Re: 24.11.7 patches review and test

[Kevin Traynor <ktraynor@redhat.com>]

On 8/4/26 14:46, Ali Alnubani wrote:
> On 7/27/26 12:43 PM, luca.boccassi@gmail.com wrote:
> 
>> Hi all,
>>
>> Here is a list of patches targeted for stable release 24.11.7.The 
>> planned date for the final release is 2026/08/10.
>>
>> Please help with testing and validation of your use cases and report
>> any issues/results with reply-all to this mail. For the final release
>> the fixes and reported validations will be added to the release notes.
>>
>> A release candidate tarball can be found at:
>>
>>      https://dpdk.org/browse/dpdk-stable/tag/?id=v24.11.7-rc1
>>
>> These patches are located at branch 24.11 of dpdk-stable repo:
>>      https://dpdk.org/browse/dpdk-stable/
>>
>> Thanks.
>>
>> Luca Boccassi
>>
>> ---
> [..]
>> Maxime Peim (1):
>>        net/mlx5: prepend implicit items in sync flow creation
> [..]
> 
> Hello Maxime, Luca,
> 
> Commit 471350f6cd ("net/mlx5: prepend implicit items in sync flow 
> creation") on 24.11 regresses eSwitch HWS when representor matching is 
> disabled (repr_matching_en=0): traffic that should hit PF NIC Rx rules 
> after FDB transfer is no longer received.
> 
> The backport dropped the priv->sh->config.repr_matching checks that 
> previously gated implicit REPRESENTED_PORT / REG_C_0 prepending. That is 
> correct on upstream main after repr_matching_en was removed 
> (c41f621786), but on 24.11 the option still exists and, when off, TX 
> representor tagging is not set up. Unconditionally prepending those 
> matches then scopes NIC rules incorrectly.
> 

I think we should be ok for 23.11 as the commit that was being fixed was 
introduced in 24.07 so the backport is not relevant.

e909fd3a90 S F net/mlx5: prepend implicit items in sync flow creation 
(24.07)

and ok for 25.11 as c41f621786 is present.

$ git tag --contains c41f621786 | grep v25.11
v25.11


> Dariusz volunteered to send a fix.
> 
> We will send our full report soon.
> 
> Regards,
> Ali
>