From e643692138cfa33528f054b071ba2583509bb217 Mon Sep 17 00:00:00 2001
From: Jan Kara <jack@suse.cz>
Date: Wed, 20 Mar 2013 14:39:05 +0100
Subject: [PATCH 1/4] ext3: fix data=journal fast mount/umount hang
In data=journal mode, if we unmount the file system before a
transaction has a chance to complete, when the journal inode is being
evicted, we can end up calling into log_wait_commit() for the
last transaction, after the journalling machinery has been shut down.
That triggers the WARN_ONCE in __log_start_commit().
Arguably we should adjust ext3_should_journal_data() to return FALSE
for the journal inode, but the only place it matters is
ext3_evict_inode(), and so it's to save a bit of CPU time, and to make
the patch much more obviously correct by inspection(tm), we'll fix it
by explicitly not trying to waiting for a journal commit when we are
evicting the journal inode, since it's guaranteed to never succeed in
this case.
This can be easily replicated via:
mount -t ext3 -o data=journal /dev/vdb /vdb ; umount /vdb
This is a port of ext4 fix from Ted Ts'o.
Signed-off-by: Jan Kara <jack@suse.cz>
---
fs/ext3/inode.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c
index d512c4bc4ad7..d706dbfa6220 100644
--- a/fs/ext3/inode.c
+++ b/fs/ext3/inode.c
@@ -218,7 +218,8 @@ void ext3_evict_inode (struct inode *inode)
*/
if (inode->i_nlink && ext3_should_journal_data(inode) &&
EXT3_SB(inode->i_sb)->s_journal &&
- (S_ISLNK(inode->i_mode) || S_ISREG(inode->i_mode))) {
+ (S_ISLNK(inode->i_mode) || S_ISREG(inode->i_mode)) &&
+ inode->i_ino != EXT3_JOURNAL_INO) {
tid_t commit_tid = atomic_read(&ei->i_datasync_tid);
journal_t *journal = EXT3_SB(inode->i_sb)->s_journal;
From e678a4f0f50d4fa4f7aaa6de8eb3f071513356a0 Mon Sep 17 00:00:00 2001
From: Jan Kara <jack@suse.cz>
Date: Wed, 27 Mar 2013 17:30:59 +0100
Subject: [PATCH 2/4] jbd: don't wait (forever) for stale tid caused by
wraparound
In the case where an inode has a very stale transaction id (tid) in
i_datasync_tid or i_sync_tid, it's possible that after a very large
(2**31) number of transactions, that the tid number space might wrap,
causing tid_geq()'s calculations to fail.
Commit d9b0193 "jbd: fix fsync() tid wraparound bug" attempted to fix
this problem, but it only avoided kjournald spinning forever by fixing
the logic in jbd_log_start_commit().
Signed-off-by: Jan Kara <jack@suse.cz>
---
fs/jbd/journal.c | 11 +++++++++++
1 file changed, 11 insertions(+)
diff --git a/fs/jbd/journal.c b/fs/jbd/journal.c
index 81cc7eaff863..81880c6d6b5e 100644
--- a/fs/jbd/journal.c
+++ b/fs/jbd/journal.c
@@ -564,6 +564,16 @@ int log_wait_commit(journal_t *journal, tid_t tid)
spin_unlock(&journal->j_state_lock);
#endif
spin_lock(&journal->j_state_lock);
+ /*
+ * Not running or committing trans? Must be already committed. This
+ * saves us from waiting for a *long* time when tid overflows.
