Merge tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4

Pull ext4 updates from Ted Ts'o:
 "Performance improvements in SEEK_DATA and xattr scalability
  improvements, plus a lot of clean ups and bug fixes"

* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4: (38 commits)
  ext4: clean up error handling in the MMP support
  jbd2: do not fail journal because of frozen_buffer allocation failure
  ext4: use __GFP_NOFAIL in ext4_free_blocks()
  ext4: fix compile error while opening the macro DOUBLE_CHECK
  ext4: print ext4 mount option data_err=abort correctly
  ext4: fix NULL pointer dereference in ext4_mark_inode_dirty()
  ext4: drop unneeded BUFFER_TRACE in ext4_delete_inline_entry()
  ext4: fix misspellings in comments.
  jbd2: fix FS corruption possibility in jbd2_journal_destroy() on umount path
  ext4: more efficient SEEK_DATA implementation
  ext4: cleanup handling of bh->b_state in DAX mmap
  ext4: return hole from ext4_map_blocks()
  ext4: factor out determining of hole size
  ext4: fix setting of referenced bit in ext4_es_lookup_extent()
  ext4: remove i_ioend_count
  ext4: simplify io_end handling for AIO DIO
  ext4: move trans handling and completion deferal out of _ext4_get_block
  ext4: rename and split get blocks functions
  ext4: use i_mutex to serialize unaligned AIO DIO
  ext4: pack ioend structure better
  ...

4 files changed, 90 insertions, 98 deletions

diff --git a/fs/ext2/ext2.h b/fs/ext2/ext2.h
index 4c69c94..170939f 100644
--- a/fs/ext2/ext2.h
+++ b/fs/ext2/ext2.h
@@ -61,6 +61,8 @@ struct ext2_block_alloc_info {
 #define rsv_start rsv_window._rsv_start
 #define rsv_end rsv_window._rsv_end
 
+struct mb_cache;
+
 /*
  * second extended-fs super-block data in memory
  */
@@ -111,6 +113,7 @@ struct ext2_sb_info {
 	 * of the mount options.
 	 */
 	spinlock_t s_lock;
+	struct mb_cache *s_mb_cache;
 };
 
 static inline spinlock_t *
diff --git a/fs/ext2/super.c b/fs/ext2/super.c
index 2a18841..b78caf2 100644
--- a/fs/ext2/super.c
+++ b/fs/ext2/super.c
@@ -131,7 +131,10 @@ static void ext2_put_super (struct super_block * sb)
 
 	dquot_disable(sb, -1, DQUOT_USAGE_ENABLED | DQUOT_LIMITS_ENABLED);
 
-	ext2_xattr_put_super(sb);
+	if (sbi->s_mb_cache) {
+		ext2_xattr_destroy_cache(sbi->s_mb_cache);
+		sbi->s_mb_cache = NULL;
+	}
 	if (!(sb->s_flags & MS_RDONLY)) {
 		struct ext2_super_block *es = sbi->s_es;
 
@@ -1104,6 +1107,14 @@ static int ext2_fill_super(struct super_block *sb, void *data, int silent)
 		ext2_msg(sb, KERN_ERR, "error: insufficient memory");
 		goto failed_mount3;
 	}
+
+#ifdef CONFIG_EXT2_FS_XATTR
+	sbi->s_mb_cache = ext2_xattr_create_cache();
+	if (!sbi->s_mb_cache) {
+		ext2_msg(sb, KERN_ERR, "Failed to create an mb_cache");
+		goto failed_mount3;
+	}
+#endif
 	/*
 	 * set up enough so that it can read an inode
 	 */
@@ -1149,6 +1160,8 @@ cantfind_ext2:
 			sb->s_id);
 	goto failed_mount;
 failed_mount3:
+	if (sbi->s_mb_cache)
+		ext2_xattr_destroy_cache(sbi->s_mb_cache);
 	percpu_counter_destroy(&sbi->s_freeblocks_counter);
 	percpu_counter_destroy(&sbi->s_freeinodes_counter);
 	percpu_counter_destroy(&sbi->s_dirs_counter);
@@ -1555,20 +1568,17 @@ MODULE_ALIAS_FS("ext2");
 
 static int __init init_ext2_fs(void)
 {
-	int err = init_ext2_xattr();
-	if (err)
-		return err;
+	int err;
+
 	err = init_inodecache();
 	if (err)
-		goto out1;
+		return err;
         err = register_filesystem(&ext2_fs_type);
 	if (err)
 		goto out;
 	return 0;
 out:
 	destroy_inodecache();
-out1:
-	exit_ext2_xattr();
 	return err;
 }
 
