Plug buffer head leak on submit for the 7.2 kernel

Starting with the 7.2 kernel, buffer heads no longer lose a reference
during the submit-wait procedure. I didn't notice that change, so now I
would expect all buffer heads to be leaking. The fact that I still can't
tell the difference makes me believe, like I mentioned before, that
somehow the kernel is still disposing of the buffer heads. Either way,
fix this.

Signed-off-by: Ernesto A. Fernández <ernesto@corellium.com>

Makefile

@@ -14,9 +14,9 @@ apfs-y := btree.o compress.o dir.o extents.o file.o inode.o key.o libzbitmap.o \
 	  spaceman.o super.o symlink.o transaction.o unicode.o xattr.o xfield.o
 
 # If you want mounts to be writable by default, run the build as:
-#   make CONFIG=-DCONFIG_APFS_RW_ALWAYS
+#   make APFS_CONFIG=-DCONFIG_APFS_RW_ALWAYS
 # This is risky and not generally recommended.
-ccflags-y += $(CONFIG)
+ccflags-y += $(APFS_CONFIG)
 
 default:
 	./genver.sh

README.rst

@@ -57,10 +57,15 @@ Now you can just cd to the linux-apfs-rw directory and run::
 	make
 
 The resulting module is the apfs.ko file. Before you can use it you must insert
-it into the kernel, as well as its dependencies. Again as root::
+it into the kernel. Again as root::
+
+	insmod apfs.ko
+
+If you get an "unknown symbol in module" error, run::
 
 	modprobe libcrc32c
-	insmod apfs.ko
+
+and then try the insmod command again.
 
 Mount
 =====

apfs.h

@@ -44,10 +44,33 @@ static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags
 #endif
 
 /* Compatibility wrapper around submit_bh() */
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(6, 0, 0) || RHEL_VERSION_GE(9, 2)
-#define apfs_submit_bh(op, op_flags, bh) submit_bh(op | op_flags, bh)
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(7, 2, 0)
+#define apfs_submit_bh(op, op_flags, bh)				\
+do {									\
+	put_bh(bh);							\
+	if (op == REQ_OP_READ)						\
+		bh_submit(bh, op | op_flags, bh_end_read);		\
+	else								\
+		bh_submit(bh, op | op_flags, bh_end_write);		\
+} while (0)
+#elif LINUX_VERSION_CODE >= KERNEL_VERSION(6, 0, 0) || RHEL_VERSION_GE(9, 2)
+#define apfs_submit_bh(op, op_flags, bh)				\
+do {									\
+	if (op == REQ_OP_READ)						\
+		bh->b_end_io = end_buffer_read_sync;			\
+	else								\
+		bh->b_end_io = end_buffer_write_sync;			\
+	submit_bh(op | op_flags, bh);					\
+} while (0)
 #else
-#define apfs_submit_bh(op, op_flags, bh) submit_bh(op, op_flags, bh)
+#define apfs_submit_bh(op, op_flags, bh)				\
+do {									\
+	if (op == REQ_OP_READ)						\
+		bh->b_end_io = end_buffer_read_sync;			\
+	else								\
+		bh->b_end_io = end_buffer_write_sync;			\
+	submit_bh(op, op_flags, bh);					\
+} while (0)
 #endif
 
 /*
@@ -380,6 +403,7 @@ struct apfs_sb_info {
 	unsigned int s_vol_nr;		/* Index of the volume in the sb list */
 	kuid_t s_uid;			/* uid to override on-disk uid */
 	kgid_t s_gid;			/* gid to override on-disk gid */
+	unsigned int s_mount_opt;
 
 	struct apfs_crypto_state_val *s_dflt_pfk; /* default per-file key */
 
@@ -1040,7 +1064,9 @@ extern int apfs_setattr(struct mnt_idmap *idmap,
 			struct dentry *dentry, struct iattr *iattr);
 #endif
 
