[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <568F9EC6.8070708@suse.cz>
Date: Fri, 8 Jan 2016 12:34:30 +0100
From: Vlastimil Babka <vbabka@...e.cz>
To: David Rientjes <rientjes@...gle.com>
Cc: linux-mm@...ck.org, linux-kernel@...r.kernel.org,
Andrew Morton <akpm@...ux-foundation.org>,
Joonsoo Kim <iamjoonsoo.kim@....com>,
Minchan Kim <minchan@...nel.org>,
Sasha Levin <sasha.levin@...cle.com>,
"Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>,
Mel Gorman <mgorman@...e.de>, Michal Hocko <mhocko@...e.cz>
Subject: Re: [PATCH v2 9/9] mm, oom: print symbolic gfp_flags in oom warning
On 01/07/2016 10:29 PM, David Rientjes wrote:
> On Tue, 24 Nov 2015, Vlastimil Babka wrote:
>
>> It would be useful to translate gfp_flags into string representation when
>> printing in case of an OOM, especially as the flags have been undergoing some
>> changes recently and the script ./scripts/gfp-translate needs a matching source
>> version to be accurate.
>>
>> Example output:
>>
>> a.out invoked oom-killer: order=0, oom_score_adj=0, gfp_mask=0x24280ca(GFP_HIGHUSER_MOVABLE|GFP_ZERO)
>>
>
> Is there a way that we can keep the order of the fields so that anything
> parsing the kernel log for oom kills doesn't break?
Yes, this is possible with the new printk handling of flags, please look
at v3:
http://marc.info/?l=linux-mm&m=145042944710510&w=2
There I changed the print just to have order first and gfp_mask next, as
it seemed more logical. But it doesn't need to be that way and I can
post V4 keeping the original order of variables. But do you think the
flags expansion is safe to add there, or should I put it on separate line?
Thanks
> The messages printed
> to the kernel log are the only (current) way to determine that the kernel
> killed something so we should be careful not to break anything parsing
> them, and this is a common line to look for.
>
Powered by blists - more mailing lists