All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Rafael J. Wysocki" <rjw@rjwysocki.net>
To: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>,
	linux-kernel@vger.kernel.org,
	Linux PM list <linux-pm@vger.kernel.org>
Subject: Re: [RFC][PATCH] irq: Rework IRQF_NO_SUSPENDED
Date: Fri, 25 Jul 2014 14:47:25 +0200	[thread overview]
Message-ID: <1805200.1exkRoNuC2@vostro.rjw.lan> (raw)
In-Reply-To: <alpine.DEB.2.10.1407251135590.23352@nanos>

On Friday, July 25, 2014 11:40:48 AM Thomas Gleixner wrote:
> On Thu, 24 Jul 2014, Peter Zijlstra wrote:
> > @@ -29,14 +29,20 @@ void suspend_device_irqs(void)
> >  	for_each_irq_desc(irq, desc) {
> >  		unsigned long flags;
> >  
> > +		/*
> > +		 * Ideally this would be a global state, but we cannot
> > +		 * for the trainwreck that is IRQD_WAKEUP_STATE.
> > +		 */
> >  		raw_spin_lock_irqsave(&desc->lock, flags);
> > -		__disable_irq(desc, irq, true);
> > +		if (!irqd_has_set(&desc->irq_data, IRQD_WAKEUP_STATE))
> > +			desc->istate |= IRQS_SUSPENDED;
> >  		raw_spin_unlock_irqrestore(&desc->lock, flags);
> >  	}
> >  
> > -	for_each_irq_desc(irq, desc)
> > +	for_each_irq_desc(irq, desc) {
> >  		if (desc->istate & IRQS_SUSPENDED)
> >  			synchronize_irq(irq);
> > +	}
> >  }
> 
> So, instead of disabling the interrupt you just mark it
> suspended. Good luck with level triggered interrupt lines then.
> 
> Assume the interrupt fires after you marked it suspended. Then the
> flow handler will call handle_irq_event() which will do nothing and
> return handled. So the flow handler will reenable the interrupt line,
> which will cause the interrupt to fire immediately again after the
> RETI. Guess how much progress the system is going to make when that
> happens.

Good point.

So it looks like we really need the "suspend" thing to either disable
the interrupt entirely (in which case all handlers for all actions
will not be invoked after it's been suspended) or leave it enabled
(causing all handlers to be invoked all the time).

I'm not sure if we can do much beyond what's already in your tree
about that, then.

Well, perhaps the patch failing the irq request in case of the
IRQF_NO_SUSPEND mismatch is a bit too drastic.  Instead, we could
just print a warning in __setup_irq() in that case and then check
IRQF_NO_SUSPEND for all actions in __disable_irq().

What about this?

---
 kernel/irq/manage.c |   22 +++++++++++++++++++---
 1 file changed, 19 insertions(+), 3 deletions(-)

Index: linux-pm/kernel/irq/manage.c
===================================================================
--- linux-pm.orig/kernel/irq/manage.c
+++ linux-pm/kernel/irq/manage.c
@@ -385,8 +385,17 @@ setup_affinity(unsigned int irq, struct
 void __disable_irq(struct irq_desc *desc, unsigned int irq, bool suspend)
 {
 	if (suspend) {
-		if (!desc->action || (desc->action->flags & IRQF_NO_SUSPEND)
-		    || irqd_has_set(&desc->irq_data, IRQD_WAKEUP_STATE))
+		struct irqaction *action = desc->action;
+		unsigned int flags;
+
+		if (!action || irqd_has_set(&desc->irq_data, IRQD_WAKEUP_STATE))
+			return;
+		flags = IRQF_NO_SUSPEND;
+		do {
+			flags &= action->flags;
+			action = action->next;
+		} while (action);
+		if (flags)
 			return;
 		desc->istate |= IRQS_SUSPENDED;
 	}
@@ -1079,7 +1088,7 @@ __setup_irq(unsigned int irq, struct irq
 		 */
 