@@ -1576,7 +1586,6 @@ static void __exit exit_ext2_fs(void)
 {
 	unregister_filesystem(&ext2_fs_type);
 	destroy_inodecache();
-	exit_ext2_xattr();
 }
 
 MODULE_AUTHOR("Remy Card and others");
diff --git a/fs/ext2/xattr.c b/fs/ext2/xattr.c
index f57a7ab..1a5e3bf 100644
--- a/fs/ext2/xattr.c
+++ b/fs/ext2/xattr.c
@@ -90,14 +90,12 @@
 static int ext2_xattr_set2(struct inode *, struct buffer_head *,
 			   struct ext2_xattr_header *);
 
-static int ext2_xattr_cache_insert(struct buffer_head *);
+static int ext2_xattr_cache_insert(struct mb_cache *, struct buffer_head *);
 static struct buffer_head *ext2_xattr_cache_find(struct inode *,
 						 struct ext2_xattr_header *);
 static void ext2_xattr_rehash(struct ext2_xattr_header *,
 			      struct ext2_xattr_entry *);
 
-static struct mb_cache *ext2_xattr_cache;
-
 static const struct xattr_handler *ext2_xattr_handler_map[] = {
 	[EXT2_XATTR_INDEX_USER]		     = &ext2_xattr_user_handler,
 #ifdef CONFIG_EXT2_FS_POSIX_ACL
@@ -152,6 +150,7 @@ ext2_xattr_get(struct inode *inode, int name_index, const char *name,
 	size_t name_len, size;
 	char *end;
 	int error;
+	struct mb_cache *ext2_mb_cache = EXT2_SB(inode->i_sb)->s_mb_cache;
 
 	ea_idebug(inode, "name=%d.%s, buffer=%p, buffer_size=%ld",
 		  name_index, name, buffer, (long)buffer_size);
@@ -196,7 +195,7 @@ bad_block:	ext2_error(inode->i_sb, "ext2_xattr_get",
 			goto found;
 		entry = next;
 	}
-	if (ext2_xattr_cache_insert(bh))
+	if (ext2_xattr_cache_insert(ext2_mb_cache, bh))
 		ea_idebug(inode, "cache insert failed");
 	error = -ENODATA;
 	goto cleanup;
@@ -209,7 +208,7 @@ found:
 	    le16_to_cpu(entry->e_value_offs) + size > inode->i_sb->s_blocksize)
 		goto bad_block;
 
-	if (ext2_xattr_cache_insert(bh))
+	if (ext2_xattr_cache_insert(ext2_mb_cache, bh))
 		ea_idebug(inode, "cache insert failed");
 	if (buffer) {
 		error = -ERANGE;
@@ -247,6 +246,7 @@ ext2_xattr_list(struct dentry *dentry, char *buffer, size_t buffer_size)
 	char *end;
 	size_t rest = buffer_size;
 	int error;
+	struct mb_cache *ext2_mb_cache = EXT2_SB(inode->i_sb)->s_mb_cache;
 
 	ea_idebug(inode, "buffer=%p, buffer_size=%ld",
 		  buffer, (long)buffer_size);
@@ -281,7 +281,7 @@ bad_block:	ext2_error(inode->i_sb, "ext2_xattr_list",
 			goto bad_block;
 		entry = next;
 	}
-	if (ext2_xattr_cache_insert(bh))
+	if (ext2_xattr_cache_insert(ext2_mb_cache, bh))
 		ea_idebug(inode, "cache insert failed");
 
