All of lore.kernel.org
 help / color / mirror / Atom feed
From: Leon Romanovsky <leon@kernel.org>
To: Doug Ledford <dledford@redhat.com>, Jason Gunthorpe <jgg@mellanox.com>
Cc: Artemy Kovalyov <artemyko@mellanox.com>,
	linux-rdma@vger.kernel.org, Yishai Hadas <yishaih@mellanox.com>
Subject: [PATCH rdma-next 4/9] RDMA/mlx5: Always remove MRs from the cache before destroying them
Date: Thu, 27 Feb 2020 14:33:55 +0200	[thread overview]
Message-ID: <20200227123400.97758-5-leon@kernel.org> (raw)
In-Reply-To: <20200227123400.97758-1-leon@kernel.org>

From: Jason Gunthorpe <jgg@mellanox.com>

The cache bucket tracks the total number of MRs that exists, both inside
and outside of the cache. Removing a MR from the cache (by setting
cache_ent to NULL) without updating total_mrs will cause the tracking to
leak and be inflated.

Further fix the rereg_mr path to always destroy the MR. reg_create will
always overwrite all the MR data in mlx5_ib_mr, so the MR must be
completely destroyed, in all cases, before this function can be
called. Detach the MR from the cache and unconditionally destroy it to
avoid leaking HW mkeys.

Fixes: afd1417404fb ("IB/mlx5: Use direct mkey destroy command upon UMR unreg failure")
Fixes: 56e11d628c5d ("IB/mlx5: Added support for re-registration of MRs")
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
---
 drivers/infiniband/hw/mlx5/mr.c | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index 68e5a3955b1d..c951bb14de56 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -464,6 +464,16 @@ static struct mlx5_ib_mr *alloc_cached_mr(struct mlx5_cache_ent *req_ent)
 	return mr;
 }

+static void detach_mr_from_cache(struct mlx5_ib_mr *mr)
+{
+	struct mlx5_cache_ent *ent = mr->cache_ent;
+
+	mr->cache_ent = NULL;
+	spin_lock_irq(&ent->lock);
+	ent->total_mrs--;
+	spin_unlock_irq(&ent->lock);
+}
+
 void mlx5_mr_cache_free(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
 {
 	struct mlx5_cache_ent *ent = mr->cache_ent;
@@ -473,7 +483,7 @@ void mlx5_mr_cache_free(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
 		return;

 	if (mlx5_mr_cache_invalidate(mr)) {
-		mr->cache_ent = NULL;
+		detach_mr_from_cache(mr);
 		destroy_mkey(dev, mr);
 		if (ent->available_mrs < ent->limit)
 			queue_work(dev->cache.wq, &ent->work);
@@ -1432,9 +1442,8 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
 		 * UMR can't be used - MKey needs to be replaced.
 		 */
 		if (mr->cache_ent)
-			err = mlx5_mr_cache_invalidate(mr);
-		else
-			err = destroy_mkey(dev, mr);
+			detach_mr_from_cache(mr);
+		err = destroy_mkey(dev, mr);
 		if (err)
 			goto err;

@@ -1446,8 +1455,6 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
 			mr = to_mmr(ib_mr);
 			goto err;
 		}
-
-		mr->cache_ent = NULL;
 	} else {
 		/*
 		 * Send a UMR WQE
--
2.24.1


  parent reply	other threads:[~2020-02-27 12:34 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-02-27 12:33 [PATCH rdma-next 0/9] MR cache fixes and refactoring Leon Romanovsky
2020-02-27 12:33 ` [PATCH mlx5-next 1/9] RDMA/mlx5: Move asynchronous mkey creation to mlx5_ib Leon Romanovsky
2020-02-27 19:41   ` Saeed Mahameed
2020-02-27 20:00     ` Jason Gunthorpe
2020-02-27 20:42       ` Saeed Mahameed
2020-02-27 12:33 ` [PATCH rdma-next 2/9] RDMA/mlx5: Rename the tracking variables for the MR cache Leon Romanovsky
2020-02-27 12:33 ` [PATCH rdma-next 3/9] RDMA/mlx5: Simplify how the MR cache bucket is located Leon Romanovsky
2020-02-27 12:33 ` Leon Romanovsky [this message]
2020-02-27 12:33 ` [PATCH rdma-next 5/9] RDMA/mlx5: Fix MR cache size and limit debugfs Leon Romanovsky
2020-02-27 12:33 ` [PATCH rdma-next 6/9] RDMA/mlx5: Lock access to ent->available_mrs/limit when doing queue_work Leon Romanovsky
2020-02-27 12:33 ` [PATCH rdma-next 7/9] RDMA/mlx5: Fix locking in MR cache work queue Leon Romanovsky
2020-02-27 12:33 ` [PATCH rdma-next 8/9] RDMA/mlx5: Revise how the hysteresis scheme works for cache filling Leon Romanovsky
2020-02-27 12:34 ` [PATCH rdma-next 9/9] RDMA/mlx5: Allow MRs to be created in the cache synchronously Leon Romanovsky

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200227123400.97758-5-leon@kernel.org \
    --to=leon@kernel.org \
    --cc=artemyko@mellanox.com \
    --cc=dledford@redhat.com \
    --cc=jgg@mellanox.com \
    --cc=linux-rdma@vger.kernel.org \
    --cc=yishaih@mellanox.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.