-#if LINUX_VERSION_CODE < KERNEL_VERSION(6, 6, 0)
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(7, 0, 0)
+extern int apfs_update_time(struct inode *inode, enum fs_update_time time, unsigned int flags);
+#elif LINUX_VERSION_CODE < KERNEL_VERSION(6, 6, 0)
 extern int apfs_update_time(struct inode *inode, struct timespec64 *time, int flags);
 #else
 extern int apfs_update_time(struct inode *inode, int flags);
@@ -1216,6 +1242,33 @@ static inline void apfs_assert_query_is_valid(const struct apfs_query *query)
 #define page_has_buffers(page)	folio_buffers(page_folio(page))
 #endif
 
+static inline int apfs_inode_state_read_once(struct inode *inode)
+{
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(6, 19, 0)
+	return inode_state_read_once(inode);
+#else
+	return inode->i_state;
+#endif
+}
+
+static inline void apfs_inode_state_set_raw(struct inode *inode, int flags)
+{
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(6, 19, 0)
+	inode_state_set_raw(inode, flags);
+#else
+	inode->i_state |= flags;
+#endif
+}
+
+static inline void apfs_inode_state_clear_raw(struct inode *inode, int flags)
+{
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(6, 19, 0)
+	inode_state_clear_raw(inode, flags);
+#else
+	inode->i_state &= ~flags;
+#endif
+}
+
 /*
  * TODO: the following are modified variants of buffer head functions that will
  * work with the shared block device for the container. The correct approach
@@ -1294,7 +1347,7 @@ __apfs_getblk(struct super_block *sb, sector_t block)
 		block -= nxi->nx_tier2_bno;
 	}
 
-#if LINUX_VERSION_CODE < KERNEL_VERSION(6, 7, 0)
+#if LINUX_VERSION_CODE < KERNEL_VERSION(6, 7, 0) && !RHEL_VERSION_GE(9, 8)
 	return __getblk_gfp(info->blki_bdev, block, sb->s_blocksize, __GFP_MOVABLE);
 #else
 	return bdev_getblk(info->blki_bdev, block, sb->s_blocksize, __GFP_MOVABLE);

dir.c

@@ -1255,9 +1255,10 @@ static void __apfs_undo_unlink(struct dentry *dentry)
 {
 	struct inode *inode = d_inode(dentry);
 
-	inode->i_state |= I_LINKABLE; /* Silence warning about nlink 0->1 */
+	/* Silence warning about nlink 0->1 */
+	apfs_inode_state_set_raw(inode, I_LINKABLE);
 	inc_nlink(inode);
-	inode->i_state &= ~I_LINKABLE;
+	apfs_inode_state_clear_raw(inode, I_LINKABLE);
 
 	apfs_undo_delete_dentry(dentry);
 }

dkms.conf

@@ -1,5 +1,5 @@
 PACKAGE_NAME="linux-apfs-rw"
-PACKAGE_VERSION="0.3.17"
+PACKAGE_VERSION="0.3.21"
 
 BUILT_MODULE_NAME[0]="apfs"
 DEST_MODULE_LOCATION[0]="/extra"

extents.c

@@ -2298,7 +2298,6 @@ int apfs_nonsparse_dstream_read(struct apfs_dstream_info *dstream, void *buf, si
 		if (!buffer_uptodate(bh)) {
 			get_bh(bh);
 			lock_buffer(bh);
-			bh->b_end_io = end_buffer_read_sync;
 			apfs_submit_bh(REQ_OP_READ, 0, bh);
 		}
 	}
@@ -2365,7 +2364,6 @@ void apfs_nonsparse_dstream_preread(struct apfs_dstream_info *dstream)
 		if (!buffer_uptodate(bh)) {
 			get_bh(bh);
 			lock_buffer(bh);
-			bh->b_end_io = end_buffer_read_sync;
 			apfs_submit_bh(REQ_OP_READ, 0, bh);
 		}
 		brelse(bh);

