From: Peter Zijlstra <peterz@infradead.org>
To: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: linux-kernel@vger.kernel.org, Thomas Gleixner <tglx@linutronix.de>
Subject: Re: [PATCH v2 3/4] irq_work: Handle some irq_work in a per-CPU thread on PREEMPT_RT
Date: Thu, 7 Oct 2021 10:50:23 +0200 [thread overview]
Message-ID: <20211007085023.GP174703@worktop.programming.kicks-ass.net> (raw)
In-Reply-To: <20211006111852.1514359-4-bigeasy@linutronix.de>
On Wed, Oct 06, 2021 at 01:18:51PM +0200, Sebastian Andrzej Siewior wrote:
> @@ -104,7 +141,24 @@ bool irq_work_queue_on(struct irq_work *work, int cpu)
> if (cpu != smp_processor_id()) {
> /* Arch remote IPI send/receive backend aren't NMI safe */
> WARN_ON_ONCE(in_nmi());
> - __smp_call_single_queue(cpu, &work->node.llist);
> +
> + /*
> + * On PREEMPT_RT the items which are not marked as
> + * IRQ_WORK_HARD_IRQ are added to the lazy list and a HARD work
> + * item is used on the remote CPU to wake the thread.
> + */
> + if (IS_ENABLED(CONFIG_PREEMPT_RT) &&
> + !(atomic_read(&work->node.a_flags) & IRQ_WORK_HARD_IRQ) &&
> + llist_add(&work->node.llist, &per_cpu(lazy_list, cpu))) {
So if the llist not empty, we'll fail here, and go to the else branch
> + struct irq_work *wake_work;
> +
> + wake_work = &per_cpu(irq_work_pending, cpu);
> + if (irq_work_claim(wake_work))
> + __smp_call_single_queue(cpu,
> + &wake_work->node.llist);
> + } else {
And do this,.. that seems wrong.
> + __smp_call_single_queue(cpu, &work->node.llist);
> + }
> } else {
> __irq_work_queue_local(work);
> }
How's this instead?
(work rename due to there already being a percpu variable of that same
name on a number or archs, per 0day)
Index: linux-2.6/kernel/irq_work.c
===================================================================
--- linux-2.6.orig/kernel/irq_work.c
+++ linux-2.6/kernel/irq_work.c
@@ -39,7 +39,7 @@ static void irq_work_wake(struct irq_wor
wake_irq_workd();
}
-static DEFINE_PER_CPU(struct irq_work, irq_work_pending) =
+static DEFINE_PER_CPU(struct irq_work, irq_work_wakeup) =
IRQ_WORK_INIT_HARD(irq_work_wake);
static int irq_workd_should_run(unsigned int cpu)
@@ -148,20 +148,21 @@ bool irq_work_queue_on(struct irq_work *
* item is used on the remote CPU to wake the thread.
*/
if (IS_ENABLED(CONFIG_PREEMPT_RT) &&
- !(atomic_read(&work->node.a_flags) & IRQ_WORK_HARD_IRQ) &&
- llist_add(&work->node.llist, &per_cpu(lazy_list, cpu))) {
- struct irq_work *wake_work;
-
- wake_work = &per_cpu(irq_work_pending, cpu);
- if (irq_work_claim(wake_work))
- __smp_call_single_queue(cpu,
- &wake_work->node.llist);
- } else {
- __smp_call_single_queue(cpu, &work->node.llist);
+ !(atomic_read(&work->node.a_flags) & IRQ_WORK_HARD_IRQ)) {
+
+ if (!llist_add(&work->node.llist, &per_cpu(lazy_list, cpu)))
+ goto out;
+
+ work = &per_cpu(irq_work_wakeup, cpu);
+ if (!irq_work_claim(wake_work))
+ goto out;
}
+
+ __smp_call_single_queue(cpu, &work->node.llist);
} else {
__irq_work_queue_local(work);
}
+out:
preempt_enable();
return true;
next prev parent reply other threads:[~2021-10-07 8:50 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-10-06 11:18 [PATCH v2 0/4] irq_work: PREEMPT_RT bits Sebastian Andrzej Siewior
2021-10-06 11:18 ` [PATCH v2 1/4] sched/rt: Annotate the RT balancing logic irqwork as IRQ_WORK_HARD_IRQ Sebastian Andrzej Siewior
2021-10-13 22:13 ` Steven Rostedt
2021-10-15 9:44 ` [tip: sched/core] " tip-bot2 for Sebastian Andrzej Siewior
2021-10-06 11:18 ` [PATCH v2 2/4] irq_work: Allow irq_work_sync() to sleep if irq_work() no IRQ support Sebastian Andrzej Siewior
2021-10-15 9:44 ` [tip: sched/core] " tip-bot2 for Sebastian Andrzej Siewior
2021-10-06 11:18 ` [PATCH v2 3/4] irq_work: Handle some irq_work in a per-CPU thread on PREEMPT_RT Sebastian Andrzej Siewior
2021-10-06 11:35 ` Sebastian Andrzej Siewior
2021-10-06 16:26 ` kernel test robot
2021-10-07 8:50 ` Peter Zijlstra [this message]
2021-10-07 9:08 ` Sebastian Andrzej Siewior
2021-10-07 9:26 ` [PATCH v3 " Sebastian Andrzej Siewior
2021-10-15 9:44 ` [tip: sched/core] " tip-bot2 for Sebastian Andrzej Siewior
2021-10-06 11:18 ` [PATCH v2 4/4] irq_work: Also rcuwait for !IRQ_WORK_HARD_IRQ " Sebastian Andrzej Siewior
2021-10-15 9:44 ` [tip: sched/core] " tip-bot2 for Sebastian Andrzej Siewior
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20211007085023.GP174703@worktop.programming.kicks-ass.net \
--to=peterz@infradead.org \
--cc=bigeasy@linutronix.de \
--cc=linux-kernel@vger.kernel.org \
--cc=tglx@linutronix.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).