Merge tag 'drm-intel-next-2012-06-21-merged' of git://people.freedesktop.org/~danvet/drm-intel into drm-core-next

Daniel writes:
 New -next pull request. Highlights:
 - Remaining vlv patches from Jesse et al.
 - Some hw workarounds from Jesse
 - hw context support from Ben
 - full uncore sharing on ivb
 - prep work to move the gtt code from intel-gtt.c to drm/i915 for gen6+
 - some backlight code improvements
 - leftovers for the timeout ioctl (we've forgotten the getparam)
 - ibx transcoder workarounds
 - some smaller fixlets and improvements
 - the new version of the "dont rely on HPD exclusively for VGA" patch

 Wrt regressions QA reported quite a few this time around.
 - The piglit/kernel-test fallout all has patches that are just awaiting
   review and merging into the next -next cycle.
 - Which just leaves a bunch of bugs about new modelines that don't work.
   It looks like these are all due to the new 16:9/16:10 modes in 3.5
   (yeah, only in this manual testing cycle did the git branch used by QA
   contain a backmerge of mainline with these patches).  Although I haven't
   yet confirmed this by letting our QA test the revert of that series.
 - Wrt bugs in general I'm trying to fight down some of our long-standing
   backlight issues (not regressions), but this seems to be a game of
   "you move, you lose" ... :("

Dropped merge bits since this had an -rc4 merge in it to fix some ugly
conflicts.

1 files changed, 16 insertions, 16 deletions

diff --git a/fs/xfs/xfs_sync.c b/fs/xfs/xfs_sync.c
index c9d3409c5ca..1e9ee064dbb 100644
--- a/fs/xfs/xfs_sync.c
+++ b/fs/xfs/xfs_sync.c
@@ -386,23 +386,23 @@ xfs_sync_worker(
 	 * We shouldn't write/force the log if we are in the mount/unmount
 	 * process or on a read only filesystem. The workqueue still needs to be
 	 * active in both cases, however, because it is used for inode reclaim
-	 * during these times.  Use the s_umount semaphore to provide exclusion
-	 * with unmount.
+	 * during these times.  Use the MS_ACTIVE flag to avoid doing anything
+	 * during mount.  Doing work during unmount is avoided by calling
+	 * cancel_delayed_work_sync on this work queue before tearing down
+	 * the ail and the log in xfs_log_unmount.
 	 */
-	if (down_read_trylock(&mp->m_super->s_umount)) {
-		if (!(mp->m_flags & XFS_MOUNT_RDONLY)) {
-			/* dgc: errors ignored here */
-			if (mp->m_super->s_frozen == SB_UNFROZEN &&
-			    xfs_log_need_covered(mp))
-				error = xfs_fs_log_dummy(mp);
-			else
-				xfs_log_force(mp, 0);
-
-			/* start pushing all the metadata that is currently
-			 * dirty */
-			xfs_ail_push_all(mp->m_ail);
-		}
-		up_read(&mp->m_super->s_umount);
+	if (!(mp->m_super->s_flags & MS_ACTIVE) &&
+	    !(mp->m_flags & XFS_MOUNT_RDONLY)) {
+		/* dgc: errors ignored here */
+		if (mp->m_super->s_frozen == SB_UNFROZEN &&
+		    xfs_log_need_covered(mp))
+			error = xfs_fs_log_dummy(mp);
+		else
+			xfs_log_force(mp, 0);
+
+		/* start pushing all the metadata that is currently
+		 * dirty */
+		xfs_ail_push_all(mp->m_ail);
 	}
 
 	/* queue us up again */