| 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
| ||
|
Message-ID: <87y5or1x7l.fsf@sejong.aot.lge.com>
Date: Thu, 17 May 2012 09:13:18 +0900
From: Namhyung Kim <namhyung.kim@....com>
To: Peter Zijlstra <peterz@...radead.org>
Cc: Ingo Molnar <mingo@...nel.org>,
Linus Torvalds <torvalds@...ux-foundation.org>,
Arnaldo Carvalho de Melo <acme@...radead.org>,
linux-kernel@...r.kernel.org,
Arjan van de Ven <arjan@...radead.org>,
David Ahern <dsahern@...il.com>,
Frederic Weisbecker <fweisbec@...il.com>,
Mike Galbraith <efault@....de>,
Paul Mackerras <paulus@...ba.org>,
Stephane Eranian <eranian@...gle.com>, arnaldo.melo@...il.com,
Arnaldo Carvalho de Melo <acme@...hat.com>
Subject: Re: [GIT PULL 0/8] Annotation weekly ponies delivery
Hi,
On Wed, 16 May 2012 14:57:49 +0900, Namhyung Kim wrote:
> In fact, my perf stat -g was already broken - it cannot count at
> all on my SNB box. I'll investigate it more.
>
It's like a matter of the number of the hw counters. The perf stat -g
uses 6 hw events, but I can use up to 5 hw events on my SNB machine:
$ ./perf stat -g -e cycles,cache-references,cache-misses,instructions,branches sleep 1
Performance counter stats for 'sleep 1':
999,008 cycles # 0.000 GHz
10,963 cache-references
2,630 cache-misses # 23.990 % of all cache refs
635,225 instructions # 0.64 insns per cycle
128,014 branches
1.001466421 seconds time elapsed
But I adding branch-misses event outputs:
Performance counter stats for 'sleep 1':
<not counted> cycles
<not counted> cache-references
<not counted> cache-misses
<not counted> instructions
<not counted> branches
<not counted> branch-misses
1.000742533 seconds time elapsed
However adding ref-cycles is fine (Is it related to fixed counters?):
Performance counter stats for 'sleep 1':
1,027,263 cycles # 0.000 GHz
12,435 cache-references
2,747 cache-misses # 22.091 % of all cache refs
650,100 instructions # 0.63 insns per cycle
130,765 branches
2,367,488 ref-cycles
1.001557088 seconds time elapsed
BTW, dmesg says I have 7, but a counter was taken by NMI watchdog:
$ dmesg | grep -A32 Performance
[ 0.137227] Performance Events: PEBS fmt1+, 16-deep LBR, SandyBridge events, Intel PMU driver.
[ 0.137232] ... version: 3
[ 0.137233] ... bit width: 48
[ 0.137234] ... generic registers: 4
[ 0.137235] ... value mask: 0000ffffffffffff
[ 0.137236] ... max period: 000000007fffffff
[ 0.137237] ... fixed-purpose events: 3
[ 0.137238] ... event mask: 000000070000000f
[ 0.143371] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.149251] Booting Node 0, Processors #1
[ 0.162396] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.166265] #2
[ 0.179401] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.183263] #3
[ 0.196399] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.200263] #4
[ 0.213398] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.217268] #5
[ 0.230404] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.234263] #6
[ 0.247402] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.251268] #7
[ 0.264405] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.268266] #8
[ 0.281403] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.285268] #9
[ 0.298405] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.302264] #10
[ 0.315401] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.319266] #11 Ok.
[ 0.332404] NMI watchdog: enabled, takes one hw-pmu counter.
[ 0.332435] Brought up 12 CPUs
[ 0.332437] Total of 12 processors activated (76799.44 BogoMIPS).
Thanks,
Namhyung
--
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/
Powered by blists - more mailing lists