linux-ext4.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jan Kara <jack@suse.cz>
To: Ted Tso <tytso@mit.edu>
Cc: <linux-ext4@vger.kernel.org>,
	Adrian Hunter <adrian.hunter@intel.com>, Jan Kara <jack@suse.cz>
Subject: [PATCH] jbd2: Avoid long hold times of j_state_lock while committing a transaction
Date: Tue, 16 Oct 2018 11:42:48 +0200	[thread overview]
Message-ID: <20181016094248.21359-1-jack@suse.cz> (raw)

We can hold j_state_lock for writing at the beginning of
jbd2_journal_commit_transaction() for a rather long time (reportedly for
30 ms) due cleaning revoke bits of all revoked buffers under it. The
handling of revoke tables as well as cleaning of t_reserved_list, and
checkpoint lists does not need j_state_lock for anything. Furthermore
the transaction is in T_LOCKED state and we waited for all outstanding
handles so nobody is going to be adding anything to the transaction.

Just drop the lock for unnecessary operations.

Reported-and-tested-by: Adrian Hunter <adrian.hunter@intel.com>
Suggested-by: "Theodore Y. Ts'o" <tytso@mit.edu>
Signed-off-by: Jan Kara <jack@suse.cz>
---
 fs/jbd2/commit.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
index 150cc030b4d7..356b75fa3101 100644
--- a/fs/jbd2/commit.c
+++ b/fs/jbd2/commit.c
@@ -422,6 +422,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
 				       stats.run.rs_locked);
 	stats.run.rs_running = jbd2_time_diff(commit_transaction->t_start,
 					      stats.run.rs_locked);
+	write_unlock(&journal->j_state_lock);
 
 	spin_lock(&commit_transaction->t_handle_lock);
 	while (atomic_read(&commit_transaction->t_updates)) {
@@ -431,9 +432,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
 					TASK_UNINTERRUPTIBLE);
 		if (atomic_read(&commit_transaction->t_updates)) {
 			spin_unlock(&commit_transaction->t_handle_lock);
-			write_unlock(&journal->j_state_lock);
 			schedule();
-			write_lock(&journal->j_state_lock);
 			spin_lock(&commit_transaction->t_handle_lock);
 		}
 		finish_wait(&journal->j_wait_updates, &wait);
@@ -505,6 +504,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
 	atomic_sub(atomic_read(&journal->j_reserved_credits),
 		   &commit_transaction->t_outstanding_credits);
 
+	write_lock(&journal->j_state_lock);
 	trace_jbd2_commit_flushing(journal, commit_transaction);
 	stats.run.rs_flushing = jiffies;
 	stats.run.rs_locked = jbd2_time_diff(stats.run.rs_locked,
-- 
2.16.4

             reply	other threads:[~2018-10-16 17:32 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-10-16  9:42 Jan Kara [this message]
2018-10-30 14:44 ` [PATCH] jbd2: Avoid long hold times of j_state_lock while committing a transaction Theodore Y. Ts'o
2018-11-06  5:20 ` Theodore Y. Ts'o
2018-11-06 10:22   ` Jan Kara
2018-11-06 16:47     ` Theodore Y. Ts'o
2018-11-06 17:14       ` Hunter, Adrian
2018-11-06 17:21       ` Jan Kara
2018-11-08 12:30         ` Jan Kara

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20181016094248.21359-1-jack@suse.cz \
    --to=jack@suse.cz \
    --cc=adrian.hunter@intel.com \
    --cc=linux-ext4@vger.kernel.org \
    --cc=tytso@mit.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).