLSL_STORAGE_CENTROID 0x00004000 +#define HLSL_STORAGE_NOPERSPECTIVE 0x00008000 #define HLSL_TYPE_MODIFIERS_MASK (HLSL_MODIFIER_PRECISE | HLSL_MODIFIER_VOLATILE | \ HLSL_MODIFIER_CONST | HLSL_MODIFIER_ROW_MAJOR | \ HLSL_MODIFIER_COLUMN_MAJOR) +#define HLSL_INTERPOLATION_MODIFIERS_MASK (HLSL_STORAGE_NOINTERPOLATION | HLSL_STORAGE_CENTROID | \ + HLSL_STORAGE_NOPERSPECTIVE) + #define HLSL_MODIFIERS_MAJORITY_MASK (HLSL_MODIFIER_ROW_MAJOR | HLSL_MODIFIER_COLUMN_MAJOR) #define HLSL_ARRAY_ELEMENTS_COUNT_IMPLICIT 0 @@ -369,7 +372,7 @@ struct hlsl_attribute /* Reservation of a register and/or an offset for objects inside constant buffers, to be used as a * starting point of their allocation. They are available through the register(·) and the * packoffset(·) syntaxes, respectivelly. - * The costant buffer offset is measured register components. */ + * The constant buffer offset is measured register components. */ struct hlsl_reg_reservation { char reg_type; @@ -422,6 +425,9 @@ struct hlsl_ir_var * It may be less than the allocation size, e.g. for texture arrays. */ unsigned int bind_count[HLSL_REGSET_LAST_OBJECT + 1]; + /* Whether the shader performs dereferences with non-constant offsets in the variable. */ + bool indexable; + uint32_t is_input_semantic : 1; uint32_t is_output_semantic : 1; uint32_t is_uniform : 1; @@ -444,7 +450,7 @@ struct hlsl_ir_function const char *name; /* Tree containing function definitions, stored as hlsl_ir_function_decl structures, which would * be more than one in case of function overloading. */ - struct rb_tree overloads; + struct list overloads; }; struct hlsl_ir_function_decl @@ -454,8 +460,8 @@ struct hlsl_ir_function_decl struct hlsl_ir_var *return_var; struct vkd3d_shader_location loc; - /* Item entry in hlsl_ir_function.overloads. The parameters' types are used as key. */ - struct rb_entry entry; + /* Item entry in hlsl_ir_function.overloads. */ + struct list entry; /* Function to which this declaration corresponds. */ struct hlsl_ir_function *func; @@ -497,6 +503,22 @@ struct hlsl_ir_loop unsigned int next_index; /* liveness index of the end of the loop */ }; +struct hlsl_ir_switch_case +{ + unsigned int value; + bool is_default; + struct hlsl_block body; + struct list entry; + struct vkd3d_shader_location loc; +}; + +struct hlsl_ir_switch +{ + struct hlsl_ir_node node; + struct hlsl_src selector; + struct list cases; +}; + enum hlsl_ir_expr_op { HLSL_OP0_VOID, @@ -504,6 +526,7 @@ enum hlsl_ir_expr_op HLSL_OP1_ABS, HLSL_OP1_BIT_NOT, HLSL_OP1_CAST, + HLSL_OP1_CEIL, HLSL_OP1_COS, HLSL_OP1_COS_REDUCED, /* Reduced range [-pi, pi] */ HLSL_OP1_DSX, @@ -556,7 +579,10 @@ enum hlsl_ir_expr_op /* MOVC(a, b, c) returns c if a is bitwise zero and b otherwise. * TERNARY(a, b, c) returns c if a == 0 and b otherwise. * They differ for floating point numbers, because - * -0.0 == 0.0, but it is not bitwise zero. */ + * -0.0 == 0.0, but it is not bitwise zero. CMP(a, b, c) returns b + if a >= 0, and c otherwise. It's used only for SM1-SM3 targets, while + SM4+ is using MOVC in such cases. */ + HLSL_OP3_CMP, HLSL_OP3_MOVC, HLSL_OP3_TERNARY, }; @@ -577,6 +603,10 @@ enum hlsl_ir_jump_type HLSL_IR_JUMP_DISCARD_NEG, HLSL_IR_JUMP_DISCARD_NZ, HLSL_IR_JUMP_RETURN, + /* UNRESOLVED_CONTINUE type is used by the parser when 'continue' statement is found, + it never reaches code generation, and is resolved to CONTINUE type once iteration + and loop exit logic was properly applied. */ + HLSL_IR_JUMP_UNRESOLVED_CONTINUE, }; struct hlsl_ir_jump @@ -615,17 +645,25 @@ struct hlsl_deref unsigned int path_len; struct hlsl_src *path; - /* Single instruction node of data type uint used to represent the register offset (in register - * components, within the pertaining regset), from the start of the variable, of the part - * referenced. - * The path is lowered to this single offset -- whose value may vary between SM1 and SM4 -- - * before writing the bytecode. + /* Before writing the bytecode, deref paths are lowered into an offset (within the pertaining + * regset) from the start of the variable, to the part of the variable that is referenced. + * This offset is stored using two fields, one for a variable part and other for a constant + * part, which are added together: + * - rel_offset: An offset given by an instruction node, in whole registers. + * - const_offset: A constant number of register components. * Since the type information cannot longer be retrieved from the offset alone, the type is - * stored in the data_type field. */ - struct hlsl_src offset; + * stored in the data_type field, which remains NULL if the deref hasn't been lowered yet. */ + struct hlsl_src rel_offset; + unsigned int const_offset; struct hlsl_type *data_type; }; +/* Whether the path has been lowered to an offset or not. */ +static inline bool hlsl_deref_is_lowered(const struct hlsl_deref *deref) +{ + return !!deref->data_type; +} + struct hlsl_ir_load { struct hlsl_ir_node node; @@ -641,6 +679,7 @@ enum hlsl_resource_load_type HLSL_RESOURCE_SAMPLE_LOD, HLSL_RESOURCE_SAMPLE_LOD_BIAS, HLSL_RESOURCE_SAMPLE_GRAD, + HLSL_RESOURCE_SAMPLE_PROJ, HLSL_RESOURCE_GATHER_RED, HLSL_RESOURCE_GATHER_GREEN, HLSL_RESOURCE_GATHER_BLUE, @@ -702,6 +741,10 @@ struct hlsl_scope struct rb_tree types; /* Scope containing this scope. This value is NULL for the global scope. */ struct hlsl_scope *upper; + /* The scope was created for the loop statement. */ + bool loop; + /* The scope was created for the switch statement. */ + bool _switch; }; struct hlsl_profile_info @@ -854,13 +897,8 @@ struct hlsl_ctx uint32_t in_state_block : 1; /* Whether the numthreads() attribute has been provided in the entry-point function. */ uint32_t found_numthreads : 1; -}; -enum hlsl_error_level -{ - HLSL_LEVEL_ERROR = 0, - HLSL_LEVEL_WARNING, - HLSL_LEVEL_NOTE, + bool semantic_compat_mapping; }; struct hlsl_resource_load_params @@ -944,6 +982,12 @@ static inline struct hlsl_ir_index *hlsl_ir_index(const struct hlsl_ir_node *nod return CONTAINING_RECORD(node, struct hlsl_ir_index, node); } +static inline struct hlsl_ir_switch *hlsl_ir_switch(const struct hlsl_ir_node *node) +{ + assert(node->type == HLSL_IR_SWITCH); + return CONTAINING_RECORD(node, struct hlsl_ir_switch, node); +} + static inline void hlsl_block_init(struct hlsl_block *block) { list_init(&block->instrs); @@ -1061,6 +1105,11 @@ static inline struct hlsl_type *hlsl_get_numeric_type(const struct hlsl_ctx *ctx return hlsl_get_matrix_type(ctx, base_type, dimx, dimy); } +static inline bool hlsl_is_numeric_type(const struct hlsl_type *type) +{ + return type->class <= HLSL_CLASS_LAST_NUMERIC; +} + static inline unsigned int hlsl_sampler_dim_count(enum hlsl_sampler_dim dim) { switch (dim) @@ -1117,6 +1166,9 @@ bool hlsl_copy_deref(struct hlsl_ctx *ctx, struct hlsl_deref *deref, const struc void hlsl_cleanup_deref(struct hlsl_deref *deref); void hlsl_cleanup_semantic(struct hlsl_semantic *semantic); +void hlsl_cleanup_ir_switch_cases(struct list *cases); +void hlsl_free_ir_switch_case(struct hlsl_ir_switch_case *c); + void hlsl_replace_node(struct hlsl_ir_node *old, struct hlsl_ir_node *new); void hlsl_free_attribute(struct hlsl_attribute *attr); @@ -1126,7 +1178,10 @@ void hlsl_free_type(struct hlsl_type *type); void hlsl_free_var(struct hlsl_ir_var *decl); struct hlsl_ir_function *hlsl_get_function(struct hlsl_ctx *ctx, const char *name); -struct hlsl_ir_function_decl *hlsl_get_func_decl(struct hlsl_ctx *ctx, const char *name); +struct hlsl_ir_function_decl *hlsl_get_first_func_decl(struct hlsl_ctx *ctx, const char *name); +struct hlsl_ir_function_decl *hlsl_get_func_decl(struct hlsl_ctx *ctx, const char *name, + const struct hlsl_func_parameters *parameters); +const struct hlsl_profile_info *hlsl_get_target_info(const char *target); struct hlsl_type *hlsl_get_type(struct hlsl_scope *scope, const char *name, bool recursive, bool case_insensitive); struct hlsl_ir_var *hlsl_get_var(struct hlsl_scope *scope, const char *name); @@ -1210,6 +1265,10 @@ struct hlsl_ir_node *hlsl_new_unary_expr(struct hlsl_ctx *ctx, enum hlsl_ir_expr struct hlsl_ir_var *hlsl_new_var(struct hlsl_ctx *ctx, const char *name, struct hlsl_type *type, const struct vkd3d_shader_location *loc, const struct hlsl_semantic *semantic, unsigned int modifiers, const struct hlsl_reg_reservation *reg_reservation); +struct hlsl_ir_switch_case *hlsl_new_switch_case(struct hlsl_ctx *ctx, unsigned int value, bool is_default, + struct hlsl_block *body, const struct vkd3d_shader_location *loc); +struct hlsl_ir_node *hlsl_new_switch(struct hlsl_ctx *ctx, struct hlsl_ir_node *selector, + struct list *cases, const struct vkd3d_shader_location *loc); void hlsl_error(struct hlsl_ctx *ctx, const struct vkd3d_shader_location *loc, enum vkd3d_shader_error error, const char *fmt, ...) VKD3D_PRINTF_FUNC(4, 5); @@ -1232,13 +1291,12 @@ unsigned int hlsl_type_get_array_element_reg_size(const struct hlsl_type *type, struct hlsl_type *hlsl_type_get_component_type(struct hlsl_ctx *ctx, struct hlsl_type *type, unsigned int index); unsigned int hlsl_type_get_component_offset(struct hlsl_ctx *ctx, struct hlsl_type *type, - enum hlsl_regset regset, unsigned int index); + unsigned int index, enum hlsl_regset *regset); bool hlsl_type_is_row_major(const struct hlsl_type *type); unsigned int hlsl_type_minor_size(const struct hlsl_type *type); unsigned int hlsl_type_major_size(const struct hlsl_type *type); unsigned int hlsl_type_element_count(const struct hlsl_type *type); bool hlsl_type_is_resource(const struct hlsl_type *type); -enum hlsl_regset hlsl_type_get_regset(const struct hlsl_type *type); unsigned int hlsl_type_get_sm4_offset(const struct hlsl_type *type, unsigned int offset); bool hlsl_types_are_equal(const struct hlsl_type *t1, const struct hlsl_type *t2); @@ -1251,6 +1309,7 @@ unsigned int hlsl_map_swizzle(unsigned int swizzle, unsigned int writemask); unsigned int hlsl_swizzle_from_writemask(unsigned int writemask); struct hlsl_type *hlsl_deref_get_type(struct hlsl_ctx *ctx, const struct hlsl_deref *deref); +enum hlsl_regset hlsl_deref_get_regset(struct hlsl_ctx *ctx, const struct hlsl_deref *deref); bool hlsl_component_index_range_from_deref(struct hlsl_ctx *ctx, const struct hlsl_deref *deref, unsigned int *start, unsigned int *count); bool hlsl_regset_index_from_deref(struct hlsl_ctx *ctx, const struct hlsl_deref *deref, @@ -1273,7 +1332,7 @@ int hlsl_sm1_write(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry_fun bool hlsl_sm4_usage_from_semantic(struct hlsl_ctx *ctx, const struct hlsl_semantic *semantic, bool output, D3D_NAME *usage); bool hlsl_sm4_register_from_semantic(struct hlsl_ctx *ctx, const struct hlsl_semantic *semantic, - bool output, enum vkd3d_shader_register_type *type, enum vkd3d_sm4_swizzle_type *swizzle_type, bool *has_idx); + bool output, enum vkd3d_shader_register_type *type, bool *has_idx); int hlsl_sm4_write(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry_func, struct vkd3d_shader_code *out); struct hlsl_ir_function_decl *hlsl_compile_internal_function(struct hlsl_ctx *ctx, const char *name, const char *hlsl); diff --git a/libs/vkd3d/libs/vkd3d-shader/hlsl.l b/libs/vkd3d/libs/vkd3d-shader/hlsl.l index e9ae3ccf3d3..401fba60422 100644 --- a/libs/vkd3d/libs/vkd3d-shader/hlsl.l +++ b/libs/vkd3d/libs/vkd3d-shader/hlsl.l @@ -46,14 +46,13 @@ static void update_location(struct hlsl_ctx *ctx, YYLTYPE *loc); %x pp pp_line pp_pragma pp_ignore -RESERVED1 auto|case|catch|char|class|const_cast|default|delete|dynamic_cast|enum +RESERVED1 auto|catch|char|class|const_cast|delete|dynamic_cast|enum RESERVED2 explicit|friend|goto|long|mutable|new|operator|private|protected|public RESERVED3 reinterpret_cast|short|signed|sizeof|static_cast|template|this|throw|try RESERVED4 typename|union|unsigned|using|virtual WS [ \t] NEWLINE (\n)|(\r\n) -DOUBLESLASHCOMMENT "//"[^\n]* STRING \"[^\"]*\" IDENTIFIER [A-Za-z_][A-Za-z0-9_]* @@ -73,12 +72,16 @@ ANY (.) BlendState {return KW_BLENDSTATE; } break {return KW_BREAK; } Buffer {return KW_BUFFER; } +case {return KW_CASE; } cbuffer {return KW_CBUFFER; } +centroid {return KW_CENTROID; } +column_major {return KW_COLUMN_MAJOR; } compile {return KW_COMPILE; } const {return KW_CONST; } continue {return KW_CONTINUE; } DepthStencilState {return KW_DEPTHSTENCILSTATE; } DepthStencilView {return KW_DEPTHSTENCILVIEW; } +default {return KW_DEFAULT; } discard {return KW_DISCARD; } do {return KW_DO; } double {return KW_DOUBLE; } @@ -95,28 +98,32 @@ inout {return KW_INOUT; } matrix {return KW_MATRIX; } namespace {return KW_NAMESPACE; } nointerpolation {return KW_NOINTERPOLATION; } +noperspective {return KW_NOPERSPECTIVE; } out {return KW_OUT; } packoffset {return KW_PACKOFFSET; } pass {return KW_PASS; } PixelShader {return KW_PIXELSHADER; } precise {return KW_PRECISE; } RasterizerState {return KW_RASTERIZERSTATE; } +register {return KW_REGISTER; } RenderTargetView {return KW_RENDERTARGETVIEW; } return {return KW_RETURN; } -register {return KW_REGISTER; } +row_major {return KW_ROW_MAJOR; } RWBuffer {return KW_RWBUFFER; } RWStructuredBuffer {return KW_RWSTRUCTUREDBUFFER; } RWTexture1D {return KW_RWTEXTURE1D; } +RWTexture1DArray {return KW_RWTEXTURE1DARRAY; } RWTexture2D {return KW_RWTEXTURE2D; } +RWTexture2DArray {return KW_RWTEXTURE2DARRAY; } RWTexture3D {return KW_RWTEXTURE3D; } sampler {return KW_SAMPLER; } sampler1D {return KW_SAMPLER1D; } sampler2D {return KW_SAMPLER2D; } sampler3D {return KW_SAMPLER3D; } -samplerCUBE {return KW_SAMPLERCUBE; } -sampler_state {return KW_SAMPLER_STATE; } SamplerComparisonState {return KW_SAMPLERCOMPARISONSTATE;} +samplerCUBE {return KW_SAMPLERCUBE; } SamplerState {return KW_SAMPLER; } +sampler_state {return KW_SAMPLER_STATE; } shared {return KW_SHARED; } stateblock {return KW_STATEBLOCK; } stateblock_state {return KW_STATEBLOCK_STATE; } @@ -125,21 +132,22 @@ string {return KW_STRING; } struct {return KW_STRUCT; } switch {return KW_SWITCH; } tbuffer {return KW_TBUFFER; } -technique {return KW_TECHNIQUE; } +(?i:technique) {return KW_TECHNIQUE; } technique10 {return KW_TECHNIQUE10; } +technique11 {return KW_TECHNIQUE11; } texture {return KW_TEXTURE; } -texture1D {return KW_TEXTURE1D; } Texture1D {return KW_TEXTURE1D; } +texture1D {return KW_TEXTURE1D; } Texture1DArray {return KW_TEXTURE1DARRAY; } -texture2D {return KW_TEXTURE2D; } Texture2D {return KW_TEXTURE2D; } +texture2D {return KW_TEXTURE2D; } Texture2DArray {return KW_TEXTURE2DARRAY; } Texture2DMS {return KW_TEXTURE2DMS; } Texture2DMSArray {return KW_TEXTURE2DMSARRAY; } -texture3D {return KW_TEXTURE3D; } Texture3D {return KW_TEXTURE3D; } -textureCUBE {return KW_TEXTURECUBE; } +texture3D {return KW_TEXTURE3D; } TextureCube {return KW_TEXTURECUBE; } +textureCUBE {return KW_TEXTURECUBE; } TextureCubeArray {return KW_TEXTURECUBEARRAY; } true {return KW_TRUE; } typedef {return KW_TYPEDEF; } @@ -159,7 +167,6 @@ while {return KW_WHILE; } \<\<= {return OP_LEFTSHIFTASSIGN; } \>\> {return OP_RIGHTSHIFT; } \>\>= {return OP_RIGHTSHIFTASSIGN; } -\.\.\. {return OP_ELLIPSIS; } \<= {return OP_LE; } \>= {return OP_GE; } != {return OP_NE; } @@ -171,13 +178,6 @@ while {return KW_WHILE; } &= {return OP_ANDASSIGN; } \|= {return OP_ORASSIGN; } ^= {return OP_XORASSIGN; } -## {return OP_UNKNOWN1; } -#@ {return OP_UNKNOWN2; } -:: {return OP_UNKNOWN3; } -\-\> {return OP_UNKNOWN4; } - -column_major {return KW_COLUMN_MAJOR; } -row_major {return KW_ROW_MAJOR; } {IDENTIFIER} { struct hlsl_ctx *ctx = yyget_extra(yyscanner); @@ -203,20 +203,30 @@ row_major {return KW_ROW_MAJOR; } yylval->floatval = atof(yytext); return C_FLOAT; } -0x[0-9a-fA-F]+ { +0x[0-9a-fA-F]+[lL]? { yylval->intval = vkd3d_parse_integer(yytext); return C_INTEGER; } -0[0-7]+ { +0[0-7]+[lL]? { yylval->intval = vkd3d_parse_integer(yytext); return C_INTEGER; } -[0-9]+ { +[0-9]+[lL]? { yylval->intval = vkd3d_parse_integer(yytext); return C_INTEGER; } - -{DOUBLESLASHCOMMENT} {} +0x[0-9a-fA-F]+([uU]|[uU][lL]|[lL][uU]) { + yylval->intval = vkd3d_parse_integer(yytext); + return C_UNSIGNED; + } +0[0-7]+([uU]|[uU][lL]|[lL][uU]) { + yylval->intval = vkd3d_parse_integer(yytext); + return C_UNSIGNED; + } +[0-9]+([uU]|[uU][lL]|[lL][uU]) { + yylval->intval = vkd3d_parse_integer(yytext); + return C_UNSIGNED; + } {WS}+ {} {NEWLINE} { diff --git a/libs/vkd3d/libs/vkd3d-shader/hlsl.y b/libs/vkd3d/libs/vkd3d-shader/hlsl.y index fb6d485ea69..67b01293683 100644 --- a/libs/vkd3d/libs/vkd3d-shader/hlsl.y +++ b/libs/vkd3d/libs/vkd3d-shader/hlsl.y @@ -162,6 +162,12 @@ static void destroy_block(struct hlsl_block *block) vkd3d_free(block); } +static void destroy_switch_cases(struct list *cases) +{ + hlsl_cleanup_ir_switch_cases(cases); + vkd3d_free(cases); +} + static bool hlsl_types_are_componentwise_compatible(struct hlsl_ctx *ctx, struct hlsl_type *src, struct hlsl_type *dst) { @@ -205,7 +211,7 @@ static bool hlsl_types_are_componentwise_equal(struct hlsl_ctx *ctx, struct hlsl return true; } -static bool type_contains_only_numerics(struct hlsl_type *type) +static bool type_contains_only_numerics(const struct hlsl_type *type) { unsigned int i; @@ -220,12 +226,12 @@ static bool type_contains_only_numerics(struct hlsl_type *type) } return true; } - return type->class <= HLSL_CLASS_LAST_NUMERIC; + return hlsl_is_numeric_type(type); } static bool explicit_compatible_data_types(struct hlsl_ctx *ctx, struct hlsl_type *src, struct hlsl_type *dst) { - if (src->class <= HLSL_CLASS_LAST_NUMERIC && src->dimx == 1 && src->dimy == 1 && type_contains_only_numerics(dst)) + if (hlsl_is_numeric_type(src) && src->dimx == 1 && src->dimy == 1 && type_contains_only_numerics(dst)) return true; if (src->class == HLSL_CLASS_MATRIX && dst->class == HLSL_CLASS_MATRIX @@ -245,10 +251,10 @@ static bool explicit_compatible_data_types(struct hlsl_ctx *ctx, struct hlsl_typ static bool implicit_compatible_data_types(struct hlsl_ctx *ctx, struct hlsl_type *src, struct hlsl_type *dst) { - if ((src->class <= HLSL_CLASS_LAST_NUMERIC) != (dst->class <= HLSL_CLASS_LAST_NUMERIC)) + if (hlsl_is_numeric_type(src) != hlsl_is_numeric_type(dst)) return false; - if (src->class <= HLSL_CLASS_LAST_NUMERIC) + if (hlsl_is_numeric_type(src)) { /* Scalar vars can be converted to any other numeric data type */ if (src->dimx == 1 && src->dimy == 1) @@ -305,7 +311,7 @@ static struct hlsl_ir_node *add_cast(struct hlsl_ctx *ctx, struct hlsl_block *bl struct hlsl_ir_var *var; unsigned int dst_idx; - broadcast = src_type->class <= HLSL_CLASS_LAST_NUMERIC && src_type->dimx == 1 && src_type->dimy == 1; + broadcast = hlsl_is_numeric_type(src_type) && src_type->dimx == 1 && src_type->dimy == 1; matrix_cast = !broadcast && dst_comp_count != src_comp_count && src_type->class == HLSL_CLASS_MATRIX && dst_type->class == HLSL_CLASS_MATRIX; assert(src_comp_count >= dst_comp_count || broadcast); @@ -464,6 +470,72 @@ static bool attribute_list_has_duplicates(const struct parse_attribute_list *att return false; } +static void resolve_loop_continue(struct hlsl_ctx *ctx, struct hlsl_block *block, enum loop_type type, + struct hlsl_block *cond, struct hlsl_block *iter) +{ + struct hlsl_ir_node *instr, *next; + + LIST_FOR_EACH_ENTRY_SAFE(instr, next, &block->instrs, struct hlsl_ir_node, entry) + { + if (instr->type == HLSL_IR_IF) + { + struct hlsl_ir_if *iff = hlsl_ir_if(instr); + + resolve_loop_continue(ctx, &iff->then_block, type, cond, iter); + resolve_loop_continue(ctx, &iff->else_block, type, cond, iter); + } + else if (instr->type == HLSL_IR_JUMP) + { + struct hlsl_ir_jump *jump = hlsl_ir_jump(instr); + struct hlsl_block block; + + if (jump->type != HLSL_IR_JUMP_UNRESOLVED_CONTINUE) + continue; + + if (type == LOOP_DO_WHILE) + { + if (!hlsl_clone_block(ctx, &block, cond)) + return; + if (!append_conditional_break(ctx, &block)) + { + hlsl_block_cleanup(&block); + return; + } + list_move_before(&instr->entry, &block.instrs); + } + else if (type == LOOP_FOR) + { + if (!hlsl_clone_block(ctx, &block, iter)) + return; + list_move_before(&instr->entry, &block.instrs); + } + jump->type = HLSL_IR_JUMP_CONTINUE; + } + } +} + +static void check_loop_attributes(struct hlsl_ctx *ctx, const struct parse_attribute_list *attributes, + const struct vkd3d_shader_location *loc) +{ + bool has_unroll = false, has_loop = false, has_fastopt = false; + unsigned int i; + + for (i = 0; i < attributes->count; ++i) + { + const char *name = attributes->attrs[i]->name; + + has_loop |= !strcmp(name, "loop"); + has_unroll |= !strcmp(name, "unroll"); + has_fastopt |= !strcmp(name, "fastopt"); + } + + if (has_unroll && has_loop) + hlsl_error(ctx, loc, VKD3D_SHADER_ERROR_HLSL_INVALID_SYNTAX, "Unroll attribute can't be used with 'loop' attribute."); + + if (has_unroll && has_fastopt) + hlsl_error(ctx, loc, VKD3D_SHADER_ERROR_HLSL_INVALID_SYNTAX, "Unroll attribute can't be used with 'fastopt' attribute."); +} + static struct hlsl_block *create_loop(struct hlsl_ctx *ctx, enum loop_type type, const struct parse_attribute_list *attributes, struct hlsl_block *init, struct hlsl_block *cond, struct hlsl_block *iter, struct hlsl_block *body, const struct vkd3d_shader_location *loc) @@ -474,6 +546,8 @@ static struct hlsl_block *create_loop(struct hlsl_ctx *ctx, enum loop_type type, if (attribute_list_has_duplicates(attributes)) hlsl_error(ctx, loc, VKD3D_SHADER_ERROR_HLSL_INVALID_SYNTAX, "Found duplicate attribute."); + check_loop_attributes(ctx, attributes, loc); + /* Ignore unroll(0) attribute, and any invalid attribute. */ for (i = 0; i < attributes->count; ++i) { @@ -489,8 +563,11 @@ static struct hlsl_block *create_loop(struct hlsl_ctx *ctx, enum loop_type type, hlsl_warning(ctx, loc, VKD3D_SHADER_ERROR_HLSL_NOT_IMPLEMENTED, "Loop unrolling is not implemented."); } } - else if (!strcmp(attr->name, "loop") - || !strcmp(attr->name, "fastopt") + else if (!strcmp(attr->name, "loop")) + { + /* TODO: this attribute will be used to disable unrolling, once it's implememented. */ + } + else if (!strcmp(attr->name, "fastopt") || !strcmp(attr->name, "allow_uav_condition")) { hlsl_fixme(ctx, loc, "Unhandled attribute '%s'.", attr->name); @@ -501,6 +578,8 @@ static struct hlsl_block *create_loop(struct hlsl_ctx *ctx, enum loop_type type, } } + resolve_loop_continue(ctx, body, type, cond, iter); + if (!init && !(init = make_empty_block(ctx))) goto oom; @@ -1079,22 +1158,6 @@ static struct hlsl_reg_reservation parse_packoffset(struct hlsl_ctx *ctx, const return reservation; } -static struct hlsl_ir_function_decl *get_func_decl(struct rb_tree *funcs, - const char *name, const struct hlsl_func_parameters *parameters) -{ - struct hlsl_ir_function *func; - struct rb_entry *entry; - - if ((entry = rb_get(funcs, name))) - { - func = RB_ENTRY_VALUE(entry, struct hlsl_ir_function, entry); - - if ((entry = rb_get(&func->overloads, parameters))) - return RB_ENTRY_VALUE(entry, struct hlsl_ir_function_decl, entry); - } - return NULL; -} - static struct hlsl_block *make_block(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr) { struct hlsl_block *block; @@ -1134,6 +1197,7 @@ static unsigned int evaluate_static_expression_as_uint(struct hlsl_ctx *ctx, str case HLSL_IR_RESOURCE_LOAD: case HLSL_IR_RESOURCE_STORE: case HLSL_IR_STORE: + case HLSL_IR_SWITCH: hlsl_error(ctx, &node->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_SYNTAX, "Expected literal expression."); } @@ -1228,7 +1292,7 @@ static enum hlsl_base_type expr_common_base_type(enum hlsl_base_type t1, enum hl static bool expr_common_shape(struct hlsl_ctx *ctx, struct hlsl_type *t1, struct hlsl_type *t2, const struct vkd3d_shader_location *loc, enum hlsl_type_class *type, unsigned int *dimx, unsigned int *dimy) { - if (t1->class > HLSL_CLASS_LAST_NUMERIC) + if (!hlsl_is_numeric_type(t1)) { struct vkd3d_string_buffer *string; @@ -1239,7 +1303,7 @@ static bool expr_common_shape(struct hlsl_ctx *ctx, struct hlsl_type *t1, struct return false; } - if (t2->class > HLSL_CLASS_LAST_NUMERIC) + if (!hlsl_is_numeric_type(t2)) { struct vkd3d_string_buffer *string; @@ -1711,7 +1775,7 @@ static struct hlsl_ir_node *add_assignment(struct hlsl_ctx *ctx, struct hlsl_blo return NULL; } - if (lhs_type->class <= HLSL_CLASS_LAST_NUMERIC) + if (hlsl_is_numeric_type(lhs_type)) writemask = (1 << lhs_type->dimx) - 1; if (!(rhs = add_implicit_conversion(ctx, block, rhs, lhs_type, &rhs->loc))) @@ -1941,7 +2005,7 @@ static bool type_has_object_components(struct hlsl_type *type, bool must_be_in_s static bool type_has_numeric_components(struct hlsl_type *type) { - if (type->class <= HLSL_CLASS_LAST_NUMERIC) + if (hlsl_is_numeric_type(type)) return true; if (type->class == HLSL_CLASS_ARRAY) return type_has_numeric_components(type->e.array.type); @@ -2106,7 +2170,7 @@ static void declare_var(struct hlsl_ctx *ctx, struct parse_variable_def *v) "Target profile doesn't support objects as struct members in uniform variables."); } - if ((func = hlsl_get_func_decl(ctx, var->name))) + if ((func = hlsl_get_first_func_decl(ctx, var->name))) { hlsl_error(ctx, &var->loc, VKD3D_SHADER_ERROR_HLSL_REDEFINED, "'%s' is already defined as a function.", var->name); @@ -2261,56 +2325,27 @@ static struct hlsl_block *initialize_vars(struct hlsl_ctx *ctx, struct list *var return initializers; } -struct find_function_call_args -{ - struct hlsl_ctx *ctx; - const struct parse_initializer *params; - struct hlsl_ir_function_decl *decl; - unsigned int compatible_overloads_count; -}; - -static void find_function_call_exact(struct rb_entry *entry, void *context) -{ - struct hlsl_ir_function_decl *decl = RB_ENTRY_VALUE(entry, struct hlsl_ir_function_decl, entry); - struct find_function_call_args *args = context; - unsigned int i; - - if (decl->parameters.count != args->params->args_count) - return; - - for (i = 0; i < decl->parameters.count; ++i) - { - if (!hlsl_types_are_equal(decl->parameters.vars[i]->data_type, args->params->args[i]->data_type)) - return; - } - args->decl = decl; -} - -static void find_function_call_compatible(struct rb_entry *entry, void *context) +static bool func_is_compatible_match(struct hlsl_ctx *ctx, + const struct hlsl_ir_function_decl *decl, const struct parse_initializer *args) { - struct hlsl_ir_function_decl *decl = RB_ENTRY_VALUE(entry, struct hlsl_ir_function_decl, entry); - struct find_function_call_args *args = context; unsigned int i; - if (decl->parameters.count != args->params->args_count) - return; + if (decl->parameters.count != args->args_count) + return false; for (i = 0; i < decl->parameters.count; ++i) { - if (!implicit_compatible_data_types(args->ctx, args->params->args[i]->data_type, - decl->parameters.vars[i]->data_type)) - return; + if (!implicit_compatible_data_types(ctx, args->args[i]->data_type, decl->parameters.vars[i]->data_type)) + return false; } - - args->compatible_overloads_count++; - args->decl = decl; + return true; } static struct hlsl_ir_function_decl *find_function_call(struct hlsl_ctx *ctx, - const char *name, const struct parse_initializer *params, + const char *name, const struct parse_initializer *args, const struct vkd3d_shader_location *loc) { - struct find_function_call_args args = {.ctx = ctx, .params = params}; + struct hlsl_ir_function_decl *decl, *compatible_match = NULL; struct hlsl_ir_function *func; struct rb_entry *entry; @@ -2318,16 +2353,20 @@ static struct hlsl_ir_function_decl *find_function_call(struct hlsl_ctx *ctx, return NULL; func = RB_ENTRY_VALUE(entry, struct hlsl_ir_function, entry); - rb_for_each_entry(&func->overloads, find_function_call_exact, &args); - if (!args.decl) + LIST_FOR_EACH_ENTRY(decl, &func->overloads, struct hlsl_ir_function_decl, entry) { - rb_for_each_entry(&func->overloads, find_function_call_compatible, &args); - if (args.compatible_overloads_count > 1) + if (func_is_compatible_match(ctx, decl, args)) { - hlsl_fixme(ctx, loc, "Prioritize between multiple compatible function overloads."); + if (compatible_match) + { + hlsl_fixme(ctx, loc, "Prioritize between multiple compatible function overloads."); + break; + } + compatible_match = decl; } } - return args.decl; + + return compatible_match; } static struct hlsl_ir_node *hlsl_new_void_expr(struct hlsl_ctx *ctx, const struct vkd3d_shader_location *loc) @@ -2669,6 +2708,17 @@ static bool intrinsic_asuint(struct hlsl_ctx *ctx, return add_expr(ctx, params->instrs, HLSL_OP1_REINTERPRET, operands, data_type, loc); } +static bool intrinsic_ceil(struct hlsl_ctx *ctx, + const struct parse_initializer *params, const struct vkd3d_shader_location *loc) +{ + struct hlsl_ir_node *arg; + + if (!(arg = intrinsic_float_convert_arg(ctx, params, params->args[0], loc))) + return false; + + return !!add_unary_arithmetic_expr(ctx, params->instrs, HLSL_OP1_CEIL, arg, loc); +} + static bool intrinsic_clamp(struct hlsl_ctx *ctx, const struct parse_initializer *params, const struct vkd3d_shader_location *loc) { @@ -2828,6 +2878,22 @@ static bool intrinsic_ddy_coarse(struct hlsl_ctx *ctx, return !!add_unary_arithmetic_expr(ctx, params->instrs, HLSL_OP1_DSY_COARSE, arg, loc); } +static bool intrinsic_degrees(struct hlsl_ctx *ctx, + const struct parse_initializer *params, const struct vkd3d_shader_location *loc) +{ + struct hlsl_ir_node *arg, *deg; + + if (!(arg = intrinsic_float_convert_arg(ctx, params, params->args[0], loc))) + return false; + + /* 1 rad = 180/pi degree = 57.2957795 degree */ + if (!(deg = hlsl_new_float_constant(ctx, 57.2957795f, loc))) + return false; + hlsl_block_add_instr(params->instrs, deg); + + return !!add_binary_arithmetic_expr(ctx, params->instrs, HLSL_OP2_MUL, arg, deg, loc); +} + static bool intrinsic_ddy_fine(struct hlsl_ctx *ctx, const struct parse_initializer *params, const struct vkd3d_shader_location *loc) { @@ -2961,6 +3027,33 @@ static bool intrinsic_frac(struct hlsl_ctx *ctx, return !!add_unary_arithmetic_expr(ctx, params->instrs, HLSL_OP1_FRACT, arg, loc); } +static bool intrinsic_fwidth(struct hlsl_ctx *ctx, + const struct parse_initializer *params, const struct vkd3d_shader_location *loc) +{ + struct hlsl_ir_function_decl *func; + struct hlsl_type *type; + char *body; + + static const char template[] = + "%s fwidth(%s x)\n" + "{\n" + " return abs(ddx(x)) + abs(ddy(x));\n" + "}"; + + if (!elementwise_intrinsic_float_convert_args(ctx, params, loc)) + return false; + type = params->args[0]->data_type; + + if (!(body = hlsl_sprintf_alloc(ctx, template, type->name, type->name))) + return false; + func = hlsl_compile_internal_function(ctx, "fwidth", body); + vkd3d_free(body); + if (!func) + return false; + + return add_user_call(ctx, func, params, loc); +} + static bool intrinsic_ldexp(struct hlsl_ctx *ctx, const struct parse_initializer *params, const struct vkd3d_shader_location *loc) { @@ -3078,6 +3171,7 @@ static bool intrinsic_log(struct hlsl_ctx *ctx, /* ln(2) */ if (!(coeff = hlsl_new_float_constant(ctx, 0.69314718055f, loc))) return false; + hlsl_block_add_instr(params->instrs, coeff); return !!add_binary_arithmetic_expr(ctx, params->instrs, HLSL_OP2_MUL, log, coeff, loc); } @@ -3096,6 +3190,7 @@ static bool intrinsic_log10(struct hlsl_ctx *ctx, /* 1 / log2(10) */ if (!(coeff = hlsl_new_float_constant(ctx, 0.301029996f, loc))) return false; + hlsl_block_add_instr(params->instrs, coeff); return !!add_binary_arithmetic_expr(ctx, params->instrs, HLSL_OP2_MUL, log, coeff, loc); } @@ -3264,6 +3359,22 @@ static bool intrinsic_pow(struct hlsl_ctx *ctx, return !!add_pow_expr(ctx, params->instrs, params->args[0], params->args[1], loc); } +static bool intrinsic_radians(struct hlsl_ctx *ctx, + const struct parse_initializer *params, const struct vkd3d_shader_location *loc) +{ + struct hlsl_ir_node *arg, *rad; + + if (!(arg = intrinsic_float_convert_arg(ctx, params, params->args[0], loc))) + return false; + + /* 1 degree = pi/180 rad = 0.0174532925f rad */ + if (!(rad = hlsl_new_float_constant(ctx, 0.0174532925f, loc))) + return false; + hlsl_block_add_instr(params->instrs, rad); + + return !!add_binary_arithmetic_expr(ctx, params->instrs, HLSL_OP2_MUL, arg, rad, loc); +} + static bool intrinsic_reflect(struct hlsl_ctx *ctx, const struct parse_initializer *params, const struct vkd3d_shader_location *loc) { @@ -3423,10 +3534,24 @@ static bool intrinsic_step(struct hlsl_ctx *ctx, return !!add_implicit_conversion(ctx, params->instrs, ge, type, loc); } +static bool intrinsic_tan(struct hlsl_ctx *ctx, + const struct parse_initializer *params, const struct vkd3d_shader_location *loc) +{ + struct hlsl_ir_node *arg = params->args[0], *sin, *cos; + + if (!(sin = add_unary_arithmetic_expr(ctx, params->instrs, HLSL_OP1_SIN, arg, loc))) + return false; + + if (!(cos = add_unary_arithmetic_expr(ctx, params->instrs, HLSL_OP1_COS, arg, loc))) + return false; + + return !!add_binary_arithmetic_expr(ctx, params->instrs, HLSL_OP2_DIV, sin, cos, loc); +} + static bool intrinsic_tex(struct hlsl_ctx *ctx, const struct parse_initializer *params, const struct vkd3d_shader_location *loc, const char *name, enum hlsl_sampler_dim dim) { - struct hlsl_resource_load_params load_params = {.type = HLSL_RESOURCE_SAMPLE}; + struct hlsl_resource_load_params load_params = { 0 }; const struct hlsl_type *sampler_type; struct hlsl_ir_node *coords, *load; @@ -3455,10 +3580,74 @@ static bool intrinsic_tex(struct hlsl_ctx *ctx, const struct parse_initializer * hlsl_release_string_buffer(ctx, string); } - if (!(coords = add_implicit_conversion(ctx, params->instrs, params->args[1], - hlsl_get_vector_type(ctx, HLSL_TYPE_FLOAT, hlsl_sampler_dim_count(dim)), loc))) + if (!strcmp(name, "tex2Dlod")) { - return false; + struct hlsl_ir_node *lod, *c; + + load_params.type = HLSL_RESOURCE_SAMPLE_LOD; + + if (!(c = hlsl_new_swizzle(ctx, HLSL_SWIZZLE(X, Y, Z, W), hlsl_sampler_dim_count(dim), params->args[1], loc))) + return false; + hlsl_block_add_instr(params->instrs, c); + + if (!(coords = add_implicit_conversion(ctx, params->instrs, c, hlsl_get_vector_type(ctx, HLSL_TYPE_FLOAT, + hlsl_sampler_dim_count(dim)), loc))) + { + return false; + } + + if (!(lod = hlsl_new_swizzle(ctx, HLSL_SWIZZLE(W, W, W, W), 1, params->args[1], loc))) + return false; + hlsl_block_add_instr(params->instrs, lod); + + if (!(load_params.lod = add_implicit_conversion(ctx, params->instrs, lod, + hlsl_get_scalar_type(ctx, HLSL_TYPE_FLOAT), loc))) + { + return false; + } + } + else if (!strcmp(name, "tex2Dproj") + || !strcmp(name, "tex3Dproj") + || !strcmp(name, "texCUBEproj")) + { + if (!(coords = add_implicit_conversion(ctx, params->instrs, params->args[1], + hlsl_get_vector_type(ctx, HLSL_TYPE_FLOAT, 4), loc))) + { + return false; + } + + if (shader_profile_version_ge(ctx, 4, 0)) + { + unsigned int count = hlsl_sampler_dim_count(dim); + struct hlsl_ir_node *divisor; + + if (!(divisor = hlsl_new_swizzle(ctx, HLSL_SWIZZLE(W, W, W, W), count, coords, loc))) + return false; + hlsl_block_add_instr(params->instrs, divisor); + + if (!(coords = hlsl_new_swizzle(ctx, HLSL_SWIZZLE(X, Y, Z, W), count, coords, loc))) + return false; + hlsl_block_add_instr(params->instrs, coords); + + if (!(coords = add_binary_arithmetic_expr(ctx, params->instrs, HLSL_OP2_DIV, coords, divisor, loc))) + return false; + + load_params.type = HLSL_RESOURCE_SAMPLE; + } + else + { + load_params.type = HLSL_RESOURCE_SAMPLE_PROJ; + } + } + else + { + load_params.type = HLSL_RESOURCE_SAMPLE; + + if (!(coords = add_implicit_conversion(ctx, params->instrs, params->args[1], + hlsl_get_vector_type(ctx, HLSL_TYPE_FLOAT, hlsl_sampler_dim_count(dim)), loc))) + { + return false; + } } /* tex1D() functions never produce 1D resource declarations. For newer profiles half offset @@ -3517,18 +3706,42 @@ static bool intrinsic_tex2D(struct hlsl_ctx *ctx, return intrinsic_tex(ctx, params, loc, "tex2D", HLSL_SAMPLER_DIM_2D); } +static bool intrinsic_tex2Dlod(struct hlsl_ctx *ctx, + const struct parse_initializer *params, const struct vkd3d_shader_location *loc) +{ + return intrinsic_tex(ctx, params, loc, "tex2Dlod", HLSL_SAMPLER_DIM_2D); +} + +static bool intrinsic_tex2Dproj(struct hlsl_ctx *ctx, + const struct parse_initializer *params, const struct vkd3d_shader_location *loc) +{ + return intrinsic_tex(ctx, params, loc, "tex2Dproj", HLSL_SAMPLER_DIM_2D); +} + static bool intrinsic_tex3D(struct hlsl_ctx *ctx, const struct parse_initializer *params, const struct vkd3d_shader_location *loc) { return intrinsic_tex(ctx, params, loc, "tex3D", HLSL_SAMPLER_DIM_3D); } +static bool intrinsic_tex3Dproj(struct hlsl_ctx *ctx, + const struct parse_initializer *params, const struct vkd3d_shader_location *loc) +{ + return intrinsic_tex(ctx, params, loc, "tex3Dproj", HLSL_SAMPLER_DIM_3D); +} + static bool intrinsic_texCUBE(struct hlsl_ctx *ctx, const struct parse_initializer *params, const struct vkd3d_shader_location *loc) { return intrinsic_tex(ctx, params, loc, "texCUBE", HLSL_SAMPLER_DIM_CUBE); } +static bool intrinsic_texCUBEproj(struct hlsl_ctx *ctx, + const struct parse_initializer *params, const struct vkd3d_shader_location *loc) +{ + return intrinsic_tex(ctx, params, loc, "texCUBEproj", HLSL_SAMPLER_DIM_CUBE); +} + static bool intrinsic_transpose(struct hlsl_ctx *ctx, const struct parse_initializer *params, const struct vkd3d_shader_location *loc) { @@ -3659,6 +3872,7 @@ intrinsic_functions[] = {"any", 1, true, intrinsic_any}, {"asfloat", 1, true, intrinsic_asfloat}, {"asuint", -1, true, intrinsic_asuint}, + {"ceil", 1, true, intrinsic_ceil}, {"clamp", 3, true, intrinsic_clamp}, {"clip", 1, true, intrinsic_clip}, {"cos", 1, true, intrinsic_cos}, @@ -3669,6 +3883,7 @@ intrinsic_functions[] = {"ddy", 1, true, intrinsic_ddy}, {"ddy_coarse", 1, true, intrinsic_ddy_coarse}, {"ddy_fine", 1, true, intrinsic_ddy_fine}, + {"degrees", 1, true, intrinsic_degrees}, {"distance", 2, true, intrinsic_distance}, {"dot", 2, true, intrinsic_dot}, {"exp", 1, true, intrinsic_exp}, @@ -3676,6 +3891,7 @@ intrinsic_functions[] = {"floor", 1, true, intrinsic_floor}, {"fmod", 2, true, intrinsic_fmod}, {"frac", 1, true, intrinsic_frac}, + {"fwidth", 1, true, intrinsic_fwidth}, {"ldexp", 2, true, intrinsic_ldexp}, {"length", 1, true, intrinsic_length}, {"lerp", 3, true, intrinsic_lerp}, @@ -3688,6 +3904,7 @@ intrinsic_functions[] = {"mul", 2, true, intrinsic_mul}, {"normalize", 1, true, intrinsic_normalize}, {"pow", 2, true, intrinsic_pow}, + {"radians", 1, true, intrinsic_radians}, {"reflect", 2, true, intrinsic_reflect}, {"round", 1, true, intrinsic_round}, {"rsqrt", 1, true, intrinsic_rsqrt}, @@ -3697,10 +3914,15 @@ intrinsic_functions[] = {"smoothstep", 3, true, intrinsic_smoothstep}, {"sqrt", 1, true, intrinsic_sqrt}, {"step", 2, true, intrinsic_step}, + {"tan", 1, true, intrinsic_tan}, {"tex1D", -1, false, intrinsic_tex1D}, {"tex2D", -1, false, intrinsic_tex2D}, + {"tex2Dlod", 2, false, intrinsic_tex2Dlod}, + {"tex2Dproj", 2, false, intrinsic_tex2Dproj}, {"tex3D", -1, false, intrinsic_tex3D}, + {"tex3Dproj", 2, false, intrinsic_tex3Dproj}, {"texCUBE", -1, false, intrinsic_texCUBE}, + {"texCUBEproj", 2, false, intrinsic_texCUBEproj}, {"transpose", 1, true, intrinsic_transpose}, {"trunc", 1, true, intrinsic_trunc}, }; @@ -3740,7 +3962,7 @@ static struct hlsl_block *add_call(struct hlsl_ctx *ctx, const char *name, for (i = 0; i < args->args_count; ++i) { - if (args->args[i]->data_type->class > HLSL_CLASS_LAST_NUMERIC) + if (!hlsl_is_numeric_type(args->args[i]->data_type)) { struct vkd3d_string_buffer *string; @@ -4544,6 +4766,66 @@ static void validate_texture_format_type(struct hlsl_ctx *ctx, struct hlsl_type } } +static bool check_continue(struct hlsl_ctx *ctx, const struct hlsl_scope *scope, const struct vkd3d_shader_location *loc) +{ + if (scope->_switch) + { + hlsl_error(ctx, loc, VKD3D_SHADER_ERROR_HLSL_INVALID_SYNTAX, + "The 'continue' statement is not allowed in 'switch' statements."); + return false; + } + + if (scope->loop) + return true; + + if (scope->upper) + return check_continue(ctx, scope->upper, loc); + + hlsl_error(ctx, loc, VKD3D_SHADER_ERROR_HLSL_INVALID_SYNTAX, "The 'continue' statement is only allowed in loops."); + return false; +} + +static bool is_break_allowed(const struct hlsl_scope *scope) +{ + if (scope->loop || scope->_switch) + return true; + + return scope->upper ? is_break_allowed(scope->upper) : false; +} + +static void check_duplicated_switch_cases(struct hlsl_ctx *ctx, const struct hlsl_ir_switch_case *check, struct list *cases) +{ + struct hlsl_ir_switch_case *c; + bool found_duplicate = false; + + LIST_FOR_EACH_ENTRY(c, cases, struct hlsl_ir_switch_case, entry) + { + if (check->is_default) + { + if ((found_duplicate = c->is_default)) + { + hlsl_error(ctx, &check->loc, VKD3D_SHADER_ERROR_HLSL_DUPLICATE_SWITCH_CASE, + "Found multiple 'default' statements."); + hlsl_note(ctx, &c->loc, VKD3D_SHADER_LOG_ERROR, "The 'default' statement was previously found here."); + } + } + else + { + if (c->is_default) continue; + if ((found_duplicate = (c->value == check->value))) + { + hlsl_error(ctx, &check->loc, VKD3D_SHADER_ERROR_HLSL_DUPLICATE_SWITCH_CASE, + "Found duplicate 'case' statement."); + hlsl_note(ctx, &c->loc, VKD3D_SHADER_LOG_ERROR, "The same 'case %d' statement was previously found here.", + c->value); + } + } + + if (found_duplicate) + break; + } +} + } %locations @@ -4582,16 +4864,20 @@ static void validate_texture_format_type(struct hlsl_ctx *ctx, struct hlsl_type enum hlsl_sampler_dim sampler_dim; struct hlsl_attribute *attr; struct parse_attribute_list attr_list; + struct hlsl_ir_switch_case *switch_case; } %token KW_BLENDSTATE %token KW_BREAK %token KW_BUFFER +%token KW_CASE %token KW_CBUFFER +%token KW_CENTROID %token KW_COLUMN_MAJOR %token KW_COMPILE %token KW_CONST %token KW_CONTINUE +%token KW_DEFAULT %token KW_DEPTHSTENCILSTATE %token KW_DEPTHSTENCILVIEW %token KW_DISCARD @@ -4610,6 +4896,7 @@ static void validate_texture_format_type(struct hlsl_ctx *ctx, struct hlsl_type %token KW_MATRIX %token KW_NAMESPACE %token KW_NOINTERPOLATION +%token KW_NOPERSPECTIVE %token KW_OUT %token KW_PACKOFFSET %token KW_PASS @@ -4623,7 +4910,9 @@ static void validate_texture_format_type(struct hlsl_ctx *ctx, struct hlsl_type %token KW_RWBUFFER %token KW_RWSTRUCTUREDBUFFER %token KW_RWTEXTURE1D +%token KW_RWTEXTURE1DARRAY %token KW_RWTEXTURE2D +%token KW_RWTEXTURE2DARRAY %token KW_RWTEXTURE3D %token KW_SAMPLER %token KW_SAMPLER1D @@ -4642,6 +4931,7 @@ static void validate_texture_format_type(struct hlsl_ctx *ctx, struct hlsl_type %token KW_TBUFFER %token KW_TECHNIQUE %token KW_TECHNIQUE10 +%token KW_TECHNIQUE11 %token KW_TEXTURE %token KW_TEXTURE1D %token KW_TEXTURE1DARRAY @@ -4670,7 +4960,6 @@ static void validate_texture_format_type(struct hlsl_ctx *ctx, struct hlsl_type %token OP_LEFTSHIFTASSIGN %token OP_RIGHTSHIFT %token OP_RIGHTSHIFTASSIGN -%token OP_ELLIPSIS %token OP_LE %token OP_GE %token OP_NE @@ -4682,19 +4971,17 @@ static void validate_texture_format_type(struct hlsl_ctx *ctx, struct hlsl_type %token OP_ANDASSIGN %token OP_ORASSIGN %token OP_XORASSIGN -%token OP_UNKNOWN1 -%token OP_UNKNOWN2 -%token OP_UNKNOWN3 -%token OP_UNKNOWN4 %token C_FLOAT %token C_INTEGER +%token C_UNSIGNED %token PRE_LINE %type type_specs %type variables_def %type variables_def_typed +%type switch_cases %token VAR_IDENTIFIER %token NEW_IDENTIFIER @@ -4737,6 +5024,7 @@ static void validate_texture_format_type(struct hlsl_ctx *ctx, struct hlsl_type %type statement %type statement_list %type struct_declaration_without_vars +%type switch_statement %type unary_expr %type boolean @@ -4762,6 +5050,7 @@ static void validate_texture_format_type(struct hlsl_ctx *ctx, struct hlsl_type %type any_identifier %type var_identifier +%type technique_name %type parameter @@ -4775,6 +5064,8 @@ static void validate_texture_format_type(struct hlsl_ctx *ctx, struct hlsl_type %type semantic +%type switch_case + %type field_type %type named_struct_spec %type unnamed_struct_spec @@ -4801,8 +5092,50 @@ hlsl_prog: destroy_block($2); } | hlsl_prog preproc_directive + | hlsl_prog technique | hlsl_prog ';' +technique_name: + %empty + { + $$ = NULL; + } + | any_identifier + +pass_list: + %empty + +technique9: + KW_TECHNIQUE technique_name '{' pass_list '}' + { + hlsl_fixme(ctx, &@$, "Unsupported \'technique\' declaration."); + } + +technique10: + KW_TECHNIQUE10 technique_name '{' pass_list '}' + { + if (ctx->profile->type == VKD3D_SHADER_TYPE_EFFECT && ctx->profile->major_version == 2) + hlsl_error(ctx, &@1, VKD3D_SHADER_ERROR_HLSL_INVALID_SYNTAX, + "The 'technique10' keyword is invalid for this profile."); + + hlsl_fixme(ctx, &@$, "Unsupported \'technique10\' declaration."); + } + +technique11: + KW_TECHNIQUE11 technique_name '{' pass_list '}' + { + if (ctx->profile->type == VKD3D_SHADER_TYPE_EFFECT && ctx->profile->major_version == 2) + hlsl_error(ctx, &@1, VKD3D_SHADER_ERROR_HLSL_INVALID_SYNTAX, + "The 'technique11' keyword is invalid for this profile."); + + hlsl_fixme(ctx, &@$, "Unsupported \'technique11\' declaration."); + } + +technique: + technique9 + | technique10 + | technique11 + buffer_declaration: buffer_type any_identifier colon_attribute { @@ -4952,7 +5285,7 @@ field: if (!(type = apply_type_modifiers(ctx, $2, &modifiers, true, &@1))) YYABORT; - if (modifiers & ~HLSL_STORAGE_NOINTERPOLATION) + if (modifiers & ~HLSL_INTERPOLATION_MODIFIERS_MASK) { struct vkd3d_string_buffer *string; @@ -5132,7 +5465,7 @@ func_prototype_no_attrs: hlsl_error(ctx, &@5, VKD3D_SHADER_ERROR_HLSL_INVALID_RESERVATION, "packoffset() is not allowed on functions."); - if (($$.decl = get_func_decl(&ctx->functions, $3, &$5))) + if (($$.decl = hlsl_get_func_decl(ctx, $3, &$5))) { const struct hlsl_func_parameters *params = &$$.decl->parameters; size_t i; @@ -5249,6 +5582,20 @@ scope_start: hlsl_push_scope(ctx); } +loop_scope_start: + %empty + { + hlsl_push_scope(ctx); + ctx->cur_scope->loop = true; + } + +switch_scope_start: + %empty + { + hlsl_push_scope(ctx); + ctx->cur_scope->_switch = true; + } + var_identifier: VAR_IDENTIFIER | NEW_IDENTIFIER @@ -5441,10 +5788,18 @@ uav_type: { $$ = HLSL_SAMPLER_DIM_1D; } + | KW_RWTEXTURE1DARRAY + { + $$ = HLSL_SAMPLER_DIM_1DARRAY; + } | KW_RWTEXTURE2D { $$ = HLSL_SAMPLER_DIM_2D; } + | KW_RWTEXTURE2DARRAY + { + $$ = HLSL_SAMPLER_DIM_2DARRAY; + } | KW_RWTEXTURE3D { $$ = HLSL_SAMPLER_DIM_3D; @@ -5591,7 +5946,9 @@ type_no_void: { case HLSL_SAMPLER_DIM_BUFFER: case HLSL_SAMPLER_DIM_1D: + case HLSL_SAMPLER_DIM_1DARRAY: case HLSL_SAMPLER_DIM_2D: + case HLSL_SAMPLER_DIM_2DARRAY: case HLSL_SAMPLER_DIM_3D: if ($3->class == HLSL_CLASS_ARRAY) { @@ -5901,6 +6258,14 @@ var_modifiers: { $$ = add_modifiers(ctx, $2, HLSL_STORAGE_NOINTERPOLATION, &@1); } + | KW_CENTROID var_modifiers + { + $$ = add_modifiers(ctx, $2, HLSL_STORAGE_CENTROID, &@1); + } + | KW_NOPERSPECTIVE var_modifiers + { + $$ = add_modifiers(ctx, $2, HLSL_STORAGE_NOPERSPECTIVE, &@1); + } | KW_PRECISE var_modifiers { $$ = add_modifiers(ctx, $2, HLSL_MODIFIER_PRECISE, &@1); @@ -6059,9 +6424,39 @@ statement: | jump_statement | selection_statement | loop_statement + | switch_statement jump_statement: - KW_RETURN expr ';' + KW_BREAK ';' + { + struct hlsl_ir_node *jump; + + if (!is_break_allowed(ctx->cur_scope)) + { + hlsl_error(ctx, &@1, VKD3D_SHADER_ERROR_HLSL_INVALID_SYNTAX, + "The 'break' statement must be used inside of a loop or a switch."); + } + + if (!($$ = make_empty_block(ctx))) + YYABORT; + if (!(jump = hlsl_new_jump(ctx, HLSL_IR_JUMP_BREAK, NULL, &@1))) + YYABORT; + hlsl_block_add_instr($$, jump); + } + | KW_CONTINUE ';' + { + struct hlsl_ir_node *jump; + + check_continue(ctx, ctx->cur_scope, &@1); + + if (!($$ = make_empty_block(ctx))) + YYABORT; + + if (!(jump = hlsl_new_jump(ctx, HLSL_IR_JUMP_UNRESOLVED_CONTINUE, NULL, &@1))) + YYABORT; + hlsl_block_add_instr($$, jump); + } + | KW_RETURN expr ';' { $$ = $2; if (!add_return(ctx, $$, node_from_block($$), &@1)) @@ -6150,25 +6545,127 @@ if_body: } loop_statement: - attribute_list_optional KW_WHILE '(' expr ')' statement + attribute_list_optional loop_scope_start KW_WHILE '(' expr ')' statement { - $$ = create_loop(ctx, LOOP_WHILE, &$1, NULL, $4, NULL, $6, &@2); + $$ = create_loop(ctx, LOOP_WHILE, &$1, NULL, $5, NULL, $7, &@3); + hlsl_pop_scope(ctx); } - | attribute_list_optional KW_DO statement KW_WHILE '(' expr ')' ';' + | attribute_list_optional loop_scope_start KW_DO statement KW_WHILE '(' expr ')' ';' { - $$ = create_loop(ctx, LOOP_DO_WHILE, &$1, NULL, $6, NULL, $3, &@2); + $$ = create_loop(ctx, LOOP_DO_WHILE, &$1, NULL, $7, NULL, $4, &@3); + hlsl_pop_scope(ctx); } - | attribute_list_optional KW_FOR '(' scope_start expr_statement expr_statement expr_optional ')' statement + | attribute_list_optional loop_scope_start KW_FOR '(' expr_statement expr_statement expr_optional ')' statement { - $$ = create_loop(ctx, LOOP_FOR, &$1, $5, $6, $7, $9, &@2); + $$ = create_loop(ctx, LOOP_FOR, &$1, $5, $6, $7, $9, &@3); hlsl_pop_scope(ctx); } - | attribute_list_optional KW_FOR '(' scope_start declaration expr_statement expr_optional ')' statement + | attribute_list_optional loop_scope_start KW_FOR '(' declaration expr_statement expr_optional ')' statement { - $$ = create_loop(ctx, LOOP_FOR, &$1, $5, $6, $7, $9, &@2); + $$ = create_loop(ctx, LOOP_FOR, &$1, $5, $6, $7, $9, &@3); + hlsl_pop_scope(ctx); + } + +switch_statement: + attribute_list_optional switch_scope_start KW_SWITCH '(' expr ')' '{' switch_cases '}' + { + struct hlsl_ir_node *selector = node_from_block($5); + struct hlsl_ir_node *s; + + if (!(selector = add_implicit_conversion(ctx, $5, selector, hlsl_get_scalar_type(ctx, HLSL_TYPE_UINT), &@5))) + { + destroy_switch_cases($8); + destroy_block($5); + YYABORT; + } + + s = hlsl_new_switch(ctx, selector, $8, &@3); + + destroy_switch_cases($8); + + if (!s) + { + destroy_block($5); + YYABORT; + } + + $$ = $5; + hlsl_block_add_instr($$, s); + hlsl_pop_scope(ctx); } +switch_case: + KW_CASE expr ':' statement_list + { + struct hlsl_ir_switch_case *c; + unsigned int value; + + value = evaluate_static_expression_as_uint(ctx, $2, &@2); + + c = hlsl_new_switch_case(ctx, value, false, $4, &@2); + + destroy_block($2); + destroy_block($4); + + if (!c) + YYABORT; + $$ = c; + } + | KW_CASE expr ':' + { + struct hlsl_ir_switch_case *c; + unsigned int value; + + value = evaluate_static_expression_as_uint(ctx, $2, &@2); + + c = hlsl_new_switch_case(ctx, value, false, NULL, &@2); + + destroy_block($2); + + if (!c) + YYABORT; + $$ = c; + } + | KW_DEFAULT ':' statement_list + { + struct hlsl_ir_switch_case *c; + + c = hlsl_new_switch_case(ctx, 0, true, $3, &@1); + + destroy_block($3); + + if (!c) + YYABORT; + $$ = c; + } + | KW_DEFAULT ':' + { + struct hlsl_ir_switch_case *c; + + if (!(c = hlsl_new_switch_case(ctx, 0, true, NULL, &@1))) + YYABORT; + $$ = c; + } + +switch_cases: + switch_case + { + struct hlsl_ir_switch_case *c = LIST_ENTRY($1, struct hlsl_ir_switch_case, entry); + if (!($$ = make_empty_list(ctx))) + { + hlsl_free_ir_switch_case(c); + YYABORT; + } + list_add_head($$, &$1->entry); + } + | switch_cases switch_case + { + $$ = $1; + check_duplicated_switch_cases(ctx, $2, $$); + list_add_tail($$, &$2->entry); + } + expr_optional: %empty { @@ -6213,6 +6710,15 @@ primary_expr: if (!($$ = make_block(ctx, c))) YYABORT; } + | C_UNSIGNED + { + struct hlsl_ir_node *c; + + if (!(c = hlsl_new_uint_constant(ctx, $1, &@1))) + YYABORT; + if (!($$ = make_block(ctx, c))) + YYABORT; + } | boolean { struct hlsl_ir_node *c; @@ -6316,7 +6822,7 @@ postfix_expr: YYABORT; $$ = $1; } - else if (node->data_type->class <= HLSL_CLASS_LAST_NUMERIC) + else if (hlsl_is_numeric_type(node->data_type)) { struct hlsl_ir_node *swizzle; @@ -6359,7 +6865,7 @@ postfix_expr: free_parse_initializer(&$4); YYABORT; } - if ($2->class > HLSL_CLASS_LAST_NUMERIC) + if (!hlsl_is_numeric_type($2)) { struct vkd3d_string_buffer *string; diff --git a/libs/vkd3d/libs/vkd3d-shader/hlsl_codegen.c b/libs/vkd3d/libs/vkd3d-shader/hlsl_codegen.c index be024842164..5a70878bca7 100644 --- a/libs/vkd3d/libs/vkd3d-shader/hlsl_codegen.c +++ b/libs/vkd3d/libs/vkd3d-shader/hlsl_codegen.c @@ -23,30 +23,21 @@ /* TODO: remove when no longer needed, only used for new_offset_instr_from_deref() */ static struct hlsl_ir_node *new_offset_from_path_index(struct hlsl_ctx *ctx, struct hlsl_block *block, - struct hlsl_type *type, struct hlsl_ir_node *offset, struct hlsl_ir_node *idx, - enum hlsl_regset regset, const struct vkd3d_shader_location *loc) + struct hlsl_type *type, struct hlsl_ir_node *base_offset, struct hlsl_ir_node *idx, + enum hlsl_regset regset, unsigned int *offset_component, const struct vkd3d_shader_location *loc) { struct hlsl_ir_node *idx_offset = NULL; struct hlsl_ir_node *c; - hlsl_block_init(block); - switch (type->class) { case HLSL_CLASS_VECTOR: - idx_offset = idx; + *offset_component += hlsl_ir_constant(idx)->value.u[0].u; break; case HLSL_CLASS_MATRIX: { - if (!(c = hlsl_new_uint_constant(ctx, 4, loc))) - return NULL; - hlsl_block_add_instr(block, c); - - if (!(idx_offset = hlsl_new_binary_expr(ctx, HLSL_OP2_MUL, c, idx))) - return NULL; - hlsl_block_add_instr(block, idx_offset); - + idx_offset = idx; break; } @@ -54,6 +45,12 @@ static struct hlsl_ir_node *new_offset_from_path_index(struct hlsl_ctx *ctx, str { unsigned int size = hlsl_type_get_array_element_reg_size(type->e.array.type, regset); + if (regset == HLSL_REGSET_NUMERIC) + { + assert(size % 4 == 0); + size /= 4; + } + if (!(c = hlsl_new_uint_constant(ctx, size, loc))) return NULL; hlsl_block_add_instr(block, c); @@ -69,8 +66,16 @@ static struct hlsl_ir_node *new_offset_from_path_index(struct hlsl_ctx *ctx, str { unsigned int field_idx = hlsl_ir_constant(idx)->value.u[0].u; struct hlsl_struct_field *field = &type->e.record.fields[field_idx]; + unsigned int field_offset = field->reg_offset[regset]; + + if (regset == HLSL_REGSET_NUMERIC) + { + assert(*offset_component == 0); + *offset_component = field_offset % 4; + field_offset /= 4; + } - if (!(c = hlsl_new_uint_constant(ctx, field->reg_offset[regset], loc))) + if (!(c = hlsl_new_uint_constant(ctx, field_offset, loc))) return NULL; hlsl_block_add_instr(block, c); @@ -83,27 +88,33 @@ static struct hlsl_ir_node *new_offset_from_path_index(struct hlsl_ctx *ctx, str vkd3d_unreachable(); } - if (offset) + if (idx_offset) { - if (!(idx_offset = hlsl_new_binary_expr(ctx, HLSL_OP2_ADD, offset, idx_offset))) + if (!(base_offset = hlsl_new_binary_expr(ctx, HLSL_OP2_ADD, base_offset, idx_offset))) return NULL; - hlsl_block_add_instr(block, idx_offset); + hlsl_block_add_instr(block, base_offset); } - return idx_offset; + return base_offset; } /* TODO: remove when no longer needed, only used for replace_deref_path_with_offset() */ static struct hlsl_ir_node *new_offset_instr_from_deref(struct hlsl_ctx *ctx, struct hlsl_block *block, - const struct hlsl_deref *deref, const struct vkd3d_shader_location *loc) + const struct hlsl_deref *deref, unsigned int *offset_component, const struct vkd3d_shader_location *loc) { - enum hlsl_regset regset = hlsl_type_get_regset(deref->data_type); - struct hlsl_ir_node *offset = NULL; + enum hlsl_regset regset = hlsl_deref_get_regset(ctx, deref); + struct hlsl_ir_node *offset; struct hlsl_type *type; unsigned int i; + *offset_component = 0; + hlsl_block_init(block); + if (!(offset = hlsl_new_uint_constant(ctx, 0, loc))) + return NULL; + hlsl_block_add_instr(block, offset); + assert(deref->var); type = deref->var->data_type; @@ -111,9 +122,14 @@ static struct hlsl_ir_node *new_offset_instr_from_deref(struct hlsl_ctx *ctx, st { struct hlsl_block idx_block; + hlsl_block_init(&idx_block); + if (!(offset = new_offset_from_path_index(ctx, &idx_block, type, offset, deref->path[i].node, - regset, loc))) + regset, offset_component, loc))) + { + hlsl_block_cleanup(&idx_block); return NULL; + } hlsl_block_add_block(block, &idx_block); @@ -127,14 +143,13 @@ static struct hlsl_ir_node *new_offset_instr_from_deref(struct hlsl_ctx *ctx, st static bool replace_deref_path_with_offset(struct hlsl_ctx *ctx, struct hlsl_deref *deref, struct hlsl_ir_node *instr) { - struct hlsl_type *type; + unsigned int offset_component; struct hlsl_ir_node *offset; struct hlsl_block block; + struct hlsl_type *type; assert(deref->var); - - /* register offsets shouldn't be used before this point is reached. */ - assert(!deref->offset.node); + assert(!hlsl_deref_is_lowered(deref)); type = hlsl_deref_get_type(ctx, deref); @@ -148,16 +163,35 @@ static bool replace_deref_path_with_offset(struct hlsl_ctx *ctx, struct hlsl_der deref->data_type = type; - if (!(offset = new_offset_instr_from_deref(ctx, &block, deref, &instr->loc))) + if (!(offset = new_offset_instr_from_deref(ctx, &block, deref, &offset_component, &instr->loc))) return false; list_move_before(&instr->entry, &block.instrs); hlsl_cleanup_deref(deref); - hlsl_src_from_node(&deref->offset, offset); + hlsl_src_from_node(&deref->rel_offset, offset); + deref->const_offset = offset_component; return true; } +static bool clean_constant_deref_offset_srcs(struct hlsl_ctx *ctx, struct hlsl_deref *deref, + struct hlsl_ir_node *instr) +{ + if (deref->rel_offset.node && deref->rel_offset.node->type == HLSL_IR_CONSTANT) + { + enum hlsl_regset regset = hlsl_deref_get_regset(ctx, deref); + + if (regset == HLSL_REGSET_NUMERIC) + deref->const_offset += 4 * hlsl_ir_constant(deref->rel_offset.node)->value.u[0].u; + else + deref->const_offset += hlsl_ir_constant(deref->rel_offset.node)->value.u[0].u; + hlsl_src_remove(&deref->rel_offset); + return true; + } + return false; +} + + /* Split uniforms into two variables representing the constant and temp * registers, and copy the former to the latter, so that writes to uniforms * work. */ @@ -195,7 +229,7 @@ static void prepend_uniform_copy(struct hlsl_ctx *ctx, struct hlsl_block *block, static void validate_field_semantic(struct hlsl_ctx *ctx, struct hlsl_struct_field *field) { - if (!field->semantic.name && hlsl_get_multiarray_element_type(field->type)->class <= HLSL_CLASS_LAST_NUMERIC + if (!field->semantic.name && hlsl_is_numeric_type(hlsl_get_multiarray_element_type(field->type)) && !field->semantic.reported_missing) { hlsl_error(ctx, &field->loc, VKD3D_SHADER_ERROR_HLSL_MISSING_SEMANTIC, @@ -249,7 +283,7 @@ static struct hlsl_ir_var *add_semantic_var(struct hlsl_ctx *ctx, struct hlsl_ir { hlsl_error(ctx, loc, VKD3D_SHADER_ERROR_HLSL_INVALID_SEMANTIC, "Output semantic \"%s%u\" is used multiple times.", semantic->name, index); - hlsl_note(ctx, &ext_var->loc, HLSL_LEVEL_ERROR, + hlsl_note(ctx, &ext_var->loc, VKD3D_SHADER_LOG_ERROR, "First use of \"%s%u\" is here.", semantic->name, index); semantic->reported_duplicated_output_next_index = index + 1; } @@ -262,7 +296,7 @@ static struct hlsl_ir_var *add_semantic_var(struct hlsl_ctx *ctx, struct hlsl_ir hlsl_error(ctx, loc, VKD3D_SHADER_ERROR_HLSL_INVALID_SEMANTIC, "Input semantic \"%s%u\" is used multiple times with incompatible types.", semantic->name, index); - hlsl_note(ctx, &ext_var->loc, HLSL_LEVEL_ERROR, + hlsl_note(ctx, &ext_var->loc, VKD3D_SHADER_LOG_ERROR, "First declaration of \"%s%u\" is here.", semantic->name, index); semantic->reported_duplicated_input_incompatible_next_index = index + 1; } @@ -305,7 +339,7 @@ static void prepend_input_copy(struct hlsl_ctx *ctx, struct hlsl_block *block, s struct hlsl_ir_node *c; unsigned int i; - if (type->class > HLSL_CLASS_LAST_NUMERIC) + if (!hlsl_is_numeric_type(type)) { struct vkd3d_string_buffer *string; if (!(string = hlsl_type_to_string(ctx, type))) @@ -377,6 +411,8 @@ static void prepend_input_copy_recurse(struct hlsl_ctx *ctx, struct hlsl_block * for (i = 0; i < hlsl_type_element_count(type); ++i) { + unsigned int element_modifiers = modifiers; + if (type->class == HLSL_CLASS_ARRAY) { elem_semantic_index = semantic_index @@ -391,6 +427,17 @@ static void prepend_input_copy_recurse(struct hlsl_ctx *ctx, struct hlsl_block * semantic = &field->semantic; elem_semantic_index = semantic->index; loc = &field->loc; + element_modifiers |= field->storage_modifiers; + + /* TODO: 'sample' modifier is not supported yet */ + + /* 'nointerpolation' always takes precedence, next the same is done for 'sample', + remaining modifiers are combined. */ + if (element_modifiers & HLSL_STORAGE_NOINTERPOLATION) + { + element_modifiers &= ~HLSL_INTERPOLATION_MODIFIERS_MASK; + element_modifiers |= HLSL_STORAGE_NOINTERPOLATION; + } } if (!(c = hlsl_new_uint_constant(ctx, i, &var->loc))) @@ -402,7 +449,7 @@ static void prepend_input_copy_recurse(struct hlsl_ctx *ctx, struct hlsl_block * return; list_add_after(&c->entry, &element_load->node.entry); - prepend_input_copy_recurse(ctx, block, element_load, modifiers, semantic, elem_semantic_index); + prepend_input_copy_recurse(ctx, block, element_load, element_modifiers, semantic, elem_semantic_index); } } else @@ -434,7 +481,7 @@ static void append_output_copy(struct hlsl_ctx *ctx, struct hlsl_block *block, s struct hlsl_ir_node *c; unsigned int i; - if (type->class > HLSL_CLASS_LAST_NUMERIC) + if (!hlsl_is_numeric_type(type)) { struct vkd3d_string_buffer *string; if (!(string = hlsl_type_to_string(ctx, type))) @@ -562,7 +609,19 @@ bool hlsl_transform_ir(struct hlsl_ctx *ctx, bool (*func)(struct hlsl_ctx *ctx, progress |= hlsl_transform_ir(ctx, func, &iff->else_block, context); } else if (instr->type == HLSL_IR_LOOP) + { progress |= hlsl_transform_ir(ctx, func, &hlsl_ir_loop(instr)->body, context); + } + else if (instr->type == HLSL_IR_SWITCH) + { + struct hlsl_ir_switch *s = hlsl_ir_switch(instr); + struct hlsl_ir_switch_case *c; + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + progress |= hlsl_transform_ir(ctx, func, &c->body, context); + } + } progress |= func(ctx, instr, context); } @@ -822,6 +881,30 @@ static bool lower_return(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *fun } } } + else if (instr->type == HLSL_IR_SWITCH) + { + struct hlsl_ir_switch *s = hlsl_ir_switch(instr); + struct hlsl_ir_switch_case *c; + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + has_early_return |= lower_return(ctx, func, &c->body, true); + } + + if (has_early_return) + { + if (in_loop) + { + /* For a 'switch' nested in a loop append a break after the 'switch'. */ + insert_early_return_break(ctx, func, instr); + } + else + { + cf_instr = instr; + break; + } + } + } } if (return_instr) @@ -989,7 +1072,7 @@ static bool lower_matrix_swizzles(struct hlsl_ctx *ctx, struct hlsl_ir_node *ins * For the latter case, this pass takes care of lowering hlsl_ir_indexes into individual * hlsl_ir_loads, or individual hlsl_ir_resource_loads, in case the indexing is a * resource access. */ -static bool lower_index_loads(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_index_loads(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { struct hlsl_ir_node *val, *store; struct hlsl_deref var_deref; @@ -1023,8 +1106,7 @@ static bool lower_index_loads(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, if (!(load = hlsl_new_resource_load(ctx, ¶ms, &instr->loc))) return false; - list_add_before(&instr->entry, &load->entry); - hlsl_replace_node(instr, load); + hlsl_block_add_instr(block, load); return true; } @@ -1034,7 +1116,7 @@ static bool lower_index_loads(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, if (!(store = hlsl_new_simple_store(ctx, var, val))) return false; - list_add_before(&instr->entry, &store->entry); + hlsl_block_add_instr(block, store); if (hlsl_index_is_noncontiguous(index)) { @@ -1054,38 +1136,36 @@ static bool lower_index_loads(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, if (!(c = hlsl_new_uint_constant(ctx, i, &instr->loc))) return false; - list_add_before(&instr->entry, &c->entry); + hlsl_block_add_instr(block, c); if (!(load = hlsl_new_load_index(ctx, &var_deref, c, &instr->loc))) return false; - list_add_before(&instr->entry, &load->node.entry); + hlsl_block_add_instr(block, &load->node); if (!(load = hlsl_new_load_index(ctx, &load->src, index->idx.node, &instr->loc))) return false; - list_add_before(&instr->entry, &load->node.entry); + hlsl_block_add_instr(block, &load->node); if (!(store = hlsl_new_store_index(ctx, &row_deref, c, &load->node, 0, &instr->loc))) return false; - list_add_before(&instr->entry, &store->entry); + hlsl_block_add_instr(block, store); } if (!(load = hlsl_new_var_load(ctx, var, &instr->loc))) return false; - list_add_before(&instr->entry, &load->node.entry); - hlsl_replace_node(instr, &load->node); + hlsl_block_add_instr(block, &load->node); } else { if (!(load = hlsl_new_load_index(ctx, &var_deref, index->idx.node, &instr->loc))) return false; - list_add_before(&instr->entry, &load->node.entry); - hlsl_replace_node(instr, &load->node); + hlsl_block_add_instr(block, &load->node); } return true; } /* Lower casts from vec1 to vecN to swizzles. */ -static bool lower_broadcasts(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_broadcasts(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { const struct hlsl_type *src_type, *dst_type; struct hlsl_type *dst_scalar_type; @@ -1101,25 +1181,22 @@ static bool lower_broadcasts(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, v if (src_type->class <= HLSL_CLASS_VECTOR && dst_type->class <= HLSL_CLASS_VECTOR && src_type->dimx == 1) { - struct hlsl_ir_node *replacement, *new_cast, *swizzle; + struct hlsl_ir_node *new_cast, *swizzle; dst_scalar_type = hlsl_get_scalar_type(ctx, dst_type->base_type); /* We need to preserve the cast since it might be doing more than just * turning the scalar into a vector. */ if (!(new_cast = hlsl_new_cast(ctx, cast->operands[0].node, dst_scalar_type, &cast->node.loc))) return false; - list_add_after(&cast->node.entry, &new_cast->entry); - replacement = new_cast; + hlsl_block_add_instr(block, new_cast); if (dst_type->dimx != 1) { - if (!(swizzle = hlsl_new_swizzle(ctx, HLSL_SWIZZLE(X, X, X, X), dst_type->dimx, replacement, &cast->node.loc))) + if (!