netdev - Re: [PATCH net-next] net: warn if NAPI instance wasn't shut down

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: <CANn89iKstzu3O2eD1OAhR=Vc1EVyesg14DAp0P42i8juJuH_gA@mail.gmail.com>
Date: Tue, 4 Feb 2025 06:24:49 +0100
From: Eric Dumazet <edumazet@...gle.com>
To: Jakub Kicinski <kuba@...nel.org>
Cc: davem@...emloft.net, netdev@...r.kernel.org, pabeni@...hat.com, 
	andrew+netdev@...n.ch, horms@...nel.org
Subject: Re: [PATCH net-next] net: warn if NAPI instance wasn't shut down

On Mon, Feb 3, 2025 at 10:58 PM Jakub Kicinski <kuba@...nel.org> wrote:
>
> Drivers should always disable a NAPI instance before removing it.
> If they don't the instance may be queued for polling.
> Since commit 86e25f40aa1e ("net: napi: Add napi_config")
> we also remove the NAPI from the busy polling hash table
> in napi_disable(), so not disabling would leave a stale
> entry there.
>
> Use of busy polling is relatively uncommon so bugs may be lurking
> in the drivers. Add an explicit warning.
>
> Signed-off-by: Jakub Kicinski <kuba@...nel.org>
> ---
>  net/core/dev.c | 3 +++
>  1 file changed, 3 insertions(+)
>
> diff --git a/net/core/dev.c b/net/core/dev.c
> index c0021cbd28fc..2b141f20b13b 100644
> --- a/net/core/dev.c
> +++ b/net/core/dev.c
> @@ -7071,6 +7071,9 @@ void __netif_napi_del_locked(struct napi_struct *napi)
>         if (!test_and_clear_bit(NAPI_STATE_LISTED, &napi->state))
>                 return;
>
> +       /* Make sure NAPI is disabled (or was never enabled). */
> +       WARN_ON(!test_bit(NAPI_STATE_SCHED, &napi->state));
> +
>         if (napi->config) {
>                 napi->index = -1;
>                 napi->config = NULL;
> --
> 2.48.1
>

This makes sense. Although a WARN_ON_ONCE() might avoid some noise.

Reviewed-by: Eric Dumazet <edumazet@...gle.com>