NULL mp->m_log in 2.6.31 xfs_log_move_tail

[Christoph Hellwig]

On Mon, Nov 09, 2009 at 03:39:32PM -0500, Ed Cashin wrote:
> A colleague has seen oopses in 2.6.31 when an XFS is mounted on an AoE
> target that becomes unresponsive and is marked as "down" by the aoe
> driver.  The aoe driver starts failing all new I/O requests after
> failing all current requests when the device is down.
> 
> I looked at the trace (included below) and put in the following check:

Given that you seem to be able to reproduce it can you see if the patch
below helps:

Index: linux-2.6/fs/xfs/xfs_log.c
===================================================================
--- linux-2.6.orig/fs/xfs/xfs_log.c	2009-11-09 22:09:08.858026060 +0100
+++ linux-2.6/fs/xfs/xfs_log.c	2009-11-09 22:13:13.958255857 +0100
@@ -1602,6 +1602,8 @@ xlog_dealloc_log(xlog_t *log)
 	xlog_in_core_t	*iclog, *next_iclog;
 	int		i;
 
+	xfs_flush_buftarg(log->l_mp->m_logdev_targp, 1);
+
 	iclog = log->l_iclog;
 	for (i=0; i<log->l_iclog_bufs; i++) {
 		sv_destroy(&iclog->ic_force_wait);