linux-kernel - Re: [RFC] oom, but there is enough memory

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]

Date:	Thu, 31 Mar 2016 11:30:12 +0200
From:	Michal Hocko <mhocko@...nel.org>
To:	Xishi Qiu <qiuxishi@...wei.com>
Cc:	Linux MM <linux-mm@...ck.org>, LKML <linux-kernel@...r.kernel.org>
Subject: Re: [RFC] oom, but there is enough memory

On Thu 31-03-16 17:16:00, Xishi Qiu wrote:
> It triggers a lot of ooms, but there is enough memory(many large blocks).
> And at last "Kernel panic - not syncing: Out of memory and no killable processes..."
> 
> I find almost the every call trace include "pagefault_out_of_memory" and "gfp_mask=0x0".
> If it does oom, why not it triger in mm core path? 

It seems that somebody in the page fault path has returned with
VM_FAULT_OOM without invoking the page allocator and kept returning the
same error until there is nothing killable and so the oom killer panics.

[...]
> <4>[63651.040374s][pid:2912,cpu3,sh]DMA free:550600kB min:5244kB low:27580kB high:28892kB active_anon:343060kB inactive_anon:1224kB active_file:107596kB inactive_file:465156kB unevictable:1040kB isolated(anon):0kB isolated(file):0kB present:2016252kB managed:1720040kB mlocked:1040kB dirty:40kB writeback:0kB mapped:200420kB shmem:1312kB slab_reclaimable:27048kB slab_unreclaimable:73300kB kernel_stack:15248kB pagetables:14484kB unstable:0kB bounce:0kB free_cma:30896kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no

This is rather weird. DMA zone with 2GB? What kind of architecture is
this?
-- 
Michal Hocko
SUSE Labs