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

Pull ext4 updates from Ted Ts'o:
 "Various bug fixes and cleanups for ext4; no new features this cycle"

* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4: (29 commits)
  ext4: remove unnecessary wbc parameter from ext4_bio_write_page
  ext4: avoid s_mb_prefetch to be zero in individual scenarios
  ext4: defer saving error info from atomic context
  ext4: simplify ext4 error translation
  ext4: move functions in super.c
  ext4: make ext4_abort() use __ext4_error()
  ext4: standardize error message in ext4_protect_reserved_inode()
  ext4: remove redundant sb checksum recomputation
  ext4: don't remount read-only with errors=continue on reboot
  ext4: fix deadlock with fs freezing and EA inodes
  jbd2: add a helper to find out number of fast commit blocks
  ext4: make fast_commit.h byte identical with e2fsprogs/fast_commit.h
  ext4: fix fall-through warnings for Clang
  ext4: add docs about fast commit idempotence
  ext4: remove the unused EXT4_CURRENT_REV macro
  ext4: fix an IS_ERR() vs NULL check
  ext4: check for invalid block size early when mounting a file system
  ext4: fix a memory leak of ext4_free_data
  ext4: delete nonsensical (commented-out) code inside ext4_xattr_block_set()
  ext4: update ext4_data_block_valid related comments
  ...

Documentation/filesystems/ext4/journal.rst

@@ -681,3 +681,53 @@ Here is the list of supported tags and their meanings:
      - Stores the TID of the commit, CRC of the fast commit of which this tag
        represents the end of
 
+Fast Commit Replay Idempotence
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Fast commits tags are idempotent in nature provided the recovery code follows
+certain rules. The guiding principle that the commit path follows while
+committing is that it stores the result of a particular operation instead of
+storing the procedure.
+
+Let's consider this rename operation: 'mv /a /b'. Let's assume dirent '/a'
+was associated with inode 10. During fast commit, instead of storing this
+operation as a procedure "rename a to b", we store the resulting file system
+state as a "series" of outcomes:
+
+- Link dirent b to inode 10
+- Unlink dirent a
+- Inode 10 with valid refcount
+
+Now when recovery code runs, it needs "enforce" this state on the file
+system. This is what guarantees idempotence of fast commit replay.
+
+Let's take an example of a procedure that is not idempotent and see how fast
+commits make it idempotent. Consider following sequence of operations:
+
+1) rm A
+2) mv B A
+3) read A
+
+If we store this sequence of operations as is then the replay is not idempotent.
+Let's say while in replay, we crash after (2). During the second replay,
+file A (which was actually created as a result of "mv B A" operation) would get
+deleted. Thus, file named A would be absent when we try to read A. So, this
+sequence of operations is not idempotent. However, as mentioned above, instead
+of storing the procedure fast commits store the outcome of each procedure. Thus
+the fast commit log for above procedure would be as follows:
+
+(Let's assume dirent A was linked to inode 10 and dirent B was linked to
+inode 11 before the replay)
+
+1) Unlink A
+2) Link A to inode 11
+3) Unlink B
+4) Inode 11
+
+If we crash after (3) we will have file A linked to inode 11. During the second
+replay, we will remove file A (inode 11). But we will create it back and make
+it point to inode 11. We won't find B, so we'll just skip that step. At this
+point, the refcount for inode 11 is not reliable, but that gets fixed by the
+replay of last inode 11 tag. Thus, by converting a non-idempotent procedure
+into a series of idempotent outcomes, fast commits ensured idempotence during
+the replay.

fs/ext4/balloc.c

@@ -185,7 +185,7 @@ static int ext4_init_block_bitmap(struct super_block *sb,
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 	ext4_fsblk_t start, tmp;
 
-	J_ASSERT_BH(bh, buffer_locked(bh));
+	ASSERT(buffer_locked(bh));
 
 	/* If checksum is bad mark all blocks used to prevent allocation
 	 * essentially implementing a per-group read-only flag. */

fs/ext4/block_validity.c

@@ -176,12 +176,10 @@ static int ext4_protect_reserved_inode(struct super_block *sb,
 			err = add_system_zone(system_blks, map.m_pblk, n, ino);
 			if (err < 0) {
 				if (err == -EFSCORRUPTED) {
-					__ext4_error(sb, __func__, __LINE__,
-						     -err, map.m_pblk,
-						     "blocks %llu-%llu from inode %u overlap system zone",
-						     map.m_pblk,
-						     map.m_pblk + map.m_len - 1,
-						     ino);
+					EXT4_ERROR_INODE_ERR(inode, -err,
+						"blocks %llu-%llu from inode overlap system zone",
+						map.m_pblk,
+						map.m_pblk + map.m_len - 1);
 				}
 				break;
 			}
@@ -206,7 +204,7 @@ static void ext4_destroy_system_zone(struct rcu_head *rcu)
  *
  * The update of system_blks pointer in this function is protected by
  * sb->s_umount semaphore. However we have to be careful as we can be
- * racing with ext4_data_block_valid() calls reading system_blks rbtree
+ * racing with ext4_inode_block_valid() calls reading system_blks rbtree
  * protected only by RCU. That's why we first build the rbtree and then
  * swap it in place.
  */
@@ -258,7 +256,7 @@ int ext4_setup_system_zone(struct super_block *sb)
 
