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: <20240828093719.3KJWbR6Y@linutronix.de>
Date: Wed, 28 Aug 2024 11:37:19 +0200
From: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
To: "Brandt, Oliver - Lenze" <oliver.brandt@...ze.com>
Cc: "tglx@...utronix.de" <tglx@...utronix.de>,
	"peterz@...radead.org" <peterz@...radead.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] irq_work: Avoid unnecessary "IRQ work" interrupts

On 2024-08-27 15:45:42 [+0000], Brandt, Oliver - Lenze wrote:
> It was noticed that newer kernels (>= 6.1) raises a lot of "IRQ work"
> interrupts where older ones (<= 5.10) haven't used this at all on our
> system.
> 
> Root cause seems to be commit b4c6f86ec2f6 ('irq_work: Handle some
> irq_work in a per-CPU thread on PREEMPT_RT'). This commit tries to avoid
> calling irq_work callbacks from hardirq context as much as possible.
> Therefore interrupts marked as IRQ_WORK_LAZY and (on PREEMT_RT) interrupts
> not marked as IRQ_WORK_HARD_IRQ should be handled from an per-CPU thread.
> 
> Running the remaining interrupts from hardirq context is triggered using
> irq_work_raise(). But on PREEMPT_RT irq_work_raise() will be called for
> all interrupts not marked as IRQ_WORK_LAZY which results in unnecessary
> "IRQ work" interrupts.

Good catch I think.

> Fixes: b4c6f86ec2f6 ('irq_work: Handle some irq_work in a per-CPU thread on PREEMPT_RT')
> Signed-off-by: Oliver Brandt <oliver.brandt@...ze.com>
> ---
>  kernel/irq_work.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/irq_work.c b/kernel/irq_work.c
> index 2f4fb336dda1..df08b7dde7d5 100644
> --- a/kernel/irq_work.c
> +++ b/kernel/irq_work.c
> @@ -108,7 +108,7 @@ static void __irq_work_queue_local(struct irq_work *work)
>                 return;
>  
>         /* If the work is "lazy", handle it from next tick if any */
> -       if (!lazy_work || tick_nohz_tick_stopped())
> +       if (!(lazy_work || rt_lazy_work) || tick_nohz_tick_stopped())
>                 irq_work_raise(work);

Looking at this I *think* rt_lazy_work was needed earlier due to
different code but not anymore. Couldn't you just remove rt_lazy_work
and set lazy_work in the RT path? That should work.

>  }
>  

Sebastian

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