+ */
+ if (!((journal->j_running_transaction &&
+ journal->j_running_transaction->t_tid == tid) ||
+ (journal->j_committing_transaction &&
+ journal->j_committing_transaction->t_tid == tid)))
+ goto out_unlock;
+
if (!tid_geq(journal->j_commit_waited, tid))
journal->j_commit_waited = tid;
while (tid_gt(tid, journal->j_commit_sequence)) {
@@ -575,6 +585,7 @@ int log_wait_commit(journal_t *journal, tid_t tid)
!tid_gt(tid, journal->j_commit_sequence));
spin_lock(&journal->j_state_lock);
}
+out_unlock:
spin_unlock(&journal->j_state_lock);
if (unlikely(is_journal_aborted(journal))) {
From e162b2f835eeda0d255bd463753b5eb823735205 Mon Sep 17 00:00:00 2001
From: Zheng Liu <wenqing.lz@taobao.com>
Date: Fri, 12 Apr 2013 02:24:06 +0000
Subject: [PATCH 3/4] jbd: use kmem_cache_zalloc instead of
kmem_cache_alloc/memset
Now jbd_alloc_handle is only called by new_handle. So this commit
uses kmem_cache_zalloc instead of kmem_cache_alloc/memset.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com>
Signed-off-by: Jan Kara <jack@suse.cz>
---
fs/jbd/transaction.c | 1 -
include/linux/jbd.h | 2 +-
2 files changed, 1 insertion(+), 2 deletions(-)
diff --git a/fs/jbd/transaction.c b/fs/jbd/transaction.c
index 071d6905f0dd..e3e255c0a509 100644
--- a/fs/jbd/transaction.c
+++ b/fs/jbd/transaction.c
@@ -245,7 +245,6 @@ static handle_t *new_handle(int nblocks)
handle_t *handle = jbd_alloc_handle(GFP_NOFS);
if (!handle)
return NULL;
- memset(handle, 0, sizeof(*handle));
handle->h_buffer_credits = nblocks;
handle->h_ref = 1;
diff --git a/include/linux/jbd.h b/include/linux/jbd.h
index c8f32975f0e4..7e0b622503c4 100644
--- a/include/linux/jbd.h
+++ b/include/linux/jbd.h
@@ -887,7 +887,7 @@ extern struct kmem_cache *jbd_handle_cache;
static inline handle_t *jbd_alloc_handle(gfp_t gfp_flags)
{
- return kmem_cache_alloc(jbd_handle_cache, gfp_flags);
+ return kmem_cache_zalloc(jbd_handle_cache, gfp_flags);
}
static inline void jbd_free_handle(handle_t *handle)
From 8bb9da943a62f4ffaabba1ad745990260fbb39df Mon Sep 17 00:00:00 2001
From: Zheng Liu <wenqing.lz@taobao.com>
Date: Mon, 29 Apr 2013 17:08:51 +0800
Subject: [PATCH 4/4] jbd: use kmem_cache_zalloc for allocating journal head
This commit tries to use kmem_cache_zalloc instead of kmem_cache_alloc/
memset when a new journal head is alloctated.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com>
Cc: Jan Kara <jack@suse.cz>
Signed-off-by: Jan Kara <jack@suse.cz>
---
fs/jbd/journal.c | 8 +++-----
1 file changed, 3 insertions(+), 5 deletions(-)
diff --git a/fs/jbd/journal.c b/fs/jbd/journal.c
index 81880c6d6b5e..726a4432cbb1 100644
--- a/fs/jbd/journal.c
+++ b/fs/jbd/journal.c
@@ -1856,7 +1856,7 @@ static struct journal_head *journal_alloc_journal_head(void)
#ifdef CONFIG_JBD_DEBUG
atomic_inc(&nr_journal_heads);
#endif
- ret = kmem_cache_alloc(journal_head_cache, GFP_NOFS);
+ ret = kmem_cache_zalloc(journal_head_cache, GFP_NOFS);
if (ret == NULL) {
jbd_debug(1, "out of memory for journal_head\n");
printk_ratelimited(KERN_NOTICE "ENOMEM in %s, retrying.\n",
@@ -1864,7 +1864,7 @@ static struct journal_head *journal_alloc_journal_head(void)
while (ret == NULL) {
yield();
- ret = kmem_cache_alloc(journal_head_cache, GFP_NOFS);
+ ret = kmem_cache_zalloc(journal_head_cache, GFP_NOFS);
}
}
return ret;
@@ -1926,10 +1926,8 @@ struct journal_head *journal_add_journal_head(struct buffer_head *bh)
struct journal_head *new_jh = NULL;
repeat:
- if (!buffer_jbd(bh)) {
+ if (!buffer_jbd(bh))
new_jh = journal_alloc_journal_head();
- memset(new_jh, 0, sizeof(*new_jh));
- }
jbd_lock_bh_journal_head(bh);
if (buffer_jbd(bh)) {