Re: [PATCH] xfs: call xfs_idestroy_fork() in xfs_ilock() critical section

From: Waiman Long
Date: Wed Apr 22 2015 - 16:28:46 EST


On 04/22/2015 03:11 PM, Brian Foster wrote:
On Wed, Apr 22, 2015 at 01:33:41PM -0400, Waiman Long wrote:
The commit f7be2d7f594cbc ("xfs: push down inactive transaction
mgmt for truncate") refactored the xfs_inactive() function
in fs/xfs/xfs_inode.c. However, it also moved the call to
xfs_idestroy_fork() from inside the xfs_ilock() critical section to
outside. That was causing memory corruption and strange failures like
deferencing NULL pointers in some circumstances.

This patch moves the xfs_idestroy_fork() call back into an xfs_ilock()
critical section to avoid memory corruption problem.

Signed-off-by: Waiman Long<Waiman.Long@xxxxxx>
---
Interesting... so from your previous mail we have an inactive/reclaim
racing with an xfs_iflush_fork() of the attr fork, or something of that
nature? Is there a specific reproducer or is it some kind of stress
test?

Good catch in any case, it looks like a deviation from the previous
code...

I am not sure what kind of races are going on. I was running the AIM7 workload for performance comparison purpose. I hit the error when running the disk workload with xfs filesystem. The smaller the ramdisk that I used, the easier it was to reproduce the error. I think I haven't run it for quite a while so I did not notice any problem or I might have just ignored it in some previous runs.

I did check some other call sites of xfs_idestroy_fork() and they are under xfs_ilock(). So I suppose it is not safe to call it outside of the critical section. This patch did indeed fix the problem that I saw when running the disk workload.

Cheers,
Longman


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/