Merge tag 'drm-xe-next-2025-01-10' of https://gitlab.freedesktop.org/drm/xe/kernel into drm-next

Driver Changes:
- SRIOV VF: Avoid reading inaccessible registers (Jakub, Marcin)
 - Introduce RPa frequency information (Rodrigo)
 - Remove unnecessary force wakes on SLPC code (Vinay)
 - Fix all typos in xe (Nitin)
 - Adding steering info support for GuC register lists (Jesus)
 - Remove unused xe_pciids.h harder, add missing PCI ID (Jani)

Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/Z4E0tlTAA6MZ7PF2@intel.com

drivers/gpu/drm/xe/Kconfig.debug

@@ -66,7 +66,7 @@ config DRM_XE_DEBUG_MEM
 	bool "Enable passing SYS/VRAM addresses to user space"
 	default n
 	help
-	  Pass object location trough uapi. Intended for extended
+	  Pass object location through uapi. Intended for extended
 	  testing and development only.
 
 	  Recommended for driver developers only.
@@ -104,5 +104,5 @@ config DRM_XE_USERPTR_INVAL_INJECT
          Choose this option when debugging error paths that
 	 are hit during checks for userptr invalidations.
 
-	 Recomended for driver developers only.
+	 Recommended for driver developers only.
 	 If in doubt, say "N".

drivers/gpu/drm/xe/abi/guc_capture_abi.h

@@ -25,7 +25,7 @@ enum guc_state_capture_type {
 
 #define GUC_STATE_CAPTURE_TYPE_MAX	(GUC_STATE_CAPTURE_TYPE_ENGINE_INSTANCE + 1)
 
-/* Class indecies for capture_class and capture_instance arrays */
+/* Class indices for capture_class and capture_instance arrays */
 enum guc_capture_list_class_type {
 	GUC_CAPTURE_LIST_CLASS_RENDER_COMPUTE = 0,
 	GUC_CAPTURE_LIST_CLASS_VIDEO = 1,

drivers/gpu/drm/xe/abi/guc_klvs_abi.h

@@ -132,7 +132,7 @@ enum  {
  * _`GUC_KLV_VGT_POLICY_SCHED_IF_IDLE` : 0x8001
  *      This config sets whether strict scheduling is enabled whereby any VF
  *      that doesn’t have work to submit is still allocated a fixed execution
- *      time-slice to ensure active VFs execution is always consitent even
+ *      time-slice to ensure active VFs execution is always consistent even
  *      during other VF reprovisiong / rebooting events. Changing this KLV
  *      impacts all VFs and takes effect on the next VF-Switch event.
  *
@@ -207,7 +207,7 @@ enum  {
  *      of and this will never be perfectly-exact (accumulated nano-second
  *      granularity) since the GPUs clock time runs off a different crystal
  *      from the CPUs clock. Changing this KLV on a VF that is currently
- *      running a context wont take effect until a new context is scheduled in.
+ *      running a context won't take effect until a new context is scheduled in.
  *      That said, when the PF is changing this value from 0x0 to
  *      a non-zero value, it might never take effect if the VF is running an
  *      infinitely long compute or shader kernel. In such a scenario, the
@@ -227,7 +227,7 @@ enum  {
  *      HW is capable and this will never be perfectly-exact (accumulated
  *      nano-second granularity) since the GPUs clock time runs off a
  *      different crystal from the CPUs clock. Changing this KLV on a VF
- *      that is currently running a context wont take effect until a new
+ *      that is currently running a context won't take effect until a new
  *      context is scheduled in.
  *      That said, when the PF is changing this value from 0x0 to
  *      a non-zero value, it might never take effect if the VF is running an

drivers/gpu/drm/xe/regs/xe_reg_defs.h

@@ -13,7 +13,7 @@
 /**
  * struct xe_reg - Register definition
  *
- * Register defintion to be used by the individual register. Although the same
+ * Register definition to be used by the individual register. Although the same
  * definition is used for xe_reg and xe_reg_mcr, they use different internal
  * APIs for accesses.
  */

drivers/gpu/drm/xe/regs/xe_regs.h

@@ -44,12 +44,16 @@
 
 #define MTL_RP_STATE_CAP			XE_REG(0x138000)
 