(swizzle = hlsl_new_swizzle(ctx, HLSL_SWIZZLE(X, X, X, X), dst_type->dimx, new_cast, &cast->node.loc))) return false; - list_add_after(&new_cast->entry, &swizzle->entry); - replacement = swizzle; + hlsl_block_add_instr(block, swizzle); } - hlsl_replace_node(&cast->node, replacement); return true; } @@ -1632,6 +1709,19 @@ static void copy_propagation_invalidate_from_block(struct hlsl_ctx *ctx, struct break; } + case HLSL_IR_SWITCH: + { + struct hlsl_ir_switch *s = hlsl_ir_switch(instr); + struct hlsl_ir_switch_case *c; + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + copy_propagation_invalidate_from_block(ctx, state, &c->body); + } + + break; + } + default: break; } @@ -1680,6 +1770,28 @@ static bool copy_propagation_process_loop(struct hlsl_ctx *ctx, struct hlsl_ir_l return progress; } +static bool copy_propagation_process_switch(struct hlsl_ctx *ctx, struct hlsl_ir_switch *s, + struct copy_propagation_state *state) +{ + struct copy_propagation_state inner_state; + struct hlsl_ir_switch_case *c; + bool progress = false; + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + copy_propagation_state_init(ctx, &inner_state, state); + progress |= copy_propagation_transform_block(ctx, &c->body, &inner_state); + copy_propagation_state_destroy(&inner_state); + } + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + copy_propagation_invalidate_from_block(ctx, state, &c->body); + } + + return progress; +} + static bool copy_propagation_transform_block(struct hlsl_ctx *ctx, struct hlsl_block *block, struct copy_propagation_state *state) { @@ -1718,6 +1830,10 @@ static bool copy_propagation_transform_block(struct hlsl_ctx *ctx, struct hlsl_b progress |= copy_propagation_process_loop(ctx, hlsl_ir_loop(instr), state); break; + case HLSL_IR_SWITCH: + progress |= copy_propagation_process_switch(ctx, hlsl_ir_switch(instr), state); + break; + default: break; } @@ -1981,7 +2097,7 @@ static bool split_matrix_copies(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr return true; } -static bool lower_narrowing_casts(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_narrowing_casts(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { const struct hlsl_type *src_type, *dst_type; struct hlsl_type *dst_vector_type; @@ -2004,12 +2120,12 @@ static bool lower_narrowing_casts(struct hlsl_ctx *ctx, struct hlsl_ir_node *ins * narrowing the vector. */ if (!(new_cast = hlsl_new_cast(ctx, cast->operands[0].node, dst_vector_type, &cast->node.loc))) return false; - list_add_after(&cast->node.entry, &new_cast->entry); + hlsl_block_add_instr(block, new_cast); + if (!(swizzle = hlsl_new_swizzle(ctx, HLSL_SWIZZLE(X, Y, Z, W), dst_type->dimx, new_cast, &cast->node.loc))) return false; - list_add_after(&new_cast->entry, &swizzle->entry); + hlsl_block_add_instr(block, swizzle); - hlsl_replace_node(&cast->node, swizzle); return true; } @@ -2068,7 +2184,138 @@ static bool remove_trivial_swizzles(struct hlsl_ctx *ctx, struct hlsl_ir_node *i return true; } -static bool lower_nonconstant_vector_derefs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool remove_trivial_conditional_branches(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +{ + struct hlsl_ir_constant *condition; + struct hlsl_ir_if *iff; + + if (instr->type != HLSL_IR_IF) + return false; + iff = hlsl_ir_if(instr); + if (iff->condition.node->type != HLSL_IR_CONSTANT) + return false; + condition = hlsl_ir_constant(iff->condition.node); + + list_move_before(&instr->entry, condition->value.u[0].u ? &iff->then_block.instrs : &iff->else_block.instrs); + list_remove(&instr->entry); + hlsl_free_instr(instr); + + return true; +} + +static bool normalize_switch_cases(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +{ + struct hlsl_ir_switch_case *c, *def = NULL; + bool missing_terminal_break = false; + struct hlsl_ir_node *node; + struct hlsl_ir_jump *jump; + struct hlsl_ir_switch *s; + + if (instr->type != HLSL_IR_SWITCH) + return false; + s = hlsl_ir_switch(instr); + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + bool terminal_break = false; + + if (list_empty(&c->body.instrs)) + { + terminal_break = !!list_next(&s->cases, &c->entry); + } + else + { + node = LIST_ENTRY(list_tail(&c->body.instrs), struct hlsl_ir_node, entry); + if (node->type == HLSL_IR_JUMP) + { + jump = hlsl_ir_jump(node); + terminal_break = jump->type == HLSL_IR_JUMP_BREAK; + } + } + + missing_terminal_break |= !terminal_break; + + if (!terminal_break) + { + if (c->is_default) + { + hlsl_error(ctx, &c->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_SYNTAX, + "The 'default' case block is not terminated with 'break' or 'return'."); + } + else + { + hlsl_error(ctx, &c->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_SYNTAX, + "Switch case block '%u' is not terminated with 'break' or 'return'.", c->value); + } + } + } + + if (missing_terminal_break) + return true; + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + if (c->is_default) + { + def = c; + + /* Remove preceding empty cases. */ + while (list_prev(&s->cases, &def->entry)) + { + c = LIST_ENTRY(list_prev(&s->cases, &def->entry), struct hlsl_ir_switch_case, entry); + if (!list_empty(&c->body.instrs)) + break; + hlsl_free_ir_switch_case(c); + } + + if (list_empty(&def->body.instrs)) + { + /* Remove following empty cases. */ + while (list_next(&s->cases, &def->entry)) + { + c = LIST_ENTRY(list_next(&s->cases, &def->entry), struct hlsl_ir_switch_case, entry); + if (!list_empty(&c->body.instrs)) + break; + hlsl_free_ir_switch_case(c); + } + + /* Merge with the next case. */ + if (list_next(&s->cases, &def->entry)) + { + c = LIST_ENTRY(list_next(&s->cases, &def->entry), struct hlsl_ir_switch_case, entry); + c->is_default = true; + hlsl_free_ir_switch_case(def); + def = c; + } + } + + break; + } + } + + if (def) + { + list_remove(&def->entry); + } + else + { + struct hlsl_ir_node *jump; + + if (!(def = hlsl_new_switch_case(ctx, 0, true, NULL, &s->node.loc))) + return true; + if (!(jump = hlsl_new_jump(ctx, HLSL_IR_JUMP_BREAK, NULL, &s->node.loc))) + { + hlsl_free_ir_switch_case(def); + return true; + } + hlsl_block_add_instr(&def->body, jump); + } + list_add_tail(&s->cases, &def->entry); + + return true; +} + +static bool lower_nonconstant_vector_derefs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { struct hlsl_ir_node *idx; struct hlsl_deref *deref; @@ -2099,11 +2346,11 @@ static bool lower_nonconstant_vector_derefs(struct hlsl_ctx *ctx, struct hlsl_ir if (!(vector_load = hlsl_new_load_parent(ctx, deref, &instr->loc))) return false; - list_add_before(&instr->entry, &vector_load->node.entry); + hlsl_block_add_instr(block, &vector_load->node); if (!(swizzle = hlsl_new_swizzle(ctx, HLSL_SWIZZLE(X, X, X, X), type->dimx, idx, &instr->loc))) return false; - list_add_before(&instr->entry, &swizzle->entry); + hlsl_block_add_instr(block, swizzle); value.u[0].u = 0; value.u[1].u = 1; @@ -2111,18 +2358,18 @@ static bool lower_nonconstant_vector_derefs(struct hlsl_ctx *ctx, struct hlsl_ir value.u[3].u = 3; if (!(c = hlsl_new_constant(ctx, hlsl_get_vector_type(ctx, HLSL_TYPE_UINT, type->dimx), &value, &instr->loc))) return false; - list_add_before(&instr->entry, &c->entry); + hlsl_block_add_instr(block, c); operands[0] = swizzle; operands[1] = c; if (!(eq = hlsl_new_expr(ctx, HLSL_OP2_EQUAL, operands, hlsl_get_vector_type(ctx, HLSL_TYPE_BOOL, type->dimx), &instr->loc))) return false; - list_add_before(&instr->entry, &eq->entry); + hlsl_block_add_instr(block, eq); if (!(eq = hlsl_new_cast(ctx, eq, type, &instr->loc))) return false; - list_add_before(&instr->entry, &eq->entry); + hlsl_block_add_instr(block, eq); op = HLSL_OP2_DOT; if (type->dimx == 1) @@ -2134,8 +2381,7 @@ static bool lower_nonconstant_vector_derefs(struct hlsl_ctx *ctx, struct hlsl_ir operands[1] = eq; if (!(dot = hlsl_new_expr(ctx, op, operands, instr->data_type, &instr->loc))) return false; - list_add_before(&instr->entry, &dot->entry); - hlsl_replace_node(instr, dot); + hlsl_block_add_instr(block, dot); return true; } @@ -2173,6 +2419,7 @@ static bool lower_combined_samples(struct hlsl_ctx *ctx, struct hlsl_ir_node *in case HLSL_RESOURCE_SAMPLE: case HLSL_RESOURCE_SAMPLE_LOD: case HLSL_RESOURCE_SAMPLE_LOD_BIAS: + case HLSL_RESOURCE_SAMPLE_PROJ: break; } if (load->sampler.var) @@ -2184,7 +2431,7 @@ static bool lower_combined_samples(struct hlsl_ctx *ctx, struct hlsl_ir_node *in return false; } - assert(hlsl_type_get_regset(load->resource.var->data_type) == HLSL_REGSET_SAMPLERS); + assert(hlsl_deref_get_regset(ctx, &load->resource) == HLSL_REGSET_SAMPLERS); if (!(name = hlsl_get_string_buffer(ctx))) return false; @@ -2275,10 +2522,10 @@ static bool sort_synthetic_separated_samplers_first(struct hlsl_ctx *ctx) } /* Lower DIV to RCP + MUL. */ -static bool lower_division(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_division(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { + struct hlsl_ir_node *rcp, *mul; struct hlsl_ir_expr *expr; - struct hlsl_ir_node *rcp; if (instr->type != HLSL_IR_EXPR) return false; @@ -2288,18 +2535,20 @@ static bool lower_division(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, voi if (!(rcp = hlsl_new_unary_expr(ctx, HLSL_OP1_RCP, expr->operands[1].node, &instr->loc))) return false; - list_add_before(&expr->node.entry, &rcp->entry); - expr->op = HLSL_OP2_MUL; - hlsl_src_remove(&expr->operands[1]); - hlsl_src_from_node(&expr->operands[1], rcp); + hlsl_block_add_instr(block, rcp); + + if (!(mul = hlsl_new_binary_expr(ctx, HLSL_OP2_MUL, expr->operands[0].node, rcp))) + return false; + hlsl_block_add_instr(block, mul); + return true; } /* Lower SQRT to RSQ + RCP. */ -static bool lower_sqrt(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_sqrt(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { + struct hlsl_ir_node *rsq, *rcp; struct hlsl_ir_expr *expr; - struct hlsl_ir_node *rsq; if (instr->type != HLSL_IR_EXPR) return false; @@ -2309,15 +2558,16 @@ static bool lower_sqrt(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *c if (!(rsq = hlsl_new_unary_expr(ctx, HLSL_OP1_RSQ, expr->operands[0].node, &instr->loc))) return false; - list_add_before(&expr->node.entry, &rsq->entry); - expr->op = HLSL_OP1_RCP; - hlsl_src_remove(&expr->operands[0]); - hlsl_src_from_node(&expr->operands[0], rsq); + hlsl_block_add_instr(block, rsq); + + if (!(rcp = hlsl_new_unary_expr(ctx, HLSL_OP1_RCP, rsq, &instr->loc))) + return false; + hlsl_block_add_instr(block, rcp); return true; } /* Lower DP2 to MUL + ADD */ -static bool lower_dot(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_dot(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { struct hlsl_ir_node *arg1, *arg2, *mul, *replacement, *zero, *add_x, *add_y; struct hlsl_ir_expr *expr; @@ -2338,7 +2588,7 @@ static bool lower_dot(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *co if (!(zero = hlsl_new_float_constant(ctx, 0.0f, &expr->node.loc))) return false; - list_add_before(&instr->entry, &zero->entry); + hlsl_block_add_instr(block, zero); operands[0] = arg1; operands[1] = arg2; @@ -2351,27 +2601,26 @@ static bool lower_dot(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *co { if (!(mul = hlsl_new_binary_expr(ctx, HLSL_OP2_MUL, expr->operands[0].node, expr->operands[1].node))) return false; - list_add_before(&instr->entry, &mul->entry); + hlsl_block_add_instr(block, mul); if (!(add_x = hlsl_new_swizzle(ctx, HLSL_SWIZZLE(X, X, X, X), instr->data_type->dimx, mul, &expr->node.loc))) return false; - list_add_before(&instr->entry, &add_x->entry); + hlsl_block_add_instr(block, add_x); if (!(add_y = hlsl_new_swizzle(ctx, HLSL_SWIZZLE(Y, Y, Y, Y), instr->data_type->dimx, mul, &expr->node.loc))) return false; - list_add_before(&instr->entry, &add_y->entry); + hlsl_block_add_instr(block, add_y); if (!(replacement = hlsl_new_binary_expr(ctx, HLSL_OP2_ADD, add_x, add_y))) return false; } - list_add_before(&instr->entry, &replacement->entry); + hlsl_block_add_instr(block, replacement); - hlsl_replace_node(instr, replacement); return true; } /* Lower ABS to MAX */ -static bool lower_abs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_abs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { struct hlsl_ir_node *arg, *neg, *replacement; struct hlsl_ir_expr *expr; @@ -2385,18 +2634,17 @@ static bool lower_abs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *co if (!(neg = hlsl_new_unary_expr(ctx, HLSL_OP1_NEG, arg, &instr->loc))) return false; - list_add_before(&instr->entry, &neg->entry); + hlsl_block_add_instr(block, neg); if (!(replacement = hlsl_new_binary_expr(ctx, HLSL_OP2_MAX, neg, arg))) return false; - list_add_before(&instr->entry, &replacement->entry); + hlsl_block_add_instr(block, replacement); - hlsl_replace_node(instr, replacement); return true; } /* Lower ROUND using FRC, ROUND(x) -> ((x + 0.5) - FRC(x + 0.5)). */ -static bool lower_round(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_round(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { struct hlsl_ir_node *arg, *neg, *sum, *frc, *half, *replacement; struct hlsl_type *type = instr->data_type; @@ -2417,33 +2665,89 @@ static bool lower_round(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void * half_value.u[i].f = 0.5f; if (!(half = hlsl_new_constant(ctx, type, &half_value, &expr->node.loc))) return false; - - list_add_before(&instr->entry, &half->entry); + hlsl_block_add_instr(block, half); if (!(sum = hlsl_new_binary_expr(ctx, HLSL_OP2_ADD, arg, half))) return false; - list_add_before(&instr->entry, &sum->entry); + hlsl_block_add_instr(block, sum); if (!(frc = hlsl_new_unary_expr(ctx, HLSL_OP1_FRACT, sum, &instr->loc))) return false; - list_add_before(&instr->entry, &frc->entry); + hlsl_block_add_instr(block, frc); if (!(neg = hlsl_new_unary_expr(ctx, HLSL_OP1_NEG, frc, &instr->loc))) return false; - list_add_before(&instr->entry, &neg->entry); + hlsl_block_add_instr(block, neg); if (!(replacement = hlsl_new_binary_expr(ctx, HLSL_OP2_ADD, sum, neg))) return false; - list_add_before(&instr->entry, &replacement->entry); + hlsl_block_add_instr(block, replacement); + + return true; +} + +/* Lower CEIL to FRC */ +static bool lower_ceil(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) +{ + struct hlsl_ir_node *arg, *neg, *sum, *frc; + struct hlsl_ir_expr *expr; + + if (instr->type != HLSL_IR_EXPR) + return false; + + expr = hlsl_ir_expr(instr); + arg = expr->operands[0].node; + if (expr->op != HLSL_OP1_CEIL) + return false; + + if (!(neg = hlsl_new_unary_expr(ctx, HLSL_OP1_NEG, arg, &instr->loc))) + return false; + hlsl_block_add_instr(block, neg); + + if (!(frc = hlsl_new_unary_expr(ctx, HLSL_OP1_FRACT, neg, &instr->loc))) + return false; + hlsl_block_add_instr(block, frc); + + if (!(sum = hlsl_new_binary_expr(ctx, HLSL_OP2_ADD, frc, arg))) + return false; + hlsl_block_add_instr(block, sum); + + return true; +} + +/* Lower FLOOR to FRC */ +static bool lower_floor(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) +{ + struct hlsl_ir_node *arg, *neg, *sum, *frc; + struct hlsl_ir_expr *expr; + + if (instr->type != HLSL_IR_EXPR) + return false; + + expr = hlsl_ir_expr(instr); + arg = expr->operands[0].node; + if (expr->op != HLSL_OP1_FLOOR) + return false; + + if (!(frc = hlsl_new_unary_expr(ctx, HLSL_OP1_FRACT, arg, &instr->loc))) + return false; + hlsl_block_add_instr(block, frc); + + if (!(neg = hlsl_new_unary_expr(ctx, HLSL_OP1_NEG, frc, &instr->loc))) + return false; + hlsl_block_add_instr(block, neg); + + if (!(sum = hlsl_new_binary_expr(ctx, HLSL_OP2_ADD, neg, arg))) + return false; + hlsl_block_add_instr(block, sum); - hlsl_replace_node(instr, replacement); return true; } /* Use 'movc' for the ternary operator. */ -static bool lower_ternary(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_ternary(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { - struct hlsl_ir_node *operands[HLSL_MAX_OPERANDS], *replacement; + struct hlsl_ir_node *operands[HLSL_MAX_OPERANDS] = { 0 }, *replacement; struct hlsl_ir_node *zero, *cond, *first, *second; struct hlsl_constant_value zero_value = { 0 }; struct hlsl_ir_expr *expr; @@ -2460,39 +2764,63 @@ static bool lower_ternary(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void first = expr->operands[1].node; second = expr->operands[2].node; - if (cond->data_type->base_type == HLSL_TYPE_FLOAT) + if (ctx->profile->major_version < 4 && ctx->profile->type == VKD3D_SHADER_TYPE_PIXEL) { - if (!(zero = hlsl_new_constant(ctx, cond->data_type, &zero_value, &instr->loc))) + struct hlsl_ir_node *abs, *neg; + + if (!(abs = hlsl_new_unary_expr(ctx, HLSL_OP1_ABS, cond, &instr->loc))) return false; - list_add_tail(&instr->entry, &zero->entry); + hlsl_block_add_instr(block, abs); - memset(operands, 0, sizeof(operands)); - operands[0] = zero; - operands[1] = cond; - type = cond->data_type; - type = hlsl_get_numeric_type(ctx, type->class, HLSL_TYPE_BOOL, type->dimx, type->dimy); - if (!(cond = hlsl_new_expr(ctx, HLSL_OP2_NEQUAL, operands, type, &instr->loc))) + if (!(neg = hlsl_new_unary_expr(ctx, HLSL_OP1_NEG, abs, &instr->loc))) return false; - list_add_before(&instr->entry, &cond->entry); - } + hlsl_block_add_instr(block, neg); - memset(operands, 0, sizeof(operands)); - operands[0] = cond; - operands[1] = first; - operands[2] = second; - if (!(replacement = hlsl_new_expr(ctx, HLSL_OP3_MOVC, operands, first->data_type, &instr->loc))) + operands[0] = neg; + operands[1] = second; + operands[2] = first; + if (!(replacement = hlsl_new_expr(ctx, HLSL_OP3_CMP, operands, first->data_type, &instr->loc))) + return false; + } + else if (ctx->profile->major_version < 4 && ctx->profile->type == VKD3D_SHADER_TYPE_VERTEX) + { + hlsl_fixme(ctx, &instr->loc, "Ternary operator is not implemented for %s profile.", ctx->profile->name); return false; - list_add_before(&instr->entry, &replacement->entry); + } + else + { + if (cond->data_type->base_type == HLSL_TYPE_FLOAT) + { + if (!(zero = hlsl_new_constant(ctx, cond->data_type, &zero_value, &instr->loc))) + return false; + hlsl_block_add_instr(block, zero); + + operands[0] = zero; + operands[1] = cond; + type = cond->data_type; + type = hlsl_get_numeric_type(ctx, type->class, HLSL_TYPE_BOOL, type->dimx, type->dimy); + if (!(cond = hlsl_new_expr(ctx, HLSL_OP2_NEQUAL, operands, type, &instr->loc))) + return false; + hlsl_block_add_instr(block, cond); + } + + memset(operands, 0, sizeof(operands)); + operands[0] = cond; + operands[1] = first; + operands[2] = second; + if (!(replacement = hlsl_new_expr(ctx, HLSL_OP3_MOVC, operands, first->data_type, &instr->loc))) + return false; + } - hlsl_replace_node(instr, replacement); + hlsl_block_add_instr(block, replacement); return true; } -static bool lower_casts_to_bool(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_casts_to_bool(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { struct hlsl_type *type = instr->data_type, *arg_type; static const struct hlsl_constant_value zero_value; - struct hlsl_ir_node *zero; + struct hlsl_ir_node *zero, *neq; struct hlsl_ir_expr *expr; if (instr->type != HLSL_IR_EXPR) @@ -2512,10 +2840,12 @@ static bool lower_casts_to_bool(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr zero = hlsl_new_constant(ctx, arg_type, &zero_value, &instr->loc); if (!zero) return false; - list_add_before(&instr->entry, &zero->entry); + hlsl_block_add_instr(block, zero); - expr->op = HLSL_OP2_NEQUAL; - hlsl_src_from_node(&expr->operands[1], zero); + if (!(neq = hlsl_new_binary_expr(ctx, HLSL_OP2_NEQUAL, expr->operands[0].node, zero))) + return false; + neq->data_type = expr->node.data_type; + hlsl_block_add_instr(block, neq); return true; } @@ -2523,36 +2853,19 @@ static bool lower_casts_to_bool(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr struct hlsl_ir_node *hlsl_add_conditional(struct hlsl_ctx *ctx, struct hlsl_block *instrs, struct hlsl_ir_node *condition, struct hlsl_ir_node *if_true, struct hlsl_ir_node *if_false) { - struct hlsl_block then_block, else_block; - struct hlsl_ir_node *iff, *store; - struct hlsl_ir_load *load; - struct hlsl_ir_var *var; + struct hlsl_ir_node *operands[HLSL_MAX_OPERANDS]; + struct hlsl_ir_node *cond; assert(hlsl_types_are_equal(if_true->data_type, if_false->data_type)); - if (!(var = hlsl_new_synthetic_var(ctx, "conditional", if_true->data_type, &condition->loc))) - return NULL; - - hlsl_block_init(&then_block); - hlsl_block_init(&else_block); - - if (!(store = hlsl_new_simple_store(ctx, var, if_true))) - return NULL; - hlsl_block_add_instr(&then_block, store); - - if (!(store = hlsl_new_simple_store(ctx, var, if_false))) - return NULL; - hlsl_block_add_instr(&else_block, store); - - if (!(iff = hlsl_new_if(ctx, condition, &then_block, &else_block, &condition->loc))) - return NULL; - hlsl_block_add_instr(instrs, iff); - - if (!(load = hlsl_new_var_load(ctx, var, &condition->loc))) - return NULL; - hlsl_block_add_instr(instrs, &load->node); + operands[0] = condition; + operands[1] = if_true; + operands[2] = if_false; + if (!(cond = hlsl_new_expr(ctx, HLSL_OP3_TERNARY, operands, if_true->data_type, &condition->loc))) + return false; + hlsl_block_add_instr(instrs, cond); - return &load->node; + return cond; } static bool lower_int_division(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) @@ -2683,10 +2996,10 @@ static bool lower_int_modulus(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, return hlsl_add_conditional(ctx, block, and, neg, cast3); } -static bool lower_int_abs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_int_abs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { struct hlsl_type *type = instr->data_type; - struct hlsl_ir_node *arg, *neg; + struct hlsl_ir_node *arg, *neg, *max; struct hlsl_ir_expr *expr; if (instr->type != HLSL_IR_EXPR) @@ -2704,15 +3017,16 @@ static bool lower_int_abs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void if (!(neg = hlsl_new_unary_expr(ctx, HLSL_OP1_NEG, arg, &instr->loc))) return false; - list_add_before(&instr->entry, &neg->entry); + hlsl_block_add_instr(block, neg); - expr->op = HLSL_OP2_MAX; - hlsl_src_from_node(&expr->operands[1], neg); + if (!(max = hlsl_new_binary_expr(ctx, HLSL_OP2_MAX, arg, neg))) + return false; + hlsl_block_add_instr(block, max); return true; } -static bool lower_int_dot(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) +static bool lower_int_dot(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, struct hlsl_block *block) { struct hlsl_ir_node *arg1, *arg2, *mult, *comps[4] = {0}, *res; struct hlsl_type *type = instr->data_type; @@ -2738,7 +3052,7 @@ static bool lower_int_dot(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void if (!(mult = hlsl_new_binary_expr(ctx, is_bool ? HLSL_OP2_LOGIC_AND : HLSL_OP2_MUL, arg1, arg2))) return false; - list_add_before(&instr->entry, &mult->entry); + hlsl_block_add_instr(block, mult); for (i = 0; i < dimx; ++i) { @@ -2746,7 +3060,7 @@ static bool lower_int_dot(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void if (!(comps[i] = hlsl_new_swizzle(ctx, s, 1, mult, &instr->loc))) return false; - list_add_before(&instr->entry, &comps[i]->entry); + hlsl_block_add_instr(block, comps[i]); } res = comps[0]; @@ -2754,10 +3068,9 @@ static bool lower_int_dot(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void { if (!(res = hlsl_new_binary_expr(ctx, is_bool ? HLSL_OP2_LOGIC_OR : HLSL_OP2_ADD, res, comps[i]))) return false; - list_add_before(&instr->entry, &res->entry); + hlsl_block_add_instr(block, res); } - hlsl_replace_node(instr, res); return true; } @@ -2922,6 +3235,7 @@ static bool dce(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) case HLSL_IR_JUMP: case HLSL_IR_LOOP: case HLSL_IR_RESOURCE_STORE: + case HLSL_IR_SWITCH: break; } @@ -2949,26 +3263,45 @@ static unsigned int index_instructions(struct hlsl_block *block, unsigned int in index = index_instructions(&hlsl_ir_loop(instr)->body, index); hlsl_ir_loop(instr)->next_index = index; } + else if (instr->type == HLSL_IR_SWITCH) + { + struct hlsl_ir_switch *s = hlsl_ir_switch(instr); + struct hlsl_ir_switch_case *c; + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + index = index_instructions(&c->body, index); + } + } } return index; } -static void dump_function_decl(struct rb_entry *entry, void *context) +static void dump_function(struct rb_entry *entry, void *context) { - struct hlsl_ir_function_decl *func = RB_ENTRY_VALUE(entry, struct hlsl_ir_function_decl, entry); + struct hlsl_ir_function *func = RB_ENTRY_VALUE(entry, struct hlsl_ir_function, entry); + struct hlsl_ir_function_decl *decl; struct hlsl_ctx *ctx = context; - if (func->has_body) - hlsl_dump_function(ctx, func); + LIST_FOR_EACH_ENTRY(decl, &func->overloads, struct hlsl_ir_function_decl, entry) + { + if (decl->has_body) + hlsl_dump_function(ctx, decl); + } } -static void dump_function(struct rb_entry *entry, void *context) +static bool mark_indexable_vars(struct hlsl_ctx *ctx, struct hlsl_deref *deref, + struct hlsl_ir_node *instr) { - struct hlsl_ir_function *func = RB_ENTRY_VALUE(entry, struct hlsl_ir_function, entry); - struct hlsl_ctx *ctx = context; + if (!deref->rel_offset.node) + return false; + + assert(deref->var); + assert(deref->rel_offset.node->type != HLSL_IR_CONSTANT); + deref->var->indexable = true; - rb_for_each_entry(&func->overloads, dump_function_decl, ctx); + return true; } static char get_regset_name(enum hlsl_regset regset) @@ -2993,31 +3326,39 @@ static void allocate_register_reservations(struct hlsl_ctx *ctx) LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry) { - enum hlsl_regset regset; + unsigned int r; if (!hlsl_type_is_resource(var->data_type)) continue; - regset = hlsl_type_get_regset(var->data_type); - if (var->reg_reservation.reg_type && var->regs[regset].allocation_size) + if (var->reg_reservation.reg_type) { - if (var->reg_reservation.reg_type != get_regset_name(regset)) - { - struct vkd3d_string_buffer *type_string; - - type_string = hlsl_type_to_string(ctx, var->data_type); - hlsl_error(ctx, &var->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_RESERVATION, - "Object of type '%s' must be bound to register type '%c'.", - type_string->buffer, get_regset_name(regset)); - hlsl_release_string_buffer(ctx, type_string); - } - else + for (r = 0; r <= HLSL_REGSET_LAST_OBJECT; ++r) { - var->regs[regset].allocated = true; - var->regs[regset].id = var->reg_reservation.reg_index; - TRACE("Allocated reserved %s to %c%u-%c%u.\n", var->name, var->reg_reservation.reg_type, - var->reg_reservation.reg_index, var->reg_reservation.reg_type, - var->reg_reservation.reg_index + var->regs[regset].allocation_size); + if (var->regs[r].allocation_size > 0) + { + if (var->reg_reservation.reg_type != get_regset_name(r)) + { + struct vkd3d_string_buffer *type_string; + + /* We can throw this error because resources can only span across a single + * regset, but we have to check for multiple regsets if we support register + * reservations for structs for SM5. */ + type_string = hlsl_type_to_string(ctx, var->data_type); + hlsl_error(ctx, &var->loc, VKD3D_SHADER_ERROR_HLSL_INVALID_RESERVATION, + "Object of type '%s' must be bound to register type '%c'.", + type_string->buffer, get_regset_name(r)); + hlsl_release_string_buffer(ctx, type_string); + } + else + { + var->regs[r].allocated = true; + var->regs[r].id = var->reg_reservation.reg_index; + TRACE("Allocated reserved %s to %c%u-%c%u.\n", var->name, var->reg_reservation.reg_type, + var->reg_reservation.reg_index, var->reg_reservation.reg_type, + var->reg_reservation.reg_index + var->regs[r].allocation_size); + } + } } } } @@ -3051,8 +3392,8 @@ static void compute_liveness_recurse(struct hlsl_block *block, unsigned int loop if (!var->first_write) var->first_write = loop_first ? min(instr->index, loop_first) : instr->index; store->rhs.node->last_read = last_read; - if (store->lhs.offset.node) - store->lhs.offset.node->last_read = last_read; + if (store->lhs.rel_offset.node) + store->lhs.rel_offset.node->last_read = last_read; break; } case HLSL_IR_EXPR: @@ -3079,8 +3420,8 @@ static void compute_liveness_recurse(struct hlsl_block *block, unsigned int loop var = load->src.var; var->last_read = max(var->last_read, last_read); - if (load->src.offset.node) - load->src.offset.node->last_read = last_read; + if (load->src.rel_offset.node) + load->src.rel_offset.node->last_read = last_read; break; } case HLSL_IR_LOOP: @@ -3097,14 +3438,14 @@ static void compute_liveness_recurse(struct hlsl_block *block, unsigned int loop var = load->resource.var; var->last_read = max(var->last_read, last_read); - if (load->resource.offset.node) - load->resource.offset.node->last_read = last_read; + if (load->resource.rel_offset.node) + load->resource.rel_offset.node->last_read = last_read; if ((var = load->sampler.var)) { var->last_read = max(var->last_read, last_read); - if (load->sampler.offset.node) - load->sampler.offset.node->last_read = last_read; + if (load->sampler.rel_offset.node) + load->sampler.rel_offset.node->last_read = last_read; } if (load->coords.node) @@ -3129,8 +3470,8 @@ static void compute_liveness_recurse(struct hlsl_block *block, unsigned int loop var = store->resource.var; var->last_read = max(var->last_read, last_read); - if (store->resource.offset.node) - store->resource.offset.node->last_read = last_read; + if (store->resource.rel_offset.node) + store->resource.rel_offset.node->last_read = last_read; store->coords.node->last_read = last_read; store->value.node->last_read = last_read; break; @@ -3158,6 +3499,16 @@ static void compute_liveness_recurse(struct hlsl_block *block, unsigned int loop jump->condition.node->last_read = last_read; break; } + case HLSL_IR_SWITCH: + { + struct hlsl_ir_switch *s = hlsl_ir_switch(instr); + struct hlsl_ir_switch_case *c; + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + compute_liveness_recurse(&c->body, loop_first, loop_last); + s->selector.node->last_read = last_read; + break; + } case HLSL_IR_CONSTANT: break; } @@ -3191,18 +3542,20 @@ static void compute_liveness(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl struct register_allocator { - size_t count, capacity; - - /* Highest register index that has been allocated. - * Used to declare sm4 temp count. */ - uint32_t max_reg; - struct allocation { uint32_t reg; unsigned int writemask; unsigned int first_write, last_read; } *allocations; + size_t count, capacity; + + /* Indexable temps are allocated separately and always keep their index regardless of their + * lifetime. */ + size_t indexable_count; + + /* Total number of registers allocated so far. Used to declare sm4 temp count. */ + uint32_t reg_count; }; static unsigned int get_available_writemask(const struct register_allocator *allocator, @@ -3245,7 +3598,7 @@ static void record_allocation(struct hlsl_ctx *ctx, struct register_allocator *a allocation->first_write = first_write; allocation->last_read = last_read; - allocator->max_reg = max(allocator->max_reg, reg_idx); + allocator->reg_count = max(allocator->reg_count, reg_idx + 1); } /* reg_size is the number of register components to be reserved, while component_count is the number @@ -3284,13 +3637,19 @@ static struct hlsl_reg allocate_register(struct hlsl_ctx *ctx, struct register_a static bool is_range_available(const struct register_allocator *allocator, unsigned int first_write, unsigned int last_read, uint32_t reg_idx, unsigned int reg_size) { + unsigned int last_reg_mask = (1u << (reg_size % 4)) - 1; + unsigned int writemask; uint32_t i; for (i = 0; i < (reg_size / 4); ++i) { - if (get_available_writemask(allocator, first_write, last_read, reg_idx + i) != VKD3DSP_WRITEMASK_ALL) + writemask = get_available_writemask(allocator, first_write, last_read, reg_idx + i); + if (writemask != VKD3DSP_WRITEMASK_ALL) return false; } + writemask = get_available_writemask(allocator, first_write, last_read, reg_idx + (reg_size / 4)); + if ((writemask & last_reg_mask) != last_reg_mask) + return false; return true; } @@ -3309,6 +3668,8 @@ static struct hlsl_reg allocate_range(struct hlsl_ctx *ctx, struct register_allo for (i = 0; i < reg_size / 4; ++i) record_allocation(ctx, allocator, reg_idx + i, VKD3DSP_WRITEMASK_ALL, first_write, last_read); + if (reg_size % 4) + record_allocation(ctx, allocator, reg_idx + (reg_size / 4), (1u << (reg_size % 4)) - 1, first_write, last_read); ret.id = reg_idx; ret.allocation_size = align(reg_size, 4) / 4; @@ -3356,7 +3717,7 @@ static bool track_object_components_sampler_dim(struct hlsl_ctx *ctx, struct hls load = hlsl_ir_resource_load(instr); var = load->resource.var; - regset = hlsl_type_get_regset(hlsl_deref_get_type(ctx, &load->resource)); + regset = hlsl_deref_get_regset(ctx, &load->resource); if (!hlsl_regset_index_from_deref(ctx, &load->resource, regset, &index)) return false; @@ -3401,7 +3762,8 @@ static bool track_object_components_usage(struct hlsl_ctx *ctx, struct hlsl_ir_n load = hlsl_ir_resource_load(instr); var = load->resource.var; - regset = hlsl_type_get_regset(hlsl_deref_get_type(ctx, &load->resource)); + regset = hlsl_deref_get_regset(ctx, &load->resource); + if (!hlsl_regset_index_from_deref(ctx, &load->resource, regset, &index)) return false; @@ -3448,11 +3810,23 @@ static void allocate_variable_temp_register(struct hlsl_ctx *ctx, if (!var->regs[HLSL_REGSET_NUMERIC].allocated && var->last_read) { - var->regs[HLSL_REGSET_NUMERIC] = allocate_numeric_registers_for_type(ctx, allocator, - var->first_write, var->last_read, var->data_type); + if (var->indexable) + { + var->regs[HLSL_REGSET_NUMERIC].id = allocator->indexable_count++; + var->regs[HLSL_REGSET_NUMERIC].allocation_size = 1; + var->regs[HLSL_REGSET_NUMERIC].writemask = 0; + var->regs[HLSL_REGSET_NUMERIC].allocated = true; + + TRACE("Allocated %s to x%u[].\n", var->name, var->regs[HLSL_REGSET_NUMERIC].id); + } + else + { + var->regs[HLSL_REGSET_NUMERIC] = allocate_numeric_registers_for_type(ctx, allocator, + var->first_write, var->last_read, var->data_type); - TRACE("Allocated %s to %s (liveness %u-%u).\n", var->name, debug_register('r', - var->regs[HLSL_REGSET_NUMERIC], var->data_type), var->first_write, var->last_read); + TRACE("Allocated %s to %s (liveness %u-%u).\n", var->name, debug_register('r', + var->regs[HLSL_REGSET_NUMERIC], var->data_type), var->first_write, var->last_read); + } } } @@ -3508,6 +3882,18 @@ static void allocate_temp_registers_recurse(struct hlsl_ctx *ctx, break; } + case HLSL_IR_SWITCH: + { + struct hlsl_ir_switch *s = hlsl_ir_switch(instr); + struct hlsl_ir_switch_case *c; + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + allocate_temp_registers_recurse(ctx, &c->body, allocator); + } + break; + } + default: break; } @@ -3556,7 +3942,7 @@ static void allocate_const_registers_recurse(struct hlsl_ctx *ctx, constant->reg = allocate_numeric_registers_for_type(ctx, allocator, 1, UINT_MAX, type); TRACE("Allocated constant @%u to %s.\n", instr->index, debug_register('c', constant->reg, type)); - assert(type->class <= HLSL_CLASS_LAST_NUMERIC); + assert(hlsl_is_numeric_type(type)); assert(type->dimy == 1); assert(constant->reg.writemask); @@ -3617,6 +4003,18 @@ static void allocate_const_registers_recurse(struct hlsl_ctx *ctx, break; } + case HLSL_IR_SWITCH: + { + struct hlsl_ir_switch *s = hlsl_ir_switch(instr); + struct hlsl_ir_switch_case *c; + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + allocate_const_registers_recurse(ctx, &c->body, allocator); + } + break; + } + default: break; } @@ -3675,7 +4073,7 @@ static void allocate_temp_registers(struct hlsl_ctx *ctx, struct hlsl_ir_functio } allocate_temp_registers_recurse(ctx, &entry_func->body, &allocator); - ctx->temp_count = allocator.max_reg + 1; + ctx->temp_count = allocator.reg_count; vkd3d_free(allocator.allocations); } @@ -3728,7 +4126,7 @@ static void allocate_semantic_register(struct hlsl_ctx *ctx, struct hlsl_ir_var "Invalid semantic '%s'.", var->semantic.name); return; } - if ((builtin = hlsl_sm4_register_from_semantic(ctx, &var->semantic, output, &type, NULL, &has_idx))) + if ((builtin = hlsl_sm4_register_from_semantic(ctx, &var->semantic, output, &type, &has_idx))) reg = has_idx ? var->semantic.index : 0; } @@ -4190,30 +4588,25 @@ bool hlsl_regset_index_from_deref(struct hlsl_ctx *ctx, const struct hlsl_deref bool hlsl_offset_from_deref(struct hlsl_ctx *ctx, const struct hlsl_deref *deref, unsigned int *offset) { - struct hlsl_ir_node *offset_node = deref->offset.node; - enum hlsl_regset regset; + enum hlsl_regset regset = hlsl_deref_get_regset(ctx, deref); + struct hlsl_ir_node *offset_node = deref->rel_offset.node; unsigned int size; - if (!offset_node) - { - *offset = 0; - return true; - } + *offset = deref->const_offset; - /* We should always have generated a cast to UINT. */ - assert(offset_node->data_type->class == HLSL_CLASS_SCALAR - && offset_node->data_type->base_type == HLSL_TYPE_UINT); - - if (offset_node->type != HLSL_IR_CONSTANT) + if (offset_node) + { + /* We should always have generated a cast to UINT. */ + assert(offset_node->data_type->class == HLSL_CLASS_SCALAR + && offset_node->data_type->base_type == HLSL_TYPE_UINT); + assert(offset_node->type != HLSL_IR_CONSTANT); return false; - - *offset = hlsl_ir_constant(offset_node)->value.u[0].u; - regset = hlsl_type_get_regset(deref->data_type); + } size = deref->var->data_type->reg_size[regset]; if (*offset >= size) { - hlsl_error(ctx, &deref->offset.node->loc, VKD3D_SHADER_ERROR_HLSL_OFFSET_OUT_OF_BOUNDS, + hlsl_error(ctx, &offset_node->loc, VKD3D_SHADER_ERROR_HLSL_OFFSET_OUT_OF_BOUNDS, "Dereference is out of bounds. %u/%u", *offset, size); return false; } @@ -4228,8 +4621,8 @@ unsigned int hlsl_offset_from_deref_safe(struct hlsl_ctx *ctx, const struct hlsl if (hlsl_offset_from_deref(ctx, deref, &offset)) return offset; - hlsl_fixme(ctx, &deref->offset.node->loc, "Dereference with non-constant offset of type %s.", - hlsl_node_type_to_string(deref->offset.node->type)); + hlsl_fixme(ctx, &deref->rel_offset.node->loc, "Dereference with non-constant offset of type %s.", + hlsl_node_type_to_string(deref->rel_offset.node->type)); return 0; } @@ -4241,7 +4634,7 @@ struct hlsl_reg hlsl_reg_from_deref(struct hlsl_ctx *ctx, const struct hlsl_dere unsigned int offset = hlsl_offset_from_deref_safe(ctx, deref); assert(deref->data_type); - assert(deref->data_type->class <= HLSL_CLASS_LAST_NUMERIC); + assert(hlsl_is_numeric_type(deref->data_type)); ret.id += offset / 4; @@ -4319,6 +4712,62 @@ static bool type_has_object_components(struct hlsl_type *type) return false; } +static void remove_unreachable_code(struct hlsl_ctx *ctx, struct hlsl_block *body) +{ + struct hlsl_ir_node *instr, *next; + struct hlsl_block block; + struct list *start; + + LIST_FOR_EACH_ENTRY_SAFE(instr, next, &body->instrs, struct hlsl_ir_node, entry) + { + if (instr->type == HLSL_IR_IF) + { + struct hlsl_ir_if *iff = hlsl_ir_if(instr); + + remove_unreachable_code(ctx, &iff->then_block); + remove_unreachable_code(ctx, &iff->else_block); + } + else if (instr->type == HLSL_IR_LOOP) + { + struct hlsl_ir_loop *loop = hlsl_ir_loop(instr); + + remove_unreachable_code(ctx, &loop->body); + } + else if (instr->type == HLSL_IR_SWITCH) + { + struct hlsl_ir_switch *s = hlsl_ir_switch(instr); + struct hlsl_ir_switch_case *c; + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + remove_unreachable_code(ctx, &c->body); + } + } + } + + /* Remove instructions past unconditional jumps. */ + LIST_FOR_EACH_ENTRY(instr, &body->instrs, struct hlsl_ir_node, entry) + { + struct hlsl_ir_jump *jump; + + if (instr->type != HLSL_IR_JUMP) + continue; + + jump = hlsl_ir_jump(instr); + if (jump->type != HLSL_IR_JUMP_BREAK && jump->type != HLSL_IR_JUMP_CONTINUE) + continue; + + if (!(start = list_next(&body->instrs, &instr->entry))) + break; + + hlsl_block_init(&block); + list_move_slice_tail(&block.instrs, start, list_tail(&body->instrs)); + hlsl_block_cleanup(&block); + + break; + } +} + int hlsl_emit_bytecode(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry_func, enum vkd3d_shader_target_type target_type, struct vkd3d_shader_code *out) { @@ -4345,7 +4794,7 @@ int hlsl_emit_bytecode(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry while (hlsl_transform_ir(ctx, lower_calls, body, NULL)); lower_ir(ctx, lower_matrix_swizzles, body); - hlsl_transform_ir(ctx, lower_index_loads, body, NULL); + lower_ir(ctx, lower_index_loads, body); LIST_FOR_EACH_ENTRY(var, &ctx->globals->vars, struct hlsl_ir_var, scope_entry) { @@ -4408,7 +4857,7 @@ int hlsl_emit_bytecode(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry { hlsl_transform_ir(ctx, lower_discard_neg, body, NULL); } - hlsl_transform_ir(ctx, lower_broadcasts, body, NULL); + lower_ir(ctx, lower_broadcasts, body); while (hlsl_transform_ir(ctx, fold_redundant_casts, body, NULL)); do { @@ -4418,12 +4867,12 @@ int hlsl_emit_bytecode(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry while (progress); hlsl_transform_ir(ctx, split_matrix_copies, body, NULL); - hlsl_transform_ir(ctx, lower_narrowing_casts, body, NULL); - hlsl_transform_ir(ctx, lower_casts_to_bool, body, NULL); - hlsl_transform_ir(ctx, lower_int_dot, body, NULL); + lower_ir(ctx, lower_narrowing_casts, body); + lower_ir(ctx, lower_casts_to_bool, body); + lower_ir(ctx, lower_int_dot, body); lower_ir(ctx, lower_int_division, body); lower_ir(ctx, lower_int_modulus, body); - hlsl_transform_ir(ctx, lower_int_abs, body, NULL); + lower_ir(ctx, lower_int_abs, body); lower_ir(ctx, lower_float_modulus, body); hlsl_transform_ir(ctx, fold_redundant_casts, body, NULL); do @@ -4433,12 +4882,15 @@ int hlsl_emit_bytecode(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry progress |= hlsl_copy_propagation_execute(ctx, body); progress |= hlsl_transform_ir(ctx, fold_swizzle_chains, body, NULL); progress |= hlsl_transform_ir(ctx, remove_trivial_swizzles, body, NULL); + progress |= hlsl_transform_ir(ctx, remove_trivial_conditional_branches, body, NULL); } while (progress); + remove_unreachable_code(ctx, body); + hlsl_transform_ir(ctx, normalize_switch_cases, body, NULL); - hlsl_transform_ir(ctx, lower_nonconstant_vector_derefs, body, NULL); - hlsl_transform_ir(ctx, lower_casts_to_bool, body, NULL); - hlsl_transform_ir(ctx, lower_int_dot, body, NULL); + lower_ir(ctx, lower_nonconstant_vector_derefs, body); + lower_ir(ctx, lower_casts_to_bool, body); + lower_ir(ctx, lower_int_dot, body); hlsl_transform_ir(ctx, validate_static_object_references, body, NULL); hlsl_transform_ir(ctx, track_object_components_sampler_dim, body, NULL); @@ -4447,24 +4899,26 @@ int hlsl_emit_bytecode(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry hlsl_transform_ir(ctx, track_object_components_usage, body, NULL); sort_synthetic_separated_samplers_first(ctx); - if (profile->major_version >= 4) - hlsl_transform_ir(ctx, lower_ternary, body, NULL); + lower_ir(ctx, lower_ternary, body); if (profile->major_version < 4) { - hlsl_transform_ir(ctx, lower_division, body, NULL); - hlsl_transform_ir(ctx, lower_sqrt, body, NULL); - hlsl_transform_ir(ctx, lower_dot, body, NULL); - hlsl_transform_ir(ctx, lower_round, body, NULL); + lower_ir(ctx, lower_division, body); + lower_ir(ctx, lower_sqrt, body); + lower_ir(ctx, lower_dot, body); + lower_ir(ctx, lower_round, body); + lower_ir(ctx, lower_ceil, body); + lower_ir(ctx, lower_floor, body); } if (profile->major_version < 2) { - hlsl_transform_ir(ctx, lower_abs, body, NULL); + lower_ir(ctx, lower_abs, body); } /* TODO: move forward, remove when no longer needed */ transform_derefs(ctx, replace_deref_path_with_offset, body); while (hlsl_transform_ir(ctx, hlsl_fold_constant_exprs, body, NULL)); + transform_derefs(ctx, clean_constant_deref_offset_srcs, body); do compute_liveness(ctx, entry_func); @@ -4475,6 +4929,8 @@ int hlsl_emit_bytecode(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry if (TRACE_ON()) rb_for_each_entry(&ctx->functions, dump_function, ctx); + transform_derefs(ctx, mark_indexable_vars, body); + calculate_resource_register_counts(ctx); allocate_register_reservations(ctx); diff --git a/libs/vkd3d/libs/vkd3d-shader/hlsl_constant_ops.c b/libs/vkd3d/libs/vkd3d-shader/hlsl_constant_ops.c index 41a72ab6c0d..b76b1fce507 100644 --- a/libs/vkd3d/libs/vkd3d-shader/hlsl_constant_ops.c +++ b/libs/vkd3d/libs/vkd3d-shader/hlsl_constant_ops.c @@ -63,6 +63,82 @@ static bool fold_abs(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, return true; } +static uint32_t float_to_uint(float x) +{ + if (isnan(x) || x <= 0) + return 0; + + if (x >= 4294967296.0f) + return UINT32_MAX; + + return x; +} + +static int32_t float_to_int(float x) +{ + if (isnan(x)) + return 0; + + if (x <= -2147483648.0f) + return INT32_MIN; + + if (x >= 2147483648.0f) + return INT32_MAX; + + return x; +} + +static uint32_t double_to_uint(double x) +{ + if (isnan(x) || x <= 0) + return 0; + + if (x >= 4294967296.0) + return UINT32_MAX; + + return x; +} + +static int32_t double_to_int(double x) +{ + if (isnan(x)) + return 0; + + if (x <= -2147483648.0) + return INT32_MIN; + + if (x >= 2147483648.0) + return INT32_MAX; + + return x; +} + +static bool fold_bit_not(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, + const struct hlsl_type *dst_type, const struct hlsl_ir_constant *src) +{ + enum hlsl_base_type type = dst_type->base_type; + unsigned int k; + + assert(type == src->node.data_type->base_type); + + for (k = 0; k < dst_type->dimx; ++k) + { + switch (type) + { + case HLSL_TYPE_INT: + case HLSL_TYPE_UINT: + case HLSL_TYPE_BOOL: + dst->u[k].u = ~src->value.u[k].u; + break; + + default: + vkd3d_unreachable(); + } + } + + return true; +} + static bool fold_cast(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, const struct hlsl_type *dst_type, const struct hlsl_ir_constant *src) { @@ -86,15 +162,15 @@ static bool fold_cast(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, { case HLSL_TYPE_FLOAT: case HLSL_TYPE_HALF: - u = src->value.u[k].f; - i = src->value.u[k].f; + u = float_to_uint(src->value.u[k].f); + i = float_to_int(src->value.u[k].f); f = src->value.u[k].f; d = src->value.u[k].f; break; case HLSL_TYPE_DOUBLE: - u = src->value.u[k].d; - i = src->value.u[k].d; + u = double_to_uint(src->value.u[k].d); + i = double_to_int(src->value.u[k].d); f = src->value.u[k].d; d = src->value.u[k].d; break; @@ -152,6 +228,111 @@ static bool fold_cast(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, return true; } +static bool fold_ceil(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, + const struct hlsl_type *dst_type, const struct hlsl_ir_constant *src) +{ + enum hlsl_base_type type = dst_type->base_type; + unsigned int k; + + assert(type == src->node.data_type->base_type); + + for (k = 0; k < dst_type->dimx; ++k) + { + switch (type) + { + case HLSL_TYPE_FLOAT: + case HLSL_TYPE_HALF: + dst->u[k].f = ceilf(src->value.u[k].f); + break; + + default: + FIXME("Fold 'ceil' for type %s.\n", debug_hlsl_type(ctx, dst_type)); + return false; + } + } + + return true; +} + +static bool fold_exp2(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, + const struct hlsl_type *dst_type, const struct hlsl_ir_constant *src) +{ + enum hlsl_base_type type = dst_type->base_type; + unsigned int k; + + assert(type == src->node.data_type->base_type); + + for (k = 0; k < dst_type->dimx; ++k) + { + switch (type) + { + case HLSL_TYPE_FLOAT: + case HLSL_TYPE_HALF: + dst->u[k].f = exp2f(src->value.u[k].f); + break; + + default: + FIXME("Fold 'exp2' for type %s.\n", debug_hlsl_type(ctx, dst_type)); + return false; + } + } + + return true; +} + +static bool fold_floor(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, + const struct hlsl_type *dst_type, const struct hlsl_ir_constant *src) +{ + enum hlsl_base_type type = dst_type->base_type; + unsigned int k; + + assert(type == src->node.data_type->base_type); + + for (k = 0; k < dst_type->dimx; ++k) + { + switch (type) + { + case HLSL_TYPE_FLOAT: + case HLSL_TYPE_HALF: + dst->u[k].f = floorf(src->value.u[k].f); + break; + + default: + FIXME("Fold 'floor' for type %s.\n", debug_hlsl_type(ctx, dst_type)); + return false; + } + } + + return true; +} + +static bool fold_fract(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, + const struct hlsl_type *dst_type, const struct hlsl_ir_constant *src) +{ + enum hlsl_base_type type = dst_type->base_type; + unsigned int k; + float i; + + assert(type == src->node.data_type->base_type); + + for (k = 0; k < dst_type->dimx; ++k) + { + switch (type) + { + case HLSL_TYPE_FLOAT: + case HLSL_TYPE_HALF: + dst->u[k].f = modff(src->value.u[k].f, &i); + break; + + default: + FIXME("Fold 'fract' for type %s.\n", debug_hlsl_type(ctx, dst_type)); + return false; + } + } + + return true; +} + static bool fold_log2(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, const struct hlsl_type *dst_type, const struct hlsl_ir_constant *src, const struct vkd3d_shader_location *loc) { @@ -231,6 +412,30 @@ static bool fold_neg(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, return true; } +static bool fold_not(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, + const struct hlsl_type *dst_type, const struct hlsl_ir_constant *src) +{ + enum hlsl_base_type type = dst_type->base_type; + unsigned int k; + + assert(type == src->node.data_type->base_type); + + for (k = 0; k < dst_type->dimx; ++k) + { + switch (type) + { + case HLSL_TYPE_BOOL: + dst->u[k].u = ~src->value.u[k].u; + break; + + default: + FIXME("Fold logic 'not' for type %s.\n", debug_hlsl_type(ctx, dst_type)); + return false; + } + } + return true; +} + static bool fold_rcp(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, const struct hlsl_type *dst_type, const struct hlsl_ir_constant *src, const struct vkd3d_shader_location *loc) { @@ -276,6 +481,72 @@ static bool fold_rcp(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, cons return true; } +static bool fold_rsq(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, const struct hlsl_type *dst_type, + const struct hlsl_ir_constant *src, const struct vkd3d_shader_location *loc) +{ + enum hlsl_base_type type = dst_type->base_type; + unsigned int k; + + assert(type == src->node.data_type->base_type); + + for (k = 0; k < dst_type->dimx; ++k) + { + switch (type) + { + case HLSL_TYPE_FLOAT: + case HLSL_TYPE_HALF: + if (ctx->profile->major_version >= 4) + { + if (src->value.u[k].f < 0.0f) + hlsl_warning(ctx, loc, VKD3D_SHADER_WARNING_HLSL_IMAGINARY_NUMERIC_RESULT, + "Imaginary square root result."); + else if (src->value.u[k].f == 0.0f) + hlsl_warning(ctx, loc, VKD3D_SHADER_WARNING_HLSL_IMAGINARY_NUMERIC_RESULT, + "Floating point division by zero."); + } + dst->u[k].f = 1.0f / sqrtf(src->value.u[k].f); + if (ctx->profile->major_version < 4 && !isfinite(dst->u[k].f)) + { + hlsl_error(ctx, loc, VKD3D_SHADER_ERROR_HLSL_NON_FINITE_RESULT, + "Infinities and NaNs are not allowed by the shader model."); + } + break; + + default: + FIXME("Fold 'rsq' for type %s.\n", debug_hlsl_type(ctx, dst_type)); + return false; + } + } + + return true; +} + +static bool fold_sat(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, const struct hlsl_type *dst_type, + const struct hlsl_ir_constant *src) +{ + enum hlsl_base_type type = dst_type->base_type; + unsigned int k; + + assert(type == src->node.data_type->base_type); + + for (k = 0; k < dst_type->dimx; ++k) + { + switch (type) + { + case HLSL_TYPE_FLOAT: + case HLSL_TYPE_HALF: + dst->u[k].f = min(max(0.0f, src->value.u[k].f), 1.0f); + break; + + default: + FIXME("Fold 'sat' for type %s.\n", debug_hlsl_type(ctx, dst_type)); + return false; + } + } + + return true; +} + static bool fold_sqrt(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, const struct hlsl_type *dst_type, const struct hlsl_ir_constant *src, const struct vkd3d_shader_location *loc) { @@ -679,6 +950,36 @@ static bool fold_less(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, con return true; } +static bool fold_lshift(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, const struct hlsl_type *dst_type, + const struct hlsl_ir_constant *src1, const struct hlsl_ir_constant *src2) +{ + unsigned int k; + + assert(dst_type->base_type == src1->node.data_type->base_type); + assert(src2->node.data_type->base_type == HLSL_TYPE_INT); + + for (k = 0; k < dst_type->dimx; ++k) + { + unsigned int shift = src2->value.u[k].u % 32; + + switch (src1->node.data_type->base_type) + { + case HLSL_TYPE_INT: + dst->u[k].i = src1->value.u[k].i << shift; + break; + + case HLSL_TYPE_UINT: + dst->u[k].u = src1->value.u[k].u << shift; + break; + + default: + vkd3d_unreachable(); + } + } + + return true; +} + static bool fold_max(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, const struct hlsl_type *dst_type, const struct hlsl_ir_constant *src1, const struct hlsl_ir_constant *src2) { @@ -869,6 +1170,70 @@ static bool fold_nequal(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, c return true; } +static bool fold_ternary(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, const struct hlsl_type *dst_type, + const struct hlsl_ir_constant *src1, const struct hlsl_ir_constant *src2, const struct hlsl_ir_constant *src3) +{ + unsigned int k; + + assert(dst_type->base_type == src2->node.data_type->base_type); + assert(dst_type->base_type == src3->node.data_type->base_type); + + for (k = 0; k < dst_type->dimx; ++k) + { + switch (src1->node.data_type->base_type) + { + case HLSL_TYPE_FLOAT: + case HLSL_TYPE_HALF: + dst->u[k] = src1->value.u[k].f != 0.0f ? src2->value.u[k] : src3->value.u[k]; + break; + + case HLSL_TYPE_DOUBLE: + dst->u[k] = src1->value.u[k].d != 0.0 ? src2->value.u[k] : src3->value.u[k]; + break; + + case HLSL_TYPE_INT: + case HLSL_TYPE_UINT: + case HLSL_TYPE_BOOL: + dst->u[k] = src1->value.u[k].u ? src2->value.u[k] : src3->value.u[k]; + break; + + default: + vkd3d_unreachable(); + } + } + return true; +} + +static bool fold_rshift(struct hlsl_ctx *ctx, struct hlsl_constant_value *dst, const struct hlsl_type *dst_type, + const struct hlsl_ir_constant *src1, const struct hlsl_ir_constant *src2) +{ + unsigned int k; + + assert(dst_type->base_type == src1->node.data_type->base_type); + assert(src2->node.data_type->base_type == HLSL_TYPE_INT); + + for (k = 0; k < dst_type->dimx; ++k) + { + unsigned int shift = src2->value.u[k].u % 32; + + switch (src1->node.data_type->base_type) + { + case HLSL_TYPE_INT: + dst->u[k].i = src1->value.u[k].i >> shift; + break; + + case HLSL_TYPE_UINT: + dst->u[k].u = src1->value.u[k].u >> shift; + break; + + default: + vkd3d_unreachable(); + } + } + + return true; +} + bool hlsl_fold_constant_exprs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, void *context) { struct hlsl_ir_constant *arg1, *arg2 = NULL, *arg3 = NULL; @@ -908,10 +1273,30 @@ bool hlsl_fold_constant_exprs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, success = fold_abs(ctx, &res, instr->data_type, arg1); break; + case HLSL_OP1_BIT_NOT: + success = fold_bit_not(ctx, &res, instr->data_type, arg1); + break; + case HLSL_OP1_CAST: success = fold_cast(ctx, &res, instr->data_type, arg1); break; + case HLSL_OP1_CEIL: + success = fold_ceil(ctx, &res, instr->data_type, arg1); + break; + + case HLSL_OP1_EXP2: + success = fold_exp2(ctx, &res, instr->data_type, arg1); + break; + + case HLSL_OP1_FLOOR: + success = fold_floor(ctx, &res, instr->data_type, arg1); + break; + + case HLSL_OP1_FRACT: + success = fold_fract(ctx, &res, instr->data_type, arg1); + break; + case HLSL_OP1_LOG2: success = fold_log2(ctx, &res, instr->data_type, arg1, &instr->loc); break; @@ -920,10 +1305,22 @@ bool hlsl_fold_constant_exprs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, success = fold_neg(ctx, &res, instr->data_type, arg1); break; + case HLSL_OP1_LOGIC_NOT: + success = fold_not(ctx, &res, instr->data_type, arg1); + break; + case HLSL_OP1_RCP: success = fold_rcp(ctx, &res, instr->data_type, arg1, &instr->loc); break; + case HLSL_OP1_RSQ: + success = fold_rsq(ctx, &res, instr->data_type, arg1, &instr->loc); + break; + + case HLSL_OP1_SAT: + success = fold_sat(ctx, &res, instr->data_type, arg1); + break; + case HLSL_OP1_SQRT: success = fold_sqrt(ctx, &res, instr->data_type, arg1, &instr->loc); break; @@ -966,6 +1363,10 @@ bool hlsl_fold_constant_exprs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, success = fold_less(ctx, &res, instr->data_type, arg1, arg2); break; + case HLSL_OP2_LSHIFT: + success = fold_lshift(ctx, &res, instr->data_type, arg1, arg2); + break; + case HLSL_OP2_MAX: success = fold_max(ctx, &res, instr->data_type, arg1, arg2); break; @@ -986,10 +1387,18 @@ bool hlsl_fold_constant_exprs(struct hlsl_ctx *ctx, struct hlsl_ir_node *instr, success = fold_nequal(ctx, &res, instr->data_type, arg1, arg2); break; + case HLSL_OP2_RSHIFT: + success = fold_rshift(ctx, &res, instr->data_type, arg1, arg2); + break; + case HLSL_OP3_DP2ADD: success = fold_dp2add(ctx, &res, instr->data_type, arg1, arg2, arg3); break; + case HLSL_OP3_TERNARY: + success = fold_ternary(ctx, &res, instr->data_type, arg1, arg2, arg3); + break; + default: FIXME("Fold \"%s\" expression.\n", debug_hlsl_expr_op(expr->op)); success = false; diff --git a/libs/vkd3d/libs/vkd3d-shader/ir.c b/libs/vkd3d/libs/vkd3d-shader/ir.c index d2bfb933edc..acdb660ea82 100644 --- a/libs/vkd3d/libs/vkd3d-shader/ir.c +++ b/libs/vkd3d/libs/vkd3d-shader/ir.c @@ -31,11 +31,9 @@ static bool shader_instruction_is_dcl(const struct vkd3d_shader_instruction *ins static void vkd3d_shader_instruction_make_nop(struct vkd3d_shader_instruction *ins) { - ins->handler_idx = VKD3DSIH_NOP; - ins->dst_count = 0; - ins->src_count = 0; - ins->dst = NULL; - ins->src = NULL; + struct vkd3d_shader_location location = ins->location; + + vsir_instruction_init(ins, &location, VKD3DSIH_NOP); } static void shader_register_eliminate_phase_addressing(struct vkd3d_shader_register *reg, @@ -64,17 +62,7 @@ static void shader_instruction_eliminate_phase_instance_id(struct vkd3d_shader_i reg = (struct vkd3d_shader_register *)&ins->src[i].reg; if (shader_register_is_phase_instance_id(reg)) { - reg->type = VKD3DSPR_IMMCONST; - reg->precision = VKD3D_SHADER_REGISTER_PRECISION_DEFAULT; - reg->non_uniform = false; - reg->idx[0].offset = ~0u; - reg->idx[0].rel_addr = NULL; - reg->idx[1].offset = ~0u; - reg->idx[1].rel_addr = NULL; - reg->idx[2].offset = ~0u; - reg->idx[2].rel_addr = NULL; - reg->idx_count = 0; - reg->immconst_type = VKD3D_IMMCONST_SCALAR; + vsir_register_init(reg, VKD3DSPR_IMMCONST, reg->data_type, 0); reg->u.immconst_uint[0] = instance_id; continue; } @@ -161,6 +149,7 @@ struct hull_flattener unsigned int instance_count; unsigned int phase_body_idx; enum vkd3d_shader_opcode phase; + struct vkd3d_shader_location last_ret_location; }; static bool flattener_is_in_fork_or_join_phase(const struct hull_flattener *flattener) @@ -218,10 +207,15 @@ static void flattener_eliminate_phase_related_dcls(struct hull_flattener *normal { /* Leave only the first temp declaration and set it to the max count later. */ if (!normaliser->max_temp_count) + { + normaliser->max_temp_count = ins->declaration.count; normaliser->temp_dcl_idx = index; + } else + { + normaliser->max_temp_count = max(normaliser->max_temp_count, ins->declaration.count); vkd3d_shader_instruction_make_nop(ins); - normaliser->max_temp_count = max(normaliser->max_temp_count, ins->declaration.count); + } return; } @@ -233,6 +227,7 @@ static void flattener_eliminate_phase_related_dcls(struct hull_flattener *normal if (ins->handler_idx == VKD3DSIH_RET) { + normaliser->last_ret_location = ins->location; vkd3d_shader_instruction_make_nop(ins); if (locations->count >= ARRAY_SIZE(locations->locations)) { @@ -296,7 +291,7 @@ static enum vkd3d_result flattener_flatten_phases(struct hull_flattener *normali return VKD3D_OK; } -void shader_register_init(struct vkd3d_shader_register *reg, enum vkd3d_shader_register_type reg_type, +void vsir_register_init(struct vkd3d_shader_register *reg, enum vkd3d_shader_register_type reg_type, enum vkd3d_data_type data_type, unsigned int idx_count) { reg->type = reg_type; @@ -305,17 +300,23 @@ void shader_register_init(struct vkd3d_shader_register *reg, enum vkd3d_shader_r reg->data_type = data_type; reg->idx[0].offset = ~0u; reg->idx[0].rel_addr = NULL; + reg->idx[0].is_in_bounds = false; reg->idx[1].offset = ~0u; reg->idx[1].rel_addr = NULL; + reg->idx[1].is_in_bounds = false; reg->idx[2].offset = ~0u; reg->idx[2].rel_addr = NULL; + reg->idx[2].is_in_bounds = false; reg->idx_count = idx_count; - reg->immconst_type = VKD3D_IMMCONST_SCALAR; + reg->dimension = VSIR_DIMENSION_SCALAR; + reg->alignment = 0; } -void shader_instruction_init(struct vkd3d_shader_instruction *ins, enum vkd3d_shader_opcode handler_idx) +void vsir_instruction_init(struct vkd3d_shader_instruction *ins, const struct vkd3d_shader_location *location, + enum vkd3d_shader_opcode handler_idx) { memset(ins, 0, sizeof(*ins)); + ins->location = *location; ins->handler_idx = handler_idx; } @@ -343,7 +344,7 @@ static enum vkd3d_result instruction_array_flatten_hull_shader_phases(struct vkd if (!shader_instruction_array_reserve(&flattener.instructions, flattener.instructions.count + 1)) return VKD3D_ERROR_OUT_OF_MEMORY; - shader_instruction_init(&instructions->elements[instructions->count++], VKD3DSIH_RET); + vsir_instruction_init(&instructions->elements[instructions->count++], &flattener.last_ret_location, VKD3DSIH_RET); } *src_instructions = flattener.instructions; @@ -370,7 +371,7 @@ static struct vkd3d_shader_src_param *instruction_array_create_outpointid_param( if (!