[PATCH 18/18] staging: lustre: osc: Lustre returns EINTR from writes when SA_RESTART is set

[James Simmons <jsimmons@infradead.org>]

When Lustre is in a read or write system call and receives a
SIGALRM, it will return EINTR if interrupted in osc_enter_cache.
This prevents the system call from being restarted if
SA_RESTART is set in the handler.

This patch changes behavior in this location to return ERESTARTSYS
when a signal arrives during the call to l_wait_event.

Signed-off-by: Patrick Farrell <paf@cray.com>
Signed-off-by: James Simmons <uja.ornl@gmail.com>
ntel-bug-id: https://jira.hpdd.intel.com/browse/LU-3581
Reviewed-on: http://review.whamcloud.com/7002
Reviewed-by: Rahul Deshmukh <rahul_deshmukh@xyratex.com>
Reviewed-by: Cory Spitz <spitzcor@cray.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: James Simmons <jsimmons@infradead.org>
---
 drivers/staging/lustre/lustre/osc/osc_cache.c |   13 ++++++++++++-
 1 files changed, 12 insertions(+), 1 deletions(-)

diff --git a/drivers/staging/lustre/lustre/osc/osc_cache.c b/drivers/staging/lustre/lustre/osc/osc_cache.c
index f09b89d..956d57b 100644
--- a/drivers/staging/lustre/lustre/osc/osc_cache.c
+++ b/drivers/staging/lustre/lustre/osc/osc_cache.c
@@ -1626,11 +1626,22 @@ static int osc_enter_cache(const struct lu_env *env, struct client_obd *cli,
 
 		/* l_wait_event is interrupted by signal, or timed out */
 		if (rc < 0) {
-			if (rc == -ETIMEDOUT) {
+			switch (rc) {
+			case -ETIMEDOUT:
 				OSC_DUMP_GRANT(D_ERROR, cli,
 					       "try to reserve %d.\n", bytes);
 				osc_extent_tree_dump(D_ERROR, osc);
 				rc = -EDQUOT;
+				break;
+			case -EINTR:
+				/* Ensures restartability - LU-3581 */
+				rc = -ERESTARTSYS;
+				break;
+			default:
+				CDEBUG(D_CACHE, "%s: event for cache space @ %p never arrived due to %d\n",
+				       cli->cl_import->imp_obd->obd_name,
+				       &ocw, rc);
+				break;
 			}
 			list_del_init(&ocw.ocw_entry);
 			goto out;
-- 
1.7.1