 	/*
 	 * System blks rbtree complete, announce it once to prevent racing
-	 * with ext4_data_block_valid() accessing the rbtree at the same
+	 * with ext4_inode_block_valid() accessing the rbtree at the same
 	 * time.
 	 */
 	rcu_assign_pointer(sbi->s_system_blks, system_blks);
@@ -278,7 +276,7 @@ int ext4_setup_system_zone(struct super_block *sb)
  *
  * The update of system_blks pointer in this function is protected by
  * sb->s_umount semaphore. However we have to be careful as we can be
- * racing with ext4_data_block_valid() calls reading system_blks rbtree
+ * racing with ext4_inode_block_valid() calls reading system_blks rbtree
  * protected only by RCU. So we first clear the system_blks pointer and
  * then free the rbtree only after RCU grace period expires.
  */

fs/ext4/ext4.h

@@ -98,6 +98,16 @@
 #define ext_debug(ino, fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
 #endif
 
+#define ASSERT(assert)						\
+do {									\
+	if (unlikely(!(assert))) {					\
+		printk(KERN_EMERG					\
+		       "Assertion failure in %s() at %s:%d: '%s'\n",	\
+		       __func__, __FILE__, __LINE__, #assert);		\
+		BUG();							\
+	}								\
+} while (0)
+
 /* data type for block offset of block group */
 typedef int ext4_grpblk_t;
 
@@ -1619,6 +1629,27 @@ struct ext4_sb_info {
 	errseq_t s_bdev_wb_err;
 	spinlock_t s_bdev_wb_lock;
 
+	/* Information about errors that happened during this mount */
+	spinlock_t s_error_lock;
+	int s_add_error_count;
+	int s_first_error_code;
+	__u32 s_first_error_line;
+	__u32 s_first_error_ino;
+	__u64 s_first_error_block;
+	const char *s_first_error_func;
+	time64_t s_first_error_time;
+	int s_last_error_code;
+	__u32 s_last_error_line;
+	__u32 s_last_error_ino;
+	__u64 s_last_error_block;
+	const char *s_last_error_func;
+	time64_t s_last_error_time;
+	/*
+	 * If we are in a context where we cannot update error information in
+	 * the on-disk superblock, we queue this work to do it.
+	 */
+	struct work_struct s_error_work;
+
 	/* Ext4 fast commit stuff */
 	atomic_t s_fc_subtid;
 	atomic_t s_fc_ineligible_updates;
@@ -1858,7 +1889,6 @@ static inline bool ext4_verity_in_progress(struct inode *inode)
 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
 
-#define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
 
 #define EXT4_GOOD_OLD_INODE_SIZE 128
@@ -2952,9 +2982,9 @@ extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
 					     ext4_group_t block_group,
 					     unsigned int flags);
 
-extern __printf(6, 7)
-void __ext4_error(struct super_block *, const char *, unsigned int, int, __u64,
-		  const char *, ...);
+extern __printf(7, 8)
+void __ext4_error(struct super_block *, const char *, unsigned int, bool,
+		  int, __u64, const char *, ...);
 extern __printf(6, 7)
 void __ext4_error_inode(struct inode *, const char *, unsigned int,
 			ext4_fsblk_t, int, const char *, ...);
@@ -2963,9 +2993,6 @@ void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
 		     const char *, ...);
 extern void __ext4_std_error(struct super_block *, const char *,
 			     unsigned int, int);
-extern __printf(5, 6)
-void __ext4_abort(struct super_block *, const char *, unsigned int, int,
-		  const char *, ...);
 extern __printf(4, 5)
 void __ext4_warning(struct super_block *, const char *, unsigned int,
 		    const char *, ...);
@@ -2995,6 +3022,9 @@ void __ext4_grp_locked_error(const char *, unsigned int,
 #define EXT4_ERROR_FILE(file, block, fmt, a...)				\
 	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
 
+#define ext4_abort(sb, err, fmt, a...)					\
+	__ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
+
 #ifdef CONFIG_PRINTK
 
 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
@@ -3005,11 +3035,11 @@ void __ext4_grp_locked_error(const char *, unsigned int,
 #define ext4_error_file(file, func, line, block, fmt, ...)		\
 	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
 #define ext4_error(sb, fmt, ...)					\
-	__ext4_error((sb), __func__, __LINE__, 0, 0, (fmt), ##__VA_ARGS__)
+	__ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt),	\
+		##__VA_ARGS__)
 #define ext4_error_err(sb, err, fmt, ...)				\
-	__ext4_error((sb), __func__, __LINE__, (err), 0, (fmt), ##__VA_ARGS__)
-#define ext4_abort(sb, err, fmt, ...)					\
-	__ext4_abort((sb), __func__, __LINE__, (err), (fmt), ##__VA_ARGS__)
+	__ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt),	\
+		##__VA_ARGS__)
 #define ext4_warning(sb, fmt, ...)					\
 	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
 #define ext4_warning_inode(inode, fmt, ...)				\
@@ -3042,17 +3072,12 @@ do {									\
 #define ext4_error(sb, fmt, ...)					\
 do {									\
 	no_printk(fmt, ##__VA_ARGS__);					\
-	__ext4_error(sb, "", 0, 0, 0, " ");				\
+	__ext4_error(sb, "", 0, false, 0, 0, " ");			\
 } while (0)
 #define ext4_error_err(sb, err, fmt, ...)				\
 do {									\
 	no_printk(fmt, ##__VA_ARGS__);					\
-	__ext4_error(sb, "", 0, err, 0, " ");				\
-} while (0)
-#define ext4_abort(sb, err, fmt, ...)					\
-do {									\
-	no_printk(fmt, ##__VA_ARGS__);					\
-	__ext4_abort(sb, "", 0, err, " ");				\
+	__ext4_error(sb, "", 0, false, err, 0, " ");			\
 } while (0)
 #define ext4_warning(sb, fmt, ...)					\
 do {									\
@@ -3361,6 +3386,21 @@ static inline void ext4_unlock_group(struct super_block *sb,
 	spin_unlock(ext4_group_lock_ptr(sb, group));
 }
 