(rel_addr = shader_src_param_allocator_get(&instructions->src_params, 1))) return NULL; - shader_register_init(&rel_addr->reg, VKD3DSPR_OUTPOINTID, VKD3D_DATA_UINT, 0); + vsir_register_init(&rel_addr->reg, VKD3DSPR_OUTPOINTID, VKD3D_DATA_UINT, 0); rel_addr->swizzle = 0; rel_addr->modifiers = 0; @@ -400,11 +401,12 @@ static void shader_dst_param_io_init(struct vkd3d_shader_dst_param *param, const param->write_mask = e->mask; param->modifiers = 0; param->shift = 0; - shader_register_init(¶m->reg, reg_type, vkd3d_data_type_from_component_type(e->component_type), idx_count); + vsir_register_init(¶m->reg, reg_type, vkd3d_data_type_from_component_type(e->component_type), idx_count); } static enum vkd3d_result control_point_normaliser_emit_hs_input(struct control_point_normaliser *normaliser, - const struct shader_signature *s, unsigned int input_control_point_count, unsigned int dst) + const struct shader_signature *s, unsigned int input_control_point_count, unsigned int dst, + const struct vkd3d_shader_location *location) { struct vkd3d_shader_instruction *ins; struct vkd3d_shader_dst_param *param; @@ -422,7 +424,7 @@ static enum vkd3d_result control_point_normaliser_emit_hs_input(struct control_p normaliser->instructions.count += count; ins = &normaliser->instructions.elements[dst]; - shader_instruction_init(ins, VKD3DSIH_HS_CONTROL_POINT_PHASE); + vsir_instruction_init(ins, location, VKD3DSIH_HS_CONTROL_POINT_PHASE); ins->flags = 1; ++ins; @@ -434,19 +436,20 @@ static enum vkd3d_result control_point_normaliser_emit_hs_input(struct control_p if (e->sysval_semantic != VKD3D_SHADER_SV_NONE) { - shader_instruction_init(ins, VKD3DSIH_DCL_INPUT_SIV); + vsir_instruction_init(ins, location, VKD3DSIH_DCL_INPUT_SIV); param = &ins->declaration.register_semantic.reg; ins->declaration.register_semantic.sysval_semantic = vkd3d_siv_from_sysval(e->sysval_semantic); } else { - shader_instruction_init(ins, VKD3DSIH_DCL_INPUT); + vsir_instruction_init(ins, location, VKD3DSIH_DCL_INPUT); param = &ins->declaration.dst; } shader_dst_param_io_init(param, e, VKD3DSPR_INPUT, 2); param->reg.idx[0].offset = input_control_point_count; - param->reg.idx[1].offset = i; + param->reg.idx[1].offset = e->register_index; + param->write_mask = e->mask; ++ins; } @@ -511,7 +514,7 @@ static enum vkd3d_result instruction_array_normalise_hull_shader_control_point_i case VKD3DSIH_HS_FORK_PHASE: case VKD3DSIH_HS_JOIN_PHASE: ret = control_point_normaliser_emit_hs_input(&normaliser, input_signature, - input_control_point_count, i); + input_control_point_count, i, &ins->location); *src_instructions = normaliser.instructions; return ret; default: @@ -547,6 +550,8 @@ struct io_normaliser uint8_t input_range_map[MAX_REG_OUTPUT][VKD3D_VEC4_SIZE]; uint8_t output_range_map[MAX_REG_OUTPUT][VKD3D_VEC4_SIZE]; uint8_t pc_range_map[MAX_REG_OUTPUT][VKD3D_VEC4_SIZE]; + + bool use_vocp; }; static bool io_normaliser_is_in_fork_or_join_phase(const struct io_normaliser *normaliser) @@ -578,6 +583,12 @@ static unsigned int shader_signature_find_element_for_reg(const struct shader_si vkd3d_unreachable(); } +struct signature_element *vsir_signature_find_element_for_reg(const struct shader_signature *signature, + unsigned int reg_idx, unsigned int write_mask) +{ + return &signature->elements[shader_signature_find_element_for_reg(signature, reg_idx, write_mask)]; +} + static unsigned int range_map_get_register_count(uint8_t range_map[][VKD3D_VEC4_SIZE], unsigned int register_idx, unsigned int write_mask) { @@ -816,12 +827,6 @@ static bool shader_signature_merge(struct shader_signature *s, uint8_t range_map return true; } -static bool sysval_semantic_is_tess_factor(enum vkd3d_shader_sysval_semantic sysval_semantic) -{ - return sysval_semantic >= VKD3D_SHADER_SV_TESS_FACTOR_QUADEDGE - && sysval_semantic <= VKD3D_SHADER_SV_TESS_FACTOR_LINEDEN; -} - static unsigned int shader_register_normalise_arrayed_addressing(struct vkd3d_shader_register *reg, unsigned int id_idx, unsigned int register_index) { @@ -870,11 +875,13 @@ static bool shader_dst_param_io_normalise(struct vkd3d_shader_dst_param *dst_par else if (reg->type == VKD3DSPR_OUTPUT || dst_param->reg.type == VKD3DSPR_COLOROUT) { signature = normaliser->output_signature; + reg->type = VKD3DSPR_OUTPUT; dcl_params = normaliser->output_dcl_params; } else if (dst_param->reg.type == VKD3DSPR_INCONTROLPOINT || dst_param->reg.type == VKD3DSPR_INPUT) { signature = normaliser->input_signature; + reg->type = VKD3DSPR_INPUT; dcl_params = normaliser->input_dcl_params; } else @@ -922,7 +929,7 @@ static bool shader_dst_param_io_normalise(struct vkd3d_shader_dst_param *dst_par id_idx = 1; } - if ((e->register_count > 1 || sysval_semantic_is_tess_factor(e->sysval_semantic))) + if ((e->register_count > 1 || vsir_sysval_semantic_is_tess_factor(e->sysval_semantic))) { if (is_io_dcl) { @@ -974,15 +981,13 @@ static void shader_src_param_io_normalise(struct vkd3d_shader_src_param *src_par signature = normaliser->patch_constant_signature; break; case VKD3DSPR_INCONTROLPOINT: - if (normaliser->shader_type == VKD3D_SHADER_TYPE_HULL) - reg->type = VKD3DSPR_INPUT; + reg->type = VKD3DSPR_INPUT; /* fall through */ case VKD3DSPR_INPUT: signature = normaliser->input_signature; break; case VKD3DSPR_OUTCONTROLPOINT: - if (normaliser->shader_type == VKD3D_SHADER_TYPE_HULL) - reg->type = VKD3DSPR_OUTPUT; + reg->type = VKD3DSPR_OUTPUT; /* fall through */ case VKD3DSPR_OUTPUT: signature = normaliser->output_signature; @@ -997,7 +1002,7 @@ static void shader_src_param_io_normalise(struct vkd3d_shader_src_param *src_par element_idx = shader_signature_find_element_for_reg(signature, reg_idx, write_mask); e = &signature->elements[element_idx]; - if ((e->register_count > 1 || sysval_semantic_is_tess_factor(e->sysval_semantic))) + if ((e->register_count > 1 || vsir_sysval_semantic_is_tess_factor(e->sysval_semantic))) id_idx = shader_register_normalise_arrayed_addressing(reg, id_idx, e->register_index); reg->idx[id_idx].offset = element_idx; reg->idx_count = id_idx + 1; @@ -1014,7 +1019,6 @@ static void shader_instruction_normalise_io_params(struct vkd3d_shader_instructi struct io_normaliser *normaliser) { struct vkd3d_shader_register *reg; - bool keep = true; unsigned int i; switch (ins->handler_idx) @@ -1023,6 +1027,10 @@ static void shader_instruction_normalise_io_params(struct vkd3d_shader_instructi if (normaliser->shader_type == VKD3D_SHADER_TYPE_HULL) { reg = &ins->declaration.dst.reg; + + if (reg->type == VKD3DSPR_OUTCONTROLPOINT) + normaliser->use_vocp = true; + /* We don't need to keep OUTCONTROLPOINT or PATCHCONST input declarations since their * equivalents were declared earlier, but INCONTROLPOINT may be the first occurrence. */ if (reg->type == VKD3DSPR_OUTCONTROLPOINT || reg->type == VKD3DSPR_PATCHCONST) @@ -1033,15 +1041,16 @@ static void shader_instruction_normalise_io_params(struct vkd3d_shader_instructi /* fall through */ case VKD3DSIH_DCL_INPUT_PS: case VKD3DSIH_DCL_OUTPUT: - keep = shader_dst_param_io_normalise(&ins->declaration.dst, true, normaliser); + if (!shader_dst_param_io_normalise(&ins->declaration.dst, true, normaliser)) + vkd3d_shader_instruction_make_nop(ins); break; case VKD3DSIH_DCL_INPUT_SGV: case VKD3DSIH_DCL_INPUT_SIV: case VKD3DSIH_DCL_INPUT_PS_SGV: case VKD3DSIH_DCL_INPUT_PS_SIV: case VKD3DSIH_DCL_OUTPUT_SIV: - keep = shader_dst_param_io_normalise(&ins->declaration.register_semantic.reg, true, - normaliser); + if (!shader_dst_param_io_normalise(&ins->declaration.register_semantic.reg, true, normaliser)) + vkd3d_shader_instruction_make_nop(ins); break; case VKD3DSIH_HS_CONTROL_POINT_PHASE: case VKD3DSIH_HS_FORK_PHASE: @@ -1060,29 +1069,24 @@ static void shader_instruction_normalise_io_params(struct vkd3d_shader_instructi shader_src_param_io_normalise((struct vkd3d_shader_src_param *)&ins->src[i], normaliser); break; } - - if (!keep) - shader_instruction_init(ins, VKD3DSIH_NOP); } -static enum vkd3d_result instruction_array_normalise_io_registers(struct vkd3d_shader_instruction_array *instructions, - enum vkd3d_shader_type shader_type, struct shader_signature *input_signature, - struct shader_signature *output_signature, struct shader_signature *patch_constant_signature) +static enum vkd3d_result shader_normalise_io_registers(struct vkd3d_shader_parser *parser) { - struct io_normaliser normaliser = {*instructions}; + struct io_normaliser normaliser = {parser->instructions}; struct vkd3d_shader_instruction *ins; bool has_control_point_phase; unsigned int i, j; normaliser.phase = VKD3DSIH_INVALID; - normaliser.shader_type = shader_type; - normaliser.input_signature = input_signature; - normaliser.output_signature = output_signature; - normaliser.patch_constant_signature = patch_constant_signature; + normaliser.shader_type = parser->shader_version.type; + normaliser.input_signature = &parser->shader_desc.input_signature; + normaliser.output_signature = &parser->shader_desc.output_signature; + normaliser.patch_constant_signature = &parser->shader_desc.patch_constant_signature; - for (i = 0, has_control_point_phase = false; i < instructions->count; ++i) + for (i = 0, has_control_point_phase = false; i < parser->instructions.count; ++i) { - ins = &instructions->elements[i]; + ins = &parser->instructions.elements[i]; switch (ins->handler_idx) { @@ -1121,11 +1125,11 @@ static enum vkd3d_result instruction_array_normalise_io_registers(struct vkd3d_s } } - if (!shader_signature_merge(input_signature, normaliser.input_range_map, false) - || !shader_signature_merge(output_signature, normaliser.output_range_map, false) - || !shader_signature_merge(patch_constant_signature, normaliser.pc_range_map, true)) + if (!shader_signature_merge(&parser->shader_desc.input_signature, normaliser.input_range_map, false) + || !shader_signature_merge(&parser->shader_desc.output_signature, normaliser.output_range_map, false) + || !shader_signature_merge(&parser->shader_desc.patch_constant_signature, normaliser.pc_range_map, true)) { - *instructions = normaliser.instructions; + parser->instructions = normaliser.instructions; return VKD3D_ERROR_OUT_OF_MEMORY; } @@ -1133,7 +1137,8 @@ static enum vkd3d_result instruction_array_normalise_io_registers(struct vkd3d_s for (i = 0; i < normaliser.instructions.count; ++i) shader_instruction_normalise_io_params(&normaliser.instructions.elements[i], &normaliser); - *instructions = normaliser.instructions; + parser->instructions = normaliser.instructions; + parser->shader_desc.use_vocp = normaliser.use_vocp; return VKD3D_OK; } @@ -1207,7 +1212,7 @@ static void shader_register_normalise_flat_constants(struct vkd3d_shader_src_par { param->reg.type = VKD3DSPR_IMMCONST; param->reg.idx_count = 0; - param->reg.immconst_type = VKD3D_IMMCONST_VEC4; + param->reg.dimension = VSIR_DIMENSION_VEC4; for (j = 0; j < 4; ++j) param->reg.u.immconst_uint[j] = normaliser->defs[i].value[j]; return; @@ -1260,6 +1265,164 @@ static enum vkd3d_result instruction_array_normalise_flat_constants(struct vkd3d return VKD3D_OK; } +static void remove_dead_code(struct vkd3d_shader_parser *parser) +{ + size_t i, depth = 0; + bool dead = false; + + for (i = 0; i < parser->instructions.count; ++i) + { + struct vkd3d_shader_instruction *ins = &parser->instructions.elements[i]; + + switch (ins->handler_idx) + { + case VKD3DSIH_IF: + case VKD3DSIH_LOOP: + case VKD3DSIH_SWITCH: + if (dead) + { + vkd3d_shader_instruction_make_nop(ins); + ++depth; + } + break; + + case VKD3DSIH_ENDIF: + case VKD3DSIH_ENDLOOP: + case VKD3DSIH_ENDSWITCH: + case VKD3DSIH_ELSE: + if (dead) + { + if (depth > 0) + { + if (ins->handler_idx != VKD3DSIH_ELSE) + --depth; + vkd3d_shader_instruction_make_nop(ins); + } + else + { + dead = false; + } + } + break; + + /* `depth' is counted with respect to where the dead code + * segment began. So it starts at zero and it signals the + * termination of the dead code segment when it would + * become negative. */ + case VKD3DSIH_BREAK: + case VKD3DSIH_RET: + case VKD3DSIH_CONTINUE: + if (dead) + { + vkd3d_shader_instruction_make_nop(ins); + } + else + { + dead = true; + depth = 0; + } + break; + + /* If `case' or `default' appears at zero depth, it means + * that they are a possible target for the corresponding + * switch, so the code is live again. */ + case VKD3DSIH_CASE: + case VKD3DSIH_DEFAULT: + if (dead) + { + if (depth == 0) + dead = false; + else + vkd3d_shader_instruction_make_nop(ins); + } + break; + + /* Phase instructions can only appear in hull shaders and + * outside of any block. When a phase returns, control is + * moved to the following phase, so they make code live + * again. */ + case VKD3DSIH_HS_CONTROL_POINT_PHASE: + case VKD3DSIH_HS_FORK_PHASE: + case VKD3DSIH_HS_JOIN_PHASE: + dead = false; + break; + + default: + if (dead) + vkd3d_shader_instruction_make_nop(ins); + break; + } + } +} + +static enum vkd3d_result normalise_combined_samplers(struct vkd3d_shader_parser *parser) +{ + unsigned int i; + + for (i = 0; i < parser->instructions.count; ++i) + { + struct vkd3d_shader_instruction *ins = &parser->instructions.elements[i]; + struct vkd3d_shader_src_param *srcs; + + switch (ins->handler_idx) + { + case VKD3DSIH_TEX: + if (!(srcs = shader_src_param_allocator_get(&parser->instructions.src_params, 3))) + return VKD3D_ERROR_OUT_OF_MEMORY; + memset(srcs, 0, sizeof(*srcs) * 3); + + ins->handler_idx = VKD3DSIH_SAMPLE; + + srcs[0] = ins->src[0]; + + srcs[1].reg.type = VKD3DSPR_RESOURCE; + srcs[1].reg.idx[0] = ins->src[1].reg.idx[0]; + srcs[1].reg.idx[1] = ins->src[1].reg.idx[0]; + srcs[1].reg.idx_count = 2; + srcs[1].reg.data_type = VKD3D_DATA_RESOURCE; + srcs[1].swizzle = VKD3D_SHADER_NO_SWIZZLE; + + srcs[2].reg.type = VKD3DSPR_SAMPLER; + srcs[2].reg.idx[0] = ins->src[1].reg.idx[0]; + srcs[2].reg.idx[1] = ins->src[1].reg.idx[0]; + srcs[2].reg.idx_count = 2; + srcs[2].reg.data_type = VKD3D_DATA_SAMPLER; + + ins->src = srcs; + ins->src_count = 3; + break; + + case VKD3DSIH_TEXBEM: + case VKD3DSIH_TEXBEML: + case VKD3DSIH_TEXCOORD: + case VKD3DSIH_TEXDEPTH: + case VKD3DSIH_TEXDP3: + case VKD3DSIH_TEXDP3TEX: + case VKD3DSIH_TEXLDD: + case VKD3DSIH_TEXLDL: + case VKD3DSIH_TEXM3x2PAD: + case VKD3DSIH_TEXM3x2TEX: + case VKD3DSIH_TEXM3x3DIFF: + case VKD3DSIH_TEXM3x3PAD: + case VKD3DSIH_TEXM3x3SPEC: + case VKD3DSIH_TEXM3x3TEX: + case VKD3DSIH_TEXM3x3VSPEC: + case VKD3DSIH_TEXREG2AR: + case VKD3DSIH_TEXREG2GB: + case VKD3DSIH_TEXREG2RGB: + vkd3d_shader_parser_error(parser, VKD3D_SHADER_ERROR_VSIR_NOT_IMPLEMENTED, + "Aborting due to not yet implemented feature: " + "Combined sampler instruction %#x.", ins->handler_idx); + return VKD3D_ERROR_NOT_IMPLEMENTED; + + default: + break; + } + } + + return VKD3D_OK; +} + enum vkd3d_result vkd3d_shader_normalise(struct vkd3d_shader_parser *parser, const struct vkd3d_shader_compile_info *compile_info) { @@ -1280,15 +1443,366 @@ enum vkd3d_result vkd3d_shader_normalise(struct vkd3d_shader_parser *parser, &parser->shader_desc.input_signature); } if (result >= 0) - result = instruction_array_normalise_io_registers(instructions, parser->shader_version.type, - &parser->shader_desc.input_signature, &parser->shader_desc.output_signature, - &parser->shader_desc.patch_constant_signature); + result = shader_normalise_io_registers(parser); if (result >= 0) result = instruction_array_normalise_flat_constants(parser); + if (result >= 0) + remove_dead_code(parser); + + if (result >= 0) + result = normalise_combined_samplers(parser); + if (result >= 0 && TRACE_ON()) vkd3d_shader_trace(instructions, &parser->shader_version); + if (result >= 0 && !parser->failed) + vsir_validate(parser); + + if (result >= 0 && parser->failed) + result = VKD3D_ERROR_INVALID_SHADER; + return result; } + +struct validation_context +{ + struct vkd3d_shader_parser *parser; + size_t instruction_idx; + bool dcl_temps_found; + unsigned int temp_count; + enum vkd3d_shader_opcode phase; + + enum vkd3d_shader_opcode *blocks; + size_t depth; + size_t blocks_capacity; +}; + +static void VKD3D_PRINTF_FUNC(3, 4) validator_error(struct validation_context *ctx, + enum vkd3d_shader_error error, const char *format, ...) +{ + struct vkd3d_string_buffer buf; + va_list args; + + vkd3d_string_buffer_init(&buf); + + va_start(args, format); + vkd3d_string_buffer_vprintf(&buf, format, args); + va_end(args); + + vkd3d_shader_parser_error(ctx->parser, error, "instruction %zu: %s", ctx->instruction_idx + 1, buf.buffer); + ERR("VSIR validation error: instruction %zu: %s\n", ctx->instruction_idx + 1, buf.buffer); + + vkd3d_string_buffer_cleanup(&buf); +} + +static void vsir_validate_src_param(struct validation_context *ctx, + const struct vkd3d_shader_src_param *src); + +static void vsir_validate_register(struct validation_context *ctx, + const struct vkd3d_shader_register *reg) +{ + unsigned int i, temp_count = ctx->temp_count; + + /* SM1-3 shaders do not include a DCL_TEMPS instruction. */ + if (ctx->parser->shader_version.major <= 3) + temp_count = ctx->parser->shader_desc.temp_count; + + if (reg->type >= VKD3DSPR_COUNT) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_REGISTER_TYPE, "Invalid register type %#x.", + reg->type); + + if (reg->precision >= VKD3D_SHADER_REGISTER_PRECISION_COUNT) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_PRECISION, "Invalid register precision %#x.", + reg->precision); + + if (reg->data_type >= VKD3D_DATA_COUNT) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_DATA_TYPE, "Invalid register data type %#x.", + reg->data_type); + + if (reg->dimension >= VSIR_DIMENSION_COUNT) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_DIMENSION, "Invalid register dimension %#x.", + reg->dimension); + + if (reg->idx_count > ARRAY_SIZE(reg->idx)) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INDEX_COUNT, "Invalid register index count %u.", + reg->idx_count); + + for (i = 0; i < min(reg->idx_count, ARRAY_SIZE(reg->idx)); ++i) + { + const struct vkd3d_shader_src_param *param = reg->idx[i].rel_addr; + if (reg->idx[i].rel_addr) + vsir_validate_src_param(ctx, param); + } + + switch (reg->type) + { + case VKD3DSPR_TEMP: + if (reg->idx_count != 1) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INDEX_COUNT, "Invalid index count %u for a TEMP register.", + reg->idx_count); + + if (reg->idx_count >= 1 && reg->idx[0].rel_addr) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INDEX, "Non-NULL relative address for a TEMP register."); + + if (reg->idx_count >= 1 && reg->idx[0].offset >= temp_count) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INDEX, "TEMP register index %u exceeds the maximum count %u.", + reg->idx[0].offset, temp_count); + break; + + case VKD3DSPR_NULL: + if (reg->idx_count != 0) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INDEX_COUNT, "Invalid index count %u for a NULL register.", + reg->idx_count); + break; + + case VKD3DSPR_IMMCONST: + if (reg->idx_count != 0) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INDEX_COUNT, "Invalid index count %u for a IMMCONST register.", + reg->idx_count); + break; + + case VKD3DSPR_IMMCONST64: + if (reg->idx_count != 0) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INDEX_COUNT, "Invalid index count %u for a IMMCONST64 register.", + reg->idx_count); + break; + + default: + break; + } +} + +static void vsir_validate_dst_param(struct validation_context *ctx, + const struct vkd3d_shader_dst_param *dst) +{ + vsir_validate_register(ctx, &dst->reg); + + if (dst->write_mask & ~VKD3DSP_WRITEMASK_ALL) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_WRITE_MASK, "Destination has invalid write mask %#x.", + dst->write_mask); + + if (dst->modifiers & ~VKD3DSPDM_MASK) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_MODIFIERS, "Destination has invalid modifiers %#x.", + dst->modifiers); + + switch (dst->shift) + { + case 0: + case 1: + case 2: + case 3: + case 13: + case 14: + case 15: + break; + + default: + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_SHIFT, "Destination has invalid shift %#x.", + dst->shift); + } +} + +static void vsir_validate_src_param(struct validation_context *ctx, + const struct vkd3d_shader_src_param *src) +{ + vsir_validate_register(ctx, &src->reg); + + if (src->swizzle & ~0x03030303u) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_SWIZZLE, "Source has invalid swizzle %#x.", + src->swizzle); + + if (src->modifiers >= VKD3DSPSM_COUNT) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_MODIFIERS, "Source has invalid modifiers %#x.", + src->modifiers); +} + +static void vsir_validate_dst_count(struct validation_context *ctx, + const struct vkd3d_shader_instruction *instruction, unsigned int count) +{ + if (instruction->dst_count != count) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_DEST_COUNT, + "Invalid destination count %u for an instruction of type %#x, expected %u.", + instruction->dst_count, instruction->handler_idx, count); +} + +static void vsir_validate_src_count(struct validation_context *ctx, + const struct vkd3d_shader_instruction *instruction, unsigned int count) +{ + if (instruction->src_count != count) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_SOURCE_COUNT, + "Invalid source count %u for an instruction of type %#x, expected %u.", + instruction->src_count, instruction->handler_idx, count); +} + +static void vsir_validate_instruction(struct validation_context *ctx) +{ + const struct vkd3d_shader_instruction *instruction = &ctx->parser->instructions.elements[ctx->instruction_idx]; + size_t i; + + ctx->parser->location = instruction->location; + + for (i = 0; i < instruction->dst_count; ++i) + vsir_validate_dst_param(ctx, &instruction->dst[i]); + + for (i = 0; i < instruction->src_count; ++i) + vsir_validate_src_param(ctx, &instruction->src[i]); + + if (instruction->handler_idx >= VKD3DSIH_INVALID) + { + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_HANDLER, "Invalid instruction handler %#x.", + instruction->handler_idx); + } + + switch (instruction->handler_idx) + { + case VKD3DSIH_HS_DECLS: + case VKD3DSIH_HS_CONTROL_POINT_PHASE: + case VKD3DSIH_HS_FORK_PHASE: + case VKD3DSIH_HS_JOIN_PHASE: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, 0); + if (ctx->parser->shader_version.type != VKD3D_SHADER_TYPE_HULL) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_HANDLER, "Phase instruction %#x is only valid in a hull shader.", + instruction->handler_idx); + if (ctx->depth != 0) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INSTRUCTION_NESTING, "Phase instruction %#x must appear to top level.", + instruction->handler_idx); + ctx->phase = instruction->handler_idx; + ctx->dcl_temps_found = false; + ctx->temp_count = 0; + return; + + default: + break; + } + + if (ctx->parser->shader_version.type == VKD3D_SHADER_TYPE_HULL && + ctx->phase == VKD3DSIH_INVALID) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_HANDLER, "Instruction %#x appear before any phase instruction in a hull shader.", + instruction->handler_idx); + + switch (instruction->handler_idx) + { + case VKD3DSIH_DCL_TEMPS: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, 0); + if (ctx->dcl_temps_found) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_DUPLICATE_DCL_TEMPS, "Duplicate DCL_TEMPS instruction."); + if (instruction->declaration.count > ctx->parser->shader_desc.temp_count) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_DCL_TEMPS, "Invalid DCL_TEMPS count %u, expected at most %u.", + instruction->declaration.count, ctx->parser->shader_desc.temp_count); + ctx->dcl_temps_found = true; + ctx->temp_count = instruction->declaration.count; + break; + + case VKD3DSIH_IF: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, 1); + if (!vkd3d_array_reserve((void **)&ctx->blocks, &ctx->blocks_capacity, ctx->depth + 1, sizeof(*ctx->blocks))) + return; + ctx->blocks[ctx->depth++] = instruction->handler_idx; + break; + + case VKD3DSIH_IFC: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, 2); + if (!vkd3d_array_reserve((void **)&ctx->blocks, &ctx->blocks_capacity, ctx->depth + 1, sizeof(*ctx->blocks))) + return; + ctx->blocks[ctx->depth++] = VKD3DSIH_IF; + break; + + case VKD3DSIH_ELSE: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, 0); + if (ctx->depth == 0 || ctx->blocks[ctx->depth - 1] != VKD3DSIH_IF) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INSTRUCTION_NESTING, "ELSE instruction doesn't terminate IF block."); + else + ctx->blocks[ctx->depth - 1] = instruction->handler_idx; + break; + + case VKD3DSIH_ENDIF: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, 0); + if (ctx->depth == 0 || (ctx->blocks[ctx->depth - 1] != VKD3DSIH_IF && ctx->blocks[ctx->depth - 1] != VKD3DSIH_ELSE)) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INSTRUCTION_NESTING, "ENDIF instruction doesn't terminate IF/ELSE block."); + else + --ctx->depth; + break; + + case VKD3DSIH_LOOP: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, ctx->parser->shader_version.major <= 3 ? 2 : 0); + if (!vkd3d_array_reserve((void **)&ctx->blocks, &ctx->blocks_capacity, ctx->depth + 1, sizeof(*ctx->blocks))) + return; + ctx->blocks[ctx->depth++] = instruction->handler_idx; + break; + + case VKD3DSIH_ENDLOOP: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, 0); + if (ctx->depth == 0 || ctx->blocks[ctx->depth - 1] != VKD3DSIH_LOOP) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INSTRUCTION_NESTING, "ENDLOOP instruction doesn't terminate LOOP block."); + else + --ctx->depth; + break; + + case VKD3DSIH_REP: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, 1); + if (!vkd3d_array_reserve((void **)&ctx->blocks, &ctx->blocks_capacity, ctx->depth + 1, sizeof(*ctx->blocks))) + return; + ctx->blocks[ctx->depth++] = instruction->handler_idx; + break; + + case VKD3DSIH_ENDREP: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, 0); + if (ctx->depth == 0 || ctx->blocks[ctx->depth - 1] != VKD3DSIH_REP) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INSTRUCTION_NESTING, "ENDREP instruction doesn't terminate REP block."); + else + --ctx->depth; + break; + + case VKD3DSIH_SWITCH: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, 1); + if (!vkd3d_array_reserve((void **)&ctx->blocks, &ctx->blocks_capacity, ctx->depth + 1, sizeof(*ctx->blocks))) + return; + ctx->blocks[ctx->depth++] = instruction->handler_idx; + break; + + case VKD3DSIH_ENDSWITCH: + vsir_validate_dst_count(ctx, instruction, 0); + vsir_validate_src_count(ctx, instruction, 0); + if (ctx->depth == 0 || ctx->blocks[ctx->depth - 1] != VKD3DSIH_SWITCH) + validator_error(ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INSTRUCTION_NESTING, "ENDSWITCH instruction doesn't terminate SWITCH block."); + else + --ctx->depth; + break; + + default: + break; + } +} + +void vsir_validate(struct vkd3d_shader_parser *parser) +{ + struct validation_context ctx = + { + .parser = parser, + .phase = VKD3DSIH_INVALID, + }; + + if (!(parser->config_flags & VKD3D_SHADER_CONFIG_FLAG_FORCE_VALIDATION)) + return; + + for (ctx.instruction_idx = 0; ctx.instruction_idx < parser->instructions.count; ++ctx.instruction_idx) + vsir_validate_instruction(&ctx); + + if (ctx.depth != 0) + validator_error(&ctx, VKD3D_SHADER_ERROR_VSIR_INVALID_INSTRUCTION_NESTING, "%zu nested blocks were not closed.", ctx.depth); + + free(ctx.blocks); +} diff --git a/libs/vkd3d/libs/vkd3d-shader/preproc.h b/libs/vkd3d/libs/vkd3d-shader/preproc.h index 4860cf5f90e..e1cb75e177c 100644 --- a/libs/vkd3d/libs/vkd3d-shader/preproc.h +++ b/libs/vkd3d/libs/vkd3d-shader/preproc.h @@ -22,7 +22,7 @@ #define __VKD3D_SHADER_PREPROC_H #include "vkd3d_shader_private.h" -#include "wine/rbtree.h" +#include "rbtree.h" struct preproc_if_state { diff --git a/libs/vkd3d/libs/vkd3d-shader/spirv.c b/libs/vkd3d/libs/vkd3d-shader/spirv.c index 638764b81bc..be149a0cf34 100644 --- a/libs/vkd3d/libs/vkd3d-shader/spirv.c +++ b/libs/vkd3d/libs/vkd3d-shader/spirv.c @@ -18,7 +18,7 @@ */ #include "vkd3d_shader_private.h" -#include "wine/rbtree.h" +#include "rbtree.h" #include #include @@ -173,7 +173,13 @@ enum vkd3d_shader_input_sysval_semantic vkd3d_siv_from_sysval_indexed(enum vkd3d { switch (sysval) { + case VKD3D_SHADER_SV_COVERAGE: + case VKD3D_SHADER_SV_DEPTH: + case VKD3D_SHADER_SV_DEPTH_GREATER_EQUAL: + case VKD3D_SHADER_SV_DEPTH_LESS_EQUAL: case VKD3D_SHADER_SV_NONE: + case VKD3D_SHADER_SV_STENCIL_REF: + case VKD3D_SHADER_SV_TARGET: return VKD3D_SIV_NONE; case VKD3D_SHADER_SV_POSITION: return VKD3D_SIV_POSITION; @@ -181,6 +187,16 @@ enum vkd3d_shader_input_sysval_semantic vkd3d_siv_from_sysval_indexed(enum vkd3d return VKD3D_SIV_CLIP_DISTANCE; case VKD3D_SHADER_SV_CULL_DISTANCE: return VKD3D_SIV_CULL_DISTANCE; + case VKD3D_SHADER_SV_INSTANCE_ID: + return VKD3D_SIV_INSTANCE_ID; + case VKD3D_SHADER_SV_IS_FRONT_FACE: + return VKD3D_SIV_IS_FRONT_FACE; + case VKD3D_SHADER_SV_PRIMITIVE_ID: + return VKD3D_SIV_PRIMITIVE_ID; + case VKD3D_SHADER_SV_RENDER_TARGET_ARRAY_INDEX: + return VKD3D_SIV_RENDER_TARGET_ARRAY_INDEX; + case VKD3D_SHADER_SV_SAMPLE_INDEX: + return VKD3D_SIV_SAMPLE_INDEX; case VKD3D_SHADER_SV_TESS_FACTOR_QUADEDGE: return VKD3D_SIV_QUAD_U0_TESS_FACTOR + index; case VKD3D_SHADER_SV_TESS_FACTOR_QUADINT: @@ -193,6 +209,10 @@ enum vkd3d_shader_input_sysval_semantic vkd3d_siv_from_sysval_indexed(enum vkd3d return VKD3D_SIV_LINE_DETAIL_TESS_FACTOR; case VKD3D_SHADER_SV_TESS_FACTOR_LINEDEN: return VKD3D_SIV_LINE_DENSITY_TESS_FACTOR; + case VKD3D_SHADER_SV_VERTEX_ID: + return VKD3D_SIV_VERTEX_ID; + case VKD3D_SHADER_SV_VIEWPORT_ARRAY_INDEX: + return VKD3D_SIV_VIEWPORT_ARRAY_INDEX; default: FIXME("Unhandled sysval %#x, index %u.\n", sysval, index); return VKD3D_SIV_NONE; @@ -1130,6 +1150,20 @@ static uint32_t vkd3d_spirv_get_op_type_pointer(struct vkd3d_spirv_builder *buil vkd3d_spirv_build_op_type_pointer); } +static uint32_t vkd3d_spirv_build_op_constant_bool(struct vkd3d_spirv_builder *builder, + uint32_t result_type, uint32_t value) +{ + return vkd3d_spirv_build_op_tr(builder, &builder->global_stream, + value ? SpvOpConstantTrue : SpvOpConstantFalse, result_type); +} + +static uint32_t vkd3d_spirv_get_op_constant_bool(struct vkd3d_spirv_builder *builder, + uint32_t result_type, uint32_t value) +{ + return vkd3d_spirv_build_once2(builder, value ? SpvOpConstantTrue : SpvOpConstantFalse, result_type, value, + vkd3d_spirv_build_op_constant_bool); +} + /* Types larger than 32-bits are not supported. */ static uint32_t vkd3d_spirv_build_op_constant(struct vkd3d_spirv_builder *builder, uint32_t result_type, uint32_t value) @@ -1215,10 +1249,14 @@ static uint32_t vkd3d_spirv_build_op_function_call(struct vkd3d_spirv_builder *b SpvOpFunctionCall, result_type, function_id, arguments, argument_count); } -static uint32_t vkd3d_spirv_build_op_undef(struct vkd3d_spirv_builder *builder, - struct vkd3d_spirv_stream *stream, uint32_t type_id) +static uint32_t vkd3d_spirv_build_op_undef(struct vkd3d_spirv_builder *builder, uint32_t type_id) +{ + return vkd3d_spirv_build_op_tr(builder, &builder->global_stream, SpvOpUndef, type_id); +} + +static uint32_t vkd3d_spirv_get_op_undef(struct vkd3d_spirv_builder *builder, uint32_t type_id) { - return vkd3d_spirv_build_op_tr(builder, stream, SpvOpUndef, type_id); + return vkd3d_spirv_build_once1(builder, SpvOpUndef, type_id, vkd3d_spirv_build_op_undef); } static uint32_t vkd3d_spirv_build_op_access_chain(struct vkd3d_spirv_builder *builder, @@ -1407,13 +1445,6 @@ static uint32_t vkd3d_spirv_build_op_udiv(struct vkd3d_spirv_builder *builder, SpvOpUDiv, result_type, operand0, operand1); } -static uint32_t vkd3d_spirv_build_op_umod(struct vkd3d_spirv_builder *builder, - uint32_t result_type, uint32_t operand0, uint32_t operand1) -{ - return vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, - SpvOpUMod, result_type, operand0, operand1); -} - static uint32_t vkd3d_spirv_build_op_isub(struct vkd3d_spirv_builder *builder, uint32_t result_type, uint32_t operand0, uint32_t operand1) { @@ -1710,6 +1741,15 @@ static uint32_t vkd3d_spirv_build_op_glsl_std450_cos(struct vkd3d_spirv_builder return vkd3d_spirv_build_op_glsl_std450_tr1(builder, GLSLstd450Cos, result_type, operand); } +static uint32_t vkd3d_spirv_build_op_glsl_std450_max(struct vkd3d_spirv_builder *builder, + uint32_t result_type, uint32_t x, uint32_t y) +{ + uint32_t glsl_std450_id = vkd3d_spirv_get_glsl_std450_instr_set(builder); + uint32_t operands[] = {x, y}; + return vkd3d_spirv_build_op_ext_inst(builder, result_type, glsl_std450_id, + GLSLstd450NMax, operands, ARRAY_SIZE(operands)); +} + static uint32_t vkd3d_spirv_build_op_glsl_std450_nclamp(struct vkd3d_spirv_builder *builder, uint32_t result_type, uint32_t x, uint32_t min, uint32_t max) { @@ -1776,6 +1816,8 @@ static uint32_t vkd3d_spirv_get_type_id_for_data_type(struct vkd3d_spirv_builder break; case VKD3D_DATA_DOUBLE: return vkd3d_spirv_get_op_type_float(builder, 64); + case VKD3D_DATA_BOOL: + return vkd3d_spirv_get_op_type_bool(builder); default: FIXME("Unhandled data type %#x.\n", data_type); return 0; @@ -2099,14 +2141,22 @@ static void vkd3d_symbol_make_register(struct vkd3d_symbol *symbol, symbol->type = VKD3D_SYMBOL_REGISTER; memset(&symbol->key, 0, sizeof(symbol->key)); symbol->key.reg.type = reg->type; - if (vkd3d_shader_register_is_input(reg) || vkd3d_shader_register_is_output(reg) - || vkd3d_shader_register_is_patch_constant(reg)) + + switch (reg->type) { - symbol->key.reg.idx = reg->idx_count ? reg->idx[reg->idx_count - 1].offset : ~0u; - assert(!reg->idx_count || symbol->key.reg.idx != ~0u); + case VKD3DSPR_INPUT: + case VKD3DSPR_OUTPUT: + case VKD3DSPR_PATCHCONST: + symbol->key.reg.idx = reg->idx_count ? reg->idx[reg->idx_count - 1].offset : ~0u; + assert(!reg->idx_count || symbol->key.reg.idx != ~0u); + break; + + case VKD3DSPR_IMMCONSTBUFFER: + break; + + default: + symbol->key.reg.idx = reg->idx_count ? reg->idx[0].offset : ~0u; } - else if (reg->type != VKD3DSPR_IMMCONSTBUFFER) - symbol->key.reg.idx = reg->idx_count ? reg->idx[0].offset : ~0u; } static void vkd3d_symbol_set_register_info(struct vkd3d_symbol *symbol, @@ -2251,6 +2301,12 @@ struct vkd3d_hull_shader_variables uint32_t patch_constants_id; }; +struct ssa_register_info +{ + enum vkd3d_data_type data_type; + uint32_t id; +}; + struct spirv_compiler { struct vkd3d_spirv_builder spirv_builder; @@ -2262,6 +2318,7 @@ struct spirv_compiler bool strip_debug; bool ssbo_uavs; bool uav_read_without_format; + SpvExecutionMode fragment_coordinate_origin; struct rb_tree symbol_table; uint32_t temp_id; @@ -2299,7 +2356,6 @@ struct spirv_compiler uint32_t array_element_mask; } *output_info; uint32_t private_output_variable[MAX_REG_OUTPUT + 1]; /* 1 entry for oDepth */ - uint32_t private_output_variable_array_idx[MAX_REG_OUTPUT + 1]; /* 1 entry for oDepth */ uint32_t private_output_variable_write_mask[MAX_REG_OUTPUT + 1]; /* 1 entry for oDepth */ uint32_t epilogue_function_id; @@ -2323,6 +2379,9 @@ struct spirv_compiler bool write_tess_geom_point_size; struct vkd3d_string_buffer_cache string_buffers; + + struct ssa_register_info *ssa_register_info; + unsigned int ssa_register_count; }; static bool is_in_default_phase(const struct spirv_compiler *compiler) @@ -2370,6 +2429,8 @@ static void spirv_compiler_destroy(struct spirv_compiler *compiler) shader_signature_cleanup(&compiler->output_signature); shader_signature_cleanup(&compiler->patch_constant_signature); + vkd3d_free(compiler->ssa_register_info); + vkd3d_free(compiler); } @@ -2422,6 +2483,7 @@ static struct spirv_compiler *spirv_compiler_create(const struct vkd3d_shader_ve compiler->formatting = VKD3D_SHADER_COMPILE_OPTION_FORMATTING_INDENT | VKD3D_SHADER_COMPILE_OPTION_FORMATTING_HEADER; compiler->write_tess_geom_point_size = true; + compiler->fragment_coordinate_origin = SpvExecutionModeOriginUpperLeft; for (i = 0; i < compile_info->option_count; ++i) { @@ -2446,9 +2508,6 @@ static struct spirv_compiler *spirv_compiler_create(const struct vkd3d_shader_ve compiler->formatting = option->value; break; - default: - WARN("Ignoring unrecognised option %#x with value %#x.\n", option->name, option->value); - case VKD3D_SHADER_COMPILE_OPTION_API_VERSION: break; @@ -2464,6 +2523,19 @@ static struct spirv_compiler *spirv_compiler_create(const struct vkd3d_shader_ve case VKD3D_SHADER_COMPILE_OPTION_WRITE_TESS_GEOM_POINT_SIZE: compiler->write_tess_geom_point_size = option->value; break; + + case VKD3D_SHADER_COMPILE_OPTION_FRAGMENT_COORDINATE_ORIGIN: + if (option->value == VKD3D_SHADER_COMPILE_OPTION_FRAGMENT_COORDINATE_ORIGIN_UPPER_LEFT) + compiler->fragment_coordinate_origin = SpvExecutionModeOriginUpperLeft; + else if (option->value == VKD3D_SHADER_COMPILE_OPTION_FRAGMENT_COORDINATE_ORIGIN_LOWER_LEFT) + compiler->fragment_coordinate_origin = SpvExecutionModeOriginLowerLeft; + else + WARN("Ignoring unrecognised value %#x for option %#x.\n", option->value, option->name); + break; + + default: + WARN("Ignoring unrecognised option %#x with value %#x.\n", option->name, option->value); + break; } } @@ -2848,9 +2920,16 @@ static uint32_t spirv_compiler_get_constant(struct spirv_compiler *compiler, case VKD3D_SHADER_COMPONENT_INT: case VKD3D_SHADER_COMPONENT_FLOAT: break; + case VKD3D_SHADER_COMPONENT_BOOL: + if (component_count == 1) + return vkd3d_spirv_get_op_constant_bool(builder, type_id, *values); + FIXME("Unsupported vector of bool.\n"); + spirv_compiler_error(compiler, VKD3D_SHADER_ERROR_SPV_INVALID_TYPE, + "Vectors of bool type are not supported."); + return vkd3d_spirv_get_op_undef(builder, type_id); default: FIXME("Unhandled component_type %#x.\n", component_type); - return vkd3d_spirv_build_op_undef(builder, &builder->global_stream, type_id); + return vkd3d_spirv_get_op_undef(builder, type_id); } if (component_count == 1) @@ -2879,7 +2958,7 @@ static uint32_t spirv_compiler_get_constant64(struct spirv_compiler *compiler, if (component_type != VKD3D_SHADER_COMPONENT_DOUBLE) { FIXME("Unhandled component_type %#x.\n", component_type); - return vkd3d_spirv_build_op_undef(builder, &builder->global_stream, type_id); + return vkd3d_spirv_get_op_undef(builder, type_id); } if (component_count == 1) @@ -2975,13 +3054,12 @@ static bool spirv_compiler_get_register_name(char *buffer, unsigned int buffer_s case VKD3DSPR_INPUT: snprintf(buffer, buffer_size, "v%u", idx); break; - case VKD3DSPR_INCONTROLPOINT: - snprintf(buffer, buffer_size, "vicp%u", idx); - break; case VKD3DSPR_OUTPUT: - case VKD3DSPR_COLOROUT: snprintf(buffer, buffer_size, "o%u", idx); break; + case VKD3DSPR_COLOROUT: + snprintf(buffer, buffer_size, "oC%u", idx); + break; case VKD3DSPR_DEPTHOUT: case VKD3DSPR_DEPTHOUTGE: case VKD3DSPR_DEPTHOUTLE: @@ -3318,21 +3396,6 @@ static bool spirv_compiler_get_register_info(const struct spirv_compiler *compil return true; } -static bool register_is_descriptor(const struct vkd3d_shader_register *reg) -{ - switch (reg->type) - { - case VKD3DSPR_SAMPLER: - case VKD3DSPR_RESOURCE: - case VKD3DSPR_CONSTBUFFER: - case VKD3DSPR_UAV: - return true; - - default: - return false; - } -} - static bool spirv_compiler_enable_descriptor_indexing(struct spirv_compiler *compiler, enum vkd3d_shader_register_type reg_type, enum vkd3d_shader_resource_type resource_type) { @@ -3458,10 +3521,13 @@ static void spirv_compiler_emit_dereference_register(struct spirv_compiler *comp FIXME("Relative addressing not implemented.\n"); /* Handle arrayed registers, e.g. v[3][0]. */ - if (reg->idx_count > 1 && !register_is_descriptor(reg)) + if (reg->idx_count > 1 && !vsir_register_is_descriptor(reg)) indexes[index_count++] = spirv_compiler_emit_register_addressing(compiler, ®->idx[0]); } + if (reg->alignment) + WARN("Ignoring alignment %u.\n", reg->alignment); + if (index_count) { component_count = vkd3d_write_mask_component_count(register_info->write_mask); @@ -3496,6 +3562,14 @@ static bool vkd3d_swizzle_is_equal(unsigned int dst_write_mask, return vkd3d_compact_swizzle(VKD3D_SHADER_NO_SWIZZLE, dst_write_mask) == vkd3d_compact_swizzle(swizzle, write_mask); } +static bool vkd3d_swizzle_is_scalar(unsigned int swizzle) +{ + unsigned int component_idx = vkd3d_swizzle_get_component(swizzle, 0); + return vkd3d_swizzle_get_component(swizzle, 1) == component_idx + && vkd3d_swizzle_get_component(swizzle, 2) == component_idx + && vkd3d_swizzle_get_component(swizzle, 3) == component_idx; +} + static uint32_t spirv_compiler_emit_swizzle(struct spirv_compiler *compiler, uint32_t val_id, unsigned int val_write_mask, enum vkd3d_shader_component_type component_type, unsigned int swizzle, unsigned int write_mask) @@ -3576,7 +3650,7 @@ static uint32_t spirv_compiler_emit_load_constant(struct spirv_compiler *compile assert(reg->type == VKD3DSPR_IMMCONST); - if (reg->immconst_type == VKD3D_IMMCONST_SCALAR) + if (reg->dimension == VSIR_DIMENSION_SCALAR) { for (i = 0; i < component_count; ++i) values[i] = *reg->u.immconst_uint; @@ -3603,7 +3677,7 @@ static uint32_t spirv_compiler_emit_load_constant64(struct spirv_compiler *compi assert(reg->type == VKD3DSPR_IMMCONST64); - if (reg->immconst_type == VKD3D_IMMCONST_SCALAR) + if (reg->dimension == VSIR_DIMENSION_SCALAR) { for (i = 0; i < component_count; ++i) values[i] = *reg->u.immconst_uint64; @@ -3631,7 +3705,7 @@ static uint32_t spirv_compiler_emit_load_undef(struct spirv_compiler *compiler, assert(reg->type == VKD3DSPR_UNDEF); type_id = vkd3d_spirv_get_type_id_for_data_type(builder, reg->data_type, component_count); - return vkd3d_spirv_build_op_undef(builder, &builder->global_stream, type_id); + return vkd3d_spirv_get_op_undef(builder, type_id); } static uint32_t spirv_compiler_emit_load_scalar(struct spirv_compiler *compiler, @@ -3682,6 +3756,55 @@ static uint32_t spirv_compiler_emit_load_scalar(struct spirv_compiler *compiler, return val_id; } +static const struct ssa_register_info *spirv_compiler_get_ssa_register_info(const struct spirv_compiler *compiler, + const struct vkd3d_shader_register *reg) +{ + assert(reg->idx[0].offset < compiler->ssa_register_count); + assert(reg->idx_count == 1); + return &compiler->ssa_register_info[reg->idx[0].offset]; +} + +static void spirv_compiler_set_ssa_register_info(const struct spirv_compiler *compiler, + const struct vkd3d_shader_register *reg, uint32_t val_id) +{ + unsigned int i = reg->idx[0].offset; + assert(i < compiler->ssa_register_count); + compiler->ssa_register_info[i].data_type = reg->data_type; + compiler->ssa_register_info[i].id = val_id; +} + +static uint32_t spirv_compiler_emit_load_ssa_reg(struct spirv_compiler *compiler, + const struct vkd3d_shader_register *reg, enum vkd3d_shader_component_type component_type, + unsigned int swizzle) +{ + struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; + enum vkd3d_shader_component_type reg_component_type; + const struct ssa_register_info *ssa; + unsigned int component_idx; + uint32_t type_id, val_id; + + ssa = spirv_compiler_get_ssa_register_info(compiler, reg); + val_id = ssa->id; + assert(val_id); + assert(vkd3d_swizzle_is_scalar(swizzle)); + + if (reg->dimension == VSIR_DIMENSION_SCALAR) + { + reg_component_type = vkd3d_component_type_from_data_type(ssa->data_type); + if (component_type != reg_component_type) + { + type_id = vkd3d_spirv_get_type_id(builder, component_type, 1); + val_id = vkd3d_spirv_build_op_bitcast(builder, type_id, val_id); + } + + return val_id; + } + + type_id = vkd3d_spirv_get_type_id(builder, component_type, 1); + component_idx = vkd3d_swizzle_get_component(swizzle, 0); + return vkd3d_spirv_build_op_composite_extract1(builder, type_id, val_id, component_idx); +} + static uint32_t spirv_compiler_emit_load_reg(struct spirv_compiler *compiler, const struct vkd3d_shader_register *reg, DWORD swizzle, DWORD write_mask) { @@ -3701,10 +3824,14 @@ static uint32_t spirv_compiler_emit_load_reg(struct spirv_compiler *compiler, component_count = vkd3d_write_mask_component_count(write_mask); component_type = vkd3d_component_type_from_data_type(reg->data_type); + + if (reg->type == VKD3DSPR_SSA) + return spirv_compiler_emit_load_ssa_reg(compiler, reg, component_type, swizzle); + if (!spirv_compiler_get_register_info(compiler, reg, ®_info)) { type_id = vkd3d_spirv_get_type_id(builder, component_type, component_count); - return vkd3d_spirv_build_op_undef(builder, &builder->global_stream, type_id); + return vkd3d_spirv_get_op_undef(builder, type_id); } assert(reg_info.component_type != VKD3D_SHADER_COMPONENT_DOUBLE); spirv_compiler_emit_dereference_register(compiler, reg, ®_info); @@ -3777,7 +3904,7 @@ static uint32_t spirv_compiler_emit_neg(struct spirv_compiler *compiler, type_id = spirv_compiler_get_type_id_for_reg(compiler, reg, write_mask); if (reg->data_type == VKD3D_DATA_FLOAT || reg->data_type == VKD3D_DATA_DOUBLE) return vkd3d_spirv_build_op_fnegate(builder, type_id, val_id); - else if (reg->data_type == VKD3D_DATA_INT) + else if (reg->data_type == VKD3D_DATA_INT || reg->data_type == VKD3D_DATA_UINT) return vkd3d_spirv_build_op_snegate(builder, type_id, val_id); FIXME("Unhandled data type %#x.\n", reg->data_type); @@ -3912,6 +4039,12 @@ static void spirv_compiler_emit_store_reg(struct spirv_compiler *compiler, assert(!register_is_constant_or_undef(reg)); + if (reg->type == VKD3DSPR_SSA) + { + spirv_compiler_set_ssa_register_info(compiler, reg, val_id); + return; + } + if (!spirv_compiler_get_register_info(compiler, reg, ®_info)) return; spirv_compiler_emit_dereference_register(compiler, reg, ®_info); @@ -4108,17 +4241,41 @@ static uint32_t spirv_compiler_emit_int_to_bool(struct spirv_compiler *compiler, } static uint32_t spirv_compiler_emit_bool_to_int(struct spirv_compiler *compiler, - unsigned int component_count, uint32_t val_id) + unsigned int component_count, uint32_t val_id, bool signedness) { struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; uint32_t type_id, true_id, false_id; - true_id = spirv_compiler_get_constant_uint_vector(compiler, 0xffffffff, component_count); + true_id = spirv_compiler_get_constant_uint_vector(compiler, signedness ? 0xffffffff : 1, component_count); false_id = spirv_compiler_get_constant_uint_vector(compiler, 0, component_count); type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_UINT, component_count); return vkd3d_spirv_build_op_select(builder, type_id, val_id, true_id, false_id); } +static uint32_t spirv_compiler_emit_bool_to_float(struct spirv_compiler *compiler, + unsigned int component_count, uint32_t val_id, bool signedness) +{ + struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; + uint32_t type_id, true_id, false_id; + + true_id = spirv_compiler_get_constant_float_vector(compiler, signedness ? -1.0f : 1.0f, component_count); + false_id = spirv_compiler_get_constant_float_vector(compiler, 0.0f, component_count); + type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_FLOAT, component_count); + return vkd3d_spirv_build_op_select(builder, type_id, val_id, true_id, false_id); +} + +static uint32_t spirv_compiler_emit_bool_to_double(struct spirv_compiler *compiler, + unsigned int component_count, uint32_t val_id, bool signedness) +{ + struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; + uint32_t type_id, true_id, false_id; + + true_id = spirv_compiler_get_constant_double_vector(compiler, signedness ? -1.0 : 1.0, component_count); + false_id = spirv_compiler_get_constant_double_vector(compiler, 0.0, component_count); + type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_DOUBLE, component_count); + return vkd3d_spirv_build_op_select(builder, type_id, val_id, true_id, false_id); +} + typedef uint32_t (*vkd3d_spirv_builtin_fixup_pfn)(struct spirv_compiler *compiler, uint32_t val_id); @@ -4161,7 +4318,7 @@ static uint32_t sv_instance_id_fixup(struct spirv_compiler *compiler, static uint32_t sv_front_face_fixup(struct spirv_compiler *compiler, uint32_t front_facing_id) { - return spirv_compiler_emit_bool_to_int(compiler, 1, front_facing_id); + return spirv_compiler_emit_bool_to_int(compiler, 1, front_facing_id, true); } /* frag_coord.w = 1.0f / frag_coord.w */ @@ -4195,47 +4352,41 @@ struct vkd3d_spirv_builtin */ static const struct { - enum vkd3d_shader_input_sysval_semantic sysval; + enum vkd3d_shader_sysval_semantic sysval; struct vkd3d_spirv_builtin builtin; enum vkd3d_shader_spirv_environment environment; } vkd3d_system_value_builtins[] = { - {VKD3D_SIV_VERTEX_ID, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInVertexId}, + {VKD3D_SHADER_SV_VERTEX_ID, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInVertexId}, VKD3D_SHADER_SPIRV_ENVIRONMENT_OPENGL_4_5}, - {VKD3D_SIV_INSTANCE_ID, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInInstanceId}, + {VKD3D_SHADER_SV_INSTANCE_ID, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInInstanceId}, VKD3D_SHADER_SPIRV_ENVIRONMENT_OPENGL_4_5}, - {VKD3D_SIV_POSITION, {VKD3D_SHADER_COMPONENT_FLOAT, 4, SpvBuiltInPosition}}, - {VKD3D_SIV_VERTEX_ID, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInVertexIndex, sv_vertex_id_fixup}}, - {VKD3D_SIV_INSTANCE_ID, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInInstanceIndex, sv_instance_id_fixup}}, + {VKD3D_SHADER_SV_POSITION, {VKD3D_SHADER_COMPONENT_FLOAT, 4, SpvBuiltInPosition}}, + {VKD3D_SHADER_SV_VERTEX_ID, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInVertexIndex, sv_vertex_id_fixup}}, + {VKD3D_SHADER_SV_INSTANCE_ID, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInInstanceIndex, sv_instance_id_fixup}}, - {VKD3D_SIV_PRIMITIVE_ID, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInPrimitiveId}}, + {VKD3D_SHADER_SV_PRIMITIVE_ID, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInPrimitiveId}}, - {VKD3D_SIV_RENDER_TARGET_ARRAY_INDEX, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInLayer}}, - {VKD3D_SIV_VIEWPORT_ARRAY_INDEX, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInViewportIndex}}, + {VKD3D_SHADER_SV_RENDER_TARGET_ARRAY_INDEX, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInLayer}}, + {VKD3D_SHADER_SV_VIEWPORT_ARRAY_INDEX, {VKD3D_SHADER_COMPONENT_INT, 1, SpvBuiltInViewportIndex}}, - {VKD3D_SIV_IS_FRONT_FACE, {VKD3D_SHADER_COMPONENT_BOOL, 1, SpvBuiltInFrontFacing, sv_front_face_fixup}}, + {VKD3D_SHADER_SV_IS_FRONT_FACE, {VKD3D_SHADER_COMPONENT_BOOL, 1, SpvBuiltInFrontFacing, sv_front_face_fixup}}, - {VKD3D_SIV_SAMPLE_INDEX, {VKD3D_SHADER_COMPONENT_UINT, 1, SpvBuiltInSampleId}}, + {VKD3D_SHADER_SV_SAMPLE_INDEX, {VKD3D_SHADER_COMPONENT_UINT, 1, SpvBuiltInSampleId}}, - {VKD3D_SIV_CLIP_DISTANCE, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInClipDistance, NULL, 1}}, - {VKD3D_SIV_CULL_DISTANCE, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInCullDistance, NULL, 1}}, + {VKD3D_SHADER_SV_CLIP_DISTANCE, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInClipDistance, NULL, 1}}, + {VKD3D_SHADER_SV_CULL_DISTANCE, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInCullDistance, NULL, 1}}, - {VKD3D_SIV_QUAD_U0_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4, 0}}, - {VKD3D_SIV_QUAD_V0_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4, 1}}, - {VKD3D_SIV_QUAD_U1_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4, 2}}, - {VKD3D_SIV_QUAD_V1_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4, 3}}, - {VKD3D_SIV_QUAD_U_INNER_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelInner, NULL, 2, 0}}, - {VKD3D_SIV_QUAD_V_INNER_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelInner, NULL, 2, 1}}, + {VKD3D_SHADER_SV_TESS_FACTOR_QUADEDGE, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4}}, + {VKD3D_SHADER_SV_TESS_FACTOR_QUADINT, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelInner, NULL, 2}}, - {VKD3D_SIV_TRIANGLE_U_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4, 0}}, - {VKD3D_SIV_TRIANGLE_V_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4, 1}}, - {VKD3D_SIV_TRIANGLE_W_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4, 2}}, - {VKD3D_SIV_TRIANGLE_INNER_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelInner, NULL, 2, 0}}, + {VKD3D_SHADER_SV_TESS_FACTOR_TRIEDGE, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4}}, + {VKD3D_SHADER_SV_TESS_FACTOR_TRIINT, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelInner, NULL, 2}}, - {VKD3D_SIV_LINE_DENSITY_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4, 0}}, - {VKD3D_SIV_LINE_DETAIL_TESS_FACTOR, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4, 1}}, + {VKD3D_SHADER_SV_TESS_FACTOR_LINEDEN, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4, 0}}, + {VKD3D_SHADER_SV_TESS_FACTOR_LINEDET, {VKD3D_SHADER_COMPONENT_FLOAT, 1, SpvBuiltInTessLevelOuter, NULL, 4, 1}}, }; static const struct vkd3d_spirv_builtin vkd3d_pixel_shader_position_builtin = { @@ -4299,7 +4450,7 @@ static void spirv_compiler_emit_register_execution_mode(struct spirv_compiler *c } static const struct vkd3d_spirv_builtin *get_spirv_builtin_for_sysval( - const struct spirv_compiler *compiler, enum vkd3d_shader_input_sysval_semantic sysval) + const struct spirv_compiler *compiler, enum vkd3d_shader_sysval_semantic sysval) { enum vkd3d_shader_spirv_environment environment; unsigned int i; @@ -4308,7 +4459,7 @@ static const struct vkd3d_spirv_builtin *get_spirv_builtin_for_sysval( return NULL; /* In pixel shaders, SV_Position is mapped to SpvBuiltInFragCoord. */ - if (sysval == VKD3D_SIV_POSITION && compiler->shader_type == VKD3D_SHADER_TYPE_PIXEL) + if (sysval == VKD3D_SHADER_SV_POSITION && compiler->shader_type == VKD3D_SHADER_TYPE_PIXEL) return &vkd3d_pixel_shader_position_builtin; environment = spirv_compiler_get_target_environment(compiler); @@ -4340,7 +4491,7 @@ static const struct vkd3d_spirv_builtin *get_spirv_builtin_for_register( } static const struct vkd3d_spirv_builtin *vkd3d_get_spirv_builtin(const struct spirv_compiler *compiler, - enum vkd3d_shader_register_type reg_type, enum vkd3d_shader_input_sysval_semantic sysval) + enum vkd3d_shader_register_type reg_type, enum vkd3d_shader_sysval_semantic sysval) { const struct vkd3d_spirv_builtin *builtin; @@ -4349,8 +4500,7 @@ static const struct vkd3d_spirv_builtin *vkd3d_get_spirv_builtin(const struct sp if ((builtin = get_spirv_builtin_for_register(reg_type))) return builtin; - if (sysval != VKD3D_SIV_NONE || (reg_type != VKD3DSPR_OUTPUT && reg_type != VKD3DSPR_COLOROUT - && reg_type != VKD3DSPR_PATCHCONST)) + if (sysval != VKD3D_SHADER_SV_NONE || (reg_type != VKD3DSPR_OUTPUT && reg_type != VKD3DSPR_PATCHCONST)) FIXME("Unhandled builtin (register type %#x, sysval %#x).\n", reg_type, sysval); return NULL; } @@ -4361,11 +4511,7 @@ static uint32_t spirv_compiler_get_invocation_id(struct spirv_compiler *compiler assert(compiler->shader_type == VKD3D_SHADER_TYPE_HULL); - memset(&r, 0, sizeof(r)); - r.type = VKD3DSPR_OUTPOINTID; - r.idx[0].offset = ~0u; - r.idx[1].offset = ~0u; - r.idx_count = 0; + vsir_register_init(&r, VKD3DSPR_OUTPOINTID, VKD3D_DATA_FLOAT, 0); return spirv_compiler_get_register_id(compiler, &r); } @@ -4486,7 +4632,7 @@ static uint32_t spirv_compiler_emit_builtin_variable_v(struct spirv_compiler *co assert(size_count <= ARRAY_SIZE(sizes)); memcpy(sizes, array_sizes, size_count * sizeof(sizes[0])); array_sizes = sizes; - sizes[0] = max(sizes[0], builtin->spirv_array_size); + sizes[size_count - 1] = max(sizes[size_count - 1], builtin->spirv_array_size); id = spirv_compiler_emit_array_variable(compiler, &builder->global_stream, storage_class, builtin->component_type, builtin->component_count, array_sizes, size_count); @@ -4524,27 +4670,8 @@ static unsigned int shader_signature_next_location(const struct shader_signature return max_row; } -static unsigned int shader_register_get_io_indices(const struct vkd3d_shader_register *reg, - unsigned int *array_sizes) -{ - unsigned int i, element_idx; - - array_sizes[0] = 0; - array_sizes[1] = 0; - element_idx = reg->idx[0].offset; - for (i = 1; i < reg->idx_count; ++i) - { - array_sizes[1] = array_sizes[0]; - array_sizes[0] = element_idx; - element_idx = reg->idx[i].offset; - } - - return element_idx; -} - static uint32_t spirv_compiler_emit_input(struct spirv_compiler *compiler, - const struct vkd3d_shader_dst_param *dst, enum vkd3d_shader_input_sysval_semantic sysval, - enum vkd3d_shader_interpolation_mode interpolation_mode) + const struct vkd3d_shader_dst_param *dst) { struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; const struct vkd3d_shader_register *reg = &dst->reg; @@ -4552,18 +4679,18 @@ static uint32_t spirv_compiler_emit_input(struct spirv_compiler *compiler, const struct signature_element *signature_element; const struct shader_signature *shader_signature; enum vkd3d_shader_component_type component_type; - uint32_t type_id, ptr_type_id, float_type_id; const struct vkd3d_spirv_builtin *builtin; + enum vkd3d_shader_sysval_semantic sysval; unsigned int write_mask, reg_write_mask; struct vkd3d_symbol *symbol = NULL; uint32_t val_id, input_id, var_id; + uint32_t type_id, float_type_id; struct vkd3d_symbol reg_symbol; SpvStorageClass storage_class; struct rb_entry *entry = NULL; bool use_private_var = false; unsigned int array_sizes[2]; unsigned int element_idx; - uint32_t i, index; assert(!reg->idx_count || !reg->idx[0].rel_addr); assert(reg->idx_count < 2 || !reg->idx[1].rel_addr); @@ -4571,15 +4698,22 @@ static uint32_t spirv_compiler_emit_input(struct spirv_compiler *compiler, shader_signature = reg->type == VKD3DSPR_PATCHCONST ? &compiler->patch_constant_signature : &compiler->input_signature; - element_idx = shader_register_get_io_indices(reg, array_sizes); + element_idx = reg->idx[reg->idx_count - 1].offset; signature_element = &shader_signature->elements[element_idx]; - - if ((compiler->shader_type == VKD3D_SHADER_TYPE_HULL || compiler->shader_type == VKD3D_SHADER_TYPE_GEOMETRY) - && !sysval && signature_element->sysval_semantic) - sysval = vkd3d_siv_from_sysval(signature_element->sysval_semantic); + sysval = signature_element->sysval_semantic; + /* The Vulkan spec does not explicitly forbid passing varyings from the + * TCS to the TES via builtins. However, Mesa doesn't seem to handle it + * well, and we don't actually need them to be in builtins. */ + if (compiler->shader_type == VKD3D_SHADER_TYPE_DOMAIN && reg->type != VKD3DSPR_PATCHCONST) + sysval = VKD3D_SHADER_SV_NONE; builtin = get_spirv_builtin_for_sysval(compiler, sysval); + array_sizes[0] = (reg->type == VKD3DSPR_PATCHCONST ? 0 : compiler->input_control_point_count); + array_sizes[1] = signature_element->register_count; + if (array_sizes[1] == 1 && !vsir_sysval_semantic_is_tess_factor(signature_element->sysval_semantic)) + array_sizes[1] = 0; + write_mask = signature_element->mask; if (builtin) @@ -4644,7 +4778,7 @@ static uint32_t spirv_compiler_emit_input(struct spirv_compiler *compiler, if (component_idx) vkd3d_spirv_build_op_decorate1(builder, input_id, SpvDecorationComponent, component_idx); - spirv_compiler_emit_interpolation_decorations(compiler, input_id, interpolation_mode); + spirv_compiler_emit_interpolation_decorations(compiler, input_id, signature_element->interpolation_mode); } var_id = input_id; @@ -4666,47 +4800,27 @@ static uint32_t spirv_compiler_emit_input(struct spirv_compiler *compiler, if (use_private_var) { + struct vkd3d_shader_register dst_reg = *reg; + dst_reg.data_type = VKD3D_DATA_FLOAT; + type_id = vkd3d_spirv_get_type_id(builder, component_type, input_component_count); - for (i = 0; i < max(array_sizes[0], 1); ++i) - { - struct vkd3d_shader_register dst_reg = *reg; - dst_reg.data_type = VKD3D_DATA_FLOAT; - val_id = input_id; - if (array_sizes[0]) - { - ptr_type_id = vkd3d_spirv_get_op_type_pointer(builder, SpvStorageClassInput, type_id); - index = spirv_compiler_get_constant_uint(compiler, i); - val_id = vkd3d_spirv_build_op_in_bounds_access_chain1(builder, ptr_type_id, input_id, index); - dst_reg.idx[0].offset = i; - } - else if (builtin && builtin->spirv_array_size) - { - /* The D3D builtin is not an array, but the SPIR-V builtin is, - * so we'll need to index into the SPIR-V builtin when loading - * it. This happens when reading TessLevel in domain shaders. */ - ptr_type_id = vkd3d_spirv_get_op_type_pointer(builder, SpvStorageClassInput, type_id); - index = spirv_compiler_get_constant_uint(compiler, builtin->member_idx); - val_id = vkd3d_spirv_build_op_in_bounds_access_chain1(builder, ptr_type_id, input_id, index); - dst_reg.idx[0].offset = element_idx + i; - } - val_id = vkd3d_spirv_build_op_load(builder, type_id, val_id, SpvMemoryAccessMaskNone); + val_id = vkd3d_spirv_build_op_load(builder, type_id, input_id, SpvMemoryAccessMaskNone); - if (builtin && builtin->fixup_pfn) - val_id = builtin->fixup_pfn(compiler, val_id); + if (builtin && builtin->fixup_pfn) + val_id = builtin->fixup_pfn(compiler, val_id); - if (component_type != VKD3D_SHADER_COMPONENT_FLOAT) - { - float_type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_FLOAT, input_component_count); - val_id = vkd3d_spirv_build_op_bitcast(builder, float_type_id, val_id); - } + if (component_type != VKD3D_SHADER_COMPONENT_FLOAT) + { + float_type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_FLOAT, input_component_count); + val_id = vkd3d_spirv_build_op_bitcast(builder, float_type_id, val_id); + } - val_id = spirv_compiler_emit_swizzle(compiler, val_id, - vkd3d_write_mask_from_component_count(input_component_count), - VKD3D_SHADER_COMPONENT_FLOAT, VKD3D_SHADER_NO_SWIZZLE, dst->write_mask >> component_idx); + val_id = spirv_compiler_emit_swizzle(compiler, val_id, + vkd3d_write_mask_from_component_count(input_component_count), + VKD3D_SHADER_COMPONENT_FLOAT, VKD3D_SHADER_NO_SWIZZLE, dst->write_mask >> component_idx); - spirv_compiler_emit_store_reg(compiler, &dst_reg, dst->write_mask, val_id); - } + spirv_compiler_emit_store_reg(compiler, &dst_reg, dst->write_mask, val_id); } return input_id; @@ -4747,30 +4861,6 @@ static void spirv_compiler_emit_input_register(struct spirv_compiler *compiler, spirv_compiler_emit_register_debug_name(builder, input_id, reg); } -static void spirv_compiler_emit_shader_phase_input(struct spirv_compiler *compiler, - const struct vkd3d_shader_dst_param *dst) -{ - const struct vkd3d_shader_register *reg = &dst->reg; - - switch (reg->type) - { - case VKD3DSPR_INPUT: - case VKD3DSPR_INCONTROLPOINT: - case VKD3DSPR_PATCHCONST: - spirv_compiler_emit_input(compiler, dst, VKD3D_SIV_NONE, VKD3DSIM_NONE); - return; - case VKD3DSPR_PRIMID: - spirv_compiler_emit_input_register(compiler, dst); - return; - case VKD3DSPR_OUTPOINTID: /* Emitted in spirv_compiler_emit_initial_declarations(). */ - case VKD3DSPR_OUTCONTROLPOINT: /* See spirv_compiler_leave_shader_phase(). */ - return; - default: - FIXME("Unhandled shader phase input register %#x.\n", reg->type); - return; - } -} - static unsigned int get_shader_output_swizzle(const struct spirv_compiler *compiler, unsigned int register_idx) { @@ -4835,7 +4925,7 @@ static void spirv_compiler_emit_shader_signature_outputs(struct spirv_compiler * if (clip_distance_mask) { count = vkd3d_popcount(clip_distance_mask); - builtin = get_spirv_builtin_for_sysval(compiler, VKD3D_SIV_CLIP_DISTANCE); + builtin = get_spirv_builtin_for_sysval(compiler, VKD3D_SHADER_SV_CLIP_DISTANCE); clip_distance_id = spirv_compiler_emit_builtin_variable(compiler, builtin, SpvStorageClassOutput, count); } @@ -4843,7 +4933,7 @@ static void spirv_compiler_emit_shader_signature_outputs(struct spirv_compiler * if (cull_distance_mask) { count = vkd3d_popcount(cull_distance_mask); - builtin = get_spirv_builtin_for_sysval(compiler, VKD3D_SIV_CULL_DISTANCE); + builtin = get_spirv_builtin_for_sysval(compiler, VKD3D_SHADER_SV_CULL_DISTANCE); cull_distance_id = spirv_compiler_emit_builtin_variable(compiler, builtin, SpvStorageClassOutput, count); } @@ -4904,7 +4994,7 @@ static void spirv_compiler_emit_output_register(struct spirv_compiler *compiler, } static uint32_t spirv_compiler_emit_shader_phase_builtin_variable(struct spirv_compiler *compiler, - const struct vkd3d_spirv_builtin *builtin) + const struct vkd3d_spirv_builtin *builtin, const unsigned int *array_sizes, unsigned int size_count) { struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; uint32_t *variable_id, id; @@ -4919,7 +5009,7 @@ static uint32_t spirv_compiler_emit_shader_phase_builtin_variable(struct spirv_c if (variable_id && *variable_id) return *variable_id; - id = spirv_compiler_emit_builtin_variable(compiler, builtin, SpvStorageClassOutput, 0); + id = spirv_compiler_emit_builtin_variable_v(compiler, builtin, SpvStorageClassOutput, array_sizes, size_count); if (is_in_fork_or_join_phase(compiler)) vkd3d_spirv_build_op_decorate(builder, id, SpvDecorationPatch, NULL, 0); @@ -4928,8 +5018,7 @@ static uint32_t spirv_compiler_emit_shader_phase_builtin_variable(struct spirv_c return id; } -static void spirv_compiler_emit_output(struct spirv_compiler *compiler, - const struct vkd3d_shader_dst_param *dst, enum vkd3d_shader_input_sysval_semantic sysval) +static void spirv_compiler_emit_output(struct spirv_compiler *compiler, const struct vkd3d_shader_dst_param *dst) { struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; const struct vkd3d_shader_register *reg = &dst->reg; @@ -4938,6 +5027,7 @@ static void spirv_compiler_emit_output(struct spirv_compiler *compiler, enum vkd3d_shader_component_type component_type; const struct shader_signature *shader_signature; const struct vkd3d_spirv_builtin *builtin; + enum vkd3d_shader_sysval_semantic sysval; unsigned int write_mask, reg_write_mask; bool use_private_variable = false; struct vkd3d_symbol reg_symbol; @@ -4951,8 +5041,16 @@ static void spirv_compiler_emit_output(struct spirv_compiler *compiler, shader_signature = is_patch_constant ? &compiler->patch_constant_signature : &compiler->output_signature; - element_idx = shader_register_get_io_indices(reg, array_sizes); + element_idx = reg->idx[reg->idx_count - 1].offset; signature_element = &shader_signature->elements[element_idx]; + sysval = signature_element->sysval_semantic; + /* Don't use builtins for TCS -> TES varyings. See spirv_compiler_emit_input(). */ + if (compiler->shader_type == VKD3D_SHADER_TYPE_HULL && !is_patch_constant) + sysval = VKD3D_SHADER_SV_NONE; + array_sizes[0] = (reg->type == VKD3DSPR_PATCHCONST ? 0 : compiler->output_control_point_count); + array_sizes[1] = signature_element->register_count; + if (array_sizes[1] == 1 && !vsir_sysval_semantic_is_tess_factor(signature_element->sysval_semantic)) + array_sizes[1] = 0; builtin = vkd3d_get_spirv_builtin(compiler, dst->reg.type, sysval); @@ -4997,7 +5095,7 @@ static void spirv_compiler_emit_output(struct spirv_compiler *compiler, else if (builtin) { if (spirv_compiler_get_current_shader_phase(compiler)) - id = spirv_compiler_emit_shader_phase_builtin_variable(compiler, builtin); + id = spirv_compiler_emit_shader_phase_builtin_variable(compiler, builtin, array_sizes, 2); else id = spirv_compiler_emit_builtin_variable_v(compiler, builtin, storage_class, array_sizes, 2); @@ -5015,6 +5113,9 @@ static void spirv_compiler_emit_output(struct spirv_compiler *compiler, if (is_patch_constant) location += shader_signature_next_location(&compiler->output_signature); + else if (compiler->shader_type == VKD3D_SHADER_TYPE_PIXEL + && signature_element->sysval_semantic == VKD3D_SHADER_SV_TARGET) + location = signature_element->semantic_index; id = spirv_compiler_emit_array_variable(compiler, &builder->global_stream, storage_class, component_type, output_component_count, array_sizes, 2); @@ -5073,16 +5174,15 @@ static void spirv_compiler_emit_output(struct spirv_compiler *compiler, static uint32_t spirv_compiler_get_output_array_index(struct spirv_compiler *compiler, const struct signature_element *e) { - enum vkd3d_shader_input_sysval_semantic sysval; + enum vkd3d_shader_sysval_semantic sysval = e->sysval_semantic; const struct vkd3d_spirv_builtin *builtin; - sysval = vkd3d_siv_from_sysval_indexed(e->sysval_semantic, e->semantic_index); builtin = get_spirv_builtin_for_sysval(compiler, sysval); switch (sysval) { - case VKD3D_SIV_LINE_DETAIL_TESS_FACTOR: - case VKD3D_SIV_LINE_DENSITY_TESS_FACTOR: + case VKD3D_SHADER_SV_TESS_FACTOR_LINEDEN: + case VKD3D_SHADER_SV_TESS_FACTOR_LINEDET: return builtin->member_idx; default: return e->semantic_index; @@ -5196,7 +5296,6 @@ static void spirv_compiler_emit_shader_epilogue_function(struct spirv_compiler * STATIC_ASSERT(ARRAY_SIZE(compiler->private_output_variable) == ARRAY_SIZE(param_id)); STATIC_ASSERT(ARRAY_SIZE(compiler->private_output_variable) == ARRAY_SIZE(param_type_id)); - STATIC_ASSERT(ARRAY_SIZE(compiler->private_output_variable) == ARRAY_SIZE(compiler->private_output_variable_array_idx)); STATIC_ASSERT(ARRAY_SIZE(compiler->private_output_variable) == ARRAY_SIZE(compiler->private_output_variable_write_mask)); is_patch_constant = is_in_fork_or_join_phase(compiler); @@ -5252,7 +5351,6 @@ static void spirv_compiler_emit_shader_epilogue_function(struct spirv_compiler * vkd3d_spirv_build_op_function_end(builder); memset(compiler->private_output_variable, 0, sizeof(compiler->private_output_variable)); - memset(compiler->private_output_variable_array_idx, 0, sizeof(compiler->private_output_variable_array_idx)); memset(compiler->private_output_variable_write_mask, 0, sizeof(compiler->private_output_variable_write_mask)); compiler->epilogue_function_id = 0; } @@ -5262,10 +5360,7 @@ static void spirv_compiler_emit_hull_shader_builtins(struct spirv_compiler *comp struct vkd3d_shader_dst_param dst; memset(&dst, 0, sizeof(dst)); - dst.reg.type = VKD3DSPR_OUTPOINTID; - dst.reg.idx[0].offset = ~0u; - dst.reg.idx[1].offset = ~0u; - dst.reg.idx_count = 0; + vsir_register_init(&dst.reg, VKD3DSPR_OUTPOINTID, VKD3D_DATA_FLOAT, 0); dst.write_mask = VKD3DSP_WRITEMASK_0; spirv_compiler_emit_input_register(compiler, &dst); } @@ -5293,7 +5388,7 @@ static void spirv_compiler_emit_initial_declarations(struct spirv_compiler *comp break; case VKD3D_SHADER_TYPE_PIXEL: vkd3d_spirv_set_execution_model(builder, SpvExecutionModelFragment); - spirv_compiler_emit_execution_mode(compiler, SpvExecutionModeOriginUpperLeft, NULL, 0); + spirv_compiler_emit_execution_mode(compiler, compiler->fragment_coordinate_origin, NULL, 0); break; case VKD3D_SHADER_TYPE_COMPUTE: vkd3d_spirv_set_execution_model(builder, SpvExecutionModelGLCompute); @@ -5329,7 +5424,7 @@ static size_t spirv_compiler_get_current_function_location(struct spirv_compiler static void spirv_compiler_emit_dcl_global_flags(struct spirv_compiler *compiler, const struct vkd3d_shader_instruction *instruction) { - unsigned int flags = instruction->flags; + enum vkd3d_shader_global_flags flags = instruction->declaration.global_flags; if (flags & VKD3DSGF_FORCE_EARLY_DEPTH_STENCIL) { @@ -5344,9 +5439,9 @@ static void spirv_compiler_emit_dcl_global_flags(struct spirv_compiler *compiler } if (flags & ~(VKD3DSGF_REFACTORING_ALLOWED | VKD3DSGF_ENABLE_RAW_AND_STRUCTURED_BUFFERS)) - FIXME("Unhandled global flags %#x.\n", flags); + FIXME("Unhandled global flags %#"PRIx64".\n", (uint64_t)flags); else - WARN("Unhandled global flags %#x.\n", flags); + WARN("Unhandled global flags %#"PRIx64".\n", (uint64_t)flags); } static void spirv_compiler_emit_temps(struct spirv_compiler *compiler, uint32_t count) @@ -5375,6 +5470,18 @@ static void spirv_compiler_emit_temps(struct spirv_compiler *compiler, uint32_t vkd3d_spirv_end_function_stream_insertion(builder); } +static void spirv_compiler_allocate_ssa_register_ids(struct spirv_compiler *compiler, unsigned int count) +{ + assert(!compiler->ssa_register_info); + if (!(compiler->ssa_register_info = vkd3d_calloc(count, sizeof(*compiler->ssa_register_info)))) + { + ERR("Failed to allocate SSA register value id array, count %u.\n", count); + spirv_compiler_error(compiler, VKD3D_SHADER_ERROR_SPV_OUT_OF_MEMORY, + "Failed to allocate SSA register value id array of count %u.", count); + } + compiler->ssa_register_count = count; +} + static void spirv_compiler_emit_dcl_indexable_temp(struct spirv_compiler *compiler, const struct vkd3d_shader_instruction *instruction) { @@ -5388,11 +5495,17 @@ static void spirv_compiler_emit_dcl_indexable_temp(struct spirv_compiler *compil if (temp->component_count != 4) FIXME("Unhandled component count %u.\n", temp->component_count); - memset(®, 0, sizeof(reg)); - reg.type = VKD3DSPR_IDXTEMP; + vsir_register_init(®, VKD3DSPR_IDXTEMP, VKD3D_DATA_FLOAT, 1); reg.idx[0].offset = temp->register_idx; - reg.idx[1].offset = ~0u; - reg.idx_count = 1; + + if (temp->alignment) + WARN("Ignoring alignment %u.\n", temp->alignment); + if (temp->initialiser) + { + FIXME("Initialisers are not supported.