 	/* list the attribute names */
@@ -483,22 +483,23 @@ bad_block:		ext2_error(sb, "ext2_xattr_set",
 	/* Here we know that we can set the new attribute. */
 
 	if (header) {
-		struct mb_cache_entry *ce;
-
 		/* assert(header == HDR(bh)); */
-		ce = mb_cache_entry_get(ext2_xattr_cache, bh->b_bdev,
-					bh->b_blocknr);
 		lock_buffer(bh);
 		if (header->h_refcount == cpu_to_le32(1)) {
+			__u32 hash = le32_to_cpu(header->h_hash);
+
 			ea_bdebug(bh, "modifying in-place");
-			if (ce)
-				mb_cache_entry_free(ce);
+			/*
+			 * This must happen under buffer lock for
+			 * ext2_xattr_set2() to reliably detect modified block
+			 */
+			mb_cache_entry_delete_block(EXT2_SB(sb)->s_mb_cache,
+						    hash, bh->b_blocknr);
+
 			/* keep the buffer locked while modifying it. */
 		} else {
 			int offset;
 
-			if (ce)
-				mb_cache_entry_release(ce);
 			unlock_buffer(bh);
 			ea_bdebug(bh, "cloning");
 			header = kmalloc(bh->b_size, GFP_KERNEL);
@@ -626,6 +627,7 @@ ext2_xattr_set2(struct inode *inode, struct buffer_head *old_bh,
 	struct super_block *sb = inode->i_sb;
 	struct buffer_head *new_bh = NULL;
 	int error;
+	struct mb_cache *ext2_mb_cache = EXT2_SB(sb)->s_mb_cache;
 
 	if (header) {
 		new_bh = ext2_xattr_cache_find(inode, header);
@@ -653,7 +655,7 @@ ext2_xattr_set2(struct inode *inode, struct buffer_head *old_bh,
 			   don't need to change the reference count. */
 			new_bh = old_bh;
 			get_bh(new_bh);
-			ext2_xattr_cache_insert(new_bh);
+			ext2_xattr_cache_insert(ext2_mb_cache, new_bh);
 		} else {
 			/* We need to allocate a new block */
 			ext2_fsblk_t goal = ext2_group_first_block_no(sb,
@@ -674,7 +676,7 @@ ext2_xattr_set2(struct inode *inode, struct buffer_head *old_bh,
 			memcpy(new_bh->b_data, header, new_bh->b_size);
 			set_buffer_uptodate(new_bh);
 			unlock_buffer(new_bh);
-			ext2_xattr_cache_insert(new_bh);
+			ext2_xattr_cache_insert(ext2_mb_cache, new_bh);
 			
 			ext2_xattr_update_super_block(sb);
 		}
@@ -707,19 +709,21 @@ ext2_xattr_set2(struct inode *inode, struct buffer_head *old_bh,
 
 	error = 0;
 	if (old_bh && old_bh != new_bh) {
-		struct mb_cache_entry *ce;
-
 		/*
 		 * If there was an old block and we are no longer using it,
 		 * release the old block.
 		 */
-		ce = mb_cache_entry_get(ext2_xattr_cache, old_bh->b_bdev,
-					old_bh->b_blocknr);
 		lock_buffer(old_bh);
 		if (HDR(old_bh)->h_refcount == cpu_to_le32(1)) {
+			__u32 hash = le32_to_cpu(HDR(old_bh)->h_hash);
+
+			/*
+			 * This must happen under buffer lock for
+			 * ext2_xattr_set2() to reliably detect freed block
+			 */
+			mb_cache_entry_delete_block(ext2_mb_cache,
+						    hash, old_bh->b_blocknr);
 			/* Free the old block. */
-			if (ce)
-				mb_cache_entry_free(ce);
 			ea_bdebug(old_bh, "freeing");
 			ext2_free_blocks(inode, old_bh->b_blocknr, 1);
 			mark_inode_dirty(inode);
@@ -730,8 +734,6 @@ ext2_xattr_set2(struct inode *inode, struct buffer_head *old_bh,
 		} else {
 			/* Decrement the refcount only. */
 			le32_add_cpu(&HDR(old_bh)->h_refcount, -1);
-			if (ce)
-				mb_cache_entry_release(ce);
 			dquot_free_block_nodirty(inode, 1);
 			mark_inode_dirty(inode);
 			mark_buffer_dirty(old_bh);
@@ -757,7 +759,6 @@ void
 ext2_xattr_delete_inode(struct inode *inode)
 {
 	struct buffer_head *bh = NULL;
-	struct mb_cache_entry *ce;
 