 #define IRQF_MISMATCH \
-	(IRQF_TRIGGER_MASK | IRQF_ONESHOT | IRQF_NO_SUSPEND)
+	(IRQF_TRIGGER_MASK | IRQF_ONESHOT)
 
 		if (!((old->flags & new->flags) & IRQF_SHARED) ||
 		    ((old->flags ^ new->flags) & IRQF_MISMATCH))
@@ -1090,6 +1099,13 @@ __setup_irq(unsigned int irq, struct irq
 		    (new->flags & IRQF_PERCPU))
 			goto mismatch;
 
+		if ((old->flags & IRQF_NO_SUSPEND) !=
+		    (new->flags & IRQF_NO_SUSPEND)) {
+			pr_err("irq %d: IRQF_NO_SUSPEND mismatch %08x (%s) vs. %08x (%s)\n",
+			       irq, new->flags, new->name, old->flags, old->name);
+			pr_err("irq %d: IRQF_NO_SUSPEND will be ignored\n");
+		}
+
 		/* add new interrupt at end of irq queue */
 		do {
 			/*


  parent reply	other threads:[~2014-07-25 12:29 UTC|newest]

Thread overview: 75+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-07-24 21:26 [RFC][PATCH] irq: Rework IRQF_NO_SUSPENDED Peter Zijlstra
2014-07-24 22:02 ` Rafael J. Wysocki
2014-07-24 23:10 ` Rafael J. Wysocki
2014-07-25  5:58   ` Peter Zijlstra
2014-07-29 19:20     ` Brian Norris
2014-07-29 19:28       ` Peter Zijlstra
2014-07-29 20:41         ` Brian Norris
2014-07-25  9:27   ` Thomas Gleixner
2014-07-25 12:49     ` Rafael J. Wysocki
2014-07-25 13:55       ` Thomas Gleixner
2014-07-25  9:40 ` Thomas Gleixner
2014-07-25 12:40   ` Peter Zijlstra
2014-07-25 13:25     ` Peter Zijlstra
2014-07-25 17:03       ` Rafael J. Wysocki
2014-07-25 16:58         ` Peter Zijlstra
2014-07-25 21:00         ` Thomas Gleixner
2014-07-25 22:25           ` Rafael J. Wysocki
2014-07-25 23:07             ` Rafael J. Wysocki
2014-07-26 11:49             ` Rafael J. Wysocki
2014-07-26 11:53               ` Rafael J. Wysocki
2014-07-28  6:49               ` Peter Zijlstra
2014-07-28 12:33                 ` Thomas Gleixner
2014-07-28 13:04                   ` Peter Zijlstra
2014-07-28 21:53                   ` Rafael J. Wysocki
2014-07-28 23:01                     ` Rafael J. Wysocki
2014-07-29 12:46                       ` Thomas Gleixner
2014-07-29 13:33                         ` Rafael J. Wysocki
2014-07-30 21:46                           ` [PATCH 0/3] irq / PM: wakeup interrupt interface for drivers (was: Re: [RFC][PATCH] irq: Rework IRQF_NO_SUSPENDED) Rafael J. Wysocki
2014-07-30 21:51                             ` [PATCH 1/3] irq / PM: New driver interface for wakeup interrupts Rafael J. Wysocki
2014-07-30 22:56                               ` Thomas Gleixner
2014-07-31  0:12                                 ` Thomas Gleixner
2014-07-31  2:14                                   ` Rafael J. Wysocki
2014-07-31 10:44                                     ` Thomas Gleixner
2014-07-31 18:36                                       ` Rafael J. Wysocki
2014-07-31 20:12                                         ` Alan Stern
2014-07-31 20:12                                           ` Alan Stern
2014-07-31 21:04                                           ` Rafael J. Wysocki
2014-07-31 23:41                                             ` Thomas Gleixner
2014-08-01  0:51                                               ` Rafael J. Wysocki
2014-08-01 14:41                                               ` Alan Stern
2014-08-01 14:41                                                 ` Alan Stern
2014-07-31 22:16                                         ` Thomas Gleixner
2014-08-01  0:08                                           ` Rafael J. Wysocki
2014-08-01  1:24                                             ` Rafael J. Wysocki
2014-08-01  9:40                                             ` [PATCH 1/3] irq / PM: New driver interface for wakeup interruptsn Thomas Gleixner