\n"); + spirv_compiler_error(compiler, VKD3D_SHADER_ERROR_SPV_NOT_IMPLEMENTED, + "Initialisers for indexable temps are not supported."); + } function_location = spirv_compiler_get_current_function_location(compiler); vkd3d_spirv_begin_function_stream_insertion(builder, function_location); @@ -5461,13 +5574,14 @@ static void spirv_compiler_emit_push_constant_buffers(struct spirv_compiler *com struct_id = vkd3d_spirv_build_op_type_struct(builder, member_ids, count); vkd3d_spirv_build_op_decorate(builder, struct_id, SpvDecorationBlock, NULL, 0); - vkd3d_spirv_build_op_name(builder, struct_id, "push_cb"); + vkd3d_spirv_build_op_name(builder, struct_id, "push_cb_struct"); vkd3d_free(member_ids); pointer_type_id = vkd3d_spirv_get_op_type_pointer(builder, storage_class, struct_id); var_id = vkd3d_spirv_build_op_variable(builder, &builder->global_stream, pointer_type_id, storage_class, 0); compiler->push_constants_var_id = var_id; + vkd3d_spirv_build_op_name(builder, var_id, "push_cb"); for (i = 0, j = 0; i < compiler->shader_interface.push_constant_buffer_count; ++i) { @@ -5571,17 +5685,17 @@ static void spirv_compiler_emit_cbv_declaration(struct spirv_compiler *compiler, const SpvStorageClass storage_class = SpvStorageClassUniform; struct vkd3d_push_constant_buffer_binding *push_cb; struct vkd3d_descriptor_variable_info var_info; + struct vkd3d_shader_register reg; struct vkd3d_symbol reg_symbol; unsigned int size; - struct vkd3d_shader_register reg = - { - .type = VKD3DSPR_CONSTBUFFER, - .idx[0].offset = register_id, - .idx_count = 1, - }; + vsir_register_init(®, VKD3DSPR_CONSTBUFFER, VKD3D_DATA_FLOAT, 3); + reg.idx[0].offset = register_id; + reg.idx[1].offset = range->first; + reg.idx[2].offset = range->last; - size = size_in_bytes / (VKD3D_VEC4_SIZE * sizeof(uint32_t)); + size = align(size_in_bytes, VKD3D_VEC4_SIZE * sizeof(uint32_t)); + size /= VKD3D_VEC4_SIZE * sizeof(uint32_t); if ((push_cb = spirv_compiler_find_push_constant_buffer(compiler, range))) { @@ -5629,23 +5743,25 @@ static void spirv_compiler_emit_dcl_immediate_constant_buffer(struct spirv_compi struct vkd3d_symbol reg_symbol; unsigned int i; - if (!(elements = vkd3d_calloc(icb->vec4_count, sizeof(*elements)))) + assert(icb->data_type == VKD3D_DATA_FLOAT); + assert(icb->component_count == VKD3D_VEC4_SIZE); + + if (!(elements = vkd3d_calloc(icb->element_count, sizeof(*elements)))) return; - for (i = 0; i < icb->vec4_count; ++i) + for (i = 0; i < icb->element_count; ++i) elements[i] = spirv_compiler_get_constant(compiler, VKD3D_SHADER_COMPONENT_FLOAT, VKD3D_VEC4_SIZE, &icb->data[4 * i]); type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_FLOAT, VKD3D_VEC4_SIZE); - length_id = spirv_compiler_get_constant_uint(compiler, icb->vec4_count); + length_id = spirv_compiler_get_constant_uint(compiler, icb->element_count); type_id = vkd3d_spirv_get_op_type_array(builder, type_id, length_id); - const_id = vkd3d_spirv_build_op_constant_composite(builder, type_id, elements, icb->vec4_count); + const_id = vkd3d_spirv_build_op_constant_composite(builder, type_id, elements, icb->element_count); ptr_type_id = vkd3d_spirv_get_op_type_pointer(builder, SpvStorageClassPrivate, type_id); icb_id = vkd3d_spirv_build_op_variable(builder, &builder->global_stream, ptr_type_id, SpvStorageClassPrivate, const_id); vkd3d_spirv_build_op_name(builder, icb_id, "icb"); vkd3d_free(elements); - memset(®, 0, sizeof(reg)); - reg.type = VKD3DSPR_IMMCONSTBUFFER; + vsir_register_init(®, VKD3DSPR_IMMCONSTBUFFER, VKD3D_DATA_FLOAT, 0); vkd3d_symbol_make_register(®_symbol, ®); vkd3d_symbol_set_register_info(®_symbol, icb_id, SpvStorageClassPrivate, VKD3D_SHADER_COMPONENT_FLOAT, VKD3DSP_WRITEMASK_ALL); @@ -5658,15 +5774,12 @@ static void spirv_compiler_emit_sampler_declaration(struct spirv_compiler *compi const SpvStorageClass storage_class = SpvStorageClassUniformConstant; struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; struct vkd3d_descriptor_variable_info var_info; + struct vkd3d_shader_register reg; struct vkd3d_symbol reg_symbol; uint32_t type_id, var_id; - const struct vkd3d_shader_register reg = - { - .type = VKD3DSPR_SAMPLER, - .idx[0].offset = register_id, - .idx_count = 1, - }; + vsir_register_init(®, VKD3DSPR_SAMPLER, VKD3D_DATA_FLOAT, 1); + reg.idx[0].offset = register_id; vkd3d_symbol_make_sampler(®_symbol, ®); reg_symbol.info.sampler.range = *range; @@ -5847,6 +5960,7 @@ static void spirv_compiler_emit_combined_sampler_declarations(struct spirv_compi current->sampler_index == VKD3D_SHADER_DUMMY_SAMPLER_INDEX ? 0 : current->sampler_space, current->sampler_index); symbol.id = var_id; + symbol.descriptor_array = NULL; symbol.info.resource.range = *resource_range; symbol.info.resource.sampled_type = sampled_type; symbol.info.resource.type_id = image_type_id; @@ -5872,13 +5986,10 @@ static void spirv_compiler_emit_resource_declaration(struct spirv_compiler *comp const struct vkd3d_spirv_resource_type *resource_type_info; enum vkd3d_shader_component_type sampled_type; struct vkd3d_symbol resource_symbol; + struct vkd3d_shader_register reg; - struct vkd3d_shader_register reg = - { - .type = is_uav ? VKD3DSPR_UAV : VKD3DSPR_RESOURCE, - .idx[0].offset = register_id, - .idx_count = 1, - }; + vsir_register_init(®, is_uav ? VKD3DSPR_UAV : VKD3DSPR_RESOURCE, VKD3D_DATA_FLOAT, 1); + reg.idx[0].offset = register_id; if (resource_type == VKD3D_SHADER_RESOURCE_TEXTURE_2DMS && sample_count == 1) resource_type = VKD3D_SHADER_RESOURCE_TEXTURE_2D; @@ -6030,36 +6141,17 @@ static void spirv_compiler_emit_dcl_input(struct spirv_compiler *compiler, { const struct vkd3d_shader_dst_param *dst = &instruction->declaration.dst; - if (spirv_compiler_get_current_shader_phase(compiler)) - spirv_compiler_emit_shader_phase_input(compiler, dst); - else if (vkd3d_shader_register_is_input(&dst->reg) || dst->reg.type == VKD3DSPR_PATCHCONST) - spirv_compiler_emit_input(compiler, dst, VKD3D_SIV_NONE, VKD3DSIM_NONE); - else + /* OUTPOINTID is handled in spirv_compiler_emit_hull_shader_builtins(). */ + if (dst->reg.type == VKD3DSPR_INPUT || dst->reg.type == VKD3DSPR_PATCHCONST) + spirv_compiler_emit_input(compiler, dst); + else if (dst->reg.type != VKD3DSPR_OUTPOINTID) spirv_compiler_emit_input_register(compiler, dst); - - if (dst->reg.type == VKD3DSPR_OUTCONTROLPOINT) - compiler->use_vocp = true; -} - -static void spirv_compiler_emit_dcl_input_ps(struct spirv_compiler *compiler, - const struct vkd3d_shader_instruction *instruction) -{ - spirv_compiler_emit_input(compiler, &instruction->declaration.dst, VKD3D_SIV_NONE, instruction->flags); -} - -static void spirv_compiler_emit_dcl_input_ps_sysval(struct spirv_compiler *compiler, - const struct vkd3d_shader_instruction *instruction) -{ - const struct vkd3d_shader_register_semantic *semantic = &instruction->declaration.register_semantic; - - spirv_compiler_emit_input(compiler, &semantic->reg, semantic->sysval_semantic, instruction->flags); } static void spirv_compiler_emit_dcl_input_sysval(struct spirv_compiler *compiler, const struct vkd3d_shader_instruction *instruction) { - spirv_compiler_emit_input(compiler, &instruction->declaration.register_semantic.reg, - instruction->declaration.register_semantic.sysval_semantic, VKD3DSIM_NONE); + spirv_compiler_emit_input(compiler, &instruction->declaration.register_semantic.reg); } static void spirv_compiler_emit_dcl_output(struct spirv_compiler *compiler, @@ -6067,9 +6159,9 @@ static void spirv_compiler_emit_dcl_output(struct spirv_compiler *compiler, { const struct vkd3d_shader_dst_param *dst = &instruction->declaration.dst; - if (vkd3d_shader_register_is_output(&dst->reg) - || (is_in_fork_or_join_phase(compiler) && vkd3d_shader_register_is_patch_constant(&dst->reg))) - spirv_compiler_emit_output(compiler, dst, VKD3D_SIV_NONE); + if (dst->reg.type == VKD3DSPR_OUTPUT + || (is_in_fork_or_join_phase(compiler) && dst->reg.type == VKD3DSPR_PATCHCONST)) + spirv_compiler_emit_output(compiler, dst); else spirv_compiler_emit_output_register(compiler, dst); } @@ -6077,13 +6169,7 @@ static void spirv_compiler_emit_dcl_output(struct spirv_compiler *compiler, static void spirv_compiler_emit_dcl_output_siv(struct spirv_compiler *compiler, const struct vkd3d_shader_instruction *instruction) { - enum vkd3d_shader_input_sysval_semantic sysval; - const struct vkd3d_shader_dst_param *dst; - - dst = &instruction->declaration.register_semantic.reg; - sysval = instruction->declaration.register_semantic.sysval_semantic; - - spirv_compiler_emit_output(compiler, dst, sysval); + spirv_compiler_emit_output(compiler, &instruction->declaration.register_semantic.reg); } static void spirv_compiler_emit_dcl_stream(struct spirv_compiler *compiler, @@ -6305,7 +6391,6 @@ static void spirv_compiler_leave_shader_phase(struct spirv_compiler *compiler) * Control point phase has separate output registers. */ memset(compiler->output_info, 0, signature->element_count * sizeof(*compiler->output_info)); memset(compiler->private_output_variable, 0, sizeof(compiler->private_output_variable)); - memset(compiler->private_output_variable_array_idx, 0, sizeof(compiler->private_output_variable_array_idx)); memset(compiler->private_output_variable_write_mask, 0, sizeof(compiler->private_output_variable_write_mask)); } } @@ -6361,20 +6446,13 @@ static void spirv_compiler_emit_default_control_point_phase(struct spirv_compile invocation_id = spirv_compiler_emit_load_invocation_id(compiler); memset(&invocation, 0, sizeof(invocation)); - invocation.reg.type = VKD3DSPR_OUTPOINTID; - invocation.reg.data_type = VKD3D_DATA_INT; - invocation.reg.idx[0].offset = ~0u; - invocation.reg.idx[1].offset = ~0u; - invocation.reg.idx[2].offset = ~0u; - invocation.reg.idx_count = 0; + vsir_register_init(&invocation.reg, VKD3DSPR_OUTPOINTID, VKD3D_DATA_INT, 0); invocation.swizzle = VKD3D_SHADER_NO_SWIZZLE; - memset(&input_reg, 0, sizeof(input_reg)); - input_reg.type = VKD3DSPR_INPUT; - input_reg.data_type = VKD3D_DATA_FLOAT; + vsir_register_init(&input_reg, VKD3DSPR_INPUT, VKD3D_DATA_FLOAT, 2); + input_reg.idx[0].offset = 0; input_reg.idx[0].rel_addr = &invocation; - input_reg.idx[2].offset = ~0u; - input_reg.idx_count = 2; + input_reg.idx[1].offset = 0; input_id = spirv_compiler_get_register_id(compiler, &input_reg); assert(input_signature->element_count == output_signature->element_count); @@ -6441,36 +6519,17 @@ static void spirv_compiler_emit_hull_shader_barrier(struct spirv_compiler *compi static void spirv_compiler_emit_shader_epilogue_invocation(struct spirv_compiler *compiler) { struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; - uint32_t void_id, type_id, ptr_type_id, function_id; uint32_t arguments[MAX_REG_OUTPUT]; + uint32_t void_id, function_id; unsigned int i, count; if ((function_id = compiler->epilogue_function_id)) { void_id = vkd3d_spirv_get_op_type_void(builder); - type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_FLOAT, 4); - ptr_type_id = vkd3d_spirv_get_op_type_pointer(builder, SpvStorageClassPrivate, type_id); for (i = 0, count = 0; i < ARRAY_SIZE(compiler->private_output_variable); ++i) { if (compiler->private_output_variable[i]) - { - uint32_t argument_id = compiler->private_output_variable[i]; - unsigned int argument_idx = count++; - - if (compiler->private_output_variable_array_idx[i]) - { - uint32_t tmp_id; - - tmp_id = vkd3d_spirv_build_op_access_chain1(builder, ptr_type_id, - argument_id, compiler->private_output_variable_array_idx[i]); - tmp_id = vkd3d_spirv_build_op_load(builder, type_id, tmp_id, SpvMemoryAccessMaskNone); - argument_id = vkd3d_spirv_build_op_variable(builder, - &builder->global_stream, ptr_type_id, SpvStorageClassPrivate, 0); - vkd3d_spirv_build_op_store(builder, argument_id, tmp_id, SpvMemoryAccessMaskNone); - } - - arguments[argument_idx] = argument_id; - } + arguments[count++] = compiler->private_output_variable[i]; } vkd3d_spirv_build_op_function_call(builder, void_id, function_id, arguments, count); @@ -6520,9 +6579,8 @@ static SpvOp spirv_compiler_map_alu_instruction(const struct vkd3d_shader_instru {VKD3DSIH_DTOF, SpvOpFConvert}, {VKD3DSIH_DTOI, SpvOpConvertFToS}, {VKD3DSIH_DTOU, SpvOpConvertFToU}, + {VKD3DSIH_FREM, SpvOpFRem}, {VKD3DSIH_FTOD, SpvOpFConvert}, - {VKD3DSIH_FTOI, SpvOpConvertFToS}, - {VKD3DSIH_FTOU, SpvOpConvertFToU}, {VKD3DSIH_IADD, SpvOpIAdd}, {VKD3DSIH_INEG, SpvOpSNegate}, {VKD3DSIH_ISHL, SpvOpShiftLeftLogical}, @@ -6548,7 +6606,55 @@ static SpvOp spirv_compiler_map_alu_instruction(const struct vkd3d_shader_instru return SpvOpMax; } -static void spirv_compiler_emit_alu_instruction(struct spirv_compiler *compiler, +static SpvOp spirv_compiler_map_logical_instruction(const struct vkd3d_shader_instruction *instruction) +{ + switch (instruction->handler_idx) + { + case VKD3DSIH_AND: + return SpvOpLogicalAnd; + case VKD3DSIH_OR: + return SpvOpLogicalOr; + case VKD3DSIH_XOR: + return SpvOpLogicalNotEqual; + default: + return SpvOpMax; + } +} + +static void spirv_compiler_emit_bool_cast(struct spirv_compiler *compiler, + const struct vkd3d_shader_instruction *instruction) +{ + const struct vkd3d_shader_dst_param *dst = instruction->dst; + const struct vkd3d_shader_src_param *src = instruction->src; + uint32_t val_id; + + assert(src->reg.data_type == VKD3D_DATA_BOOL && dst->reg.data_type != VKD3D_DATA_BOOL); + + val_id = spirv_compiler_emit_load_src(compiler, src, dst->write_mask); + if (dst->reg.data_type == VKD3D_DATA_FLOAT) + { + val_id = spirv_compiler_emit_bool_to_float(compiler, 1, val_id, instruction->handler_idx == VKD3DSIH_ITOF); + } + else if (dst->reg.data_type == VKD3D_DATA_DOUBLE) + { + /* ITOD is not supported. Frontends which emit bool casts must use ITOF for double. */ + val_id = spirv_compiler_emit_bool_to_double(compiler, 1, val_id, instruction->handler_idx == VKD3DSIH_ITOF); + } + else if (dst->reg.data_type == VKD3D_DATA_UINT) + { + val_id = spirv_compiler_emit_bool_to_int(compiler, 1, val_id, instruction->handler_idx == VKD3DSIH_ITOI); + } + else + { + WARN("Unhandled data type %u.\n", dst->reg.data_type); + spirv_compiler_error(compiler, VKD3D_SHADER_ERROR_SPV_INVALID_TYPE, + "Register data type %u is unhandled.", dst->reg.data_type); + } + + spirv_compiler_emit_store_dst(compiler, dst, val_id); +} + +static enum vkd3d_result spirv_compiler_emit_alu_instruction(struct spirv_compiler *compiler, const struct vkd3d_shader_instruction *instruction) { struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; @@ -6556,14 +6662,36 @@ static void spirv_compiler_emit_alu_instruction(struct spirv_compiler *compiler, const struct vkd3d_shader_src_param *src = instruction->src; uint32_t src_ids[SPIRV_MAX_SRC_COUNT]; uint32_t type_id, val_id; + SpvOp op = SpvOpMax; unsigned int i; - SpvOp op; - op = spirv_compiler_map_alu_instruction(instruction); + if (src->reg.data_type == VKD3D_DATA_BOOL) + { + if (dst->reg.data_type == VKD3D_DATA_BOOL) + { + /* VSIR supports logic ops AND/OR/XOR on bool values. */ + op = spirv_compiler_map_logical_instruction(instruction); + } + else if (instruction->handler_idx == VKD3DSIH_ITOF || instruction->handler_idx == VKD3DSIH_UTOF + || instruction->handler_idx == VKD3DSIH_ITOI || instruction->handler_idx == VKD3DSIH_UTOU) + { + /* VSIR supports cast from bool to signed/unsigned integer types and floating point types, + * where bool is treated as a 1-bit integer and a signed 'true' value converts to -1. */ + spirv_compiler_emit_bool_cast(compiler, instruction); + return VKD3D_OK; + } + } + else + { + op = spirv_compiler_map_alu_instruction(instruction); + } + if (op == SpvOpMax) { ERR("Unexpected instruction %#x.\n", instruction->handler_idx); - return; + spirv_compiler_error(compiler, VKD3D_SHADER_ERROR_SPV_INVALID_HANDLER, + "Encountered invalid/unhandled instruction handler %#x.", instruction->handler_idx); + return VKD3D_ERROR_INVALID_SHADER; } assert(instruction->dst_count == 1); @@ -6595,6 +6723,7 @@ static void spirv_compiler_emit_alu_instruction(struct spirv_compiler *compiler, vkd3d_spirv_build_op_decorate(builder, val_id, SpvDecorationNoContraction, NULL, 0); spirv_compiler_emit_store_dst(compiler, dst, val_id); + return VKD3D_OK; } static enum GLSLstd450 spirv_compiler_map_ext_glsl_instruction( @@ -6694,7 +6823,8 @@ static void spirv_compiler_emit_mov(struct spirv_compiler *compiler, uint32_t components[VKD3D_VEC4_SIZE]; unsigned int i, component_count; - if (register_is_constant_or_undef(&src->reg) || dst->modifiers || src->modifiers) + if (register_is_constant_or_undef(&src->reg) || src->reg.type == VKD3DSPR_SSA || dst->reg.type == VKD3DSPR_SSA + || dst->modifiers || src->modifiers) goto general_implementation; spirv_compiler_get_register_info(compiler, &dst->reg, &dst_reg_info); @@ -6760,8 +6890,9 @@ static void spirv_compiler_emit_movc(struct spirv_compiler *compiler, component_count = vkd3d_write_mask_component_count(dst->write_mask); type_id = spirv_compiler_get_type_id_for_dst(compiler, dst); - condition_id = spirv_compiler_emit_int_to_bool(compiler, - VKD3D_SHADER_CONDITIONAL_OP_NZ, component_count, condition_id); + if (src[0].reg.data_type != VKD3D_DATA_BOOL) + condition_id = spirv_compiler_emit_int_to_bool(compiler, + VKD3D_SHADER_CONDITIONAL_OP_NZ, component_count, condition_id); val_id = vkd3d_spirv_build_op_select(builder, type_id, condition_id, src1_id, src2_id); spirv_compiler_emit_store_dst(compiler, dst, val_id); @@ -6899,7 +7030,7 @@ static void spirv_compiler_emit_imul(struct spirv_compiler *compiler, uint32_t type_id, val_id, src0_id, src1_id; if (dst[0].reg.type != VKD3DSPR_NULL) - FIXME("Extended multiplies not implemented.\n"); /* SpvOpSMulExtended */ + FIXME("Extended multiplies not implemented.\n"); /* SpvOpSMulExtended/SpvOpUMulExtended */ if (dst[1].reg.type == VKD3DSPR_NULL) return; @@ -6935,7 +7066,7 @@ static void spirv_compiler_emit_imad(struct spirv_compiler *compiler, spirv_compiler_emit_store_dst(compiler, dst, val_id); } -static void spirv_compiler_emit_udiv(struct spirv_compiler *compiler, +static void spirv_compiler_emit_int_div(struct spirv_compiler *compiler, const struct vkd3d_shader_instruction *instruction) { uint32_t type_id, val_id, src0_id, src1_id, condition_id, uint_max_id; @@ -6943,6 +7074,10 @@ static void spirv_compiler_emit_udiv(struct spirv_compiler *compiler, const struct vkd3d_shader_dst_param *dst = instruction->dst; const struct vkd3d_shader_src_param *src = instruction->src; unsigned int component_count = 0; + SpvOp div_op, mod_op; + + div_op = instruction->handler_idx == VKD3DSIH_IDIV ? SpvOpSDiv : SpvOpUDiv; + mod_op = instruction->handler_idx == VKD3DSIH_IDIV ? SpvOpSRem : SpvOpUMod; if (dst[0].reg.type != VKD3DSPR_NULL) { @@ -6957,7 +7092,7 @@ static void spirv_compiler_emit_udiv(struct spirv_compiler *compiler, uint_max_id = spirv_compiler_get_constant_uint_vector(compiler, 0xffffffff, component_count); - val_id = vkd3d_spirv_build_op_udiv(builder, type_id, src0_id, src1_id); + val_id = vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, div_op, type_id, src0_id, src1_id); /* The SPIR-V spec says: "The resulting value is undefined if Operand 2 is 0." */ val_id = vkd3d_spirv_build_op_select(builder, type_id, condition_id, val_id, uint_max_id); @@ -6980,7 +7115,7 @@ static void spirv_compiler_emit_udiv(struct spirv_compiler *compiler, 0xffffffff, component_count); } - val_id = vkd3d_spirv_build_op_umod(builder, type_id, src0_id, src1_id); + val_id = vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, mod_op, type_id, src0_id, src1_id); /* The SPIR-V spec says: "The resulting value is undefined if Operand 2 is 0." */ val_id = vkd3d_spirv_build_op_select(builder, type_id, condition_id, val_id, uint_max_id); @@ -6988,6 +7123,106 @@ static void spirv_compiler_emit_udiv(struct spirv_compiler *compiler, } } +static void spirv_compiler_emit_ftoi(struct spirv_compiler *compiler, + const struct vkd3d_shader_instruction *instruction) +{ + uint32_t src_id, int_min_id, int_max_id, zero_id, float_max_id, condition_id, val_id; + struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; + const struct vkd3d_shader_dst_param *dst = instruction->dst; + const struct vkd3d_shader_src_param *src = instruction->src; + uint32_t src_type_id, dst_type_id, condition_type_id; + enum vkd3d_shader_component_type component_type; + unsigned int component_count; + + assert(instruction->dst_count == 1); + assert(instruction->src_count == 1); + + /* OpConvertFToI has undefined results if the result cannot be represented + * as a signed integer, but Direct3D expects the result to saturate, + * and for NaN to yield zero. */ + + component_count = vkd3d_write_mask_component_count(dst->write_mask); + src_type_id = spirv_compiler_get_type_id_for_reg(compiler, &src->reg, dst->write_mask); + dst_type_id = spirv_compiler_get_type_id_for_dst(compiler, dst); + src_id = spirv_compiler_emit_load_src(compiler, src, dst->write_mask); + + if (src->reg.data_type == VKD3D_DATA_DOUBLE) + { + int_min_id = spirv_compiler_get_constant_double_vector(compiler, -2147483648.0, component_count); + float_max_id = spirv_compiler_get_constant_double_vector(compiler, 2147483648.0, component_count); + } + else + { + int_min_id = spirv_compiler_get_constant_float_vector(compiler, -2147483648.0f, component_count); + float_max_id = spirv_compiler_get_constant_float_vector(compiler, 2147483648.0f, component_count); + } + + val_id = vkd3d_spirv_build_op_glsl_std450_max(builder, src_type_id, src_id, int_min_id); + + /* VSIR allows the destination of a signed conversion to be unsigned. */ + component_type = vkd3d_component_type_from_data_type(dst->reg.data_type); + + int_max_id = spirv_compiler_get_constant_vector(compiler, component_type, component_count, INT_MAX); + condition_type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_BOOL, component_count); + condition_id = vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, + SpvOpFOrdGreaterThanEqual, condition_type_id, val_id, float_max_id); + + val_id = vkd3d_spirv_build_op_tr1(builder, &builder->function_stream, SpvOpConvertFToS, dst_type_id, val_id); + val_id = vkd3d_spirv_build_op_select(builder, dst_type_id, condition_id, int_max_id, val_id); + + zero_id = spirv_compiler_get_constant_vector(compiler, component_type, component_count, 0); + condition_id = vkd3d_spirv_build_op_tr1(builder, &builder->function_stream, SpvOpIsNan, condition_type_id, src_id); + val_id = vkd3d_spirv_build_op_select(builder, dst_type_id, condition_id, zero_id, val_id); + + spirv_compiler_emit_store_dst(compiler, dst, val_id); +} + +static void spirv_compiler_emit_ftou(struct spirv_compiler *compiler, + const struct vkd3d_shader_instruction *instruction) +{ + uint32_t src_id, zero_id, uint_max_id, float_max_id, condition_id, val_id; + struct vkd3d_spirv_builder *builder = &compiler->spirv_builder; + const struct vkd3d_shader_dst_param *dst = instruction->dst; + const struct vkd3d_shader_src_param *src = instruction->src; + uint32_t src_type_id, dst_type_id, condition_type_id; + unsigned int component_count; + + assert(instruction->dst_count == 1); + assert(instruction->src_count == 1); + + /* OpConvertFToU has undefined results if the result cannot be represented + * as an unsigned integer, but Direct3D expects the result to saturate, + * and for NaN to yield zero. */ + + component_count = vkd3d_write_mask_component_count(dst->write_mask); + src_type_id = spirv_compiler_get_type_id_for_reg(compiler, &src->reg, dst->write_mask); + dst_type_id = spirv_compiler_get_type_id_for_dst(compiler, dst); + src_id = spirv_compiler_emit_load_src(compiler, src, dst->write_mask); + + if (src->reg.data_type == VKD3D_DATA_DOUBLE) + { + zero_id = spirv_compiler_get_constant_double_vector(compiler, 0.0, component_count); + float_max_id = spirv_compiler_get_constant_double_vector(compiler, 4294967296.0, component_count); + } + else + { + zero_id = spirv_compiler_get_constant_float_vector(compiler, 0.0f, component_count); + float_max_id = spirv_compiler_get_constant_float_vector(compiler, 4294967296.0f, component_count); + } + + val_id = vkd3d_spirv_build_op_glsl_std450_max(builder, src_type_id, src_id, zero_id); + + uint_max_id = spirv_compiler_get_constant_uint_vector(compiler, UINT_MAX, component_count); + condition_type_id = vkd3d_spirv_get_type_id(builder, VKD3D_SHADER_COMPONENT_BOOL, component_count); + condition_id = vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, + SpvOpFOrdGreaterThanEqual, condition_type_id, val_id, float_max_id); + + val_id = vkd3d_spirv_build_op_tr1(builder, &builder->function_stream, SpvOpConvertFToU, dst_type_id, val_id); + val_id = vkd3d_spirv_build_op_select(builder, dst_type_id, condition_id, uint_max_id, val_id); + + spirv_compiler_emit_store_dst(compiler, dst, val_id); +} + static void spirv_compiler_emit_bitfield_instruction(struct spirv_compiler *compiler, const struct vkd3d_shader_instruction *instruction) { @@ -7123,18 +7358,22 @@ static void spirv_compiler_emit_comparison_instruction(struct spirv_compiler *co switch (instruction->handler_idx) { - case VKD3DSIH_DEQ: - case VKD3DSIH_EQ: op = SpvOpFOrdEqual; break; - case VKD3DSIH_DGE: - case VKD3DSIH_GE: op = SpvOpFOrdGreaterThanEqual; break; + case VKD3DSIH_DEQO: + case VKD3DSIH_EQO: op = SpvOpFOrdEqual; break; + case VKD3DSIH_EQU: op = SpvOpFUnordEqual; break; + case VKD3DSIH_DGEO: + case VKD3DSIH_GEO: op = SpvOpFOrdGreaterThanEqual; break; + case VKD3DSIH_GEU: op = SpvOpFUnordGreaterThanEqual; break; case VKD3DSIH_IEQ: op = SpvOpIEqual; break; case VKD3DSIH_IGE: op = SpvOpSGreaterThanEqual; break; case VKD3DSIH_ILT: op = SpvOpSLessThan; break; case VKD3DSIH_INE: op = SpvOpINotEqual; break; case VKD3DSIH_DLT: - case VKD3DSIH_LT: op = SpvOpFOrdLessThan; break; + case VKD3DSIH_LTO: op = SpvOpFOrdLessThan; break; + case VKD3DSIH_LTU: op = SpvOpFUnordLessThan; break; + case VKD3DSIH_NEO: op = SpvOpFOrdNotEqual; break; case VKD3DSIH_DNE: - case VKD3DSIH_NE: op = SpvOpFUnordNotEqual; break; + case VKD3DSIH_NEU: op = SpvOpFUnordNotEqual; break; case VKD3DSIH_UGE: op = SpvOpUGreaterThanEqual; break; case VKD3DSIH_ULT: op = SpvOpULessThan; break; default: @@ -7151,7 +7390,8 @@ static void spirv_compiler_emit_comparison_instruction(struct spirv_compiler *co result_id = vkd3d_spirv_build_op_tr2(builder, &builder->function_stream, op, type_id, src0_id, src1_id); - result_id = spirv_compiler_emit_bool_to_int(compiler, component_count, result_id); + if (dst->reg.data_type != VKD3D_DATA_BOOL) + result_id = spirv_compiler_emit_bool_to_int(compiler, component_count, result_id, true); spirv_compiler_emit_store_reg(compiler, &dst->reg, dst->write_mask, result_id); } @@ -9175,16 +9415,12 @@ static int spirv_compiler_handle_instruction(struct spirv_compiler *compiler, case VKD3DSIH_DCL_TGSM_STRUCTURED: spirv_compiler_emit_dcl_tgsm_structured(compiler, instruction); break; + case VKD3DSIH_DCL_INPUT_PS: case VKD3DSIH_DCL_INPUT: spirv_compiler_emit_dcl_input(compiler, instruction); break; - case VKD3DSIH_DCL_INPUT_PS: - spirv_compiler_emit_dcl_input_ps(compiler, instruction); - break; case VKD3DSIH_DCL_INPUT_PS_SGV: case VKD3DSIH_DCL_INPUT_PS_SIV: - spirv_compiler_emit_dcl_input_ps_sysval(compiler, instruction); - break; case VKD3DSIH_DCL_INPUT_SGV: case VKD3DSIH_DCL_INPUT_SIV: spirv_compiler_emit_dcl_input_sysval(compiler, instruction); @@ -9210,11 +9446,7 @@ static int spirv_compiler_handle_instruction(struct spirv_compiler *compiler, case VKD3DSIH_DCL_GS_INSTANCES: spirv_compiler_emit_dcl_gs_instances(compiler, instruction); break; - case VKD3DSIH_DCL_INPUT_CONTROL_POINT_COUNT: - compiler->input_control_point_count = instruction->declaration.count; - break; case VKD3DSIH_DCL_OUTPUT_CONTROL_POINT_COUNT: - compiler->output_control_point_count = instruction->declaration.count; spirv_compiler_emit_output_vertex_count(compiler, instruction); break; case VKD3DSIH_DCL_TESSELLATOR_DOMAIN: @@ -9256,25 +9488,24 @@ static int spirv_compiler_handle_instruction(struct spirv_compiler *compiler, case VKD3DSIH_DIV: case VKD3DSIH_DMUL: case VKD3DSIH_DTOF: - case VKD3DSIH_DTOI: - case VKD3DSIH_DTOU: + case VKD3DSIH_FREM: case VKD3DSIH_FTOD: - case VKD3DSIH_FTOI: - case VKD3DSIH_FTOU: case VKD3DSIH_IADD: case VKD3DSIH_INEG: case VKD3DSIH_ISHL: case VKD3DSIH_ISHR: case VKD3DSIH_ITOD: case VKD3DSIH_ITOF: + case VKD3DSIH_ITOI: case VKD3DSIH_MUL: case VKD3DSIH_NOT: case VKD3DSIH_OR: case VKD3DSIH_USHR: case VKD3DSIH_UTOD: case VKD3DSIH_UTOF: + case VKD3DSIH_UTOU: case VKD3DSIH_XOR: - spirv_compiler_emit_alu_instruction(compiler, instruction); + ret = spirv_compiler_emit_alu_instruction(compiler, instruction); break; case VKD3DSIH_DFMA: case VKD3DSIH_DMAX: @@ -9313,26 +9544,40 @@ static int spirv_compiler_handle_instruction(struct spirv_compiler *compiler, spirv_compiler_emit_sincos(compiler, instruction); break; case VKD3DSIH_IMUL: + case VKD3DSIH_UMUL: spirv_compiler_emit_imul(compiler, instruction); break; case VKD3DSIH_IMAD: spirv_compiler_emit_imad(compiler, instruction); break; + case VKD3DSIH_IDIV: case VKD3DSIH_UDIV: - spirv_compiler_emit_udiv(compiler, instruction); + spirv_compiler_emit_int_div(compiler, instruction); break; - case VKD3DSIH_DEQ: - case VKD3DSIH_DGE: + case VKD3DSIH_DTOI: + case VKD3DSIH_FTOI: + spirv_compiler_emit_ftoi(compiler, instruction); + break; + case VKD3DSIH_DTOU: + case VKD3DSIH_FTOU: + spirv_compiler_emit_ftou(compiler, instruction); + break; + case VKD3DSIH_DEQO: + case VKD3DSIH_DGEO: case VKD3DSIH_DLT: case VKD3DSIH_DNE: - case VKD3DSIH_EQ: - case VKD3DSIH_GE: + case VKD3DSIH_EQO: + case VKD3DSIH_EQU: + case VKD3DSIH_GEO: + case VKD3DSIH_GEU: case VKD3DSIH_IEQ: case VKD3DSIH_IGE: case VKD3DSIH_ILT: case VKD3DSIH_INE: - case VKD3DSIH_LT: - case VKD3DSIH_NE: + case VKD3DSIH_LTO: + case VKD3DSIH_LTU: + case VKD3DSIH_NEO: + case VKD3DSIH_NEU: case VKD3DSIH_UGE: case VKD3DSIH_ULT: spirv_compiler_emit_comparison_instruction(compiler, instruction); @@ -9467,6 +9712,7 @@ static int spirv_compiler_handle_instruction(struct spirv_compiler *compiler, case VKD3DSIH_DCL: case VKD3DSIH_DCL_CONSTANT_BUFFER: case VKD3DSIH_DCL_HS_MAX_TESSFACTOR: + case VKD3DSIH_DCL_INPUT_CONTROL_POINT_COUNT: case VKD3DSIH_DCL_RESOURCE_RAW: case VKD3DSIH_DCL_RESOURCE_STRUCTURED: case VKD3DSIH_DCL_SAMPLER: @@ -9480,6 +9726,9 @@ static int spirv_compiler_handle_instruction(struct spirv_compiler *compiler, break; default: FIXME("Unhandled instruction %#x.\n", instruction->handler_idx); + spirv_compiler_error(compiler, VKD3D_SHADER_ERROR_SPV_INVALID_HANDLER, + "Encountered invalid/unhandled instruction handler %#x.", instruction->handler_idx); + break; } return ret; @@ -9543,6 +9792,8 @@ static int spirv_compiler_generate_spirv(struct spirv_compiler *compiler, if (parser->shader_desc.temp_count) spirv_compiler_emit_temps(compiler, parser->shader_desc.temp_count); + if (parser->shader_desc.ssa_count) + spirv_compiler_allocate_ssa_register_ids(compiler, parser->shader_desc.ssa_count); spirv_compiler_emit_descriptor_declarations(compiler); @@ -9561,6 +9812,10 @@ static int spirv_compiler_generate_spirv(struct spirv_compiler *compiler, memset(&shader_desc->input_signature, 0, sizeof(shader_desc->input_signature)); memset(&shader_desc->output_signature, 0, sizeof(shader_desc->output_signature)); memset(&shader_desc->patch_constant_signature, 0, sizeof(shader_desc->patch_constant_signature)); + compiler->use_vocp = parser->shader_desc.use_vocp; + + compiler->input_control_point_count = shader_desc->input_control_point_count; + compiler->output_control_point_count = shader_desc->output_control_point_count; if (compiler->shader_type != VKD3D_SHADER_TYPE_HULL) spirv_compiler_emit_shader_signature_outputs(compiler); diff --git a/libs/vkd3d/libs/vkd3d-shader/tpf.c b/libs/vkd3d/libs/vkd3d-shader/tpf.c index 045fb6c5f64..61d14e30d3c 100644 --- a/libs/vkd3d/libs/vkd3d-shader/tpf.c +++ b/libs/vkd3d/libs/vkd3d-shader/tpf.c @@ -146,6 +146,9 @@ STATIC_ASSERT(SM4_MAX_SRC_COUNT <= SPIRV_MAX_SRC_COUNT); #define VKD3D_SM4_SWIZZLE_SHIFT 4 #define VKD3D_SM4_SWIZZLE_MASK (0xffu << VKD3D_SM4_SWIZZLE_SHIFT) +#define VKD3D_SM4_SCALAR_DIM_SHIFT 4 +#define VKD3D_SM4_SCALAR_DIM_MASK (0x3u << VKD3D_SM4_SCALAR_DIM_SHIFT) + #define VKD3D_SM4_VERSION_MAJOR(version) (((version) >> 4) & 0xf) #define VKD3D_SM4_VERSION_MINOR(version) (((version) >> 0) & 0xf) @@ -383,6 +386,8 @@ enum vkd3d_sm4_opcode VKD3D_SM5_OP_SAMPLE_GRAD_CL_S = 0xe8, VKD3D_SM5_OP_SAMPLE_C_CL_S = 0xe9, VKD3D_SM5_OP_CHECK_ACCESS_FULLY_MAPPED = 0xea, + + VKD3D_SM4_OP_COUNT, }; enum vkd3d_sm4_instruction_modifier @@ -510,6 +515,9 @@ enum vkd3d_sm4_swizzle_type VKD3D_SM4_SWIZZLE_NONE = 0x0, /* swizzle bitfield contains a mask */ VKD3D_SM4_SWIZZLE_VEC4 = 0x1, VKD3D_SM4_SWIZZLE_SCALAR = 0x2, + + VKD3D_SM4_SWIZZLE_DEFAULT = ~0u - 1, + VKD3D_SM4_SWIZZLE_INVALID = ~0u, }; enum vkd3d_sm4_dimension @@ -519,6 +527,38 @@ enum vkd3d_sm4_dimension VKD3D_SM4_DIMENSION_VEC4 = 0x2, }; +static enum vsir_dimension vsir_dimension_from_sm4_dimension(enum vkd3d_sm4_dimension dim) +{ + switch (dim) + { + case VKD3D_SM4_DIMENSION_NONE: + return VSIR_DIMENSION_NONE; + case VKD3D_SM4_DIMENSION_SCALAR: + return VSIR_DIMENSION_SCALAR; + case VKD3D_SM4_DIMENSION_VEC4: + return VSIR_DIMENSION_VEC4; + default: + FIXME("Unknown SM4 dimension %#x.\n", dim); + return VSIR_DIMENSION_NONE; + } +} + +static enum vkd3d_sm4_dimension sm4_dimension_from_vsir_dimension(enum vsir_dimension dim) +{ + switch (dim) + { + case VSIR_DIMENSION_NONE: + return VKD3D_SM4_DIMENSION_NONE; + case VSIR_DIMENSION_SCALAR: + return VKD3D_SM4_DIMENSION_SCALAR; + case VSIR_DIMENSION_VEC4: + return VKD3D_SM4_DIMENSION_VEC4; + case VSIR_DIMENSION_COUNT: + vkd3d_unreachable(); + } + vkd3d_unreachable(); +} + enum vkd3d_sm4_resource_type { VKD3D_SM4_RESOURCE_BUFFER = 0x1, @@ -575,6 +615,7 @@ struct sm4_index_range_array struct vkd3d_sm4_lookup_tables { + const struct vkd3d_sm4_opcode_info *opcode_info_from_sm4[VKD3D_SM4_OP_COUNT]; const struct vkd3d_sm4_register_type_info *register_type_info_from_sm4[VKD3D_SM4_REGISTER_TYPE_COUNT]; const struct vkd3d_sm4_register_type_info *register_type_info_from_vkd3d[VKD3DSPR_COUNT]; }; @@ -583,8 +624,6 @@ struct vkd3d_shader_sm4_parser { const uint32_t *start, *end, *ptr; - unsigned int output_map[MAX_REG_OUTPUT]; - enum vkd3d_shader_opcode phase; bool has_control_point_phase; unsigned int input_register_masks[MAX_REG_OUTPUT]; @@ -620,16 +659,18 @@ static const enum vkd3d_primitive_type output_primitive_type_table[] = /* VKD3D_SM4_OUTPUT_PT_TRIANGLESTRIP */ VKD3D_PT_TRIANGLESTRIP, }; -static const enum vkd3d_primitive_type input_primitive_type_table[] = +static const struct { - /* UNKNOWN */ VKD3D_PT_UNDEFINED, - /* VKD3D_SM4_INPUT_PT_POINT */ VKD3D_PT_POINTLIST, - /* VKD3D_SM4_INPUT_PT_LINE */ VKD3D_PT_LINELIST, - /* VKD3D_SM4_INPUT_PT_TRIANGLE */ VKD3D_PT_TRIANGLELIST, - /* UNKNOWN */ VKD3D_PT_UNDEFINED, - /* UNKNOWN */ VKD3D_PT_UNDEFINED, - /* VKD3D_SM4_INPUT_PT_LINEADJ */ VKD3D_PT_LINELIST_ADJ, - /* VKD3D_SM4_INPUT_PT_TRIANGLEADJ */ VKD3D_PT_TRIANGLELIST_ADJ, + unsigned int control_point_count; + enum vkd3d_primitive_type vkd3d_type; +} +input_primitive_type_table[] = +{ + [VKD3D_SM4_INPUT_PT_POINT] = {1, VKD3D_PT_POINTLIST}, + [VKD3D_SM4_INPUT_PT_LINE] = {2, VKD3D_PT_LINELIST}, + [VKD3D_SM4_INPUT_PT_TRIANGLE] = {3, VKD3D_PT_TRIANGLELIST}, + [VKD3D_SM4_INPUT_PT_LINEADJ] = {4, VKD3D_PT_LINELIST_ADJ}, + [VKD3D_SM4_INPUT_PT_TRIANGLEADJ] = {6, VKD3D_PT_TRIANGLELIST_ADJ}, }; static const enum vkd3d_shader_resource_type resource_type_table[] = @@ -751,7 +792,9 @@ static void shader_sm4_read_shader_data(struct vkd3d_shader_instruction *ins, ui ins->handler_idx = VKD3DSIH_INVALID; return; } - icb->vec4_count = icb_size / 4; + icb->data_type = VKD3D_DATA_FLOAT; + icb->component_count = VKD3D_VEC4_SIZE; + icb->element_count = icb_size / VKD3D_VEC4_SIZE; memcpy(icb->data, tokens, sizeof(*tokens) * icb_size); shader_instruction_array_add_icb(&priv->p.instructions, icb); ins->declaration.icb = icb; @@ -849,6 +892,8 @@ static void shader_sm4_read_dcl_constant_buffer(struct vkd3d_shader_instruction ins->declaration.cb.size = *tokens++; shader_sm4_read_register_space(priv, &tokens, end, &ins->declaration.cb.range.space); } + + ins->declaration.cb.size *= VKD3D_VEC4_SIZE * sizeof(float); } static void shader_sm4_read_dcl_sampler(struct vkd3d_shader_instruction *ins, uint32_t opcode, uint32_t opcode_token, @@ -994,6 +1039,7 @@ static void shader_sm4_read_dcl_input_primitive(struct vkd3d_shader_instruction { ins->declaration.primitive_type.type = VKD3D_PT_PATCH; ins->declaration.primitive_type.patch_vertex_count = primitive_type - VKD3D_SM5_INPUT_PT_PATCH1 + 1; + priv->p.shader_desc.input_control_point_count = ins->declaration.primitive_type.patch_vertex_count; } else if (primitive_type >= ARRAY_SIZE(input_primitive_type_table)) { @@ -1001,7 +1047,8 @@ static void shader_sm4_read_dcl_input_primitive(struct vkd3d_shader_instruction } else { - ins->declaration.primitive_type.type = input_primitive_type_table[primitive_type]; + ins->declaration.primitive_type.type = input_primitive_type_table[primitive_type].vkd3d_type; + priv->p.shader_desc.input_control_point_count = input_primitive_type_table[primitive_type].control_point_count; } if (ins->declaration.primitive_type.type == VKD3D_PT_UNDEFINED) @@ -1033,16 +1080,31 @@ static void shader_sm4_read_declaration_register_semantic(struct vkd3d_shader_in static void shader_sm4_read_dcl_input_ps(struct vkd3d_shader_instruction *ins, uint32_t opcode, uint32_t opcode_token, const uint32_t *tokens, unsigned int token_count, struct vkd3d_shader_sm4_parser *priv) { + struct vkd3d_shader_dst_param *dst = &ins->declaration.dst; + ins->flags = (opcode_token & VKD3D_SM4_INTERPOLATION_MODE_MASK) >> VKD3D_SM4_INTERPOLATION_MODE_SHIFT; - shader_sm4_read_dst_param(priv, &tokens, &tokens[token_count], VKD3D_DATA_FLOAT, &ins->declaration.dst); + if (shader_sm4_read_dst_param(priv, &tokens, &tokens[token_count], VKD3D_DATA_FLOAT, dst)) + { + struct signature_element *e = vsir_signature_find_element_for_reg( + &priv->p.shader_desc.input_signature, dst->reg.idx[dst->reg.idx_count - 1].offset, dst->write_mask); + + e->interpolation_mode = ins->flags; + } } static void shader_sm4_read_dcl_input_ps_siv(struct vkd3d_shader_instruction *ins, uint32_t opcode, uint32_t opcode_token, const uint32_t *tokens, unsigned int token_count, struct vkd3d_shader_sm4_parser *priv) { + struct vkd3d_shader_dst_param *dst = &ins->declaration.register_semantic.reg; + ins->flags = (opcode_token & VKD3D_SM4_INTERPOLATION_MODE_MASK) >> VKD3D_SM4_INTERPOLATION_MODE_SHIFT; - shader_sm4_read_dst_param(priv, &tokens, &tokens[token_count], VKD3D_DATA_FLOAT, - &ins->declaration.register_semantic.reg); + if (shader_sm4_read_dst_param(priv, &tokens, &tokens[token_count], VKD3D_DATA_FLOAT, dst)) + { + struct signature_element *e = vsir_signature_find_element_for_reg( + &priv->p.shader_desc.input_signature, dst->reg.idx[dst->reg.idx_count - 1].offset, dst->write_mask); + + e->interpolation_mode = ins->flags; + } ins->declaration.register_semantic.sysval_semantic = *tokens; } @@ -1051,13 +1113,15 @@ static void shader_sm4_read_dcl_indexable_temp(struct vkd3d_shader_instruction * { ins->declaration.indexable_temp.register_idx = *tokens++; ins->declaration.indexable_temp.register_size = *tokens++; + ins->declaration.indexable_temp.alignment = 0; + ins->declaration.indexable_temp.data_type = VKD3D_DATA_FLOAT; ins->declaration.indexable_temp.component_count = *tokens; } static void shader_sm4_read_dcl_global_flags(struct vkd3d_shader_instruction *ins, uint32_t opcode, uint32_t opcode_token, const uint32_t *tokens, unsigned int token_count, struct vkd3d_shader_sm4_parser *priv) { - ins->flags = (opcode_token & VKD3D_SM4_GLOBAL_FLAGS_MASK) >> VKD3D_SM4_GLOBAL_FLAGS_SHIFT; + ins->declaration.global_flags = (opcode_token & VKD3D_SM4_GLOBAL_FLAGS_MASK) >> VKD3D_SM4_GLOBAL_FLAGS_SHIFT; } static void shader_sm5_read_fcall(struct vkd3d_shader_instruction *ins, uint32_t opcode, uint32_t opcode_token, @@ -1096,6 +1160,11 @@ static void shader_sm5_read_control_point_count(struct vkd3d_shader_instruction { ins->declaration.count = (opcode_token & VKD3D_SM5_CONTROL_POINT_COUNT_MASK) >> VKD3D_SM5_CONTROL_POINT_COUNT_SHIFT; + + if (opcode == VKD3D_SM5_OP_DCL_INPUT_CONTROL_POINT_COUNT) + priv->p.shader_desc.input_control_point_count = ins->declaration.count; + else + priv->p.shader_desc.output_control_point_count = ins->declaration.count; } static void shader_sm5_read_dcl_tessellator_domain(struct vkd3d_shader_instruction *ins, uint32_t opcode, @@ -1211,291 +1280,13 @@ static void shader_sm5_read_sync(struct vkd3d_shader_instruction *ins, uint32_t ins->flags = (opcode_token & VKD3D_SM5_SYNC_FLAGS_MASK) >> VKD3D_SM5_SYNC_FLAGS_SHIFT; } -/* - * d -> VKD3D_DATA_DOUBLE - * f -> VKD3D_DATA_FLOAT - * i -> VKD3D_DATA_INT - * u -> VKD3D_DATA_UINT - * O -> VKD3D_DATA_OPAQUE - * R -> VKD3D_DATA_RESOURCE - * S -> VKD3D_DATA_SAMPLER - * U -> VKD3D_DATA_UAV - */ -static const struct vkd3d_sm4_opcode_info opcode_table[] = -{ - {VKD3D_SM4_OP_ADD, VKD3DSIH_ADD, "f", "ff"}, - {VKD3D_SM4_OP_AND, VKD3DSIH_AND, "u", "uu"}, - {VKD3D_SM4_OP_BREAK, VKD3DSIH_BREAK, "", ""}, - {VKD3D_SM4_OP_BREAKC, VKD3DSIH_BREAKP, "", "u", - shader_sm4_read_conditional_op}, - {VKD3D_SM4_OP_CASE, VKD3DSIH_CASE, "", "u", - shader_sm4_read_case_condition}, - {VKD3D_SM4_OP_CONTINUE, VKD3DSIH_CONTINUE, "", ""}, - {VKD3D_SM4_OP_CONTINUEC, VKD3DSIH_CONTINUEP, "", "u", - shader_sm4_read_conditional_op}, - {VKD3D_SM4_OP_CUT, VKD3DSIH_CUT, "", ""}, - {VKD3D_SM4_OP_DEFAULT, VKD3DSIH_DEFAULT, "", ""}, - {VKD3D_SM4_OP_DERIV_RTX, VKD3DSIH_DSX, "f", "f"}, - {VKD3D_SM4_OP_DERIV_RTY, VKD3DSIH_DSY, "f", "f"}, - {VKD3D_SM4_OP_DISCARD, VKD3DSIH_DISCARD, "", "u", - shader_sm4_read_conditional_op}, - {VKD3D_SM4_OP_DIV, VKD3DSIH_DIV, "f", "ff"}, - {VKD3D_SM4_OP_DP2, VKD3DSIH_DP2, "f", "ff"}, - {VKD3D_SM4_OP_DP3, VKD3DSIH_DP3, "f", "ff"}, - {VKD3D_SM4_OP_DP4, VKD3DSIH_DP4, "f", "ff"}, - {VKD3D_SM4_OP_ELSE, VKD3DSIH_ELSE, "", ""}, - {VKD3D_SM4_OP_EMIT, VKD3DSIH_EMIT, "", ""}, - {VKD3D_SM4_OP_ENDIF, VKD3DSIH_ENDIF, "", ""}, - {VKD3D_SM4_OP_ENDLOOP, VKD3DSIH_ENDLOOP, "", ""}, - {VKD3D_SM4_OP_ENDSWITCH, VKD3DSIH_ENDSWITCH, "", ""}, - {VKD3D_SM4_OP_EQ, VKD3DSIH_EQ, "u", "ff"}, - {VKD3D_SM4_OP_EXP, VKD3DSIH_EXP, "f", "f"}, - {VKD3D_SM4_OP_FRC, VKD3DSIH_FRC, "f", "f"}, - {VKD3D_SM4_OP_FTOI, VKD3DSIH_FTOI, "i", "f"}, - {VKD3D_SM4_OP_FTOU, VKD3DSIH_FTOU, "u", "f"}, - {VKD3D_SM4_OP_GE, VKD3DSIH_GE, "u", "ff"}, - {VKD3D_SM4_OP_IADD, VKD3DSIH_IADD, "i", "ii"}, - {VKD3D_SM4_OP_IF, VKD3DSIH_IF, "", "u", - shader_sm4_read_conditional_op}, - {VKD3D_SM4_OP_IEQ, VKD3DSIH_IEQ, "u", "ii"}, - {VKD3D_SM4_OP_IGE, VKD3DSIH_IGE, "u", "ii"}, - {VKD3D_SM4_OP_ILT, VKD3DSIH_ILT, "u", "ii"}, - {VKD3D_SM4_OP_IMAD, VKD3DSIH_IMAD, "i", "iii"}, - {VKD3D_SM4_OP_IMAX, VKD3DSIH_IMAX, "i", "ii"}, - {VKD3D_SM4_OP_IMIN, VKD3DSIH_IMIN, "i", "ii"}, - {VKD3D_SM4_OP_IMUL, VKD3DSIH_IMUL, "ii", "ii"}, - {VKD3D_SM4_OP_INE, VKD3DSIH_INE, "u", "ii"}, - {VKD3D_SM4_OP_INEG, VKD3DSIH_INEG, "i", "i"}, - {VKD3D_SM4_OP_ISHL, VKD3DSIH_ISHL, "i", "ii"}, - {VKD3D_SM4_OP_ISHR, VKD3DSIH_ISHR, "i", "ii"}, - {VKD3D_SM4_OP_ITOF, VKD3DSIH_ITOF, "f", "i"}, - {VKD3D_SM4_OP_LABEL, VKD3DSIH_LABEL, "", "O"}, - {VKD3D_SM4_OP_LD, VKD3DSIH_LD, "u", "iR"}, - {VKD3D_SM4_OP_LD2DMS, VKD3DSIH_LD2DMS, "u", "iRi"}, - {VKD3D_SM4_OP_LOG, VKD3DSIH_LOG, "f", "f"}, - {VKD3D_SM4_OP_LOOP, VKD3DSIH_LOOP, "", ""}, - {VKD3D_SM4_OP_LT, VKD3DSIH_LT, "u", "ff"}, - {VKD3D_SM4_OP_MAD, VKD3DSIH_MAD, "f", "fff"}, - {VKD3D_SM4_OP_MIN, VKD3DSIH_MIN, "f", "ff"}, - {VKD3D_SM4_OP_MAX, VKD3DSIH_MAX, "f", "ff"}, - {VKD3D_SM4_OP_SHADER_DATA, VKD3DSIH_DCL_IMMEDIATE_CONSTANT_BUFFER, "", "", - shader_sm4_read_shader_data}, - {VKD3D_SM4_OP_MOV, VKD3DSIH_MOV, "f", "f"}, - {VKD3D_SM4_OP_MOVC, VKD3DSIH_MOVC, "f", "uff"}, - {VKD3D_SM4_OP_MUL, VKD3DSIH_MUL, "f", "ff"}, - {VKD3D_SM4_OP_NE, VKD3DSIH_NE, "u", "ff"}, - {VKD3D_SM4_OP_NOP, VKD3DSIH_NOP, "", ""}, - {VKD3D_SM4_OP_NOT, VKD3DSIH_NOT, "u", "u"}, - {VKD3D_SM4_OP_OR, VKD3DSIH_OR, "u", "uu"}, - {VKD3D_SM4_OP_RESINFO, VKD3DSIH_RESINFO, "f", "iR"}, - {VKD3D_SM4_OP_RET, VKD3DSIH_RET, "", ""}, - {VKD3D_SM4_OP_RETC, VKD3DSIH_RETP, "", "u", - shader_sm4_read_conditional_op}, - {VKD3D_SM4_OP_ROUND_NE, VKD3DSIH_ROUND_NE, "f", "f"}, - {VKD3D_SM4_OP_ROUND_NI, VKD3DSIH_ROUND_NI, "f", "f"}, - {VKD3D_SM4_OP_ROUND_PI, VKD3DSIH_ROUND_PI, "f", "f"}, - {VKD3D_SM4_OP_ROUND_Z, VKD3DSIH_ROUND_Z, "f", "f"}, - {VKD3D_SM4_OP_RSQ, VKD3DSIH_RSQ, "f", "f"}, - {VKD3D_SM4_OP_SAMPLE, VKD3DSIH_SAMPLE, "u", "fRS"}, - {VKD3D_SM4_OP_SAMPLE_C, VKD3DSIH_SAMPLE_C, "f", "fRSf"}, - {VKD3D_SM4_OP_SAMPLE_C_LZ, VKD3DSIH_SAMPLE_C_LZ, "f", "fRSf"}, - {VKD3D_SM4_OP_SAMPLE_LOD, VKD3DSIH_SAMPLE_LOD, "u", "fRSf"}, - {VKD3D_SM4_OP_SAMPLE_GRAD, VKD3DSIH_SAMPLE_GRAD, "u", "fRSff"}, - {VKD3D_SM4_OP_SAMPLE_B, VKD3DSIH_SAMPLE_B, "u", "fRSf"}, - {VKD3D_SM4_OP_SQRT, VKD3DSIH_SQRT, "f", "f"}, - {VKD3D_SM4_OP_SWITCH, VKD3DSIH_SWITCH, "", "i"}, - {VKD3D_SM4_OP_SINCOS, VKD3DSIH_SINCOS, "ff", "f"}, - {VKD3D_SM4_OP_UDIV, VKD3DSIH_UDIV, "uu", "uu"}, - {VKD3D_SM4_OP_ULT, VKD3DSIH_ULT, "u", "uu"}, - {VKD3D_SM4_OP_UGE, VKD3DSIH_UGE, "u", "uu"}, - {VKD3D_SM4_OP_UMUL, VKD3DSIH_UMUL, "uu", "uu"}, - {VKD3D_SM4_OP_UMAX, VKD3DSIH_UMAX, "u", "uu"}, - {VKD3D_SM4_OP_UMIN, VKD3DSIH_UMIN, "u", "uu"}, - {VKD3D_SM4_OP_USHR, VKD3DSIH_USHR, "u", "uu"}, - {VKD3D_SM4_OP_UTOF, VKD3DSIH_UTOF, "f", "u"}, - {VKD3D_SM4_OP_XOR, VKD3DSIH_XOR, "u", "uu"}, - {VKD3D_SM4_OP_DCL_RESOURCE, VKD3DSIH_DCL, "", "", - shader_sm4_read_dcl_resource}, - {VKD3D_SM4_OP_DCL_CONSTANT_BUFFER, VKD3DSIH_DCL_CONSTANT_BUFFER, "", "", - shader_sm4_read_dcl_constant_buffer}, - {VKD3D_SM4_OP_DCL_SAMPLER, VKD3DSIH_DCL_SAMPLER, "", "", - shader_sm4_read_dcl_sampler}, - {VKD3D_SM4_OP_DCL_INDEX_RANGE, VKD3DSIH_DCL_INDEX_RANGE, "", "", - shader_sm4_read_dcl_index_range}, - {VKD3D_SM4_OP_DCL_OUTPUT_TOPOLOGY, VKD3DSIH_DCL_OUTPUT_TOPOLOGY, "", "", - shader_sm4_read_dcl_output_topology}, - {VKD3D_SM4_OP_DCL_INPUT_PRIMITIVE, VKD3DSIH_DCL_INPUT_PRIMITIVE, "", "", - shader_sm4_read_dcl_input_primitive}, - {VKD3D_SM4_OP_DCL_VERTICES_OUT, VKD3DSIH_DCL_VERTICES_OUT, "", "", - shader_sm4_read_declaration_count}, - {VKD3D_SM4_OP_DCL_INPUT, VKD3DSIH_DCL_INPUT, "", "", - shader_sm4_read_declaration_dst}, - {VKD3D_SM4_OP_DCL_INPUT_SGV, VKD3DSIH_DCL_INPUT_SGV, "", "", - shader_sm4_read_declaration_register_semantic}, - {VKD3D_SM4_OP_DCL_INPUT_SIV, VKD3DSIH_DCL_INPUT_SIV, "", "", - shader_sm4_read_declaration_register_semantic}, - {VKD3D_SM4_OP_DCL_INPUT_PS, VKD3DSIH_DCL_INPUT_PS, "", "", - shader_sm4_read_dcl_input_ps}, - {VKD3D_SM4_OP_DCL_INPUT_PS_SGV, VKD3DSIH_DCL_INPUT_PS_SGV, "", "", - shader_sm4_read_declaration_register_semantic}, - {VKD3D_SM4_OP_DCL_INPUT_PS_SIV, VKD3DSIH_DCL_INPUT_PS_SIV, "", "", - shader_sm4_read_dcl_input_ps_siv}, - {VKD3D_SM4_OP_DCL_OUTPUT, VKD3DSIH_DCL_OUTPUT, "", "", - shader_sm4_read_declaration_dst}, - {VKD3D_SM4_OP_DCL_OUTPUT_SIV, VKD3DSIH_DCL_OUTPUT_SIV, "", "", - shader_sm4_read_declaration_register_semantic}, - {VKD3D_SM4_OP_DCL_TEMPS, VKD3DSIH_DCL_TEMPS, "", "", - shader_sm4_read_declaration_count}, - {VKD3D_SM4_OP_DCL_INDEXABLE_TEMP, VKD3DSIH_DCL_INDEXABLE_TEMP, "", "", - shader_sm4_read_dcl_indexable_temp}, - {VKD3D_SM4_OP_DCL_GLOBAL_FLAGS, VKD3DSIH_DCL_GLOBAL_FLAGS, "", "", - shader_sm4_read_dcl_global_flags}, - {VKD3D_SM4_OP_LOD, VKD3DSIH_LOD, "f", "fRS"}, - {VKD3D_SM4_OP_GATHER4, VKD3DSIH_GATHER4, "u", "fRS"}, - {VKD3D_SM4_OP_SAMPLE_POS, VKD3DSIH_SAMPLE_POS, "f", "Ru"}, - {VKD3D_SM4_OP_SAMPLE_INFO, VKD3DSIH_SAMPLE_INFO, "f", "R"}, - {VKD3D_SM5_OP_HS_DECLS, VKD3DSIH_HS_DECLS, "", ""}, - {VKD3D_SM5_OP_HS_CONTROL_POINT_PHASE, VKD3DSIH_HS_CONTROL_POINT_PHASE, "", ""}, - {VKD3D_SM5_OP_HS_FORK_PHASE, VKD3DSIH_HS_FORK_PHASE, "", ""}, - {VKD3D_SM5_OP_HS_JOIN_PHASE, VKD3DSIH_HS_JOIN_PHASE, "", ""}, - {VKD3D_SM5_OP_EMIT_STREAM, VKD3DSIH_EMIT_STREAM, "", "f"}, - {VKD3D_SM5_OP_CUT_STREAM, VKD3DSIH_CUT_STREAM, "", "f"}, - {VKD3D_SM5_OP_FCALL, VKD3DSIH_FCALL, "", "O", - shader_sm5_read_fcall}, - {VKD3D_SM5_OP_BUFINFO, VKD3DSIH_BUFINFO, "i", "U"}, - {VKD3D_SM5_OP_DERIV_RTX_COARSE, VKD3DSIH_DSX_COARSE, "f", "f"}, - {VKD3D_SM5_OP_DERIV_RTX_FINE, VKD3DSIH_DSX_FINE, "f", "f"}, - {VKD3D_SM5_OP_DERIV_RTY_COARSE, VKD3DSIH_DSY_COARSE, "f", "f"}, - {VKD3D_SM5_OP_DERIV_RTY_FINE, VKD3DSIH_DSY_FINE, "f", "f"}, - {VKD3D_SM5_OP_GATHER4_C, VKD3DSIH_GATHER4_C, "f", "fRSf"}, - {VKD3D_SM5_OP_GATHER4_PO, VKD3DSIH_GATHER4_PO, "f", "fiRS"}, - {VKD3D_SM5_OP_GATHER4_PO_C, VKD3DSIH_GATHER4_PO_C, "f", "fiRSf"}, - {VKD3D_SM5_OP_RCP, VKD3DSIH_RCP, "f", "f"}, - {VKD3D_SM5_OP_F32TOF16, VKD3DSIH_F32TOF16, "u", "f"}, - {VKD3D_SM5_OP_F16TOF32, VKD3DSIH_F16TOF32, "f", "u"}, - {VKD3D_SM5_OP_COUNTBITS, VKD3DSIH_COUNTBITS, "u", "u"}, - {VKD3D_SM5_OP_FIRSTBIT_HI, VKD3DSIH_FIRSTBIT_HI, "u", "u"}, - {VKD3D_SM5_OP_FIRSTBIT_LO, VKD3DSIH_FIRSTBIT_LO, "u", "u"}, - {VKD3D_SM5_OP_FIRSTBIT_SHI, VKD3DSIH_FIRSTBIT_SHI, "u", "i"}, - {VKD3D_SM5_OP_UBFE, VKD3DSIH_UBFE, "u", "iiu"}, - {VKD3D_SM5_OP_IBFE, VKD3DSIH_IBFE, "i", "iii"}, - {VKD3D_SM5_OP_BFI, VKD3DSIH_BFI, "u", "iiuu"}, - {VKD3D_SM5_OP_BFREV, VKD3DSIH_BFREV, "u", "u"}, - {VKD3D_SM5_OP_SWAPC, VKD3DSIH_SWAPC, "ff", "uff"}, - {VKD3D_SM5_OP_DCL_STREAM, VKD3DSIH_DCL_STREAM, "", "O"}, - {VKD3D_SM5_OP_DCL_FUNCTION_BODY, VKD3DSIH_DCL_FUNCTION_BODY, "", "", - shader_sm5_read_dcl_function_body}, - {VKD3D_SM5_OP_DCL_FUNCTION_TABLE, VKD3DSIH_DCL_FUNCTION_TABLE, "", "", - shader_sm5_read_dcl_function_table}, - {VKD3D_SM5_OP_DCL_INTERFACE, VKD3DSIH_DCL_INTERFACE, "", "", - shader_sm5_read_dcl_interface}, - {VKD3D_SM5_OP_DCL_INPUT_CONTROL_POINT_COUNT, VKD3DSIH_DCL_INPUT_CONTROL_POINT_COUNT, "", "", - shader_sm5_read_control_point_count}, - {VKD3D_SM5_OP_DCL_OUTPUT_CONTROL_POINT_COUNT, VKD3DSIH_DCL_OUTPUT_CONTROL_POINT_COUNT, "", "", - shader_sm5_read_control_point_count}, - {VKD3D_SM5_OP_DCL_TESSELLATOR_DOMAIN, VKD3DSIH_DCL_TESSELLATOR_DOMAIN, "", "", - shader_sm5_read_dcl_tessellator_domain}, - {VKD3D_SM5_OP_DCL_TESSELLATOR_PARTITIONING, VKD3DSIH_DCL_TESSELLATOR_PARTITIONING, "", "", - shader_sm5_read_dcl_tessellator_partitioning}, - {VKD3D_SM5_OP_DCL_TESSELLATOR_OUTPUT_PRIMITIVE, VKD3DSIH_DCL_TESSELLATOR_OUTPUT_PRIMITIVE, "", "", - shader_sm5_read_dcl_tessellator_output_primitive}, - {VKD3D_SM5_OP_DCL_HS_MAX_TESSFACTOR, VKD3DSIH_DCL_HS_MAX_TESSFACTOR, "", "", - shader_sm5_read_dcl_hs_max_tessfactor}, - {VKD3D_SM5_OP_DCL_HS_FORK_PHASE_INSTANCE_COUNT, VKD3DSIH_DCL_HS_FORK_PHASE_INSTANCE_COUNT, "", "", - shader_sm4_read_declaration_count}, - {VKD3D_SM5_OP_DCL_HS_JOIN_PHASE_INSTANCE_COUNT, VKD3DSIH_DCL_HS_JOIN_PHASE_INSTANCE_COUNT, "", "", - shader_sm4_read_declaration_count}, - {VKD3D_SM5_OP_DCL_THREAD_GROUP, VKD3DSIH_DCL_THREAD_GROUP, "", "", - shader_sm5_read_dcl_thread_group}, - {VKD3D_SM5_OP_DCL_UAV_TYPED, VKD3DSIH_DCL_UAV_TYPED, "", "", - shader_sm4_read_dcl_resource}, - {VKD3D_SM5_OP_DCL_UAV_RAW, VKD3DSIH_DCL_UAV_RAW, "", "", - shader_sm5_read_dcl_uav_raw}, - {VKD3D_SM5_OP_DCL_UAV_STRUCTURED, VKD3DSIH_DCL_UAV_STRUCTURED, "", "", - shader_sm5_read_dcl_uav_structured}, - {VKD3D_SM5_OP_DCL_TGSM_RAW, VKD3DSIH_DCL_TGSM_RAW, "", "", - shader_sm5_read_dcl_tgsm_raw}, - {VKD3D_SM5_OP_DCL_TGSM_STRUCTURED, VKD3DSIH_DCL_TGSM_STRUCTURED, "", "", - shader_sm5_read_dcl_tgsm_structured}, - {VKD3D_SM5_OP_DCL_RESOURCE_RAW, VKD3DSIH_DCL_RESOURCE_RAW, "", "", - shader_sm5_read_dcl_resource_raw}, - {VKD3D_SM5_OP_DCL_RESOURCE_STRUCTURED, VKD3DSIH_DCL_RESOURCE_STRUCTURED, "", "", - shader_sm5_read_dcl_resource_structured}, - {VKD3D_SM5_OP_LD_UAV_TYPED, VKD3DSIH_LD_UAV_TYPED, "u", "iU"}, - {VKD3D_SM5_OP_STORE_UAV_TYPED, VKD3DSIH_STORE_UAV_TYPED, "U", "iu"}, - {VKD3D_SM5_OP_LD_RAW, VKD3DSIH_LD_RAW, "u", "iU"}, - {VKD3D_SM5_OP_STORE_RAW, VKD3DSIH_STORE_RAW, "U", "uu"}, - {VKD3D_SM5_OP_LD_STRUCTURED, VKD3DSIH_LD_STRUCTURED, "u", "iiR"}, - {VKD3D_SM5_OP_STORE_STRUCTURED, VKD3DSIH_STORE_STRUCTURED, "U", "iiu"}, - {VKD3D_SM5_OP_ATOMIC_AND, VKD3DSIH_ATOMIC_AND, "U", "iu"}, - {VKD3D_SM5_OP_ATOMIC_OR, VKD3DSIH_ATOMIC_OR, "U", "iu"}, - {VKD3D_SM5_OP_ATOMIC_XOR, VKD3DSIH_ATOMIC_XOR, "U", "iu"}, - {VKD3D_SM5_OP_ATOMIC_CMP_STORE, VKD3DSIH_ATOMIC_CMP_STORE, "U", "iuu"}, - {VKD3D_SM5_OP_ATOMIC_IADD, VKD3DSIH_ATOMIC_IADD, "U", "ii"}, - {VKD3D_SM5_OP_ATOMIC_IMAX, VKD3DSIH_ATOMIC_IMAX, "U", "ii"}, - {VKD3D_SM5_OP_ATOMIC_IMIN, VKD3DSIH_ATOMIC_IMIN, "U", "ii"}, - {VKD3D_SM5_OP_ATOMIC_UMAX, VKD3DSIH_ATOMIC_UMAX, "U", "iu"}, - {VKD3D_SM5_OP_ATOMIC_UMIN, VKD3DSIH_ATOMIC_UMIN, "U", "iu"}, - {VKD3D_SM5_OP_IMM_ATOMIC_ALLOC, VKD3DSIH_IMM_ATOMIC_ALLOC, "u", "U"}, - {VKD3D_SM5_OP_IMM_ATOMIC_CONSUME, VKD3DSIH_IMM_ATOMIC_CONSUME, "u", "U"}, - {VKD3D_SM5_OP_IMM_ATOMIC_IADD, VKD3DSIH_IMM_ATOMIC_IADD, "uU", "ii"}, - {VKD3D_SM5_OP_IMM_ATOMIC_AND, VKD3DSIH_IMM_ATOMIC_AND, "uU", "iu"}, - {VKD3D_SM5_OP_IMM_ATOMIC_OR, VKD3DSIH_IMM_ATOMIC_OR, "uU", "iu"}, - {VKD3D_SM5_OP_IMM_ATOMIC_XOR, VKD3DSIH_IMM_ATOMIC_XOR, "uU", "iu"}, - {VKD3D_SM5_OP_IMM_ATOMIC_EXCH, VKD3DSIH_IMM_ATOMIC_EXCH, "uU", "iu"}, - {VKD3D_SM5_OP_IMM_ATOMIC_CMP_EXCH, VKD3DSIH_IMM_ATOMIC_CMP_EXCH, "uU", "iuu"}, - {VKD3D_SM5_OP_IMM_ATOMIC_IMAX, VKD3DSIH_IMM_ATOMIC_IMAX, "iU", "ii"}, - {VKD3D_SM5_OP_IMM_ATOMIC_IMIN, VKD3DSIH_IMM_ATOMIC_IMIN, "iU", "ii"}, - {VKD3D_SM5_OP_IMM_ATOMIC_UMAX, VKD3DSIH_IMM_ATOMIC_UMAX, "uU", "iu"}, - {VKD3D_SM5_OP_IMM_ATOMIC_UMIN, VKD3DSIH_IMM_ATOMIC_UMIN, "uU", "iu"}, - {VKD3D_SM5_OP_SYNC, VKD3DSIH_SYNC, "", "", - shader_sm5_read_sync}, - {VKD3D_SM5_OP_DADD, VKD3DSIH_DADD, "d", "dd"}, - {VKD3D_SM5_OP_DMAX, VKD3DSIH_DMAX, "d", "dd"}, - {VKD3D_SM5_OP_DMIN, VKD3DSIH_DMIN, "d", "dd"}, - {VKD3D_SM5_OP_DMUL, VKD3DSIH_DMUL, "d", "dd"}, - {VKD3D_SM5_OP_DEQ, VKD3DSIH_DEQ, "u", "dd"}, - {VKD3D_SM5_OP_DGE, VKD3DSIH_DGE, "u", "dd"}, - {VKD3D_SM5_OP_DLT, VKD3DSIH_DLT, "u", "dd"}, - {VKD3D_SM5_OP_DNE, VKD3DSIH_DNE, "u", "dd"}, - {VKD3D_SM5_OP_DMOV, VKD3DSIH_DMOV, "d", "d"}, - {VKD3D_SM5_OP_DMOVC, VKD3DSIH_DMOVC, "d", "udd"}, - {VKD3D_SM5_OP_DTOF, VKD3DSIH_DTOF, "f", "d"}, - {VKD3D_SM5_OP_FTOD, VKD3DSIH_FTOD, "d", "f"}, - {VKD3D_SM5_OP_EVAL_SAMPLE_INDEX, VKD3DSIH_EVAL_SAMPLE_INDEX, "f", "fi"}, - {VKD3D_SM5_OP_EVAL_CENTROID, VKD3DSIH_EVAL_CENTROID, "f", "f"}, - {VKD3D_SM5_OP_DCL_GS_INSTANCES, VKD3DSIH_DCL_GS_INSTANCES, "", "", - shader_sm4_read_declaration_count}, - {VKD3D_SM5_OP_DDIV, VKD3DSIH_DDIV, "d", "dd"}, - {VKD3D_SM5_OP_DFMA, VKD3DSIH_DFMA, "d", "ddd"}, - {VKD3D_SM5_OP_DRCP, VKD3DSIH_DRCP, "d", "d"}, - {VKD3D_SM5_OP_MSAD, VKD3DSIH_MSAD, "u", "uuu"}, - {VKD3D_SM5_OP_DTOI, VKD3DSIH_DTOI, "i", "d"}, - {VKD3D_SM5_OP_DTOU, VKD3DSIH_DTOU, "u", "d"}, - {VKD3D_SM5_OP_ITOD, VKD3DSIH_ITOD, "d", "i"}, - {VKD3D_SM5_OP_UTOD, VKD3DSIH_UTOD, "d", "u"}, - {VKD3D_SM5_OP_GATHER4_S, VKD3DSIH_GATHER4_S, "uu", "fRS"}, - {VKD3D_SM5_OP_GATHER4_C_S, VKD3DSIH_GATHER4_C_S, "fu", "fRSf"}, - {VKD3D_SM5_OP_GATHER4_PO_S, VKD3DSIH_GATHER4_PO_S, "fu", "fiRS"}, - {VKD3D_SM5_OP_GATHER4_PO_C_S, VKD3DSIH_GATHER4_PO_C_S, "fu", "fiRSf"}, - {VKD3D_SM5_OP_LD_S, VKD3DSIH_LD_S, "uu", "iR"}, - {VKD3D_SM5_OP_LD2DMS_S, VKD3DSIH_LD2DMS_S, "uu", "iRi"}, - {VKD3D_SM5_OP_LD_UAV_TYPED_S, VKD3DSIH_LD_UAV_TYPED_S, "uu", "iU"}, - {VKD3D_SM5_OP_LD_RAW_S, VKD3DSIH_LD_RAW_S, "uu", "iU"}, - {VKD3D_SM5_OP_LD_STRUCTURED_S, VKD3DSIH_LD_STRUCTURED_S, "uu", "iiR"}, - {VKD3D_SM5_OP_SAMPLE_LOD_S, VKD3DSIH_SAMPLE_LOD_S, "uu", "fRSf"}, - {VKD3D_SM5_OP_SAMPLE_C_LZ_S, VKD3DSIH_SAMPLE_C_LZ_S, "fu", "fRSf"}, - {VKD3D_SM5_OP_SAMPLE_CL_S, VKD3DSIH_SAMPLE_CL_S, "uu", "fRSf"}, - {VKD3D_SM5_OP_SAMPLE_B_CL_S, VKD3DSIH_SAMPLE_B_CL_S, "uu", "fRSff"}, - {VKD3D_SM5_OP_SAMPLE_GRAD_CL_S, VKD3DSIH_SAMPLE_GRAD_CL_S, "uu", "fRSfff"}, - {VKD3D_SM5_OP_SAMPLE_C_CL_S, VKD3DSIH_SAMPLE_C_CL_S, "fu", "fRSff"}, - {VKD3D_SM5_OP_CHECK_ACCESS_FULLY_MAPPED, VKD3DSIH_CHECK_ACCESS_FULLY_MAPPED, "u", "u"}, -}; - struct vkd3d_sm4_register_type_info { enum vkd3d_sm4_register_type sm4_type; enum vkd3d_shader_register_type vkd3d_type; + + /* Swizzle type to be used for src registers when their dimension is VKD3D_SM4_DIMENSION_VEC4. */ + enum vkd3d_sm4_swizzle_type default_src_swizzle_type; }; static const enum vkd3d_shader_register_precision register_precision_table[] = @@ -1515,69 +1306,344 @@ struct tpf_writer struct vkd3d_sm4_lookup_tables lookup; }; -static const struct vkd3d_sm4_opcode_info *get_opcode_info(enum vkd3d_sm4_opcode opcode) -{ - unsigned int i; - - for (i = 0; i < sizeof(opcode_table) / sizeof(*opcode_table); ++i) - { - if (opcode == opcode_table[i].opcode) - return &opcode_table[i]; - } - - return NULL; -} - static void init_sm4_lookup_tables(struct vkd3d_sm4_lookup_tables *lookup) { - const struct vkd3d_sm4_register_type_info *info; unsigned int i; + /* + * d -> VKD3D_DATA_DOUBLE + * f -> VKD3D_DATA_FLOAT + * i -> VKD3D_DATA_INT + * u -> VKD3D_DATA_UINT + * O -> VKD3D_DATA_OPAQUE + * R -> VKD3D_DATA_RESOURCE + * S -> VKD3D_DATA_SAMPLER + * U -> VKD3D_DATA_UAV + */ + static const struct vkd3d_sm4_opcode_info opcode_table[] = + { + {VKD3D_SM4_OP_ADD, VKD3DSIH_ADD, "f", "ff"}, + {VKD3D_SM4_OP_AND, VKD3DSIH_AND, "u", "uu"}, + {VKD3D_SM4_OP_BREAK, VKD3DSIH_BREAK, "", ""}, + {VKD3D_SM4_OP_BREAKC, VKD3DSIH_BREAKP, "", "u", + shader_sm4_read_conditional_op}, + {VKD3D_SM4_OP_CASE, VKD3DSIH_CASE, "", "u", + shader_sm4_read_case_condition}, + {VKD3D_SM4_OP_CONTINUE, VKD3DSIH_CONTINUE, "", ""}, + {VKD3D_SM4_OP_CONTINUEC, VKD3DSIH_CONTINUEP, "", "u", + shader_sm4_read_conditional_op}, + {VKD3D_SM4_OP_CUT, VKD3DSIH_CUT, "", ""}, + {VKD3D_SM4_OP_DEFAULT, VKD3DSIH_DEFAULT, "", ""}, + {VKD3D_SM4_OP_DERIV_RTX, VKD3DSIH_DSX, "f", "f"}, + {VKD3D_SM4_OP_DERIV_RTY, VKD3DSIH_DSY, "f", "f"}, + {VKD3D_SM4_OP_DISCARD, VKD3DSIH_DISCARD, "", "u", + shader_sm4_read_conditional_op}, + {VKD3D_SM4_OP_DIV, VKD3DSIH_DIV, "f", "ff"}, + {VKD3D_SM4_OP_DP2, VKD3DSIH_DP2, "f", "ff"}, + {VKD3D_SM4_OP_DP3, VKD3DSIH_DP3, "f", "ff"}, + {VKD3D_SM4_OP_DP4, VKD3DSIH_DP4, "f", "ff"}, + {VKD3D_SM4_OP_ELSE, VKD3DSIH_ELSE, "", ""}, + {VKD3D_SM4_OP_EMIT, VKD3DSIH_EMIT, "", ""}, + {VKD3D_SM4_OP_ENDIF, VKD3DSIH_ENDIF, "", ""}, + {VKD3D_SM4_OP_ENDLOOP, VKD3DSIH_ENDLOOP, "", ""}, + {VKD3D_SM4_OP_ENDSWITCH, VKD3DSIH_ENDSWITCH, "", ""}, + {VKD3D_SM4_OP_EQ, VKD3DSIH_EQO, "u", "ff"}, + {VKD3D_SM4_OP_EXP, VKD3DSIH_EXP, "f", "f"}, + {VKD3D_SM4_OP_FRC, VKD3DSIH_FRC, "f", "f"}, + {VKD3D_SM4_OP_FTOI, VKD3DSIH_FTOI, "i", "f"}, + {VKD3D_SM4_OP_FTOU, VKD3DSIH_FTOU, "u", "f"}, + {VKD3D_SM4_OP_GE, VKD3DSIH_GEO, "u", "ff"}, + {VKD3D_SM4_OP_IADD, VKD3DSIH_IADD, "i", "ii"}, + {VKD3D_SM4_OP_IF, VKD3DSIH_IF, "", "u", + shader_sm4_read_conditional_op}, + {VKD3D_SM4_OP_IEQ, VKD3DSIH_IEQ, "u", "ii"}, + {VKD3D_SM4_OP_IGE, VKD3DSIH_IGE, "u", "ii"}, + {VKD3D_SM4_OP_ILT, VKD3DSIH_ILT, "u", "ii"}, + {VKD3D_SM4_OP_IMAD, VKD3DSIH_IMAD, "i", "iii"}, + {VKD3D_SM4_OP_IMAX, VKD3DSIH_IMAX, "i", "ii"}, + {VKD3D_SM4_OP_IMIN, VKD3DSIH_IMIN, "i", "ii"}, + {VKD3D_SM4_OP_IMUL, VKD3DSIH_IMUL, "ii", "ii"}, + {VKD3D_SM4_OP_INE, VKD3DSIH_INE, "u", "ii"}, + {VKD3D_SM4_OP_INEG, VKD3DSIH_INEG, "i", "i"}, + {VKD3D_SM4_OP_ISHL, VKD3DSIH_ISHL, "i", "ii"}, + {VKD3D_SM4_OP_ISHR, VKD3DSIH_ISHR, "i", "ii"}, + {VKD3D_SM4_OP_ITOF, VKD3DSIH_ITOF, "f", "i"}, + {VKD3D_SM4_OP_LABEL, VKD3DSIH_LABEL, "", "O"}, + {VKD3D_SM4_OP_LD, VKD3DSIH_LD, "u", "iR"}, + {VKD3D_SM4_OP_LD2DMS, VKD3DSIH_LD2DMS, "u", "iRi"}, + {VKD3D_SM4_OP_LOG, VKD3DSIH_LOG, "f", "f"}, + {VKD3D_SM4_OP_LOOP, VKD3DSIH_LOOP, "", ""}, + {VKD3D_SM4_OP_LT, VKD3DSIH_LTO, "u", "ff"}, + {VKD3D_SM4_OP_MAD, VKD3DSIH_MAD, "f", "fff"}, + {VKD3D_SM4_OP_MIN, VKD3DSIH_MIN, "f", "ff"}, + {VKD3D_SM4_OP_MAX, VKD3DSIH_MAX, "f", "ff"}, + {VKD3D_SM4_OP_SHADER_DATA, VKD3DSIH_DCL_IMMEDIATE_CONSTANT_BUFFER, "", "", + shader_sm4_read_shader_data}, + {VKD3D_SM4_OP_MOV, VKD3DSIH_MOV, "f", "f"}, + {VKD3D_SM4_OP_MOVC, VKD3DSIH_MOVC, "f", "uff"}, + {VKD3D_SM4_OP_MUL, VKD3DSIH_MUL, "f", "ff"}, + {VKD3D_SM4_OP_NE, VKD3DSIH_NEU, "u", "ff"}, + {VKD3D_SM4_OP_NOP, VKD3DSIH_NOP, "", ""}, + {VKD3D_SM4_OP_NOT, VKD3DSIH_NOT, "u", "u"}, + {VKD3D_SM4_OP_OR, VKD3DSIH_OR, "u", "uu"}, + {VKD3D_SM4_OP_RESINFO, VKD3DSIH_RESINFO, "f", "iR"}, + {VKD3D_SM4_OP_RET, VKD3DSIH_RET, "", ""}, + {VKD3D_SM4_OP_RETC, VKD3DSIH_RETP, "", "u", + shader_sm4_read_conditional_op}, + {VKD3D_SM4_OP_ROUND_NE, VKD3DSIH_ROUND_NE, "f", "f"}, + {VKD3D_SM4_OP_ROUND_NI, VKD3DSIH_ROUND_NI, "f", "f"}, + {VKD3D_SM4_OP_ROUND_PI, VKD3DSIH_ROUND_PI, "f", "f"}, + {VKD3D_SM4_OP_ROUND_Z, VKD3DSIH_ROUND_Z, "f", "f"}, + {VKD3D_SM4_OP_RSQ, VKD3DSIH_RSQ, "f", "f"}, + {VKD3D_SM4_OP_SAMPLE, VKD3DSIH_SAMPLE, "u", "fRS"}, + {VKD3D_SM4_OP_SAMPLE_C, VKD3DSIH_SAMPLE_C, "f", "fRSf"}, + {VKD3D_SM4_OP_SAMPLE_C_LZ, VKD3DSIH_SAMPLE_C_LZ, "f", "fRSf"}, + {VKD3D_SM4_OP_SAMPLE_LOD, VKD3DSIH_SAMPLE_LOD, "u", "fRSf"}, + {VKD3D_SM4_OP_SAMPLE_GRAD, VKD3DSIH_SAMPLE_GRAD, "u", "fRSff"}, + {VKD3D_SM4_OP_SAMPLE_B, VKD3DSIH_SAMPLE_B, "u", "fRSf"}, + {VKD3D_SM4_OP_SQRT, VKD3DSIH_SQRT, "f", "f"}, + {VKD3D_SM4_OP_SWITCH, VKD3DSIH_SWITCH, "", "i"}, + {VKD3D_SM4_OP_SINCOS, VKD3DSIH_SINCOS, "ff", "f"}, + {VKD3D_SM4_OP_UDIV, VKD3DSIH_UDIV, "uu", "uu"}, + {VKD3D_SM4_OP_ULT, VKD3DSIH_ULT, "u", "uu"}, + {VKD3D_SM4_OP_UGE, VKD3DSIH_UGE, "u", "uu"}, + {VKD3D_SM4_OP_UMUL, VKD3DSIH_UMUL, "uu", "uu"}, + {VKD3D_SM4_OP_UMAX, VKD3DSIH_UMAX, "u", "uu"}, + {VKD3D_SM4_OP_UMIN, VKD3DSIH_UMIN, "u", "uu"}, + {VKD3D_SM4_OP_USHR, VKD3DSIH_USHR, "u", "uu"}, + {VKD3D_SM4_OP_UTOF, VKD3DSIH_UTOF, "f", "u"}, + {VKD3D_SM4_OP_XOR, VKD3DSIH_XOR, "u", "uu"}, + {VKD3D_SM4_OP_DCL_RESOURCE, VKD3DSIH_DCL, "", "", + shader_sm4_read_dcl_resource}, + {VKD3D_SM4_OP_DCL_CONSTANT_BUFFER, VKD3DSIH_DCL_CONSTANT_BUFFER, "", "", + shader_sm4_read_dcl_constant_buffer}, + {VKD3D_SM4_OP_DCL_SAMPLER, VKD3DSIH_DCL_SAMPLER, "", "", + shader_sm4_read_dcl_sampler}, + {VKD3D_SM4_OP_DCL_INDEX_RANGE, VKD3DSIH_DCL_INDEX_RANGE, "", "", + shader_sm4_read_dcl_index_range}, + {VKD3D_SM4_OP_DCL_OUTPUT_TOPOLOGY, VKD3DSIH_DCL_OUTPUT_TOPOLOGY, "", "", + shader_sm4_read_dcl_output_topology}, + {VKD3D_SM4_OP_DCL_INPUT_PRIMITIVE, VKD3DSIH_DCL_INPUT_PRIMITIVE, "", "", + shader_sm4_read_dcl_input_primitive}, + {VKD3D_SM4_OP_DCL_VERTICES_OUT, VKD3DSIH_DCL_VERTICES_OUT, "", "", + shader_sm4_read_declaration_count}, + {VKD3D_SM4_OP_DCL_INPUT, VKD3DSIH_DCL_INPUT, "", "", + shader_sm4_read_declaration_dst}, + {VKD3D_SM4_OP_DCL_INPUT_SGV, VKD3DSIH_DCL_INPUT_SGV, "", "", + shader_sm4_read_declaration_register_semantic}, + {VKD3D_SM4_OP_DCL_INPUT_SIV, VKD3DSIH_DCL_INPUT_SIV, "", "", + shader_sm4_read_declaration_register_semantic}, + {VKD3D_SM4_OP_DCL_INPUT_PS, VKD3DSIH_DCL_INPUT_PS, "", "", + shader_sm4_read_dcl_input_ps}, + {VKD3D_SM4_OP_DCL_INPUT_PS_SGV, VKD3DSIH_DCL_INPUT_PS_SGV, "", "", + shader_sm4_read_declaration_register_semantic}, + {VKD3D_SM4_OP_DCL_INPUT_PS_SIV, VKD3DSIH_DCL_INPUT_PS_SIV, "", "", + shader_sm4_read_dcl_input_ps_siv}, + {VKD3D_SM4_OP_DCL_OUTPUT, VKD3DSIH_DCL_OUTPUT, "", "", + shader_sm4_read_declaration_dst}, + {VKD3D_SM4_OP_DCL_OUTPUT_SIV, VKD3DSIH_DCL_OUTPUT_SIV, "", "", + shader_sm4_read_declaration_register_semantic}, + {VKD3D_SM4_OP_DCL_TEMPS, VKD3DSIH_DCL_TEMPS, "", "", + shader_sm4_read_declaration_count}, + {VKD3D_SM4_OP_DCL_INDEXABLE_TEMP, VKD3DSIH_DCL_INDEXABLE_TEMP, "", "", + shader_sm4_read_dcl_indexable_temp}, + {VKD3D_SM4_OP_DCL_GLOBAL_FLAGS, VKD3DSIH_DCL_GLOBAL_FLAGS, "", "", + shader_sm4_read_dcl_global_flags}, + {VKD3D_SM4_OP_LOD, VKD3DSIH_LOD, "f", "fRS"}, + {VKD3D_SM4_OP_GATHER4, VKD3DSIH_GATHER4, "u", "fRS"}, + {VKD3D_SM4_OP_SAMPLE_POS, VKD3DSIH_SAMPLE_POS, "f", "Ru"}, + {VKD3D_SM4_OP_SAMPLE_INFO, VKD3DSIH_SAMPLE_INFO, "f", "R"}, + {VKD3D_SM5_OP_HS_DECLS, VKD3DSIH_HS_DECLS, "", ""}, + {VKD3D_SM5_OP_HS_CONTROL_POINT_PHASE, VKD3DSIH_HS_CONTROL_POINT_PHASE, "", ""}, + {VKD3D_SM5_OP_HS_FORK_PHASE, VKD3DSIH_HS_FORK_PHASE, "", ""}, + {VKD3D_SM5_OP_HS_JOIN_PHASE, VKD3DSIH_HS_JOIN_PHASE, "", ""}, + {VKD3D_SM5_OP_EMIT_STREAM, VKD3DSIH_EMIT_STREAM, "", "f"}, + {VKD3D_SM5_OP_CUT_STREAM, VKD3DSIH_CUT_STREAM, "", "f"}, + {VKD3D_SM5_OP_FCALL, VKD3DSIH_FCALL, "", "O", + shader_sm5_read_fcall}, + {VKD3D_SM5_OP_BUFINFO, VKD3DSIH_BUFINFO, "i", "U"}, + {VKD3D_SM5_OP_DERIV_RTX_COARSE, VKD3DSIH_DSX_COARSE, "f", "f"}, + {VKD3D_SM5_OP_DERIV_RTX_FINE, VKD3DSIH_DSX_FINE, "f", "f"}, + {VKD3D_SM5_OP_DERIV_RTY_COARSE, VKD3DSIH_DSY_COARSE, "f", "f"}, + {VKD3D_SM5_OP_DERIV_RTY_FINE, VKD3DSIH_DSY_FINE, "f", "f"}, + {VKD3D_SM5_OP_GATHER4_C, VKD3DSIH_GATHER4_C, "f", "fRSf"}, + {VKD3D_SM5_OP_GATHER4_PO, VKD3DSIH_GATHER4_PO, "f", "fiRS"}, + {VKD3D_SM5_OP_GATHER4_PO_C, VKD3DSIH_GATHER4_PO_C, "f", "fiRSf"}, + {VKD3D_SM5_OP_RCP, VKD3DSIH_RCP, "f", "f"}, + {VKD3D_SM5_OP_F32TOF16, VKD3DSIH_F32TOF16, "u", "f"}, + {VKD3D_SM5_OP_F16TOF32, VKD3DSIH_F16TOF32, "f", "u"}, + {VKD3D_SM5_OP_COUNTBITS, VKD3DSIH_COUNTBITS, "u", "u"}, + {VKD3D_SM5_OP_FIRSTBIT_HI, VKD3DSIH_FIRSTBIT_HI, "u", "u"}, + {VKD3D_SM5_OP_FIRSTBIT_LO, VKD3DSIH_FIRSTBIT_LO, "u", "u"}, + {VKD3D_SM5_OP_FIRSTBIT_SHI, VKD3DSIH_FIRSTBIT_SHI, "u", "i"}, + {VKD3D_SM5_OP_UBFE, VKD3DSIH_UBFE, "u", "iiu"}, + {VKD3D_SM5_OP_IBFE, VKD3DSIH_IBFE, "i", "iii"}, + {VKD3D_SM5_OP_BFI, VKD3DSIH_BFI, "u", "iiuu"}, + {VKD3D_SM5_OP_BFREV, VKD3DSIH_BFREV, "u", "u"}, + {VKD3D_SM5_OP_SWAPC, VKD3DSIH_SWAPC, "ff", "uff"}, + {VKD3D_SM5_OP_DCL_STREAM, VKD3DSIH_DCL_STREAM, "", "O"}, + {VKD3D_SM5_OP_DCL_FUNCTION_BODY, VKD3DSIH_DCL_FUNCTION_BODY, "", "", + shader_sm5_read_dcl_function_body}, + {VKD3D_SM5_OP_DCL_FUNCTION_TABLE, VKD3DSIH_DCL_FUNCTION_TABLE, "", "", + shader_sm5_read_dcl_function_table}, + {VKD3D_SM5_OP_DCL_INTERFACE, VKD3DSIH_DCL_INTERFACE, "", "", + shader_sm5_read_dcl_interface}, + {VKD3D_SM5_OP_DCL_INPUT_CONTROL_POINT_COUNT, VKD3DSIH_DCL_INPUT_CONTROL_POINT_COUNT, "", "", + shader_sm5_read_control_point_count}, + {VKD3D_SM5_OP_DCL_OUTPUT_CONTROL_POINT_COUNT, VKD3DSIH_DCL_OUTPUT_CONTROL_POINT_COUNT, "", "", + shader_sm5_read_control_point_count}, + {VKD3D_SM5_OP_DCL_TESSELLATOR_DOMAIN, VKD3DSIH_DCL_TESSELLATOR_DOMAIN, "", "", + shader_sm5_read_dcl_tessellator_domain}, + {VKD3D_SM5_OP_DCL_TESSELLATOR_PARTITIONING, VKD3DSIH_DCL_TESSELLATOR_PARTITIONING, "", "", + shader_sm5_read_dcl_tessellator_partitioning}, + {VKD3D_SM5_OP_DCL_TESSELLATOR_OUTPUT_PRIMITIVE, VKD3DSIH_DCL_TESSELLATOR_OUTPUT_PRIMITIVE, "", "", + shader_sm5_read_dcl_tessellator_output_primitive}, + {VKD3D_SM5_OP_DCL_HS_MAX_TESSFACTOR, VKD3DSIH_DCL_HS_MAX_TESSFACTOR, "", "", + shader_sm5_read_dcl_hs_max_tessfactor}, + {VKD3D_SM5_OP_DCL_HS_FORK_PHASE_INSTANCE_COUNT, VKD3DSIH_DCL_HS_FORK_PHASE_INSTANCE_COUNT, "", "", + shader_sm4_read_declaration_count}, + {VKD3D_SM5_OP_DCL_HS_JOIN_PHASE_INSTANCE_COUNT, VKD3DSIH_DCL_HS_JOIN_PHASE_INSTANCE_COUNT, "", "", + shader_sm4_read_declaration_count}, + {VKD3D_SM5_OP_DCL_THREAD_GROUP, VKD3DSIH_DCL_THREAD_GROUP, "", "", + shader_sm5_read_dcl_thread_group}, + {VKD3D_SM5_OP_DCL_UAV_TYPED, VKD3DSIH_DCL_UAV_TYPED, "", "", + shader_sm4_read_dcl_resource}, + {VKD3D_SM5_OP_DCL_UAV_RAW, VKD3DSIH_DCL_UAV_RAW, "", "", + shader_sm5_read_dcl_uav_raw}, + {VKD3D_SM5_OP_DCL_UAV_STRUCTURED, VKD3DSIH_DCL_UAV_STRUCTURED, "", "", + shader_sm5_read_dcl_uav_structured}, + {VKD3D_SM5_OP_DCL_TGSM_RAW, VKD3DSIH_DCL_TGSM_RAW, "", "", + shader_sm5_read_dcl_tgsm_raw}, + {VKD3D_SM5_OP_DCL_TGSM_STRUCTURED, VKD3DSIH_DCL_TGSM_STRUCTURED, "", "", + shader_sm5_read_dcl_tgsm_structured}, + {VKD3D_SM5_OP_DCL_RESOURCE_RAW, VKD3DSIH_DCL_RESOURCE_RAW, "", "", + shader_sm5_read_dcl_resource_raw}, + {VKD3D_SM5_OP_DCL_RESOURCE_STRUCTURED, VKD3DSIH_DCL_RESOURCE_STRUCTURED, "", "", + shader_sm5_read_dcl_resource_structured}, + {VKD3D_SM5_OP_LD_UAV_TYPED, VKD3DSIH_LD_UAV_TYPED, "u", "iU"}, + {VKD3D_SM5_OP_STORE_UAV_TYPED, VKD3DSIH_STORE_UAV_TYPED, "U", "iu"}, + {VKD3D_SM5_OP_LD_RAW, VKD3DSIH_LD_RAW, "u", "iU"}, + {VKD3D_SM5_OP_STORE_RAW, VKD3DSIH_STORE_RAW, "U", "uu"}, + {VKD3D_SM5_OP_LD_STRUCTURED, VKD3DSIH_LD_STRUCTURED, "u", "iiR"}, + {VKD3D_SM5_OP_STORE_STRUCTURED, VKD3DSIH_STORE_STRUCTURED, "U", "iiu"}, + {VKD3D_SM5_OP_ATOMIC_AND, VKD3DSIH_ATOMIC_AND, "U", "iu"}, + {VKD3D_SM5_OP_ATOMIC_OR, VKD3DSIH_ATOMIC_OR, "U", "iu"}, + {VKD3D_SM5_OP_ATOMIC_XOR, VKD3DSIH_ATOMIC_XOR, "U", "iu"}, + {VKD3D_SM5_OP_ATOMIC_CMP_STORE, VKD3DSIH_ATOMIC_CMP_STORE, "U", "iuu"}, + {VKD3D_SM5_OP_ATOMIC_IADD, VKD3DSIH_ATOMIC_IADD, "U", "ii"}, + {VKD3D_SM5_OP_ATOMIC_IMAX, VKD3DSIH_ATOMIC_IMAX, "U", "ii"}, + {VKD3D_SM5_OP_ATOMIC_IMIN, VKD3DSIH_ATOMIC_IMIN, "U", "ii"}, + {VKD3D_SM5_OP_ATOMIC_UMAX, VKD3DSIH_ATOMIC_UMAX, "U", "iu"}, + {VKD3D_SM5_OP_ATOMIC_UMIN, VKD3DSIH_ATOMIC_UMIN, "U", "iu"}, + {VKD3D_SM5_OP_IMM_ATOMIC_ALLOC, VKD3DSIH_IMM_ATOMIC_ALLOC, "u", "U"}, + {VKD3D_SM5_OP_IMM_ATOMIC_CONSUME, VKD3DSIH_IMM_ATOMIC_CONSUME, "u", "U"}, + {VKD3D_SM5_OP_IMM_ATOMIC_IADD, VKD3DSIH_IMM_ATOMIC_IADD, "uU", "ii"}, + {VKD3D_SM5_OP_IMM_ATOMIC_AND, VKD3DSIH_IMM_ATOMIC_AND, "uU", "iu"}, + {VKD3D_SM5_OP_IMM_ATOMIC_OR, VKD3DSIH_IMM_ATOMIC_OR, "uU", "iu"}, + {VKD3D_SM5_OP_IMM_ATOMIC_XOR, VKD3DSIH_IMM_ATOMIC_XOR, "uU", "iu"}, + {VKD3D_SM5_OP_IMM_ATOMIC_EXCH, VKD3DSIH_IMM_ATOMIC_EXCH, "uU", "iu"}, + {VKD3D_SM5_OP_IMM_ATOMIC_CMP_EXCH, VKD3DSIH_IMM_ATOMIC_CMP_EXCH, "uU", "iuu"}, + {VKD3D_SM5_OP_IMM_ATOMIC_IMAX, VKD3DSIH_IMM_ATOMIC_IMAX, "iU", "ii"}, + {VKD3D_SM5_OP_IMM_ATOMIC_IMIN, VKD3DSIH_IMM_ATOMIC_IMIN, "iU", "ii"}, + {VKD3D_SM5_OP_IMM_ATOMIC_UMAX, VKD3DSIH_IMM_ATOMIC_UMAX, "uU", "iu"}, + {VKD3D_SM5_OP_IMM_ATOMIC_UMIN, VKD3DSIH_IMM_ATOMIC_UMIN, "uU", "iu"}, + {VKD3D_SM5_OP_SYNC, VKD3DSIH_SYNC, "", "", + shader_sm5_read_sync}, + {VKD3D_SM5_OP_DADD, VKD3DSIH_DADD, "d", "dd"}, + {VKD3D_SM5_OP_DMAX, VKD3DSIH_DMAX, "d", "dd"}, + {VKD3D_SM5_OP_DMIN, VKD3DSIH_DMIN, "d", "dd"}, + {VKD3D_SM5_OP_DMUL, VKD3DSIH_DMUL, "d", "dd"}, + {VKD3D_SM5_OP_DEQ, VKD3DSIH_DEQO, "u", "dd"}, + {VKD3D_SM5_OP_DGE, VKD3DSIH_DGEO, "u", "dd"}, + {VKD3D_SM5_OP_DLT, VKD3DSIH_DLT, "u", "dd"}, + {VKD3D_SM5_OP_DNE, VKD3DSIH_DNE, "u", "dd"}, + {VKD3D_SM5_OP_DMOV, VKD3DSIH_DMOV, "d", "d"}, + {VKD3D_SM5_OP_DMOVC, VKD3DSIH_DMOVC, "d", "udd"}, + {VKD3D_SM5_OP_DTOF, VKD3DSIH_DTOF, "f", "d"}, + {VKD3D_SM5_OP_FTOD, VKD3DSIH_FTOD, "d", "f"}, + {VKD3D_SM5_OP_EVAL_SAMPLE_INDEX, VKD3DSIH_EVAL_SAMPLE_INDEX, "f", "fi"}, + {VKD3D_SM5_OP_EVAL_CENTROID, VKD3DSIH_EVAL_CENTROID, "f", "f"}, + {VKD3D_SM5_OP_DCL_GS_INSTANCES, VKD3DSIH_DCL_GS_INSTANCES, "", "", + shader_sm4_read_declaration_count}, + {VKD3D_SM5_OP_DDIV, VKD3DSIH_DDIV, "d", "dd"}, + {VKD3D_SM5_OP_DFMA, VKD3DSIH_DFMA, "d", "ddd"}, + {VKD3D_SM5_OP_DRCP, VKD3DSIH_DRCP, "d", "d"}, + {VKD3D_SM5_OP_MSAD, VKD3DSIH_MSAD, "u", "uuu"}, + {VKD3D_SM5_OP_DTOI, VKD3DSIH_DTOI, "i", "d"}, + {VKD3D_SM5_OP_DTOU, VKD3DSIH_DTOU, "u", "d"}, + {VKD3D_SM5_OP_ITOD, VKD3DSIH_ITOD, "d", "i"}, + {VKD3D_SM5_OP_UTOD, VKD3DSIH_UTOD, "d", "u"}, + {VKD3D_SM5_OP_GATHER4_S, VKD3DSIH_GATHER4_S, "uu", "fRS"}, + {VKD3D_SM5_OP_GATHER4_C_S, VKD3DSIH_GATHER4_C_S, "fu", "fRSf"}, + {VKD3D_SM5_OP_GATHER4_PO_S, VKD3DSIH_GATHER4_PO_S, "fu", "fiRS"}, + {VKD3D_SM5_OP_GATHER4_PO_C_S, VKD3DSIH_GATHER4_PO_C_S, "fu", "fiRSf"}, + {VKD3D_SM5_OP_LD_S, VKD3DSIH_LD_S, "uu", "iR"}, + {VKD3D_SM5_OP_LD2DMS_S, VKD3DSIH_LD2DMS_S, "uu", "iRi"}, + {VKD3D_SM5_OP_LD_UAV_TYPED_S, VKD3DSIH_LD_UAV_TYPED_S, "uu", "iU"}, + {VKD3D_SM5_OP_LD_RAW_S, VKD3DSIH_LD_RAW_S, "uu", "iU"}, + {VKD3D_SM5_OP_LD_STRUCTURED_S, VKD3DSIH_LD_STRUCTURED_S, "uu", "iiR"}, + {VKD3D_SM5_OP_SAMPLE_LOD_S, VKD3DSIH_SAMPLE_LOD_S, "uu", "fRSf"}, + {VKD3D_SM5_OP_SAMPLE_C_LZ_S, VKD3DSIH_SAMPLE_C_LZ_S, "fu", "fRSf"}, + {VKD3D_SM5_OP_SAMPLE_CL_S, VKD3DSIH_SAMPLE_CL_S, "uu", "fRSf"}, + {VKD3D_SM5_OP_SAMPLE_B_CL_S, VKD3DSIH_SAMPLE_B_CL_S, "uu", "fRSff"}, + {VKD3D_SM5_OP_SAMPLE_GRAD_CL_S, VKD3DSIH_SAMPLE_GRAD_CL_S, "uu", "fRSfff"}, + {VKD3D_SM5_OP_SAMPLE_C_CL_S, VKD3DSIH_SAMPLE_C_CL_S, "fu", "fRSff"}, + {VKD3D_SM5_OP_CHECK_ACCESS_FULLY_MAPPED, VKD3DSIH_CHECK_ACCESS_FULLY_MAPPED, "u", "u"}, + }; + static