From e85c81ba8859a4c839bcd69c5d83b32954133a5b Mon Sep 17 00:00:00 2001
From: Xin Yin <yinxin.x@bytedance.com>
Date: Mon, 17 Jan 2022 17:36:54 +0800
Subject: [PATCH] ext4: fast commit may not fallback for ineligible commit
Git-commit: e85c81ba8859a4c839bcd69c5d83b32954133a5b
Patch-mainline: v5.17-rc3
References: bsc#1207590
For the follow scenario:
1. jbd start commit transaction n
2. task A get new handle for transaction n+1
3. task A do some ineligible actions and mark FC_INELIGIBLE
4. jbd complete transaction n and clean FC_INELIGIBLE
5. task A call fsync
In this case fast commit will not fallback to full commit and
transaction n+1 also not handled by jbd.
Make ext4_fc_mark_ineligible() also record transaction tid for
latest ineligible case, when call ext4_fc_cleanup() check
current transaction tid, if small than latest ineligible tid
do not clear the EXT4_MF_FC_INELIGIBLE.
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Reported-by: Ritesh Harjani <riteshh@linux.ibm.com>
Suggested-by: Harshad Shirwadkar <harshadshirwadkar@gmail.com>
Signed-off-by: Xin Yin <yinxin.x@bytedance.com>
Link: https://lore.kernel.org/r/20220117093655.35160-2-yinxin.x@bytedance.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org
Acked-by: Jan Kara <jack@suse.cz>
---
fs/ext4/ext4.h | 3 ++-
fs/ext4/extents.c | 4 ++--
fs/ext4/fast_commit.c | 33 +++++++++++++++++++++++++--------
fs/ext4/inode.c | 4 ++--
fs/ext4/ioctl.c | 4 ++--
fs/ext4/namei.c | 4 ++--
fs/ext4/super.c | 1 +
fs/ext4/xattr.c | 6 +++---
fs/jbd2/commit.c | 2 +-
fs/jbd2/journal.c | 2 +-
include/linux/jbd2.h | 2 +-
11 files changed, 42 insertions(+), 23 deletions(-)
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -1703,6 +1703,7 @@ struct ext4_sb_info {
spinlock_t s_fc_lock;
struct buffer_head *s_fc_bh;
struct ext4_fc_stats s_fc_stats;
+ tid_t s_fc_ineligible_tid;
#ifdef CONFIG_EXT4_DEBUG
int s_fc_debug_max_replay;
#endif
@@ -2871,7 +2872,7 @@ void __ext4_fc_track_create(handle_t *ha
struct dentry *dentry);
void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
-void ext4_fc_mark_ineligible(struct super_block *sb, int reason);
+void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
void ext4_fc_start_update(struct inode *inode);
void ext4_fc_stop_update(struct inode *inode);
void ext4_fc_del(struct inode *inode);
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -5336,7 +5336,7 @@ static int ext4_collapse_range(struct in
ret = PTR_ERR(handle);
goto out_mmap;
}
- ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_FALLOC_RANGE);
+ ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_FALLOC_RANGE, handle);
down_write(&EXT4_I(inode)->i_data_sem);
ext4_discard_preallocations(inode, 0);
@@ -5476,7 +5476,7 @@ static int ext4_insert_range(struct inod
ret = PTR_ERR(handle);
goto out_mmap;
}
- ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_FALLOC_RANGE);
+ ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_FALLOC_RANGE, handle);
/* Expand file to avoid data loss if there is error while shifting */
inode->i_size += len;
--- a/fs/ext4/fast_commit.c
+++ b/fs/ext4/fast_commit.c
@@ -302,18 +302,32 @@ restart:
}
/*
- * Mark file system as fast commit ineligible. This means that next commit
- * operation would result in a full jbd2 commit.
+ * Mark file system as fast commit ineligible, and record latest
+ * ineligible transaction tid. This means until the recorded
+ * transaction, commit operation would result in a full jbd2 commit.
*/
-void ext4_fc_mark_ineligible(struct super_block *sb, int reason)
+void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle)
{
struct ext4_sb_info *sbi = EXT4_SB(sb);
+ tid_t tid;
if (!test_opt2(sb, JOURNAL_FAST_COMMIT) ||
(EXT4_SB(sb)->s_mount_state & EXT4_FC_REPLAY))
return;
ext4_set_mount_flag(sb, EXT4_MF_FC_INELIGIBLE);
+ if (handle && !IS_ERR(handle))
+ tid = handle->h_transaction->t_tid;
+ else {
+ read_lock(&sbi->s_journal->j_state_lock);
+ tid = sbi->s_journal->j_running_transaction ?
