Re: perfevents: irq loop stuck!

[Peter Zijlstra <peterz@infradead.org>]

[-- Attachment #1: Type: text/plain, Size: 1559 bytes --]

On Fri, May 16, 2014 at 12:25:28AM -0400, Vince Weaver wrote:
> anyway I'm not sure if it's worth tracking this more if it's possible to 
> mostly fix the case by fixing the sample_period bounds.

Right, so lets start with that, if it triggers again, we'll have another
look.

FWIW I ran with the below patch over night, and while trinity completely
shat itself going OOM -- so I'm not sure how long it ran, it didn't
trigger the stuck interrupt loop.

Will do more runs now that I'm there to hold its hand once more.

---
Subject: perf: Limit perf_event_attr::sample_period to 63 bits
From: Peter Zijlstra <peterz@infradead.org>
Date: Thu May 15 20:23:48 CEST 2014

Vince reported that using a large sample_period (one with bit 63 set)
results in wreckage since while the sample_period is fundamentally
unsigned (negative periods don't make sense) the way we implement
things very much rely on signed logic.

So limit sample_period to 63 bits to avoid tripping over this.

Reported-by: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/n/tip-p25fhunibl4y3qi0zuqmyf4b@git.kernel.org
---
 kernel/events/core.c |    3 +++
 1 file changed, 3 insertions(+)

--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -7058,6 +7058,9 @@ SYSCALL_DEFINE5(perf_event_open,
 	if (attr.freq) {
 		if (attr.sample_freq > sysctl_perf_event_sample_rate)
 			return -EINVAL;
+	} else {
+		if (attr.sample_period & (1ULL << 63))
+			return -EINVAL;
 	}
 
 	/*

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]