inode.c

@@ -556,7 +556,6 @@ int __apfs_write_begin(struct file *file, struct address_space *mapping, loff_t
 			if (buffer_mapped(bh) && !buffer_uptodate(bh)) {
 				get_bh(bh);
 				lock_buffer(bh);
-				bh->b_end_io = end_buffer_read_sync;
 				apfs_submit_bh(REQ_OP_READ, 0, bh);
 				wait_on_buffer(bh);
 				if (!buffer_uptodate(bh)) {
@@ -1037,7 +1036,7 @@ struct inode *apfs_iget(struct super_block *sb, u64 cnid)
 	inode = apfs_iget_locked(sb, cnid);
 	if (!inode)
 		return ERR_PTR(-ENOMEM);
-	if (!(inode->i_state & I_NEW))
+	if (!(apfs_inode_state_read_once(inode) & I_NEW))
 		return inode;
 
 	down_read(&nxi->nx_big_sem);
@@ -2109,7 +2108,9 @@ fail:
 }
 
 /* TODO: this only seems to be necessary because ->write_inode() isn't firing */
-#if LINUX_VERSION_CODE < KERNEL_VERSION(6, 6, 0)
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(7, 0, 0)
+int apfs_update_time(struct inode *inode, enum fs_update_time time, unsigned int flags)
+#elif LINUX_VERSION_CODE < KERNEL_VERSION(6, 6, 0)
 int apfs_update_time(struct inode *inode, struct timespec64 *time, int flags)
 #else
 int apfs_update_time(struct inode *inode, int flags)
@@ -2123,7 +2124,9 @@ int apfs_update_time(struct inode *inode, int flags)
 		return err;
 	apfs_inode_join_transaction(sb, inode);
 
-#if LINUX_VERSION_CODE < KERNEL_VERSION(6, 6, 0) && !RHEL_VERSION_GE(9, 6)
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(7, 0, 0)
+	generic_update_time(inode, time, flags);
+#elif LINUX_VERSION_CODE < KERNEL_VERSION(6, 6, 0) && !RHEL_VERSION_GE(9, 6)
 	generic_update_time(inode, time, flags);
 #else
 	generic_update_time(inode, flags);

lzfse/lzfse_internal.h

@@ -70,7 +70,9 @@ typedef int64_t lzfse_offset;
 typedef int32_t lzfse_offset;
 #endif
 
+#if !defined(__APPLE__) || !defined(__MACH__)
 typedef uint64_t uintmax_t;
+#endif
 
 /*! @abstract Decoder state object for lzfse compressed blocks. */
 typedef struct {

spaceman.c

@@ -308,6 +308,7 @@ static int apfs_write_single_ip_bitmap(struct super_block *sb, char *bitmap, u32
 	err = apfs_transaction_join(sb, bh);
 	if (err)
 		goto fail;
+	put_bh(bh);
 	bh = NULL;
 
 	spaceman->sm_ip_bmaps[idx].dirty = false;

transaction.c

@@ -468,7 +468,7 @@ int apfs_transaction_flush_all_inodes(struct super_block *sb)
 		curr_err = apfs_update_inode(inode, NULL /* new_name */);
 		if (curr_err)
 			err = curr_err;
-		inode->i_state &= ~I_DIRTY_ALL;
+		apfs_inode_state_clear_raw(inode, I_DIRTY_ALL);
 
 		/*
 		 * The same inode may get dirtied again as soon as we release
@@ -640,6 +640,9 @@ static int apfs_write_ephemeral_objects(struct super_block *sb)
 		}
 		data_len += obj_blkcnt;
 	}
+	brelse(cpm_bh);
+	cpm = NULL;
+	cpm_bh = NULL;
 
 	/*
 	 * The checkpoint superblock can't be set until the very end of the
@@ -721,7 +724,6 @@ static int apfs_transaction_commit_nx(struct super_block *sb)
 		clear_buffer_dirty(bh);
 		get_bh(bh);
 		lock_buffer(bh);
-		bh->b_end_io = end_buffer_write_sync;
 		apfs_submit_bh(REQ_OP_WRITE, REQ_SYNC, bh);
 	}
 	list_for_each_entry_safe(bhi, tmp, &nx_trans->t_buffers, list) {