[PATCH] dumpoops: ignore "Oops" messages in syslog rhbz#726707

[Jiri Moskovcak]

This patch makes abrt-dump-oops ignore the messages with "Oops" prefix, 
it's not a standard warning and doesn't have any predefined format, so 
it's hard to properly parse it and in some cases (when it's followed by 
other kerneloops) it confuses the logic in abrt-dump-oops. After a 
discussion with kernel devs I propose to not catch these messages.

Proposed patch and testing syslog attached. Run-tested, please review.
--Jirka

[Nikola Pajkovsky]

Jiri Moskovcak <jmoskovc(a)redhat.com&gt; writes:

>  This patch makes abrt-dump-oops ignore the messages with
> "Oops"
>  prefix, it's not a standard warning and doesn't have any predefined
>  format, so it's hard to properly parse it and in some cases (when it's
>  followed by other kerneloops) it confuses the logic in
>  abrt-dump-oops. After a discussion with kernel devs I propose to not
>  catch these messages.
>
>  Proposed patch and testing syslog attached. Run-tested, please review.
>  --Jirka
>
> >From 59a82b8209e42078a8c9a71b87031093882b5930 Mon Sep 17 00:00:00 2001
>  From: Jiri Moskovcak <jmoskovc(a)redhat.com&gt;
>  Date: Mon, 12 Dec 2011 09:34:46 +0100
>  Subject: [PATCH] ignore "Oops" messages, rhbz#726707
>
>  - "Oops" is not a standard warning messages and we should ignore it
>  - it also doesn't have a calltrace which seems to confuse our parser
>    which is the case in rhzb#726707
>  ---
>   src/plugins/abrt-dump-oops.c |    3 ---
>   1 files changed, 0 insertions(+), 3 deletions(-)
>
>  diff --git a/src/plugins/abrt-dump-oops.c b/src/plugins/abrt-dump-oops.c
>  index 43a3c84..2900435 100644
>  --- a/src/plugins/abrt-dump-oops.c
>  +++ b/src/plugins/abrt-dump-oops.c
>  @@ -244,9 +244,6 @@ next_line:
>               else if (strstr(curline, "IRQ handler type mismatch"))
>                   oopsstart = i;
>   
>  -            if (i >= 3 && strstr(curline, "Oops:"))
>  -                oopsstart = i-3;
>  -
>               if (oopsstart >= 0)
>               {
>                   /* debug information */ 
ok to me. Could you add this oops into unit test?

-- 
Nikola

[Denys Vlasenko]

On Mon, 2011-12-12 at 09:39 +0100, Jiri Moskovcak wrote:
>  This patch makes abrt-dump-oops ignore the messages with
> "Oops" prefix, 
>  it's not a standard warning and doesn't have any predefined format, so 
>  it's hard to properly parse it and in some cases (when it's followed by 
>  other kerneloops) it confuses the logic in abrt-dump-oops. After a 
>  discussion with kernel devs I propose to not catch these messages.
>  
>  Proposed patch and testing syslog attached. Run-tested, please review. 
Looks good to me