Suspend has this annoying behavior in XFS where it freezes workqueues in some arbitrary order. This is a problem because the sync call can cause an AIL push, which may have to perform a RMW cycle on an inode cluster buffer if that buffer has been reclaimed. When this happens, the AIL issues a buffer read for which the io completion ends up on the xfs_buf workqueue. If /that/ workqueue has already been frozen, the suspend will timeout because we froze the workqueues in the wrong order. It seems suspicious to be freezing IO helper threads[1], so let's just not do that anymore. Prior to this patch, 4.10 fails to suspend on my laptop about 80% of the time; afterwards, it works every time. I've not done much suspend-and-crash testing on it though. [1] https://lwn.net/Articles/705269/ Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx> --- fs/xfs/xfs_super.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c index 7af5ca9..216ab89 100644 --- a/fs/xfs/xfs_super.c +++ b/fs/xfs/xfs_super.c @@ -842,7 +842,7 @@ xfs_init_mount_workqueues( struct xfs_mount *mp) { mp->m_buf_workqueue = alloc_workqueue("xfs-buf/%s", - WQ_MEM_RECLAIM|WQ_FREEZABLE, 1, mp->m_fsname); + WQ_MEM_RECLAIM, 1, mp->m_fsname); if (!mp->m_buf_workqueue) goto out; -- To unsubscribe from this list: send the line "unsubscribe linux-xfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
