]> Gentwo Git Trees - linux/.git/commit
net/mlx5: pagealloc: Fix reclaim race during command interface teardown
authorShay Drory <shayd@nvidia.com>
Sun, 28 Sep 2025 21:02:08 +0000 (00:02 +0300)
committerJakub Kicinski <kuba@kernel.org>
Tue, 30 Sep 2025 01:51:25 +0000 (18:51 -0700)
commit79a0e32b32ac4e4f9e4bb22be97f371c8c116c88
treedb7ee3ffcaa7e9317df3d1ad5f03d4f4aa0ab4f2
parentb1f0349bd6d320c382df2e7f6fc2ac95c85f2b18
net/mlx5: pagealloc: Fix reclaim race during command interface teardown

The reclaim_pages_cmd() function sends a command to the firmware to
reclaim pages if the command interface is active.

A race condition can occur if the command interface goes down (e.g., due
to a PCI error) while the mlx5_cmd_do() call is in flight. In this
case, mlx5_cmd_do() will return an error. The original code would
propagate this error immediately, bypassing the software-based page
reclamation logic that is supposed to run when the command interface is
down.

Fix this by checking whether mlx5_cmd_do() returns -ENXIO, which mark
that command interface is down. If this is the case, fall through to
the software reclamation path. If the command failed for any another
reason, or finished successfully, return as before.

Fixes: b898ce7bccf1 ("net/mlx5: cmdif, Avoid skipping reclaim pages if FW is not accessible")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c