+#ifdef CONFIG_QUOTA
+static inline bool ext4_quota_capable(struct super_block *sb)
+{
+	return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
+}
+
+static inline bool ext4_is_quota_journalled(struct super_block *sb)
+{
+	struct ext4_sb_info *sbi = EXT4_SB(sb);
+
+	return (ext4_has_feature_quota(sb) ||
+		sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
+}
+#endif
+
 /*
  * Block validity checking
  */
@@ -3609,7 +3649,6 @@ extern void ext4_io_submit(struct ext4_io_submit *io);
 extern int ext4_bio_write_page(struct ext4_io_submit *io,
 			       struct page *page,
 			       int len,
-			       struct writeback_control *wbc,
 			       bool keep_towrite);
 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);

fs/ext4/ext4_jbd2.c

@@ -296,8 +296,8 @@ int __ext4_forget(const char *where, unsigned int line, handle_t *handle,
 	if (err) {
 		ext4_journal_abort_handle(where, line, __func__,
 					  bh, handle, err);
-		__ext4_abort(inode->i_sb, where, line, -err,
-			   "error %d when attempting revoke", err);
+		__ext4_error(inode->i_sb, where, line, true, -err, 0,
+			     "error %d when attempting revoke", err);
 	}
 	BUFFER_TRACE(bh, "exit");
 	return err;

fs/ext4/ext4_jbd2.h

@@ -86,17 +86,14 @@
 #ifdef CONFIG_QUOTA
 /* Amount of blocks needed for quota update - we know that the structure was
  * allocated so we need to update only data block */
-#define EXT4_QUOTA_TRANS_BLOCKS(sb) ((test_opt(sb, QUOTA) ||\
-		ext4_has_feature_quota(sb)) ? 1 : 0)
+#define EXT4_QUOTA_TRANS_BLOCKS(sb) ((ext4_quota_capable(sb)) ? 1 : 0)
 /* Amount of blocks needed for quota insert/delete - we do some block writes
  * but inode, sb and group updates are done only once */
-#define EXT4_QUOTA_INIT_BLOCKS(sb) ((test_opt(sb, QUOTA) ||\
-		ext4_has_feature_quota(sb)) ?\
+#define EXT4_QUOTA_INIT_BLOCKS(sb) ((ext4_quota_capable(sb)) ?\
 		(DQUOT_INIT_ALLOC*(EXT4_SINGLEDATA_TRANS_BLOCKS(sb)-3)\
 		 +3+DQUOT_INIT_REWRITE) : 0)
 
-#define EXT4_QUOTA_DEL_BLOCKS(sb) ((test_opt(sb, QUOTA) ||\
-		ext4_has_feature_quota(sb)) ?\
+#define EXT4_QUOTA_DEL_BLOCKS(sb) ((ext4_quota_capable(sb)) ?\
 		(DQUOT_DEL_ALLOC*(EXT4_SINGLEDATA_TRANS_BLOCKS(sb)-3)\
 		 +3+DQUOT_DEL_REWRITE) : 0)
 #else

fs/ext4/extents.c

@@ -5815,8 +5815,8 @@ int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
 	int ret;
 
 	path = ext4_find_extent(inode, start, NULL, 0);
-	if (!path)
-		return -EINVAL;
+	if (IS_ERR(path))
+		return PTR_ERR(path);
 	ex = path[path->p_depth].p_ext;
 	if (!ex) {
 		ret = -EFSCORRUPTED;
@@ -5988,7 +5988,6 @@ int ext4_ext_replay_set_iblocks(struct inode *inode)
 			kfree(path);
 			break;
 		}
-		ex = path2[path2->p_depth].p_ext;
 		for (i = 0; i <= max(path->p_depth, path2->p_depth); i++) {
 			cmp1 = cmp2 = 0;
 			if (i <= path->p_depth)

