pcp
[Top] [All Lists]

Re: [pcp] linux PMDA regression?

To: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>, PCP <pcp@xxxxxxxxxxx>
Subject: Re: [pcp] linux PMDA regression?
From: Mark Goodwin <mgoodwin@xxxxxxxxxx>
Date: Tue, 20 Jan 2015 07:19:41 +1100
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <54BD5F60.9000107@xxxxxxxxxxxxxxxx>
References: <54BD5F60.9000107@xxxxxxxxxxxxxxxx>
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.3.0
On 01/20/2015 06:47 AM, Ken McDonell wrote:
I'm seeing qa/957 failing on several hosts.

e.g. on a single CPU machine (vm23) AND using a local context ...

$ valgrind pminfo -L -f kernel.percpu.cpu.guest
==5903== Memcheck, a memory error detector
==5903== Copyright (C) 2002-2013, and GNU GPL'd, by Julian Seward et al.
==5903== Using Valgrind-3.9.0 and LibVEX; rerun with -h for copyright info
==5903== Command: pminfo -L -f kernel.percpu.cpu.guest
==5903==
==5903== Invalid write of size 4
==5903==    at 0x44B85B7: refresh_proc_stat (proc_stat.c:183)

git blame shows that's in code I recently changed to fix the
per-node CPU metrics for single CPU systems (commit 4ea1e0b7).
Mostly likely issues with the two arrays that we were discussing.

thanks for the report, I'll fix it up.

Regards
-- Mark


<Prev in Thread] Current Thread [Next in Thread>