netdev - Re: [PATCH net-next] net: Restore napi threaded state only when it is enabled

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20250722183647.1fd15767@kernel.org>
Date: Tue, 22 Jul 2025 18:36:47 -0700
From: Jakub Kicinski <kuba@...nel.org>
To: Samiullah Khawaja <skhawaja@...gle.com>
Cc: "David S . Miller " <davem@...emloft.net>, Eric Dumazet
 <edumazet@...gle.com>, Paolo Abeni <pabeni@...hat.com>,
 almasrymina@...gle.com, willemb@...gle.com, netdev@...r.kernel.org
Subject: Re: [PATCH net-next] net: Restore napi threaded state only when it
 is enabled

On Mon, 21 Jul 2025 23:36:08 +0000 Samiullah Khawaja wrote:
> Commit 2677010e7793 ("Add support to set NAPI threaded for individual
> NAPI") added support to enable/disable threaded napi using netlink. This
> also extended the napi config save/restore functionality to set the napi
> threaded state. This breaks the netdev reset when threaded napi is

"This breaks the netdev reset" is very vague.

> enabled at device level as the napi_restore_config tries to stop the
> kthreads as napi->config->thread is false when threaded is enabled at
> device level.

My reading of the commit message is that the WARN triggers, but
looking at the code I think you mean that we fail to update the
config when we set at the device level?

> The napi_restore_config should only restore the napi threaded state when
> threaded is enabled at NAPI level.
> 
> The issue can be reproduced on virtio-net device using qemu. To
> reproduce the issue run following,
> 
>   echo 1 > /sys/class/net/threaded
>   ethtool -L eth0 combined 1

Maybe we should add that as a test under tools/testing/drivers/net -
it will run against netdevsim but also all drivers we test which
currently means virtio and fbnic, but hopefully soon more. Up to you.

I'm not sure I agree with the semantics, tho. IIUC you're basically
making the code prefer threaded option. If user enables threading
for the device, then disables it for a NAPI - reconfiguration will
lose the fact that specific NAPI instance was supposed to have threading
disabled. Why not update napi->config in netif_set_threaded() instead?
-- 
pw-bot: cr