2014-08-01 13:45                                               ` Rafael J. Wysocki
2014-08-01 13:43                                                 ` Thomas Gleixner
2014-08-01 14:29                                                   ` Rafael J. Wysocki
2014-08-02  1:31                                                     ` Rafael J. Wysocki
2014-08-03 13:42                                                       ` Rafael J. Wysocki
2014-08-04  3:38                                                         ` Rafael J. Wysocki
2014-08-05 15:22                                                     ` [PATCH 0/5] irq / PM: Shared IRQs vs IRQF_NO_SUSPEND and suspend-to-idle wakeup Rafael J. Wysocki
2014-08-05 15:24                                                       ` [PATCH 1/5] PM / sleep: Mechanism for aborting system suspends unconditionally Rafael J. Wysocki
2014-08-05 23:29                                                         ` [Update][PATCH " Rafael J. Wysocki
2014-08-05 15:25                                                       ` [PATCH 2/5] irq / PM: Fix IRQF_NO_SUSPEND problem with shared interrupts Rafael J. Wysocki
2014-08-05 15:26                                                       ` [PATCH 3/5] irq / PM: Make wakeup interrupts wake up from suspend-to-idle Rafael J. Wysocki
2014-08-08  1:58                                                         ` [Update][PATCH " Rafael J. Wysocki
2014-08-09  0:28                                                           ` Rafael J. Wysocki
2014-08-05 15:27                                                       ` [PATCH 4/5] x86 / PM: Set IRQCHIP_SKIP_SET_WAKE for IOAPIC IRQ chip objects Rafael J. Wysocki
2014-08-05 15:28                                                       ` [PATCH 5/5] PCI / PM: Make PCIe PME interrupts wake up from suspend-to-idle Rafael J. Wysocki
2014-08-05 16:12                                                       ` [PATCH 0/5] irq / PM: Shared IRQs vs IRQF_NO_SUSPEND and suspend-to-idle wakeup Peter Zijlstra
2014-08-08  2:09                                                       ` Rafael J. Wysocki
2014-07-31 22:54                                         ` [PATCH 1/3] irq / PM: New driver interface for wakeup interrupts Thomas Gleixner
2014-07-30 21:51                             ` [PATCH 2/3] PCI / PM: Make PCIe PME interrupts wake up from "freeze" sleep state Rafael J. Wysocki
2014-07-30 21:52                             ` [PATCH 3/3] gpio-keys / PM: use enable/disable_device_irq_wake() Rafael J. Wysocki
2014-07-28 21:27                 ` [RFC][PATCH] irq: Rework IRQF_NO_SUSPENDED Rafael J. Wysocki
2014-07-27 15:53             ` Rafael J. Wysocki
2014-07-27 22:00               ` [PATCH, v2] Rafael J. Wysocki
2014-07-28 12:11                 ` Thomas Gleixner
2014-07-28 21:17                   ` [PATCH, v3] irq / PM: Fix IRQF_NO_SUSPEND problem with shared interrupts (was: Re: [PATCH, v2]) Rafael J. Wysocki
2014-07-29  7:28                     ` [PATCH, v4] irq / PM: Fix IRQF_NO_SUSPEND problem with shared interrupts Rafael J. Wysocki
2014-07-29 13:46                       ` [PATCH, v5] " Rafael J. Wysocki
2014-07-30  0:54                         ` [PATCH, v6] " Rafael J. Wysocki
2014-07-25 12:47   ` Rafael J. Wysocki [this message]
2014-07-25 13:22     ` [RFC][PATCH] irq: Rework IRQF_NO_SUSPENDED Peter Zijlstra

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=1805200.1exkRoNuC2@vostro.rjw.lan \
    --to=rjw@rjwysocki.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pm@vger.kernel.org \
    --cc=peterz@infradead.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.