fs/ext4/fast_commit.c

@@ -103,8 +103,69 @@
  *
  * Replay code should thus check for all the valid tails in the FC area.
  *
+ * Fast Commit Replay Idempotence
+ * ------------------------------
+ *
+ * Fast commits tags are idempotent in nature provided the recovery code follows
+ * certain rules. The guiding principle that the commit path follows while
+ * committing is that it stores the result of a particular operation instead of
+ * storing the procedure.
+ *
+ * Let's consider this rename operation: 'mv /a /b'. Let's assume dirent '/a'
+ * was associated with inode 10. During fast commit, instead of storing this
+ * operation as a procedure "rename a to b", we store the resulting file system
+ * state as a "series" of outcomes:
+ *
+ * - Link dirent b to inode 10
+ * - Unlink dirent a
+ * - Inode <10> with valid refcount
+ *
+ * Now when recovery code runs, it needs "enforce" this state on the file
+ * system. This is what guarantees idempotence of fast commit replay.
+ *
+ * Let's take an example of a procedure that is not idempotent and see how fast
+ * commits make it idempotent. Consider following sequence of operations:
+ *
+ *     rm A;    mv B A;    read A
+ *  (x)     (y)        (z)
+ *
+ * (x), (y) and (z) are the points at which we can crash. If we store this
+ * sequence of operations as is then the replay is not idempotent. Let's say
+ * while in replay, we crash at (z). During the second replay, file A (which was
+ * actually created as a result of "mv B A" operation) would get deleted. Thus,
+ * file named A would be absent when we try to read A. So, this sequence of
+ * operations is not idempotent. However, as mentioned above, instead of storing
+ * the procedure fast commits store the outcome of each procedure. Thus the fast
+ * commit log for above procedure would be as follows:
+ *
+ * (Let's assume dirent A was linked to inode 10 and dirent B was linked to
+ * inode 11 before the replay)
+ *
+ *    [Unlink A]   [Link A to inode 11]   [Unlink B]   [Inode 11]
+ * (w)          (x)                    (y)          (z)
+ *
+ * If we crash at (z), we will have file A linked to inode 11. During the second
+ * replay, we will remove file A (inode 11). But we will create it back and make
+ * it point to inode 11. We won't find B, so we'll just skip that step. At this
+ * point, the refcount for inode 11 is not reliable, but that gets fixed by the
+ * replay of last inode 11 tag. Crashes at points (w), (x) and (y) get handled
+ * similarly. Thus, by converting a non-idempotent procedure into a series of
+ * idempotent outcomes, fast commits ensured idempotence during the replay.
+ *
  * TODOs
  * -----
+ *
+ * 0) Fast commit replay path hardening: Fast commit replay code should use
+ *    journal handles to make sure all the updates it does during the replay
+ *    path are atomic. With that if we crash during fast commit replay, after
+ *    trying to do recovery again, we will find a file system where fast commit
+ *    area is invalid (because new full commit would be found). In order to deal
+ *    with that, fast commit replay code should ensure that the "FC_REPLAY"
+ *    superblock state is persisted before starting the replay, so that after
+ *    the crash, fast commit recovery code can look at that flag and perform
+ *    fast commit recovery even if that area is invalidated by later full
+ *    commits.
+ *
  * 1) Make fast commit atomic updates more fine grained. Today, a fast commit
  *    eligible update must be protected within ext4_fc_start_update() and
  *    ext4_fc_stop_update(). These routines are called at much higher
@@ -1220,18 +1281,6 @@ static void ext4_fc_cleanup(journal_t *journal, int full)
 
 /* Ext4 Replay Path Routines */
 
-/* Get length of a particular tlv */
-static inline int ext4_fc_tag_len(struct ext4_fc_tl *tl)
-{
-	return le16_to_cpu(tl->fc_len);
-}
-
-/* Get a pointer to "value" of a tlv */
-static inline u8 *ext4_fc_tag_val(struct ext4_fc_tl *tl)
-{
-	return (u8 *)tl + sizeof(*tl);
-}
-
 /* Helper struct for dentry replay routines */
 struct dentry_info_args {
 	int parent_ino, dname_len, ino, inode_len;
@@ -1770,32 +1819,6 @@ ext4_fc_replay_del_range(struct super_block *sb, struct ext4_fc_tl *tl)
 	return 0;
 }
 
-static inline const char *tag2str(u16 tag)
-{
-	switch (tag) {
-	case EXT4_FC_TAG_LINK:
-		return "TAG_ADD_ENTRY";
-	case EXT4_FC_TAG_UNLINK:
-		return "TAG_DEL_ENTRY";
-	case EXT4_FC_TAG_ADD_RANGE:
-		return "TAG_ADD_RANGE";
-	case EXT4_FC_TAG_CREAT:
-		return "TAG_CREAT_DENTRY";
-	case EXT4_FC_TAG_DEL_RANGE:
-		return "TAG_DEL_RANGE";
-	case EXT4_FC_TAG_INODE:
-		return "TAG_INODE";
-	case EXT4_FC_TAG_PAD:
-		return "TAG_PAD";
-	case EXT4_FC_TAG_TAIL:
-		return "TAG_TAIL";
-	case EXT4_FC_TAG_HEAD:
-		return "TAG_HEAD";
-	default:
-		return "TAG_ERROR";
-	}
-}
-
 static void ext4_fc_set_bitmaps_and_counters(struct super_block *sb)
 {
 	struct ext4_fc_replay_state *state;

fs/ext4/fast_commit.h

@@ -3,6 +3,11 @@
 #ifndef __FAST_COMMIT_H__
 #define __FAST_COMMIT_H__
 
+/*
+ * Note this file is present in e2fsprogs/lib/ext2fs/fast_commit.h and
+ * linux/fs/ext4/fast_commit.h. These file should always be byte identical.
+ */
+
 /* Fast commit tags */
 #define EXT4_FC_TAG_ADD_RANGE		0x0001
 #define EXT4_FC_TAG_DEL_RANGE		0x0002
@@ -50,7 +55,7 @@ struct ext4_fc_del_range {
 struct ext4_fc_dentry_info {
 	__le32 fc_parent_ino;
 	__le32 fc_ino;
-	u8 fc_dname[0];
+	__u8 fc_dname[0];
 };
 