const struct vkd3d_sm4_register_type_info register_type_table[] = { - {VKD3D_SM4_RT_TEMP, VKD3DSPR_TEMP}, - {VKD3D_SM4_RT_INPUT, VKD3DSPR_INPUT}, - {VKD3D_SM4_RT_OUTPUT, VKD3DSPR_OUTPUT}, - {VKD3D_SM4_RT_INDEXABLE_TEMP, VKD3DSPR_IDXTEMP}, - {VKD3D_SM4_RT_IMMCONST, VKD3DSPR_IMMCONST}, - {VKD3D_SM4_RT_IMMCONST64, VKD3DSPR_IMMCONST64}, - {VKD3D_SM4_RT_SAMPLER, VKD3DSPR_SAMPLER}, - {VKD3D_SM4_RT_RESOURCE, VKD3DSPR_RESOURCE}, - {VKD3D_SM4_RT_CONSTBUFFER, VKD3DSPR_CONSTBUFFER}, - {VKD3D_SM4_RT_IMMCONSTBUFFER, VKD3DSPR_IMMCONSTBUFFER}, - {VKD3D_SM4_RT_PRIMID, VKD3DSPR_PRIMID}, - {VKD3D_SM4_RT_DEPTHOUT, VKD3DSPR_DEPTHOUT}, - {VKD3D_SM4_RT_NULL, VKD3DSPR_NULL}, - {VKD3D_SM4_RT_RASTERIZER, VKD3DSPR_RASTERIZER}, - {VKD3D_SM4_RT_OMASK, VKD3DSPR_SAMPLEMASK}, - {VKD3D_SM5_RT_STREAM, VKD3DSPR_STREAM}, - {VKD3D_SM5_RT_FUNCTION_BODY, VKD3DSPR_FUNCTIONBODY}, - {VKD3D_SM5_RT_FUNCTION_POINTER, VKD3DSPR_FUNCTIONPOINTER}, - {VKD3D_SM5_RT_OUTPUT_CONTROL_POINT_ID, VKD3DSPR_OUTPOINTID}, - {VKD3D_SM5_RT_FORK_INSTANCE_ID, VKD3DSPR_FORKINSTID}, - {VKD3D_SM5_RT_JOIN_INSTANCE_ID, VKD3DSPR_JOININSTID}, - {VKD3D_SM5_RT_INPUT_CONTROL_POINT, VKD3DSPR_INCONTROLPOINT}, - {VKD3D_SM5_RT_OUTPUT_CONTROL_POINT, VKD3DSPR_OUTCONTROLPOINT}, - {VKD3D_SM5_RT_PATCH_CONSTANT_DATA, VKD3DSPR_PATCHCONST}, - {VKD3D_SM5_RT_DOMAIN_LOCATION, VKD3DSPR_TESSCOORD}, - {VKD3D_SM5_RT_UAV, VKD3DSPR_UAV}, - {VKD3D_SM5_RT_SHARED_MEMORY, VKD3DSPR_GROUPSHAREDMEM}, - {VKD3D_SM5_RT_THREAD_ID, VKD3DSPR_THREADID}, - {VKD3D_SM5_RT_THREAD_GROUP_ID, VKD3DSPR_THREADGROUPID}, - {VKD3D_SM5_RT_LOCAL_THREAD_ID, VKD3DSPR_LOCALTHREADID}, - {VKD3D_SM5_RT_COVERAGE, VKD3DSPR_COVERAGE}, - {VKD3D_SM5_RT_LOCAL_THREAD_INDEX, VKD3DSPR_LOCALTHREADINDEX}, - {VKD3D_SM5_RT_GS_INSTANCE_ID, VKD3DSPR_GSINSTID}, - {VKD3D_SM5_RT_DEPTHOUT_GREATER_EQUAL, VKD3DSPR_DEPTHOUTGE}, - {VKD3D_SM5_RT_DEPTHOUT_LESS_EQUAL, VKD3DSPR_DEPTHOUTLE}, - {VKD3D_SM5_RT_OUTPUT_STENCIL_REF, VKD3DSPR_OUTSTENCILREF}, + {VKD3D_SM4_RT_TEMP, VKD3DSPR_TEMP, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM4_RT_INPUT, VKD3DSPR_INPUT, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM4_RT_OUTPUT, VKD3DSPR_OUTPUT, VKD3D_SM4_SWIZZLE_INVALID}, + {VKD3D_SM4_RT_INDEXABLE_TEMP, VKD3DSPR_IDXTEMP, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM4_RT_IMMCONST, VKD3DSPR_IMMCONST, VKD3D_SM4_SWIZZLE_NONE}, + {VKD3D_SM4_RT_IMMCONST64, VKD3DSPR_IMMCONST64, VKD3D_SM4_SWIZZLE_NONE}, + {VKD3D_SM4_RT_SAMPLER, VKD3DSPR_SAMPLER, VKD3D_SM4_SWIZZLE_SCALAR}, + {VKD3D_SM4_RT_RESOURCE, VKD3DSPR_RESOURCE, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM4_RT_CONSTBUFFER, VKD3DSPR_CONSTBUFFER, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM4_RT_IMMCONSTBUFFER, VKD3DSPR_IMMCONSTBUFFER, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM4_RT_PRIMID, VKD3DSPR_PRIMID, VKD3D_SM4_SWIZZLE_NONE}, + {VKD3D_SM4_RT_DEPTHOUT, VKD3DSPR_DEPTHOUT, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM4_RT_NULL, VKD3DSPR_NULL, VKD3D_SM4_SWIZZLE_INVALID}, + {VKD3D_SM4_RT_RASTERIZER, VKD3DSPR_RASTERIZER, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM4_RT_OMASK, VKD3DSPR_SAMPLEMASK, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_STREAM, VKD3DSPR_STREAM, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_FUNCTION_BODY, VKD3DSPR_FUNCTIONBODY, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_FUNCTION_POINTER, VKD3DSPR_FUNCTIONPOINTER, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_OUTPUT_CONTROL_POINT_ID, VKD3DSPR_OUTPOINTID, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_FORK_INSTANCE_ID, VKD3DSPR_FORKINSTID, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_JOIN_INSTANCE_ID, VKD3DSPR_JOININSTID, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_INPUT_CONTROL_POINT, VKD3DSPR_INCONTROLPOINT, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_OUTPUT_CONTROL_POINT, VKD3DSPR_OUTCONTROLPOINT, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_PATCH_CONSTANT_DATA, VKD3DSPR_PATCHCONST, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_DOMAIN_LOCATION, VKD3DSPR_TESSCOORD, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_UAV, VKD3DSPR_UAV, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_SHARED_MEMORY, VKD3DSPR_GROUPSHAREDMEM, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_THREAD_ID, VKD3DSPR_THREADID, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_THREAD_GROUP_ID, VKD3DSPR_THREADGROUPID, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_LOCAL_THREAD_ID, VKD3DSPR_LOCALTHREADID, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_COVERAGE, VKD3DSPR_COVERAGE, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_LOCAL_THREAD_INDEX, VKD3DSPR_LOCALTHREADINDEX,VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_GS_INSTANCE_ID, VKD3DSPR_GSINSTID, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_DEPTHOUT_GREATER_EQUAL, VKD3DSPR_DEPTHOUTGE, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_DEPTHOUT_LESS_EQUAL, VKD3DSPR_DEPTHOUTLE, VKD3D_SM4_SWIZZLE_VEC4}, + {VKD3D_SM5_RT_OUTPUT_STENCIL_REF, VKD3DSPR_OUTSTENCILREF, VKD3D_SM4_SWIZZLE_VEC4}, }; memset(lookup, 0, sizeof(*lookup)); + for (i = 0; i < ARRAY_SIZE(opcode_table); ++i) + { + const struct vkd3d_sm4_opcode_info *info = &opcode_table[i]; + + lookup->opcode_info_from_sm4[info->opcode] = info; + } + for (i = 0; i < ARRAY_SIZE(register_type_table); ++i) { - info = ®ister_type_table[i]; + const struct vkd3d_sm4_register_type_info *info = ®ister_type_table[i]; + lookup->register_type_info_from_sm4[info->sm4_type] = info; lookup->register_type_info_from_vkd3d[info->vkd3d_type] = info; } @@ -1590,6 +1656,14 @@ static void tpf_writer_init(struct tpf_writer *tpf, struct hlsl_ctx *ctx, struct init_sm4_lookup_tables(&tpf->lookup); } +static const struct vkd3d_sm4_opcode_info *get_info_from_sm4_opcode( + const struct vkd3d_sm4_lookup_tables *lookup, enum vkd3d_sm4_opcode sm4_opcode) +{ + if (sm4_opcode >= VKD3D_SM4_OP_COUNT) + return NULL; + return lookup->opcode_info_from_sm4[sm4_opcode]; +} + static const struct vkd3d_sm4_register_type_info *get_info_from_sm4_register_type( const struct vkd3d_sm4_lookup_tables *lookup, enum vkd3d_sm4_register_type sm4_type) { @@ -1606,29 +1680,14 @@ static const struct vkd3d_sm4_register_type_info *get_info_from_vkd3d_register_t return lookup->register_type_info_from_vkd3d[vkd3d_type]; } -static void map_register(const struct vkd3d_shader_sm4_parser *sm4, struct vkd3d_shader_register *reg) +static enum vkd3d_sm4_swizzle_type vkd3d_sm4_get_default_swizzle_type( + const struct vkd3d_sm4_lookup_tables *lookup, enum vkd3d_shader_register_type vkd3d_type) { - switch (sm4->p.shader_version.type) - { - case VKD3D_SHADER_TYPE_PIXEL: - if (reg->type == VKD3DSPR_OUTPUT) - { - unsigned int reg_idx = reg->idx[0].offset; + const struct vkd3d_sm4_register_type_info *register_type_info = + get_info_from_vkd3d_register_type(lookup, vkd3d_type); - if (reg_idx >= ARRAY_SIZE(sm4->output_map)) - { - /* Validated later */ - break; - } - - reg->type = VKD3DSPR_COLOROUT; - reg->idx[0].offset = sm4->output_map[reg_idx]; - } - break; - - default: - break; - } + assert(register_type_info); + return register_type_info->default_src_swizzle_type; } static enum vkd3d_data_type map_data_type(char t) @@ -1694,29 +1753,16 @@ static bool shader_sm4_read_reg_idx(struct vkd3d_shader_sm4_parser *priv, const return true; } -static bool sm4_register_is_descriptor(enum vkd3d_sm4_register_type register_type) -{ - switch (register_type) - { - case VKD3D_SM4_RT_SAMPLER: - case VKD3D_SM4_RT_RESOURCE: - case VKD3D_SM4_RT_CONSTBUFFER: - case VKD3D_SM5_RT_UAV: - return true; - - default: - return false; - } -} - static bool shader_sm4_read_param(struct vkd3d_shader_sm4_parser *priv, const uint32_t **ptr, const uint32_t *end, enum vkd3d_data_type data_type, struct vkd3d_shader_register *param, enum vkd3d_shader_src_modifier *modifier) { const struct vkd3d_sm4_register_type_info *register_type_info; + enum vkd3d_shader_register_type vsir_register_type; enum vkd3d_sm4_register_precision precision; enum vkd3d_sm4_register_type register_type; enum vkd3d_sm4_extended_operand_type type; enum vkd3d_sm4_register_modifier m; + enum vkd3d_sm4_dimension sm4_dimension; uint32_t token, order, extended; if (*ptr >= end) @@ -1731,15 +1777,18 @@ static bool shader_sm4_read_param(struct vkd3d_shader_sm4_parser *priv, const ui if (!register_type_info) { FIXME("Unhandled register type %#x.\n", register_type); - param->type = VKD3DSPR_TEMP; + vsir_register_type = VKD3DSPR_TEMP; } else { - param->type = register_type_info->vkd3d_type; + vsir_register_type = register_type_info->vkd3d_type; } + + order = (token & VKD3D_SM4_REGISTER_ORDER_MASK) >> VKD3D_SM4_REGISTER_ORDER_SHIFT; + + vsir_register_init(param, vsir_register_type, data_type, order); param->precision = VKD3D_SHADER_REGISTER_PRECISION_DEFAULT; param->non_uniform = false; - param->data_type = data_type; *modifier = VKD3DSPSM_NONE; if (token & VKD3D_SM4_EXTENDED_OPERAND) @@ -1775,10 +1824,11 @@ static bool shader_sm4_read_param(struct vkd3d_shader_sm4_parser *priv, const ui *modifier = VKD3DSPSM_ABSNEG; break; + case VKD3D_SM4_REGISTER_MODIFIER_NONE: + break; + default: FIXME("Unhandled register modifier %#x.\n", m); - /* fall-through */ - case VKD3D_SM4_REGISTER_MODIFIER_NONE: break; } @@ -1809,14 +1859,7 @@ static bool shader_sm4_read_param(struct vkd3d_shader_sm4_parser *priv, const ui } } - order = (token & VKD3D_SM4_REGISTER_ORDER_MASK) >> VKD3D_SM4_REGISTER_ORDER_SHIFT; - - if (order < 1) - { - param->idx[0].offset = ~0u; - param->idx[0].rel_addr = NULL; - } - else + if (order >= 1) { DWORD addressing = (token & VKD3D_SM4_ADDRESSING_MASK0) >> VKD3D_SM4_ADDRESSING_SHIFT0; if (!(shader_sm4_read_reg_idx(priv, ptr, end, addressing, ¶m->idx[0]))) @@ -1826,12 +1869,7 @@ static bool shader_sm4_read_param(struct vkd3d_shader_sm4_parser *priv, const ui } } - if (order < 2) - { - param->idx[1].offset = ~0u; - param->idx[1].rel_addr = NULL; - } - else + if (order >= 2) { DWORD addressing = (token & VKD3D_SM4_ADDRESSING_MASK1) >> VKD3D_SM4_ADDRESSING_SHIFT1; if (!(shader_sm4_read_reg_idx(priv, ptr, end, addressing, ¶m->idx[1]))) @@ -1841,12 +1879,7 @@ static bool shader_sm4_read_param(struct vkd3d_shader_sm4_parser *priv, const ui } } - if (order < 3) - { - param->idx[2].offset = ~0u; - param->idx[2].rel_addr = NULL; - } - else + if (order >= 3) { DWORD addressing = (token & VKD3D_SM4_ADDRESSING_MASK2) >> VKD3D_SM4_ADDRESSING_SHIFT2; if (!(shader_sm4_read_reg_idx(priv, ptr, end, addressing, ¶m->idx[2]))) @@ -1862,17 +1895,16 @@ static bool shader_sm4_read_param(struct vkd3d_shader_sm4_parser *priv, const ui return false; } - param->idx_count = order; + sm4_dimension = (token & VKD3D_SM4_DIMENSION_MASK) >> VKD3D_SM4_DIMENSION_SHIFT; + param->dimension = vsir_dimension_from_sm4_dimension(sm4_dimension); if (register_type == VKD3D_SM4_RT_IMMCONST || register_type == VKD3D_SM4_RT_IMMCONST64) { - enum vkd3d_sm4_dimension dimension = (token & VKD3D_SM4_DIMENSION_MASK) >> VKD3D_SM4_DIMENSION_SHIFT; unsigned int dword_count; - switch (dimension) + switch (param->dimension) { - case VKD3D_SM4_DIMENSION_SCALAR: - param->immconst_type = VKD3D_IMMCONST_SCALAR; + case VSIR_DIMENSION_SCALAR: dword_count = 1 + (register_type == VKD3D_SM4_RT_IMMCONST64); if (end - *ptr < dword_count) { @@ -1883,8 +1915,7 @@ static bool shader_sm4_read_param(struct vkd3d_shader_sm4_parser *priv, const ui *ptr += dword_count; break; - case VKD3D_SM4_DIMENSION_VEC4: - param->immconst_type = VKD3D_IMMCONST_VEC4; + case VSIR_DIMENSION_VEC4: if (end - *ptr < VKD3D_VEC4_SIZE) { WARN("Invalid ptr %p, end %p.\n", *ptr, end); @@ -1895,11 +1926,11 @@ static bool shader_sm4_read_param(struct vkd3d_shader_sm4_parser *priv, const ui break; default: - FIXME("Unhandled dimension %#x.\n", dimension); + FIXME("Unhandled dimension %#x.\n", param->dimension); break; } } - else if (!shader_is_sm_5_1(priv) && sm4_register_is_descriptor(register_type)) + else if (!shader_is_sm_5_1(priv) && vsir_register_is_descriptor(param)) { /* SM5.1 places a symbol identifier in idx[0] and moves * other values up one slot. Normalize to SM5.1. */ @@ -1908,8 +1939,6 @@ static bool shader_sm4_read_param(struct vkd3d_shader_sm4_parser *priv, const ui ++param->idx_count; } - map_register(priv, param); - return true; } @@ -1938,6 +1967,16 @@ static uint32_t swizzle_from_sm4(uint32_t s) return vkd3d_shader_create_swizzle(s & 0x3, (s >> 2) & 0x3, (s >> 4) & 0x3, (s >> 6) & 0x3); } +static uint32_t swizzle_to_sm4(uint32_t s) +{ + uint32_t ret = 0; + ret |= ((vkd3d_swizzle_get_component(s, 0)) & 0x3); + ret |= ((vkd3d_swizzle_get_component(s, 1)) & 0x3) << 2; + ret |= ((vkd3d_swizzle_get_component(s, 2)) & 0x3) << 4; + ret |= ((vkd3d_swizzle_get_component(s, 3)) & 0x3) << 6; + return ret; +} + static bool register_is_input_output(const struct vkd3d_shader_register *reg) { switch (reg->type) @@ -2319,7 +2358,7 @@ static void shader_sm4_read_instruction(struct vkd3d_shader_sm4_parser *sm4, str } --len; - if (!(opcode_info = get_opcode_info(opcode))) + if (!(opcode_info = get_info_from_sm4_opcode(&sm4->lookup, opcode))) { FIXME("Unrecognized opcode %#x, opcode_token 0x%08x.\n", opcode, opcode_token); ins->handler_idx = VKD3DSIH_INVALID; @@ -2327,7 +2366,7 @@ static void shader_sm4_read_instruction(struct vkd3d_shader_sm4_parser *sm4, str return; } - ins->handler_idx = opcode_info->handler_idx; + vsir_instruction_init(ins, &sm4->p.location, opcode_info->handler_idx); if (ins->handler_idx == VKD3DSIH_HS_CONTROL_POINT_PHASE || ins->handler_idx == VKD3DSIH_HS_FORK_PHASE || ins->handler_idx == VKD3DSIH_HS_JOIN_PHASE) sm4->phase = ins->handler_idx; @@ -2430,7 +2469,6 @@ static bool shader_sm4_init(struct vkd3d_shader_sm4_parser *sm4, const uint32_t { struct vkd3d_shader_version version; uint32_t version_token, token_count; - unsigned int i; if (byte_code_size / sizeof(*byte_code) < 2) { @@ -2490,23 +2528,6 @@ static bool shader_sm4_init(struct vkd3d_shader_sm4_parser *sm4, const uint32_t return false; sm4->ptr = sm4->start; - memset(sm4->output_map, 0xff, sizeof(sm4->output_map)); - for (i = 0; i < output_signature->element_count; ++i) - { - struct signature_element *e = &output_signature->elements[i]; - - if (version.type == VKD3D_SHADER_TYPE_PIXEL - && ascii_strcasecmp(e->semantic_name, "SV_Target")) - continue; - if (e->register_index >= ARRAY_SIZE(sm4->output_map)) - { - WARN("Invalid output index %u.\n", e->register_index); - continue; - } - - sm4->output_map[e->register_index] = e->semantic_index; - } - init_sm4_lookup_tables(&sm4->lookup); return true; @@ -2642,9 +2663,19 @@ int vkd3d_shader_sm4_parser_create(const struct vkd3d_shader_compile_info *compi if (sm4->p.shader_version.type == VKD3D_SHADER_TYPE_HULL && !sm4->has_control_point_phase && !sm4->p.failed) shader_sm4_validate_default_phase_index_ranges(sm4); + if (!sm4->p.failed) + vsir_validate(&sm4->p); + + if (sm4->p.failed) + { + WARN("Failed to parse shader.\n"); + shader_sm4_destroy(&sm4->p); + return VKD3D_ERROR_INVALID_SHADER; + } + *parser = &sm4->p; - return sm4->p.failed ? VKD3D_ERROR_INVALID_SHADER : VKD3D_OK; + return VKD3D_OK; } static void write_sm4_block(const struct tpf_writer *tpf, const struct hlsl_block *block); @@ -2664,7 +2695,7 @@ static bool type_is_integer(const struct hlsl_type *type) } bool hlsl_sm4_register_from_semantic(struct hlsl_ctx *ctx, const struct hlsl_semantic *semantic, - bool output, enum vkd3d_shader_register_type *type, enum vkd3d_sm4_swizzle_type *swizzle_type, bool *has_idx) + bool output, enum vkd3d_shader_register_type *type, bool *has_idx) { unsigned int i; @@ -2673,25 +2704,24 @@ bool hlsl_sm4_register_from_semantic(struct hlsl_ctx *ctx, const struct hlsl_sem const char *semantic; bool output; enum vkd3d_shader_type shader_type; - enum vkd3d_sm4_swizzle_type swizzle_type; enum vkd3d_shader_register_type type; bool has_idx; } register_table[] = { - {"sv_dispatchthreadid", false, VKD3D_SHADER_TYPE_COMPUTE, VKD3D_SM4_SWIZZLE_VEC4, VKD3DSPR_THREADID, false}, - {"sv_groupid", false, VKD3D_SHADER_TYPE_COMPUTE, VKD3D_SM4_SWIZZLE_VEC4, VKD3DSPR_THREADGROUPID, false}, - {"sv_groupthreadid", false, VKD3D_SHADER_TYPE_COMPUTE, VKD3D_SM4_SWIZZLE_VEC4, VKD3DSPR_LOCALTHREADID, false}, + {"sv_dispatchthreadid", false, VKD3D_SHADER_TYPE_COMPUTE, VKD3DSPR_THREADID, false}, + {"sv_groupid", false, VKD3D_SHADER_TYPE_COMPUTE, VKD3DSPR_THREADGROUPID, false}, + {"sv_groupthreadid", false, VKD3D_SHADER_TYPE_COMPUTE, VKD3DSPR_LOCALTHREADID, false}, - {"sv_primitiveid", false, VKD3D_SHADER_TYPE_GEOMETRY, VKD3D_SM4_SWIZZLE_NONE, VKD3DSPR_PRIMID, false}, + {"sv_primitiveid", false, VKD3D_SHADER_TYPE_GEOMETRY, VKD3DSPR_PRIMID, false}, /* Put sv_target in this table, instead of letting it fall through to * default varying allocation, so that the register index matches the * usage index. */ - {"color", true, VKD3D_SHADER_TYPE_PIXEL, VKD3D_SM4_SWIZZLE_VEC4, VKD3DSPR_OUTPUT, true}, - {"depth", true, VKD3D_SHADER_TYPE_PIXEL, VKD3D_SM4_SWIZZLE_VEC4, VKD3DSPR_DEPTHOUT, false}, - {"sv_depth", true, VKD3D_SHADER_TYPE_PIXEL, VKD3D_SM4_SWIZZLE_VEC4, VKD3DSPR_DEPTHOUT, false}, - {"sv_target", true, VKD3D_SHADER_TYPE_PIXEL, VKD3D_SM4_SWIZZLE_VEC4, VKD3DSPR_OUTPUT, true}, + {"color", true, VKD3D_SHADER_TYPE_PIXEL, VKD3DSPR_OUTPUT, true}, + {"depth", true, VKD3D_SHADER_TYPE_PIXEL, VKD3DSPR_DEPTHOUT, false}, + {"sv_depth", true, VKD3D_SHADER_TYPE_PIXEL, VKD3DSPR_DEPTHOUT, false}, + {"sv_target", true, VKD3D_SHADER_TYPE_PIXEL, VKD3DSPR_OUTPUT, true}, }; for (i = 0; i < ARRAY_SIZE(register_table); ++i) @@ -2702,8 +2732,6 @@ bool hlsl_sm4_register_from_semantic(struct hlsl_ctx *ctx, const struct hlsl_sem { if (type) *type = register_table[i].type; - if (swizzle_type) - *swizzle_type = register_table[i].swizzle_type; *has_idx = register_table[i].has_idx; return true; } @@ -2722,7 +2750,7 @@ bool hlsl_sm4_usage_from_semantic(struct hlsl_ctx *ctx, const struct hlsl_semant const char *name; bool output; enum vkd3d_shader_type shader_type; - D3DDECLUSAGE usage; + D3D_NAME usage; } semantics[] = { @@ -2753,20 +2781,21 @@ bool hlsl_sm4_usage_from_semantic(struct hlsl_ctx *ctx, const struct hlsl_semant {"position", true, VKD3D_SHADER_TYPE_VERTEX, D3D_NAME_POSITION}, {"sv_position", true, VKD3D_SHADER_TYPE_VERTEX, D3D_NAME_POSITION}, }; + bool needs_compat_mapping = ascii_strncasecmp(semantic->name, "sv_", 3); for (i = 0; i < ARRAY_SIZE(semantics); ++i) { if (!ascii_strcasecmp(semantic->name, semantics[i].name) && output == semantics[i].output - && ctx->profile->type == semantics[i].shader_type - && !ascii_strncasecmp(semantic->name, "sv_", 3)) + && (ctx->semantic_compat_mapping == needs_compat_mapping || !needs_compat_mapping) + && ctx->profile->type == semantics[i].shader_type) { *usage = semantics[i].usage; return true; } } - if (!ascii_strncasecmp(semantic->name, "sv_", 3)) + if (!needs_compat_mapping) return false; *usage = D3D_NAME_UNDEFINED; @@ -2815,7 +2844,7 @@ static void write_sm4_signature(struct hlsl_ctx *ctx, struct dxbc_writer *dxbc, continue; usage_idx = var->semantic.index; - if (hlsl_sm4_register_from_semantic(ctx, &var->semantic, output, NULL, NULL, &has_idx)) + if (hlsl_sm4_register_from_semantic(ctx, &var->semantic, output, NULL, &has_idx)) { reg_idx = has_idx ? var->semantic.index : ~0u; } @@ -3202,8 +3231,7 @@ static struct extern_resource *sm4_get_extern_resources(struct hlsl_ctx *ctx, un if (!hlsl_type_is_resource(component_type)) continue; - regset = hlsl_type_get_regset(component_type); - regset_offset = hlsl_type_get_component_offset(ctx, var->data_type, regset, k); + regset_offset = hlsl_type_get_component_offset(ctx, var->data_type, k, ®set); if (regset_offset > var->regs[regset].allocation_size) continue; @@ -3249,38 +3277,43 @@ static struct extern_resource *sm4_get_extern_resources(struct hlsl_ctx *ctx, un } else { + unsigned int r; + if (!hlsl_type_is_resource(var->data_type)) continue; - regset = hlsl_type_get_regset(var->data_type); - if (!var->regs[regset].allocated) - continue; - if (!(hlsl_array_reserve(ctx, (void **)&extern_resources, &capacity, *count + 1, - sizeof(*extern_resources)))) + for (r = 0; r <= HLSL_REGSET_LAST; ++r) { - sm4_free_extern_resources(extern_resources, *count); - *count = 0; - return NULL; - } + if (!var->regs[r].allocated) + continue; - if (!(name = hlsl_strdup(ctx, string_skip_tag(var->name)))) - { - sm4_free_extern_resources(extern_resources, *count); - *count = 0; - return NULL; - } + if (!(hlsl_array_reserve(ctx, (void **)&extern_resources, &capacity, *count + 1, + sizeof(*extern_resources)))) + { + sm4_free_extern_resources(extern_resources, *count); + *count = 0; + return NULL; + } + + if (!(name = hlsl_strdup(ctx, string_skip_tag(var->name)))) + { + sm4_free_extern_resources(extern_resources, *count); + *count = 0; + return NULL; + } - extern_resources[*count].var = var; + extern_resources[*count].var = var; - extern_resources[*count].name = name; - extern_resources[*count].data_type = var->data_type; - extern_resources[*count].is_user_packed = !!var->reg_reservation.reg_type; + extern_resources[*count].name = name; + extern_resources[*count].data_type = var->data_type; + extern_resources[*count].is_user_packed = !!var->reg_reservation.reg_type; - extern_resources[*count].regset = regset; - extern_resources[*count].id = var->regs[regset].id; - extern_resources[*count].bind_count = var->bind_count[regset]; + extern_resources[*count].regset = r; + extern_resources[*count].id = var->regs[r].id; + extern_resources[*count].bind_count = var->bind_count[r]; - ++*count; + ++*count; + } } } @@ -3427,7 +3460,8 @@ static void write_sm4_rdef(struct hlsl_ctx *ctx, struct dxbc_writer *dxbc) LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry) { - if (var->is_uniform && var->buffer == cbuffer) + if (var->is_uniform && var->buffer == cbuffer + && var->data_type->class != HLSL_CLASS_OBJECT) ++var_count; } @@ -3461,7 +3495,8 @@ static void write_sm4_rdef(struct hlsl_ctx *ctx, struct dxbc_writer *dxbc) LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry) { - if (var->is_uniform && var->buffer == cbuffer) + if (var->is_uniform && var->buffer == cbuffer + && var->data_type->class != HLSL_CLASS_OBJECT) { uint32_t flags = 0; @@ -3488,7 +3523,8 @@ static void write_sm4_rdef(struct hlsl_ctx *ctx, struct dxbc_writer *dxbc) j = 0; LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry) { - if (var->is_uniform && var->buffer == cbuffer) + if (var->is_uniform && var->buffer == cbuffer + && var->data_type->class != HLSL_CLASS_OBJECT) { const unsigned int var_size = (profile->major_version >= 5 ? 10 : 6); size_t var_offset = vars_start + j * var_size * sizeof(uint32_t); @@ -3577,61 +3613,100 @@ static uint32_t sm4_encode_instruction_modifier(const struct sm4_instruction_mod return word; } -struct sm4_register -{ - enum vkd3d_shader_register_type type; - struct vkd3d_shader_register_index idx[2]; - unsigned int idx_count; - enum vkd3d_sm4_dimension dim; - uint32_t immconst_uint[4]; - unsigned int mod; -}; - struct sm4_instruction { enum vkd3d_sm4_opcode opcode; + uint32_t extra_bits; struct sm4_instruction_modifier modifiers[1]; unsigned int modifier_count; - struct sm4_dst_register - { - struct sm4_register reg; - unsigned int writemask; - } dsts[2]; + struct vkd3d_shader_dst_param dsts[2]; unsigned int dst_count; - struct sm4_src_register - { - struct sm4_register reg; - enum vkd3d_sm4_swizzle_type swizzle_type; - unsigned int swizzle; - } srcs[5]; + struct vkd3d_shader_src_param srcs[5]; unsigned int src_count; unsigned int byte_stride; uint32_t idx[3]; unsigned int idx_count; + + struct vkd3d_shader_src_param idx_srcs[7]; + unsigned int idx_src_count; }; -static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct sm4_register *reg, - unsigned int *writemask, enum vkd3d_sm4_swizzle_type *swizzle_type, - const struct hlsl_deref *deref) +static void sm4_register_from_node(struct vkd3d_shader_register *reg, uint32_t *writemask, + const struct hlsl_ir_node *instr) +{ + assert(instr->reg.allocated); + reg->type = VKD3DSPR_TEMP; + reg->dimension = VSIR_DIMENSION_VEC4; + reg->idx[0].offset = instr->reg.id; + reg->idx_count = 1; + *writemask = instr->reg.writemask; +} + +static void sm4_numeric_register_from_deref(struct hlsl_ctx *ctx, struct vkd3d_shader_register *reg, + enum vkd3d_shader_register_type type, uint32_t *writemask, const struct hlsl_deref *deref, + struct sm4_instruction *sm4_instr) +{ + const struct hlsl_ir_var *var = deref->var; + unsigned int offset_const_deref; + + reg->type = type; + reg->idx[0].offset = var->regs[HLSL_REGSET_NUMERIC].id; + reg->dimension = VSIR_DIMENSION_VEC4; + + assert(var->regs[HLSL_REGSET_NUMERIC].allocated); + + if (!var->indexable) + { + offset_const_deref = hlsl_offset_from_deref_safe(ctx, deref); + reg->idx[0].offset += offset_const_deref / 4; + reg->idx_count = 1; + } + else + { + offset_const_deref = deref->const_offset; + reg->idx[1].offset = offset_const_deref / 4; + reg->idx_count = 2; + + if (deref->rel_offset.node) + { + struct vkd3d_shader_src_param *idx_src; + unsigned int idx_writemask; + + assert(sm4_instr->idx_src_count < ARRAY_SIZE(sm4_instr->idx_srcs)); + idx_src = &sm4_instr->idx_srcs[sm4_instr->idx_src_count++]; + memset(idx_src, 0, sizeof(*idx_src)); + + reg->idx[1].rel_addr = idx_src; + sm4_register_from_node(&idx_src->reg, &idx_writemask, deref->rel_offset.node); + assert(idx_writemask != 0); + idx_src->swizzle = swizzle_from_sm4(hlsl_swizzle_from_writemask(idx_writemask)); + } + } + + *writemask = 0xf & (0xf << (offset_const_deref % 4)); + if (var->regs[HLSL_REGSET_NUMERIC].writemask) + *writemask = hlsl_combine_writemasks(var->regs[HLSL_REGSET_NUMERIC].writemask, *writemask); +} + +static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct vkd3d_shader_register *reg, + uint32_t *writemask, const struct hlsl_deref *deref, struct sm4_instruction *sm4_instr) { const struct hlsl_type *data_type = hlsl_deref_get_type(ctx, deref); const struct hlsl_ir_var *var = deref->var; if (var->is_uniform) { - enum hlsl_regset regset = hlsl_type_get_regset(data_type); + enum hlsl_regset regset = hlsl_deref_get_regset(ctx, deref); if (regset == HLSL_REGSET_TEXTURES) { reg->type = VKD3DSPR_RESOURCE; - reg->dim = VKD3D_SM4_DIMENSION_VEC4; - if (swizzle_type) - *swizzle_type = VKD3D_SM4_SWIZZLE_VEC4; + reg->dimension = VSIR_DIMENSION_VEC4; reg->idx[0].offset = var->regs[HLSL_REGSET_TEXTURES].id; reg->idx[0].offset += hlsl_offset_from_deref_safe(ctx, deref); assert(regset == HLSL_REGSET_TEXTURES); @@ -3641,9 +3716,7 @@ static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct sm4_register *r else if (regset == HLSL_REGSET_UAVS) { reg->type = VKD3DSPR_UAV; - reg->dim = VKD3D_SM4_DIMENSION_VEC4; - if (swizzle_type) - *swizzle_type = VKD3D_SM4_SWIZZLE_VEC4; + reg->dimension = VSIR_DIMENSION_VEC4; reg->idx[0].offset = var->regs[HLSL_REGSET_UAVS].id; reg->idx[0].offset += hlsl_offset_from_deref_safe(ctx, deref); assert(regset == HLSL_REGSET_UAVS); @@ -3653,9 +3726,7 @@ static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct sm4_register *r else if (regset == HLSL_REGSET_SAMPLERS) { reg->type = VKD3DSPR_SAMPLER; - reg->dim = VKD3D_SM4_DIMENSION_NONE; - if (swizzle_type) - *swizzle_type = VKD3D_SM4_SWIZZLE_NONE; + reg->dimension = VSIR_DIMENSION_NONE; reg->idx[0].offset = var->regs[HLSL_REGSET_SAMPLERS].id; reg->idx[0].offset += hlsl_offset_from_deref_safe(ctx, deref); assert(regset == HLSL_REGSET_SAMPLERS); @@ -3668,9 +3739,7 @@ static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct sm4_register *r assert(data_type->class <= HLSL_CLASS_VECTOR); reg->type = VKD3DSPR_CONSTBUFFER; - reg->dim = VKD3D_SM4_DIMENSION_VEC4; - if (swizzle_type) - *swizzle_type = VKD3D_SM4_SWIZZLE_VEC4; + reg->dimension = VSIR_DIMENSION_VEC4; reg->idx[0].offset = var->buffer->reg.id; reg->idx[1].offset = offset / 4; reg->idx_count = 2; @@ -3681,7 +3750,7 @@ static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct sm4_register *r { bool has_idx; - if (hlsl_sm4_register_from_semantic(ctx, &var->semantic, false, ®->type, swizzle_type, &has_idx)) + if (hlsl_sm4_register_from_semantic(ctx, &var->semantic, false, ®->type, &has_idx)) { unsigned int offset = hlsl_offset_from_deref_safe(ctx, deref); @@ -3691,7 +3760,7 @@ static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct sm4_register *r reg->idx_count = 1; } - reg->dim = VKD3D_SM4_DIMENSION_VEC4; + reg->dimension = VSIR_DIMENSION_VEC4; *writemask = ((1u << data_type->dimx) - 1) << (offset % 4); } else @@ -3700,9 +3769,7 @@ static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct sm4_register *r assert(hlsl_reg.allocated); reg->type = VKD3DSPR_INPUT; - reg->dim = VKD3D_SM4_DIMENSION_VEC4; - if (swizzle_type) - *swizzle_type = VKD3D_SM4_SWIZZLE_VEC4; + reg->dimension = VSIR_DIMENSION_VEC4; reg->idx[0].offset = hlsl_reg.id; reg->idx_count = 1; *writemask = hlsl_reg.writemask; @@ -3712,7 +3779,7 @@ static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct sm4_register *r { bool has_idx; - if (hlsl_sm4_register_from_semantic(ctx, &var->semantic, true, ®->type, swizzle_type, &has_idx)) + if (hlsl_sm4_register_from_semantic(ctx, &var->semantic, true, ®->type, &has_idx)) { unsigned int offset = hlsl_offset_from_deref_safe(ctx, deref); @@ -3723,9 +3790,9 @@ static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct sm4_register *r } if (reg->type == VKD3DSPR_DEPTHOUT) - reg->dim = VKD3D_SM4_DIMENSION_SCALAR; + reg->dimension = VSIR_DIMENSION_SCALAR; else - reg->dim = VKD3D_SM4_DIMENSION_VEC4; + reg->dimension = VSIR_DIMENSION_VEC4; *writemask = ((1u << data_type->dimx) - 1) << (offset % 4); } else @@ -3734,7 +3801,7 @@ static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct sm4_register *r assert(hlsl_reg.allocated); reg->type = VKD3DSPR_OUTPUT; - reg->dim = VKD3D_SM4_DIMENSION_VEC4; + reg->dimension = VSIR_DIMENSION_VEC4; reg->idx[0].offset = hlsl_reg.id; reg->idx_count = 1; *writemask = hlsl_reg.writemask; @@ -3742,77 +3809,61 @@ static void sm4_register_from_deref(struct hlsl_ctx *ctx, struct sm4_register *r } else { - struct hlsl_reg hlsl_reg = hlsl_reg_from_deref(ctx, deref); + enum vkd3d_shader_register_type type = deref->var->indexable ? VKD3DSPR_IDXTEMP : VKD3DSPR_TEMP; - assert(hlsl_reg.allocated); - reg->type = VKD3DSPR_TEMP; - reg->dim = VKD3D_SM4_DIMENSION_VEC4; - if (swizzle_type) - *swizzle_type = VKD3D_SM4_SWIZZLE_VEC4; - reg->idx[0].offset = hlsl_reg.id; - reg->idx_count = 1; - *writemask = hlsl_reg.writemask; + sm4_numeric_register_from_deref(ctx, reg, type, writemask, deref, sm4_instr); } } -static void sm4_src_from_deref(struct hlsl_ctx *ctx, struct sm4_src_register *src, - const struct hlsl_deref *deref, unsigned int map_writemask) +static void sm4_src_from_deref(const struct tpf_writer *tpf, struct vkd3d_shader_src_param *src, + const struct hlsl_deref *deref, unsigned int map_writemask, struct sm4_instruction *sm4_instr) { - unsigned int writemask; + unsigned int hlsl_swizzle; + uint32_t writemask; - sm4_register_from_deref(ctx, &src->reg, &writemask, &src->swizzle_type, deref); - if (src->swizzle_type == VKD3D_SM4_SWIZZLE_VEC4) - src->swizzle = hlsl_map_swizzle(hlsl_swizzle_from_writemask(writemask), map_writemask); -} - -static void sm4_register_from_node(struct sm4_register *reg, unsigned int *writemask, - enum vkd3d_sm4_swizzle_type *swizzle_type, const struct hlsl_ir_node *instr) -{ - assert(instr->reg.allocated); - reg->type = VKD3DSPR_TEMP; - reg->dim = VKD3D_SM4_DIMENSION_VEC4; - *swizzle_type = VKD3D_SM4_SWIZZLE_VEC4; - reg->idx[0].offset = instr->reg.id; - reg->idx_count = 1; - *writemask = instr->reg.writemask; + sm4_register_from_deref(tpf->ctx, &src->reg, &writemask, deref, sm4_instr); + if (vkd3d_sm4_get_default_swizzle_type(&tpf->lookup, src->reg.type) == VKD3D_SM4_SWIZZLE_VEC4) + { + hlsl_swizzle = hlsl_map_swizzle(hlsl_swizzle_from_writemask(writemask), map_writemask); + src->swizzle = swizzle_from_sm4(hlsl_swizzle); + } } -static void sm4_dst_from_node(struct sm4_dst_register *dst, const struct hlsl_ir_node *instr) +static void sm4_dst_from_node(struct vkd3d_shader_dst_param *dst, const struct hlsl_ir_node *instr) { - unsigned int swizzle_type; - - sm4_register_from_node(&dst->reg, &dst->writemask, &swizzle_type, instr); + sm4_register_from_node(&dst->reg, &dst->write_mask, instr); } -static void sm4_src_from_constant_value(struct sm4_src_register *src, +static void sm4_src_from_constant_value(struct vkd3d_shader_src_param *src, const struct hlsl_constant_value *value, unsigned int width, unsigned int map_writemask) { - src->swizzle_type = VKD3D_SM4_SWIZZLE_NONE; + src->swizzle = 0; src->reg.type = VKD3DSPR_IMMCONST; if (width == 1) { - src->reg.dim = VKD3D_SM4_DIMENSION_SCALAR; - src->reg.immconst_uint[0] = value->u[0].u; + src->reg.dimension = VSIR_DIMENSION_SCALAR; + src->reg.u.immconst_uint[0] = value->u[0].u; } else { unsigned int i, j = 0; - src->reg.dim = VKD3D_SM4_DIMENSION_VEC4; + src->reg.dimension = VSIR_DIMENSION_VEC4; for (i = 0; i < 4; ++i) { if ((map_writemask & (1u << i)) && (j < width)) - src->reg.immconst_uint[i] = value->u[j++].u; + src->reg.u.immconst_uint[i] = value->u[j++].u; else - src->reg.immconst_uint[i] = 0; + src->reg.u.immconst_uint[i] = 0; } } } -static void sm4_src_from_node(struct sm4_src_register *src, - const struct hlsl_ir_node *instr, unsigned int map_writemask) +static void sm4_src_from_node(const struct tpf_writer *tpf, struct vkd3d_shader_src_param *src, + const struct hlsl_ir_node *instr, uint32_t map_writemask) { - unsigned int writemask; + unsigned int hlsl_swizzle; + uint32_t writemask; if (instr->type == HLSL_IR_CONSTANT) { @@ -3822,139 +3873,211 @@ static void sm4_src_from_node(struct sm4_src_register *src, return; } - sm4_register_from_node(&src->reg, &writemask, &src->swizzle_type, instr); - if (src->swizzle_type == VKD3D_SM4_SWIZZLE_VEC4) - src->swizzle = hlsl_map_swizzle(hlsl_swizzle_from_writemask(writemask), map_writemask); + sm4_register_from_node(&src->reg, &writemask, instr); + if (vkd3d_sm4_get_default_swizzle_type(&tpf->lookup, src->reg.type) == VKD3D_SM4_SWIZZLE_VEC4) + { + hlsl_swizzle = hlsl_map_swizzle(hlsl_swizzle_from_writemask(writemask), map_writemask); + src->swizzle = swizzle_from_sm4(hlsl_swizzle); + } } -static void sm4_write_dst_register(const struct tpf_writer *tpf, const struct sm4_dst_register *dst) +static unsigned int sm4_get_index_addressing_from_reg(const struct vkd3d_shader_register *reg, + unsigned int i) +{ + if (reg->idx[i].rel_addr) + { + if (reg->idx[i].offset == 0) + return VKD3D_SM4_ADDRESSING_RELATIVE; + else + return VKD3D_SM4_ADDRESSING_RELATIVE | VKD3D_SM4_ADDRESSING_OFFSET; + } + + return 0; +} + +static uint32_t sm4_encode_register(const struct tpf_writer *tpf, const struct vkd3d_shader_register *reg, + enum vkd3d_sm4_swizzle_type sm4_swizzle_type, uint32_t sm4_swizzle) { const struct vkd3d_sm4_register_type_info *register_type_info; - struct vkd3d_bytecode_buffer *buffer = tpf->buffer; - uint32_t sm4_reg_type, reg_dim; + uint32_t sm4_reg_type, sm4_reg_dim; uint32_t token = 0; - unsigned int j; - register_type_info = get_info_from_vkd3d_register_type(&tpf->lookup, dst->reg.type); + register_type_info = get_info_from_vkd3d_register_type(&tpf->lookup, reg->type); if (!register_type_info) { - FIXME("Unhandled vkd3d-shader register type %#x.\n", dst->reg.type); + FIXME("Unhandled vkd3d-shader register type %#x.\n", reg->type); sm4_reg_type = VKD3D_SM4_RT_TEMP; + if (sm4_swizzle_type == VKD3D_SM4_SWIZZLE_DEFAULT) + sm4_swizzle_type = VKD3D_SM4_SWIZZLE_VEC4; } else { sm4_reg_type = register_type_info->sm4_type; + if (sm4_swizzle_type == VKD3D_SM4_SWIZZLE_DEFAULT) + sm4_swizzle_type = register_type_info->default_src_swizzle_type; } - - reg_dim = dst->reg.dim; + sm4_reg_dim = sm4_dimension_from_vsir_dimension(reg->dimension); token |= sm4_reg_type << VKD3D_SM4_REGISTER_TYPE_SHIFT; - token |= dst->reg.idx_count << VKD3D_SM4_REGISTER_ORDER_SHIFT; - token |= reg_dim << VKD3D_SM4_DIMENSION_SHIFT; - if (reg_dim == VKD3D_SM4_DIMENSION_VEC4) - token |= dst->writemask << VKD3D_SM4_WRITEMASK_SHIFT; - put_u32(buffer, token); + token |= reg->idx_count << VKD3D_SM4_REGISTER_ORDER_SHIFT; + token |= sm4_reg_dim << VKD3D_SM4_DIMENSION_SHIFT; + if (reg->idx_count > 0) + token |= sm4_get_index_addressing_from_reg(reg, 0) << VKD3D_SM4_ADDRESSING_SHIFT0; + if (reg->idx_count > 1) + token |= sm4_get_index_addressing_from_reg(reg, 1) << VKD3D_SM4_ADDRESSING_SHIFT1; + if (reg->idx_count > 2) + token |= sm4_get_index_addressing_from_reg(reg, 2) << VKD3D_SM4_ADDRESSING_SHIFT2; - for (j = 0; j < dst->reg.idx_count; ++j) + if (sm4_reg_dim == VKD3D_SM4_DIMENSION_VEC4) { - put_u32(buffer, dst->reg.idx[j].offset); - assert(!dst->reg.idx[j].rel_addr); + token |= (uint32_t)sm4_swizzle_type << VKD3D_SM4_SWIZZLE_TYPE_SHIFT; + + switch (sm4_swizzle_type) + { + case VKD3D_SM4_SWIZZLE_NONE: + assert(sm4_swizzle || register_is_constant(reg)); + token |= (sm4_swizzle << VKD3D_SM4_WRITEMASK_SHIFT) & VKD3D_SM4_WRITEMASK_MASK; + break; + + case VKD3D_SM4_SWIZZLE_VEC4: + token |= (sm4_swizzle << VKD3D_SM4_SWIZZLE_SHIFT) & VKD3D_SM4_SWIZZLE_MASK; + break; + + case VKD3D_SM4_SWIZZLE_SCALAR: + token |= (sm4_swizzle << VKD3D_SM4_SCALAR_DIM_SHIFT) & VKD3D_SM4_SCALAR_DIM_MASK; + break; + + default: + vkd3d_unreachable(); + } } + + return token; } -static void sm4_write_src_register(const struct tpf_writer *tpf, const struct sm4_src_register *src) +static void sm4_write_register_index(const struct tpf_writer *tpf, const struct vkd3d_shader_register *reg, + unsigned int j) { - const struct vkd3d_sm4_register_type_info *register_type_info; + unsigned int addressing = sm4_get_index_addressing_from_reg(reg, j); struct vkd3d_bytecode_buffer *buffer = tpf->buffer; - uint32_t sm4_reg_type, reg_dim; - uint32_t token = 0; - unsigned int j; + unsigned int k; - register_type_info = get_info_from_vkd3d_register_type(&tpf->lookup, src->reg.type); - if (!register_type_info) + if (addressing & VKD3D_SM4_ADDRESSING_RELATIVE) { - FIXME("Unhandled vkd3d-shader register type %#x.\n", src->reg.type); - sm4_reg_type = VKD3D_SM4_RT_TEMP; + const struct vkd3d_shader_src_param *idx_src = reg->idx[j].rel_addr; + uint32_t idx_src_token; + + assert(idx_src); + assert(!idx_src->modifiers); + assert(idx_src->reg.type != VKD3DSPR_IMMCONST); + idx_src_token = sm4_encode_register(tpf, &idx_src->reg, VKD3D_SM4_SWIZZLE_SCALAR, idx_src->swizzle); + + put_u32(buffer, idx_src_token); + for (k = 0; k < idx_src->reg.idx_count; ++k) + { + put_u32(buffer, idx_src->reg.idx[k].offset); + assert(!idx_src->reg.idx[k].rel_addr); + } } else { - sm4_reg_type = register_type_info->sm4_type; + put_u32(tpf->buffer, reg->idx[j].offset); } +} - reg_dim = src->reg.dim; +static void sm4_write_dst_register(const struct tpf_writer *tpf, const struct vkd3d_shader_dst_param *dst) +{ + struct vkd3d_bytecode_buffer *buffer = tpf->buffer; + uint32_t token = 0; + unsigned int j; - token |= sm4_reg_type << VKD3D_SM4_REGISTER_TYPE_SHIFT; - token |= src->reg.idx_count << VKD3D_SM4_REGISTER_ORDER_SHIFT; - token |= reg_dim << VKD3D_SM4_DIMENSION_SHIFT; - if (reg_dim == VKD3D_SM4_DIMENSION_VEC4) - { - token |= (uint32_t)src->swizzle_type << VKD3D_SM4_SWIZZLE_TYPE_SHIFT; - token |= src->swizzle << VKD3D_SM4_SWIZZLE_SHIFT; - } - if (src->reg.mod) - token |= VKD3D_SM4_EXTENDED_OPERAND; + token = sm4_encode_register(tpf, &dst->reg, VKD3D_SM4_SWIZZLE_NONE, dst->write_mask); put_u32(buffer, token); - if (src->reg.mod) - put_u32(buffer, (src->reg.mod << VKD3D_SM4_REGISTER_MODIFIER_SHIFT) - | VKD3D_SM4_EXTENDED_OPERAND_MODIFIER); + for (j = 0; j < dst->reg.idx_count; ++j) + sm4_write_register_index(tpf, &dst->reg, j); +} - for (j = 0; j < src->reg.idx_count; ++j) +static void sm4_write_src_register(const struct tpf_writer *tpf, const struct vkd3d_shader_src_param *src) +{ + struct vkd3d_bytecode_buffer *buffer = tpf->buffer; + uint32_t token = 0, mod_token = 0; + unsigned int j; + + token = sm4_encode_register(tpf, &src->reg, VKD3D_SM4_SWIZZLE_DEFAULT, swizzle_to_sm4(src->swizzle)); + + switch (src->modifiers) { - put_u32(buffer, src->reg.idx[j].offset); - assert(!src->reg.idx[j].rel_addr); + case VKD3DSPSM_NONE: + mod_token = VKD3D_SM4_REGISTER_MODIFIER_NONE; + break; + + case VKD3DSPSM_ABS: + mod_token = (VKD3D_SM4_REGISTER_MODIFIER_ABS << VKD3D_SM4_REGISTER_MODIFIER_SHIFT) + | VKD3D_SM4_EXTENDED_OPERAND_MODIFIER; + break; + + case VKD3DSPSM_NEG: + mod_token = (VKD3D_SM4_REGISTER_MODIFIER_NEGATE << VKD3D_SM4_REGISTER_MODIFIER_SHIFT) + | VKD3D_SM4_EXTENDED_OPERAND_MODIFIER; + break; + + case VKD3DSPSM_ABSNEG: + mod_token = (VKD3D_SM4_REGISTER_MODIFIER_ABS_NEGATE << VKD3D_SM4_REGISTER_MODIFIER_SHIFT) + | VKD3D_SM4_EXTENDED_OPERAND_MODIFIER; + break; + + default: + ERR("Unhandled register modifier %#x.\n", src->modifiers); + vkd3d_unreachable(); + break; } + if (src->modifiers) + { + token |= VKD3D_SM4_EXTENDED_OPERAND; + put_u32(buffer, token); + put_u32(buffer, mod_token); + } + else + { + put_u32(buffer, token); + } + + for (j = 0; j < src->reg.idx_count; ++j) + sm4_write_register_index(tpf, &src->reg, j); + if (src->reg.type == VKD3DSPR_IMMCONST) { - put_u32(buffer, src->reg.immconst_uint[0]); - if (reg_dim == VKD3D_SM4_DIMENSION_VEC4) + put_u32(buffer, src->reg.u.immconst_uint[0]); + if (src->reg.dimension == VSIR_DIMENSION_VEC4) { - put_u32(buffer, src->reg.immconst_uint[1]); - put_u32(buffer, src->reg.immconst_uint[2]); - put_u32(buffer, src->reg.immconst_uint[3]); + put_u32(buffer, src->reg.u.immconst_uint[1]); + put_u32(buffer, src->reg.u.immconst_uint[2]); + put_u32(buffer, src->reg.u.immconst_uint[3]); } } } -static uint32_t sm4_register_order(const struct sm4_register *reg) -{ - uint32_t order = 1; - if (reg->type == VKD3DSPR_IMMCONST) - order += reg->dim == VKD3D_SM4_DIMENSION_VEC4 ? 4 : 1; - order += reg->idx_count; - if (reg->mod) - ++order; - return order; -} - static void write_sm4_instruction(const struct tpf_writer *tpf, const struct sm4_instruction *instr) { struct vkd3d_bytecode_buffer *buffer = tpf->buffer; - uint32_t token = instr->opcode; - unsigned int size = 1, i, j; - - size += instr->modifier_count; - for (i = 0; i < instr->dst_count; ++i) - size += sm4_register_order(&instr->dsts[i].reg); - for (i = 0; i < instr->src_count; ++i) - size += sm4_register_order(&instr->srcs[i].reg); - size += instr->idx_count; - if (instr->byte_stride) - ++size; - - token |= (size << VKD3D_SM4_INSTRUCTION_LENGTH_SHIFT); + uint32_t token = instr->opcode | instr->extra_bits; + unsigned int size, i, j; + size_t token_position; if (instr->modifier_count > 0) token |= VKD3D_SM4_INSTRUCTION_MODIFIER; - put_u32(buffer, token); + + token_position = put_u32(buffer, 0); for (i = 0; i < instr->modifier_count; ++i) { - token = sm4_encode_instruction_modifier(&instr->modifiers[i]); + uint32_t modifier_token = sm4_encode_instruction_modifier(&instr->modifiers[i]); + if (instr->modifier_count > i + 1) - token |= VKD3D_SM4_INSTRUCTION_MODIFIER; - put_u32(buffer, token); + modifier_token |= VKD3D_SM4_INSTRUCTION_MODIFIER; + put_u32(buffer, modifier_token); } for (i = 0; i < instr->dst_count; ++i) @@ -3968,6 +4091,10 @@ static void write_sm4_instruction(const struct tpf_writer *tpf, const struct sm4 for (j = 0; j < instr->idx_count; ++j) put_u32(buffer, instr->idx[j]); + + size = (bytecode_get_size(buffer) - token_position) / sizeof(uint32_t); + token |= (size << VKD3D_SM4_INSTRUCTION_LENGTH_SHIFT); + set_u32(buffer, token_position, token); } static bool encode_texel_offset_as_aoffimmi(struct sm4_instruction *instr, @@ -4003,13 +4130,12 @@ static void write_sm4_dcl_constant_buffer(const struct tpf_writer *tpf, const st { .opcode = VKD3D_SM4_OP_DCL_CONSTANT_BUFFER, - .srcs[0].reg.dim = VKD3D_SM4_DIMENSION_VEC4, + .srcs[0].reg.dimension = VSIR_DIMENSION_VEC4, .srcs[0].reg.type = VKD3DSPR_CONSTBUFFER, .srcs[0].reg.idx[0].offset = cbuffer->reg.id, .srcs[0].reg.idx[1].offset = (cbuffer->used_size + 3) / 4, .srcs[0].reg.idx_count = 2, - .srcs[0].swizzle_type = VKD3D_SM4_SWIZZLE_VEC4, - .srcs[0].swizzle = HLSL_SWIZZLE(X, Y, Z, W), + .srcs[0].swizzle = VKD3D_SHADER_NO_SWIZZLE, .src_count = 1, }; write_sm4_instruction(tpf, &instr); @@ -4031,7 +4157,7 @@ static void write_sm4_dcl_samplers(const struct tpf_writer *tpf, const struct ex component_type = hlsl_type_get_component_type(tpf->ctx, resource->data_type, 0); if (component_type->sampler_dim == HLSL_SAMPLER_DIM_COMPARISON) - instr.opcode |= VKD3D_SM4_SAMPLER_COMPARISON << VKD3D_SM4_SAMPLER_MODE_SHIFT; + instr.extra_bits |= VKD3D_SM4_SAMPLER_COMPARISON << VKD3D_SM4_SAMPLER_MODE_SHIFT; assert(resource->regset == HLSL_REGSET_SAMPLERS); @@ -4090,12 +4216,12 @@ static void write_sm4_dcl_textures(const struct tpf_writer *tpf, const struct ex { instr.opcode = VKD3D_SM4_OP_DCL_RESOURCE; } - instr.opcode |= (sm4_resource_dimension(component_type) << VKD3D_SM4_RESOURCE_TYPE_SHIFT); + instr.extra_bits |= (sm4_resource_dimension(component_type) << VKD3D_SM4_RESOURCE_TYPE_SHIFT); if (component_type->sampler_dim == HLSL_SAMPLER_DIM_2DMS || component_type->sampler_dim == HLSL_SAMPLER_DIM_2DMSARRAY) { - instr.opcode |= component_type->sample_count << VKD3D_SM4_RESOURCE_SAMPLE_COUNT_SHIFT; + instr.extra_bits |= component_type->sample_count << VKD3D_SM4_RESOURCE_SAMPLE_COUNT_SHIFT; } write_sm4_instruction(tpf, &instr); @@ -4111,11 +4237,11 @@ static void write_sm4_dcl_semantic(const struct tpf_writer *tpf, const struct hl struct sm4_instruction instr = { - .dsts[0].reg.dim = VKD3D_SM4_DIMENSION_VEC4, + .dsts[0].reg.dimension = VSIR_DIMENSION_VEC4, .dst_count = 1, }; - if (hlsl_sm4_register_from_semantic(tpf->ctx, &var->semantic, output, &instr.dsts[0].reg.type, NULL, &has_idx)) + if (hlsl_sm4_register_from_semantic(tpf->ctx, &var->semantic, output, &instr.dsts[0].reg.type, &has_idx)) { if (has_idx) { @@ -4126,18 +4252,18 @@ static void write_sm4_dcl_semantic(const struct tpf_writer *tpf, const struct hl { instr.dsts[0].reg.idx_count = 0; } - instr.dsts[0].writemask = (1 << var->data_type->dimx) - 1; + instr.dsts[0].write_mask = (1 << var->data_type->dimx) - 1; } else { instr.dsts[0].reg.type = output ? VKD3DSPR_OUTPUT : VKD3DSPR_INPUT; instr.dsts[0].reg.idx[0].offset = var->regs[HLSL_REGSET_NUMERIC].id; instr.dsts[0].reg.idx_count = 1; - instr.dsts[0].writemask = var->regs[HLSL_REGSET_NUMERIC].writemask; + instr.dsts[0].write_mask = var->regs[HLSL_REGSET_NUMERIC].writemask; } if (instr.dsts[0].reg.type == VKD3DSPR_DEPTHOUT) - instr.dsts[0].reg.dim = VKD3D_SM4_DIMENSION_SCALAR; + instr.dsts[0].reg.dimension = VSIR_DIMENSION_SCALAR; hlsl_sm4_usage_from_semantic(tpf->ctx, &var->semantic, output, &usage); if (usage == ~0u) @@ -4170,9 +4296,35 @@ static void write_sm4_dcl_semantic(const struct tpf_writer *tpf, const struct hl enum vkd3d_shader_interpolation_mode mode = VKD3DSIM_LINEAR; if ((var->storage_modifiers & HLSL_STORAGE_NOINTERPOLATION) || type_is_integer(var->data_type)) + { mode = VKD3DSIM_CONSTANT; + } + else + { + static const struct + { + unsigned int modifiers; + enum vkd3d_shader_interpolation_mode mode; + } + modes[] = + { + { HLSL_STORAGE_CENTROID | HLSL_STORAGE_NOPERSPECTIVE, VKD3DSIM_LINEAR_NOPERSPECTIVE_CENTROID }, + { HLSL_STORAGE_NOPERSPECTIVE, VKD3DSIM_LINEAR_NOPERSPECTIVE }, + { HLSL_STORAGE_CENTROID, VKD3DSIM_LINEAR_CENTROID }, + }; + unsigned int i; - instr.opcode |= mode << VKD3D_SM4_INTERPOLATION_MODE_SHIFT; + for (i = 0; i < ARRAY_SIZE(modes); ++i) + { + if ((var->storage_modifiers & modes[i].modifiers) == modes[i].modifiers) + { + mode = modes[i].mode; + break; + } + } + } + + instr.extra_bits |= mode << VKD3D_SM4_INTERPOLATION_MODE_SHIFT; } } else @@ -4215,6 +4367,20 @@ static void write_sm4_dcl_temps(const struct tpf_writer *tpf, uint32_t temp_coun write_sm4_instruction(tpf, &instr); } +static void write_sm4_dcl_indexable_temp(const struct tpf_writer *tpf, uint32_t idx, + uint32_t size, uint32_t comp_count) +{ + struct sm4_instruction instr = + { + .opcode = VKD3D_SM4_OP_DCL_INDEXABLE_TEMP, + + .idx = {idx, size, comp_count}, + .idx_count = 3, + }; + + write_sm4_instruction(tpf, &instr); +} + static void write_sm4_dcl_thread_group(const struct tpf_writer *tpf, const uint32_t thread_count[3]) { struct sm4_instruction instr = @@ -4241,7 +4407,7 @@ static void write_sm4_ret(const struct tpf_writer *tpf) } static void write_sm4_unary_op(const struct tpf_writer *tpf, enum vkd3d_sm4_opcode opcode, - const struct hlsl_ir_node *dst, const struct hlsl_ir_node *src, unsigned int src_mod) + const struct hlsl_ir_node *dst, const struct hlsl_ir_node *src, enum vkd3d_shader_src_modifier src_mod) { struct sm4_instruction instr; @@ -4251,8 +4417,8 @@ static void write_sm4_unary_op(const struct tpf_writer *tpf, enum vkd3d_sm4_opco sm4_dst_from_node(&instr.dsts[0], dst); instr.dst_count = 1; - sm4_src_from_node(&instr.srcs[0], src, instr.dsts[0].writemask); - instr.srcs[0].reg.mod = src_mod; + sm4_src_from_node(tpf, &instr.srcs[0], src, instr.dsts[0].write_mask); + instr.srcs[0].modifiers = src_mod; instr.src_count = 1; write_sm4_instruction(tpf, &instr); @@ -4270,11 +4436,11 @@ static void write_sm4_unary_op_with_two_destinations(const struct tpf_writer *tp sm4_dst_from_node(&instr.dsts[dst_idx], dst); assert(1 - dst_idx >= 0); instr.dsts[1 - dst_idx].reg.type = VKD3DSPR_NULL; - instr.dsts[1 - dst_idx].reg.dim = VKD3D_SM4_DIMENSION_NONE; + instr.dsts[1 - dst_idx].reg.dimension = VSIR_DIMENSION_NONE; instr.dsts[1 - dst_idx].reg.idx_count = 0; instr.dst_count = 2; - sm4_src_from_node(&instr.srcs[0], src, instr.dsts[dst_idx].writemask); + sm4_src_from_node(tpf, &instr.srcs[0], src, instr.dsts[dst_idx].write_mask); instr.src_count = 1; write_sm4_instruction(tpf, &instr); @@ -4291,8 +4457,8 @@ static void write_sm4_binary_op(const struct tpf_writer *tpf, enum vkd3d_sm4_opc sm4_dst_from_node(&instr.dsts[0], dst); instr.dst_count = 1; - sm4_src_from_node(&instr.srcs[0], src1, instr.dsts[0].writemask); - sm4_src_from_node(&instr.srcs[1], src2, instr.dsts[0].writemask); + sm4_src_from_node(tpf, &instr.srcs[0], src1, instr.dsts[0].write_mask); + sm4_src_from_node(tpf, &instr.srcs[1], src2, instr.dsts[0].write_mask); instr.src_count = 2; write_sm4_instruction(tpf, &instr); @@ -4310,8 +4476,8 @@ static void write_sm4_binary_op_dot(const struct tpf_writer *tpf, enum vkd3d_sm4 sm4_dst_from_node(&instr.dsts[0], dst); instr.dst_count = 1; - sm4_src_from_node(&instr.srcs[0], src1, VKD3DSP_WRITEMASK_ALL); - sm4_src_from_node(&instr.srcs[1], src2, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[0], src1, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[1], src2, VKD3DSP_WRITEMASK_ALL); instr.src_count = 2; write_sm4_instruction(tpf, &instr); @@ -4330,12 +4496,12 @@ static void write_sm4_binary_op_with_two_destinations(const struct tpf_writer *t sm4_dst_from_node(&instr.dsts[dst_idx], dst); assert(1 - dst_idx >= 0); instr.dsts[1 - dst_idx].reg.type = VKD3DSPR_NULL; - instr.dsts[1 - dst_idx].reg.dim = VKD3D_SM4_DIMENSION_NONE; + instr.dsts[1 - dst_idx].reg.dimension = VSIR_DIMENSION_NONE; instr.dsts[1 - dst_idx].reg.idx_count = 0; instr.dst_count = 2; - sm4_src_from_node(&instr.srcs[0], src1, instr.dsts[dst_idx].writemask); - sm4_src_from_node(&instr.srcs[1], src2, instr.dsts[dst_idx].writemask); + sm4_src_from_node(tpf, &instr.srcs[0], src1, instr.dsts[dst_idx].write_mask); + sm4_src_from_node(tpf, &instr.srcs[1], src2, instr.dsts[dst_idx].write_mask); instr.src_count = 2; write_sm4_instruction(tpf, &instr); @@ -4353,9 +4519,9 @@ static void write_sm4_ternary_op(const struct tpf_writer *tpf, enum vkd3d_sm4_op sm4_dst_from_node(&instr.dsts[0], dst); instr.dst_count = 1; - sm4_src_from_node(&instr.srcs[0], src1, instr.dsts[0].writemask); - sm4_src_from_node(&instr.srcs[1], src2, instr.dsts[0].writemask); - sm4_src_from_node(&instr.srcs[2], src3, instr.dsts[0].writemask); + sm4_src_from_node(tpf, &instr.srcs[0], src1, instr.dsts[0].write_mask); + sm4_src_from_node(tpf, &instr.srcs[1], src2, instr.dsts[0].write_mask); + sm4_src_from_node(tpf, &instr.srcs[2], src3, instr.dsts[0].write_mask); instr.src_count = 3; write_sm4_instruction(tpf, &instr); @@ -4369,7 +4535,7 @@ static void write_sm4_ld(const struct tpf_writer *tpf, const struct hlsl_ir_node const struct hlsl_type *resource_type = hlsl_deref_get_type(tpf->ctx, resource); bool multisampled = resource_type->base_type == HLSL_TYPE_TEXTURE && (resource_type->sampler_dim == HLSL_SAMPLER_DIM_2DMS || resource_type->sampler_dim == HLSL_SAMPLER_DIM_2DMSARRAY); - bool uav = (hlsl_type_get_regset(resource_type) == HLSL_REGSET_UAVS); + bool uav = (hlsl_deref_get_regset(tpf->ctx, resource) == HLSL_REGSET_UAVS); unsigned int coords_writemask = VKD3DSP_WRITEMASK_ALL; struct sm4_instruction instr; @@ -4404,9 +4570,9 @@ static void write_sm4_ld(const struct tpf_writer *tpf, const struct hlsl_ir_node coords_writemask = VKD3DSP_WRITEMASK_0 | VKD3DSP_WRITEMASK_1 | VKD3DSP_WRITEMASK_3; } - sm4_src_from_node(&instr.srcs[0], coords, coords_writemask); + sm4_src_from_node(tpf, &instr.srcs[0], coords, coords_writemask); - sm4_src_from_deref(tpf->ctx, &instr.srcs[1], resource, instr.dsts[0].writemask); + sm4_src_from_deref(tpf, &instr.srcs[1], resource, instr.dsts[0].write_mask, &instr); instr.src_count = 2; @@ -4414,16 +4580,15 @@ static void write_sm4_ld(const struct tpf_writer *tpf, const struct hlsl_ir_node { if (sample_index->type == HLSL_IR_CONSTANT) { - struct sm4_register *reg = &instr.srcs[2].reg; + struct vkd3d_shader_register *reg = &instr.srcs[2].reg; struct hlsl_ir_constant *index; index = hlsl_ir_constant(sample_index); memset(&instr.srcs[2], 0, sizeof(instr.srcs[2])); - instr.srcs[2].swizzle_type = VKD3D_SM4_SWIZZLE_NONE; reg->type = VKD3DSPR_IMMCONST; - reg->dim = VKD3D_SM4_DIMENSION_SCALAR; - reg->immconst_uint[0] = index->value.u[0].u; + reg->dimension = VSIR_DIMENSION_SCALAR; + reg->u.immconst_uint[0] = index->value.u[0].u; } else if (tpf->ctx->profile->major_version == 4 && tpf->ctx->profile->minor_version == 0) { @@ -4431,7 +4596,7 @@ static void write_sm4_ld(const struct tpf_writer *tpf, const struct hlsl_ir_node } else { - sm4_src_from_node(&instr.srcs[2], sample_index, 0); + sm4_src_from_node(tpf, &instr.srcs[2], sample_index, 0); } ++instr.src_count; @@ -4493,27 +4658,27 @@ static void write_sm4_sample(const struct tpf_writer *tpf, const struct hlsl_ir_ sm4_dst_from_node(&instr.dsts[0], dst); instr.dst_count = 1; - sm4_src_from_node(&instr.srcs[0], coords, VKD3DSP_WRITEMASK_ALL); - sm4_src_from_deref(tpf->ctx, &instr.srcs[1], resource, instr.dsts[0].writemask); - sm4_src_from_deref(tpf->ctx, &instr.srcs[2], sampler, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[0], coords, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_deref(tpf, &instr.srcs[1], resource, instr.dsts[0].write_mask, &instr); + sm4_src_from_deref(tpf, &instr.srcs[2], sampler, VKD3DSP_WRITEMASK_ALL, &instr); instr.src_count = 3; if (load->load_type == HLSL_RESOURCE_SAMPLE_LOD || load->load_type == HLSL_RESOURCE_SAMPLE_LOD_BIAS) { - sm4_src_from_node(&instr.srcs[3], load->lod.node, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[3], load->lod.node, VKD3DSP_WRITEMASK_ALL); ++instr.src_count; } else if (load->load_type == HLSL_RESOURCE_SAMPLE_GRAD) { - sm4_src_from_node(&instr.srcs[3], load->ddx.node, VKD3DSP_WRITEMASK_ALL); - sm4_src_from_node(&instr.srcs[4], load->ddy.node, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[3], load->ddx.node, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[4], load->ddy.node, VKD3DSP_WRITEMASK_ALL); instr.src_count += 2; } else if (load->load_type == HLSL_RESOURCE_SAMPLE_CMP || load->load_type == HLSL_RESOURCE_SAMPLE_CMP_LZ) { - sm4_src_from_node(&instr.srcs[3], load->cmp.node, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[3], load->cmp.node, VKD3DSP_WRITEMASK_ALL); ++instr.src_count; } @@ -4531,12 +4696,12 @@ static void write_sm4_sampleinfo(const struct tpf_writer *tpf, const struct hlsl memset(&instr, 0, sizeof(instr)); instr.opcode = VKD3D_SM4_OP_SAMPLE_INFO; if (dst->data_type->base_type == HLSL_TYPE_UINT) - instr.opcode |= VKD3DSI_SAMPLE_INFO_UINT << VKD3D_SM4_INSTRUCTION_FLAGS_SHIFT; + instr.extra_bits |= VKD3DSI_SAMPLE_INFO_UINT << VKD3D_SM4_INSTRUCTION_FLAGS_SHIFT; sm4_dst_from_node(&instr.dsts[0], dst); instr.dst_count = 1; - sm4_src_from_deref(tpf->ctx, &instr.srcs[0], resource, instr.dsts[0].writemask); + sm4_src_from_deref(tpf, &instr.srcs[0], resource, instr.dsts[0].write_mask, &instr); instr.src_count = 1; write_sm4_instruction(tpf, &instr); @@ -4553,13 +4718,13 @@ static void write_sm4_resinfo(const struct tpf_writer *tpf, const struct hlsl_ir memset(&instr, 0, sizeof(instr)); instr.opcode = VKD3D_SM4_OP_RESINFO; if (dst->data_type->base_type == HLSL_TYPE_UINT) - instr.opcode |= VKD3DSI_RESINFO_UINT << VKD3D_SM4_INSTRUCTION_FLAGS_SHIFT; + instr.extra_bits |= VKD3DSI_RESINFO_UINT << VKD3D_SM4_INSTRUCTION_FLAGS_SHIFT; sm4_dst_from_node(&instr.dsts[0], dst); instr.dst_count = 1; - sm4_src_from_node(&instr.srcs[0], load->lod.node, VKD3DSP_WRITEMASK_ALL); - sm4_src_from_deref(tpf->ctx, &instr.srcs[1], resource, instr.dsts[0].writemask); + sm4_src_from_node(tpf, &instr.srcs[0], load->lod.node, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_deref(tpf, &instr.srcs[1], resource, instr.dsts[0].write_mask, &instr); instr.src_count = 2; write_sm4_instruction(tpf, &instr); @@ -4581,11 +4746,10 @@ static void write_sm4_cast_from_bool(const struct tpf_writer *tpf, const struct sm4_dst_from_node(&instr.dsts[0], &expr->node); instr.dst_count = 1; - sm4_src_from_node(&instr.srcs[0], arg, instr.dsts[0].writemask); - instr.srcs[1].swizzle_type = VKD3D_SM4_SWIZZLE_NONE; + sm4_src_from_node(tpf, &instr.srcs[0], arg, instr.dsts[0].write_mask); instr.srcs[1].reg.type = VKD3DSPR_IMMCONST; - instr.srcs[1].reg.dim = VKD3D_SM4_DIMENSION_SCALAR; - instr.srcs[1].reg.immconst_uint[0] = mask; + instr.srcs[1].reg.dimension = VSIR_DIMENSION_SCALAR; + instr.srcs[1].reg.u.immconst_uint[0] = mask; instr.src_count = 2; write_sm4_instruction(tpf, &instr); @@ -4708,11 +4872,11 @@ static void write_sm4_store_uav_typed(const struct tpf_writer *tpf, const struct memset(&instr, 0, sizeof(instr)); instr.opcode = VKD3D_SM5_OP_STORE_UAV_TYPED; - sm4_register_from_deref(tpf->ctx, &instr.dsts[0].reg, &instr.dsts[0].writemask, NULL, dst); + sm4_register_from_deref(tpf->ctx, &instr.dsts[0].reg, &instr.dsts[0].write_mask, dst, &instr); instr.dst_count = 1; - sm4_src_from_node(&instr.srcs[0], coords, VKD3DSP_WRITEMASK_ALL); - sm4_src_from_node(&instr.srcs[1], value, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[0], coords, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[1], value, VKD3DSP_WRITEMASK_ALL); instr.src_count = 2; write_sm4_instruction(tpf, &instr); @@ -4737,7 +4901,7 @@ static void write_sm4_expr(const struct tpf_writer *tpf, const struct hlsl_ir_ex switch (dst_type->base_type) { case HLSL_TYPE_FLOAT: - write_sm4_unary_op(tpf, VKD3D_SM4_OP_MOV, &expr->node, arg1, VKD3D_SM4_REGISTER_MODIFIER_ABS); + write_sm4_unary_op(tpf, VKD3D_SM4_OP_MOV, &expr->node, arg1, VKD3DSPSM_ABS); break; default: @@ -4754,6 +4918,11 @@ static void write_sm4_expr(const struct tpf_writer *tpf, const struct hlsl_ir_ex write_sm4_cast(tpf, expr); break; + case HLSL_OP1_CEIL: + assert(type_is_float(dst_type)); + write_sm4_unary_op(tpf, VKD3D_SM4_OP_ROUND_PI, &expr->node, arg1, 0); + break; + case HLSL_OP1_COS: assert(type_is_float(dst_type)); write_sm4_unary_op_with_two_destinations(tpf, VKD3D_SM4_OP_SINCOS, &expr->node, 1, arg1); @@ -4818,7 +4987,7 @@ static void write_sm4_expr(const struct tpf_writer *tpf, const struct hlsl_ir_ex switch (dst_type->base_type) { case HLSL_TYPE_FLOAT: - write_sm4_unary_op(tpf, VKD3D_SM4_OP_MOV, &expr->node, arg1, VKD3D_SM4_REGISTER_MODIFIER_NEGATE); + write_sm4_unary_op(tpf, VKD3D_SM4_OP_MOV, &expr->node, arg1, VKD3DSPSM_NEG); break; case HLSL_TYPE_INT: @@ -5169,7 +5338,7 @@ static void write_sm4_if(const struct tpf_writer *tpf, const struct hlsl_ir_if * assert(iff->condition.node->data_type->dimx == 1); - sm4_src_from_node(&instr.srcs[0], iff->condition.node, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[0], iff->condition.node, VKD3DSP_WRITEMASK_ALL); write_sm4_instruction(tpf, &instr); write_sm4_block(tpf, &iff->then_block); @@ -5198,13 +5367,17 @@ static void write_sm4_jump(const struct tpf_writer *tpf, const struct hlsl_ir_ju instr.opcode = VKD3D_SM4_OP_BREAK; break; + case HLSL_IR_JUMP_CONTINUE: + instr.opcode = VKD3D_SM4_OP_CONTINUE; + break; + case HLSL_IR_JUMP_DISCARD_NZ: { instr.opcode = VKD3D_SM4_OP_DISCARD | VKD3D_SM4_CONDITIONAL_NZ; memset(&instr.srcs[0], 0, sizeof(*instr.srcs)); instr.src_count = 1; - sm4_src_from_node(&instr.srcs[0], jump->condition.node, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[0], jump->condition.node, VKD3DSP_WRITEMASK_ALL); break; } @@ -5240,7 +5413,7 @@ static void write_sm4_load(const struct tpf_writer *tpf, const struct hlsl_ir_lo sm4_dst_from_node(&instr.dsts[0], &load->node); instr.dst_count = 1; - assert(type->class <= HLSL_CLASS_LAST_NUMERIC); + assert(hlsl_is_numeric_type(type)); if (type->base_type == HLSL_TYPE_BOOL && var_is_user_input(tpf->ctx, load->src.var)) { struct hlsl_constant_value value; @@ -5250,19 +5423,19 @@ static void write_sm4_load(const struct tpf_writer *tpf, const struct hlsl_ir_lo instr.opcode = VKD3D_SM4_OP_MOVC; - sm4_src_from_deref(tpf->ctx, &instr.srcs[0], &load->src, instr.dsts[0].writemask); + sm4_src_from_deref(tpf, &instr.srcs[0], &load->src, instr.dsts[0].write_mask, &instr); memset(&value, 0xff, sizeof(value)); - sm4_src_from_constant_value(&instr.srcs[1], &value, type->dimx, instr.dsts[0].writemask); + sm4_src_from_constant_value(&instr.srcs[1], &value, type->dimx, instr.dsts[0].write_mask); memset(&value, 0, sizeof(value)); - sm4_src_from_constant_value(&instr.srcs[2], &value, type->dimx, instr.dsts[0].writemask); + sm4_src_from_constant_value(&instr.srcs[2], &value, type->dimx, instr.dsts[0].write_mask); instr.src_count = 3; } else { instr.opcode = VKD3D_SM4_OP_MOV; - sm4_src_from_deref(tpf->ctx, &instr.srcs[0], &load->src, instr.dsts[0].writemask); + sm4_src_from_deref(tpf, &instr.srcs[0], &load->src, instr.dsts[0].write_mask, &instr); instr.src_count = 1; } @@ -5285,10 +5458,10 @@ static void write_sm4_loop(const struct tpf_writer *tpf, const struct hlsl_ir_lo } static void write_sm4_gather(const struct tpf_writer *tpf, const struct hlsl_ir_node *dst, - const struct hlsl_deref *resource, const struct hlsl_deref *sampler, const struct hlsl_ir_node *coords, - unsigned int swizzle, const struct hlsl_ir_node *texel_offset) + const struct hlsl_deref *resource, const struct hlsl_deref *sampler, + const struct hlsl_ir_node *coords, DWORD swizzle, const struct hlsl_ir_node *texel_offset) { - struct sm4_src_register *src; + struct vkd3d_shader_src_param *src; struct sm4_instruction instr; memset(&instr, 0, sizeof(instr)); @@ -5298,7 +5471,7 @@ static void write_sm4_gather(const struct tpf_writer *tpf, const struct hlsl_ir_ sm4_dst_from_node(&instr.dsts[0], dst); instr.dst_count = 1; - sm4_src_from_node(&instr.srcs[instr.src_count++], coords, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[instr.src_count++], coords, VKD3DSP_WRITEMASK_ALL); if (texel_offset) { @@ -5311,16 +5484,15 @@ static void write_sm4_gather(const struct tpf_writer *tpf, const struct hlsl_ir_ return; } instr.opcode = VKD3D_SM5_OP_GATHER4_PO; - sm4_src_from_node(&instr.srcs[instr.src_count++], texel_offset, VKD3DSP_WRITEMASK_ALL); + sm4_src_from_node(tpf, &instr.srcs[instr.src_count++], texel_offset, VKD3DSP_WRITEMASK_ALL); } } - sm4_src_from_deref(tpf->ctx, &instr.srcs[instr.src_count++], resource, instr.dsts[0].writemask); + sm4_src_from_deref(tpf, &instr.srcs[instr.src_count++], resource, instr.dsts[0].write_mask, &instr); src = &instr.srcs[instr.src_count++]; - sm4_src_from_deref(tpf->ctx, src, sampler, VKD3DSP_WRITEMASK_ALL); - src->reg.dim = VKD3D_SM4_DIMENSION_VEC4; - src->swizzle_type = VKD3D_SM4_SWIZZLE_SCALAR; + sm4_src_from_deref(tpf, src, sampler, VKD3DSP_WRITEMASK_ALL, &instr); + src->reg.dimension = VSIR_DIMENSION_VEC4; src->swizzle = swizzle; write_sm4_instruction(tpf, &instr); @@ -5364,22 +5536,22 @@ static void write_sm4_resource_load(const struct tpf_writer *tpf, const struct h case HLSL_RESOURCE_GATHER_RED: write_sm4_gather(tpf, &load->node, &load->resource, &load->sampler, coords, - HLSL_SWIZZLE(X, X, X, X), texel_offset); + VKD3D_SHADER_SWIZZLE(X, X, X, X), texel_offset); break; case HLSL_RESOURCE_GATHER_GREEN: write_sm4_gather(tpf, &load->node, &load->resource, &load->sampler, coords, - HLSL_SWIZZLE(Y, Y, Y, Y), texel_offset); + VKD3D_SHADER_SWIZZLE(Y, Y, Y, Y), texel_offset); break; case HLSL_RESOURCE_GATHER_BLUE: write_sm4_gather(tpf, &load->node, &load->resource, &load->sampler, coords, - HLSL_SWIZZLE(Z, Z, Z, Z), texel_offset); + VKD3D_SHADER_SWIZZLE(Z, Z, Z, Z), texel_offset); break; case HLSL_RESOURCE_GATHER_ALPHA: write_sm4_gather(tpf, &load->node, &load->resource, &load->sampler, coords, - HLSL_SWIZZLE(W, W, W, W), texel_offset); + VKD3D_SHADER_SWIZZLE(W, W, W, W), texel_offset); break; case HLSL_RESOURCE_SAMPLE_INFO: @@ -5389,6 +5561,9 @@ static void write_sm4_resource_load(const struct tpf_writer *tpf, const struct h case HLSL_RESOURCE_RESINFO: write_sm4_resinfo(tpf, load); break; + + case HLSL_RESOURCE_SAMPLE_PROJ: + vkd3d_unreachable(); } } @@ -5415,25 +5590,66 @@ static void write_sm4_store(const struct tpf_writer *tpf, const struct hlsl_ir_s { const struct hlsl_ir_node *rhs = store->rhs.node; struct sm4_instruction instr; - unsigned int writemask; + uint32_t writemask; memset(&instr, 0, sizeof(instr)); instr.opcode = VKD3D_SM4_OP_MOV; - sm4_register_from_deref(tpf->ctx, &instr.dsts[0].reg, &writemask, NULL, &store->lhs); - instr.dsts[0].writemask = hlsl_combine_writemasks(writemask, store->writemask); + sm4_register_from_deref(tpf->ctx, &instr.dsts[0].reg, &writemask, &store->lhs, &instr); + instr.dsts[0].write_mask = hlsl_combine_writemasks(writemask, store->writemask); instr.dst_count = 1; - sm4_src_from_node(&instr.srcs[0], rhs, instr.dsts[0].writemask); + sm4_src_from_node(tpf, &instr.srcs[0], rhs, instr.dsts[0].write_mask); instr.src_count = 1; write_sm4_instruction(tpf, &instr); } +static void write_sm4_switch(const struct tpf_writer *tpf, const struct hlsl_ir_switch *s) +{ + const struct hlsl_ir_node *selector = s->selector.node; + struct hlsl_ir_switch_case *c; + struct sm4_instruction instr; + + memset(&instr, 0, sizeof(instr)); + instr.opcode = VKD3D_SM4_OP_SWITCH; + + sm4_src_from_node(tpf, &instr.srcs[0], selector, VKD3DSP_WRITEMASK_ALL); + instr.src_count = 1; + + write_sm4_instruction(tpf, &instr); + + LIST_FOR_EACH_ENTRY(c, &s->cases, struct hlsl_ir_switch_case, entry) + { + memset(&instr, 0, sizeof(instr)); + if (c->is_default) + { + instr.opcode = VKD3D_SM4_OP_DEFAULT; + } + else + { + struct hlsl_constant_value value = { .u[0].u = c->value }; + + instr.opcode = VKD3D_SM4_OP_CASE; + sm4_src_from_constant_value(&instr.srcs[0], &value, 1, VKD3DSP_WRITEMASK_ALL); + instr.src_count = 1; + } + + write_sm4_instruction(tpf, &instr); + write_sm4_block(tpf, &c->body); + } + + memset(&instr, 0, sizeof(instr)); + instr.opcode = VKD3D_SM4_OP_ENDSWITCH; + + write_sm4_instruction(tpf, &instr); +} + static void write_sm4_swizzle(const struct tpf_writer *tpf, const struct hlsl_ir_swizzle *swizzle) { + unsigned int hlsl_swizzle; struct sm4_instruction instr; - unsigned int writemask; + uint32_t writemask; memset(&instr, 0, sizeof(instr)); instr.opcode = VKD3D_SM4_OP_MOV; @@ -5441,9 +5657,10 @@ static void write_sm4_swizzle(const struct tpf_writer *tpf, const struct hlsl_ir sm4_dst_from_node(&instr.dsts[0], &swizzle->node); instr.dst_count = 1; - sm4_register_from_node(&instr.srcs[0].reg, &writemask, &instr.srcs[0].swizzle_type, swizzle->val.node); - instr.srcs[0].swizzle = hlsl_map_swizzle(hlsl_combine_swizzles(hlsl_swizzle_from_writemask(writemask), - swizzle->swizzle, swizzle->node.data_type->dimx), instr.dsts[0].writemask); + sm4_register_from_node(&instr.srcs[0].reg, &writemask, swizzle->val.node); + hlsl_swizzle = hlsl_map_swizzle(hlsl_combine_swizzles(hlsl_swizzle_from_writemask(writemask), + swizzle->swizzle, swizzle->node.data_type->dimx), instr.dsts[0].write_mask); + instr.srcs[0].swizzle = swizzle_from_sm4(hlsl_swizzle); instr.src_count = 1; write_sm4_instruction(tpf, &instr); @@ -5515,6 +5732,10 @@ static void write_sm4_block(const struct tpf_writer *tpf, const struct hlsl_bloc write_sm4_store(tpf, hlsl_ir_store(instr)); break; + case HLSL_IR_SWITCH: + write_sm4_switch(tpf, hlsl_ir_switch(instr)); + break; + case HLSL_IR_SWIZZLE: write_sm4_swizzle(tpf, hlsl_ir_swizzle(instr)); break; @@ -5533,6 +5754,7 @@ static void write_sm4_shdr(struct hlsl_ctx *ctx, struct extern_resource *extern_resources; unsigned int extern_resources_count, i; const struct hlsl_buffer *cbuffer; + const struct hlsl_scope *scope; const struct hlsl_ir_var *var; size_t token_count_position; struct tpf_writer tpf; @@ -5587,6 +5809,25 @@ static void write_sm4_shdr(struct hlsl_ctx *ctx, if (ctx->temp_count) write_sm4_dcl_temps(&tpf, ctx->temp_count); + LIST_FOR_EACH_ENTRY(scope, &ctx->scopes, struct hlsl_scope, entry) + { + LIST_FOR_EACH_ENTRY(var, &scope->vars, struct hlsl_ir_var, scope_entry) + { + if (var->is_uniform || var->is_input_semantic || var->is_output_semantic) + continue; + if (!var->regs[HLSL_REGSET_NUMERIC].allocated) + continue; + + if (var->indexable) + { + unsigned int id = var->regs[HLSL_REGSET_NUMERIC].id; + unsigned int size = align(var->data_type->reg_size[HLSL_REGSET_NUMERIC], 4) / 4; + + write_sm4_dcl_indexable_temp(&tpf, id, size, 4); + } + } + } + write_sm4_block(&tpf, &entry_func->body); write_sm4_ret(&tpf); diff --git a/libs/vkd3d/libs/vkd3d-shader/vkd3d_shader_main.c b/libs/vkd3d/libs/vkd3d-shader/vkd3d_shader_main.c index f25dbb04d69..b60dc17e902 100644 --- a/libs/vkd3d/libs/vkd3d-shader/vkd3d_shader_main.c +++ b/libs/vkd3d/libs/vkd3d-shader/vkd3d_shader_main.c @@ -18,10 +18,13 @@ #include "vkd3d_shader_private.h" #include "vkd3d_version.h" +#include "hlsl.h" #include #include +/* VKD3D_DEBUG_ENV_NAME("VKD3D_SHADER_DEBUG"); */ + static inline int char_to_int(char c) { if ('0' <= c && c <= '9') @@ -379,7 +382,7 @@ void set_u32(struct vkd3d_bytecode_buffer *buffer, size_t offset, uint32_t value memcpy(buffer->data + offset, &value, sizeof(value)); } -static void vkd3d_shader_dump_blob(const char *path, const char *prefix, +static void vkd3d_shader_dump_blob(const char *path, const char *profile, const char *suffix, const void *data, size_t size) { static LONG shader_id = 0; @@ -389,7 +392,10 @@ static void vkd3d_shader_dump_blob(const char *path, const char *prefix, id = InterlockedIncrement(&shader_id) - 1; - snprintf(filename, ARRAY_SIZE(filename), "%s/vkd3d-shader-%s-%u.