 	down_write(&EXT2_I(inode)->xattr_sem);
 	if (!EXT2_I(inode)->i_file_acl)
@@ -777,19 +778,22 @@ ext2_xattr_delete_inode(struct inode *inode)
 			EXT2_I(inode)->i_file_acl);
 		goto cleanup;
 	}
-	ce = mb_cache_entry_get(ext2_xattr_cache, bh->b_bdev, bh->b_blocknr);
 	lock_buffer(bh);
 	if (HDR(bh)->h_refcount == cpu_to_le32(1)) {
-		if (ce)
-			mb_cache_entry_free(ce);
+		__u32 hash = le32_to_cpu(HDR(bh)->h_hash);
+
+		/*
+		 * This must happen under buffer lock for ext2_xattr_set2() to
+		 * reliably detect freed block
+		 */
+		mb_cache_entry_delete_block(EXT2_SB(inode->i_sb)->s_mb_cache,
+					    hash, bh->b_blocknr);
 		ext2_free_blocks(inode, EXT2_I(inode)->i_file_acl, 1);
 		get_bh(bh);
 		bforget(bh);
 		unlock_buffer(bh);
 	} else {
 		le32_add_cpu(&HDR(bh)->h_refcount, -1);
-		if (ce)
-			mb_cache_entry_release(ce);
 		ea_bdebug(bh, "refcount now=%d",
 			le32_to_cpu(HDR(bh)->h_refcount));
 		unlock_buffer(bh);
@@ -806,18 +810,6 @@ cleanup:
 }
 
 /*
- * ext2_xattr_put_super()
- *
- * This is called when a file system is unmounted.
- */
-void
-ext2_xattr_put_super(struct super_block *sb)
-{
-	mb_cache_shrink(sb->s_bdev);
-}
-
-
-/*
  * ext2_xattr_cache_insert()
  *
  * Create a new entry in the extended attribute cache, and insert
@@ -826,28 +818,20 @@ ext2_xattr_put_super(struct super_block *sb)
  * Returns 0, or a negative error number on failure.
  */
 static int
-ext2_xattr_cache_insert(struct buffer_head *bh)
+ext2_xattr_cache_insert(struct mb_cache *cache, struct buffer_head *bh)
 {
 	__u32 hash = le32_to_cpu(HDR(bh)->h_hash);
-	struct mb_cache_entry *ce;
 	int error;
 
-	ce = mb_cache_entry_alloc(ext2_xattr_cache, GFP_NOFS);
-	if (!ce)
-		return -ENOMEM;
-	error = mb_cache_entry_insert(ce, bh->b_bdev, bh->b_blocknr, hash);
+	error = mb_cache_entry_create(cache, GFP_NOFS, hash, bh->b_blocknr, 1);
 	if (error) {
-		mb_cache_entry_free(ce);
 		if (error == -EBUSY) {
 			ea_bdebug(bh, "already in cache (%d cache entries)",
 				atomic_read(&ext2_xattr_cache->c_entry_count));
 			error = 0;
 		}
-	} else {
-		ea_bdebug(bh, "inserting [%x] (%d cache entries)", (int)hash,
-			  atomic_read(&ext2_xattr_cache->c_entry_count));
-		mb_cache_entry_release(ce);
-	}
+	} else
+		ea_bdebug(bh, "inserting [%x]", (int)hash);
 	return error;
 }
 
@@ -904,22 +888,16 @@ ext2_xattr_cache_find(struct inode *inode, struct ext2_xattr_header *header)
 {
 	__u32 hash = le32_to_cpu(header->h_hash);
 	struct mb_cache_entry *ce;
+	struct mb_cache *ext2_mb_cache = EXT2_SB(inode->i_sb)->s_mb_cache;
 
 	if (!header->h_hash)
 		return NULL;  /* never share */
 	ea_idebug(inode, "looking for cached blocks [%x]", (int)hash);
 again:
-	ce = mb_cache_entry_find_first(ext2_xattr_cache, inode->i_sb->s_bdev,
-				       hash);
+	ce = mb_cache_entry_find_first(ext2_mb_cache, hash);
 	while (ce) {
 		struct buffer_head *bh;
 
