[lustre-devel] [PATCH 09/15] lustre: obdclass: Wake up entire queue of requests on close completion

[James Simmons <jsimmons@infradead.org>]

From: Oleg Drokin <green@whamcloud.com>

Since close requests could be stuck behind normal requests and get
more slots we need to wake up entire accumulated queue waiting
for the next modrpc slot or have additional waitqueue just for
close requests.

This patch goes with the former approach.

Fixes: 7cb15d0448 ("staging: lustre: mdc: manage number of modify RPCs in flight")
WC-bug-id: https://jira.whamcloud.com/browse/LU-10948
Lustre-commit: a4e1567d67559b797 ("LU-14741 obdclass: Wake up entire queue of requests on close completion")
Signed-off-by: Oleg Drokin <green@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/43941
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Neil Brown <neilb@suse.de>
Signed-off-by: James Simmons <jsimmons@infradead.org>
---
 fs/lustre/obdclass/genops.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/fs/lustre/obdclass/genops.c b/fs/lustre/obdclass/genops.c
index bbb63b2..4e89e0a 100644
--- a/fs/lustre/obdclass/genops.c
+++ b/fs/lustre/obdclass/genops.c
@@ -1587,6 +1587,10 @@ void obd_put_mod_rpc_slot(struct client_obd *cli, u32 opc, u16 tag)
 	LASSERT(tag - 1 < OBD_MAX_RIF_MAX);
 	LASSERT(test_and_clear_bit(tag - 1, cli->cl_mod_tag_bitmap) != 0);
 	spin_unlock(&cli->cl_mod_rpcs_lock);
-	wake_up(&cli->cl_mod_rpcs_waitq);
+	/* LU-14741 - to prevent close RPCs stuck behind normal ones */
+	if (close_req)
+		wake_up_all(&cli->cl_mod_rpcs_waitq);
+	else
+		wake_up(&cli->cl_mod_rpcs_waitq);
 }
 EXPORT_SYMBOL(obd_put_mod_rpc_slot);
-- 
1.8.3.1

_______________________________________________
lustre-devel mailing list
lustre-devel@lists.lustre.org
http://lists.lustre.org/listinfo.cgi/lustre-devel-lustre.org