%s", path, prefix, id, suffix); + if (profile) + snprintf(filename, ARRAY_SIZE(filename), "%s/vkd3d-shader-%u-%s.%s", path, id, profile, suffix); + else + snprintf(filename, ARRAY_SIZE(filename), "%s/vkd3d-shader-%u.%s", path, id, suffix); if ((f = fopen(filename, "wb"))) { if (fwrite(data, 1, size, f) != size) @@ -421,9 +427,12 @@ static const char *shader_get_source_type_suffix(enum vkd3d_shader_source_type t } } -void vkd3d_shader_dump_shader(enum vkd3d_shader_source_type source_type, - enum vkd3d_shader_type shader_type, const struct vkd3d_shader_code *shader) +void vkd3d_shader_dump_shader(const struct vkd3d_shader_compile_info *compile_info) { + const struct vkd3d_shader_code *shader = &compile_info->source; + const struct vkd3d_shader_hlsl_source_info *hlsl_source_info; + const struct hlsl_profile_info *profile; + const char *profile_name = NULL; static bool enabled = true; const char *path; @@ -436,8 +445,19 @@ void vkd3d_shader_dump_shader(enum vkd3d_shader_source_type source_type, return; } - vkd3d_shader_dump_blob(path, shader_get_type_prefix(shader_type), - shader_get_source_type_suffix(source_type), shader->code, shader->size); + if (compile_info->source_type == VKD3D_SHADER_SOURCE_HLSL) + { + if (!(hlsl_source_info = vkd3d_find_struct(compile_info->next, HLSL_SOURCE_INFO))) + return; + + if (!(profile = hlsl_get_target_info(hlsl_source_info->profile))) + return; + + profile_name = profile->name; + } + + vkd3d_shader_dump_blob(path, profile_name, shader_get_source_type_suffix(compile_info->source_type), + shader->code, shader->size); } static void init_scan_signature_info(const struct vkd3d_shader_compile_info *info) @@ -452,6 +472,25 @@ static void init_scan_signature_info(const struct vkd3d_shader_compile_info *inf } } +static const struct vkd3d_debug_option vkd3d_shader_config_options[] = +{ + {"force_validation", VKD3D_SHADER_CONFIG_FLAG_FORCE_VALIDATION}, /* force validation of internal shader representations */ +}; + +static uint64_t vkd3d_shader_init_config_flags(void) +{ + uint64_t config_flags; + const char *config; + + config = getenv("VKD3D_SHADER_CONFIG"); + config_flags = vkd3d_parse_debug_options(config, vkd3d_shader_config_options, ARRAY_SIZE(vkd3d_shader_config_options)); + + if (config_flags) + TRACE("VKD3D_SHADER_CONFIG='%s'.\n", config); + + return config_flags; +} + bool vkd3d_shader_parser_init(struct vkd3d_shader_parser *parser, struct vkd3d_shader_message_context *message_context, const char *source_name, const struct vkd3d_shader_version *version, const struct vkd3d_shader_parser_ops *ops, @@ -463,6 +502,7 @@ bool vkd3d_shader_parser_init(struct vkd3d_shader_parser *parser, parser->location.column = 0; parser->shader_version = *version; parser->ops = ops; + parser->config_flags = vkd3d_shader_init_config_flags(); return shader_instruction_array_init(&parser->instructions, instruction_reserve); } @@ -577,6 +617,8 @@ static bool vkd3d_shader_signature_from_shader_signature(struct vkd3d_shader_sig struct vkd3d_shader_scan_context { + const struct vkd3d_shader_version *version; + struct vkd3d_shader_scan_descriptor_info1 *scan_descriptor_info; size_t descriptors_size; @@ -598,21 +640,48 @@ struct vkd3d_shader_scan_context size_t cf_info_count; enum vkd3d_shader_api_version api_version; + + struct vkd3d_shader_scan_combined_resource_sampler_info *combined_sampler_info; + size_t combined_samplers_size; }; +static VKD3D_PRINTF_FUNC(3, 4) void vkd3d_shader_scan_error(struct vkd3d_shader_scan_context *context, + enum vkd3d_shader_error error, const char *format, ...) +{ + va_list args; + + va_start(args, format); + vkd3d_shader_verror(context->message_context, &context->location, error, format, args); + va_end(args); +} + +static void VKD3D_PRINTF_FUNC(3, 4) vkd3d_shader_scan_warning(struct vkd3d_shader_scan_context *context, + enum vkd3d_shader_error error, const char *format, ...) +{ + va_list args; + + va_start(args, format); + vkd3d_shader_vwarning(context->message_context, &context->location, error, format, args); + va_end(args); +} + static void vkd3d_shader_scan_context_init(struct vkd3d_shader_scan_context *context, + const struct vkd3d_shader_version *version, const struct vkd3d_shader_compile_info *compile_info, struct vkd3d_shader_scan_descriptor_info1 *scan_descriptor_info, + struct vkd3d_shader_scan_combined_resource_sampler_info *combined_sampler_info, struct vkd3d_shader_message_context *message_context) { unsigned int i; memset(context, 0, sizeof(*context)); + context->version = version; context->scan_descriptor_info = scan_descriptor_info; context->message_context = message_context; context->location.source_name = compile_info->source_name; context->location.line = 2; /* Line 1 is the version token. */ context->api_version = VKD3D_SHADER_API_VERSION_1_2; + context->combined_sampler_info = combined_sampler_info; for (i = 0; i < compile_info->option_count; ++i) { @@ -762,6 +831,9 @@ static struct vkd3d_shader_descriptor_info1 *vkd3d_shader_scan_add_descriptor(st struct vkd3d_shader_scan_descriptor_info1 *info = context->scan_descriptor_info; struct vkd3d_shader_descriptor_info1 *d; + if (!info) + return NULL; + if (!vkd3d_array_reserve((void **)&info->descriptors, &context->descriptors_size, info->descriptor_count + 1, sizeof(*info->descriptors))) { @@ -789,13 +861,10 @@ static void vkd3d_shader_scan_constant_buffer_declaration(struct vkd3d_shader_sc const struct vkd3d_shader_constant_buffer *cb = &instruction->declaration.cb; struct vkd3d_shader_descriptor_info1 *d; - if (!context->scan_descriptor_info) - return; - if (!(d = vkd3d_shader_scan_add_descriptor(context, VKD3D_SHADER_DESCRIPTOR_TYPE_CBV, &cb->src.reg, &cb->range, VKD3D_SHADER_RESOURCE_BUFFER, VKD3D_SHADER_RESOURCE_DATA_UINT))) return; - d->buffer_size = cb->size * 16; + d->buffer_size = cb->size; } static void vkd3d_shader_scan_sampler_declaration(struct vkd3d_shader_scan_context *context, @@ -804,9 +873,6 @@ static void vkd3d_shader_scan_sampler_declaration(struct vkd3d_shader_scan_conte const struct vkd3d_shader_sampler *sampler = &instruction->declaration.sampler; struct vkd3d_shader_descriptor_info1 *d; - if (!context->scan_descriptor_info) - return; - if (!(d = vkd3d_shader_scan_add_descriptor(context, VKD3D_SHADER_DESCRIPTOR_TYPE_SAMPLER, &sampler->src.reg, &sampler->range, VKD3D_SHADER_RESOURCE_NONE, VKD3D_SHADER_RESOURCE_DATA_UINT))) return; @@ -815,6 +881,103 @@ static void vkd3d_shader_scan_sampler_declaration(struct vkd3d_shader_scan_conte d->flags |= VKD3D_SHADER_DESCRIPTOR_INFO_FLAG_SAMPLER_COMPARISON_MODE; } +static void vkd3d_shader_scan_combined_sampler_declaration( + struct vkd3d_shader_scan_context *context, const struct vkd3d_shader_semantic *semantic) +{ + vkd3d_shader_scan_add_descriptor(context, VKD3D_SHADER_DESCRIPTOR_TYPE_SAMPLER, &semantic->resource.reg.reg, + &semantic->resource.range, VKD3D_SHADER_RESOURCE_NONE, VKD3D_SHADER_RESOURCE_DATA_UINT); + vkd3d_shader_scan_add_descriptor(context, VKD3D_SHADER_DESCRIPTOR_TYPE_SRV, &semantic->resource.reg.reg, + &semantic->resource.range, semantic->resource_type, VKD3D_SHADER_RESOURCE_DATA_FLOAT); +} + +static void vkd3d_shader_scan_combined_sampler_usage(struct vkd3d_shader_scan_context *context, + const struct vkd3d_shader_register *resource, const struct vkd3d_shader_register *sampler) +{ + struct vkd3d_shader_scan_combined_resource_sampler_info *info; + struct vkd3d_shader_combined_resource_sampler_info *s; + unsigned resource_space = 0, sampler_space = 0; + unsigned int resource_idx, sampler_idx, i; + + if (!(info = context->combined_sampler_info)) + return; + + if (resource->type == VKD3DSPR_RESOURCE) + resource_idx = resource->idx[1].offset; + else + resource_idx = resource->idx[0].offset; + + if (!sampler) + sampler_idx = VKD3D_SHADER_DUMMY_SAMPLER_INDEX; + else if (sampler->type == VKD3DSPR_SAMPLER) + sampler_idx = sampler->idx[1].offset; + else + sampler_idx = sampler->idx[0].offset; + + if (vkd3d_shader_ver_ge(context->version, 5, 1)) + { + const struct vkd3d_shader_scan_descriptor_info1 *info = context->scan_descriptor_info; + const struct vkd3d_shader_descriptor_info1 *d; + bool dynamic_resource, dynamic_sampler; + + if ((dynamic_resource = resource->idx[1].rel_addr)) + vkd3d_shader_scan_warning(context, VKD3D_SHADER_WARNING_VSIR_DYNAMIC_DESCRIPTOR_ARRAY, + "Resource descriptor array %u is being dynamically indexed, " + "not recording a combined resource-sampler pair.", resource->idx[0].offset); + if ((dynamic_sampler = sampler && sampler->idx[1].rel_addr)) + vkd3d_shader_scan_warning(context, VKD3D_SHADER_WARNING_VSIR_DYNAMIC_DESCRIPTOR_ARRAY, + "Sampler descriptor array %u is being dynamically indexed, " + "not recording a combined resource-sampler pair.", sampler->idx[0].offset); + if (dynamic_resource || dynamic_sampler) + return; + + for (i = 0; i < info->descriptor_count; ++i) + { + d = &info->descriptors[i]; + if (d->type != VKD3D_SHADER_DESCRIPTOR_TYPE_SRV) + continue; + if (d->register_id != resource->idx[0].offset) + continue; + resource_space = d->register_space; + break; + } + + if (sampler) + { + for (i = 0; i < info->descriptor_count; ++i) + { + d = &info->descriptors[i]; + if (d->type != VKD3D_SHADER_DESCRIPTOR_TYPE_SAMPLER) + continue; + if (d->register_id != sampler->idx[0].offset) + continue; + sampler_space = d->register_space; + break; + } + } + } + + for (i = 0; i < info->combined_sampler_count; ++i) + { + s = &info->combined_samplers[i]; + if (s->resource_space == resource_space && s->resource_index == resource_idx + && s->sampler_space == sampler_space && s->sampler_index == sampler_idx) + return; + } + + if (!vkd3d_array_reserve((void **)&info->combined_samplers, &context->combined_samplers_size, + info->combined_sampler_count + 1, sizeof(*info->combined_samplers))) + { + ERR("Failed to allocate combined sampler info.\n"); + return; + } + + s = &info->combined_samplers[info->combined_sampler_count++]; + s->resource_space = resource_space; + s->resource_index = resource_idx; + s->sampler_space = sampler_space; + s->sampler_index = sampler_idx; +} + static void vkd3d_shader_scan_resource_declaration(struct vkd3d_shader_scan_context *context, const struct vkd3d_shader_resource *resource, enum vkd3d_shader_resource_type resource_type, enum vkd3d_shader_resource_data_type resource_data_type, @@ -823,9 +986,6 @@ static void vkd3d_shader_scan_resource_declaration(struct vkd3d_shader_scan_cont struct vkd3d_shader_descriptor_info1 *d; enum vkd3d_shader_descriptor_type type; - if (!context->scan_descriptor_info) - return; - if (resource->reg.reg.type == VKD3DSPR_UAV) type = VKD3D_SHADER_DESCRIPTOR_TYPE_UAV; else @@ -898,22 +1058,15 @@ static void vkd3d_shader_scan_typed_resource_declaration(struct vkd3d_shader_sca semantic->resource_type, resource_data_type, semantic->sample_count, 0, false); } -static void vkd3d_shader_scan_error(struct vkd3d_shader_scan_context *context, - enum vkd3d_shader_error error, const char *format, ...) -{ - va_list args; - - va_start(args, format); - vkd3d_shader_verror(context->message_context, &context->location, error, format, args); - va_end(args); -} - static int vkd3d_shader_scan_instruction(struct vkd3d_shader_scan_context *context, const struct vkd3d_shader_instruction *instruction) { + const struct vkd3d_shader_register *sampler_reg; struct vkd3d_shader_cf_info *cf_info; unsigned int i; + context->location = instruction->location; + switch (instruction->handler_idx) { case VKD3DSIH_DCL_CONSTANT_BUFFER: @@ -923,6 +1076,12 @@ static int vkd3d_shader_scan_instruction(struct vkd3d_shader_scan_context *conte vkd3d_shader_scan_sampler_declaration(context, instruction); break; case VKD3DSIH_DCL: + if (instruction->declaration.semantic.resource.reg.reg.type == VKD3DSPR_COMBINED_SAMPLER) + { + vkd3d_shader_scan_combined_sampler_declaration(context, &instruction->declaration.semantic); + break; + } + /* fall through */ case VKD3DSIH_DCL_UAV_TYPED: vkd3d_shader_scan_typed_resource_declaration(context, instruction); break; @@ -1053,6 +1212,58 @@ static int vkd3d_shader_scan_instruction(struct vkd3d_shader_scan_context *conte if (context->cf_info_count) context->cf_info[context->cf_info_count - 1].inside_block = false; break; + case VKD3DSIH_TEX: + if (context->version->major == 1) + sampler_reg = &instruction->dst[0].reg; + else + sampler_reg = &instruction->src[1].reg; + vkd3d_shader_scan_combined_sampler_usage(context, sampler_reg, sampler_reg); + break; + case VKD3DSIH_TEXBEM: + case VKD3DSIH_TEXBEML: + case VKD3DSIH_TEXDP3TEX: + case VKD3DSIH_TEXM3x2TEX: + case VKD3DSIH_TEXM3x3SPEC: + case VKD3DSIH_TEXM3x3TEX: + case VKD3DSIH_TEXM3x3VSPEC: + case VKD3DSIH_TEXREG2AR: + case VKD3DSIH_TEXREG2GB: + case VKD3DSIH_TEXREG2RGB: + sampler_reg = &instruction->dst[0].reg; + vkd3d_shader_scan_combined_sampler_usage(context, sampler_reg, sampler_reg); + break; + case VKD3DSIH_GATHER4: + case VKD3DSIH_GATHER4_C: + case VKD3DSIH_SAMPLE: + case VKD3DSIH_SAMPLE_B: + case VKD3DSIH_SAMPLE_C: + case VKD3DSIH_SAMPLE_C_LZ: + case VKD3DSIH_SAMPLE_GRAD: + case VKD3DSIH_SAMPLE_LOD: + vkd3d_shader_scan_combined_sampler_usage(context, &instruction->src[1].reg, &instruction->src[2].reg); + break; + case VKD3DSIH_GATHER4_PO: + case VKD3DSIH_GATHER4_PO_C: + vkd3d_shader_scan_combined_sampler_usage(context, &instruction->src[2].reg, &instruction->src[3].reg); + break; + case VKD3DSIH_LD: + case VKD3DSIH_LD2DMS: + vkd3d_shader_scan_combined_sampler_usage(context, &instruction->src[1].reg, NULL); + break; + case VKD3DSIH_BUFINFO: + case VKD3DSIH_SAMPLE_INFO: + if (instruction->src[0].reg.type == VKD3DSPR_RESOURCE) + vkd3d_shader_scan_combined_sampler_usage(context, &instruction->src[0].reg, NULL); + break; + case VKD3DSIH_LD_RAW: + case VKD3DSIH_RESINFO: + if (instruction->src[1].reg.type == VKD3DSPR_RESOURCE) + vkd3d_shader_scan_combined_sampler_usage(context, &instruction->src[1].reg, NULL); + break; + case VKD3DSIH_LD_STRUCTURED: + if (instruction->src[2].reg.type == VKD3DSPR_RESOURCE) + vkd3d_shader_scan_combined_sampler_usage(context, &instruction->src[2].reg, NULL); + break; default: break; } @@ -1083,7 +1294,6 @@ static int vkd3d_shader_scan_instruction(struct vkd3d_shader_scan_context *conte } } - ++context->location.line; return VKD3D_OK; } @@ -1124,6 +1334,7 @@ static int scan_with_parser(const struct vkd3d_shader_compile_info *compile_info struct vkd3d_shader_message_context *message_context, struct vkd3d_shader_scan_descriptor_info1 *descriptor_info1, struct vkd3d_shader_parser *parser) { + struct vkd3d_shader_scan_combined_resource_sampler_info *combined_sampler_info; struct vkd3d_shader_scan_descriptor_info1 local_descriptor_info1 = {0}; struct vkd3d_shader_scan_descriptor_info *descriptor_info; struct vkd3d_shader_scan_signature_info *signature_info; @@ -1144,7 +1355,16 @@ static int scan_with_parser(const struct vkd3d_shader_compile_info *compile_info } signature_info = vkd3d_find_struct(compile_info->next, SCAN_SIGNATURE_INFO); - vkd3d_shader_scan_context_init(&context, compile_info, descriptor_info1, message_context); + if ((combined_sampler_info = vkd3d_find_struct(compile_info->next, SCAN_COMBINED_RESOURCE_SAMPLER_INFO))) + { + combined_sampler_info->combined_samplers = NULL; + combined_sampler_info->combined_sampler_count = 0; + if (!descriptor_info1) + descriptor_info1 = &local_descriptor_info1; + } + + vkd3d_shader_scan_context_init(&context, &parser->shader_version, compile_info, + descriptor_info1, combined_sampler_info, message_context); if (TRACE_ON()) { @@ -1190,6 +1410,8 @@ static int scan_with_parser(const struct vkd3d_shader_compile_info *compile_info if (ret < 0) { + if (combined_sampler_info) + vkd3d_shader_free_scan_combined_resource_sampler_info(combined_sampler_info); if (descriptor_info) vkd3d_shader_free_scan_descriptor_info(descriptor_info); if (descriptor_info1) @@ -1276,6 +1498,8 @@ int vkd3d_shader_scan(const struct vkd3d_shader_compile_info *compile_info, char vkd3d_shader_message_context_init(&message_context, compile_info->log_level); + vkd3d_shader_dump_shader(compile_info); + switch (compile_info->source_type) { case VKD3D_SHADER_SOURCE_DXBC_TPF: @@ -1317,8 +1541,6 @@ static int vkd3d_shader_parser_compile(struct vkd3d_shader_parser *parser, struct vkd3d_shader_compile_info scan_info; int ret; - vkd3d_shader_dump_shader(compile_info->source_type, parser->shader_version.type, &compile_info->source); - scan_info = *compile_info; if ((ret = scan_with_parser(&scan_info, message_context, &scan_descriptor_info, parser)) < 0) @@ -1402,8 +1624,6 @@ static int compile_d3d_bytecode(const struct vkd3d_shader_compile_info *compile_ return ret; } - vkd3d_shader_dump_shader(compile_info->source_type, parser->shader_version.type, &compile_info->source); - if (compile_info->target_type == VKD3D_SHADER_TARGET_D3D_ASM) { ret = vkd3d_dxbc_binary_to_text(&parser->instructions, &parser->shader_version, compile_info, out); @@ -1450,6 +1670,8 @@ int vkd3d_shader_compile(const struct vkd3d_shader_compile_info *compile_info, vkd3d_shader_message_context_init(&message_context, compile_info->log_level); + vkd3d_shader_dump_shader(compile_info); + switch (compile_info->source_type) { case VKD3D_SHADER_SOURCE_DXBC_TPF: @@ -1479,6 +1701,14 @@ int vkd3d_shader_compile(const struct vkd3d_shader_compile_info *compile_info, return ret; } +void vkd3d_shader_free_scan_combined_resource_sampler_info( + struct vkd3d_shader_scan_combined_resource_sampler_info *info) +{ + TRACE("info %p.\n", info); + + vkd3d_free(info->combined_samplers); +} + void vkd3d_shader_free_scan_descriptor_info(struct vkd3d_shader_scan_descriptor_info *scan_descriptor_info) { TRACE("scan_descriptor_info %p.\n", scan_descriptor_info); @@ -1751,7 +1981,7 @@ static struct vkd3d_shader_param_node *shader_param_allocator_node_create( static void shader_param_allocator_init(struct vkd3d_shader_param_allocator *allocator, unsigned int count, unsigned int stride) { - allocator->count = max(count, 4); + allocator->count = max(count, MAX_REG_OUTPUT); allocator->stride = stride; allocator->head = NULL; allocator->current = NULL; diff --git a/libs/vkd3d/libs/vkd3d-shader/vkd3d_shader_private.h b/libs/vkd3d/libs/vkd3d-shader/vkd3d_shader_private.h index 9443df6c232..d3989672b89 100644 --- a/libs/vkd3d/libs/vkd3d-shader/vkd3d_shader_private.h +++ b/libs/vkd3d/libs/vkd3d-shader/vkd3d_shader_private.h @@ -49,7 +49,7 @@ #include "vkd3d_common.h" #include "vkd3d_memory.h" #include "vkd3d_shader.h" -#include "wine/list.h" +#include "list.h" #include #include @@ -92,6 +92,10 @@ enum vkd3d_shader_error VKD3D_SHADER_ERROR_SPV_INVALID_DESCRIPTOR_BINDING = 2002, VKD3D_SHADER_ERROR_SPV_DESCRIPTOR_IDX_UNSUPPORTED = 2003, VKD3D_SHADER_ERROR_SPV_STENCIL_EXPORT_UNSUPPORTED = 2004, + VKD3D_SHADER_ERROR_SPV_OUT_OF_MEMORY = 2005, + VKD3D_SHADER_ERROR_SPV_INVALID_TYPE = 2006, + VKD3D_SHADER_ERROR_SPV_INVALID_HANDLER = 2007, + VKD3D_SHADER_ERROR_SPV_NOT_IMPLEMENTED = 2008, VKD3D_SHADER_WARNING_SPV_INVALID_SWIZZLE = 2300, @@ -141,6 +145,7 @@ enum vkd3d_shader_error VKD3D_SHADER_ERROR_HLSL_RECURSIVE_CALL = 5025, VKD3D_SHADER_ERROR_HLSL_INCONSISTENT_SAMPLER = 5026, VKD3D_SHADER_ERROR_HLSL_NON_FINITE_RESULT = 5027, + VKD3D_SHADER_ERROR_HLSL_DUPLICATE_SWITCH_CASE = 5028, VKD3D_SHADER_WARNING_HLSL_IMPLICIT_TRUNCATION = 5300, VKD3D_SHADER_WARNING_HLSL_DIVISION_BY_ZERO = 5301, @@ -174,15 +179,43 @@ enum vkd3d_shader_error VKD3D_SHADER_ERROR_DXIL_INVALID_TYPE_ID = 8010, VKD3D_SHADER_ERROR_DXIL_INVALID_MODULE = 8011, VKD3D_SHADER_ERROR_DXIL_INVALID_OPERAND = 8012, + VKD3D_SHADER_ERROR_DXIL_UNHANDLED_INTRINSIC = 8013, + VKD3D_SHADER_ERROR_DXIL_INVALID_METADATA = 8014, + VKD3D_SHADER_ERROR_DXIL_INVALID_ENTRY_POINT = 8015, + VKD3D_SHADER_ERROR_DXIL_INVALID_SIGNATURE = 8016, + VKD3D_SHADER_ERROR_DXIL_INVALID_PROPERTIES = 8017, + VKD3D_SHADER_ERROR_DXIL_INVALID_RESOURCES = 8018, + VKD3D_SHADER_ERROR_DXIL_INVALID_RESOURCE_HANDLE = 8019, VKD3D_SHADER_WARNING_DXIL_UNKNOWN_MAGIC_NUMBER = 8300, VKD3D_SHADER_WARNING_DXIL_UNKNOWN_SHADER_TYPE = 8301, VKD3D_SHADER_WARNING_DXIL_INVALID_BLOCK_LENGTH = 8302, VKD3D_SHADER_WARNING_DXIL_INVALID_MODULE_LENGTH = 8303, VKD3D_SHADER_WARNING_DXIL_IGNORING_OPERANDS = 8304, - VKD3D_SHADER_WARNING_DXIL_UNHANDLED_INTRINSIC = 8305, + VKD3D_SHADER_WARNING_DXIL_TYPE_MISMATCH = 8305, + VKD3D_SHADER_WARNING_DXIL_ENTRY_POINT_MISMATCH = 8306, + VKD3D_SHADER_WARNING_DXIL_INVALID_MASK = 8307, + VKD3D_SHADER_WARNING_DXIL_INVALID_OPERATION = 8308, VKD3D_SHADER_ERROR_VSIR_NOT_IMPLEMENTED = 9000, + VKD3D_SHADER_ERROR_VSIR_INVALID_HANDLER = 9001, + VKD3D_SHADER_ERROR_VSIR_INVALID_REGISTER_TYPE = 9002, + VKD3D_SHADER_ERROR_VSIR_INVALID_WRITE_MASK = 9003, + VKD3D_SHADER_ERROR_VSIR_INVALID_MODIFIERS = 9004, + VKD3D_SHADER_ERROR_VSIR_INVALID_SHIFT = 9005, + VKD3D_SHADER_ERROR_VSIR_INVALID_SWIZZLE = 9006, + VKD3D_SHADER_ERROR_VSIR_INVALID_PRECISION = 9007, + VKD3D_SHADER_ERROR_VSIR_INVALID_DATA_TYPE = 9008, + VKD3D_SHADER_ERROR_VSIR_INVALID_DIMENSION = 9009, + VKD3D_SHADER_ERROR_VSIR_INVALID_INDEX_COUNT = 9010, + VKD3D_SHADER_ERROR_VSIR_INVALID_DEST_COUNT = 9011, + VKD3D_SHADER_ERROR_VSIR_INVALID_SOURCE_COUNT = 9012, + VKD3D_SHADER_ERROR_VSIR_DUPLICATE_DCL_TEMPS = 9013, + VKD3D_SHADER_ERROR_VSIR_INVALID_DCL_TEMPS = 9014, + VKD3D_SHADER_ERROR_VSIR_INVALID_INDEX = 9015, + VKD3D_SHADER_ERROR_VSIR_INVALID_INSTRUCTION_NESTING = 9016, + + VKD3D_SHADER_WARNING_VSIR_DYNAMIC_DESCRIPTOR_ARRAY = 9300, }; enum vkd3d_shader_opcode @@ -264,9 +297,9 @@ enum vkd3d_shader_opcode VKD3DSIH_DEFAULT, VKD3DSIH_DEFB, VKD3DSIH_DEFI, - VKD3DSIH_DEQ, + VKD3DSIH_DEQO, VKD3DSIH_DFMA, - VKD3DSIH_DGE, + VKD3DSIH_DGEO, VKD3DSIH_DISCARD, VKD3DSIH_DIV, VKD3DSIH_DLT, @@ -298,7 +331,8 @@ enum vkd3d_shader_opcode VKD3DSIH_ENDLOOP, VKD3DSIH_ENDREP, VKD3DSIH_ENDSWITCH, - VKD3DSIH_EQ, + VKD3DSIH_EQO, + VKD3DSIH_EQU, VKD3DSIH_EVAL_CENTROID, VKD3DSIH_EVAL_SAMPLE_INDEX, VKD3DSIH_EXP, @@ -310,6 +344,7 @@ enum vkd3d_shader_opcode VKD3DSIH_FIRSTBIT_LO, VKD3DSIH_FIRSTBIT_SHI, VKD3DSIH_FRC, + VKD3DSIH_FREM, VKD3DSIH_FTOD, VKD3DSIH_FTOI, VKD3DSIH_FTOU, @@ -321,13 +356,15 @@ enum vkd3d_shader_opcode VKD3DSIH_GATHER4_PO_C_S, VKD3DSIH_GATHER4_PO_S, VKD3DSIH_GATHER4_S, - VKD3DSIH_GE, + VKD3DSIH_GEO, + VKD3DSIH_GEU, VKD3DSIH_HS_CONTROL_POINT_PHASE, VKD3DSIH_HS_DECLS, VKD3DSIH_HS_FORK_PHASE, VKD3DSIH_HS_JOIN_PHASE, VKD3DSIH_IADD, VKD3DSIH_IBFE, + VKD3DSIH_IDIV, VKD3DSIH_IEQ, VKD3DSIH_IF, VKD3DSIH_IFC, @@ -355,6 +392,7 @@ enum vkd3d_shader_opcode VKD3DSIH_ISHR, VKD3DSIH_ITOD, VKD3DSIH_ITOF, + VKD3DSIH_ITOI, VKD3DSIH_LABEL, VKD3DSIH_LD, VKD3DSIH_LD2DMS, @@ -372,7 +410,8 @@ enum vkd3d_shader_opcode VKD3DSIH_LOGP, VKD3DSIH_LOOP, VKD3DSIH_LRP, - VKD3DSIH_LT, + VKD3DSIH_LTO, + VKD3DSIH_LTU, VKD3DSIH_M3x2, VKD3DSIH_M3x3, VKD3DSIH_M3x4, @@ -386,7 +425,8 @@ enum vkd3d_shader_opcode VKD3DSIH_MOVC, VKD3DSIH_MSAD, VKD3DSIH_MUL, - VKD3DSIH_NE, + VKD3DSIH_NEO, + VKD3DSIH_NEU, VKD3DSIH_NOP, VKD3DSIH_NOT, VKD3DSIH_NRM, @@ -462,6 +502,7 @@ enum vkd3d_shader_opcode VKD3DSIH_USHR, VKD3DSIH_UTOD, VKD3DSIH_UTOF, + VKD3DSIH_UTOU, VKD3DSIH_XOR, VKD3DSIH_INVALID, @@ -481,7 +522,7 @@ enum vkd3d_shader_register_type VKD3DSPR_CONSTINT = 7, VKD3DSPR_COLOROUT = 8, VKD3DSPR_DEPTHOUT = 9, - VKD3DSPR_SAMPLER = 10, + VKD3DSPR_COMBINED_SAMPLER = 10, VKD3DSPR_CONST2 = 11, VKD3DSPR_CONST3 = 12, VKD3DSPR_CONST4 = 13, @@ -497,6 +538,7 @@ enum vkd3d_shader_register_type VKD3DSPR_IMMCONSTBUFFER, VKD3DSPR_PRIMID, VKD3DSPR_NULL, + VKD3DSPR_SAMPLER, VKD3DSPR_RESOURCE, VKD3DSPR_UAV, VKD3DSPR_OUTPOINTID, @@ -523,6 +565,7 @@ enum vkd3d_shader_register_type VKD3DSPR_RASTERIZER, VKD3DSPR_OUTSTENCILREF, VKD3DSPR_UNDEF, + VKD3DSPR_SSA, VKD3DSPR_COUNT, @@ -537,6 +580,8 @@ enum vkd3d_shader_register_precision VKD3D_SHADER_REGISTER_PRECISION_MIN_INT_16, VKD3D_SHADER_REGISTER_PRECISION_MIN_UINT_16, + VKD3D_SHADER_REGISTER_PRECISION_COUNT, + VKD3D_SHADER_REGISTER_PRECISION_INVALID = ~0u, }; @@ -556,17 +601,30 @@ enum vkd3d_data_type VKD3D_DATA_CONTINUED, VKD3D_DATA_UNUSED, VKD3D_DATA_UINT8, + VKD3D_DATA_UINT64, + VKD3D_DATA_BOOL, + + VKD3D_DATA_COUNT, }; static inline bool data_type_is_integer(enum vkd3d_data_type data_type) { - return data_type == VKD3D_DATA_INT || data_type == VKD3D_DATA_UINT8 || data_type == VKD3D_DATA_UINT; + return data_type == VKD3D_DATA_INT || data_type == VKD3D_DATA_UINT8 || data_type == VKD3D_DATA_UINT + || data_type == VKD3D_DATA_UINT64; } -enum vkd3d_immconst_type +static inline bool data_type_is_bool(enum vkd3d_data_type data_type) { - VKD3D_IMMCONST_SCALAR, - VKD3D_IMMCONST_VEC4, + return data_type == VKD3D_DATA_BOOL; +} + +enum vsir_dimension +{ + VSIR_DIMENSION_NONE, + VSIR_DIMENSION_SCALAR, + VSIR_DIMENSION_VEC4, + + VSIR_DIMENSION_COUNT, }; enum vkd3d_shader_src_modifier @@ -585,6 +643,7 @@ enum vkd3d_shader_src_modifier VKD3DSPSM_ABS = 11, VKD3DSPSM_ABSNEG = 12, VKD3DSPSM_NOT = 13, + VKD3DSPSM_COUNT, }; #define VKD3DSP_WRITEMASK_0 0x1u /* .x r */ @@ -599,6 +658,7 @@ enum vkd3d_shader_dst_modifier VKD3DSPDM_SATURATE = 1, VKD3DSPDM_PARTIALPRECISION = 2, VKD3DSPDM_MSAMPCENTROID = 4, + VKD3DSPDM_MASK = 7, }; enum vkd3d_shader_interpolation_mode @@ -611,6 +671,8 @@ enum vkd3d_shader_interpolation_mode VKD3DSIM_LINEAR_NOPERSPECTIVE_CENTROID = 5, VKD3DSIM_LINEAR_SAMPLE = 6, VKD3DSIM_LINEAR_NOPERSPECTIVE_SAMPLE = 7, + + VKD3DSIM_COUNT = 8, }; enum vkd3d_shader_global_flags @@ -622,6 +684,32 @@ enum vkd3d_shader_global_flags VKD3DSGF_SKIP_OPTIMIZATION = 0x10, VKD3DSGF_ENABLE_MINIMUM_PRECISION = 0x20, VKD3DSGF_ENABLE_11_1_DOUBLE_EXTENSIONS = 0x40, + VKD3DSGF_ENABLE_SHADER_EXTENSIONS = 0x80, /* never emitted? */ + VKD3DSGF_BIND_FOR_DURATION = 0x100, + VKD3DSGF_ENABLE_VP_AND_RT_ARRAY_INDEX = 0x200, + VKD3DSGF_ENABLE_INNER_COVERAGE = 0x400, + VKD3DSGF_ENABLE_STENCIL_REF = 0x800, + VKD3DSGF_ENABLE_TILED_RESOURCE_INTRINSICS = 0x1000, + VKD3DSGF_ENABLE_RELAXED_TYPED_UAV_FORMATS = 0x2000, + VKD3DSGF_ENABLE_LVL_9_COMPARISON_FILTERING = 0x4000, + VKD3DSGF_ENABLE_UP_TO_64_UAVS = 0x8000, + VKD3DSGF_ENABLE_UAVS_AT_EVERY_STAGE = 0x10000, + VKD3DSGF_ENABLE_CS4_RAW_STRUCTURED_BUFFERS = 0x20000, + VKD3DSGF_ENABLE_RASTERIZER_ORDERED_VIEWS = 0x40000, + VKD3DSGF_ENABLE_WAVE_INTRINSICS = 0x80000, + VKD3DSGF_ENABLE_INT64 = 0x100000, + VKD3DSGF_ENABLE_VIEWID = 0x200000, + VKD3DSGF_ENABLE_BARYCENTRICS = 0x400000, + VKD3DSGF_FORCE_NATIVE_LOW_PRECISION = 0x800000, + VKD3DSGF_ENABLE_SHADINGRATE = 0x1000000, + VKD3DSGF_ENABLE_RAYTRACING_TIER_1_1 = 0x2000000, + VKD3DSGF_ENABLE_SAMPLER_FEEDBACK = 0x4000000, + VKD3DSGF_ENABLE_ATOMIC_INT64_ON_TYPED_RESOURCE = 0x8000000, + VKD3DSGF_ENABLE_ATOMIC_INT64_ON_GROUP_SHARED = 0x10000000, + VKD3DSGF_ENABLE_DERIVATIVES_IN_MESH_AND_AMPLIFICATION_SHADERS = 0x20000000, + VKD3DSGF_ENABLE_RESOURCE_DESCRIPTOR_HEAP_INDEXING = 0x40000000, + VKD3DSGF_ENABLE_SAMPLER_DESCRIPTOR_HEAP_INDEXING = 0x80000000, + VKD3DSGF_ENABLE_ATOMIC_INT64_ON_DESCRIPTOR_HEAP_RESOURCE = 0x100000000ull, }; enum vkd3d_shader_sync_flags @@ -633,7 +721,8 @@ enum vkd3d_shader_sync_flags enum vkd3d_shader_uav_flags { - VKD3DSUF_GLOBALLY_COHERENT = 0x2, + VKD3DSUF_GLOBALLY_COHERENT = 0x002, + VKD3DSUF_RASTERISER_ORDERED_VIEW = 0x004, VKD3DSUF_ORDER_PRESERVING_COUNTER = 0x100, }; @@ -704,22 +793,29 @@ struct vkd3d_shader_version struct vkd3d_shader_immediate_constant_buffer { - unsigned int vec4_count; + enum vkd3d_data_type data_type; + /* total count is element_count * component_count */ + unsigned int element_count; + unsigned int component_count; uint32_t data[]; }; struct vkd3d_shader_indexable_temp { - struct list entry; unsigned int register_idx; unsigned int register_size; + unsigned int alignment; + enum vkd3d_data_type data_type; unsigned int component_count; + const struct vkd3d_shader_immediate_constant_buffer *initialiser; }; struct vkd3d_shader_register_index { const struct vkd3d_shader_src_param *rel_addr; unsigned int offset; + /* address is known to fall within the object (for optimisation) */ + bool is_in_bounds; }; struct vkd3d_shader_register @@ -730,7 +826,9 @@ struct vkd3d_shader_register enum vkd3d_data_type data_type; struct vkd3d_shader_register_index idx[3]; unsigned int idx_count; - enum vkd3d_immconst_type immconst_type; + enum vsir_dimension dimension; + /* known address alignment for optimisation, or zero */ + unsigned int alignment; union { DWORD immconst_uint[VKD3D_VEC4_SIZE]; @@ -741,13 +839,28 @@ struct vkd3d_shader_register } u; }; -void shader_register_init(struct vkd3d_shader_register *reg, enum vkd3d_shader_register_type reg_type, +void vsir_register_init(struct vkd3d_shader_register *reg, enum vkd3d_shader_register_type reg_type, enum vkd3d_data_type data_type, unsigned int idx_count); +static inline bool vsir_register_is_descriptor(const struct vkd3d_shader_register *reg) +{ + switch (reg->type) + { + case VKD3DSPR_SAMPLER: + case VKD3DSPR_RESOURCE: + case VKD3DSPR_CONSTBUFFER: + case VKD3DSPR_UAV: + return true; + + default: + return false; + } +} + struct vkd3d_shader_dst_param { struct vkd3d_shader_register reg; - DWORD write_mask; + uint32_t write_mask; DWORD modifiers; DWORD shift; }; @@ -848,6 +961,7 @@ struct signature_element unsigned int mask; unsigned int used_mask; enum vkd3d_shader_minimum_precision min_precision; + enum vkd3d_shader_interpolation_mode interpolation_mode; /* Register index / location in the target shader. * If SIGNATURE_TARGET_LOCATION_UNUSED, this element should not be written. */ unsigned int target_location; @@ -860,6 +974,14 @@ struct shader_signature unsigned int element_count; }; +static inline bool vsir_sysval_semantic_is_tess_factor(enum vkd3d_shader_sysval_semantic sysval_semantic) +{ + return sysval_semantic >= VKD3D_SHADER_SV_TESS_FACTOR_QUADEDGE + && sysval_semantic <= VKD3D_SHADER_SV_TESS_FACTOR_LINEDEN; +} + +struct signature_element *vsir_signature_find_element_for_reg(const struct shader_signature *signature, + unsigned int reg_idx, unsigned int write_mask); void shader_signature_cleanup(struct shader_signature *signature); struct vkd3d_shader_desc @@ -871,12 +993,17 @@ struct vkd3d_shader_desc struct shader_signature output_signature; struct shader_signature patch_constant_signature; + unsigned int input_control_point_count, output_control_point_count; + uint32_t temp_count; + unsigned int ssa_count; struct { uint32_t used, external; } flat_constant_count[3]; + + bool use_vocp; }; struct vkd3d_shader_register_semantic @@ -968,8 +1095,15 @@ struct vkd3d_shader_primitive_type unsigned int patch_vertex_count; }; +struct vkd3d_shader_location +{ + const char *source_name; + unsigned int line, column; +}; + struct vkd3d_shader_instruction { + struct vkd3d_shader_location location; enum vkd3d_shader_opcode handler_idx; DWORD flags; unsigned int dst_count; @@ -984,6 +1118,7 @@ struct vkd3d_shader_instruction const struct vkd3d_shader_src_param *predicate; union { + enum vkd3d_shader_global_flags global_flags; struct vkd3d_shader_semantic semantic; struct vkd3d_shader_register_semantic register_semantic; struct vkd3d_shader_primitive_type primitive_type; @@ -1008,26 +1143,22 @@ struct vkd3d_shader_instruction } declaration; }; -void shader_instruction_init(struct vkd3d_shader_instruction *ins, enum vkd3d_shader_opcode handler_idx); - -static inline bool vkd3d_shader_instruction_has_texel_offset(const struct vkd3d_shader_instruction *ins) +static inline bool vkd3d_shader_ver_ge(const struct vkd3d_shader_version *v, unsigned int major, unsigned int minor) { - return ins->texel_offset.u || ins->texel_offset.v || ins->texel_offset.w; + return v->major > major || (v->major == major && v->minor >= minor); } -static inline bool vkd3d_shader_register_is_input(const struct vkd3d_shader_register *reg) +static inline bool vkd3d_shader_ver_le(const struct vkd3d_shader_version *v, unsigned int major, unsigned int minor) { - return reg->type == VKD3DSPR_INPUT || reg->type == VKD3DSPR_INCONTROLPOINT || reg->type == VKD3DSPR_OUTCONTROLPOINT; + return v->major < major || (v->major == major && v->minor <= minor); } -static inline bool vkd3d_shader_register_is_output(const struct vkd3d_shader_register *reg) -{ - return reg->type == VKD3DSPR_OUTPUT || reg->type == VKD3DSPR_COLOROUT; -} +void vsir_instruction_init(struct vkd3d_shader_instruction *ins, const struct vkd3d_shader_location *location, + enum vkd3d_shader_opcode handler_idx); -static inline bool vkd3d_shader_register_is_patch_constant(const struct vkd3d_shader_register *reg) +static inline bool vkd3d_shader_instruction_has_texel_offset(const struct vkd3d_shader_instruction *ins) { - return reg->type == VKD3DSPR_PATCHCONST; + return ins->texel_offset.u || ins->texel_offset.v || ins->texel_offset.w; } static inline bool register_is_constant(const struct vkd3d_shader_register *reg) @@ -1035,12 +1166,6 @@ static inline bool register_is_constant(const struct vkd3d_shader_register *reg) return (reg->type == VKD3DSPR_IMMCONST || reg->type == VKD3DSPR_IMMCONST64); } -struct vkd3d_shader_location -{ - const char *source_name; - unsigned int line, column; -}; - struct vkd3d_shader_param_node { struct vkd3d_shader_param_node *next; @@ -1093,6 +1218,11 @@ bool shader_instruction_array_clone_instruction(struct vkd3d_shader_instruction_ unsigned int dst, unsigned int src); void shader_instruction_array_destroy(struct vkd3d_shader_instruction_array *instructions); +enum vkd3d_shader_config_flags +{ + VKD3D_SHADER_CONFIG_FLAG_FORCE_VALIDATION = 0x00000001, +}; + struct vkd3d_shader_parser { struct vkd3d_shader_message_context *message_context; @@ -1103,6 +1233,8 @@ struct vkd3d_shader_parser struct vkd3d_shader_version shader_version; const struct vkd3d_shader_parser_ops *ops; struct vkd3d_shader_instruction_array instructions; + + uint64_t config_flags; }; struct vkd3d_shader_parser_ops @@ -1248,8 +1380,7 @@ void vkd3d_shader_vnote(struct vkd3d_shader_message_context *context, const stru void vkd3d_shader_vwarning(struct vkd3d_shader_message_context *context, const struct vkd3d_shader_location *location, enum vkd3d_shader_error error, const char *format, va_list args); -void vkd3d_shader_dump_shader(enum vkd3d_shader_source_type source_type, - enum vkd3d_shader_type shader_type, const struct vkd3d_shader_code *shader); +void vkd3d_shader_dump_shader(const struct vkd3d_shader_compile_info *compile_info); void vkd3d_shader_trace_text_(const char *text, size_t size, const char *function); #define vkd3d_shader_trace_text(text, size) \ vkd3d_shader_trace_text_(text, size, __FUNCTION__) @@ -1291,6 +1422,8 @@ int preproc_lexer_parse(const struct vkd3d_shader_compile_info *compile_info, int hlsl_compile_shader(const struct vkd3d_shader_code *hlsl, const struct vkd3d_shader_compile_info *compile_info, struct vkd3d_shader_code *out, struct vkd3d_shader_message_context *message_context); +void vsir_validate(struct vkd3d_shader_parser *parser); + static inline enum vkd3d_shader_component_type vkd3d_component_type_from_data_type( enum vkd3d_data_type data_type) { @@ -1306,6 +1439,8 @@ static inline enum vkd3d_shader_component_type vkd3d_component_type_from_data_ty return VKD3D_SHADER_COMPONENT_INT; case VKD3D_DATA_DOUBLE: return VKD3D_SHADER_COMPONENT_DOUBLE; + case VKD3D_DATA_BOOL: + return VKD3D_SHADER_COMPONENT_BOOL; default: FIXME("Unhandled data type %#x.\n", data_type); /* fall-through */ @@ -1458,23 +1593,6 @@ static inline void *vkd3d_find_struct_(const struct vkd3d_struct *chain, #define VKD3D_DXBC_HEADER_SIZE (8 * sizeof(uint32_t)) #define VKD3D_DXBC_CHUNK_ALIGNMENT sizeof(uint32_t) -#define TAG_AON9 VKD3D_MAKE_TAG('A', 'o', 'n', '9') -#define TAG_DXBC VKD3D_MAKE_TAG('D', 'X', 'B', 'C') -#define TAG_DXIL VKD3D_MAKE_TAG('D', 'X', 'I', 'L') -#define TAG_ISG1 VKD3D_MAKE_TAG('I', 'S', 'G', '1') -#define TAG_ISGN VKD3D_MAKE_TAG('I', 'S', 'G', 'N') -#define TAG_OSG1 VKD3D_MAKE_TAG('O', 'S', 'G', '1') -#define TAG_OSG5 VKD3D_MAKE_TAG('O', 'S', 'G', '5') -#define TAG_OSGN VKD3D_MAKE_TAG('O', 'S', 'G', 'N') -#define TAG_PCSG VKD3D_MAKE_TAG('P', 'C', 'S', 'G') -#define TAG_PSG1 VKD3D_MAKE_TAG('P', 'S', 'G', '1') -#define TAG_RD11 VKD3D_MAKE_TAG('R', 'D', '1', '1') -#define TAG_RDEF VKD3D_MAKE_TAG('R', 'D', 'E', 'F') -#define TAG_RTS0 VKD3D_MAKE_TAG('R', 'T', 'S', '0') -#define TAG_SHDR VKD3D_MAKE_TAG('S', 'H', 'D', 'R') -#define TAG_SHEX VKD3D_MAKE_TAG('S', 'H', 'E', 'X') -#define TAG_TEXT VKD3D_MAKE_TAG('T', 'E', 'X', 'T') - #define DXBC_MAX_SECTION_COUNT 5 struct dxbc_writer diff --git a/libs/vkd3d/libs/vkd3d/command.c b/libs/vkd3d/libs/vkd3d/command.c index 42a98763438..77d0f2751cf 100644 --- a/libs/vkd3d/libs/vkd3d/command.c +++ b/libs/vkd3d/libs/vkd3d/command.c @@ -1942,9 +1942,9 @@ static void d3d12_command_signature_decref(struct d3d12_command_signature *signa } /* ID3D12CommandList */ -static inline struct d3d12_command_list *impl_from_ID3D12GraphicsCommandList3(ID3D12GraphicsCommandList3 *iface) +static inline struct d3d12_command_list *impl_from_ID3D12GraphicsCommandList5(ID3D12GraphicsCommandList5 *iface) { - return CONTAINING_RECORD(iface, struct d3d12_command_list, ID3D12GraphicsCommandList3_iface); + return CONTAINING_RECORD(iface, struct d3d12_command_list, ID3D12GraphicsCommandList5_iface); } static void d3d12_command_list_invalidate_current_framebuffer(struct d3d12_command_list *list) @@ -2290,12 +2290,14 @@ static void d3d12_command_list_track_resource_usage(struct d3d12_command_list *l } } -static HRESULT STDMETHODCALLTYPE d3d12_command_list_QueryInterface(ID3D12GraphicsCommandList3 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_command_list_QueryInterface(ID3D12GraphicsCommandList5 *iface, REFIID iid, void **object) { TRACE("iface %p, iid %s, object %p.\n", iface, debugstr_guid(iid), object); - if (IsEqualGUID(iid, &IID_ID3D12GraphicsCommandList3) + if (IsEqualGUID(iid, &IID_ID3D12GraphicsCommandList5) + || IsEqualGUID(iid, &IID_ID3D12GraphicsCommandList4) + || IsEqualGUID(iid, &IID_ID3D12GraphicsCommandList3) || IsEqualGUID(iid, &IID_ID3D12GraphicsCommandList2) || IsEqualGUID(iid, &IID_ID3D12GraphicsCommandList1) || IsEqualGUID(iid, &IID_ID3D12GraphicsCommandList) @@ -2304,7 +2306,7 @@ static HRESULT STDMETHODCALLTYPE d3d12_command_list_QueryInterface(ID3D12Graphic || IsEqualGUID(iid, &IID_ID3D12Object) || IsEqualGUID(iid, &IID_IUnknown)) { - ID3D12GraphicsCommandList3_AddRef(iface); + ID3D12GraphicsCommandList5_AddRef(iface); *object = iface; return S_OK; } @@ -2315,9 +2317,9 @@ static HRESULT STDMETHODCALLTYPE d3d12_command_list_QueryInterface(ID3D12Graphic return E_NOINTERFACE; } -static ULONG STDMETHODCALLTYPE d3d12_command_list_AddRef(ID3D12GraphicsCommandList3 *iface) +static ULONG STDMETHODCALLTYPE d3d12_command_list_AddRef(ID3D12GraphicsCommandList5 *iface) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); ULONG refcount = InterlockedIncrement(&list->refcount); TRACE("%p increasing refcount to %u.\n", list, refcount); @@ -2330,9 +2332,9 @@ static void vkd3d_pipeline_bindings_cleanup(struct vkd3d_pipeline_bindings *bind vkd3d_free(bindings->vk_uav_counter_views); } -static ULONG STDMETHODCALLTYPE d3d12_command_list_Release(ID3D12GraphicsCommandList3 *iface) +static ULONG STDMETHODCALLTYPE d3d12_command_list_Release(ID3D12GraphicsCommandList5 *iface) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); ULONG refcount = InterlockedDecrement(&list->refcount); TRACE("%p decreasing refcount to %u.\n", list, refcount); @@ -2358,66 +2360,66 @@ static ULONG STDMETHODCALLTYPE d3d12_command_list_Release(ID3D12GraphicsCommandL return refcount; } -static HRESULT STDMETHODCALLTYPE d3d12_command_list_GetPrivateData(ID3D12GraphicsCommandList3 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_command_list_GetPrivateData(ID3D12GraphicsCommandList5 *iface, REFGUID guid, UINT *data_size, void *data) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, guid %s, data_size %p, data %p.\n", iface, debugstr_guid(guid), data_size, data); return vkd3d_get_private_data(&list->private_store, guid, data_size, data); } -static HRESULT STDMETHODCALLTYPE d3d12_command_list_SetPrivateData(ID3D12GraphicsCommandList3 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_command_list_SetPrivateData(ID3D12GraphicsCommandList5 *iface, REFGUID guid, UINT data_size, const void *data) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, guid %s, data_size %u, data %p.\n", iface, debugstr_guid(guid), data_size, data); return vkd3d_set_private_data(&list->private_store, guid, data_size, data); } -static HRESULT STDMETHODCALLTYPE d3d12_command_list_SetPrivateDataInterface(ID3D12GraphicsCommandList3 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_command_list_SetPrivateDataInterface(ID3D12GraphicsCommandList5 *iface, REFGUID guid, const IUnknown *data) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, guid %s, data %p.\n", iface, debugstr_guid(guid), data); return vkd3d_set_private_data_interface(&list->private_store, guid, data); } -static HRESULT STDMETHODCALLTYPE d3d12_command_list_SetName(ID3D12GraphicsCommandList3 *iface, const WCHAR *name) +static HRESULT STDMETHODCALLTYPE d3d12_command_list_SetName(ID3D12GraphicsCommandList5 *iface, const WCHAR *name) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, name %s.\n", iface, debugstr_w(name, list->device->wchar_size)); return name ? S_OK : E_INVALIDARG; } -static HRESULT STDMETHODCALLTYPE d3d12_command_list_GetDevice(ID3D12GraphicsCommandList3 *iface, REFIID iid, void **device) +static HRESULT STDMETHODCALLTYPE d3d12_command_list_GetDevice(ID3D12GraphicsCommandList5 *iface, REFIID iid, void **device) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, iid %s, device %p.\n", iface, debugstr_guid(iid), device); return d3d12_device_query_interface(list->device, iid, device); } -static D3D12_COMMAND_LIST_TYPE STDMETHODCALLTYPE d3d12_command_list_GetType(ID3D12GraphicsCommandList3 *iface) +static D3D12_COMMAND_LIST_TYPE STDMETHODCALLTYPE d3d12_command_list_GetType(ID3D12GraphicsCommandList5 *iface) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p.\n", iface); return list->type; } -static HRESULT STDMETHODCALLTYPE d3d12_command_list_Close(ID3D12GraphicsCommandList3 *iface) +static HRESULT STDMETHODCALLTYPE d3d12_command_list_Close(ID3D12GraphicsCommandList5 *iface) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct vkd3d_vk_device_procs *vk_procs; VkResult vr; @@ -2461,7 +2463,7 @@ static HRESULT STDMETHODCALLTYPE d3d12_command_list_Close(ID3D12GraphicsCommandL static void d3d12_command_list_reset_state(struct d3d12_command_list *list, ID3D12PipelineState *initial_pipeline_state) { - ID3D12GraphicsCommandList3 *iface = &list->ID3D12GraphicsCommandList3_iface; + ID3D12GraphicsCommandList5 *iface = &list->ID3D12GraphicsCommandList5_iface; memset(list->strides, 0, sizeof(list->strides)); list->primitive_topology = D3D_PRIMITIVE_TOPOLOGY_POINTLIST; @@ -2497,14 +2499,14 @@ static void d3d12_command_list_reset_state(struct d3d12_command_list *list, list->descriptor_heap_count = 0; - ID3D12GraphicsCommandList3_SetPipelineState(iface, initial_pipeline_state); + ID3D12GraphicsCommandList5_SetPipelineState(iface, initial_pipeline_state); } -static HRESULT STDMETHODCALLTYPE d3d12_command_list_Reset(ID3D12GraphicsCommandList3 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_command_list_Reset(ID3D12GraphicsCommandList5 *iface, ID3D12CommandAllocator *allocator, ID3D12PipelineState *initial_pipeline_state) { struct d3d12_command_allocator *allocator_impl = unsafe_impl_from_ID3D12CommandAllocator(allocator); - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); HRESULT hr; TRACE("iface %p, allocator %p, initial_pipeline_state %p.\n", @@ -2531,10 +2533,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_command_list_Reset(ID3D12GraphicsCommandL return hr; } -static void STDMETHODCALLTYPE d3d12_command_list_ClearState(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_ClearState(ID3D12GraphicsCommandList5 *iface, ID3D12PipelineState *pipeline_state) { - FIXME("iface %p, pipline_state %p stub!\n", iface, pipeline_state); + FIXME("iface %p, pipeline_state %p stub!\n", iface, pipeline_state); } static bool d3d12_command_list_has_depth_stencil_view(struct d3d12_command_list *list) @@ -3390,11 +3392,11 @@ static void d3d12_command_list_check_index_buffer_strip_cut_value(struct d3d12_c } } -static void STDMETHODCALLTYPE d3d12_command_list_DrawInstanced(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_DrawInstanced(ID3D12GraphicsCommandList5 *iface, UINT vertex_count_per_instance, UINT instance_count, UINT start_vertex_location, UINT start_instance_location) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct vkd3d_vk_device_procs *vk_procs; TRACE("iface %p, vertex_count_per_instance %u, instance_count %u, " @@ -3414,11 +3416,11 @@ static void STDMETHODCALLTYPE d3d12_command_list_DrawInstanced(ID3D12GraphicsCom instance_count, start_vertex_location, start_instance_location)); } -static void STDMETHODCALLTYPE d3d12_command_list_DrawIndexedInstanced(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_DrawIndexedInstanced(ID3D12GraphicsCommandList5 *iface, UINT index_count_per_instance, UINT instance_count, UINT start_vertex_location, INT base_vertex_location, UINT start_instance_location) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct vkd3d_vk_device_procs *vk_procs; TRACE("iface %p, index_count_per_instance %u, instance_count %u, start_vertex_location %u, " @@ -3440,10 +3442,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_DrawIndexedInstanced(ID3D12Grap instance_count, start_vertex_location, base_vertex_location, start_instance_location)); } -static void STDMETHODCALLTYPE d3d12_command_list_Dispatch(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_Dispatch(ID3D12GraphicsCommandList5 *iface, UINT x, UINT y, UINT z) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct vkd3d_vk_device_procs *vk_procs; TRACE("iface %p, x %u, y %u, z %u.\n", iface, x, y, z); @@ -3459,10 +3461,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_Dispatch(ID3D12GraphicsCommandL VK_CALL(vkCmdDispatch(list->vk_command_buffer, x, y, z)); } -static void STDMETHODCALLTYPE d3d12_command_list_CopyBufferRegion(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_CopyBufferRegion(ID3D12GraphicsCommandList5 *iface, ID3D12Resource *dst, UINT64 dst_offset, ID3D12Resource *src, UINT64 src_offset, UINT64 byte_count) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); struct d3d12_resource *dst_resource, *src_resource; const struct vkd3d_vk_device_procs *vk_procs; VkBufferCopy buffer_copy; @@ -3744,11 +3746,11 @@ static bool validate_d3d12_box(const D3D12_BOX *box) && box->back > box->front; } -static void STDMETHODCALLTYPE d3d12_command_list_CopyTextureRegion(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_CopyTextureRegion(ID3D12GraphicsCommandList5 *iface, const D3D12_TEXTURE_COPY_LOCATION *dst, UINT dst_x, UINT dst_y, UINT dst_z, const D3D12_TEXTURE_COPY_LOCATION *src, const D3D12_BOX *src_box) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); struct d3d12_resource *dst_resource, *src_resource; const struct vkd3d_format *src_format, *dst_format; const struct vkd3d_vk_device_procs *vk_procs; @@ -3869,10 +3871,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_CopyTextureRegion(ID3D12Graphic } } -static void STDMETHODCALLTYPE d3d12_command_list_CopyResource(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_CopyResource(ID3D12GraphicsCommandList5 *iface, ID3D12Resource *dst, ID3D12Resource *src) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); struct d3d12_resource *dst_resource, *src_resource; const struct vkd3d_format *dst_format, *src_format; const struct vkd3d_vk_device_procs *vk_procs; @@ -3939,7 +3941,7 @@ static void STDMETHODCALLTYPE d3d12_command_list_CopyResource(ID3D12GraphicsComm } } -static void STDMETHODCALLTYPE d3d12_command_list_CopyTiles(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_CopyTiles(ID3D12GraphicsCommandList5 *iface, ID3D12Resource *tiled_resource, const D3D12_TILED_RESOURCE_COORDINATE *tile_region_start_coordinate, const D3D12_TILE_REGION_SIZE *tile_region_size, ID3D12Resource *buffer, UINT64 buffer_offset, D3D12_TILE_COPY_FLAGS flags) @@ -3950,11 +3952,11 @@ static void STDMETHODCALLTYPE d3d12_command_list_CopyTiles(ID3D12GraphicsCommand buffer, buffer_offset, flags); } -static void STDMETHODCALLTYPE d3d12_command_list_ResolveSubresource(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_ResolveSubresource(ID3D12GraphicsCommandList5 *iface, ID3D12Resource *dst, UINT dst_sub_resource_idx, ID3D12Resource *src, UINT src_sub_resource_idx, DXGI_FORMAT format) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct vkd3d_format *src_format, *dst_format, *vk_format; struct d3d12_resource *dst_resource, *src_resource; const struct vkd3d_vk_device_procs *vk_procs; @@ -4017,10 +4019,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_ResolveSubresource(ID3D12Graphi VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, 1, &vk_image_resolve)); } -static void STDMETHODCALLTYPE d3d12_command_list_IASetPrimitiveTopology(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_IASetPrimitiveTopology(ID3D12GraphicsCommandList5 *iface, D3D12_PRIMITIVE_TOPOLOGY topology) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, topology %#x.\n", iface, topology); @@ -4031,11 +4033,11 @@ static void STDMETHODCALLTYPE d3d12_command_list_IASetPrimitiveTopology(ID3D12Gr d3d12_command_list_invalidate_current_pipeline(list); } -static void STDMETHODCALLTYPE d3d12_command_list_RSSetViewports(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_RSSetViewports(ID3D12GraphicsCommandList5 *iface, UINT viewport_count, const D3D12_VIEWPORT *viewports) { VkViewport vk_viewports[D3D12_VIEWPORT_AND_SCISSORRECT_OBJECT_COUNT_PER_PIPELINE]; - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct vkd3d_vk_device_procs *vk_procs; unsigned int i; @@ -4069,10 +4071,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_RSSetViewports(ID3D12GraphicsCo VK_CALL(vkCmdSetViewport(list->vk_command_buffer, 0, viewport_count, vk_viewports)); } -static void STDMETHODCALLTYPE d3d12_command_list_RSSetScissorRects(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_RSSetScissorRects(ID3D12GraphicsCommandList5 *iface, UINT rect_count, const D3D12_RECT *rects) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); VkRect2D vk_rects[D3D12_VIEWPORT_AND_SCISSORRECT_OBJECT_COUNT_PER_PIPELINE]; const struct vkd3d_vk_device_procs *vk_procs; unsigned int i; @@ -4097,10 +4099,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_RSSetScissorRects(ID3D12Graphic VK_CALL(vkCmdSetScissor(list->vk_command_buffer, 0, rect_count, vk_rects)); } -static void STDMETHODCALLTYPE d3d12_command_list_OMSetBlendFactor(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_OMSetBlendFactor(ID3D12GraphicsCommandList5 *iface, const FLOAT blend_factor[4]) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct vkd3d_vk_device_procs *vk_procs; TRACE("iface %p, blend_factor %p.\n", iface, blend_factor); @@ -4109,10 +4111,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_OMSetBlendFactor(ID3D12Graphics VK_CALL(vkCmdSetBlendConstants(list->vk_command_buffer, blend_factor)); } -static void STDMETHODCALLTYPE d3d12_command_list_OMSetStencilRef(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_OMSetStencilRef(ID3D12GraphicsCommandList5 *iface, UINT stencil_ref) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct vkd3d_vk_device_procs *vk_procs; TRACE("iface %p, stencil_ref %u.\n", iface, stencil_ref); @@ -4121,11 +4123,11 @@ static void STDMETHODCALLTYPE d3d12_command_list_OMSetStencilRef(ID3D12GraphicsC VK_CALL(vkCmdSetStencilReference(list->vk_command_buffer, VK_STENCIL_FRONT_AND_BACK, stencil_ref)); } -static void STDMETHODCALLTYPE d3d12_command_list_SetPipelineState(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetPipelineState(ID3D12GraphicsCommandList5 *iface, ID3D12PipelineState *pipeline_state) { struct d3d12_pipeline_state *state = unsafe_impl_from_ID3D12PipelineState(pipeline_state); - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, pipeline_state %p.\n", iface, pipeline_state); @@ -4176,10 +4178,10 @@ static unsigned int d3d12_find_ds_multiplanar_transition(const D3D12_RESOURCE_BA return 0; } -static void STDMETHODCALLTYPE d3d12_command_list_ResourceBarrier(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_ResourceBarrier(ID3D12GraphicsCommandList5 *iface, UINT barrier_count, const D3D12_RESOURCE_BARRIER *barriers) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); bool have_aliasing_barriers = false, have_split_barriers = false; const struct vkd3d_vk_device_procs *vk_procs; const struct vkd3d_vulkan_info *vk_info; @@ -4402,13 +4404,13 @@ static void STDMETHODCALLTYPE d3d12_command_list_ResourceBarrier(ID3D12GraphicsC WARN("Issuing split barrier(s) on D3D12_RESOURCE_BARRIER_FLAG_END_ONLY.\n"); } -static void STDMETHODCALLTYPE d3d12_command_list_ExecuteBundle(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_ExecuteBundle(ID3D12GraphicsCommandList5 *iface, ID3D12GraphicsCommandList *command_list) { FIXME("iface %p, command_list %p stub!\n", iface, command_list); } -static void STDMETHODCALLTYPE d3d12_command_list_SetDescriptorHeaps(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetDescriptorHeaps(ID3D12GraphicsCommandList5 *iface, UINT heap_count, ID3D12DescriptorHeap *const *heaps) { TRACE("iface %p, heap_count %u, heaps %p.\n", iface, heap_count, heaps); @@ -4434,10 +4436,10 @@ static void d3d12_command_list_set_root_signature(struct d3d12_command_list *lis d3d12_command_list_invalidate_root_parameters(list, bind_point); } -static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootSignature(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootSignature(ID3D12GraphicsCommandList5 *iface, ID3D12RootSignature *root_signature) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_signature %p.\n", iface, root_signature); @@ -4445,10 +4447,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootSignature(ID3D12G unsafe_impl_from_ID3D12RootSignature(root_signature)); } -static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRootSignature(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRootSignature(ID3D12GraphicsCommandList5 *iface, ID3D12RootSignature *root_signature) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_signature %p.\n", iface, root_signature); @@ -4487,10 +4489,10 @@ static void d3d12_command_list_set_descriptor_table(struct d3d12_command_list *l bindings->descriptor_table_active_mask |= (uint64_t)1 << index; } -static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootDescriptorTable(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootDescriptorTable(ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, D3D12_GPU_DESCRIPTOR_HANDLE base_descriptor) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, base_descriptor %#"PRIx64".\n", iface, root_parameter_index, base_descriptor.ptr); @@ -4499,10 +4501,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootDescriptorTable(I root_parameter_index, base_descriptor); } -static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRootDescriptorTable(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRootDescriptorTable(ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, D3D12_GPU_DESCRIPTOR_HANDLE base_descriptor) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, base_descriptor %#"PRIx64".\n", iface, root_parameter_index, base_descriptor.ptr); @@ -4524,10 +4526,10 @@ static void d3d12_command_list_set_root_constants(struct d3d12_command_list *lis c->stage_flags, c->offset + offset * sizeof(uint32_t), count * sizeof(uint32_t), data)); } -static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRoot32BitConstant(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRoot32BitConstant(ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, UINT data, UINT dst_offset) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, data 0x%08x, dst_offset %u.\n", iface, root_parameter_index, data, dst_offset); @@ -4536,10 +4538,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRoot32BitConstant(ID3 root_parameter_index, dst_offset, 1, &data); } -static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRoot32BitConstant(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRoot32BitConstant(ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, UINT data, UINT dst_offset) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, data 0x%08x, dst_offset %u.\n", iface, root_parameter_index, data, dst_offset); @@ -4548,10 +4550,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRoot32BitConstant(ID root_parameter_index, dst_offset, 1, &data); } -static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRoot32BitConstants(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRoot32BitConstants(ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, UINT constant_count, const void *data, UINT dst_offset) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, constant_count %u, data %p, dst_offset %u.\n", iface, root_parameter_index, constant_count, data, dst_offset); @@ -4560,10 +4562,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRoot32BitConstants(ID root_parameter_index, dst_offset, constant_count, data); } -static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRoot32BitConstants(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRoot32BitConstants(ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, UINT constant_count, const void *data, UINT dst_offset) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, constant_count %u, data %p, dst_offset %u.\n", iface, root_parameter_index, constant_count, data, dst_offset); @@ -4625,9 +4627,9 @@ static void d3d12_command_list_set_root_cbv(struct d3d12_command_list *list, } static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootConstantBufferView( - ID3D12GraphicsCommandList3 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) + ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, address %#"PRIx64".\n", iface, root_parameter_index, address); @@ -4636,9 +4638,9 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootConstantBufferVie } static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRootConstantBufferView( - ID3D12GraphicsCommandList3 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) + ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, address %#"PRIx64".\n", iface, root_parameter_index, address); @@ -4697,9 +4699,9 @@ static void d3d12_command_list_set_root_descriptor(struct d3d12_command_list *li } static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootShaderResourceView( - ID3D12GraphicsCommandList3 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) + ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, address %#"PRIx64".\n", iface, root_parameter_index, address); @@ -4709,9 +4711,9 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootShaderResourceVie } static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRootShaderResourceView( - ID3D12GraphicsCommandList3 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) + ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, address %#"PRIx64".