Attachment 672503 Details for Bug 888857 – patch

patch

0001-fenced-dlm_controld-gfs_controld-use-cluster_dead-fo.patch (text/plain), 2.18 KB, created by David Teigland on 2013-01-04 17:58:16 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: David Teigland

Created: 2013-01-04 17:58:16 UTC

Size: 2.18 KB

patch

obsolete

>From fde920ca5400a390064d59f1c5ba924e5d7e642d Mon Sep 17 00:00:00 2001
>From: David Teigland <teigland@redhat.com>
>Date: Fri, 4 Jan 2013 11:44:13 -0600
>Subject: [PATCH] fenced/dlm_controld/gfs_controld: use cluster_dead for
> corosync connections
>
>corosync connections are supposed to use the
>cluster_dead() function, not the client_dead()
>function.  The client_dead() function closes
>the fd directly, which in the case of corosync
>connections, fouls up the ipc, because the ipc
>is still expecting to use that fd, which the
>daemon has reused for something else.  The next
>time the daemon calls into corosync ipc, it gets
>stuck in a poll/futex loop as the lib attempts
>to use the fd that fenced has wrongly closed.
>
>bz 888857
>
>Signed-off-by: David Teigland <teigland@redhat.com>
>---
> fence/fenced/cpg.c           | 2 +-
> group/dlm_controld/cpg.c     | 2 +-
> group/gfs_controld/cpg-new.c | 2 +-
> 3 files changed, 3 insertions(+), 3 deletions(-)
>
>diff --git a/fence/fenced/cpg.c b/fence/fenced/cpg.c
>index 6634f8c..a5a4208 100644
>--- a/fence/fenced/cpg.c
>+++ b/fence/fenced/cpg.c
>@@ -1748,7 +1748,7 @@ int fd_join(struct fd *fd)
> 
> 	cpg_fd_get(cpg_handle_domain, &cpg_fd_domain);
> 
>-	ci = client_add(cpg_fd_domain, process_cpg_domain, NULL);
>+	ci = client_add(cpg_fd_domain, process_cpg_domain, cluster_dead);
> 
> 	list_add(&fd->list, &domains);
> 	fd->cpg_handle = cpg_handle_domain;
>diff --git a/group/dlm_controld/cpg.c b/group/dlm_controld/cpg.c
>index 4463f4d..6a4023b 100644
>--- a/group/dlm_controld/cpg.c
>+++ b/group/dlm_controld/cpg.c
>@@ -1822,7 +1822,7 @@ int dlm_join_lockspace(struct lockspace *ls)
> 
> 	cpg_fd_get(h, &fd);
> 
>-	ci = client_add(fd, process_cpg_lockspace, NULL);
>+	ci = client_add(fd, process_cpg_lockspace, cluster_dead);
> 
> 	list_add(&ls->list, &lockspaces);
> 
>diff --git a/group/gfs_controld/cpg-new.c b/group/gfs_controld/cpg-new.c
>index fd20250..08df0ab 100644
>--- a/group/gfs_controld/cpg-new.c
>+++ b/group/gfs_controld/cpg-new.c
>@@ -2761,7 +2761,7 @@ int gfs_join_mountgroup(struct mountgroup *mg)
> 
> 	cpg_fd_get(h, &fd);
> 
>-	ci = client_add(fd, process_cpg_mountgroup, NULL);
>+	ci = client_add(fd, process_cpg_mountgroup, cluster_dead);
> 
> 	mg->cpg_handle = h;
> 	mg->cpg_client = ci;
>-- 
>1.8.1.rc1.5.g7e0651a
>

Actions: View

Attachments on bug 888857: 672406 | 672501 | 672503