 /* Value structure for EXT4_FC_TAG_INODE and EXT4_FC_TAG_INODE_PARTIAL. */
@@ -65,19 +70,6 @@ struct ext4_fc_tail {
 	__le32 fc_crc;
 };
 
-/*
- * In memory list of dentry updates that are performed on the file
- * system used by fast commit code.
- */
-struct ext4_fc_dentry_update {
-	int fcd_op;		/* Type of update create / unlink / link */
-	int fcd_parent;		/* Parent inode number */
-	int fcd_ino;		/* Inode number */
-	struct qstr fcd_name;	/* Dirent name */
-	unsigned char fcd_iname[DNAME_INLINE_LEN];	/* Dirent name string */
-	struct list_head fcd_list;
-};
-
 /*
  * Fast commit reason codes
  */
@@ -107,6 +99,20 @@ enum {
 	EXT4_FC_REASON_MAX
 };
 
+#ifdef __KERNEL__
+/*
+ * In memory list of dentry updates that are performed on the file
+ * system used by fast commit code.
+ */
+struct ext4_fc_dentry_update {
+	int fcd_op;		/* Type of update create / unlink / link */
+	int fcd_parent;		/* Parent inode number */
+	int fcd_ino;		/* Inode number */
+	struct qstr fcd_name;	/* Dirent name */
+	unsigned char fcd_iname[DNAME_INLINE_LEN];	/* Dirent name string */
+	struct list_head fcd_list;
+};
+
 struct ext4_fc_stats {
 	unsigned int fc_ineligible_reason_count[EXT4_FC_REASON_MAX];
 	unsigned long fc_num_commits;
@@ -145,13 +151,51 @@ struct ext4_fc_replay_state {
 };
 
 #define region_last(__region) (((__region)->lblk) + ((__region)->len) - 1)
+#endif
 
 #define fc_for_each_tl(__start, __end, __tl)				\
-	for (tl = (struct ext4_fc_tl *)start;				\
-		(u8 *)tl < (u8 *)end;					\
-		tl = (struct ext4_fc_tl *)((u8 *)tl +			\
+	for (tl = (struct ext4_fc_tl *)(__start);			\
+	     (__u8 *)tl < (__u8 *)(__end);				\
+		tl = (struct ext4_fc_tl *)((__u8 *)tl +			\
 					sizeof(struct ext4_fc_tl) +	\
 					+ le16_to_cpu(tl->fc_len)))
 
+static inline const char *tag2str(__u16 tag)
+{
+	switch (tag) {
+	case EXT4_FC_TAG_LINK:
+		return "ADD_ENTRY";
+	case EXT4_FC_TAG_UNLINK:
+		return "DEL_ENTRY";
+	case EXT4_FC_TAG_ADD_RANGE:
+		return "ADD_RANGE";
+	case EXT4_FC_TAG_CREAT:
+		return "CREAT_DENTRY";
+	case EXT4_FC_TAG_DEL_RANGE:
+		return "DEL_RANGE";
+	case EXT4_FC_TAG_INODE:
+		return "INODE";
+	case EXT4_FC_TAG_PAD:
+		return "PAD";
+	case EXT4_FC_TAG_TAIL:
+		return "TAIL";
+	case EXT4_FC_TAG_HEAD:
+		return "HEAD";
+	default:
+		return "ERROR";
+	}
+}
+
+/* Get length of a particular tlv */
+static inline int ext4_fc_tag_len(struct ext4_fc_tl *tl)
+{
+	return le16_to_cpu(tl->fc_len);
+}
+
+/* Get a pointer to "value" of a tlv */
+static inline __u8 *ext4_fc_tag_val(struct ext4_fc_tl *tl)
+{
+	return (__u8 *)tl + sizeof(*tl);
+}
 
 #endif /* __FAST_COMMIT_H__ */

fs/ext4/fsync.c

@@ -136,7 +136,7 @@ int ext4_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	if (unlikely(ext4_forced_shutdown(sbi)))
 		return -EIO;
 
-	J_ASSERT(ext4_journal_current_handle() == NULL);
+	ASSERT(ext4_journal_current_handle() == NULL);
 
 	trace_ext4_sync_file_enter(file, datasync);
 

fs/ext4/indirect.c

@@ -534,8 +534,8 @@ int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
 	ext4_fsblk_t first_block = 0;
 
 	trace_ext4_ind_map_blocks_enter(inode, map->m_lblk, map->m_len, flags);
-	J_ASSERT(!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)));
-	J_ASSERT(handle != NULL || (flags & EXT4_GET_BLOCKS_CREATE) == 0);
+	ASSERT(!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)));
+	ASSERT(handle != NULL || (flags & EXT4_GET_BLOCKS_CREATE) == 0);
 	depth = ext4_block_to_path(inode, map->m_lblk, offsets,
 				   &blocks_to_boundary);
 

fs/ext4/inode.c

@@ -175,6 +175,7 @@ void ext4_evict_inode(struct inode *inode)
 	 */
 	int extra_credits = 6;
 	struct ext4_xattr_inode_array *ea_inode_array = NULL;
+	bool freeze_protected = false;
 
 	trace_ext4_evict_inode(inode);
 