\n", iface, root_parameter_index, address); @@ -4721,9 +4723,9 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRootShaderResourceVi } static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootUnorderedAccessView( - ID3D12GraphicsCommandList3 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) + ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, address %#"PRIx64".\n", iface, root_parameter_index, address); @@ -4733,9 +4735,9 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetComputeRootUnorderedAccessVi } static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRootUnorderedAccessView( - ID3D12GraphicsCommandList3 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) + ID3D12GraphicsCommandList5 *iface, UINT root_parameter_index, D3D12_GPU_VIRTUAL_ADDRESS address) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); TRACE("iface %p, root_parameter_index %u, address %#"PRIx64".\n", iface, root_parameter_index, address); @@ -4744,10 +4746,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetGraphicsRootUnorderedAccessV root_parameter_index, address); } -static void STDMETHODCALLTYPE d3d12_command_list_IASetIndexBuffer(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_IASetIndexBuffer(ID3D12GraphicsCommandList5 *iface, const D3D12_INDEX_BUFFER_VIEW *view) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct vkd3d_vk_device_procs *vk_procs; struct d3d12_resource *resource; enum VkIndexType index_type; @@ -4787,10 +4789,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_IASetIndexBuffer(ID3D12Graphics view->BufferLocation - resource->gpu_address, index_type)); } -static void STDMETHODCALLTYPE d3d12_command_list_IASetVertexBuffers(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_IASetVertexBuffers(ID3D12GraphicsCommandList5 *iface, UINT start_slot, UINT view_count, const D3D12_VERTEX_BUFFER_VIEW *views) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct vkd3d_null_resources *null_resources; struct vkd3d_gpu_va_allocator *gpu_va_allocator; VkDeviceSize offsets[ARRAY_SIZE(list->strides)]; @@ -4845,10 +4847,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_IASetVertexBuffers(ID3D12Graphi d3d12_command_list_invalidate_current_pipeline(list); } -static void STDMETHODCALLTYPE d3d12_command_list_SOSetTargets(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SOSetTargets(ID3D12GraphicsCommandList5 *iface, UINT start_slot, UINT view_count, const D3D12_STREAM_OUTPUT_BUFFER_VIEW *views) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); VkDeviceSize offsets[ARRAY_SIZE(list->so_counter_buffers)]; VkDeviceSize sizes[ARRAY_SIZE(list->so_counter_buffers)]; VkBuffer buffers[ARRAY_SIZE(list->so_counter_buffers)]; @@ -4910,11 +4912,11 @@ static void STDMETHODCALLTYPE d3d12_command_list_SOSetTargets(ID3D12GraphicsComm VK_CALL(vkCmdBindTransformFeedbackBuffersEXT(list->vk_command_buffer, first, count, buffers, offsets, sizes)); } -static void STDMETHODCALLTYPE d3d12_command_list_OMSetRenderTargets(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_OMSetRenderTargets(ID3D12GraphicsCommandList5 *iface, UINT render_target_descriptor_count, const D3D12_CPU_DESCRIPTOR_HANDLE *render_target_descriptors, BOOL single_descriptor_handle, const D3D12_CPU_DESCRIPTOR_HANDLE *depth_stencil_descriptor) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct d3d12_rtv_desc *rtv_desc; const struct d3d12_dsv_desc *dsv_desc; VkFormat prev_dsv_format; @@ -5115,12 +5117,12 @@ static void d3d12_command_list_clear(struct d3d12_command_list *list, } } -static void STDMETHODCALLTYPE d3d12_command_list_ClearDepthStencilView(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_ClearDepthStencilView(ID3D12GraphicsCommandList5 *iface, D3D12_CPU_DESCRIPTOR_HANDLE dsv, D3D12_CLEAR_FLAGS flags, float depth, UINT8 stencil, UINT rect_count, const D3D12_RECT *rects) { const union VkClearValue clear_value = {.depthStencil = {depth, stencil}}; - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct d3d12_dsv_desc *dsv_desc = d3d12_dsv_desc_from_cpu_handle(dsv); struct VkAttachmentDescription attachment_desc; struct VkAttachmentReference ds_reference; @@ -5164,10 +5166,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_ClearDepthStencilView(ID3D12Gra &clear_value, rect_count, rects); } -static void STDMETHODCALLTYPE d3d12_command_list_ClearRenderTargetView(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_ClearRenderTargetView(ID3D12GraphicsCommandList5 *iface, D3D12_CPU_DESCRIPTOR_HANDLE rtv, const FLOAT color[4], UINT rect_count, const D3D12_RECT *rects) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const struct d3d12_rtv_desc *rtv_desc = d3d12_rtv_desc_from_cpu_handle(rtv); struct VkAttachmentDescription attachment_desc; struct VkAttachmentReference color_reference; @@ -5412,11 +5414,11 @@ static const struct vkd3d_format *vkd3d_fixup_clear_uav_uint_colour(struct d3d12 } } -static void STDMETHODCALLTYPE d3d12_command_list_ClearUnorderedAccessViewUint(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_ClearUnorderedAccessViewUint(ID3D12GraphicsCommandList5 *iface, D3D12_GPU_DESCRIPTOR_HANDLE gpu_handle, D3D12_CPU_DESCRIPTOR_HANDLE cpu_handle, ID3D12Resource *resource, const UINT values[4], UINT rect_count, const D3D12_RECT *rects) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); struct vkd3d_view *descriptor, *uint_view = NULL; struct d3d12_device *device = list->device; struct vkd3d_texture_view_desc view_desc; @@ -5461,6 +5463,8 @@ static void STDMETHODCALLTYPE d3d12_command_list_ClearUnorderedAccessViewUint(ID view_desc.miplevel_count = 1; view_desc.layer_idx = view->info.texture.layer_idx; view_desc.layer_count = view->info.texture.layer_count; + view_desc.vk_image_aspect = VK_IMAGE_ASPECT_COLOR_BIT; + view_desc.usage = VK_IMAGE_USAGE_STORAGE_BIT; if (!vkd3d_create_texture_view(device, VKD3D_DESCRIPTOR_MAGIC_UAV, resource_impl->u.vk_image, &view_desc, &uint_view)) @@ -5478,11 +5482,11 @@ static void STDMETHODCALLTYPE d3d12_command_list_ClearUnorderedAccessViewUint(ID vkd3d_view_decref(uint_view, device); } -static void STDMETHODCALLTYPE d3d12_command_list_ClearUnorderedAccessViewFloat(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_ClearUnorderedAccessViewFloat(ID3D12GraphicsCommandList5 *iface, D3D12_GPU_DESCRIPTOR_HANDLE gpu_handle, D3D12_CPU_DESCRIPTOR_HANDLE cpu_handle, ID3D12Resource *resource, const float values[4], UINT rect_count, const D3D12_RECT *rects) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); struct d3d12_resource *resource_impl; VkClearColorValue colour; struct vkd3d_view *view; @@ -5498,16 +5502,16 @@ static void STDMETHODCALLTYPE d3d12_command_list_ClearUnorderedAccessViewFloat(I d3d12_command_list_clear_uav(list, resource_impl, view, &colour, rect_count, rects); } -static void STDMETHODCALLTYPE d3d12_command_list_DiscardResource(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_DiscardResource(ID3D12GraphicsCommandList5 *iface, ID3D12Resource *resource, const D3D12_DISCARD_REGION *region) { FIXME_ONCE("iface %p, resource %p, region %p stub!\n", iface, resource, region); } -static void STDMETHODCALLTYPE d3d12_command_list_BeginQuery(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_BeginQuery(ID3D12GraphicsCommandList5 *iface, ID3D12QueryHeap *heap, D3D12_QUERY_TYPE type, UINT index) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); struct d3d12_query_heap *query_heap = unsafe_impl_from_ID3D12QueryHeap(heap); const struct vkd3d_vk_device_procs *vk_procs; VkQueryControlFlags flags = 0; @@ -5534,10 +5538,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_BeginQuery(ID3D12GraphicsComman VK_CALL(vkCmdBeginQuery(list->vk_command_buffer, query_heap->vk_query_pool, index, flags)); } -static void STDMETHODCALLTYPE d3d12_command_list_EndQuery(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_EndQuery(ID3D12GraphicsCommandList5 *iface, ID3D12QueryHeap *heap, D3D12_QUERY_TYPE type, UINT index) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); struct d3d12_query_heap *query_heap = unsafe_impl_from_ID3D12QueryHeap(heap); const struct vkd3d_vk_device_procs *vk_procs; @@ -5579,12 +5583,12 @@ static size_t get_query_stride(D3D12_QUERY_TYPE type) return sizeof(uint64_t); } -static void STDMETHODCALLTYPE d3d12_command_list_ResolveQueryData(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_ResolveQueryData(ID3D12GraphicsCommandList5 *iface, ID3D12QueryHeap *heap, D3D12_QUERY_TYPE type, UINT start_index, UINT query_count, ID3D12Resource *dst_buffer, UINT64 aligned_dst_buffer_offset) { const struct d3d12_query_heap *query_heap = unsafe_impl_from_ID3D12QueryHeap(heap); - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); struct d3d12_resource *buffer = unsafe_impl_from_ID3D12Resource(dst_buffer); const struct vkd3d_vk_device_procs *vk_procs; unsigned int i, first, count; @@ -5660,10 +5664,10 @@ static void STDMETHODCALLTYPE d3d12_command_list_ResolveQueryData(ID3D12Graphics } } -static void STDMETHODCALLTYPE d3d12_command_list_SetPredication(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetPredication(ID3D12GraphicsCommandList5 *iface, ID3D12Resource *buffer, UINT64 aligned_buffer_offset, D3D12_PREDICATION_OP operation) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); struct d3d12_resource *resource = unsafe_impl_from_ID3D12Resource(buffer); const struct vkd3d_vulkan_info *vk_info = &list->device->vk_info; const struct vkd3d_vk_device_procs *vk_procs; @@ -5732,19 +5736,19 @@ static void STDMETHODCALLTYPE d3d12_command_list_SetPredication(ID3D12GraphicsCo } } -static void STDMETHODCALLTYPE d3d12_command_list_SetMarker(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetMarker(ID3D12GraphicsCommandList5 *iface, UINT metadata, const void *data, UINT size) { FIXME("iface %p, metadata %#x, data %p, size %u stub!\n", iface, metadata, data, size); } -static void STDMETHODCALLTYPE d3d12_command_list_BeginEvent(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_BeginEvent(ID3D12GraphicsCommandList5 *iface, UINT metadata, const void *data, UINT size) { FIXME("iface %p, metadata %#x, data %p, size %u stub!\n", iface, metadata, data, size); } -static void STDMETHODCALLTYPE d3d12_command_list_EndEvent(ID3D12GraphicsCommandList3 *iface) +static void STDMETHODCALLTYPE d3d12_command_list_EndEvent(ID3D12GraphicsCommandList5 *iface) { FIXME("iface %p stub!\n", iface); } @@ -5753,14 +5757,14 @@ STATIC_ASSERT(sizeof(VkDispatchIndirectCommand) == sizeof(D3D12_DISPATCH_ARGUMEN STATIC_ASSERT(sizeof(VkDrawIndexedIndirectCommand) == sizeof(D3D12_DRAW_INDEXED_ARGUMENTS)); STATIC_ASSERT(sizeof(VkDrawIndirectCommand) == sizeof(D3D12_DRAW_ARGUMENTS)); -static void STDMETHODCALLTYPE d3d12_command_list_ExecuteIndirect(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_ExecuteIndirect(ID3D12GraphicsCommandList5 *iface, ID3D12CommandSignature *command_signature, UINT max_command_count, ID3D12Resource *arg_buffer, UINT64 arg_buffer_offset, ID3D12Resource *count_buffer, UINT64 count_buffer_offset) { struct d3d12_command_signature *sig_impl = unsafe_impl_from_ID3D12CommandSignature(command_signature); struct d3d12_resource *count_impl = unsafe_impl_from_ID3D12Resource(count_buffer); struct d3d12_resource *arg_impl = unsafe_impl_from_ID3D12Resource(arg_buffer); - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); const D3D12_COMMAND_SIGNATURE_DESC *signature_desc; const struct vkd3d_vk_device_procs *vk_procs; unsigned int i; @@ -5859,7 +5863,7 @@ static void STDMETHODCALLTYPE d3d12_command_list_ExecuteIndirect(ID3D12GraphicsC d3d12_command_signature_decref(sig_impl); } -static void STDMETHODCALLTYPE d3d12_command_list_AtomicCopyBufferUINT(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_AtomicCopyBufferUINT(ID3D12GraphicsCommandList5 *iface, ID3D12Resource *dst_buffer, UINT64 dst_offset, ID3D12Resource *src_buffer, UINT64 src_offset, UINT dependent_resource_count, ID3D12Resource * const *dependent_resources, @@ -5872,7 +5876,7 @@ static void STDMETHODCALLTYPE d3d12_command_list_AtomicCopyBufferUINT(ID3D12Grap dependent_resource_count, dependent_resources, dependent_sub_resource_ranges); } -static void STDMETHODCALLTYPE d3d12_command_list_AtomicCopyBufferUINT64(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_AtomicCopyBufferUINT64(ID3D12GraphicsCommandList5 *iface, ID3D12Resource *dst_buffer, UINT64 dst_offset, ID3D12Resource *src_buffer, UINT64 src_offset, UINT dependent_resource_count, ID3D12Resource * const *dependent_resources, @@ -5885,20 +5889,20 @@ static void STDMETHODCALLTYPE d3d12_command_list_AtomicCopyBufferUINT64(ID3D12Gr dependent_resource_count, dependent_resources, dependent_sub_resource_ranges); } -static void STDMETHODCALLTYPE d3d12_command_list_OMSetDepthBounds(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_OMSetDepthBounds(ID3D12GraphicsCommandList5 *iface, FLOAT min, FLOAT max) { FIXME("iface %p, min %.8e, max %.8e stub!\n", iface, min, max); } -static void STDMETHODCALLTYPE d3d12_command_list_SetSamplePositions(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetSamplePositions(ID3D12GraphicsCommandList5 *iface, UINT sample_count, UINT pixel_count, D3D12_SAMPLE_POSITION *sample_positions) { FIXME("iface %p, sample_count %u, pixel_count %u, sample_positions %p stub!\n", iface, sample_count, pixel_count, sample_positions); } -static void STDMETHODCALLTYPE d3d12_command_list_ResolveSubresourceRegion(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_ResolveSubresourceRegion(ID3D12GraphicsCommandList5 *iface, ID3D12Resource *dst_resource, UINT dst_sub_resource_idx, UINT dst_x, UINT dst_y, ID3D12Resource *src_resource, UINT src_sub_resource_idx, D3D12_RECT *src_rect, DXGI_FORMAT format, D3D12_RESOLVE_MODE mode) @@ -5910,16 +5914,16 @@ static void STDMETHODCALLTYPE d3d12_command_list_ResolveSubresourceRegion(ID3D12 src_resource, src_sub_resource_idx, src_rect, format, mode); } -static void STDMETHODCALLTYPE d3d12_command_list_SetViewInstanceMask(ID3D12GraphicsCommandList3 *iface, UINT mask) +static void STDMETHODCALLTYPE d3d12_command_list_SetViewInstanceMask(ID3D12GraphicsCommandList5 *iface, UINT mask) { FIXME("iface %p, mask %#x stub!\n", iface, mask); } -static void STDMETHODCALLTYPE d3d12_command_list_WriteBufferImmediate(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_WriteBufferImmediate(ID3D12GraphicsCommandList5 *iface, UINT count, const D3D12_WRITEBUFFERIMMEDIATE_PARAMETER *parameters, const D3D12_WRITEBUFFERIMMEDIATE_MODE *modes) { - struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList3(iface); + struct d3d12_command_list *list = impl_from_ID3D12GraphicsCommandList5(iface); struct d3d12_resource *resource; unsigned int i; @@ -5932,13 +5936,88 @@ static void STDMETHODCALLTYPE d3d12_command_list_WriteBufferImmediate(ID3D12Grap } } -static void STDMETHODCALLTYPE d3d12_command_list_SetProtectedResourceSession(ID3D12GraphicsCommandList3 *iface, +static void STDMETHODCALLTYPE d3d12_command_list_SetProtectedResourceSession(ID3D12GraphicsCommandList5 *iface, ID3D12ProtectedResourceSession *protected_session) { FIXME("iface %p, protected_session %p stub!\n", iface, protected_session); } -static const struct ID3D12GraphicsCommandList3Vtbl d3d12_command_list_vtbl = +static void STDMETHODCALLTYPE d3d12_command_list_BeginRenderPass(ID3D12GraphicsCommandList5 *iface, + UINT count, const D3D12_RENDER_PASS_RENDER_TARGET_DESC *render_targets, + const D3D12_RENDER_PASS_DEPTH_STENCIL_DESC *depth_stencil, D3D12_RENDER_PASS_FLAGS flags) +{ + FIXME("iface %p, count %u, render_targets %p, depth_stencil %p, flags %#x stub!\n", iface, + count, render_targets, depth_stencil, flags); +} + +static void STDMETHODCALLTYPE d3d12_command_list_EndRenderPass(ID3D12GraphicsCommandList5 *iface) +{ + FIXME("iface %p stub!\n", iface); +} + +static void STDMETHODCALLTYPE d3d12_command_list_InitializeMetaCommand(ID3D12GraphicsCommandList5 *iface, + ID3D12MetaCommand *meta_command, const void *parameters_data, SIZE_T data_size_in_bytes) +{ + FIXME("iface %p, meta_command %p, parameters_data %p, data_size_in_bytes %lu stub!\n", iface, + meta_command, parameters_data, data_size_in_bytes); +} + +static void STDMETHODCALLTYPE d3d12_command_list_ExecuteMetaCommand(ID3D12GraphicsCommandList5 *iface, + ID3D12MetaCommand *meta_command, const void *parameters_data, SIZE_T data_size_in_bytes) +{ + FIXME("iface %p, meta_command %p, parameters_data %p, data_size_in_bytes %lu stub!\n", iface, + meta_command, parameters_data, data_size_in_bytes); +} + +static void STDMETHODCALLTYPE d3d12_command_list_BuildRaytracingAccelerationStructure(ID3D12GraphicsCommandList5 *iface, + const D3D12_BUILD_RAYTRACING_ACCELERATION_STRUCTURE_DESC *desc, UINT count, + const D3D12_RAYTRACING_ACCELERATION_STRUCTURE_POSTBUILD_INFO_DESC *postbuild_info_descs) +{ + FIXME("iface %p, desc %p, count %u, postbuild_info_descs %p stub!\n", iface, desc, count, postbuild_info_descs); +} + +static void STDMETHODCALLTYPE d3d12_command_list_EmitRaytracingAccelerationStructurePostbuildInfo(ID3D12GraphicsCommandList5 *iface, + const D3D12_RAYTRACING_ACCELERATION_STRUCTURE_POSTBUILD_INFO_DESC *desc, + UINT structures_count, const D3D12_GPU_VIRTUAL_ADDRESS *src_structure_data) +{ + FIXME("iface %p, desc %p, structures_count %u, src_structure_data %p stub!\n", + iface, desc, structures_count, src_structure_data); +} + +static void STDMETHODCALLTYPE d3d12_command_list_CopyRaytracingAccelerationStructure(ID3D12GraphicsCommandList5 *iface, + D3D12_GPU_VIRTUAL_ADDRESS dst_structure_data, + D3D12_GPU_VIRTUAL_ADDRESS src_structure_data, + D3D12_RAYTRACING_ACCELERATION_STRUCTURE_COPY_MODE mode) +{ + FIXME("iface %p, dst_structure_data %#"PRIx64", src_structure_data %#"PRIx64", mode %u stub!\n", + iface, dst_structure_data, src_structure_data, mode); +} + +static void STDMETHODCALLTYPE d3d12_command_list_SetPipelineState1(ID3D12GraphicsCommandList5 *iface, + ID3D12StateObject *state_object) +{ + FIXME("iface %p, state_object %p stub!\n", iface, state_object); +} + +static void STDMETHODCALLTYPE d3d12_command_list_DispatchRays(ID3D12GraphicsCommandList5 *iface, + const D3D12_DISPATCH_RAYS_DESC *desc) +{ + FIXME("iface %p, desc %p stub!\n", iface, desc); +} + +static void STDMETHODCALLTYPE d3d12_command_list_RSSetShadingRate(ID3D12GraphicsCommandList5 *iface, + D3D12_SHADING_RATE rate, const D3D12_SHADING_RATE_COMBINER *combiners) +{ + FIXME("iface %p, rate %#x, combiners %p stub!\n", iface, rate, combiners); +} + +static void STDMETHODCALLTYPE d3d12_command_list_RSSetShadingRateImage(ID3D12GraphicsCommandList5 *iface, + ID3D12Resource *rate_image) +{ + FIXME("iface %p, rate_image %p stub!\n", iface, rate_image); +} + +static const struct ID3D12GraphicsCommandList5Vtbl d3d12_command_list_vtbl = { /* IUnknown methods */ d3d12_command_list_QueryInterface, @@ -6016,6 +6095,19 @@ static const struct ID3D12GraphicsCommandList3Vtbl d3d12_command_list_vtbl = d3d12_command_list_WriteBufferImmediate, /* ID3D12GraphicsCommandList3 methods */ d3d12_command_list_SetProtectedResourceSession, + /* ID3D12GraphicsCommandList4 methods */ + d3d12_command_list_BeginRenderPass, + d3d12_command_list_EndRenderPass, + d3d12_command_list_InitializeMetaCommand, + d3d12_command_list_ExecuteMetaCommand, + d3d12_command_list_BuildRaytracingAccelerationStructure, + d3d12_command_list_EmitRaytracingAccelerationStructurePostbuildInfo, + d3d12_command_list_CopyRaytracingAccelerationStructure, + d3d12_command_list_SetPipelineState1, + d3d12_command_list_DispatchRays, + /* ID3D12GraphicsCommandList5 methods */ + d3d12_command_list_RSSetShadingRate, + d3d12_command_list_RSSetShadingRateImage, }; static struct d3d12_command_list *unsafe_impl_from_ID3D12CommandList(ID3D12CommandList *iface) @@ -6023,7 +6115,7 @@ static struct d3d12_command_list *unsafe_impl_from_ID3D12CommandList(ID3D12Comma if (!iface) return NULL; assert(iface->lpVtbl == (struct ID3D12CommandListVtbl *)&d3d12_command_list_vtbl); - return CONTAINING_RECORD(iface, struct d3d12_command_list, ID3D12GraphicsCommandList3_iface); + return CONTAINING_RECORD(iface, struct d3d12_command_list, ID3D12GraphicsCommandList5_iface); } static HRESULT d3d12_command_list_init(struct d3d12_command_list *list, struct d3d12_device *device, @@ -6032,7 +6124,7 @@ static HRESULT d3d12_command_list_init(struct d3d12_command_list *list, struct d { HRESULT hr; - list->ID3D12GraphicsCommandList3_iface.lpVtbl = &d3d12_command_list_vtbl; + list->ID3D12GraphicsCommandList5_iface.lpVtbl = &d3d12_command_list_vtbl; list->refcount = 1; list->type = type; diff --git a/libs/vkd3d/libs/vkd3d/device.c b/libs/vkd3d/libs/vkd3d/device.c index c33061073a3..79028fc3dd8 100644 --- a/libs/vkd3d/libs/vkd3d/device.c +++ b/libs/vkd3d/libs/vkd3d/device.c @@ -83,6 +83,7 @@ static const struct vkd3d_optional_extension_info optional_device_extensions[] = VK_EXTENSION(KHR_DRAW_INDIRECT_COUNT, KHR_draw_indirect_count), VK_EXTENSION(KHR_GET_MEMORY_REQUIREMENTS_2, KHR_get_memory_requirements2), VK_EXTENSION(KHR_IMAGE_FORMAT_LIST, KHR_image_format_list), + VK_EXTENSION(KHR_MAINTENANCE2, KHR_maintenance2), VK_EXTENSION(KHR_MAINTENANCE3, KHR_maintenance3), VK_EXTENSION(KHR_PUSH_DESCRIPTOR, KHR_push_descriptor), VK_EXTENSION(KHR_SAMPLER_MIRROR_CLAMP_TO_EDGE, KHR_sampler_mirror_clamp_to_edge), @@ -1521,9 +1522,7 @@ static HRESULT vkd3d_init_device_caps(struct d3d12_device *device, device->feature_options1.ExpandedComputeResourceStates = TRUE; device->feature_options1.Int64ShaderOps = features->shaderInt64; - /* Depth bounds test is enabled in D3D12_DEPTH_STENCIL_DESC1, which is not - * supported. */ - device->feature_options2.DepthBoundsTestSupported = FALSE; + device->feature_options2.DepthBoundsTestSupported = features->depthBounds; /* d3d12_command_list_SetSamplePositions() is not implemented. */ device->feature_options2.ProgrammableSamplePositionsTier = D3D12_PROGRAMMABLE_SAMPLE_POSITIONS_TIER_NOT_SUPPORTED; @@ -2456,17 +2455,21 @@ static void vkd3d_desc_object_cache_cleanup(struct vkd3d_desc_object_cache *cach } /* ID3D12Device */ -static inline struct d3d12_device *impl_from_ID3D12Device1(ID3D12Device1 *iface) +static inline struct d3d12_device *impl_from_ID3D12Device5(ID3D12Device5 *iface) { - return CONTAINING_RECORD(iface, struct d3d12_device, ID3D12Device1_iface); + return CONTAINING_RECORD(iface, struct d3d12_device, ID3D12Device5_iface); } -static HRESULT STDMETHODCALLTYPE d3d12_device_QueryInterface(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_QueryInterface(ID3D12Device5 *iface, REFIID riid, void **object) { TRACE("iface %p, riid %s, object %p.\n", iface, debugstr_guid(riid), object); - if (IsEqualGUID(riid, &IID_ID3D12Device1) + if (IsEqualGUID(riid, &IID_ID3D12Device5) + || IsEqualGUID(riid, &IID_ID3D12Device4) + || IsEqualGUID(riid, &IID_ID3D12Device3) + || IsEqualGUID(riid, &IID_ID3D12Device2) + || IsEqualGUID(riid, &IID_ID3D12Device1) || IsEqualGUID(riid, &IID_ID3D12Device) || IsEqualGUID(riid, &IID_ID3D12Object) || IsEqualGUID(riid, &IID_IUnknown)) @@ -2482,9 +2485,9 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_QueryInterface(ID3D12Device1 *ifac return E_NOINTERFACE; } -static ULONG STDMETHODCALLTYPE d3d12_device_AddRef(ID3D12Device1 *iface) +static ULONG STDMETHODCALLTYPE d3d12_device_AddRef(ID3D12Device5 *iface) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); ULONG refcount = InterlockedIncrement(&device->refcount); TRACE("%p increasing refcount to %u.\n", device, refcount); @@ -2492,9 +2495,9 @@ static ULONG STDMETHODCALLTYPE d3d12_device_AddRef(ID3D12Device1 *iface) return refcount; } -static ULONG STDMETHODCALLTYPE d3d12_device_Release(ID3D12Device1 *iface) +static ULONG STDMETHODCALLTYPE d3d12_device_Release(ID3D12Device5 *iface) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); ULONG refcount = InterlockedDecrement(&device->refcount); TRACE("%p decreasing refcount to %u.\n", device, refcount); @@ -2528,10 +2531,10 @@ static ULONG STDMETHODCALLTYPE d3d12_device_Release(ID3D12Device1 *iface) return refcount; } -static HRESULT STDMETHODCALLTYPE d3d12_device_GetPrivateData(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_GetPrivateData(ID3D12Device5 *iface, REFGUID guid, UINT *data_size, void *data) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); TRACE("iface %p, guid %s, data_size %p, data %p.\n", iface, debugstr_guid(guid), data_size, data); @@ -2539,10 +2542,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_GetPrivateData(ID3D12Device1 *ifac return vkd3d_get_private_data(&device->private_store, guid, data_size, data); } -static HRESULT STDMETHODCALLTYPE d3d12_device_SetPrivateData(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_SetPrivateData(ID3D12Device5 *iface, REFGUID guid, UINT data_size, const void *data) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); TRACE("iface %p, guid %s, data_size %u, data %p.\n", iface, debugstr_guid(guid), data_size, data); @@ -2550,19 +2553,19 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_SetPrivateData(ID3D12Device1 *ifac return vkd3d_set_private_data(&device->private_store, guid, data_size, data); } -static HRESULT STDMETHODCALLTYPE d3d12_device_SetPrivateDataInterface(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_SetPrivateDataInterface(ID3D12Device5 *iface, REFGUID guid, const IUnknown *data) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); TRACE("iface %p, guid %s, data %p.\n", iface, debugstr_guid(guid), data); return vkd3d_set_private_data_interface(&device->private_store, guid, data); } -static HRESULT STDMETHODCALLTYPE d3d12_device_SetName(ID3D12Device1 *iface, const WCHAR *name) +static HRESULT STDMETHODCALLTYPE d3d12_device_SetName(ID3D12Device5 *iface, const WCHAR *name) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); TRACE("iface %p, name %s.\n", iface, debugstr_w(name, device->wchar_size)); @@ -2570,17 +2573,17 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_SetName(ID3D12Device1 *iface, cons VK_DEBUG_REPORT_OBJECT_TYPE_DEVICE_EXT, name); } -static UINT STDMETHODCALLTYPE d3d12_device_GetNodeCount(ID3D12Device1 *iface) +static UINT STDMETHODCALLTYPE d3d12_device_GetNodeCount(ID3D12Device5 *iface) { TRACE("iface %p.\n", iface); return 1; } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandQueue(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandQueue(ID3D12Device5 *iface, const D3D12_COMMAND_QUEUE_DESC *desc, REFIID riid, void **command_queue) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_command_queue *object; HRESULT hr; @@ -2594,10 +2597,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandQueue(ID3D12Device1 * riid, command_queue); } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandAllocator(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandAllocator(ID3D12Device5 *iface, D3D12_COMMAND_LIST_TYPE type, REFIID riid, void **command_allocator) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_command_allocator *object; HRESULT hr; @@ -2611,10 +2614,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandAllocator(ID3D12Devic riid, command_allocator); } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateGraphicsPipelineState(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateGraphicsPipelineState(ID3D12Device5 *iface, const D3D12_GRAPHICS_PIPELINE_STATE_DESC *desc, REFIID riid, void **pipeline_state) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_pipeline_state *object; HRESULT hr; @@ -2628,10 +2631,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateGraphicsPipelineState(ID3D12 &IID_ID3D12PipelineState, riid, pipeline_state); } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateComputePipelineState(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateComputePipelineState(ID3D12Device5 *iface, const D3D12_COMPUTE_PIPELINE_STATE_DESC *desc, REFIID riid, void **pipeline_state) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_pipeline_state *object; HRESULT hr; @@ -2645,11 +2648,11 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateComputePipelineState(ID3D12D &IID_ID3D12PipelineState, riid, pipeline_state); } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandList(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandList(ID3D12Device5 *iface, UINT node_mask, D3D12_COMMAND_LIST_TYPE type, ID3D12CommandAllocator *command_allocator, ID3D12PipelineState *initial_pipeline_state, REFIID riid, void **command_list) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_command_list *object; HRESULT hr; @@ -2662,8 +2665,8 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandList(ID3D12Device1 *i initial_pipeline_state, &object))) return hr; - return return_interface(&object->ID3D12GraphicsCommandList3_iface, - &IID_ID3D12GraphicsCommandList3, riid, command_list); + return return_interface(&object->ID3D12GraphicsCommandList5_iface, + &IID_ID3D12GraphicsCommandList5, riid, command_list); } /* Direct3D feature levels restrict which formats can be optionally supported. */ @@ -2772,10 +2775,10 @@ bool d3d12_device_is_uma(struct d3d12_device *device, bool *coherent) return true; } -static HRESULT STDMETHODCALLTYPE d3d12_device_CheckFeatureSupport(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CheckFeatureSupport(ID3D12Device5 *iface, D3D12_FEATURE feature, void *feature_data, UINT feature_data_size) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); TRACE("iface %p, feature %#x, feature_data %p, feature_data_size %u.\n", iface, feature, feature_data, feature_data_size); @@ -2925,7 +2928,14 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CheckFeatureSupport(ID3D12Device1 if (image_features & VK_FORMAT_FEATURE_BLIT_SRC_BIT) data->Support1 |= D3D12_FORMAT_SUPPORT1_MULTISAMPLE_RESOLVE; if (image_features & VK_FORMAT_FEATURE_STORAGE_IMAGE_BIT) + { data->Support1 |= D3D12_FORMAT_SUPPORT1_TYPED_UNORDERED_ACCESS_VIEW; + if (device->vk_info.uav_read_without_format) + data->Support2 |= D3D12_FORMAT_SUPPORT2_UAV_TYPED_LOAD; + /* We effectively require shaderStorageImageWriteWithoutFormat, + * so we can just report UAV_TYPED_STORE unconditionally. */ + data->Support2 |= D3D12_FORMAT_SUPPORT2_UAV_TYPED_STORE; + } if (image_features & VK_FORMAT_FEATURE_STORAGE_IMAGE_ATOMIC_BIT) data->Support2 |= D3D12_FORMAT_SUPPORT2_UAV_ATOMIC_ADD @@ -3274,10 +3284,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CheckFeatureSupport(ID3D12Device1 } } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateDescriptorHeap(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateDescriptorHeap(ID3D12Device5 *iface, const D3D12_DESCRIPTOR_HEAP_DESC *desc, REFIID riid, void **descriptor_heap) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_descriptor_heap *object; HRESULT hr; @@ -3291,7 +3301,7 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateDescriptorHeap(ID3D12Device1 &IID_ID3D12DescriptorHeap, riid, descriptor_heap); } -static UINT STDMETHODCALLTYPE d3d12_device_GetDescriptorHandleIncrementSize(ID3D12Device1 *iface, +static UINT STDMETHODCALLTYPE d3d12_device_GetDescriptorHandleIncrementSize(ID3D12Device5 *iface, D3D12_DESCRIPTOR_HEAP_TYPE descriptor_heap_type) { TRACE("iface %p, descriptor_heap_type %#x.\n", iface, descriptor_heap_type); @@ -3314,11 +3324,11 @@ static UINT STDMETHODCALLTYPE d3d12_device_GetDescriptorHandleIncrementSize(ID3D } } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateRootSignature(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateRootSignature(ID3D12Device5 *iface, UINT node_mask, const void *bytecode, SIZE_T bytecode_length, REFIID riid, void **root_signature) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_root_signature *object; HRESULT hr; @@ -3334,10 +3344,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateRootSignature(ID3D12Device1 &IID_ID3D12RootSignature, riid, root_signature); } -static void STDMETHODCALLTYPE d3d12_device_CreateConstantBufferView(ID3D12Device1 *iface, +static void STDMETHODCALLTYPE d3d12_device_CreateConstantBufferView(ID3D12Device5 *iface, const D3D12_CONSTANT_BUFFER_VIEW_DESC *desc, D3D12_CPU_DESCRIPTOR_HANDLE descriptor) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_desc tmp = {0}; TRACE("iface %p, desc %p, descriptor %#lx.\n", iface, desc, descriptor.ptr); @@ -3346,11 +3356,11 @@ static void STDMETHODCALLTYPE d3d12_device_CreateConstantBufferView(ID3D12Device d3d12_desc_write_atomic(d3d12_desc_from_cpu_handle(descriptor), &tmp, device); } -static void STDMETHODCALLTYPE d3d12_device_CreateShaderResourceView(ID3D12Device1 *iface, +static void STDMETHODCALLTYPE d3d12_device_CreateShaderResourceView(ID3D12Device5 *iface, ID3D12Resource *resource, const D3D12_SHADER_RESOURCE_VIEW_DESC *desc, D3D12_CPU_DESCRIPTOR_HANDLE descriptor) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_desc tmp = {0}; TRACE("iface %p, resource %p, desc %p, descriptor %#lx.\n", @@ -3360,11 +3370,11 @@ static void STDMETHODCALLTYPE d3d12_device_CreateShaderResourceView(ID3D12Device d3d12_desc_write_atomic(d3d12_desc_from_cpu_handle(descriptor), &tmp, device); } -static void STDMETHODCALLTYPE d3d12_device_CreateUnorderedAccessView(ID3D12Device1 *iface, +static void STDMETHODCALLTYPE d3d12_device_CreateUnorderedAccessView(ID3D12Device5 *iface, ID3D12Resource *resource, ID3D12Resource *counter_resource, const D3D12_UNORDERED_ACCESS_VIEW_DESC *desc, D3D12_CPU_DESCRIPTOR_HANDLE descriptor) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_desc tmp = {0}; TRACE("iface %p, resource %p, counter_resource %p, desc %p, descriptor %#lx.\n", @@ -3375,7 +3385,7 @@ static void STDMETHODCALLTYPE d3d12_device_CreateUnorderedAccessView(ID3D12Devic d3d12_desc_write_atomic(d3d12_desc_from_cpu_handle(descriptor), &tmp, device); } -static void STDMETHODCALLTYPE d3d12_device_CreateRenderTargetView(ID3D12Device1 *iface, +static void STDMETHODCALLTYPE d3d12_device_CreateRenderTargetView(ID3D12Device5 *iface, ID3D12Resource *resource, const D3D12_RENDER_TARGET_VIEW_DESC *desc, D3D12_CPU_DESCRIPTOR_HANDLE descriptor) { @@ -3383,10 +3393,10 @@ static void STDMETHODCALLTYPE d3d12_device_CreateRenderTargetView(ID3D12Device1 iface, resource, desc, descriptor.ptr); d3d12_rtv_desc_create_rtv(d3d12_rtv_desc_from_cpu_handle(descriptor), - impl_from_ID3D12Device1(iface), unsafe_impl_from_ID3D12Resource(resource), desc); + impl_from_ID3D12Device5(iface), unsafe_impl_from_ID3D12Resource(resource), desc); } -static void STDMETHODCALLTYPE d3d12_device_CreateDepthStencilView(ID3D12Device1 *iface, +static void STDMETHODCALLTYPE d3d12_device_CreateDepthStencilView(ID3D12Device5 *iface, ID3D12Resource *resource, const D3D12_DEPTH_STENCIL_VIEW_DESC *desc, D3D12_CPU_DESCRIPTOR_HANDLE descriptor) { @@ -3394,13 +3404,13 @@ static void STDMETHODCALLTYPE d3d12_device_CreateDepthStencilView(ID3D12Device1 iface, resource, desc, descriptor.ptr); d3d12_dsv_desc_create_dsv(d3d12_dsv_desc_from_cpu_handle(descriptor), - impl_from_ID3D12Device1(iface), unsafe_impl_from_ID3D12Resource(resource), desc); + impl_from_ID3D12Device5(iface), unsafe_impl_from_ID3D12Resource(resource), desc); } -static void STDMETHODCALLTYPE d3d12_device_CreateSampler(ID3D12Device1 *iface, +static void STDMETHODCALLTYPE d3d12_device_CreateSampler(ID3D12Device5 *iface, const D3D12_SAMPLER_DESC *desc, D3D12_CPU_DESCRIPTOR_HANDLE descriptor) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_desc tmp = {0}; TRACE("iface %p, desc %p, descriptor %#lx.\n", iface, desc, descriptor.ptr); @@ -3409,14 +3419,14 @@ static void STDMETHODCALLTYPE d3d12_device_CreateSampler(ID3D12Device1 *iface, d3d12_desc_write_atomic(d3d12_desc_from_cpu_handle(descriptor), &tmp, device); } -static void STDMETHODCALLTYPE d3d12_device_CopyDescriptors(ID3D12Device1 *iface, +static void STDMETHODCALLTYPE d3d12_device_CopyDescriptors(ID3D12Device5 *iface, UINT dst_descriptor_range_count, const D3D12_CPU_DESCRIPTOR_HANDLE *dst_descriptor_range_offsets, const UINT *dst_descriptor_range_sizes, UINT src_descriptor_range_count, const D3D12_CPU_DESCRIPTOR_HANDLE *src_descriptor_range_offsets, const UINT *src_descriptor_range_sizes, D3D12_DESCRIPTOR_HEAP_TYPE descriptor_heap_type) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); unsigned int dst_range_idx, dst_idx, src_range_idx, src_idx; unsigned int dst_range_size, src_range_size; struct d3d12_descriptor_heap *dst_heap; @@ -3472,7 +3482,7 @@ static void STDMETHODCALLTYPE d3d12_device_CopyDescriptors(ID3D12Device1 *iface, } } -static void STDMETHODCALLTYPE d3d12_device_CopyDescriptorsSimple(ID3D12Device1 *iface, +static void STDMETHODCALLTYPE d3d12_device_CopyDescriptorsSimple(ID3D12Device5 *iface, UINT descriptor_count, const D3D12_CPU_DESCRIPTOR_HANDLE dst_descriptor_range_offset, const D3D12_CPU_DESCRIPTOR_HANDLE src_descriptor_range_offset, D3D12_DESCRIPTOR_HEAP_TYPE descriptor_heap_type) @@ -3487,10 +3497,10 @@ static void STDMETHODCALLTYPE d3d12_device_CopyDescriptorsSimple(ID3D12Device1 * } static D3D12_RESOURCE_ALLOCATION_INFO * STDMETHODCALLTYPE d3d12_device_GetResourceAllocationInfo( - ID3D12Device1 *iface, D3D12_RESOURCE_ALLOCATION_INFO *info, UINT visible_mask, + ID3D12Device5 *iface, D3D12_RESOURCE_ALLOCATION_INFO *info, UINT visible_mask, UINT count, const D3D12_RESOURCE_DESC *resource_descs) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); const D3D12_RESOURCE_DESC *desc; uint64_t requested_alignment; @@ -3563,10 +3573,10 @@ invalid: return info; } -static D3D12_HEAP_PROPERTIES * STDMETHODCALLTYPE d3d12_device_GetCustomHeapProperties(ID3D12Device1 *iface, +static D3D12_HEAP_PROPERTIES * STDMETHODCALLTYPE d3d12_device_GetCustomHeapProperties(ID3D12Device5 *iface, D3D12_HEAP_PROPERTIES *heap_properties, UINT node_mask, D3D12_HEAP_TYPE heap_type) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); bool coherent; TRACE("iface %p, heap_properties %p, node_mask 0x%08x, heap_type %#x.\n", @@ -3606,12 +3616,12 @@ static D3D12_HEAP_PROPERTIES * STDMETHODCALLTYPE d3d12_device_GetCustomHeapPrope return heap_properties; } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommittedResource(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommittedResource(ID3D12Device5 *iface, const D3D12_HEAP_PROPERTIES *heap_properties, D3D12_HEAP_FLAGS heap_flags, const D3D12_RESOURCE_DESC *desc, D3D12_RESOURCE_STATES initial_state, const D3D12_CLEAR_VALUE *optimized_clear_value, REFIID iid, void **resource) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_resource *object; HRESULT hr; @@ -3621,26 +3631,26 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommittedResource(ID3D12Devi optimized_clear_value, debugstr_guid(iid), resource); if (FAILED(hr = d3d12_committed_resource_create(device, heap_properties, heap_flags, - desc, initial_state, optimized_clear_value, &object))) + desc, initial_state, optimized_clear_value, NULL, &object))) { *resource = NULL; return hr; } - return return_interface(&object->ID3D12Resource_iface, &IID_ID3D12Resource, iid, resource); + return return_interface(&object->ID3D12Resource1_iface, &IID_ID3D12Resource1, iid, resource); } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateHeap(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateHeap(ID3D12Device5 *iface, const D3D12_HEAP_DESC *desc, REFIID iid, void **heap) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_heap *object; HRESULT hr; TRACE("iface %p, desc %p, iid %s, heap %p.\n", iface, desc, debugstr_guid(iid), heap); - if (FAILED(hr = d3d12_heap_create(device, desc, NULL, &object))) + if (FAILED(hr = d3d12_heap_create(device, desc, NULL, NULL, &object))) { *heap = NULL; return hr; @@ -3649,12 +3659,12 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateHeap(ID3D12Device1 *iface, return return_interface(&object->ID3D12Heap_iface, &IID_ID3D12Heap, iid, heap); } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreatePlacedResource(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreatePlacedResource(ID3D12Device5 *iface, ID3D12Heap *heap, UINT64 heap_offset, const D3D12_RESOURCE_DESC *desc, D3D12_RESOURCE_STATES initial_state, const D3D12_CLEAR_VALUE *optimized_clear_value, REFIID iid, void **resource) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_heap *heap_object; struct d3d12_resource *object; HRESULT hr; @@ -3670,14 +3680,14 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreatePlacedResource(ID3D12Device1 desc, initial_state, optimized_clear_value, &object))) return hr; - return return_interface(&object->ID3D12Resource_iface, &IID_ID3D12Resource, iid, resource); + return return_interface(&object->ID3D12Resource1_iface, &IID_ID3D12Resource1, iid, resource); } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateReservedResource(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateReservedResource(ID3D12Device5 *iface, const D3D12_RESOURCE_DESC *desc, D3D12_RESOURCE_STATES initial_state, const D3D12_CLEAR_VALUE *optimized_clear_value, REFIID iid, void **resource) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_resource *object; HRESULT hr; @@ -3688,14 +3698,14 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateReservedResource(ID3D12Devic desc, initial_state, optimized_clear_value, &object))) return hr; - return return_interface(&object->ID3D12Resource_iface, &IID_ID3D12Resource, iid, resource); + return return_interface(&object->ID3D12Resource1_iface, &IID_ID3D12Resource1, iid, resource); } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateSharedHandle(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateSharedHandle(ID3D12Device5 *iface, ID3D12DeviceChild *object, const SECURITY_ATTRIBUTES *attributes, DWORD access, const WCHAR *name, HANDLE *handle) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); FIXME("iface %p, object %p, attributes %p, access %#x, name %s, handle %p stub!\n", iface, object, attributes, access, debugstr_w(name, device->wchar_size), handle); @@ -3703,7 +3713,7 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateSharedHandle(ID3D12Device1 * return E_NOTIMPL; } -static HRESULT STDMETHODCALLTYPE d3d12_device_OpenSharedHandle(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_OpenSharedHandle(ID3D12Device5 *iface, HANDLE handle, REFIID riid, void **object) { FIXME("iface %p, handle %p, riid %s, object %p stub!\n", @@ -3712,10 +3722,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_OpenSharedHandle(ID3D12Device1 *if return E_NOTIMPL; } -static HRESULT STDMETHODCALLTYPE d3d12_device_OpenSharedHandleByName(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_OpenSharedHandleByName(ID3D12Device5 *iface, const WCHAR *name, DWORD access, HANDLE *handle) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); FIXME("iface %p, name %s, access %#x, handle %p stub!\n", iface, debugstr_w(name, device->wchar_size), access, handle); @@ -3723,7 +3733,7 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_OpenSharedHandleByName(ID3D12Devic return E_NOTIMPL; } -static HRESULT STDMETHODCALLTYPE d3d12_device_MakeResident(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_MakeResident(ID3D12Device5 *iface, UINT object_count, ID3D12Pageable * const *objects) { FIXME_ONCE("iface %p, object_count %u, objects %p stub!\n", @@ -3732,7 +3742,7 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_MakeResident(ID3D12Device1 *iface, return S_OK; } -static HRESULT STDMETHODCALLTYPE d3d12_device_Evict(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_Evict(ID3D12Device5 *iface, UINT object_count, ID3D12Pageable * const *objects) { FIXME_ONCE("iface %p, object_count %u, objects %p stub!\n", @@ -3741,10 +3751,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_Evict(ID3D12Device1 *iface, return S_OK; } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateFence(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateFence(ID3D12Device5 *iface, UINT64 initial_value, D3D12_FENCE_FLAGS flags, REFIID riid, void **fence) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_fence *object; HRESULT hr; @@ -3757,21 +3767,21 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateFence(ID3D12Device1 *iface, return return_interface(&object->ID3D12Fence1_iface, &IID_ID3D12Fence1, riid, fence); } -static HRESULT STDMETHODCALLTYPE d3d12_device_GetDeviceRemovedReason(ID3D12Device1 *iface) +static HRESULT STDMETHODCALLTYPE d3d12_device_GetDeviceRemovedReason(ID3D12Device5 *iface) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); TRACE("iface %p.\n", iface); return device->removed_reason; } -static void STDMETHODCALLTYPE d3d12_device_GetCopyableFootprints(ID3D12Device1 *iface, +static void STDMETHODCALLTYPE d3d12_device_GetCopyableFootprints(ID3D12Device5 *iface, const D3D12_RESOURCE_DESC *desc, UINT first_sub_resource, UINT sub_resource_count, UINT64 base_offset, D3D12_PLACED_SUBRESOURCE_FOOTPRINT *layouts, UINT *row_counts, UINT64 *row_sizes, UINT64 *total_bytes) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); unsigned int i, sub_resource_idx, miplevel_idx, row_count, row_size, row_pitch; unsigned int width, height, depth, plane_count, sub_resources_per_plane; @@ -3851,10 +3861,10 @@ static void STDMETHODCALLTYPE d3d12_device_GetCopyableFootprints(ID3D12Device1 * *total_bytes = total; } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateQueryHeap(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateQueryHeap(ID3D12Device5 *iface, const D3D12_QUERY_HEAP_DESC *desc, REFIID iid, void **heap) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_query_heap *object; HRESULT hr; @@ -3867,18 +3877,18 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateQueryHeap(ID3D12Device1 *ifa return return_interface(&object->ID3D12QueryHeap_iface, &IID_ID3D12QueryHeap, iid, heap); } -static HRESULT STDMETHODCALLTYPE d3d12_device_SetStablePowerState(ID3D12Device1 *iface, BOOL enable) +static HRESULT STDMETHODCALLTYPE d3d12_device_SetStablePowerState(ID3D12Device5 *iface, BOOL enable) { FIXME("iface %p, enable %#x stub!\n", iface, enable); return E_NOTIMPL; } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandSignature(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandSignature(ID3D12Device5 *iface, const D3D12_COMMAND_SIGNATURE_DESC *desc, ID3D12RootSignature *root_signature, REFIID iid, void **command_signature) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); struct d3d12_command_signature *object; HRESULT hr; @@ -3892,14 +3902,14 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandSignature(ID3D12Devic &IID_ID3D12CommandSignature, iid, command_signature); } -static void STDMETHODCALLTYPE d3d12_device_GetResourceTiling(ID3D12Device1 *iface, +static void STDMETHODCALLTYPE d3d12_device_GetResourceTiling(ID3D12Device5 *iface, ID3D12Resource *resource, UINT *total_tile_count, D3D12_PACKED_MIP_INFO *packed_mip_info, D3D12_TILE_SHAPE *standard_tile_shape, UINT *sub_resource_tiling_count, UINT first_sub_resource_tiling, D3D12_SUBRESOURCE_TILING *sub_resource_tilings) { const struct d3d12_resource *resource_impl = impl_from_ID3D12Resource(resource); - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); TRACE("iface %p, resource %p, total_tile_count %p, packed_mip_info %p, " "standard_title_shape %p, sub_resource_tiling_count %p, " @@ -3912,9 +3922,9 @@ static void STDMETHODCALLTYPE d3d12_device_GetResourceTiling(ID3D12Device1 *ifac sub_resource_tiling_count, first_sub_resource_tiling, sub_resource_tilings); } -static LUID * STDMETHODCALLTYPE d3d12_device_GetAdapterLuid(ID3D12Device1 *iface, LUID *luid) +static LUID * STDMETHODCALLTYPE d3d12_device_GetAdapterLuid(ID3D12Device5 *iface, LUID *luid) { - struct d3d12_device *device = impl_from_ID3D12Device1(iface); + struct d3d12_device *device = impl_from_ID3D12Device5(iface); TRACE("iface %p, luid %p.\n", iface, luid); @@ -3923,7 +3933,7 @@ static LUID * STDMETHODCALLTYPE d3d12_device_GetAdapterLuid(ID3D12Device1 *iface return luid; } -static HRESULT STDMETHODCALLTYPE d3d12_device_CreatePipelineLibrary(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_CreatePipelineLibrary(ID3D12Device5 *iface, const void *blob, SIZE_T blob_size, REFIID iid, void **lib) { FIXME("iface %p, blob %p, blob_size %lu, iid %s, lib %p stub!\n", iface, blob, blob_size, debugstr_guid(iid), lib); @@ -3931,7 +3941,7 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreatePipelineLibrary(ID3D12Device return DXGI_ERROR_UNSUPPORTED; } -static HRESULT STDMETHODCALLTYPE d3d12_device_SetEventOnMultipleFenceCompletion(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_SetEventOnMultipleFenceCompletion(ID3D12Device5 *iface, ID3D12Fence *const *fences, const UINT64 *values, UINT fence_count, D3D12_MULTIPLE_FENCE_WAIT_FLAGS flags, HANDLE event) { @@ -3941,7 +3951,7 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_SetEventOnMultipleFenceCompletion( return E_NOTIMPL; } -static HRESULT STDMETHODCALLTYPE d3d12_device_SetResidencyPriority(ID3D12Device1 *iface, +static HRESULT STDMETHODCALLTYPE d3d12_device_SetResidencyPriority(ID3D12Device5 *iface, UINT object_count, ID3D12Pageable *const *objects, const D3D12_RESIDENCY_PRIORITY *priorities) { FIXME_ONCE("iface %p, object_count %u, objects %p, priorities %p stub!\n", iface, object_count, objects, priorities); @@ -3949,7 +3959,204 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_SetResidencyPriority(ID3D12Device1 return S_OK; } -static const struct ID3D12Device1Vtbl d3d12_device_vtbl = +static HRESULT STDMETHODCALLTYPE d3d12_device_CreatePipelineState(ID3D12Device5 *iface, + const D3D12_PIPELINE_STATE_STREAM_DESC *desc, REFIID iid, void **pipeline_state) +{ + struct d3d12_device *device = impl_from_ID3D12Device5(iface); + struct d3d12_pipeline_state *object; + HRESULT hr; + + TRACE("iface %p, desc %p, iid %s, pipeline_state %p.\n", iface, desc, debugstr_guid(iid), pipeline_state); + + if (FAILED(hr = d3d12_pipeline_state_create(device, desc, &object))) + return hr; + + return return_interface(&object->ID3D12PipelineState_iface, &IID_ID3D12PipelineState, iid, pipeline_state); +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_OpenExistingHeapFromAddress(ID3D12Device5 *iface, + const void *address, REFIID iid, void **heap) +{ + FIXME("iface %p, address %p, iid %s, heap %p stub!\n", iface, address, debugstr_guid(iid), heap); + + return E_NOTIMPL; +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_OpenExistingHeapFromFileMapping(ID3D12Device5 *iface, + HANDLE file_mapping, REFIID iid, void **heap) +{ + FIXME("iface %p, file_mapping %p, iid %s, heap %p stub!\n", iface, file_mapping, debugstr_guid(iid), heap); + + return E_NOTIMPL; +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_EnqueueMakeResident(ID3D12Device5 *iface, + D3D12_RESIDENCY_FLAGS flags, UINT num_objects, ID3D12Pageable *const *objects, + ID3D12Fence *fence, UINT64 fence_value) +{ + FIXME("iface %p, flags %#x, num_objects %u, objects %p, fence %p, fence_value %#"PRIx64" stub!\n", + iface, flags, num_objects, objects, fence, fence_value); + + return E_NOTIMPL; +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandList1(ID3D12Device5 *iface, + UINT node_mask, D3D12_COMMAND_LIST_TYPE type, D3D12_COMMAND_LIST_FLAGS flags, + REFIID iid, void **command_list) +{ + FIXME("iface %p, node_mask 0x%08x, type %#x, flags %#x, iid %s, command_list %p stub!\n", + iface, node_mask, type, flags, debugstr_guid(iid), command_list); + + return E_NOTIMPL; +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateProtectedResourceSession(ID3D12Device5 *iface, + const D3D12_PROTECTED_RESOURCE_SESSION_DESC *desc, REFIID iid, void **session) +{ + FIXME("iface %p, desc %p, iid %s, session %p stub!\n", iface, desc, debugstr_guid(iid), session); + + return E_NOTIMPL; +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommittedResource1(ID3D12Device5 *iface, + const D3D12_HEAP_PROPERTIES *heap_properties, D3D12_HEAP_FLAGS heap_flags, + const D3D12_RESOURCE_DESC *desc, D3D12_RESOURCE_STATES initial_state, + const D3D12_CLEAR_VALUE *optimized_clear_value, + ID3D12ProtectedResourceSession *protected_session, REFIID iid, void **resource) +{ + struct d3d12_device *device = impl_from_ID3D12Device5(iface); + struct d3d12_resource *object; + HRESULT hr; + + TRACE("iface %p, heap_properties %p, heap_flags %#x, desc %p, initial_state %#x, " + "optimized_clear_value %p, protected_session %p, iid %s, resource %p.\n", + iface, heap_properties, heap_flags, desc, initial_state, + optimized_clear_value, protected_session, debugstr_guid(iid), resource); + + if (FAILED(hr = d3d12_committed_resource_create(device, heap_properties, heap_flags, + desc, initial_state, optimized_clear_value, protected_session, &object))) + { + *resource = NULL; + return hr; + } + + return return_interface(&object->ID3D12Resource1_iface, &IID_ID3D12Resource1, iid, resource); +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateHeap1(ID3D12Device5 *iface, + const D3D12_HEAP_DESC *desc, ID3D12ProtectedResourceSession *protected_session, + REFIID iid, void **heap) +{ + struct d3d12_device *device = impl_from_ID3D12Device5(iface); + struct d3d12_heap *object; + HRESULT hr; + + TRACE("iface %p, desc %p, protected_session %p, iid %s, heap %p.\n", + iface, desc, protected_session, debugstr_guid(iid), heap); + + if (FAILED(hr = d3d12_heap_create(device, desc, NULL, protected_session, &object))) + { + *heap = NULL; + return hr; + } + + return return_interface(&object->ID3D12Heap_iface, &IID_ID3D12Heap, iid, heap); +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateReservedResource1(ID3D12Device5 *iface, + const D3D12_RESOURCE_DESC *desc, D3D12_RESOURCE_STATES initial_state, + const D3D12_CLEAR_VALUE *optimized_clear_value, + ID3D12ProtectedResourceSession *protected_session, REFIID iid, void **resource) +{ + FIXME("iface %p, desc %p, initial_state %#x, optimized_clear_value %p, " + "protected_session %p, iid %s, resource %p stub!\n", + iface, desc, initial_state, optimized_clear_value, protected_session, + debugstr_guid(iid), resource); + + return E_NOTIMPL; +} + +static D3D12_RESOURCE_ALLOCATION_INFO * STDMETHODCALLTYPE d3d12_device_GetResourceAllocationInfo1( + ID3D12Device5 *iface, D3D12_RESOURCE_ALLOCATION_INFO *info, UINT visible_mask, + UINT count, const D3D12_RESOURCE_DESC *resource_descs, + D3D12_RESOURCE_ALLOCATION_INFO1 *info1) +{ + FIXME("iface %p, info %p, visible_mask 0x%08x, count %u, resource_descs %p, info1 %p stub!\n", + iface, info, visible_mask, count, resource_descs, info1); + + return info; +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateLifetimeTracker(ID3D12Device5 *iface, + ID3D12LifetimeOwner *owner, REFIID iid, void **tracker) +{ + FIXME("iface %p, owner %p, iid %s, tracker %p stub!\n", iface, owner, debugstr_guid(iid), tracker); + + return E_NOTIMPL; +} + +static void STDMETHODCALLTYPE d3d12_device_RemoveDevice(ID3D12Device5 *iface) +{ + FIXME("iface %p stub!\n", iface); +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_EnumerateMetaCommands(ID3D12Device5 *iface, + UINT *num_meta_commands, D3D12_META_COMMAND_DESC *command_desc) +{ + FIXME("iface %p, num_meta_commands %p, command_desc %p stub!\n", iface, + num_meta_commands, command_desc); + + return E_NOTIMPL; +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_EnumerateMetaCommandParameters(ID3D12Device5 *iface, + REFGUID command_id, D3D12_META_COMMAND_PARAMETER_STAGE stage, + UINT *size_in_bytes, UINT *parameter_count, + D3D12_META_COMMAND_PARAMETER_DESC *parameter_desc) +{ + FIXME("iface %p, command_id %s, stage %u, size_in_bytes %p, " + "parameter_count %p, parameter_desc %p stub!\n", iface, + debugstr_guid(command_id), stage, size_in_bytes, parameter_count, parameter_desc); + + return E_NOTIMPL; +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateMetaCommand(ID3D12Device5 *iface, + REFGUID command_id, UINT node_mask, const void *parameters_data, + SIZE_T data_size_in_bytes, REFIID iid, void **meta_command) +{ + FIXME("iface %p, command_id %s, node_mask %#x, parameters_data %p, " + "data_size_in_bytes %lu, iid %s, meta_command %p stub!\n", iface, + debugstr_guid(command_id), node_mask, parameters_data, + data_size_in_bytes, debugstr_guid(iid), meta_command); + + return E_NOTIMPL; +} + +static HRESULT STDMETHODCALLTYPE d3d12_device_CreateStateObject(ID3D12Device5 *iface, + const D3D12_STATE_OBJECT_DESC *desc, REFIID iid, void **state_object) +{ + FIXME("iface %p, desc %p, iid %s, state_object %p stub!\n", iface, desc, debugstr_guid(iid), state_object); + + return E_NOTIMPL; +} + +static void STDMETHODCALLTYPE d3d12_device_GetRaytracingAccelerationStructurePrebuildInfo(ID3D12Device5 *iface, + const D3D12_BUILD_RAYTRACING_ACCELERATION_STRUCTURE_INPUTS *desc, + D3D12_RAYTRACING_ACCELERATION_STRUCTURE_PREBUILD_INFO *info) +{ + FIXME("iface %p, desc %p, info %p stub!\n", iface, desc, info); +} + +static D3D12_DRIVER_MATCHING_IDENTIFIER_STATUS STDMETHODCALLTYPE d3d12_device_CheckDriverMatchingIdentifier(ID3D12Device5 *iface, + D3D12_SERIALIZED_DATA_TYPE data_type, const D3D12_SERIALIZED_DATA_DRIVER_MATCHING_IDENTIFIER *identifier) +{ + FIXME("iface %p, data_type %u, identifier %p stub!\n", iface, data_type, identifier); + + return D3D12_DRIVER_MATCHING_IDENTIFIER_UNRECOGNIZED; +} + +static const struct ID3D12Device5Vtbl d3d12_device_vtbl = { /* IUnknown methods */ d3d12_device_QueryInterface, @@ -4002,14 +4209,36 @@ static const struct ID3D12Device1Vtbl d3d12_device_vtbl = d3d12_device_CreatePipelineLibrary, d3d12_device_SetEventOnMultipleFenceCompletion, d3d12_device_SetResidencyPriority, + /* ID3D12Device2 methods */ + d3d12_device_CreatePipelineState, + /* ID3D12Device3 methods */ + d3d12_device_OpenExistingHeapFromAddress, + d3d12_device_OpenExistingHeapFromFileMapping, + d3d12_device_EnqueueMakeResident, + /* ID3D12Device4 methods */ + d3d12_device_CreateCommandList1, + d3d12_device_CreateProtectedResourceSession, + d3d12_device_CreateCommittedResource1, + d3d12_device_CreateHeap1, + d3d12_device_CreateReservedResource1, + d3d12_device_GetResourceAllocationInfo1, + /* ID3D12Device5 methods */ + d3d12_device_CreateLifetimeTracker, + d3d12_device_RemoveDevice, + d3d12_device_EnumerateMetaCommands, + d3d12_device_EnumerateMetaCommandParameters, + d3d12_device_CreateMetaCommand, + d3d12_device_CreateStateObject, + d3d12_device_GetRaytracingAccelerationStructurePrebuildInfo, + d3d12_device_CheckDriverMatchingIdentifier, }; -struct d3d12_device *unsafe_impl_from_ID3D12Device1(ID3D12Device1 *iface) +struct d3d12_device *unsafe_impl_from_ID3D12Device5(ID3D12Device5 *iface) { if (!iface) return NULL; assert(iface->lpVtbl == &d3d12_device_vtbl); - return impl_from_ID3D12Device1(iface); + return impl_from_ID3D12Device5(iface); } static HRESULT d3d12_device_init(struct d3d12_device *device, @@ -4018,7 +4247,7 @@ static HRESULT d3d12_device_init(struct d3d12_device *device, const struct vkd3d_vk_device_procs *vk_procs; HRESULT hr; - device->ID3D12Device1_iface.lpVtbl = &d3d12_device_vtbl; + device->ID3D12Device5_iface.lpVtbl = &d3d12_device_vtbl; device->refcount = 1; vkd3d_instance_incref(device->vkd3d_instance = instance); @@ -4215,28 +4444,28 @@ HRESULT vkd3d_join_thread(struct vkd3d_instance *instance, union vkd3d_thread_ha IUnknown *vkd3d_get_device_parent(ID3D12Device *device) { - struct d3d12_device *d3d12_device = impl_from_ID3D12Device1((ID3D12Device1 *)device); + struct d3d12_device *d3d12_device = impl_from_ID3D12Device5((ID3D12Device5 *)device); return d3d12_device->parent; } VkDevice vkd3d_get_vk_device(ID3D12Device *device) { - struct d3d12_device *d3d12_device = impl_from_ID3D12Device1((ID3D12Device1 *)device); + struct d3d12_device *d3d12_device = impl_from_ID3D12Device5((ID3D12Device5 *)device); return d3d12_device->vk_device; } VkPhysicalDevice vkd3d_get_vk_physical_device(ID3D12Device *device) { - struct d3d12_device *d3d12_device = impl_from_ID3D12Device1((ID3D12Device1 *)device); + struct d3d12_device *d3d12_device = impl_from_ID3D12Device5((ID3D12Device5 *)device); return d3d12_device->vk_physical_device; } struct vkd3d_instance *vkd3d_instance_from_device(ID3D12Device *device) { - struct d3d12_device *d3d12_device = impl_from_ID3D12Device1((ID3D12Device1 *)device); + struct d3d12_device *d3d12_device = impl_from_ID3D12Device5((ID3D12Device5 *)device); return d3d12_device->vkd3d_instance; } diff --git a/libs/vkd3d/libs/vkd3d/resource.c b/libs/vkd3d/libs/vkd3d/resource.c index f3842958d96..abbdfbe2015 100644 --- a/libs/vkd3d/libs/vkd3d/resource.c +++ b/libs/vkd3d/libs/vkd3d/resource.c @@ -574,11 +574,15 @@ static HRESULT d3d12_heap_init(struct d3d12_heap *heap, } HRESULT d3d12_heap_create(struct d3d12_device *device, const D3D12_HEAP_DESC *desc, - const struct d3d12_resource *resource, struct d3d12_heap **heap) + const struct d3d12_resource *resource, ID3D12ProtectedResourceSession *protected_session, + struct d3d12_heap **heap) { struct d3d12_heap *object; HRESULT hr; + if (protected_session) + FIXME("Protected session is not supported.\n"); + if (!(object = vkd3d_malloc(sizeof(*object)))) return E_OUTOFMEMORY; @@ -1254,12 +1258,13 @@ static bool d3d12_resource_init_tiles(struct d3d12_resource *resource, struct d3 } /* ID3D12Resource */ -static HRESULT STDMETHODCALLTYPE d3d12_resource_QueryInterface(ID3D12Resource *iface, +static HRESULT STDMETHODCALLTYPE d3d12_resource_QueryInterface(ID3D12Resource1 *iface, REFIID riid, void **object) { TRACE("iface %p, riid %s, object %p.\n", iface, debugstr_guid(riid), object); - if (IsEqualGUID(riid, &IID_ID3D12Resource) + if (IsEqualGUID(riid, &IID_ID3D12Resource1) + || IsEqualGUID(riid, &IID_ID3D12Resource) || IsEqualGUID(riid, &IID_ID3D12Pageable) || IsEqualGUID(riid, &IID_ID3D12DeviceChild) || IsEqualGUID(riid, &IID_ID3D12Object) @@ -1276,9 +1281,9 @@ static HRESULT STDMETHODCALLTYPE d3d12_resource_QueryInterface(ID3D12Resource *i return E_NOINTERFACE; } -static ULONG STDMETHODCALLTYPE d3d12_resource_AddRef(ID3D12Resource *iface) +static ULONG STDMETHODCALLTYPE d3d12_resource_AddRef(ID3D12Resource1 *iface) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); ULONG refcount = InterlockedIncrement(&resource->refcount); TRACE("%p increasing refcount to %u.\n", resource, refcount); @@ -1294,9 +1299,9 @@ static ULONG STDMETHODCALLTYPE d3d12_resource_AddRef(ID3D12Resource *iface) return refcount; } -static ULONG STDMETHODCALLTYPE d3d12_resource_Release(ID3D12Resource *iface) +static ULONG STDMETHODCALLTYPE d3d12_resource_Release(ID3D12Resource1 *iface) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); ULONG refcount = InterlockedDecrement(&resource->refcount); TRACE("%p decreasing refcount to %u.\n", resource, refcount); @@ -1313,39 +1318,39 @@ static ULONG STDMETHODCALLTYPE d3d12_resource_Release(ID3D12Resource *iface) return refcount; } -static HRESULT STDMETHODCALLTYPE d3d12_resource_GetPrivateData(ID3D12Resource *iface, +static HRESULT STDMETHODCALLTYPE d3d12_resource_GetPrivateData(ID3D12Resource1 *iface, REFGUID guid, UINT *data_size, void *data) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); TRACE("iface %p, guid %s, data_size %p, data %p.\n", iface, debugstr_guid(guid), data_size, data); return vkd3d_get_private_data(&resource->private_store, guid, data_size, data); } -static HRESULT STDMETHODCALLTYPE d3d12_resource_SetPrivateData(ID3D12Resource *iface, +static HRESULT STDMETHODCALLTYPE d3d12_resource_SetPrivateData(ID3D12Resource1 *iface, REFGUID guid, UINT data_size, const void *data) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); TRACE("iface %p, guid %s, data_size %u, data %p.\n", iface, debugstr_guid(guid), data_size, data); return vkd3d_set_private_data(&resource->private_store, guid, data_size, data); } -static HRESULT STDMETHODCALLTYPE d3d12_resource_SetPrivateDataInterface(ID3D12Resource *iface, +static HRESULT STDMETHODCALLTYPE d3d12_resource_SetPrivateDataInterface(ID3D12Resource1 *iface, REFGUID guid, const IUnknown *data) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); TRACE("iface %p, guid %s, data %p.\n", iface, debugstr_guid(guid), data); return vkd3d_set_private_data_interface(&resource->private_store, guid, data); } -static HRESULT STDMETHODCALLTYPE d3d12_resource_SetName(ID3D12Resource *iface, const WCHAR *name) +static HRESULT STDMETHODCALLTYPE d3d12_resource_SetName(ID3D12Resource1 *iface, const WCHAR *name) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); HRESULT hr; TRACE("iface %p, name %s.\n", iface, debugstr_w(name, resource->device->wchar_size)); @@ -1364,9 +1369,9 @@ static HRESULT STDMETHODCALLTYPE d3d12_resource_SetName(ID3D12Resource *iface, c VK_DEBUG_REPORT_OBJECT_TYPE_IMAGE_EXT, name); } -static HRESULT STDMETHODCALLTYPE d3d12_resource_GetDevice(ID3D12Resource *iface, REFIID iid, void **device) +static HRESULT STDMETHODCALLTYPE d3d12_resource_GetDevice(ID3D12Resource1 *iface, REFIID iid, void **device) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); TRACE("iface %p, iid %s, device %p.\n", iface, debugstr_guid(iid), device); @@ -1417,10 +1422,10 @@ static void d3d12_resource_flush(struct d3d12_resource *resource, uint64_t offse ERR("Failed to flush memory, vr %d.\n", vr); } -static HRESULT STDMETHODCALLTYPE d3d12_resource_Map(ID3D12Resource *iface, UINT sub_resource, +static HRESULT STDMETHODCALLTYPE d3d12_resource_Map(ID3D12Resource1 *iface, UINT sub_resource, const D3D12_RANGE *read_range, void **data) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); unsigned int sub_resource_count; TRACE("iface %p, sub_resource %u, read_range %p, data %p.\n", @@ -1466,10 +1471,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_resource_Map(ID3D12Resource *iface, UINT return S_OK; } -static void STDMETHODCALLTYPE d3d12_resource_Unmap(ID3D12Resource *iface, UINT sub_resource, +static void STDMETHODCALLTYPE d3d12_resource_Unmap(ID3D12Resource1 *iface, UINT sub_resource, const D3D12_RANGE *written_range) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); unsigned int sub_resource_count; TRACE("iface %p, sub_resource %u, written_range %p.\n", @@ -1488,10 +1493,10 @@ static void STDMETHODCALLTYPE d3d12_resource_Unmap(ID3D12Resource *iface, UINT s d3d12_resource_flush(resource, written_range->Begin, written_range->End - written_range->Begin); } -static D3D12_RESOURCE_DESC * STDMETHODCALLTYPE d3d12_resource_GetDesc(ID3D12Resource *iface, +static D3D12_RESOURCE_DESC * STDMETHODCALLTYPE d3d12_resource_GetDesc(ID3D12Resource1 *iface, D3D12_RESOURCE_DESC *resource_desc) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); TRACE("iface %p, resource_desc %p.\n", iface, resource_desc); @@ -1499,20 +1504,20 @@ static D3D12_RESOURCE_DESC * STDMETHODCALLTYPE d3d12_resource_GetDesc(ID3D12Reso return resource_desc; } -static D3D12_GPU_VIRTUAL_ADDRESS STDMETHODCALLTYPE d3d12_resource_GetGPUVirtualAddress(ID3D12Resource *iface) +static D3D12_GPU_VIRTUAL_ADDRESS STDMETHODCALLTYPE d3d12_resource_GetGPUVirtualAddress(ID3D12Resource1 *iface) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); TRACE("iface %p.