ACK: [SRU][F][PATCH 0/1] net/mlx5: Avoid processing commands before cmdif is ready (LP: 1987287)

Philip Cox philip.cox at canonical.com
Thu Sep 1 19:23:46 UTC 2022


-- 
Acked-by: Philip Cox <philip.cox at canonical.com>


On Thu, 2022-09-01 at 18:53 +0200, frank.heimes at canonical.com wrote:
> BugLink: https://bugs.launchpad.net/bugs/1987287
> 
> SRU Justification:
> 
> [Impact] 
> 
>  * If the mlx5 driver is reloading while the recovery flow is
> happening,
>    and if it receives new commands before the command interface is up
>    again, this can lead to null pointer that tries to access non-
>    initialized command structures.
> 
>  * So it's required to avoid processing commands before the command
>    interface is up again.
> 
>  * This is accomplished by a new cmdif state that helps to avoid
>    processing commands while cmdif is not ready.
> 
> [Fix]
> 
>  * backport of f7936ddd35d8 f7936ddd35d8b849daf0372770c7c9dbe7910fca
> "net/mlx5: Avoid processing commands before cmdif is ready"
> 
> [Test Plan]
> 
>  * An Ubuntu Server for s390x 18.04 or 20.04 LPAR or z/VM
> installation
>    is needed that has Mellanox cards (RoCE Express 2.1) assigned, 
>    configured and enabled and that runs a 5.4 kernel (on bionic hwe-
> 5.4).
> 
>  * Now trigger a recovery (guess that can be done at the Support
> Element)
>    and reload the driver at the same time.
> 
>  * Make sure the module/driver mlx5 is loaded and in use
>    (otherwise it can't be removed/unloaded).
> 
>  * Now remove/unload the module with:
>    sudo modprobe -r mlx5
>    and (re-)load it again with:
>    sudo modprobe mlx5
> 
>  * Due to the lack of RoCE Express 2.1 hardware,
>    IBM needs to do the verification.
> 
> [Where problems could occur]
> 
>  * In case there is an issue with 'cmdif' it might not have the
> correct
>    interface state, which:
>    - either might lead to the fact that commands are not properly
> blocked
>      and the situation is similar like before
>    - or the commands may get always blocked,
>      which render the hardware useless
>    - or might block in wrong situation,
>      which will cause unexpected issues and broken behavior.
> 
>  * Since the patch got upstream accepted with v5.7-rc7 it's
>    not new to the kernel, was already part of groovy (and above)
>    and is therefor already in use by newer Ubuntu releases.
> 
> [Other Info]
>  
>  * Since the patch is upstream since v5.7-rc7,
>    it's already included in jammy and kinetic.
> 
>  * Since the upstream patch incl. the line:
>    Fixes: e126ba97dba9 ("mlx5: Add driver for Mellanox
>    Connect-IB adapters") it looks to me that it was forgotten
>    to mark the patch for upstream stable updates.
> 
>  * Such SRUs for focal's 5.4 will automatically land in bionic's
>    hwe-5.4, too. But since this was especially requested for
>    bionic's hwe-5.4, I wanted to mention this here.
> 
> Eran Ben Elisha (1):
>   net/mlx5: Avoid processing commands before cmdif is ready
> 
>  drivers/net/ethernet/mellanox/mlx5/core/cmd.c  | 10 ++++++++++
>  drivers/net/ethernet/mellanox/mlx5/core/main.c |  4 ++++
>  include/linux/mlx5/driver.h                    |  9 +++++++++
>  3 files changed, 23 insertions(+)
> 
> -- 
> 2.25.1
> 
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/kernel-team/attachments/20220901/da8098ea/attachment.html>


More information about the kernel-team mailing list