[PATCH 5.10 038/135] RDMA/mlx5: Delay emptying a cache entry when a new MR is added to it recently

From: Greg Kroah-Hartman
Date: Tue Aug 10 2021 - 13:46:25 EST


From: Aharon Landau <aharonl@xxxxxxxxxx>

[ Upstream commit d6793ca97b76642b77629dd0783ec64782a50bdb ]

Fixing a typo that causes a cache entry to shrink immediately after adding
to it new MRs if the entry size exceeds the high limit. In doing so, the
cache misses its purpose to prevent the creation of new mkeys on the
runtime by using the cached ones.

Fixes: b9358bdbc713 ("RDMA/mlx5: Fix locking in MR cache work queue")
Link: https://lore.kernel.org/r/fcb546986be346684a016f5ca23a0567399145fa.1627370131.git.leonro@xxxxxxxxxx
Signed-off-by: Aharon Landau <aharonl@xxxxxxxxxx>
Reviewed-by: Maor Gottlieb <maorg@xxxxxxxxxx>
Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxx>
Signed-off-by: Jason Gunthorpe <jgg@xxxxxxxxxx>
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
drivers/infiniband/hw/mlx5/mr.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index 971694e781b6..19346693c1da 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -526,8 +526,8 @@ static void __cache_work_func(struct mlx5_cache_ent *ent)
*/
spin_unlock_irq(&ent->lock);
need_delay = need_resched() || someone_adding(cache) ||
- time_after(jiffies,
- READ_ONCE(cache->last_add) + 300 * HZ);
+ !time_after(jiffies,
+ READ_ONCE(cache->last_add) + 300 * HZ);
spin_lock_irq(&ent->lock);
if (ent->disabled)
goto out;
--
2.30.2