[Cluster-devel] [PATCH] gfs2: Remove dirty buffer warning from gfs2_releasepage

[Andreas Gruenbacher]

Unlike what its documentation suggests, the releasepage address space
operation can currently be called on dirty pages via shrink_active_list.
This may eventually be changed when the remaining code relying on the
current behavior has been fixed, but until then, it makes no sense to
warn on dirty buffers in gfs2_releasepage.

Signed-off-by: Andreas Gruenbacher <agruenba at redhat.com>
---
 fs/gfs2/aops.c | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)

diff --git a/fs/gfs2/aops.c b/fs/gfs2/aops.c
index 82df368..d17bf16 100644
--- a/fs/gfs2/aops.c
+++ b/fs/gfs2/aops.c
@@ -1147,6 +1147,16 @@ int gfs2_releasepage(struct page *page, gfp_t gfp_mask)
 	if (!page_has_buffers(page))
 		return 0;
 
+	/*
+	 * From xfs_vm_releasepage: mm accommodates an old ext3 case where
+	 * clean pages might not have had the dirty bit cleared.  Thus, it can
+	 * send actual dirty pages to ->releasepage() via shrink_active_list().
+	 *
+	 * As a workaround, we skip pages that contain dirty buffers below.
+	 * Once ->releasepage isn't called on dirty pages anymore, we can warn
+	 * on dirty buffers like we used to here again.
+	 */
+
 	gfs2_log_lock(sdp);
 	spin_lock(&sdp->sd_ail_lock);
 	head = bh = page_buffers(page);
@@ -1156,8 +1166,8 @@ int gfs2_releasepage(struct page *page, gfp_t gfp_mask)
 		bd = bh->b_private;
 		if (bd && bd->bd_tr)
 			goto cannot_release;
-		if (buffer_pinned(bh) || buffer_dirty(bh))
-			goto not_possible;
+		if (buffer_dirty(bh) || WARN_ON(buffer_pinned(bh)))
+			goto cannot_release;
 		bh = bh->b_this_page;
 	} while(bh != head);
 	spin_unlock(&sdp->sd_ail_lock);
@@ -1180,9 +1190,6 @@ int gfs2_releasepage(struct page *page, gfp_t gfp_mask)
 
 	return try_to_free_buffers(page);
 
-not_possible: /* Should never happen */
-	WARN_ON(buffer_dirty(bh));
-	WARN_ON(buffer_pinned(bh));
 cannot_release:
 	spin_unlock(&sdp->sd_ail_lock);
 	gfs2_log_unlock(sdp);
-- 
2.7.4