linux-kernel - Re: [RFC net-next 3/6] gve: Use napi_affinity_no

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: <CANLc=av2PUaXQ5KVPQGppOdD5neHUtUgioqOO4fA=+Qb594Z4w@mail.gmail.com>
Date: Tue, 13 Aug 2024 11:55:31 -0700
From: Shailend Chand <shailend@...gle.com>
To: Joe Damato <jdamato@...tly.com>
Cc: netdev@...r.kernel.org, Jeroen de Borst <jeroendb@...gle.com>, 
	Praveen Kaligineedi <pkaligineedi@...gle.com>, "David S. Miller" <davem@...emloft.net>, 
	Eric Dumazet <edumazet@...gle.com>, Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>, 
	Willem de Bruijn <willemb@...gle.com>, Harshitha Ramamurthy <hramamurthy@...gle.com>, 
	Ziwei Xiao <ziweixiao@...gle.com>, open list <linux-kernel@...r.kernel.org>
Subject: Re: [RFC net-next 3/6] gve: Use napi_affinity_no_change

On Mon, Aug 12, 2024 at 7:57 AM Joe Damato <jdamato@...tly.com> wrote:
>
> Use napi_affinity_no_change instead of gve's internal implementation,
> simplifying and centralizing the logic.
>
> Signed-off-by: Joe Damato <jdamato@...tly.com>
> ---
>  drivers/net/ethernet/google/gve/gve_main.c | 14 +-------------
>  1 file changed, 1 insertion(+), 13 deletions(-)
>
> diff --git a/drivers/net/ethernet/google/gve/gve_main.c b/drivers/net/ethernet/google/gve/gve_main.c
> index 661566db68c8..ad5e85b8c6a5 100644
> --- a/drivers/net/ethernet/google/gve/gve_main.c
> +++ b/drivers/net/ethernet/google/gve/gve_main.c
> @@ -298,18 +298,6 @@ static irqreturn_t gve_intr_dqo(int irq, void *arg)
>         return IRQ_HANDLED;
>  }
>
> -static int gve_is_napi_on_home_cpu(struct gve_priv *priv, u32 irq)
> -{
> -       int cpu_curr = smp_processor_id();
> -       const struct cpumask *aff_mask;
> -
> -       aff_mask = irq_get_effective_affinity_mask(irq);
> -       if (unlikely(!aff_mask))
> -               return 1;
> -
> -       return cpumask_test_cpu(cpu_curr, aff_mask);
> -}
> -
>  int gve_napi_poll(struct napi_struct *napi, int budget)
>  {
>         struct gve_notify_block *block;
> @@ -383,7 +371,7 @@ int gve_napi_poll_dqo(struct napi_struct *napi, int budget)
>                 /* Reschedule by returning budget only if already on the correct
>                  * cpu.
>                  */
> -               if (likely(gve_is_napi_on_home_cpu(priv, block->irq)))
> +               if (likely(napi_affinity_no_change(block->irq)))

Nice to centralize this code! Evolving this to cache the affinity mask
like the other drivers would probably also require a means to update
the cache when the affinity changes.

>                         return budget;
>
>                 /* If not on the cpu with which this queue's irq has affinity
> --
> 2.25.1
>