-		if (IS_ERR(ce)) {
-			if (PTR_ERR(ce) == -EAGAIN)
-				goto again;
-			break;
-		}
-
 		bh = sb_bread(inode->i_sb, ce->e_block);
 		if (!bh) {
 			ext2_error(inode->i_sb, "ext2_xattr_cache_find",
@@ -927,7 +905,21 @@ again:
 				inode->i_ino, (unsigned long) ce->e_block);
 		} else {
 			lock_buffer(bh);
-			if (le32_to_cpu(HDR(bh)->h_refcount) >
+			/*
+			 * We have to be careful about races with freeing or
+			 * rehashing of xattr block. Once we hold buffer lock
+			 * xattr block's state is stable so we can check
+			 * whether the block got freed / rehashed or not.
+			 * Since we unhash mbcache entry under buffer lock when
+			 * freeing / rehashing xattr block, checking whether
+			 * entry is still hashed is reliable.
+			 */
+			if (hlist_bl_unhashed(&ce->e_hash_list)) {
+				mb_cache_entry_put(ext2_mb_cache, ce);
+				unlock_buffer(bh);
+				brelse(bh);
+				goto again;
+			} else if (le32_to_cpu(HDR(bh)->h_refcount) >
 				   EXT2_XATTR_REFCOUNT_MAX) {
 				ea_idebug(inode, "block %ld refcount %d>%d",
 					  (unsigned long) ce->e_block,
@@ -936,13 +928,14 @@ again:
 			} else if (!ext2_xattr_cmp(header, HDR(bh))) {
 				ea_bdebug(bh, "b_count=%d",
 					  atomic_read(&(bh->b_count)));
-				mb_cache_entry_release(ce);
+				mb_cache_entry_touch(ext2_mb_cache, ce);
+				mb_cache_entry_put(ext2_mb_cache, ce);
 				return bh;
 			}
 			unlock_buffer(bh);
 			brelse(bh);
 		}
-		ce = mb_cache_entry_find_next(ce, inode->i_sb->s_bdev, hash);
+		ce = mb_cache_entry_find_next(ext2_mb_cache, ce);
 	}
 	return NULL;
 }
@@ -1015,17 +1008,15 @@ static void ext2_xattr_rehash(struct ext2_xattr_header *header,
 
 #undef BLOCK_HASH_SHIFT
 
-int __init
-init_ext2_xattr(void)
+#define HASH_BUCKET_BITS 10
+
+struct mb_cache *ext2_xattr_create_cache(void)
 {
-	ext2_xattr_cache = mb_cache_create("ext2_xattr", 6);
-	if (!ext2_xattr_cache)
-		return -ENOMEM;
-	return 0;
+	return mb_cache_create(HASH_BUCKET_BITS);
 }
 
-void
-exit_ext2_xattr(void)
+void ext2_xattr_destroy_cache(struct mb_cache *cache)
 {
-	mb_cache_destroy(ext2_xattr_cache);
+	if (cache)
+		mb_cache_destroy(cache);
 }
diff --git a/fs/ext2/xattr.h b/fs/ext2/xattr.h
index 60edf29..6f82ab1 100644
--- a/fs/ext2/xattr.h
+++ b/fs/ext2/xattr.h
@@ -53,6 +53,8 @@ struct ext2_xattr_entry {
 #define EXT2_XATTR_SIZE(size) \
 	(((size) + EXT2_XATTR_ROUND) & ~EXT2_XATTR_ROUND)
 
+struct mb_cache;
+
 # ifdef CONFIG_EXT2_FS_XATTR
 
 extern const struct xattr_handler ext2_xattr_user_handler;
@@ -65,10 +67,9 @@ extern int ext2_xattr_get(struct inode *, int, const char *, void *, size_t);
 extern int ext2_xattr_set(struct inode *, int, const char *, const void *, size_t, int);
 
 extern void ext2_xattr_delete_inode(struct inode *);
-extern void ext2_xattr_put_super(struct super_block *);
 
-extern int init_ext2_xattr(void);
-extern void exit_ext2_xattr(void);
+extern struct mb_cache *ext2_xattr_create_cache(void);
+extern void ext2_xattr_destroy_cache(struct mb_cache *cache);
 
 extern const struct xattr_handler *ext2_xattr_handlers[];
 
@@ -93,19 +94,7 @@ ext2_xattr_delete_inode(struct inode *inode)
 {
 }
 
-static inline void
-ext2_xattr_put_super(struct super_block *sb)
-{
-}
-
-static inline int
-init_ext2_xattr(void)
-{
-	return 0;
-}
-
-static inline void
-exit_ext2_xattr(void)
+static inline void ext2_xattr_destroy_cache(struct mb_cache *cache)
 {
 }