+#define MTL_GT_RPA_FREQUENCY			XE_REG(0x138008)
 #define MTL_GT_RPE_FREQUENCY			XE_REG(0x13800c)
 
 #define MTL_MEDIAP_STATE_CAP			XE_REG(0x138020)
 #define   MTL_RPN_CAP_MASK			REG_GENMASK(24, 16)
 #define   MTL_RP0_CAP_MASK			REG_GENMASK(8, 0)
 
+#define MTL_MPA_FREQUENCY			XE_REG(0x138028)
+#define   MTL_RPA_MASK				REG_GENMASK(8, 0)
+
 #define MTL_MPE_FREQUENCY			XE_REG(0x13802c)
 #define   MTL_RPE_MASK				REG_GENMASK(8, 0)
 

drivers/gpu/drm/xe/tests/xe_mocs.c

@@ -58,7 +58,7 @@ static void read_l3cc_table(struct xe_gt *gt,
 
 			mocs_dbg(gt, "reg_val=0x%x\n", reg_val);
 		} else {
-			/* Just re-use value read on previous iteration */
+			/* Just reuse value read on previous iteration */
 			reg_val >>= 16;
 		}
 

drivers/gpu/drm/xe/xe_bb.c

@@ -41,7 +41,7 @@ struct xe_bb *xe_bb_new(struct xe_gt *gt, u32 dwords, bool usm)
 	/*
 	 * We need to allocate space for the requested number of dwords,
 	 * one additional MI_BATCH_BUFFER_END dword, and additional buffer
-	 * space to accomodate the platform-specific hardware prefetch
+	 * space to accommodate the platform-specific hardware prefetch
 	 * requirements.
 	 */
 	bb->bo = xe_sa_bo_new(!usm ? tile->mem.kernel_bb_pool : gt->usm.bb_pool,

drivers/gpu/drm/xe/xe_bo.c

@@ -786,7 +786,7 @@ static int xe_bo_move(struct ttm_buffer_object *ttm_bo, bool evict,
 		 * / resume, some of the pinned memory is required for the
 		 * device to resume / use the GPU to move other evicted memory
 		 * (user memory) around. This likely could be optimized a bit
-		 * futher where we find the minimum set of pinned memory
+		 * further where we find the minimum set of pinned memory
 		 * required for resume but for simplity doing a memcpy for all
 		 * pinned memory.
 		 */
@@ -875,7 +875,7 @@ out:
  * xe_bo_evict_pinned() - Evict a pinned VRAM object to system memory
  * @bo: The buffer object to move.
  *
- * On successful completion, the object memory will be moved to sytem memory.
+ * On successful completion, the object memory will be moved to system memory.
  *
  * This is needed to for special handling of pinned VRAM object during
  * suspend-resume.
@@ -1370,7 +1370,7 @@ static const struct drm_gem_object_funcs xe_gem_object_funcs = {
 /**
  * xe_bo_alloc - Allocate storage for a struct xe_bo
  *
- * This funcition is intended to allocate storage to be used for input
+ * This function is intended to allocate storage to be used for input
  * to __xe_bo_create_locked(), in the case a pointer to the bo to be
  * created is needed before the call to __xe_bo_create_locked().
  * If __xe_bo_create_locked ends up never to be called, then the
@@ -2412,7 +2412,7 @@ int xe_bo_migrate(struct xe_bo *bo, u32 mem_type)
  * @force_alloc: Set force_alloc in ttm_operation_ctx
  *
  * On successful completion, the object memory will be moved to evict
- * placement. Ths function blocks until the object has been fully moved.
+ * placement. This function blocks until the object has been fully moved.
  *
  * Return: 0 on success. Negative error code on failure.
  */

drivers/gpu/drm/xe/xe_bo_doc.h

@@ -41,7 +41,7 @@
  * created the BO can be mmap'd (via DRM_IOCTL_XE_GEM_MMAP_OFFSET) for user
  * access and it can be bound for GPU access (via DRM_IOCTL_XE_VM_BIND). All
  * user BOs are evictable and user BOs are never pinned by XE. The allocation of
- * the backing store can be defered from creation time until first use which is
+ * the backing store can be deferred from creation time until first use which is
  * either mmap, bind, or pagefault.
  *
  * Private BOs

drivers/gpu/drm/xe/xe_devcoredump.c

@@ -48,7 +48,7 @@
  *
  * **Coredump release**:
  *   After a coredump is generated, it stays in kernel memory until released by
- *   userpace by writing anything to it, or after an internal timer expires. The
+ *   userspace by writing anything to it, or after an internal timer expires. The
  *   exact timeout may vary and should not be relied upon. Example to release
  *   a coredump:
  *

drivers/gpu/drm/xe/xe_device.c

@@ -606,7 +606,7 @@ static int probe_has_flat_ccs(struct xe_device *xe)
 	u32 reg;
 