+ sbi->s_journal->j_running_transaction->t_tid : 0;
+ read_unlock(&sbi->s_journal->j_state_lock);
+ }
+ spin_lock(&sbi->s_fc_lock);
+ if (sbi->s_fc_ineligible_tid < tid)
+ sbi->s_fc_ineligible_tid = tid;
+ spin_unlock(&sbi->s_fc_lock);
WARN_ON(reason >= EXT4_FC_REASON_MAX);
sbi->s_fc_stats.fc_ineligible_reason_count[reason]++;
}
@@ -389,7 +403,7 @@ static int __track_dentry_update(struct
mutex_unlock(&ei->i_fc_lock);
node = kmem_cache_alloc(ext4_fc_dentry_cachep, GFP_NOFS);
if (!node) {
- ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_NOMEM);
+ ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_NOMEM, NULL);
mutex_lock(&ei->i_fc_lock);
return -ENOMEM;
}
@@ -402,7 +416,7 @@ static int __track_dentry_update(struct
if (!node->fcd_name.name) {
kmem_cache_free(ext4_fc_dentry_cachep, node);
ext4_fc_mark_ineligible(inode->i_sb,
- EXT4_FC_REASON_NOMEM);
+ EXT4_FC_REASON_NOMEM, NULL);
mutex_lock(&ei->i_fc_lock);
return -ENOMEM;
}
@@ -504,7 +518,7 @@ void ext4_fc_track_inode(handle_t *handl
if (ext4_should_journal_data(inode)) {
ext4_fc_mark_ineligible(inode->i_sb,
- EXT4_FC_REASON_INODE_JOURNAL_DATA);
+ EXT4_FC_REASON_INODE_JOURNAL_DATA, handle);
return;
}
@@ -1189,7 +1203,7 @@ fallback:
* Fast commit cleanup routine. This is called after every fast commit and
* full commit. full is true if we are called after a full commit.
*/
-static void ext4_fc_cleanup(journal_t *journal, int full)
+static void ext4_fc_cleanup(journal_t *journal, int full, tid_t tid)
{
struct super_block *sb = journal->j_private;
struct ext4_sb_info *sbi = EXT4_SB(sb);
@@ -1237,7 +1251,10 @@ static void ext4_fc_cleanup(journal_t *j
&sbi->s_fc_q[FC_Q_MAIN]);
ext4_clear_mount_flag(sb, EXT4_MF_FC_COMMITTING);
- ext4_clear_mount_flag(sb, EXT4_MF_FC_INELIGIBLE);
+ if (tid >= sbi->s_fc_ineligible_tid) {
+ sbi->s_fc_ineligible_tid = 0;
+ ext4_clear_mount_flag(sb, EXT4_MF_FC_INELIGIBLE);
+ }
if (full)
sbi->s_fc_bytes = 0;
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -337,7 +337,7 @@ stop_handle:
return;
no_delete:
if (!list_empty(&EXT4_I(inode)->i_fc_list))
- ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_NOMEM);
+ ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_NOMEM, NULL);
ext4_clear_inode(inode); /* We must guarantee clearing of inode... */
}
@@ -6026,7 +6026,7 @@ int ext4_change_inode_journal_flag(struc
return PTR_ERR(handle);
ext4_fc_mark_ineligible(inode->i_sb,
- EXT4_FC_REASON_JOURNAL_FLAG_CHANGE);
+ EXT4_FC_REASON_JOURNAL_FLAG_CHANGE, handle);
err = ext4_mark_inode_dirty(handle, inode);
ext4_handle_sync(handle);
ext4_journal_stop(handle);
--- a/fs/ext4/ioctl.c
+++ b/fs/ext4/ioctl.c
@@ -169,7 +169,7 @@ static long swap_inode_boot_loader(struc
err = -EINVAL;
goto err_out;
}
- ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_SWAP_BOOT);
+ ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_SWAP_BOOT, handle);
/* Protect extent tree against block allocations via delalloc */
ext4_double_down_write_data_sem(inode, inode_bl);
@@ -1073,7 +1073,7 @@ mext_out:
err = ext4_resize_fs(sb, n_blocks_count);
if (EXT4_SB(sb)->s_journal) {
- ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_RESIZE);
+ ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_RESIZE, NULL);
jbd2_journal_lock_updates(EXT4_SB(sb)->s_journal);
err2 = jbd2_journal_flush(EXT4_SB(sb)->s_journal, 0);
jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal);
--- a/fs/ext4/namei.c
+++ b/fs/ext4/namei.c
@@ -4071,7 +4071,7 @@ static int ext4_rename(struct user_names
* dirents in directories.
