linux-kernel - Re: Interesting interaction between lguest and CFS

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: <20070605071904.GB25163@elte.hu>
Date:	Tue, 5 Jun 2007 09:19:04 +0200
From:	Ingo Molnar <mingo@...e.hu>
To:	Matt Mackall <mpm@...enic.com>
Cc:	Rusty Russell <rusty@...tcorp.com.au>, akpm@...ux-foundation.org,
	Linux-kernel@...r.kernel.org
Subject: Re: Interesting interaction between lguest and CFS


* Matt Mackall <mpm@...enic.com> wrote:

> sleep_max                :         57476665627
> block_max                :   18014060106626075

hm, this block_max looks a bit suspect, it's 003fffb1359e341b. Does your 
box make any use of cpufreq? (what CPU is it?)

> strace -f doesn't kill the latency, here's an strace:
> 
> 24936 13:39:44.469329 rt_sigprocmask(SIG_BLOCK, [USR1], NULL, 8) = 0
> 24936 13:39:44.469572 select(4, [0 3], NULL, NULL, {0, 0}) = 0 (Timeout)
> 24936 13:39:44.469677 rt_sigprocmask(SIG_UNBLOCK, [USR1], NULL, 8) = 0
> 24936 13:39:44.469765 read(7, 0xbfe6c010, 8) = -1 EINTR (Interrupted system call)
> 24936 13:39:48.699490 --- SIGUSR1 (User defined signal 1) @ 0 (0) ---

hm, this indeed implicates some wakeup problem. lguest task 24936's 
relevant stats are:

 block_start              :                   0
 sleep_max                :            21492554
 block_max                :            27044576
 exec_max                 :             4008057
 wait_max                 :          1253670288

the wait_max means it was delayed on the runqueue for 1.2 seconds. Could 
you try to get a /proc/sched_debug snapshot done exactly during the 
'delay' window - it's 4 seconds so you should in theory be able to 
trigger it by doing something like:

  sleep 3; cat /proc/sched_debug > sched_debug.txt

Click into the lguest window and trigger the delay.

	Ingo
-
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/