public inbox for linux-rdma@vger.kernel.org
 help / color / mirror / Atom feed
From: Edward Srouji <edwards@nvidia.com>
To: Leon Romanovsky <leon@kernel.org>, Jason Gunthorpe <jgg@ziepe.ca>,
	"Chiara Meiohas" <cmeiohas@nvidia.com>,
	Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>,
	Gal Pressman <galpress@amazon.com>,
	Mark Bloch <markb@mellanox.com>,
	Steve Wise <larrystevenwise@gmail.com>,
	Mark Zhang <markzhang@nvidia.com>,
	"Neta Ostrovsky" <netao@nvidia.com>,
	Patrisious Haddad <phaddad@nvidia.com>,
	"Doug Ledford" <dledford@redhat.com>,
	Matan Barak <matanb@mellanox.com>, <majd@mellanox.com>,
	Maor Gottlieb <maorg@mellanox.com>
Cc: <linux-rdma@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
	"Edward Srouji" <edwards@nvidia.com>,
	Michael Guralnik <michaelgur@nvidia.com>
Subject: [PATCH rdma-next v2 05/11] RDMA/core: Fix potential use after free in ib_destroy_cq_user()
Date: Mon, 6 Apr 2026 12:11:16 +0300	[thread overview]
Message-ID: <20260406-security-bug-fixes-v2-5-ee8815fa81b7@nvidia.com> (raw)
In-Reply-To: <20260406-security-bug-fixes-v2-0-ee8815fa81b7@nvidia.com>

From: Patrisious Haddad <phaddad@nvidia.com>

When accessing a CQ via the netlink path the only synchronization
mechanism for the said CQ is rdma_restrack_get().
Currently, rdma_restrack_del() is invoked at the end of
ib_destroy_cq_user(), which is too late, since by that point
vendor-specific resources associated with the CQ might already be
freed. This can leave a short window where the CQ remains accessible
through restrack, leading to a potential use-after-free.

Fix this by moving the rdma_restrack_del() call to the start of
ib_destroy_cq_user(), ensuring that the CQ is removed from restrack
before its internal resources are released. This guarantees that no new
users hold references to a CQ that is in the process of destruction.

In addition, this change preserves the intended asymmetric behavior
between create and destroy routines: resources are added to the
restrack at the end of successful creation, and hence shall be removed
from the restrack first thing during the destruction flow, which keeps
the lifecycle management consistent and predictable.

Fixes: 08f294a1524b ("RDMA/core: Add resource tracking for create and destroy CQs")
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Reviewed-by: Michael Guralnik <michaelgur@nvidia.com>
Signed-off-by: Edward Srouji <edwards@nvidia.com>
---
 drivers/infiniband/core/verbs.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c
index f1438d5802a3e97e22cdb607cf90a097d041a162..0e8f99807c7c0ce063ed0c1561f4ba42b485b69d 100644
--- a/drivers/infiniband/core/verbs.c
+++ b/drivers/infiniband/core/verbs.c
@@ -2256,12 +2256,16 @@ int ib_destroy_cq_user(struct ib_cq *cq, struct ib_udata *udata)
 	if (atomic_read(&cq->usecnt))
 		return -EBUSY;
 
+	rdma_restrack_del(&cq->res);
+
 	ret = cq->device->ops.destroy_cq(cq, udata);
-	if (ret)
+	if (ret) {
+		rdma_restrack_new(&cq->res, RDMA_RESTRACK_CQ);
+		rdma_restrack_add(&cq->res);
 		return ret;
+	}
 
 	ib_umem_release(cq->umem);
-	rdma_restrack_del(&cq->res);
 	kfree(cq);
 	return ret;
 }

-- 
2.49.0


  parent reply	other threads:[~2026-04-06  9:11 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-04-06  9:11 [PATCH rdma-next v2 00/11] RDMA: Stability and race condition fixes Edward Srouji
2026-04-06  9:11 ` [PATCH rdma-next v2 01/11] RDMA/mlx5: Remove DCT restrack tracking Edward Srouji
2026-04-06  9:11 ` [PATCH rdma-next v2 02/11] RDMA/mlx5: Remove raw RSS QP " Edward Srouji
2026-04-06  9:11 ` [PATCH rdma-next v2 03/11] RDMA/core: Preserve restrack resource ID on reinsertion Edward Srouji
2026-04-06 22:23   ` Jason Gunthorpe
2026-04-07  9:18     ` Patrisious Haddad
2026-04-07 14:29       ` Jason Gunthorpe
2026-04-06  9:11 ` [PATCH rdma-next v2 04/11] RDMA/core: Fix use after free in ib_query_qp() Edward Srouji
2026-04-06  9:11 ` Edward Srouji [this message]
2026-04-06  9:11 ` [PATCH rdma-next v2 06/11] RDMA/core: Fix potential use after free in ib_destroy_srq_user() Edward Srouji
2026-04-06  9:11 ` [PATCH rdma-next v2 07/11] RDMA/mlx5: Fix UAF in SRQ destroy due to race with create Edward Srouji
2026-04-06  9:11 ` [PATCH rdma-next v2 08/11] RDMA/mlx5: Fix UAF in DCT " Edward Srouji
2026-04-06  9:11 ` [PATCH rdma-next v2 09/11] IB/core: Fix IPv6 netlink message size in ib_nl_ip_send_msg() Edward Srouji
2026-04-06  9:11 ` [PATCH rdma-next v2 10/11] RDMA/core: Fix rereg_mr use-after-free race Edward Srouji
2026-04-06  9:11 ` [PATCH rdma-next v2 11/11] RDMA/mlx5: Fix null-ptr-deref in Raw Packet QP creation Edward Srouji

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=20260406-security-bug-fixes-v2-5-ee8815fa81b7@nvidia.com \
    --to=edwards@nvidia.com \
    --cc=cmeiohas@nvidia.com \
    --cc=dennis.dalessandro@cornelisnetworks.com \
    --cc=dledford@redhat.com \
    --cc=galpress@amazon.com \
    --cc=jgg@ziepe.ca \
    --cc=larrystevenwise@gmail.com \
    --cc=leon@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=majd@mellanox.com \
    --cc=maorg@mellanox.com \
    --cc=markb@mellanox.com \
    --cc=markzhang@nvidia.com \
    --cc=matanb@mellanox.com \
    --cc=michaelgur@nvidia.com \
    --cc=netao@nvidia.com \
    --cc=phaddad@nvidia.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox