Linux RDMA and InfiniBand development
 help / color / mirror / Atom feed
From: Artem Dinaburg <artem@trailofbits.com>
To: stable@vger.kernel.org
Cc: andrew+netdev@lunn.ch, davem@davemloft.net, edumazet@kernel.org,
	kuba@kernel.org, leon@kernel.org, linux-kernel@vger.kernel.org,
	linux-rdma@vger.kernel.org, mbloch@nvidia.com,
	netdev@vger.kernel.org, pabeni@redhat.com, saeedm@nvidia.com,
	tariqt@nvidia.com
Subject: [PATCH 6.6.y 1/2] net/mlx5e: Don't store mlx5e_priv in mlx5e_dev devlink priv
Date: Wed, 14 Oct 2026 15:17:36 -0400	[thread overview]
Message-ID: <20261014191738.24034-2-artem@trailofbits.com> (raw)
In-Reply-To: <20261014191738.24034-1-artem@trailofbits.com>

From: Saeed Mahameed <saeedm@nvidia.com>

[ Upstream commit 123eda2e5b1638e298e3a66bb1e64a8da92de5e1 ]

mlx5e_priv is an unstable structure that can be memset(0) if profile
attaching fails, mlx5e_priv in mlx5e_dev devlink private is used to
reference the netdev and mdev associated with that struct. Instead,
store netdev directly into mlx5e_dev and get mdev from the containing
mlx5_adev aux device structure.

This fixes a kernel oops in mlx5e_remove when switchdev mode fails due
to change profile failure.

$ devlink dev eswitch set pci/0000:00:03.0 mode switchdev
Error: mlx5_core: Failed setting eswitch to offloads.
dmesg:
workqueue: Failed to create a rescuer kthread for wq "mlx5e": -EINTR
mlx5_core 0012:03:00.1: mlx5e_netdev_init_profile:6214:(pid 37199): mlx5e_priv_init failed, err=-12
mlx5_core 0012:03:00.1 gpu3rdma1: mlx5e_netdev_change_profile: new profile init failed, -12
workqueue: Failed to create a rescuer kthread for wq "mlx5e": -EINTR
mlx5_core 0012:03:00.1: mlx5e_netdev_init_profile:6214:(pid 37199): mlx5e_priv_init failed, err=-12
mlx5_core 0012:03:00.1 gpu3rdma1: mlx5e_netdev_change_profile: failed to rollback to orig profile, -12

$ devlink dev reload pci/0000:00:03.0 ==> oops

BUG: kernel NULL pointer dereference, address: 0000000000000520
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
PGD 0 P4D 0
Oops: Oops: 0000 [#1] SMP NOPTI
CPU: 3 UID: 0 PID: 521 Comm: devlink Not tainted 6.18.0-rc5+ #117 PREEMPT(voluntary)
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-2.fc40 04/01/2014
RIP: 0010:mlx5e_remove+0x68/0x130
RSP: 0018:ffffc900034838f0 EFLAGS: 00010246
RAX: ffff88810283c380 RBX: ffff888101874400 RCX: ffffffff826ffc45
RDX: 0000000000000000 RSI: 0000000000000001 RDI: 0000000000000000
RBP: ffff888102d789c0 R08: ffff8881007137f0 R09: ffff888100264e10
R10: ffffc90003483898 R11: ffffc900034838a0 R12: ffff888100d261a0
R13: ffff888100d261a0 R14: ffff8881018749a0 R15: ffff888101874400
FS:  00007f8565fea740(0000) GS:ffff88856a759000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000520 CR3: 000000010b11a004 CR4: 0000000000370ef0
Call Trace:
 <TASK>
 device_release_driver_internal+0x19c/0x200
 bus_remove_device+0xc6/0x130
 device_del+0x160/0x3d0
 ? devl_param_driverinit_value_get+0x2d/0x90
 mlx5_detach_device+0x89/0xe0
 mlx5_unload_one_devl_locked+0x3a/0x70
 mlx5_devlink_reload_down+0xc8/0x220
 devlink_reload+0x7d/0x260
 devlink_nl_reload_doit+0x45b/0x5a0
 genl_family_rcv_msg_doit+0xe8/0x140

Fixes: ee75f1fc44dd ("net/mlx5e: Create separate devlink instance for ethernet auxiliary device")
Fixes: c4d7eb57687f ("net/mxl5e: Add change profile method")
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Link: https://patch.msgid.link/20260108212657.25090-3-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Assisted-by: LLM
[ Backport to 6.6.y: Derive mdev from the auxiliary device in the
  older remove path. The auxiliary device only binds after successful
  netdev registration, and the target uplink profile-switch paths
  retain that registration, so remove performs one unregister without
  upstream's NETREG_REGISTERED conditional. Use that hardware owner
  when clearing the uplink pointer if profile rollback clears priv.
  Apply after the pending 6.6.y backport of 3d5918477f94 (net/mlx5e:
  Fix netif state handling), which makes probe registration-failure
  unwind release resources even before netif_device_attach(). Patch 2
  of this series ("net/mlx5e: Pass netdev to mlx5e_destroy_netdev instead
  of priv") uses netdev-based destruction. ]
Signed-off-by: Artem Dinaburg <artem@trailofbits.com>
---
Hi,

Please consider this backport of the fix for CVE-2026-22996 to 6.6.y.