@@ -232,9 +233,14 @@ void ext4_evict_inode(struct inode *inode)
 
 	/*
 	 * Protect us against freezing - iput() caller didn't have to have any
-	 * protection against it
+	 * protection against it. When we are in a running transaction though,
+	 * we are already protected against freezing and we cannot grab further
+	 * protection due to lock ordering constraints.
 	 */
-	sb_start_intwrite(inode->i_sb);
+	if (!ext4_journal_current_handle()) {
+		sb_start_intwrite(inode->i_sb);
+		freeze_protected = true;
+	}
 
 	if (!IS_NOQUOTA(inode))
 		extra_credits += EXT4_MAXQUOTAS_DEL_BLOCKS(inode->i_sb);
@@ -253,7 +259,8 @@ void ext4_evict_inode(struct inode *inode)
 		 * cleaned up.
 		 */
 		ext4_orphan_del(NULL, inode);
-		sb_end_intwrite(inode->i_sb);
+		if (freeze_protected)
+			sb_end_intwrite(inode->i_sb);
 		goto no_delete;
 	}
 
@@ -294,7 +301,8 @@ void ext4_evict_inode(struct inode *inode)
 stop_handle:
 		ext4_journal_stop(handle);
 		ext4_orphan_del(NULL, inode);
-		sb_end_intwrite(inode->i_sb);
+		if (freeze_protected)
+			sb_end_intwrite(inode->i_sb);
 		ext4_xattr_inode_array_free(ea_inode_array);
 		goto no_delete;
 	}
@@ -323,7 +331,8 @@ void ext4_evict_inode(struct inode *inode)
 	else
 		ext4_free_inode(handle, inode);
 	ext4_journal_stop(handle);
-	sb_end_intwrite(inode->i_sb);
+	if (freeze_protected)
+		sb_end_intwrite(inode->i_sb);
 	ext4_xattr_inode_array_free(ea_inode_array);
 	return;
 no_delete:
@@ -830,8 +839,8 @@ struct buffer_head *ext4_getblk(handle_t *handle, struct inode *inode,
 	int create = map_flags & EXT4_GET_BLOCKS_CREATE;
 	int err;
 
-	J_ASSERT((EXT4_SB(inode->i_sb)->s_mount_state & EXT4_FC_REPLAY)
-		 || handle != NULL || create == 0);
+	ASSERT((EXT4_SB(inode->i_sb)->s_mount_state & EXT4_FC_REPLAY)
+		    || handle != NULL || create == 0);
 
 	map.m_lblk = block;
 	map.m_len = 1;
@@ -846,9 +855,9 @@ struct buffer_head *ext4_getblk(handle_t *handle, struct inode *inode,
 	if (unlikely(!bh))
 		return ERR_PTR(-ENOMEM);
 	if (map.m_flags & EXT4_MAP_NEW) {
-		J_ASSERT(create != 0);
-		J_ASSERT((EXT4_SB(inode->i_sb)->s_mount_state & EXT4_FC_REPLAY)
-			 || (handle != NULL));
+		ASSERT(create != 0);
+		ASSERT((EXT4_SB(inode->i_sb)->s_mount_state & EXT4_FC_REPLAY)
+			    || (handle != NULL));
 
 		/*
 		 * Now that we do not always journal data, we should
@@ -2055,7 +2064,7 @@ static int ext4_writepage(struct page *page,
 		unlock_page(page);
 		return -ENOMEM;
 	}
-	ret = ext4_bio_write_page(&io_submit, page, len, wbc, keep_towrite);
+	ret = ext4_bio_write_page(&io_submit, page, len, keep_towrite);
 	ext4_io_submit(&io_submit);
 	/* Drop io_end reference we got from init */
 	ext4_put_io_end_defer(io_submit.io_end);
@@ -2089,7 +2098,7 @@ static int mpage_submit_page(struct mpage_da_data *mpd, struct page *page)
 		len = size & ~PAGE_MASK;
 	else
 		len = PAGE_SIZE;
-	err = ext4_bio_write_page(&mpd->io_submit, page, len, mpd->wbc, false);
+	err = ext4_bio_write_page(&mpd->io_submit, page, len, false);
 	if (!err)
 		mpd->wbc->nr_to_write--;
 	mpd->first_page++;
@@ -4610,7 +4619,7 @@ struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
 	    (ino > le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count))) {
 		if (flags & EXT4_IGET_HANDLE)
 			return ERR_PTR(-ESTALE);
-		__ext4_error(sb, function, line, EFSCORRUPTED, 0,
+		__ext4_error(sb, function, line, false, EFSCORRUPTED, 0,
 			     "inode #%lu: comm %s: iget: illegal inode #",
 			     ino, current->comm);
 		return ERR_PTR(-EFSCORRUPTED);

fs/ext4/mballoc.c

@@ -822,24 +822,6 @@ void ext4_mb_generate_buddy(struct super_block *sb,
 	spin_unlock(&sbi->s_bal_lock);
 }
 