*/
ext4_fc_mark_ineligible(old.inode->i_sb,
- EXT4_FC_REASON_RENAME_DIR);
+ EXT4_FC_REASON_RENAME_DIR, handle);
} else {
if (new.inode)
ext4_fc_track_unlink(handle, new.dentry);
@@ -4231,7 +4231,7 @@ static int ext4_cross_rename(struct inod
if (unlikely(retval))
goto end_rename;
ext4_fc_mark_ineligible(new.inode->i_sb,
- EXT4_FC_REASON_CROSS_RENAME);
+ EXT4_FC_REASON_CROSS_RENAME, handle);
if (old.dir_bh) {
retval = ext4_rename_dir_finish(handle, &old, new.dir->i_ino);
if (retval)
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -4778,6 +4778,7 @@ static int ext4_fill_super(struct super_
sbi->s_fc_bytes = 0;
ext4_clear_mount_flag(sb, EXT4_MF_FC_INELIGIBLE);
ext4_clear_mount_flag(sb, EXT4_MF_FC_COMMITTING);
+ sbi->s_fc_ineligible_tid = 0;
spin_lock_init(&sbi->s_fc_lock);
memset(&sbi->s_fc_stats, 0, sizeof(sbi->s_fc_stats));
sbi->s_fc_replay_state.fc_regions = NULL;
--- a/fs/ext4/xattr.c
+++ b/fs/ext4/xattr.c
@@ -2408,7 +2408,7 @@ retry_inode:
if (IS_SYNC(inode))
ext4_handle_sync(handle);
}
- ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_XATTR);
+ ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_XATTR, handle);
cleanup:
brelse(is.iloc.bh);
@@ -2486,7 +2486,7 @@ retry:
if (error == 0)
error = error2;
}
- ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_XATTR);
+ ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_XATTR, NULL);
return error;
}
@@ -2920,7 +2920,7 @@ int ext4_xattr_delete_inode(handle_t *ha
error);
goto cleanup;
}
- ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_XATTR);
+ ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_XATTR, handle);
}
error = 0;
cleanup:
--- a/fs/jbd2/commit.c
+++ b/fs/jbd2/commit.c
@@ -1170,7 +1170,7 @@ restart_loop:
if (journal->j_commit_callback)
journal->j_commit_callback(journal, commit_transaction);
if (journal->j_fc_cleanup_callback)
- journal->j_fc_cleanup_callback(journal, 1);
+ journal->j_fc_cleanup_callback(journal, 1, commit_transaction->t_tid);
trace_jbd2_end_commit(journal, commit_transaction);
jbd_debug(1, "JBD2: commit %d complete, head %d\n",
--- a/fs/jbd2/journal.c
+++ b/fs/jbd2/journal.c
@@ -771,7 +771,7 @@ static int __jbd2_fc_end_commit(journal_
{
jbd2_journal_unlock_updates(journal);
if (journal->j_fc_cleanup_callback)
- journal->j_fc_cleanup_callback(journal, 0);
+ journal->j_fc_cleanup_callback(journal, 0, tid);
write_lock(&journal->j_state_lock);
journal->j_flags &= ~JBD2_FAST_COMMIT_ONGOING;
if (fallback)
--- a/include/linux/jbd2.h
+++ b/include/linux/jbd2.h
@@ -1295,7 +1295,7 @@ struct journal_s
* Clean-up after fast commit or full commit. JBD2 calls this function
* after every commit operation.
*/
- void (*j_fc_cleanup_callback)(struct journal_s *journal, int);
+ void (*j_fc_cleanup_callback)(struct journal_s *journal, int full, tid_t tid);
/**
* @j_fc_replay_callback: