From be12b299a83fc807bbaccd2bcb8ec50cbb0cb55c Mon Sep 17 00:00:00 2001 From: Joseph Qi Date: Fri, 25 Mar 2016 14:21:29 -0700 Subject: ocfs2/dlm: fix BUG in dlm_move_lockres_to_recovery_list When master handles convert request, it queues ast first and then returns status. This may happen that the ast is sent before the request status because the above two messages are sent by two threads. And right after the ast is sent, if master down, it may trigger BUG in dlm_move_lockres_to_recovery_list in the requested node because ast handler moves it to grant list without clear lock->convert_pending. So remove BUG_ON statement and check if the ast is processed in dlmconvert_remote. Signed-off-by: Joseph Qi Reported-by: Yiwen Jiang Cc: Junxiao Bi Cc: Mark Fasheh Cc: Joel Becker Cc: Tariq Saeed Cc: Junxiao Bi Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/ocfs2/dlm/dlmconvert.c | 13 +++++++++++++ fs/ocfs2/dlm/dlmrecovery.c | 1 - 2 files changed, 13 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmconvert.c b/fs/ocfs2/dlm/dlmconvert.c index 84de55ed865a..f90931335c6b 100644 --- a/fs/ocfs2/dlm/dlmconvert.c +++ b/fs/ocfs2/dlm/dlmconvert.c @@ -288,6 +288,19 @@ enum dlm_status dlmconvert_remote(struct dlm_ctxt *dlm, status = DLM_DENIED; goto bail; } + + if (lock->ml.type == type && lock->ml.convert_type == LKM_IVMODE) { + mlog(0, "last convert request returned DLM_RECOVERING, but " + "owner has already queued and sent ast to me. res %.*s, " + "(cookie=%u:%llu, type=%d, conv=%d)\n", + res->lockname.len, res->lockname.name, + dlm_get_lock_cookie_node(be64_to_cpu(lock->ml.cookie)), + dlm_get_lock_cookie_seq(be64_to_cpu(lock->ml.cookie)), + lock->ml.type, lock->ml.convert_type); + status = DLM_NORMAL; + goto bail; + } + res->state |= DLM_LOCK_RES_IN_PROGRESS; /* move lock to local convert queue */ /* do not alter lock refcount. switching lists. */ diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c index cd38488a10fc..f6b313898763 100644 --- a/fs/ocfs2/dlm/dlmrecovery.c +++ b/fs/ocfs2/dlm/dlmrecovery.c @@ -2083,7 +2083,6 @@ void dlm_move_lockres_to_recovery_list(struct dlm_ctxt *dlm, dlm_lock_get(lock); if (lock->convert_pending) { /* move converting lock back to granted */ - BUG_ON(i != DLM_CONVERTING_LIST); mlog(0, "node died with convert pending " "on %.*s. move back to granted list.\n", res->lockname.len, res->lockname.name); -- cgit v1.2.3-59-g8ed1b