diff options
| author | Benjamin Marzinski <bmarzins@redhat.com> | 2008-03-14 13:52:52 -0500 |
|---|---|---|
| committer | Steven Whitehouse <swhiteho@redhat.com> | 2008-03-31 10:41:44 +0100 |
| commit | 58e9fee13e579df44922172dbe3c9e3ba3edf7a3 (patch) | |
| tree | 7b134f28032a3cd498b2d9a18c02ea085c0725e2 /fs/gfs2/locking/dlm/thread.c | |
| parent | f5a8cd020173c455705fc0095b7d299da6f8f87b (diff) | |
[GFS2] Invalidate cache at correct point
GFS2 wasn't invalidating its cache before it called into the lock manager
with a request that could potentially drop a lock. This was leaving a
window where the lock could be actually be held by another node, but the
file's page cache would still appear valid, causing coherency problems.
This patch moves the cache invalidation to before the lock manager call
when dropping a lock. It also adds the option to the lock_dlm lock
manager to not use conversion mode deadlock avoidance, which, on a
conversion from shared to exclusive, could internally drop the lock, and
then reacquire in. GFS2 now asks lock_dlm to not do this. Instead, GFS2
manually drops the lock and reacquires it.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Diffstat (limited to 'fs/gfs2/locking/dlm/thread.c')
| -rw-r--r-- | fs/gfs2/locking/dlm/thread.c | 10 |
1 files changed, 9 insertions, 1 deletions
diff --git a/fs/gfs2/locking/dlm/thread.c b/fs/gfs2/locking/dlm/thread.c index 521694fc19d..e53db6fd28a 100644 --- a/fs/gfs2/locking/dlm/thread.c +++ b/fs/gfs2/locking/dlm/thread.c @@ -135,7 +135,15 @@ static void process_complete(struct gdlm_lock *lp) lp->lksb.sb_status, lp->lockname.ln_type, (unsigned long long)lp->lockname.ln_number, lp->flags); - return; + if (lp->lksb.sb_status == -EDEADLOCK && + lp->ls->fsflags & LM_MFLAG_CONV_NODROP) { + lp->req = lp->cur; + acb.lc_ret |= LM_OUT_CONV_DEADLK; + if (lp->cur == DLM_LOCK_IV) + lp->lksb.sb_lkid = 0; + goto out; + } else + return; } /* |