The code changes follow upstream, except the older remove path derives
mdev from the auxiliary device and omits the newer registration-state
branches; see the backport note.

The complete fix is already present in 6.12.y, 6.18.y and 7.2.y.

Build-tested only as part of a cumulative 6.6.y series on x86_64
(allmodconfig, CONFIG_WERROR=y), changed objects only; no runtime testing.

Thanks,
Artem Dinaburg

 drivers/net/ethernet/mellanox/mlx5/core/en.h      |    2 +-
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c |   22 ++++++++++++---------
 2 files changed, 14 insertions(+), 10 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -947,3 +947,3 @@
 struct mlx5e_dev {
-	struct mlx5e_priv *priv;
+	struct net_device *netdev;
 	struct devlink_port dl_port;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -5998,4 +5998,4 @@
 	struct mlx5e_dev *mlx5e_dev = auxiliary_get_drvdata(adev);
-	struct mlx5e_priv *priv = mlx5e_dev->priv;
-	struct net_device *netdev = priv->netdev;
+	struct mlx5e_priv *priv = netdev_priv(mlx5e_dev->netdev);
+	struct net_device *netdev = mlx5e_dev->netdev;
 	struct mlx5_core_dev *mdev = edev->mdev;
@@ -6022,6 +6022,7 @@
 {
+	struct mlx5_adev *edev = container_of(adev, struct mlx5_adev, adev);
 	struct mlx5e_dev *mlx5e_dev = auxiliary_get_drvdata(adev);
-	struct mlx5e_priv *priv = mlx5e_dev->priv;
-	struct net_device *netdev = priv->netdev;
-	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_priv *priv = netdev_priv(mlx5e_dev->netdev);
+	struct net_device *netdev = mlx5e_dev->netdev;
+	struct mlx5_core_dev *mdev = edev->mdev;
 
@@ -6072,2 +6073,3 @@
 	SET_NETDEV_DEVLINK_PORT(netdev, &mlx5e_dev->dl_port);
+	mlx5e_dev->netdev = netdev;
 
@@ -6076,3 +6078,2 @@
 	priv = netdev_priv(netdev);
-	mlx5e_dev->priv = priv;
 
@@ -6127,9 +6128,12 @@
 {
+	struct mlx5_adev *edev = container_of(adev, struct mlx5_adev, adev);
 	struct mlx5e_dev *mlx5e_dev = auxiliary_get_drvdata(adev);
-	struct mlx5e_priv *priv = mlx5e_dev->priv;
+	struct net_device *netdev = mlx5e_dev->netdev;
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = edev->mdev;
 	pm_message_t state = {};
 
-	mlx5_core_uplink_netdev_set(priv->mdev, NULL);
+	mlx5_core_uplink_netdev_set(mdev, NULL);
 	mlx5e_dcbnl_delete_app(priv);
-	unregister_netdev(priv->netdev);
+	unregister_netdev(netdev);
 	mlx5e_suspend(adev, state);

  reply	other threads:[~2026-10-14 19:17 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-14 19:17 [PATCH 6.6.y 0/2] CVE-2026-22996, CVE-2026-23035: backport fix series Artem Dinaburg
2026-10-14 19:17 ` Artem Dinaburg [this message]
2026-10-14 19:34   ` [PATCH 6.6.y 1/2] net/mlx5e: Don't store mlx5e_priv in mlx5e_dev devlink priv sashiko-bot
2026-10-14 19:17 ` [PATCH 6.6.y 2/2] net/mlx5e: Pass netdev to mlx5e_destroy_netdev instead of priv Artem Dinaburg
2026-10-14 19:34   ` sashiko-bot

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=20261014191738.24034-2-artem@trailofbits.com \
    --to=artem@trailofbits.com \
    --cc=andrew+netdev@lunn.ch \
    --cc=davem@davemloft.net \
    --cc=edumazet@kernel.org \
    --cc=kuba@kernel.org \
    --cc=leon@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=mbloch@nvidia.com \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=saeedm@nvidia.com \
    --cc=stable@vger.kernel.org \
    --cc=tariqt@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