[lustre-devel] [PATCH 40/50] lustre: osc: osc_extent_wait() deadlock

[James Simmons]

From: Andriy Skulysh <andriy.skulysh at hpe.com>

Thread 1:
vvp_io_write_commit
osc_io_commit_async
osc_page_cache_add
osc_extent_find
osc_extent_wait

Thread 2:
ptlrpcd_check
ptlrpc_check_set
brw_queue_work
osc_extent_make_ready
vvp_page_make_ready_start
__lock_page

We must not hold a page lock while we do osc_extent_find()

HPE-bug-id: LUS-10414
Fixes: f8a5fb036a ("lustre: vvp: dirty pages with pagevec")
WC-bug-id: https://jira.whamcloud.com/browse/LU-15477
Lustre-commit: 821a8d7b481d34a54 ("LU-15477 osc: osc_extent_wait() deadlock")
Signed-off-by: Andriy Skulysh <andriy.skulysh at hpe.com>
Reviewed-on: https://review.whamcloud.com/46281
Reviewed-by: Patrick Farrell <pfarrell at whamcloud.com>
Reviewed-by: Alexander Boyko <alexander.boyko at hpe.com>
Reviewed-by: Oleg Drokin <green at whamcloud.com>
Signed-off-by: James Simmons <jsimmons at infradead.org>
---
 fs/lustre/osc/osc_cache.c | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)

diff --git a/fs/lustre/osc/osc_cache.c b/fs/lustre/osc/osc_cache.c
index 7bd28c5..072cfac 100644
--- a/fs/lustre/osc/osc_cache.c
+++ b/fs/lustre/osc/osc_cache.c
@@ -2487,16 +2487,17 @@ int osc_queue_async_io(const struct lu_env *env, struct cl_io *io,
 		LASSERT(ergo(grants > 0, grants >= tmp));
 
 		rc = 0;
+
+		/* We must not hold a page lock while we do osc_enter_cache()
+		 * or osc_extent_find(), so we must mark dirty & unlock
+		 * any pages in the write commit pagevec.
+		 */
+		if (pagevec_count(pvec)) {
+			cb(env, io, pvec);
+			pagevec_reinit(pvec);
+		}
+
 		if (grants == 0) {
-			/* We haven't allocated grant for this page, and we
-			 * must not hold a page lock while we do enter_cache,
-			 * so we must mark dirty & unlock any pages in the
-			 * write commit pagevec.
-			 */
-			if (pagevec_count(pvec)) {
-				cb(env, io, pvec);
-				pagevec_reinit(pvec);
-			}
 			rc = osc_enter_cache(env, cli, oap, tmp);
 			if (rc == 0)
 				grants = tmp;
-- 
1.8.3.1