linux-kernel - Re: [PATCH 1/2] drivers: cpuidle: remove stale irq disabling call in cpuidle_enter

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: <1789709.hzh3CkisvM@vostro.rjw.lan>
Date:	Thu, 26 Feb 2015 00:36:10 +0100
From:	"Rafael J. Wysocki" <rjw@...ysocki.net>
To:	Lorenzo Pieralisi <lorenzo.pieralisi@....com>,
	Daniel Lezcano <daniel.lezcano@...aro.org>
Cc:	"linux-pm@...r.kernel.org" <linux-pm@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	Peter Zijlstra <peterz@...radead.org>
Subject: Re: [PATCH 1/2] drivers: cpuidle: remove stale irq disabling call in cpuidle_enter_freeze()

On Wednesday, February 25, 2015 02:39:17 PM Lorenzo Pieralisi wrote:
> On Wed, Feb 25, 2015 at 02:13:23PM +0000, Daniel Lezcano wrote:
> > On 02/24/2015 06:58 PM, Lorenzo Pieralisi wrote:
> > > On return from cpuidle_enter_freeze() irqs are re-enabled by the function
> > > caller (ie cpuidle_idle_call) in the idle loop. This patch removes a stale
> > > local_irq_disable() call and its stale comment in cpuidle_enter_freeze(),
> > > since they disagree and do not serve a useful purpose.
> > >
> > > Cc: Rafael J. Wysocki <rjw@...ysocki.net>
> > > Cc: Daniel Lezcano <daniel.lezcano@...aro.org>
> > > Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@....com>
> > > ---
> > >   drivers/cpuidle/cpuidle.c | 3 ---
> > >   1 file changed, 3 deletions(-)
> > >
> > > diff --git a/drivers/cpuidle/cpuidle.c b/drivers/cpuidle/cpuidle.c
> > > index 4d53458..f47edc6c 100644
> > > --- a/drivers/cpuidle/cpuidle.c
> > > +++ b/drivers/cpuidle/cpuidle.c
> > > @@ -144,9 +144,6 @@ void cpuidle_enter_freeze(void)
> > >   		cpuidle_enter(drv, dev, index);
> > >   	else
> > >   		arch_cpu_idle();
> > > -
> > > -	/* Interrupts are enabled again here. */
> > > -	local_irq_disable();
> > >   }
> > 
> > Hmm, I think Rafael added this prevent lockdep to raise a warning.
> 
> Ok, so the comment is there to say "at this point of execution IRQs
> are enabled", it does not refer to local_irq_disable() call effects,
> that's misleading and not necessarily nice, at least it should
> be explained.
> 
> > Otherwise, cpuidle_enter or arch_cpu_idle enables the irq again and then 
> > when exiting the cpu_idle_call, we enable them again, so leading to a 
> > lockdep WARN in trace_hardirqs_on_caller.
> 
> Would not it be better to enable irqs in cpuidle_enter_freeze() on
> returning from enter_freeze_proper() and remove the local_irq_enable()
> call in the cpuidle_idle_call() before jumping to exit_idle ?
> 
> > That said, if we have to do this, it may reveal something is wrong in 
> > the code.
> 
> I just spotted code through inspection, I have to say at the moment it
> is not very clear what it is meant to achieve, so I put together this
> patch.

So there are two code paths in cpuidle_idle_call(), the enter_freeze_proper()
one which does *not* re-enable interrupts and the one you modified which does
that.  The local_irq_disable() is to keep things consistent.

I'm not entirely against of re-arranging things here, but a patch like the
(untested) one below might be more appropriate.

Rafael (who would appreciate it if people asked questions instead of sending
patches on a hunch).


---
 drivers/cpuidle/cpuidle.c |    2 +-
 kernel/sched/idle.c       |    1 -
 2 files changed, 1 insertion(+), 2 deletions(-)

Index: linux-pm/drivers/cpuidle/cpuidle.c
===================================================================
--- linux-pm.orig/drivers/cpuidle/cpuidle.c
+++ linux-pm/drivers/cpuidle/cpuidle.c
@@ -132,6 +132,7 @@ void cpuidle_enter_freeze(void)
 	index = cpuidle_find_deepest_state(drv, dev, true);
 	if (index >= 0) {
 		enter_freeze_proper(drv, dev, index);
+		local_irq_enable();
 		return;
 	}
 
@@ -146,7 +147,6 @@ void cpuidle_enter_freeze(void)
 		arch_cpu_idle();
 
 	/* Interrupts are enabled again here. */
-	local_irq_disable();
 }
 
 /**
Index: linux-pm/kernel/sched/idle.c
===================================================================
--- linux-pm.orig/kernel/sched/idle.c
+++ linux-pm/kernel/sched/idle.c
@@ -116,7 +116,6 @@ static void cpuidle_idle_call(void)
 	 */
 	if (idle_should_freeze()) {
 		cpuidle_enter_freeze();
-		local_irq_enable();
 		goto exit_idle;
 	}
 

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/