\n", iface); return resource->gpu_address; } -static HRESULT STDMETHODCALLTYPE d3d12_resource_WriteToSubresource(ID3D12Resource *iface, +static HRESULT STDMETHODCALLTYPE d3d12_resource_WriteToSubresource(ID3D12Resource1 *iface, UINT dst_sub_resource, const D3D12_BOX *dst_box, const void *src_data, UINT src_row_pitch, UINT src_slice_pitch) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); const struct vkd3d_vk_device_procs *vk_procs; VkImageSubresource vk_sub_resource; const struct vkd3d_format *format; @@ -1593,11 +1598,11 @@ static HRESULT STDMETHODCALLTYPE d3d12_resource_WriteToSubresource(ID3D12Resourc return S_OK; } -static HRESULT STDMETHODCALLTYPE d3d12_resource_ReadFromSubresource(ID3D12Resource *iface, +static HRESULT STDMETHODCALLTYPE d3d12_resource_ReadFromSubresource(ID3D12Resource1 *iface, void *dst_data, UINT dst_row_pitch, UINT dst_slice_pitch, UINT src_sub_resource, const D3D12_BOX *src_box) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); const struct vkd3d_vk_device_procs *vk_procs; VkImageSubresource vk_sub_resource; const struct vkd3d_format *format; @@ -1678,10 +1683,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_resource_ReadFromSubresource(ID3D12Resour return S_OK; } -static HRESULT STDMETHODCALLTYPE d3d12_resource_GetHeapProperties(ID3D12Resource *iface, +static HRESULT STDMETHODCALLTYPE d3d12_resource_GetHeapProperties(ID3D12Resource1 *iface, D3D12_HEAP_PROPERTIES *heap_properties, D3D12_HEAP_FLAGS *flags) { - struct d3d12_resource *resource = impl_from_ID3D12Resource(iface); + struct d3d12_resource *resource = impl_from_ID3D12Resource1(iface); struct d3d12_heap *heap; TRACE("iface %p, heap_properties %p, flags %p.\n", @@ -1715,7 +1720,15 @@ static HRESULT STDMETHODCALLTYPE d3d12_resource_GetHeapProperties(ID3D12Resource return S_OK; } -static const struct ID3D12ResourceVtbl d3d12_resource_vtbl = +static HRESULT STDMETHODCALLTYPE d3d12_resource_GetProtectedResourceSession(ID3D12Resource1 *iface, + REFIID iid, void **session) +{ + FIXME("iface %p, iid %s, session %p stub!\n", iface, debugstr_guid(iid), session); + + return E_NOTIMPL; +} + +static const struct ID3D12Resource1Vtbl d3d12_resource_vtbl = { /* IUnknown methods */ d3d12_resource_QueryInterface, @@ -1736,13 +1749,15 @@ static const struct ID3D12ResourceVtbl d3d12_resource_vtbl = d3d12_resource_WriteToSubresource, d3d12_resource_ReadFromSubresource, d3d12_resource_GetHeapProperties, + /* ID3D12Resource1 methods */ + d3d12_resource_GetProtectedResourceSession, }; struct d3d12_resource *unsafe_impl_from_ID3D12Resource(ID3D12Resource *iface) { if (!iface) return NULL; - assert(iface->lpVtbl == &d3d12_resource_vtbl); + assert(iface->lpVtbl == (ID3D12ResourceVtbl *)&d3d12_resource_vtbl); return impl_from_ID3D12Resource(iface); } @@ -1940,7 +1955,7 @@ static HRESULT d3d12_resource_init(struct d3d12_resource *resource, struct d3d12 { HRESULT hr; - resource->ID3D12Resource_iface.lpVtbl = &d3d12_resource_vtbl; + resource->ID3D12Resource1_iface.lpVtbl = &d3d12_resource_vtbl; resource->refcount = 1; resource->internal_refcount = 1; @@ -2064,7 +2079,7 @@ static HRESULT vkd3d_allocate_resource_memory( heap_desc.Properties = *heap_properties; heap_desc.Alignment = 0; heap_desc.Flags = heap_flags; - if (SUCCEEDED(hr = d3d12_heap_create(device, &heap_desc, resource, &resource->heap))) + if (SUCCEEDED(hr = d3d12_heap_create(device, &heap_desc, resource, NULL, &resource->heap))) resource->flags |= VKD3D_RESOURCE_DEDICATED_HEAP; return hr; } @@ -2072,7 +2087,8 @@ static HRESULT vkd3d_allocate_resource_memory( HRESULT d3d12_committed_resource_create(struct d3d12_device *device, const D3D12_HEAP_PROPERTIES *heap_properties, D3D12_HEAP_FLAGS heap_flags, const D3D12_RESOURCE_DESC *desc, D3D12_RESOURCE_STATES initial_state, - const D3D12_CLEAR_VALUE *optimized_clear_value, struct d3d12_resource **resource) + const D3D12_CLEAR_VALUE *optimized_clear_value, ID3D12ProtectedResourceSession *protected_session, + struct d3d12_resource **resource) { struct d3d12_resource *object; HRESULT hr; @@ -2083,13 +2099,16 @@ HRESULT d3d12_committed_resource_create(struct d3d12_device *device, return E_INVALIDARG; } + if (protected_session) + FIXME("Protected session is not supported.\n"); + if (FAILED(hr = d3d12_resource_create(device, heap_properties, heap_flags, desc, initial_state, optimized_clear_value, &object))) return hr; if (FAILED(hr = vkd3d_allocate_resource_memory(device, object, heap_properties, heap_flags))) { - d3d12_resource_Release(&object->ID3D12Resource_iface); + d3d12_resource_Release(&object->ID3D12Resource1_iface); return hr; } @@ -2182,7 +2201,7 @@ HRESULT d3d12_placed_resource_create(struct d3d12_device *device, struct d3d12_h if (FAILED(hr = vkd3d_bind_heap_memory(device, object, heap, heap_offset))) { - d3d12_resource_Release(&object->ID3D12Resource_iface); + d3d12_resource_Release(&object->ID3D12Resource1_iface); return hr; } @@ -2206,7 +2225,7 @@ HRESULT d3d12_reserved_resource_create(struct d3d12_device *device, if (!d3d12_resource_init_tiles(object, device)) { - d3d12_resource_Release(&object->ID3D12Resource_iface); + d3d12_resource_Release(&object->ID3D12Resource1_iface); return E_OUTOFMEMORY; } @@ -2220,7 +2239,7 @@ HRESULT d3d12_reserved_resource_create(struct d3d12_device *device, HRESULT vkd3d_create_image_resource(ID3D12Device *device, const struct vkd3d_image_resource_create_info *create_info, ID3D12Resource **resource) { - struct d3d12_device *d3d12_device = unsafe_impl_from_ID3D12Device1((ID3D12Device1 *)device); + struct d3d12_device *d3d12_device = unsafe_impl_from_ID3D12Device5((ID3D12Device5 *)device); struct d3d12_resource *object; HRESULT hr; @@ -2241,7 +2260,7 @@ HRESULT vkd3d_create_image_resource(ID3D12Device *device, memset(object, 0, sizeof(*object)); - object->ID3D12Resource_iface.lpVtbl = &d3d12_resource_vtbl; + object->ID3D12Resource1_iface.lpVtbl = &d3d12_resource_vtbl; object->refcount = 1; object->internal_refcount = 1; object->desc = create_info->desc; @@ -2265,7 +2284,7 @@ HRESULT vkd3d_create_image_resource(ID3D12Device *device, TRACE("Created resource %p.\n", object); - *resource = &object->ID3D12Resource_iface; + *resource = (ID3D12Resource *)&object->ID3D12Resource1_iface; return S_OK; } @@ -2998,6 +3017,7 @@ static bool init_default_texture_view_desc(struct vkd3d_texture_view_desc *desc, desc->components.b = VK_COMPONENT_SWIZZLE_IDENTITY; desc->components.a = VK_COMPONENT_SWIZZLE_IDENTITY; desc->allowed_swizzle = false; + desc->usage = 0; return true; } @@ -3039,6 +3059,7 @@ bool vkd3d_create_texture_view(struct d3d12_device *device, uint32_t magic, VkIm { const struct vkd3d_vk_device_procs *vk_procs = &device->vk_procs; const struct vkd3d_format *format = desc->format; + VkImageViewUsageCreateInfoKHR usage_desc; struct VkImageViewCreateInfo view_desc; VkImageView vk_view = VK_NULL_HANDLE; struct vkd3d_view *object; @@ -3060,6 +3081,13 @@ bool vkd3d_create_texture_view(struct d3d12_device *device, uint32_t magic, VkIm view_desc.subresourceRange.levelCount = desc->miplevel_count; view_desc.subresourceRange.baseArrayLayer = desc->layer_idx; view_desc.subresourceRange.layerCount = desc->layer_count; + if (device->vk_info.KHR_maintenance2) + { + usage_desc.sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_USAGE_CREATE_INFO; + usage_desc.pNext = NULL; + usage_desc.usage = desc->usage; + view_desc.pNext = &usage_desc; + } if ((vr = VK_CALL(vkCreateImageView(device->vk_device, &view_desc, NULL, &vk_view))) < 0) { WARN("Failed to create Vulkan image view, vr %d.\n", vr); @@ -3196,6 +3224,7 @@ static void vkd3d_create_null_srv(struct d3d12_desc *descriptor, vkd3d_desc.components.b = VK_COMPONENT_SWIZZLE_ZERO; vkd3d_desc.components.a = VK_COMPONENT_SWIZZLE_ZERO; vkd3d_desc.allowed_swizzle = true; + vkd3d_desc.usage = VK_IMAGE_USAGE_SAMPLED_BIT; vkd3d_create_texture_view(device, VKD3D_DESCRIPTOR_MAGIC_SRV, vk_image, &vkd3d_desc, &descriptor->s.u.view); } @@ -3268,6 +3297,7 @@ void d3d12_desc_create_srv(struct d3d12_desc *descriptor, vkd3d_desc.miplevel_count = VK_REMAINING_MIP_LEVELS; vkd3d_desc.allowed_swizzle = true; + vkd3d_desc.usage = VK_IMAGE_USAGE_SAMPLED_BIT; if (desc) { @@ -3421,6 +3451,7 @@ static void vkd3d_create_null_uav(struct d3d12_desc *descriptor, vkd3d_desc.components.b = VK_COMPONENT_SWIZZLE_B; vkd3d_desc.components.a = VK_COMPONENT_SWIZZLE_A; vkd3d_desc.allowed_swizzle = false; + vkd3d_desc.usage = VK_IMAGE_USAGE_STORAGE_BIT; vkd3d_create_texture_view(device, VKD3D_DESCRIPTOR_MAGIC_UAV, vk_image, &vkd3d_desc, &descriptor->s.u.view); } @@ -3480,6 +3511,8 @@ static void vkd3d_create_texture_uav(struct d3d12_desc *descriptor, if (!init_default_texture_view_desc(&vkd3d_desc, resource, desc ? desc->Format : 0)) return; + vkd3d_desc.usage = VK_IMAGE_USAGE_STORAGE_BIT; + if (vkd3d_format_is_compressed(vkd3d_desc.format)) { WARN("UAVs cannot be created for compressed formats.\n"); @@ -3747,6 +3780,8 @@ void d3d12_rtv_desc_create_rtv(struct d3d12_rtv_desc *rtv_desc, struct d3d12_dev if (!init_default_texture_view_desc(&vkd3d_desc, resource, desc ? desc->Format : 0)) return; + vkd3d_desc.usage = VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT; + if (vkd3d_desc.format->vk_aspect_mask != VK_IMAGE_ASPECT_COLOR_BIT) { WARN("Trying to create RTV for depth/stencil format %#x.\n", vkd3d_desc.format->dxgi_format); @@ -3847,6 +3882,8 @@ void d3d12_dsv_desc_create_dsv(struct d3d12_dsv_desc *dsv_desc, struct d3d12_dev if (!init_default_texture_view_desc(&vkd3d_desc, resource, desc ? desc->Format : 0)) return; + vkd3d_desc.usage = VK_IMAGE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT; + if (!(vkd3d_desc.format->vk_aspect_mask & (VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT))) { WARN("Trying to create DSV for format %#x.\n", vkd3d_desc.format->dxgi_format); diff --git a/libs/vkd3d/libs/vkd3d/state.c b/libs/vkd3d/libs/vkd3d/state.c index 0b92cffcde3..de0e04ea1e6 100644 --- a/libs/vkd3d/libs/vkd3d/state.c +++ b/libs/vkd3d/libs/vkd3d/state.c @@ -836,7 +836,7 @@ static unsigned int vk_heap_binding_count_from_descriptor_range(const struct d3d else { /* Prefer an unsupported binding count vs a zero count, because shader compilation will fail - * to match a declaration to a zero binding, resulting in failure of pipline state creation. */ + * to match a declaration to a zero binding, resulting in failure of pipeline state creation. */ return max_count + !max_count; } } @@ -1736,6 +1736,189 @@ void vkd3d_render_pass_cache_cleanup(struct vkd3d_render_pass_cache *cache, cache->render_passes = NULL; } +static void d3d12_init_pipeline_state_desc(struct d3d12_pipeline_state_desc *desc) +{ + D3D12_DEPTH_STENCIL_DESC1 *ds_state = &desc->depth_stencil_state; + D3D12_RASTERIZER_DESC *rs_state = &desc->rasterizer_state; + D3D12_BLEND_DESC *blend_state = &desc->blend_state; + DXGI_SAMPLE_DESC *sample_desc = &desc->sample_desc; + + memset(desc, 0, sizeof(*desc)); + ds_state->DepthEnable = TRUE; + ds_state->DepthWriteMask = D3D12_DEPTH_WRITE_MASK_ALL; + ds_state->DepthFunc = D3D12_COMPARISON_FUNC_LESS; + ds_state->StencilReadMask = D3D12_DEFAULT_STENCIL_READ_MASK; + ds_state->StencilWriteMask = D3D12_DEFAULT_STENCIL_WRITE_MASK; + ds_state->FrontFace.StencilFunc = ds_state->BackFace.StencilFunc = D3D12_COMPARISON_FUNC_ALWAYS; + ds_state->FrontFace.StencilDepthFailOp = ds_state->BackFace.StencilDepthFailOp = D3D12_STENCIL_OP_KEEP; + ds_state->FrontFace.StencilPassOp = ds_state->BackFace.StencilPassOp = D3D12_STENCIL_OP_KEEP; + ds_state->FrontFace.StencilFailOp = ds_state->BackFace.StencilFailOp = D3D12_STENCIL_OP_KEEP; + + rs_state->FillMode = D3D12_FILL_MODE_SOLID; + rs_state->CullMode = D3D12_CULL_MODE_BACK; + rs_state->DepthClipEnable = TRUE; + rs_state->ConservativeRaster = D3D12_CONSERVATIVE_RASTERIZATION_MODE_OFF; + + blend_state->RenderTarget[0].RenderTargetWriteMask = D3D12_COLOR_WRITE_ENABLE_ALL; + + sample_desc->Count = 1; + sample_desc->Quality = 0; + + desc->sample_mask = D3D12_DEFAULT_SAMPLE_MASK; +} + +static void pipeline_state_desc_from_d3d12_graphics_desc(struct d3d12_pipeline_state_desc *desc, + const D3D12_GRAPHICS_PIPELINE_STATE_DESC *d3d12_desc) +{ + memset(desc, 0, sizeof(*desc)); + desc->root_signature = d3d12_desc->pRootSignature; + desc->vs = d3d12_desc->VS; + desc->ps = d3d12_desc->PS; + desc->ds = d3d12_desc->DS; + desc->hs = d3d12_desc->HS; + desc->gs = d3d12_desc->GS; + desc->stream_output = d3d12_desc->StreamOutput; + desc->blend_state = d3d12_desc->BlendState; + desc->sample_mask = d3d12_desc->SampleMask; + desc->rasterizer_state = d3d12_desc->RasterizerState; + memcpy(&desc->depth_stencil_state, &d3d12_desc->DepthStencilState, sizeof(d3d12_desc->DepthStencilState)); + desc->input_layout = d3d12_desc->InputLayout; + desc->strip_cut_value = d3d12_desc->IBStripCutValue; + desc->primitive_topology_type = d3d12_desc->PrimitiveTopologyType; + desc->rtv_formats.NumRenderTargets = d3d12_desc->NumRenderTargets; + memcpy(desc->rtv_formats.RTFormats, d3d12_desc->RTVFormats, sizeof(desc->rtv_formats.RTFormats)); + desc->dsv_format = d3d12_desc->DSVFormat; + desc->sample_desc = d3d12_desc->SampleDesc; + desc->node_mask = d3d12_desc->NodeMask; + desc->cached_pso = d3d12_desc->CachedPSO; + desc->flags = d3d12_desc->Flags; +} + +static void pipeline_state_desc_from_d3d12_compute_desc(struct d3d12_pipeline_state_desc *desc, + const D3D12_COMPUTE_PIPELINE_STATE_DESC *d3d12_desc) +{ + memset(desc, 0, sizeof(*desc)); + desc->root_signature = d3d12_desc->pRootSignature; + desc->cs = d3d12_desc->CS; + desc->node_mask = d3d12_desc->NodeMask; + desc->cached_pso = d3d12_desc->CachedPSO; + desc->flags = d3d12_desc->Flags; +} + +static HRESULT pipeline_state_desc_from_d3d12_stream_desc(struct d3d12_pipeline_state_desc *desc, + const D3D12_PIPELINE_STATE_STREAM_DESC *d3d12_desc, VkPipelineBindPoint *vk_bind_point) +{ + D3D12_PIPELINE_STATE_SUBOBJECT_TYPE subobject_type; + uint64_t defined_subobjects = 0; + const uint8_t *stream_ptr; + uint64_t subobject_bit; + size_t start, size, i; + uint8_t *desc_bytes; + + static const struct + { + size_t alignment; + size_t size; + size_t dst_offset; + } + subobject_info[] = + { +#define DCL_SUBOBJECT_INFO(type, field) {__alignof__(type), sizeof(type), offsetof(struct d3d12_pipeline_state_desc, field)} + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_ROOT_SIGNATURE] = DCL_SUBOBJECT_INFO(ID3D12RootSignature *, root_signature), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_VS] = DCL_SUBOBJECT_INFO(D3D12_SHADER_BYTECODE, vs), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_PS] = DCL_SUBOBJECT_INFO(D3D12_SHADER_BYTECODE, ps), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_DS] = DCL_SUBOBJECT_INFO(D3D12_SHADER_BYTECODE, ds), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_HS] = DCL_SUBOBJECT_INFO(D3D12_SHADER_BYTECODE, hs), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_GS] = DCL_SUBOBJECT_INFO(D3D12_SHADER_BYTECODE, gs), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_CS] = DCL_SUBOBJECT_INFO(D3D12_SHADER_BYTECODE, cs), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_STREAM_OUTPUT] = DCL_SUBOBJECT_INFO(D3D12_STREAM_OUTPUT_DESC, stream_output), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_BLEND] = DCL_SUBOBJECT_INFO(D3D12_BLEND_DESC, blend_state), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_SAMPLE_MASK] = DCL_SUBOBJECT_INFO(UINT, sample_mask), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_RASTERIZER] = DCL_SUBOBJECT_INFO(D3D12_RASTERIZER_DESC, rasterizer_state), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_DEPTH_STENCIL] = DCL_SUBOBJECT_INFO(D3D12_DEPTH_STENCIL_DESC, depth_stencil_state), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_INPUT_LAYOUT] = DCL_SUBOBJECT_INFO(D3D12_INPUT_LAYOUT_DESC, input_layout), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_IB_STRIP_CUT_VALUE] = DCL_SUBOBJECT_INFO(D3D12_INDEX_BUFFER_STRIP_CUT_VALUE, strip_cut_value), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_PRIMITIVE_TOPOLOGY] = DCL_SUBOBJECT_INFO(D3D12_PRIMITIVE_TOPOLOGY_TYPE, primitive_topology_type), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_RENDER_TARGET_FORMATS] = DCL_SUBOBJECT_INFO(D3D12_RT_FORMAT_ARRAY, rtv_formats), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_DEPTH_STENCIL_FORMAT] = DCL_SUBOBJECT_INFO(DXGI_FORMAT, dsv_format), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_SAMPLE_DESC] = DCL_SUBOBJECT_INFO(DXGI_SAMPLE_DESC, sample_desc), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_NODE_MASK] = DCL_SUBOBJECT_INFO(UINT, node_mask), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_CACHED_PSO] = DCL_SUBOBJECT_INFO(D3D12_CACHED_PIPELINE_STATE, cached_pso), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_FLAGS] = DCL_SUBOBJECT_INFO(D3D12_PIPELINE_STATE_FLAGS, flags), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_DEPTH_STENCIL1] = DCL_SUBOBJECT_INFO(D3D12_DEPTH_STENCIL_DESC1, depth_stencil_state), + [D3D12_PIPELINE_STATE_SUBOBJECT_TYPE_VIEW_INSTANCING] = DCL_SUBOBJECT_INFO(D3D12_VIEW_INSTANCING_DESC, view_instancing_desc), +#undef DCL_SUBOBJECT_INFO + }; + STATIC_ASSERT(ARRAY_SIZE(subobject_info) <= sizeof(defined_subobjects) * CHAR_BIT); + + /* Initialize defaults for undefined subobjects. */ + d3d12_init_pipeline_state_desc(desc); + + stream_ptr = d3d12_desc->pPipelineStateSubobjectStream; + desc_bytes = (uint8_t *)desc; + + for (i = 0; i < d3d12_desc->SizeInBytes; ) + { + if (!vkd3d_bound_range(0, sizeof(subobject_type), d3d12_desc->SizeInBytes - i)) + { + WARN("Invalid pipeline state stream.\n"); + return E_INVALIDARG; + } + + subobject_type = *(const D3D12_PIPELINE_STATE_SUBOBJECT_TYPE *)&stream_ptr[i]; + if (subobject_type >= ARRAY_SIZE(subobject_info)) + { + FIXME("Unhandled pipeline subobject type %#x.\n", subobject_type); + return E_INVALIDARG; + } + + subobject_bit = 1ull << subobject_type; + if (defined_subobjects & subobject_bit) + { + WARN("Duplicate pipeline subobject type %u.\n", subobject_type); + return E_INVALIDARG; + } + defined_subobjects |= subobject_bit; + + start = align(sizeof(subobject_type), subobject_info[subobject_type].alignment); + size = subobject_info[subobject_type].size; + + if (!vkd3d_bound_range(start, size, d3d12_desc->SizeInBytes - i)) + { + WARN("Invalid pipeline state stream.\n"); + return E_INVALIDARG; + } + + memcpy(&desc_bytes[subobject_info[subobject_type].dst_offset], &stream_ptr[i + start], size); + /* Stream packets are aligned to the size of pointers. */ + i += align(start + size, sizeof(void *)); + } + + /* Deduce pipeline type from specified shaders. */ + if (desc->vs.BytecodeLength && desc->vs.pShaderBytecode) + { + *vk_bind_point = VK_PIPELINE_BIND_POINT_GRAPHICS; + } + else if (desc->cs.BytecodeLength && desc->cs.pShaderBytecode) + { + *vk_bind_point = VK_PIPELINE_BIND_POINT_COMPUTE; + } + else + { + WARN("Cannot deduce pipeline type from shader stages.\n"); + return E_INVALIDARG; + } + + if (desc->vs.BytecodeLength && desc->vs.pShaderBytecode + && desc->cs.BytecodeLength && desc->cs.pShaderBytecode) + { + WARN("Invalid combination of shader stages VS and CS.\n"); + return E_INVALIDARG; + } + + return S_OK; +} + struct vkd3d_pipeline_key { D3D12_PRIMITIVE_TOPOLOGY topology; @@ -2193,7 +2376,7 @@ static HRESULT d3d12_pipeline_state_find_and_init_uav_counters(struct d3d12_pipe } static HRESULT d3d12_pipeline_state_init_compute(struct d3d12_pipeline_state *state, - struct d3d12_device *device, const D3D12_COMPUTE_PIPELINE_STATE_DESC *desc) + struct d3d12_device *device, const struct d3d12_pipeline_state_desc *desc) { const struct vkd3d_vk_device_procs *vk_procs = &device->vk_procs; struct vkd3d_shader_interface_info shader_interface; @@ -2208,14 +2391,14 @@ static HRESULT d3d12_pipeline_state_init_compute(struct d3d12_pipeline_state *st memset(&state->uav_counters, 0, sizeof(state->uav_counters)); - if (!(root_signature = unsafe_impl_from_ID3D12RootSignature(desc->pRootSignature))) + if (!(root_signature = unsafe_impl_from_ID3D12RootSignature(desc->root_signature))) { WARN("Root signature is NULL.\n"); return E_INVALIDARG; } if (FAILED(hr = d3d12_pipeline_state_find_and_init_uav_counters(state, device, root_signature, - &desc->CS, VK_SHADER_STAGE_COMPUTE_BIT))) + &desc->cs, VK_SHADER_STAGE_COMPUTE_BIT))) return hr; memset(&target_info, 0, sizeof(target_info)); @@ -2256,7 +2439,7 @@ static HRESULT d3d12_pipeline_state_init_compute(struct d3d12_pipeline_state *st vk_pipeline_layout = state->uav_counters.vk_pipeline_layout ? state->uav_counters.vk_pipeline_layout : root_signature->vk_pipeline_layout; - if (FAILED(hr = vkd3d_create_compute_pipeline(device, &desc->CS, &shader_interface, + if (FAILED(hr = vkd3d_create_compute_pipeline(device, &desc->cs, &shader_interface, vk_pipeline_layout, &state->u.compute.vk_pipeline))) { WARN("Failed to create Vulkan compute pipeline, hr %#x.\n", hr); @@ -2280,13 +2463,16 @@ static HRESULT d3d12_pipeline_state_init_compute(struct d3d12_pipeline_state *st HRESULT d3d12_pipeline_state_create_compute(struct d3d12_device *device, const D3D12_COMPUTE_PIPELINE_STATE_DESC *desc, struct d3d12_pipeline_state **state) { + struct d3d12_pipeline_state_desc pipeline_desc; struct d3d12_pipeline_state *object; HRESULT hr; + pipeline_state_desc_from_d3d12_compute_desc(&pipeline_desc, desc); + if (!(object = vkd3d_malloc(sizeof(*object)))) return E_OUTOFMEMORY; - if (FAILED(hr = d3d12_pipeline_state_init_compute(object, device, desc))) + if (FAILED(hr = d3d12_pipeline_state_init_compute(object, device, &pipeline_desc))) { vkd3d_free(object); return hr; @@ -2457,7 +2643,7 @@ static void vk_stencil_op_state_from_d3d12(struct VkStencilOpState *vk_desc, } static void ds_desc_from_d3d12(struct VkPipelineDepthStencilStateCreateInfo *vk_desc, - const D3D12_DEPTH_STENCIL_DESC *d3d12_desc) + const D3D12_DEPTH_STENCIL_DESC1 *d3d12_desc) { memset(vk_desc, 0, sizeof(*vk_desc)); vk_desc->sType = VK_STRUCTURE_TYPE_PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO; @@ -2473,7 +2659,7 @@ static void ds_desc_from_d3d12(struct VkPipelineDepthStencilStateCreateInfo *vk_ vk_desc->depthWriteEnable = VK_FALSE; vk_desc->depthCompareOp = VK_COMPARE_OP_NEVER; } - vk_desc->depthBoundsTestEnable = VK_FALSE; + vk_desc->depthBoundsTestEnable = d3d12_desc->DepthBoundsTestEnable; if ((vk_desc->stencilTestEnable = d3d12_desc->StencilEnable)) { vk_stencil_op_state_from_d3d12(&vk_desc->front, &d3d12_desc->FrontFace, @@ -2738,12 +2924,12 @@ static VkLogicOp vk_logic_op_from_d3d12(D3D12_LOGIC_OP op) } static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *state, - struct d3d12_device *device, const D3D12_GRAPHICS_PIPELINE_STATE_DESC *desc) + struct d3d12_device *device, const struct d3d12_pipeline_state_desc *desc) { unsigned int ps_output_swizzle[D3D12_SIMULTANEOUS_RENDER_TARGET_COUNT]; struct d3d12_graphics_pipeline_state *graphics = &state->u.graphics; const struct vkd3d_vk_device_procs *vk_procs = &device->vk_procs; - const D3D12_STREAM_OUTPUT_DESC *so_desc = &desc->StreamOutput; + const D3D12_STREAM_OUTPUT_DESC *so_desc = &desc->stream_output; VkVertexInputBindingDivisorDescriptionEXT *binding_divisor; const struct vkd3d_vulkan_info *vk_info = &device->vk_info; uint32_t instance_divisors[D3D12_VS_INPUT_REGISTER_COUNT]; @@ -2787,11 +2973,11 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s } shader_stages[] = { - {VK_SHADER_STAGE_VERTEX_BIT, offsetof(D3D12_GRAPHICS_PIPELINE_STATE_DESC, VS)}, - {VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT, offsetof(D3D12_GRAPHICS_PIPELINE_STATE_DESC, HS)}, - {VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT, offsetof(D3D12_GRAPHICS_PIPELINE_STATE_DESC, DS)}, - {VK_SHADER_STAGE_GEOMETRY_BIT, offsetof(D3D12_GRAPHICS_PIPELINE_STATE_DESC, GS)}, - {VK_SHADER_STAGE_FRAGMENT_BIT, offsetof(D3D12_GRAPHICS_PIPELINE_STATE_DESC, PS)}, + {VK_SHADER_STAGE_VERTEX_BIT, offsetof(struct d3d12_pipeline_state_desc, vs)}, + {VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT, offsetof(struct d3d12_pipeline_state_desc, hs)}, + {VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT, offsetof(struct d3d12_pipeline_state_desc, ds)}, + {VK_SHADER_STAGE_GEOMETRY_BIT, offsetof(struct d3d12_pipeline_state_desc, gs)}, + {VK_SHADER_STAGE_FRAGMENT_BIT, offsetof(struct d3d12_pipeline_state_desc, ps)}, }; state->ID3D12PipelineState_iface.lpVtbl = &d3d12_pipeline_state_vtbl; @@ -2802,26 +2988,26 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s memset(&input_signature, 0, sizeof(input_signature)); - for (i = desc->NumRenderTargets; i < ARRAY_SIZE(desc->RTVFormats); ++i) + for (i = desc->rtv_formats.NumRenderTargets; i < ARRAY_SIZE(desc->rtv_formats.RTFormats); ++i) { - if (desc->RTVFormats[i] != DXGI_FORMAT_UNKNOWN) + if (desc->rtv_formats.RTFormats[i] != DXGI_FORMAT_UNKNOWN) { WARN("Format must be set to DXGI_FORMAT_UNKNOWN for inactive render targets.\n"); return E_INVALIDARG; } } - if (!(root_signature = unsafe_impl_from_ID3D12RootSignature(desc->pRootSignature))) + if (!(root_signature = unsafe_impl_from_ID3D12RootSignature(desc->root_signature))) { WARN("Root signature is NULL.\n"); return E_INVALIDARG; } - sample_count = vk_samples_from_dxgi_sample_desc(&desc->SampleDesc); - if (desc->SampleDesc.Count != 1 && desc->SampleDesc.Quality) - WARN("Ignoring sample quality %u.\n", desc->SampleDesc.Quality); + sample_count = vk_samples_from_dxgi_sample_desc(&desc->sample_desc); + if (desc->sample_desc.Count != 1 && desc->sample_desc.Quality) + WARN("Ignoring sample quality %u.\n", desc->sample_desc.Quality); - rt_count = desc->NumRenderTargets; + rt_count = desc->rtv_formats.NumRenderTargets; if (rt_count > ARRAY_SIZE(graphics->blend_attachments)) { FIXME("NumRenderTargets %zu > %zu, ignoring extra formats.\n", @@ -2829,40 +3015,40 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s rt_count = ARRAY_SIZE(graphics->blend_attachments); } - graphics->om_logic_op_enable = desc->BlendState.RenderTarget[0].LogicOpEnable + graphics->om_logic_op_enable = desc->blend_state.RenderTarget[0].LogicOpEnable && device->feature_options.OutputMergerLogicOp; graphics->om_logic_op = graphics->om_logic_op_enable - ? vk_logic_op_from_d3d12(desc->BlendState.RenderTarget[0].LogicOp) + ? vk_logic_op_from_d3d12(desc->blend_state.RenderTarget[0].LogicOp) : VK_LOGIC_OP_COPY; - if (desc->BlendState.RenderTarget[0].LogicOpEnable && !graphics->om_logic_op_enable) + if (desc->blend_state.RenderTarget[0].LogicOpEnable && !graphics->om_logic_op_enable) WARN("The device does not support output merger logic ops. Ignoring logic op %#x.\n", - desc->BlendState.RenderTarget[0].LogicOp); + desc->blend_state.RenderTarget[0].LogicOp); graphics->null_attachment_mask = 0; for (i = 0; i < rt_count; ++i) { const D3D12_RENDER_TARGET_BLEND_DESC *rt_desc; - if (desc->RTVFormats[i] == DXGI_FORMAT_UNKNOWN) + if (desc->rtv_formats.RTFormats[i] == DXGI_FORMAT_UNKNOWN) { graphics->null_attachment_mask |= 1u << i; ps_output_swizzle[i] = VKD3D_SHADER_NO_SWIZZLE; graphics->rtv_formats[i] = VK_FORMAT_UNDEFINED; } - else if ((format = vkd3d_get_format(device, desc->RTVFormats[i], false))) + else if ((format = vkd3d_get_format(device, desc->rtv_formats.RTFormats[i], false))) { ps_output_swizzle[i] = vkd3d_get_rt_format_swizzle(format); graphics->rtv_formats[i] = format->vk_format; } else { - WARN("Invalid RTV format %#x.\n", desc->RTVFormats[i]); + WARN("Invalid RTV format %#x.\n", desc->rtv_formats.RTFormats[i]); hr = E_INVALIDARG; goto fail; } - rt_desc = &desc->BlendState.RenderTarget[desc->BlendState.IndependentBlendEnable ? i : 0]; - if (desc->BlendState.IndependentBlendEnable && rt_desc->LogicOpEnable) + rt_desc = &desc->blend_state.RenderTarget[desc->blend_state.IndependentBlendEnable ? i : 0]; + if (desc->blend_state.IndependentBlendEnable && rt_desc->LogicOpEnable) { WARN("IndependentBlendEnable must be FALSE when logic operations are enabled.\n"); hr = E_INVALIDARG; @@ -2881,8 +3067,14 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s graphics->rtv_formats[i] = VK_FORMAT_UNDEFINED; graphics->rt_count = rt_count; - ds_desc_from_d3d12(&graphics->ds_desc, &desc->DepthStencilState); - if (desc->DSVFormat == DXGI_FORMAT_UNKNOWN + ds_desc_from_d3d12(&graphics->ds_desc, &desc->depth_stencil_state); + if (graphics->ds_desc.depthBoundsTestEnable && !device->feature_options2.DepthBoundsTestSupported) + { + WARN("Depth bounds test not supported by device.\n"); + hr = E_INVALIDARG; + goto fail; + } + if (desc->dsv_format == DXGI_FORMAT_UNKNOWN && graphics->ds_desc.depthTestEnable && !graphics->ds_desc.depthWriteEnable && graphics->ds_desc.depthCompareOp == VK_COMPARE_OP_ALWAYS && !graphics->ds_desc.stencilTestEnable) { @@ -2891,15 +3083,16 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s } graphics->dsv_format = VK_FORMAT_UNDEFINED; - if (graphics->ds_desc.depthTestEnable || graphics->ds_desc.stencilTestEnable) + if (graphics->ds_desc.depthTestEnable || graphics->ds_desc.stencilTestEnable + || graphics->ds_desc.depthBoundsTestEnable) { - if (desc->DSVFormat == DXGI_FORMAT_UNKNOWN) + if (desc->dsv_format == DXGI_FORMAT_UNKNOWN) { WARN("DSV format is DXGI_FORMAT_UNKNOWN.\n"); graphics->dsv_format = VK_FORMAT_UNDEFINED; graphics->null_attachment_mask |= dsv_attachment_mask(graphics); } - else if ((format = vkd3d_get_format(device, desc->DSVFormat, true))) + else if ((format = vkd3d_get_format(device, desc->dsv_format, true))) { if (!(format->vk_aspect_mask & (VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT))) FIXME("Format %#x is not depth/stencil format.\n", format->dxgi_format); @@ -2908,12 +3101,12 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s } else { - WARN("Invalid DSV format %#x.\n", desc->DSVFormat); + WARN("Invalid DSV format %#x.\n", desc->dsv_format); hr = E_INVALIDARG; goto fail; } - if (!desc->PS.pShaderBytecode) + if (!desc->ps.pShaderBytecode) { if (FAILED(hr = create_shader_stage(device, &graphics->stages[graphics->stage_count], VK_SHADER_STAGE_FRAGMENT_BIT, &default_ps, NULL))) @@ -2936,7 +3129,7 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s ps_target_info.extension_count = vk_info->shader_extension_count; ps_target_info.parameters = ps_shader_parameters; ps_target_info.parameter_count = ARRAY_SIZE(ps_shader_parameters); - ps_target_info.dual_source_blending = is_dual_source_blending(&desc->BlendState.RenderTarget[0]); + ps_target_info.dual_source_blending = is_dual_source_blending(&desc->blend_state.RenderTarget[0]); ps_target_info.output_swizzles = ps_output_swizzle; ps_target_info.output_swizzle_count = rt_count; @@ -2946,11 +3139,11 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s hr = E_INVALIDARG; goto fail; } - if (ps_target_info.dual_source_blending && desc->BlendState.IndependentBlendEnable) + if (ps_target_info.dual_source_blending && desc->blend_state.IndependentBlendEnable) { - for (i = 1; i < ARRAY_SIZE(desc->BlendState.RenderTarget); ++i) + for (i = 1; i < ARRAY_SIZE(desc->blend_state.RenderTarget); ++i) { - if (desc->BlendState.RenderTarget[i].BlendEnable) + if (desc->blend_state.RenderTarget[i].BlendEnable) { WARN("Blend enable cannot be set for render target %u when dual source blending is used.\n", i); hr = E_INVALIDARG; @@ -2992,9 +3185,9 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s xfb_info.buffer_strides = so_desc->pBufferStrides; xfb_info.buffer_stride_count = so_desc->NumStrides; - if (desc->GS.pShaderBytecode) + if (desc->gs.pShaderBytecode) xfb_stage = VK_SHADER_STAGE_GEOMETRY_BIT; - else if (desc->DS.pShaderBytecode) + else if (desc->ds.pShaderBytecode) xfb_stage = VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT; else xfb_stage = VK_SHADER_STAGE_VERTEX_BIT; @@ -3046,7 +3239,7 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s case VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT: case VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT: - if (desc->PrimitiveTopologyType != D3D12_PRIMITIVE_TOPOLOGY_TYPE_PATCH) + if (desc->primitive_topology_type != D3D12_PRIMITIVE_TOPOLOGY_TYPE_PATCH) { WARN("D3D12_PRIMITIVE_TOPOLOGY_TYPE_PATCH must be used with tessellation shaders.\n"); hr = E_INVALIDARG; @@ -3088,7 +3281,7 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s ++graphics->stage_count; } - graphics->attribute_count = desc->InputLayout.NumElements; + graphics->attribute_count = desc->input_layout.NumElements; if (graphics->attribute_count > ARRAY_SIZE(graphics->attributes)) { FIXME("InputLayout.NumElements %zu > %zu, ignoring extra elements.\n", @@ -3104,13 +3297,13 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s goto fail; } - if (FAILED(hr = compute_input_layout_offsets(device, &desc->InputLayout, aligned_offsets))) + if (FAILED(hr = compute_input_layout_offsets(device, &desc->input_layout, aligned_offsets))) goto fail; graphics->instance_divisor_count = 0; for (i = 0, j = 0, mask = 0; i < graphics->attribute_count; ++i) { - const D3D12_INPUT_ELEMENT_DESC *e = &desc->InputLayout.pInputElementDescs[i]; + const D3D12_INPUT_ELEMENT_DESC *e = &desc->input_layout.pInputElementDescs[i]; const struct vkd3d_shader_signature_element *signature_element; /* TODO: DXGI_FORMAT_UNKNOWN will succeed here, which may not match @@ -3194,30 +3387,30 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s graphics->attribute_count = j; vkd3d_shader_free_shader_signature(&input_signature); - switch (desc->IBStripCutValue) + switch (desc->strip_cut_value) { case D3D12_INDEX_BUFFER_STRIP_CUT_VALUE_DISABLED: case D3D12_INDEX_BUFFER_STRIP_CUT_VALUE_0xFFFF: case D3D12_INDEX_BUFFER_STRIP_CUT_VALUE_0xFFFFFFFF: - graphics->index_buffer_strip_cut_value = desc->IBStripCutValue; + graphics->index_buffer_strip_cut_value = desc->strip_cut_value; break; default: - WARN("Invalid index buffer strip cut value %#x.\n", desc->IBStripCutValue); + WARN("Invalid index buffer strip cut value %#x.\n", desc->strip_cut_value); hr = E_INVALIDARG; goto fail; } is_dsv_format_unknown = graphics->null_attachment_mask & dsv_attachment_mask(graphics); - rs_desc_from_d3d12(&graphics->rs_desc, &desc->RasterizerState); + rs_desc_from_d3d12(&graphics->rs_desc, &desc->rasterizer_state); have_attachment = graphics->rt_count || graphics->dsv_format || is_dsv_format_unknown; - if ((!have_attachment && !(desc->PS.pShaderBytecode && desc->PS.BytecodeLength)) + if ((!have_attachment && !(desc->ps.pShaderBytecode && desc->ps.BytecodeLength)) || (graphics->xfb_enabled && so_desc->RasterizedStream == D3D12_SO_NO_RASTERIZED_STREAM)) graphics->rs_desc.rasterizerDiscardEnable = VK_TRUE; rs_stream_info_from_d3d12(&graphics->rs_stream_info, &graphics->rs_desc, so_desc, vk_info); if (vk_info->EXT_depth_clip_enable) - rs_depth_clip_info_from_d3d12(&graphics->rs_depth_clip_info, &graphics->rs_desc, &desc->RasterizerState); + rs_depth_clip_info_from_d3d12(&graphics->rs_depth_clip_info, &graphics->rs_desc, &desc->rasterizer_state); graphics->ms_desc.sType = VK_STRUCTURE_TYPE_PIPELINE_MULTISAMPLE_STATE_CREATE_INFO; graphics->ms_desc.pNext = NULL; @@ -3226,17 +3419,24 @@ static HRESULT d3d12_pipeline_state_init_graphics(struct d3d12_pipeline_state *s graphics->ms_desc.sampleShadingEnable = VK_FALSE; graphics->ms_desc.minSampleShading = 0.0f; graphics->ms_desc.pSampleMask = NULL; - if (desc->SampleMask != ~0u) + if (desc->sample_mask != ~0u) { assert(DIV_ROUND_UP(sample_count, 32) <= ARRAY_SIZE(graphics->sample_mask)); - graphics->sample_mask[0] = desc->SampleMask; + graphics->sample_mask[0] = desc->sample_mask; graphics->sample_mask[1] = 0xffffffffu; graphics->ms_desc.pSampleMask = graphics->sample_mask; } - graphics->ms_desc.alphaToCoverageEnable = desc->BlendState.AlphaToCoverageEnable; + graphics->ms_desc.alphaToCoverageEnable = desc->blend_state.AlphaToCoverageEnable; graphics->ms_desc.alphaToOneEnable = VK_FALSE; - /* We defer creating the render pass for pipelines wth DSVFormat equal to + if (desc->view_instancing_desc.ViewInstanceCount) + { + FIXME("View instancing is not supported yet.\n"); + hr = E_INVALIDARG; + goto fail; + } + + /* We defer creating the render pass for pipelines with DSVFormat equal to * DXGI_FORMAT_UNKNOWN. We take the actual DSV format from the bound DSV. */ if (is_dsv_format_unknown) graphics->render_pass = VK_NULL_HANDLE; @@ -3271,13 +3471,16 @@ fail: HRESULT d3d12_pipeline_state_create_graphics(struct d3d12_device *device, const D3D12_GRAPHICS_PIPELINE_STATE_DESC *desc, struct d3d12_pipeline_state **state) { + struct d3d12_pipeline_state_desc pipeline_desc; struct d3d12_pipeline_state *object; HRESULT hr; + pipeline_state_desc_from_d3d12_graphics_desc(&pipeline_desc, desc); + if (!(object = vkd3d_malloc(sizeof(*object)))) return E_OUTOFMEMORY; - if (FAILED(hr = d3d12_pipeline_state_init_graphics(object, device, desc))) + if (FAILED(hr = d3d12_pipeline_state_init_graphics(object, device, &pipeline_desc))) { vkd3d_free(object); return hr; @@ -3290,6 +3493,46 @@ HRESULT d3d12_pipeline_state_create_graphics(struct d3d12_device *device, return S_OK; } +HRESULT d3d12_pipeline_state_create(struct d3d12_device *device, + const D3D12_PIPELINE_STATE_STREAM_DESC *desc, struct d3d12_pipeline_state **state) +{ + struct d3d12_pipeline_state_desc pipeline_desc; + struct d3d12_pipeline_state *object; + VkPipelineBindPoint bind_point; + HRESULT hr; + + if (FAILED(hr = pipeline_state_desc_from_d3d12_stream_desc(&pipeline_desc, desc, &bind_point))) + return hr; + + if (!(object = vkd3d_calloc(1, sizeof(*object)))) + return E_OUTOFMEMORY; + + switch (bind_point) + { + case VK_PIPELINE_BIND_POINT_COMPUTE: + hr = d3d12_pipeline_state_init_compute(object, device, &pipeline_desc); + break; + + case VK_PIPELINE_BIND_POINT_GRAPHICS: + hr = d3d12_pipeline_state_init_graphics(object, device, &pipeline_desc); + break; + + default: + vkd3d_unreachable(); + } + + if (FAILED(hr)) + { + vkd3d_free(object); + return hr; + } + + TRACE("Created pipeline state %p.\n", object); + + *state = object; + return S_OK; +} + static enum VkPrimitiveTopology vk_topology_from_d3d12_topology(D3D12_PRIMITIVE_TOPOLOGY topology) { switch (topology) @@ -3605,6 +3848,36 @@ VkPipeline d3d12_pipeline_state_get_or_create_pipeline(struct d3d12_pipeline_sta return vk_pipeline; } +static int compile_hlsl_cs(const struct vkd3d_shader_code *hlsl, struct vkd3d_shader_code *dxbc) +{ + struct vkd3d_shader_hlsl_source_info hlsl_info; + struct vkd3d_shader_compile_info info; + + static const struct vkd3d_shader_compile_option options[] = + { + {VKD3D_SHADER_COMPILE_OPTION_API_VERSION, VKD3D_SHADER_API_VERSION_1_9}, + }; + + info.type = VKD3D_SHADER_STRUCTURE_TYPE_COMPILE_INFO; + info.next = &hlsl_info; + info.source = *hlsl; + info.source_type = VKD3D_SHADER_SOURCE_HLSL; + info.target_type = VKD3D_SHADER_TARGET_DXBC_TPF; + info.options = options; + info.option_count = ARRAY_SIZE(options); + info.log_level = VKD3D_SHADER_LOG_NONE; + info.source_name = NULL; + + hlsl_info.type = VKD3D_SHADER_STRUCTURE_TYPE_HLSL_SOURCE_INFO; + hlsl_info.next = NULL; + hlsl_info.entry_point = "main"; + hlsl_info.secondary_code.code = NULL; + hlsl_info.secondary_code.size = 0; + hlsl_info.profile = "cs_5_0"; + + return vkd3d_shader_compile(&info, dxbc, NULL); +} + static void vkd3d_uav_clear_pipelines_cleanup(struct vkd3d_uav_clear_pipelines *pipelines, struct d3d12_device *device) { @@ -3658,7 +3931,7 @@ HRESULT vkd3d_uav_clear_state_init(struct vkd3d_uav_clear_state *state, struct d { VkPipeline *pipeline; VkPipelineLayout *pipeline_layout; - D3D12_SHADER_BYTECODE code; + struct vkd3d_shader_code code; } pipelines[] = { @@ -3748,13 +4021,25 @@ HRESULT vkd3d_uav_clear_state_init(struct vkd3d_uav_clear_state *state, struct d for (i = 0; i < ARRAY_SIZE(pipelines); ++i) { + struct vkd3d_shader_code dxbc; + int ret; + + if ((ret = compile_hlsl_cs(&pipelines[i].code, &dxbc))) + { + ERR("Failed to compile HLSL compute shader %u, ret %d.\n", i, ret); + hr = hresult_from_vk_result(ret); + goto fail; + } + if (pipelines[i].pipeline_layout == &state->vk_pipeline_layout_buffer) binding.flags = VKD3D_SHADER_BINDING_FLAG_BUFFER; else binding.flags = VKD3D_SHADER_BINDING_FLAG_IMAGE; - if (FAILED(hr = vkd3d_create_compute_pipeline(device, &pipelines[i].code, &shader_interface, - *pipelines[i].pipeline_layout, pipelines[i].pipeline))) + hr = vkd3d_create_compute_pipeline(device, &(D3D12_SHADER_BYTECODE){dxbc.code, dxbc.size}, + &shader_interface, *pipelines[i].pipeline_layout, pipelines[i].pipeline); + vkd3d_shader_free_shader_code(&dxbc); + if (FAILED(hr)) { ERR("Failed to create compute pipeline %u, hr %#x.\n", i, hr); goto fail; diff --git a/libs/vkd3d/libs/vkd3d/vkd3d_main.c b/libs/vkd3d/libs/vkd3d/vkd3d_main.c index 159560afd8e..245edb5aeac 100644 --- a/libs/vkd3d/libs/vkd3d/vkd3d_main.c +++ b/libs/vkd3d/libs/vkd3d/vkd3d_main.c @@ -71,11 +71,11 @@ HRESULT vkd3d_create_device(const struct vkd3d_device_create_info *create_info, if (!device) { - ID3D12Device_Release(&object->ID3D12Device1_iface); + ID3D12Device_Release(&object->ID3D12Device5_iface); return S_FALSE; } - return return_interface(&object->ID3D12Device1_iface, &IID_ID3D12Device, iid, device); + return return_interface(&object->ID3D12Device5_iface, &IID_ID3D12Device, iid, device); } /* ID3D12RootSignatureDeserializer */ diff --git a/libs/vkd3d/libs/vkd3d/vkd3d_private.h b/libs/vkd3d/libs/vkd3d/vkd3d_private.h index 363a7132c3a..231bc615e5c 100644 --- a/libs/vkd3d/libs/vkd3d/vkd3d_private.h +++ b/libs/vkd3d/libs/vkd3d/vkd3d_private.h @@ -22,6 +22,9 @@ #define COBJMACROS #define NONAMELESSUNION #define VK_NO_PROTOTYPES +#ifndef CONST_VTABLE +#define CONST_VTABLE +#endif #ifdef _WIN32 # define _WIN32_WINNT 0x0600 /* for condition variables */ @@ -31,8 +34,8 @@ #include "vkd3d_blob.h" #include "vkd3d_memory.h" #include "vkd3d_utf8.h" -#include "wine/list.h" -#include "wine/rbtree.h" +#include "list.h" +#include "rbtree.h" #include "vkd3d.h" #include "vkd3d_shader.h" @@ -121,6 +124,7 @@ struct vkd3d_vulkan_info bool KHR_draw_indirect_count; bool KHR_get_memory_requirements2; bool KHR_image_format_list; + bool KHR_maintenance2; bool KHR_maintenance3; bool KHR_push_descriptor; bool KHR_sampler_mirror_clamp_to_edge; @@ -680,7 +684,7 @@ struct d3d12_heap }; HRESULT d3d12_heap_create(struct d3d12_device *device, const D3D12_HEAP_DESC *desc, - const struct d3d12_resource *resource, struct d3d12_heap **heap); + const struct d3d12_resource *resource, ID3D12ProtectedResourceSession *protected_session, struct d3d12_heap **heap); struct d3d12_heap *unsafe_impl_from_ID3D12Heap(ID3D12Heap *iface); #define VKD3D_RESOURCE_PUBLIC_FLAGS \ @@ -716,7 +720,7 @@ struct d3d12_resource_tile_info /* ID3D12Resource */ struct d3d12_resource { - ID3D12Resource ID3D12Resource_iface; + ID3D12Resource1 ID3D12Resource1_iface; LONG refcount; LONG internal_refcount; @@ -748,7 +752,12 @@ struct d3d12_resource static inline struct d3d12_resource *impl_from_ID3D12Resource(ID3D12Resource *iface) { - return CONTAINING_RECORD(iface, struct d3d12_resource, ID3D12Resource_iface); + return CONTAINING_RECORD(iface, struct d3d12_resource, ID3D12Resource1_iface); +} + +static inline struct d3d12_resource *impl_from_ID3D12Resource1(ID3D12Resource1 *iface) +{ + return CONTAINING_RECORD(iface, struct d3d12_resource, ID3D12Resource1_iface); } static inline bool d3d12_resource_is_buffer(const struct d3d12_resource *resource) @@ -771,7 +780,8 @@ void d3d12_resource_get_tiling(struct d3d12_device *device, const struct d3d12_r HRESULT d3d12_committed_resource_create(struct d3d12_device *device, const D3D12_HEAP_PROPERTIES *heap_properties, D3D12_HEAP_FLAGS heap_flags, const D3D12_RESOURCE_DESC *desc, D3D12_RESOURCE_STATES initial_state, - const D3D12_CLEAR_VALUE *optimized_clear_value, struct d3d12_resource **resource); + const D3D12_CLEAR_VALUE *optimized_clear_value, ID3D12ProtectedResourceSession *protected_session, + struct d3d12_resource **resource); HRESULT d3d12_placed_resource_create(struct d3d12_device *device, struct d3d12_heap *heap, uint64_t heap_offset, const D3D12_RESOURCE_DESC *desc, D3D12_RESOURCE_STATES initial_state, const D3D12_CLEAR_VALUE *optimized_clear_value, struct d3d12_resource **resource); @@ -835,6 +845,7 @@ struct vkd3d_texture_view_desc VkImageAspectFlags vk_image_aspect; VkComponentMapping components; bool allowed_swizzle; + VkImageUsageFlags usage; }; struct vkd3d_desc_header @@ -1090,7 +1101,7 @@ HRESULT d3d12_query_heap_create(struct d3d12_device *device, struct d3d12_query_heap *unsafe_impl_from_ID3D12QueryHeap(ID3D12QueryHeap *iface); /* A Vulkan query has to be issued at least one time before the result is - * available. In D3D12 it is legal to get query reults for not issued queries. + * available. In D3D12 it is legal to get query results for not issued queries. */ static inline bool d3d12_query_heap_is_result_available(const struct d3d12_query_heap *heap, unsigned int query_index) @@ -1308,10 +1319,38 @@ static inline bool d3d12_pipeline_state_has_unknown_dsv_format(struct d3d12_pipe return false; } +struct d3d12_pipeline_state_desc +{ + ID3D12RootSignature *root_signature; + D3D12_SHADER_BYTECODE vs; + D3D12_SHADER_BYTECODE ps; + D3D12_SHADER_BYTECODE ds; + D3D12_SHADER_BYTECODE hs; + D3D12_SHADER_BYTECODE gs; + D3D12_SHADER_BYTECODE cs; + D3D12_STREAM_OUTPUT_DESC stream_output; + D3D12_BLEND_DESC blend_state; + unsigned int sample_mask; + D3D12_RASTERIZER_DESC rasterizer_state; + D3D12_DEPTH_STENCIL_DESC1 depth_stencil_state; + D3D12_INPUT_LAYOUT_DESC input_layout; + D3D12_INDEX_BUFFER_STRIP_CUT_VALUE strip_cut_value; + D3D12_PRIMITIVE_TOPOLOGY_TYPE primitive_topology_type; + D3D12_RT_FORMAT_ARRAY rtv_formats; + DXGI_FORMAT dsv_format; + DXGI_SAMPLE_DESC sample_desc; + D3D12_VIEW_INSTANCING_DESC view_instancing_desc; + unsigned int node_mask; + D3D12_CACHED_PIPELINE_STATE cached_pso; + D3D12_PIPELINE_STATE_FLAGS flags; +}; + HRESULT d3d12_pipeline_state_create_compute(struct d3d12_device *device, const D3D12_COMPUTE_PIPELINE_STATE_DESC *desc, struct d3d12_pipeline_state **state); HRESULT d3d12_pipeline_state_create_graphics(struct d3d12_device *device, const D3D12_GRAPHICS_PIPELINE_STATE_DESC *desc, struct d3d12_pipeline_state **state); +HRESULT d3d12_pipeline_state_create(struct d3d12_device *device, + const D3D12_PIPELINE_STATE_STREAM_DESC *desc, struct d3d12_pipeline_state **state); VkPipeline d3d12_pipeline_state_get_or_create_pipeline(struct d3d12_pipeline_state *state, D3D12_PRIMITIVE_TOPOLOGY topology, const uint32_t *strides, VkFormat dsv_format, VkRenderPass *vk_render_pass); struct d3d12_pipeline_state *unsafe_impl_from_ID3D12PipelineState(ID3D12PipelineState *iface); @@ -1426,7 +1465,7 @@ enum vkd3d_pipeline_bind_point /* ID3D12CommandList */ struct d3d12_command_list { - ID3D12GraphicsCommandList3 ID3D12GraphicsCommandList3_iface; + ID3D12GraphicsCommandList5 ID3D12GraphicsCommandList5_iface; LONG refcount; D3D12_COMMAND_LIST_TYPE type; @@ -1709,7 +1748,7 @@ struct vkd3d_desc_object_cache /* ID3D12Device */ struct d3d12_device { - ID3D12Device1 ID3D12Device1_iface; + ID3D12Device5 ID3D12Device5_iface; LONG refcount; VkDevice vk_device; @@ -1775,27 +1814,27 @@ struct vkd3d_queue *d3d12_device_get_vkd3d_queue(struct d3d12_device *device, D3 bool d3d12_device_is_uma(struct d3d12_device *device, bool *coherent); void d3d12_device_mark_as_removed(struct d3d12_device *device, HRESULT reason, const char *message, ...) VKD3D_PRINTF_FUNC(3, 4); -struct d3d12_device *unsafe_impl_from_ID3D12Device1(ID3D12Device1 *iface); +struct d3d12_device *unsafe_impl_from_ID3D12Device5(ID3D12Device5 *iface); static inline HRESULT d3d12_device_query_interface(struct d3d12_device *device, REFIID iid, void **object) { - return ID3D12Device1_QueryInterface(&device->ID3D12Device1_iface, iid, object); + return ID3D12Device5_QueryInterface(&device->ID3D12Device5_iface, iid, object); } static inline ULONG d3d12_device_add_ref(struct d3d12_device *device) { - return ID3D12Device1_AddRef(&device->ID3D12Device1_iface); + return ID3D12Device5_AddRef(&device->ID3D12Device5_iface); } static inline ULONG d3d12_device_release(struct d3d12_device *device) { - return ID3D12Device1_Release(&device->ID3D12Device1_iface); + return ID3D12Device5_Release(&device->ID3D12Device5_iface); } static inline unsigned int d3d12_device_get_descriptor_handle_increment_size(struct d3d12_device *device, D3D12_DESCRIPTOR_HEAP_TYPE descriptor_type) { - return ID3D12Device1_GetDescriptorHandleIncrementSize(&device->ID3D12Device1_iface, descriptor_type); + return ID3D12Device5_GetDescriptorHandleIncrementSize(&device->ID3D12Device5_iface, descriptor_type); } /* utils */ diff --git a/libs/vkd3d/libs/vkd3d/vkd3d_shaders.h b/libs/vkd3d/libs/vkd3d/vkd3d_shaders.h index b2a90cdbf3c..3fefe0da849 100644 --- a/libs/vkd3d/libs/vkd3d/vkd3d_shaders.h +++ b/libs/vkd3d/libs/vkd3d/vkd3d_shaders.h @@ -19,370 +19,208 @@ #ifndef __VKD3D_SHADERS_H #define __VKD3D_SHADERS_H -static const uint32_t cs_uav_clear_buffer_float_code[] = -{ -#if 0 - RWBuffer dst; - - struct - { - float4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(128, 1, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (thread_id.x < u_info.dst_extent.x) - dst[u_info.dst_offset.x + thread_id.x] = u_info.clear_value; - } -#endif - 0x43425844, 0xe114ba61, 0xff6a0d0b, 0x7b25c8f4, 0xfcf7cf22, 0x00000001, 0x0000010c, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000b8, 0x00050050, 0x0000002e, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400089c, 0x0011e000, 0x00000000, 0x00005555, - 0x0200005f, 0x00020012, 0x02000068, 0x00000001, 0x0400009b, 0x00000080, 0x00000001, 0x00000001, - 0x07000022, 0x00100012, 0x00000000, 0x0002000a, 0x0020802a, 0x00000000, 0x00000001, 0x0304001f, - 0x0010000a, 0x00000000, 0x0700001e, 0x00100012, 0x00000000, 0x0002000a, 0x0020800a, 0x00000000, - 0x00000001, 0x080000a4, 0x0011e0f2, 0x00000000, 0x00100006, 0x00000000, 0x00208e46, 0x00000000, - 0x00000000, 0x01000015, 0x0100003e, -}; - -static const uint32_t cs_uav_clear_buffer_uint_code[] = -{ -#if 0 - RWBuffer dst; - - struct - { - uint4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(128, 1, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (thread_id.x < u_info.dst_extent.x) - dst[u_info.dst_offset.x + thread_id.x] = u_info.clear_value; - } -#endif - 0x43425844, 0x3afd0cfd, 0x5145c166, 0x5b9f76b8, 0xa73775cd, 0x00000001, 0x0000010c, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000b8, 0x00050050, 0x0000002e, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400089c, 0x0011e000, 0x00000000, 0x00004444, - 0x0200005f, 0x00020012, 0x02000068, 0x00000001, 0x0400009b, 0x00000080, 0x00000001, 0x00000001, - 0x07000022, 0x00100012, 0x00000000, 0x0002000a, 0x0020802a, 0x00000000, 0x00000001, 0x0304001f, - 0x0010000a, 0x00000000, 0x0700001e, 0x00100012, 0x00000000, 0x0002000a, 0x0020800a, 0x00000000, - 0x00000001, 0x080000a4, 0x0011e0f2, 0x00000000, 0x00100006, 0x00000000, 0x00208e46, 0x00000000, - 0x00000000, 0x01000015, 0x0100003e, -}; - -static const uint32_t cs_uav_clear_1d_array_float_code[] = -{ -#if 0 - RWTexture1DArray dst; - - struct - { - float4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(64, 1, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (thread_id.x < u_info.dst_extent.x) - dst[int2(u_info.dst_offset.x + thread_id.x, thread_id.y)] = u_info.clear_value; - } -#endif - 0x43425844, 0x3d73bc2d, 0x2b635f3d, 0x6bf98e92, 0xbe0aa5d9, 0x00000001, 0x0000011c, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000c8, 0x00050050, 0x00000032, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400389c, 0x0011e000, 0x00000000, 0x00005555, - 0x0200005f, 0x00020032, 0x02000068, 0x00000001, 0x0400009b, 0x00000040, 0x00000001, 0x00000001, - 0x07000022, 0x00100012, 0x00000000, 0x0002000a, 0x0020802a, 0x00000000, 0x00000001, 0x0304001f, - 0x0010000a, 0x00000000, 0x0700001e, 0x00100012, 0x00000000, 0x0002000a, 0x0020800a, 0x00000000, - 0x00000001, 0x04000036, 0x001000e2, 0x00000000, 0x00020556, 0x080000a4, 0x0011e0f2, 0x00000000, - 0x00100e46, 0x00000000, 0x00208e46, 0x00000000, 0x00000000, 0x01000015, 0x0100003e, -}; - -static const uint32_t cs_uav_clear_1d_array_uint_code[] = -{ -#if 0 - RWTexture1DArray dst; - - struct - { - uint4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(64, 1, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (thread_id.x < u_info.dst_extent.x) - dst[int2(u_info.dst_offset.x + thread_id.x, thread_id.y)] = u_info.clear_value; - } -#endif - 0x43425844, 0x2f0ca457, 0x72068b34, 0xd9dadc2b, 0xd3178c3e, 0x00000001, 0x0000011c, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000c8, 0x00050050, 0x00000032, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400389c, 0x0011e000, 0x00000000, 0x00004444, - 0x0200005f, 0x00020032, 0x02000068, 0x00000001, 0x0400009b, 0x00000040, 0x00000001, 0x00000001, - 0x07000022, 0x00100012, 0x00000000, 0x0002000a, 0x0020802a, 0x00000000, 0x00000001, 0x0304001f, - 0x0010000a, 0x00000000, 0x0700001e, 0x00100012, 0x00000000, 0x0002000a, 0x0020800a, 0x00000000, - 0x00000001, 0x04000036, 0x001000e2, 0x00000000, 0x00020556, 0x080000a4, 0x0011e0f2, 0x00000000, - 0x00100e46, 0x00000000, 0x00208e46, 0x00000000, 0x00000000, 0x01000015, 0x0100003e, -}; - -static const uint32_t cs_uav_clear_1d_float_code[] = -{ -#if 0 - RWTexture1D dst; - - struct - { - float4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(64, 1, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (thread_id.x < u_info.dst_extent.x) - dst[u_info.dst_offset.x + thread_id.x] = u_info.clear_value; - } -#endif - 0x43425844, 0x05266503, 0x4b97006f, 0x01a5cc63, 0xe617d0a1, 0x00000001, 0x0000010c, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000b8, 0x00050050, 0x0000002e, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400109c, 0x0011e000, 0x00000000, 0x00005555, - 0x0200005f, 0x00020012, 0x02000068, 0x00000001, 0x0400009b, 0x00000040, 0x00000001, 0x00000001, - 0x07000022, 0x00100012, 0x00000000, 0x0002000a, 0x0020802a, 0x00000000, 0x00000001, 0x0304001f, - 0x0010000a, 0x00000000, 0x0700001e, 0x00100012, 0x00000000, 0x0002000a, 0x0020800a, 0x00000000, - 0x00000001, 0x080000a4, 0x0011e0f2, 0x00000000, 0x00100006, 0x00000000, 0x00208e46, 0x00000000, - 0x00000000, 0x01000015, 0x0100003e, -}; - -static const uint32_t cs_uav_clear_1d_uint_code[] = -{ -#if 0 - RWTexture1D dst; - - struct - { - uint4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(64, 1, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (thread_id.x < u_info.dst_extent.x) - dst[u_info.dst_offset.x + thread_id.x] = u_info.clear_value; - } -#endif - 0x43425844, 0x19d5c8f2, 0x3ca4ac24, 0x9e258499, 0xf0463fd6, 0x00000001, 0x0000010c, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000b8, 0x00050050, 0x0000002e, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400109c, 0x0011e000, 0x00000000, 0x00004444, - 0x0200005f, 0x00020012, 0x02000068, 0x00000001, 0x0400009b, 0x00000040, 0x00000001, 0x00000001, - 0x07000022, 0x00100012, 0x00000000, 0x0002000a, 0x0020802a, 0x00000000, 0x00000001, 0x0304001f, - 0x0010000a, 0x00000000, 0x0700001e, 0x00100012, 0x00000000, 0x0002000a, 0x0020800a, 0x00000000, - 0x00000001, 0x080000a4, 0x0011e0f2, 0x00000000, 0x00100006, 0x00000000, 0x00208e46, 0x00000000, - 0x00000000, 0x01000015, 0x0100003e, -}; - -static const uint32_t cs_uav_clear_2d_array_float_code[] = -{ -#if 0 - RWTexture2DArray dst; - - struct - { - float4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(8, 8, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (all(thread_id.xy < u_info.dst_extent.xy)) - dst[int3(u_info.dst_offset.xy + thread_id.xy, thread_id.z)] = u_info.clear_value; - } -#endif - 0x43425844, 0x924d2d2c, 0xb9166376, 0x99f83871, 0x8ef65025, 0x00000001, 0x00000138, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000e4, 0x00050050, 0x00000039, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400409c, 0x0011e000, 0x00000000, 0x00005555, - 0x0200005f, 0x00020072, 0x02000068, 0x00000001, 0x0400009b, 0x00000008, 0x00000008, 0x00000001, - 0x07000022, 0x00100032, 0x00000000, 0x00020046, 0x00208ae6, 0x00000000, 0x00000001, 0x07000001, - 0x00100012, 0x00000000, 0x0010001a, 0x00000000, 0x0010000a, 0x00000000, 0x0304001f, 0x0010000a, - 0x00000000, 0x0700001e, 0x00100032, 0x00000000, 0x00020046, 0x00208046, 0x00000000, 0x00000001, - 0x04000036, 0x001000c2, 0x00000000, 0x00020aa6, 0x080000a4, 0x0011e0f2, 0x00000000, 0x00100e46, - 0x00000000, 0x00208e46, 0x00000000, 0x00000000, 0x01000015, 0x0100003e, -}; - -static const uint32_t cs_uav_clear_2d_array_uint_code[] = -{ -#if 0 - RWTexture2DArray dst; - - struct - { - uint4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(8, 8, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (all(thread_id.xy < u_info.dst_extent.xy)) - dst[int3(u_info.dst_offset.xy + thread_id.xy, thread_id.z)] = u_info.clear_value; - } -#endif - 0x43425844, 0xa92219d4, 0xa2c5e47d, 0x0d308500, 0xf32197b4, 0x00000001, 0x00000138, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000e4, 0x00050050, 0x00000039, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400409c, 0x0011e000, 0x00000000, 0x00004444, - 0x0200005f, 0x00020072, 0x02000068, 0x00000001, 0x0400009b, 0x00000008, 0x00000008, 0x00000001, - 0x07000022, 0x00100032, 0x00000000, 0x00020046, 0x00208ae6, 0x00000000, 0x00000001, 0x07000001, - 0x00100012, 0x00000000, 0x0010001a, 0x00000000, 0x0010000a, 0x00000000, 0x0304001f, 0x0010000a, - 0x00000000, 0x0700001e, 0x00100032, 0x00000000, 0x00020046, 0x00208046, 0x00000000, 0x00000001, - 0x04000036, 0x001000c2, 0x00000000, 0x00020aa6, 0x080000a4, 0x0011e0f2, 0x00000000, 0x00100e46, - 0x00000000, 0x00208e46, 0x00000000, 0x00000000, 0x01000015, 0x0100003e, -}; - -static const uint32_t cs_uav_clear_2d_float_code[] = -{ -#if 0 - RWTexture2D dst; - - struct - { - float4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(8, 8, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (all(thread_id.xy < u_info.dst_extent.xy)) - dst[u_info.dst_offset.xy + thread_id.xy] = u_info.clear_value; - } -#endif - 0x43425844, 0x6e735b3f, 0x7348c4fa, 0xb3634e42, 0x50e2d99b, 0x00000001, 0x00000128, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000d4, 0x00050050, 0x00000035, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400189c, 0x0011e000, 0x00000000, 0x00005555, - 0x0200005f, 0x00020032, 0x02000068, 0x00000001, 0x0400009b, 0x00000008, 0x00000008, 0x00000001, - 0x07000022, 0x00100032, 0x00000000, 0x00020046, 0x00208ae6, 0x00000000, 0x00000001, 0x07000001, - 0x00100012, 0x00000000, 0x0010001a, 0x00000000, 0x0010000a, 0x00000000, 0x0304001f, 0x0010000a, - 0x00000000, 0x0700001e, 0x001000f2, 0x00000000, 0x00020546, 0x00208546, 0x00000000, 0x00000001, - 0x080000a4, 0x0011e0f2, 0x00000000, 0x00100e46, 0x00000000, 0x00208e46, 0x00000000, 0x00000000, - 0x01000015, 0x0100003e, -}; - -static const uint32_t cs_uav_clear_2d_uint_code[] = -{ -#if 0 - RWTexture2D dst; - - struct - { - uint4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(8, 8, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (all(thread_id.xy < u_info.dst_extent.xy)) - dst[u_info.dst_offset.xy + thread_id.xy] = u_info.clear_value; - } -#endif - 0x43425844, 0xf01db5dd, 0xc7dc5e55, 0xb017c1a8, 0x55abd52d, 0x00000001, 0x00000128, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000d4, 0x00050050, 0x00000035, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400189c, 0x0011e000, 0x00000000, 0x00004444, - 0x0200005f, 0x00020032, 0x02000068, 0x00000001, 0x0400009b, 0x00000008, 0x00000008, 0x00000001, - 0x07000022, 0x00100032, 0x00000000, 0x00020046, 0x00208ae6, 0x00000000, 0x00000001, 0x07000001, - 0x00100012, 0x00000000, 0x0010001a, 0x00000000, 0x0010000a, 0x00000000, 0x0304001f, 0x0010000a, - 0x00000000, 0x0700001e, 0x001000f2, 0x00000000, 0x00020546, 0x00208546, 0x00000000, 0x00000001, - 0x080000a4, 0x0011e0f2, 0x00000000, 0x00100e46, 0x00000000, 0x00208e46, 0x00000000, 0x00000000, - 0x01000015, 0x0100003e, -}; - -static const uint32_t cs_uav_clear_3d_float_code[] = -{ -#if 0 - RWTexture3D dst; - - struct - { - float4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(8, 8, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (all(thread_id.xy < u_info.dst_extent.xy)) - dst[int3(u_info.dst_offset.xy, 0) + thread_id.xyz] = u_info.clear_value; - } -#endif - 0x43425844, 0x5d8f36a0, 0x30fa86a5, 0xfec7f2ef, 0xdfd76cbb, 0x00000001, 0x00000138, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000e4, 0x00050050, 0x00000039, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400289c, 0x0011e000, 0x00000000, 0x00005555, - 0x0200005f, 0x00020072, 0x02000068, 0x00000001, 0x0400009b, 0x00000008, 0x00000008, 0x00000001, - 0x07000022, 0x00100032, 0x00000000, 0x00020046, 0x00208ae6, 0x00000000, 0x00000001, 0x07000001, - 0x00100012, 0x00000000, 0x0010001a, 0x00000000, 0x0010000a, 0x00000000, 0x0304001f, 0x0010000a, - 0x00000000, 0x0700001e, 0x00100032, 0x00000000, 0x00020046, 0x00208046, 0x00000000, 0x00000001, - 0x04000036, 0x001000c2, 0x00000000, 0x00020aa6, 0x080000a4, 0x0011e0f2, 0x00000000, 0x00100e46, - 0x00000000, 0x00208e46, 0x00000000, 0x00000000, 0x01000015, 0x0100003e, -}; - -static const uint32_t cs_uav_clear_3d_uint_code[] = -{ -#if 0 - RWTexture3D dst; - - struct - { - uint4 clear_value; - int2 dst_offset; - int2 dst_extent; - } u_info; - - [numthreads(8, 8, 1)] - void main(int3 thread_id : SV_DispatchThreadID) - { - if (all(thread_id.xy < u_info.dst_extent.xy)) - dst[int3(u_info.dst_offset.xy, 0) + thread_id.xyz] = u_info.clear_value; - } -#endif - 0x43425844, 0x5b9c95b1, 0xc9bde4e3, 0x9aaff806, 0x24a1d264, 0x00000001, 0x00000138, 0x00000003, - 0x0000002c, 0x0000003c, 0x0000004c, 0x4e475349, 0x00000008, 0x00000000, 0x00000008, 0x4e47534f, - 0x00000008, 0x00000000, 0x00000008, 0x58454853, 0x000000e4, 0x00050050, 0x00000039, 0x0100086a, - 0x04000059, 0x00208e46, 0x00000000, 0x00000002, 0x0400289c, 0x0011e000, 0x00000000, 0x00004444, - 0x0200005f, 0x00020072, 0x02000068, 0x00000001, 0x0400009b, 0x00000008, 0x00000008, 0x00000001, - 0x07000022, 0x00100032, 0x00000000, 0x00020046, 0x00208ae6, 0x00000000, 0x00000001, 0x07000001, - 0x00100012, 0x00000000, 0x0010001a, 0x00000000, 0x0010000a, 0x00000000, 0x0304001f, 0x0010000a, - 0x00000000, 0x0700001e, 0x00100032, 0x00000000, 0x00020046, 0x00208046, 0x00000000, 0x00000001, - 0x04000036, 0x001000c2, 0x00000000, 0x00020aa6, 0x080000a4, 0x0011e0f2, 0x00000000, 0x00100e46, - 0x00000000, 0x00208e46, 0x00000000, 0x00000000, 0x01000015, 0x0100003e, -}; +static const char cs_uav_clear_buffer_float_code[] = + "RWBuffer dst;\n" + "\n" + "struct\n" + "{\n" + " float4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(128, 1, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (thread_id.x < u_info.dst_extent.x)\n" + " dst[u_info.dst_offset.x + thread_id.x] = u_info.clear_value;\n" + "}\n"; + +static const char cs_uav_clear_buffer_uint_code[] = + "RWBuffer dst;\n" + "\n" + "struct\n" + "{\n" + " uint4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(128, 1, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (thread_id.x < u_info.dst_extent.x)\n" + " dst[u_info.dst_offset.x + thread_id.x] = u_info.clear_value;\n" + "}\n"; + +static const char cs_uav_clear_1d_array_float_code[] = + "RWTexture1DArray dst;\n" + "\n" + "struct\n" + "{\n" + " float4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(64, 1, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (thread_id.x < u_info.dst_extent.x)\n" + " dst[int2(u_info.dst_offset.x + thread_id.x, thread_id.y)] = u_info.clear_value;\n" + "}\n"; + +static const char cs_uav_clear_1d_array_uint_code[] = + "RWTexture1DArray dst;\n" + "\n" + "struct\n" + "{\n" + " uint4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(64, 1, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (thread_id.x < u_info.dst_extent.x)\n" + " dst[int2(u_info.dst_offset.x + thread_id.x, thread_id.y)] = u_info.clear_value;\n" + "}\n"; + +static const char cs_uav_clear_1d_float_code[] = + "RWTexture1D dst;\n" + "\n" + "struct\n" + "{\n" + " float4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(64, 1, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (thread_id.x < u_info.dst_extent.x)\n" + " dst[u_info.dst_offset.x + thread_id.x] = u_info.clear_value;\n" + "}\n"; + +static const char cs_uav_clear_1d_uint_code[] = + "RWTexture1D dst;\n" + "\n" + "struct\n" + "{\n" + " uint4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(64, 1, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (thread_id.x < u_info.dst_extent.x)\n" + " dst[u_info.dst_offset.x + thread_id.x] = u_info.clear_value;\n" + "}\n"; + +static const char cs_uav_clear_2d_array_float_code[] = + "RWTexture2DArray dst;\n" + "\n" + "struct\n" + "{\n" + " float4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(8, 8, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (all(thread_id.xy < u_info.dst_extent.xy))\n" + " dst[int3(u_info.dst_offset.xy + thread_id.xy, thread_id.z)] = u_info.clear_value;\n" + "}\n"; + +static const char cs_uav_clear_2d_array_uint_code[] = + "RWTexture2DArray dst;\n" + "\n" + "struct\n" + "{\n" + " uint4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(8, 8, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (all(thread_id.xy < u_info.dst_extent.xy))\n" + " dst[int3(u_info.dst_offset.xy + thread_id.xy, thread_id.z)] = u_info.clear_value;\n" + "}\n"; + +static const char cs_uav_clear_2d_float_code[] = + "RWTexture2D dst;\n" + "\n" + "struct\n" + "{\n" + " float4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(8, 8, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (all(thread_id.xy < u_info.dst_extent.xy))\n" + " dst[u_info.dst_offset.xy + thread_id.xy] = u_info.clear_value;\n" + "}\n"; + +static const char cs_uav_clear_2d_uint_code[] = + "RWTexture2D dst;\n" + "\n" + "struct\n" + "{\n" + " uint4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(8, 8, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (all(thread_id.xy < u_info.dst_extent.xy))\n" + " dst[u_info.dst_offset.xy + thread_id.xy] = u_info.clear_value;\n" + "}\n"; + +static const char cs_uav_clear_3d_float_code[] = + "RWTexture3D dst;\n" + "\n" + "struct\n" + "{\n" + " float4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(8, 8, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (all(thread_id.xy < u_info.dst_extent.xy))\n" + " dst[int3(u_info.dst_offset.xy, 0) + thread_id.xyz] = u_info.clear_value;\n" + "}\n"; + +static const char cs_uav_clear_3d_uint_code[] = + "RWTexture3D dst;\n" + "\n" + "struct\n" + "{\n" + " uint4 clear_value;\n" + " int2 dst_offset;\n" + " int2 dst_extent;\n" + "} u_info;\n" + "\n" + "[numthreads(8, 8, 1)]\n" + "void main(int3 thread_id : SV_DispatchThreadID)\n" + "{\n" + " if (all(thread_id.xy < u_info.dst_extent.xy))\n" + " dst[int3(u_info.dst_offset.xy, 0) + thread_id.xyz] = u_info.clear_value;\n" + "}\n"; #endif /* __VKD3D_SHADERS_H */ -- 2.43.0