 	/* Always enabled/disabled, no runtime check to do */
-	if (GRAPHICS_VER(xe) < 20 || !xe->info.has_flat_ccs)
+	if (GRAPHICS_VER(xe) < 20 || !xe->info.has_flat_ccs || IS_SRIOV_VF(xe))
 		return 0;
 
 	gt = xe_root_mmio_gt(xe);
@@ -999,7 +999,7 @@ static void xe_device_wedged_fini(struct drm_device *drm, void *arg)
  * xe_device_declare_wedged - Declare device wedged
  * @xe: xe device instance
  *
- * This is a final state that can only be cleared with a mudule
+ * This is a final state that can only be cleared with a module
  * re-probe (unbind + bind).
  * In this state every IOCTL will be blocked so the GT cannot be used.
  * In general it will be called upon any critical error such as gt reset

drivers/gpu/drm/xe/xe_drm_client.c

@@ -385,7 +385,7 @@ static void show_run_ticks(struct drm_printer *p, struct drm_file *file)
  * @p: The drm_printer ptr
  * @file: The drm_file ptr
  *
- * This is callabck for drm fdinfo interface. Register this callback
+ * This is callback for drm fdinfo interface. Register this callback
  * in drm driver ops for show_fdinfo.
  *
  * Return: void

drivers/gpu/drm/xe/xe_exec.c

@@ -33,7 +33,7 @@
  *
  * In XE we avoid all of this complication by not allowing a BO list to be
  * passed into an exec, using the dma-buf implicit sync uAPI, have binds as
- * seperate operations, and using the DRM scheduler to flow control the ring.
+ * separate operations, and using the DRM scheduler to flow control the ring.
  * Let's deep dive on each of these.
  *
  * We can get away from a BO list by forcing the user to use in / out fences on

drivers/gpu/drm/xe/xe_ggtt.c

@@ -362,7 +362,7 @@ int xe_ggtt_init(struct xe_ggtt *ggtt)
 
 	/*
 	 * So we don't need to worry about 64K GGTT layout when dealing with
-	 * scratch entires, rather keep the scratch page in system memory on
+	 * scratch entries, rather keep the scratch page in system memory on
 	 * platforms where 64K pages are needed for VRAM.
 	 */
 	flags = XE_BO_FLAG_PINNED;

drivers/gpu/drm/xe/xe_gt.h

@@ -37,7 +37,7 @@ int xe_gt_record_default_lrcs(struct xe_gt *gt);
 
 /**
  * xe_gt_record_user_engines - save data related to engines available to
- * usersapce
+ * userspace
  * @gt: GT structure
  *
  * Walk the available HW engines from gt->info.engine_mask and calculate data

drivers/gpu/drm/xe/xe_gt_freq.c

@@ -115,6 +115,20 @@ static ssize_t rpe_freq_show(struct device *dev,
 }
 static DEVICE_ATTR_RO(rpe_freq);
 
+static ssize_t rpa_freq_show(struct device *dev,
+			     struct device_attribute *attr, char *buf)
+{
+	struct xe_guc_pc *pc = dev_to_pc(dev);
+	u32 freq;
+
+	xe_pm_runtime_get(dev_to_xe(dev));
+	freq = xe_guc_pc_get_rpa_freq(pc);
+	xe_pm_runtime_put(dev_to_xe(dev));
+
+	return sysfs_emit(buf, "%d\n", freq);
+}
+static DEVICE_ATTR_RO(rpa_freq);
+
 static ssize_t rpn_freq_show(struct device *dev,
 			     struct device_attribute *attr, char *buf)
 {
@@ -202,6 +216,7 @@ static const struct attribute *freq_attrs[] = {
 	&dev_attr_act_freq.attr,
 	&dev_attr_cur_freq.attr,
 	&dev_attr_rp0_freq.attr,
+	&dev_attr_rpa_freq.attr,
 	&dev_attr_rpe_freq.attr,
 	&dev_attr_rpn_freq.attr,
 	&dev_attr_min_freq.attr,

