From: Steven Whitehouse <swhiteho@redhat.com>
To: cluster-devel.redhat.com
Subject: [Cluster-devel] [PATCH 07/22] GFS2: Don't enforce min hold time when two demotes occur in rapid succession
Date: Mon, 18 Oct 2010 15:15:31 +0100 [thread overview]
Message-ID: <1287411346-17855-8-git-send-email-swhiteho@redhat.com> (raw)
In-Reply-To: <1287411346-17855-1-git-send-email-swhiteho@redhat.com>
Due to the design of the VFS, it is quite usual for operations on GFS2
to consist of a lookup (requiring a shared lock) followed by an
operation requiring an exclusive lock. If a remote node has cached an
exclusive lock, then it will receive two demote events in rapid succession
firstly for a shared lock and then to unlocked. The existing min hold time
code was triggering in this case, even if the node was otherwise idle
since the state change time was being updated by the initial demote.
This patch introduces logic to skip the min hold timer in the case that
a "double demote" of this kind has occurred. The min hold timer will
still be used in all other cases.
A new glock flag is introduced which is used to keep track of whether
there have been any newly queued holders since the last glock state
change. The min hold time is only applied if the flag is set.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Tested-by: Abhijith Das <adas@redhat.com>
diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index 9adf8f9..8e478e2 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -441,6 +441,8 @@ static void state_change(struct gfs2_glock *gl, unsigned int new_state)
else
gfs2_glock_put_nolock(gl);
}
+ if (held1 && held2 && list_empty(&gl->gl_holders))
+ clear_bit(GLF_QUEUED, &gl->gl_flags);
gl->gl_state = new_state;
gl->gl_tchange = jiffies;
@@ -1012,6 +1014,7 @@ fail:
if (unlikely((gh->gh_flags & LM_FLAG_PRIORITY) && !insert_pt))
insert_pt = &gh2->gh_list;
}
+ set_bit(GLF_QUEUED, &gl->gl_flags);
if (likely(insert_pt == NULL)) {
list_add_tail(&gh->gh_list, &gl->gl_holders);
if (unlikely(gh->gh_flags & LM_FLAG_PRIORITY))
@@ -1310,10 +1313,12 @@ void gfs2_glock_cb(struct gfs2_glock *gl, unsigned int state)
gfs2_glock_hold(gl);
holdtime = gl->gl_tchange + gl->gl_ops->go_min_hold_time;
- if (time_before(now, holdtime))
- delay = holdtime - now;
- if (test_bit(GLF_REPLY_PENDING, &gl->gl_flags))
- delay = gl->gl_ops->go_min_hold_time;
+ if (test_bit(GLF_QUEUED, &gl->gl_flags)) {
+ if (time_before(now, holdtime))
+ delay = holdtime - now;
+ if (test_bit(GLF_REPLY_PENDING, &gl->gl_flags))
+ delay = gl->gl_ops->go_min_hold_time;
+ }
spin_lock(&gl->gl_spin);
handle_callback(gl, state, delay);
@@ -1660,6 +1665,8 @@ static const char *gflags2str(char *buf, const unsigned long *gflags)
*p++ = 'I';
if (test_bit(GLF_FROZEN, gflags))
*p++ = 'F';
+ if (test_bit(GLF_QUEUED, gflags))
+ *p++ = 'q';
*p = 0;
return buf;
}
diff --git a/fs/gfs2/incore.h b/fs/gfs2/incore.h
index 578234b..b12ca10 100644
--- a/fs/gfs2/incore.h
+++ b/fs/gfs2/incore.h
@@ -196,6 +196,7 @@ enum {
GLF_REPLY_PENDING = 9,
GLF_INITIAL = 10,
GLF_FROZEN = 11,
+ GLF_QUEUED = 12,
};
struct gfs2_glock {
diff --git a/fs/gfs2/trace_gfs2.h b/fs/gfs2/trace_gfs2.h
index 148d55c..cedb0bb 100644
--- a/fs/gfs2/trace_gfs2.h
+++ b/fs/gfs2/trace_gfs2.h
@@ -39,7 +39,8 @@
{(1UL << GLF_INVALIDATE_IN_PROGRESS), "i" }, \
{(1UL << GLF_REPLY_PENDING), "r" }, \
{(1UL << GLF_INITIAL), "I" }, \
- {(1UL << GLF_FROZEN), "F" })
+ {(1UL << GLF_FROZEN), "F" }, \
+ {(1UL << GLF_QUEUED), "q" })
#ifndef NUMPTY
#define NUMPTY
--
1.7.1.1
next prev parent reply other threads:[~2010-10-18 14:15 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-10-18 14:15 [Cluster-devel] GFS2: Pre-pull patch posting Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 01/22] GFS2: New truncate sequence Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 02/22] GFS2: Remove i_disksize Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 03/22] GFS2: No longer experimental Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 04/22] GFS2: Add a bug trap in allocation code Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 05/22] GFS2: fallocate support Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 06/22] GFS2: Fix whitespace in previous patch Steven Whitehouse
2010-10-18 14:15 ` Steven Whitehouse [this message]
2010-10-18 14:15 ` [Cluster-devel] [PATCH 08/22] GFS2: Update handling of DLM return codes to match reality Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 09/22] GFS2: Use new workqueue scheme Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 10/22] GFS2: Make . and .. qstrs constant Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 11/22] GFS2: Remove ignore_local_fs mount argument Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 12/22] GFS2: Remove localcaching mount option Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 13/22] GFS2: Remove upgrade " Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 14/22] GFS2: Fix journal check for spectator mounts Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 15/22] GFS2: reserve more blocks for transactions Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 16/22] GFS2: Fix compiler warning from previous patch Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 17/22] GFS2: Fix spectator umount issue Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 18/22] GFS2: Add "norecovery" mount option as a synonym for "spectator" Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 19/22] GFS2: Improve journal allocation via sysfs Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 20/22] GFS2 fatal: filesystem consistency error on rename Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 21/22] GFS2: Fix type mapping for demote_rq interface Steven Whitehouse
2010-10-18 14:15 ` [Cluster-devel] [PATCH 22/22] GFS2: fixed typo Steven Whitehouse
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=1287411346-17855-8-git-send-email-swhiteho@redhat.com \
--to=swhiteho@redhat.com \
/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).