-static void mb_regenerate_buddy(struct ext4_buddy *e4b)
-{
-	int count;
-	int order = 1;
-	void *buddy;
-
-	while ((buddy = mb_find_buddy(e4b, order++, &count))) {
-		ext4_set_bits(buddy, 0, count);
-	}
-	e4b->bd_info->bb_fragments = 0;
-	memset(e4b->bd_info->bb_counters, 0,
-		sizeof(*e4b->bd_info->bb_counters) *
-		(e4b->bd_sb->s_blocksize_bits + 2));
-
-	ext4_mb_generate_buddy(e4b->bd_sb, e4b->bd_buddy,
-		e4b->bd_bitmap, e4b->bd_group);
-}
-
 /* The buddy information is attached the buddy cache inode
  * for convenience. The information regarding each group
  * is loaded via ext4_mb_load_buddy. The information involve
@@ -1307,22 +1289,18 @@ static void ext4_mb_unload_buddy(struct ext4_buddy *e4b)
 
 static int mb_find_order_for_block(struct ext4_buddy *e4b, int block)
 {
-	int order = 1;
-	int bb_incr = 1 << (e4b->bd_blkbits - 1);
+	int order = 1, max;
 	void *bb;
 
 	BUG_ON(e4b->bd_bitmap == e4b->bd_buddy);
 	BUG_ON(block >= (1 << (e4b->bd_blkbits + 3)));
 
-	bb = e4b->bd_buddy;
 	while (order <= e4b->bd_blkbits + 1) {
-		block = block >> 1;
-		if (!mb_test_bit(block, bb)) {
+		bb = mb_find_buddy(e4b, order, &max);
+		if (!mb_test_bit(block >> order, bb)) {
 			/* this block is part of buddy of order 'order' */
 			return order;
 		}
-		bb += bb_incr;
-		bb_incr >>= 1;
 		order++;
 	}
 	return 0;
@@ -1512,7 +1490,6 @@ static void mb_free_blocks(struct inode *inode, struct ext4_buddy *e4b,
 				sb, e4b->bd_group,
 				EXT4_GROUP_INFO_BBITMAP_CORRUPT);
 		}
-		mb_regenerate_buddy(e4b);
 		goto done;
 	}
 
@@ -2395,9 +2372,9 @@ ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
 
 				nr = sbi->s_mb_prefetch;
 				if (ext4_has_feature_flex_bg(sb)) {
-					nr = (group / sbi->s_mb_prefetch) *
-						sbi->s_mb_prefetch;
-					nr = nr + sbi->s_mb_prefetch - group;
+					nr = 1 << sbi->s_log_groups_per_flex;
+					nr -= group & (nr - 1);
+					nr = min(nr, sbi->s_mb_prefetch);
 				}
 				prefetch_grp = ext4_mb_prefetch(sb, group,
 							nr, &prefetch_ios);
@@ -2733,7 +2710,8 @@ static int ext4_mb_init_backend(struct super_block *sb)
 
 	if (ext4_has_feature_flex_bg(sb)) {
 		/* a single flex group is supposed to be read by a single IO */
-		sbi->s_mb_prefetch = 1 << sbi->s_es->s_log_groups_per_flex;
+		sbi->s_mb_prefetch = min(1 << sbi->s_es->s_log_groups_per_flex,
+			BLK_MAX_SEGMENT_SIZE >> (sb->s_blocksize_bits - 9));
 		sbi->s_mb_prefetch *= 8; /* 8 prefetch IOs in flight at most */
 	} else {
 		sbi->s_mb_prefetch = 32;
@@ -5126,6 +5104,7 @@ ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b,
 				ext4_group_first_block_no(sb, group) +
 				EXT4_C2B(sbi, cluster),
 				"Block already on to-be-freed list");
+			kmem_cache_free(ext4_free_data_cachep, new_entry);
 			return 0;
 		}
 	}

fs/ext4/namei.c

@@ -182,10 +182,6 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 	return bh;
 }
 
-#ifndef assert
-#define assert(test) J_ASSERT(test)
-#endif
-
 #ifdef DX_DEBUG
 #define dxtrace(command) command
 #else
@@ -843,7 +839,7 @@ dx_probe(struct ext4_filename *fname, struct inode *dir,
 					break;
 				}
 			}
-			assert (at == p - 1);
+			ASSERT(at == p - 1);
 		}
 
 		at = p - 1;
@@ -1259,8 +1255,8 @@ static void dx_insert_block(struct dx_frame *frame, u32 hash, ext4_lblk_t block)
 	struct dx_entry *old = frame->at, *new = old + 1;
 	int count = dx_get_count(entries);
 