drivers/gpu/drm/xe/xe_gt_mcr.c

@@ -371,7 +371,7 @@ void xe_gt_mcr_get_dss_steering(struct xe_gt *gt, unsigned int dss, u16 *group,
  * @group: steering group ID
  * @instance: steering instance ID
  *
- * Return: the coverted DSS id.
+ * Return: the converted DSS id.
  */
 u32 xe_gt_mcr_steering_info_to_dss_id(struct xe_gt *gt, u16 group, u16 instance)
 {
@@ -550,9 +550,9 @@ void xe_gt_mcr_set_implicit_defaults(struct xe_gt *gt)
  * Returns true if the caller should steer to the @group/@instance values
  * returned.  Returns false if the caller need not perform any steering
  */
-static bool xe_gt_mcr_get_nonterminated_steering(struct xe_gt *gt,
-						 struct xe_reg_mcr reg_mcr,
-						 u8 *group, u8 *instance)
+bool xe_gt_mcr_get_nonterminated_steering(struct xe_gt *gt,
+					  struct xe_reg_mcr reg_mcr,
+					  u8 *group, u8 *instance)
 {
 	const struct xe_reg reg = to_xe_reg(reg_mcr);
 	const struct xe_mmio_range *implicit_ranges;

drivers/gpu/drm/xe/xe_gt_mcr.h

@@ -26,6 +26,10 @@ void xe_gt_mcr_unicast_write(struct xe_gt *gt, struct xe_reg_mcr mcr_reg,
 void xe_gt_mcr_multicast_write(struct xe_gt *gt, struct xe_reg_mcr mcr_reg,
 			       u32 value);
 
+bool xe_gt_mcr_get_nonterminated_steering(struct xe_gt *gt,
+					  struct xe_reg_mcr reg_mcr,
+					  u8 *group, u8 *instance);
+
 void xe_gt_mcr_steering_dump(struct xe_gt *gt, struct drm_printer *p);
 void xe_gt_mcr_get_dss_steering(struct xe_gt *gt, unsigned int dss, u16 *group, u16 *instance);
 u32 xe_gt_mcr_steering_info_to_dss_id(struct xe_gt *gt, u16 group, u16 instance);

drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c

@@ -2161,7 +2161,7 @@ bool xe_gt_sriov_pf_config_is_empty(struct xe_gt *gt, unsigned int vfid)
  *
  * This function can only be called on PF.
  *
- * Return: mininum size of the buffer or the number of bytes saved,
+ * Return: minimum size of the buffer or the number of bytes saved,
  *         or a negative error code on failure.
  */
 ssize_t xe_gt_sriov_pf_config_save(struct xe_gt *gt, unsigned int vfid, void *buf, size_t size)

drivers/gpu/drm/xe/xe_guc_ads.c

@@ -29,6 +29,7 @@
 #include "xe_platform_types.h"
 #include "xe_uc_fw.h"
 #include "xe_wa.h"
+#include "xe_gt_mcr.h"
 
 /* Slack of a few additional entries per engine */
 #define ADS_REGSET_EXTRA_MAX	8
@@ -696,6 +697,20 @@ static void guc_mmio_regset_write_one(struct xe_guc_ads *ads,
 		.flags = reg.masked ? GUC_REGSET_MASKED : 0,
 	};
 
+	if (reg.mcr) {
+		struct xe_reg_mcr mcr_reg = XE_REG_MCR(reg.addr);
+		u8 group, instance;
+
+		bool steer = xe_gt_mcr_get_nonterminated_steering(ads_to_gt(ads), mcr_reg,
+								  &group, &instance);
+
+		if (steer) {
+			entry.flags |= FIELD_PREP(GUC_REGSET_STEERING_GROUP, group);
+			entry.flags |= FIELD_PREP(GUC_REGSET_STEERING_INSTANCE, instance);
+			entry.flags |= GUC_REGSET_STEERING_NEEDED;
+		}
+	}
+
 	xe_map_memcpy_to(ads_to_xe(ads), regset_map, n_entry * sizeof(entry),
 			 &entry, sizeof(entry));
 }