RDMA/mlx5: Delay emptying a cache entry when a new MR is added to it …

…recently

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: b9358bd ("RDMA/mlx5: Fix locking in MR cache work queue")
Link: https://lore.kernel.org/r/fcb546986be346684a016f5ca23a0567399145fa.1627370131.git.leonro@nvidia.com
Signed-off-by: Aharon Landau <aharonl@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>

drivers/infiniband/hw/mlx5/mr.c

@@ -531,8 +531,8 @@ 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;