-	assert(count < dx_get_limit(entries));
-	assert(old < entries + count);
+	ASSERT(count < dx_get_limit(entries));
+	ASSERT(old < entries + count);
 	memmove(new + 1, new, (char *)(entries + count) - (char *)(new));
 	dx_set_hash(new, hash);
 	dx_set_block(new, block);
@@ -2959,7 +2955,7 @@ int ext4_orphan_add(handle_t *handle, struct inode *inode)
 	 * hold i_mutex, or the inode can not be referenced from outside,
 	 * so i_nlink should not be bumped due to race
 	 */
-	J_ASSERT((S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
+	ASSERT((S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
 		  S_ISLNK(inode->i_mode)) || inode->i_nlink == 0);
 
 	BUFFER_TRACE(sbi->s_sbh, "get_write_access");

fs/ext4/page-io.c

@@ -111,9 +111,6 @@ static void ext4_finish_bio(struct bio *bio)
 		unsigned under_io = 0;
 		unsigned long flags;
 
-		if (!page)
-			continue;
-
 		if (fscrypt_is_bounce_page(page)) {
 			bounce_page = page;
 			page = fscrypt_pagecache_page(bounce_page);
@@ -438,7 +435,6 @@ static void io_submit_add_bh(struct ext4_io_submit *io,
 int ext4_bio_write_page(struct ext4_io_submit *io,
 			struct page *page,
 			int len,
-			struct writeback_control *wbc,
 			bool keep_towrite)
 {
 	struct page *bounce_page = NULL;
@@ -448,6 +444,7 @@ int ext4_bio_write_page(struct ext4_io_submit *io,
 	int ret = 0;
 	int nr_submitted = 0;
 	int nr_to_submit = 0;
+	struct writeback_control *wbc = io->io_wbc;
 
 	BUG_ON(!PageLocked(page));
 	BUG_ON(PageWriteback(page));

fs/ext4/xattr.c

@@ -1927,7 +1927,6 @@ ext4_xattr_block_set(handle_t *handle, struct inode *inode,
 	} else {
 		/* Allocate a buffer where we construct the new block. */
 		s->base = kzalloc(sb->s_blocksize, GFP_NOFS);
-		/* assert(header == s->base) */
 		error = -ENOMEM;
 		if (s->base == NULL)
 			goto cleanup;

fs/jbd2/journal.c

@@ -1869,9 +1869,7 @@ static int load_superblock(journal_t *journal)
 
 	if (jbd2_has_feature_fast_commit(journal)) {
 		journal->j_fc_last = be32_to_cpu(sb->s_maxlen);
-		num_fc_blocks = be32_to_cpu(sb->s_num_fc_blks);
-		if (!num_fc_blocks)
-			num_fc_blocks = JBD2_MIN_FC_BLOCKS;
+		num_fc_blocks = jbd2_journal_get_num_fc_blks(sb);
 		if (journal->j_last - num_fc_blocks >= JBD2_MIN_JOURNAL_BLOCKS)
 			journal->j_last = journal->j_fc_last - num_fc_blocks;
 		journal->j_fc_first = journal->j_last + 1;
@@ -2102,9 +2100,7 @@ jbd2_journal_initialize_fast_commit(journal_t *journal)
 	journal_superblock_t *sb = journal->j_superblock;
 	unsigned long long num_fc_blks;
 
-	num_fc_blks = be32_to_cpu(sb->s_num_fc_blks);
-	if (num_fc_blks == 0)
-		num_fc_blks = JBD2_MIN_FC_BLOCKS;
+	num_fc_blks = jbd2_journal_get_num_fc_blks(sb);
 	if (journal->j_last - num_fc_blks < JBD2_MIN_JOURNAL_BLOCKS)
 		return -ENOSPC;
 

include/linux/jbd2.h

@@ -68,7 +68,7 @@ extern void *jbd2_alloc(size_t size, gfp_t flags);
 extern void jbd2_free(void *ptr, size_t size);
 
 #define JBD2_MIN_JOURNAL_BLOCKS 1024
-#define JBD2_MIN_FC_BLOCKS	256
+#define JBD2_DEFAULT_FAST_COMMIT_BLOCKS 256
 
 #ifdef __KERNEL__
 
@@ -538,6 +538,7 @@ struct transaction_chp_stats_s {
  * The transaction keeps track of all of the buffers modified by a
  * running transaction, and all of the buffers committed but not yet
  * flushed to home for finished transactions.
+ * (Locking Documentation improved by LockDoc)
  */
 
 /*
@@ -658,12 +659,12 @@ struct transaction_s
 	unsigned long		t_start;
 
 	/*
-	 * When commit was requested
+	 * When commit was requested [j_state_lock]
 	 */
 	unsigned long		t_requested;
 
 	/*
-	 * Checkpointing stats [j_checkpoint_sem]
+	 * Checkpointing stats [j_list_lock]
 	 */
 	struct transaction_chp_stats_s t_chp_stats;
 
@@ -1691,6 +1692,13 @@ static inline int jbd2_journal_has_csum_v2or3(journal_t *journal)
 	return journal->j_chksum_driver != NULL;
 }
 
+static inline int jbd2_journal_get_num_fc_blks(journal_superblock_t *jsb)
+{
+	int num_fc_blocks = be32_to_cpu(jsb->s_num_fc_blks);
+
+	return num_fc_blocks ? num_fc_blocks : JBD2_DEFAULT_FAST_COMMIT_BLOCKS;
+}
+
 /*
  * Return number of free blocks in the log. Must be called under j_state_lock.
  */