[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZBKjHwr8jPl4sBFl@x130>
Date: Wed, 15 Mar 2023 22:03:27 -0700
From: Saeed Mahameed <saeed@...nel.org>
To: Jakub Kicinski <kuba@...nel.org>
Cc: Lorenzo Bianconi <lorenzo@...nel.org>, netdev@...r.kernel.org,
bpf@...r.kernel.org, davem@...emloft.net, edumazet@...gle.com,
pabeni@...hat.com, ast@...nel.org, daniel@...earbox.net,
hawk@...nel.org, john.fastabend@...il.com, saeedm@...dia.com,
leon@...nel.org, shayagr@...zon.com, akiyano@...zon.com,
darinzon@...zon.com, sgoutham@...vell.com,
lorenzo.bianconi@...hat.com, toke@...hat.com, teknoraver@...a.com,
ttoukan.linux@...il.com
Subject: Re: [PATCH net v2 7/8] net/mlx5e: take into account device
reconfiguration for xdp_features flag
On 15 Mar 19:53, Jakub Kicinski wrote:
>On Wed, 15 Mar 2023 19:46:10 -0700 Saeed Mahameed wrote:
>> >I think so.. let me send a full patch.
>>
>> We have an internal version of a fix, Tariq is finalizing some review
>> comments and we will be posting it ASAP.
>
>Ah, I already posted. Does it look different?
>
Yes, completely different, Tariq's fix is in mlx5 only.
He splits the xdp feature setting into two functions,
one to initialize the netdev's xdp before registration,
and another one to update xpd features and call the notifier in the
"after" registration set_features flows.
I like our solution more, since it's more explicit and doesn't require
patching xdp stack because mlx5 abused xdp_set_features_flag, unless other
drivers have the same issue.
>https://patchwork.kernel.org/project/netdevbpf/patch/20230316002903.492497-1-kuba@kernel.org/
>
I don't see anything wrong with your patch though.. it also looks more
elegant, i dunno, I will let you decide, here's our patch:
https://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux.git/commit/?h=testing/xdp-features-fix&id=72e2266525948ba1498e6a3f2d63ea10d5ee86f5
>I wanted to make sure that it's ready for tomorrow's PR
Powered by blists - more mailing lists