lkml.org 
[lkml]   [2008]   [Jan]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] Only print kernel debug information for OOMs caused by kernel allocations
On Mon, 28 Jan 2008 10:11:57 +0100 Andi Kleen <ak@suse.de> wrote:

> On Monday 28 January 2008 09:56, Andrew Morton wrote:
> > On Mon, 28 Jan 2008 07:10:07 +0100 Andi Kleen <ak@suse.de> wrote:
> > > On Monday 28 January 2008 06:52, Andrew Morton wrote:
> > > > On Wed, 16 Jan 2008 23:24:21 +0100 Andi Kleen <ak@suse.de> wrote:
> > > > > I recently suffered an 20+ minutes oom thrash disk to death and
> > > > > computer completely unresponsive situation on my desktop when some
> > > > > user program decided to grab all memory. It eventually recovered, but
> > > > > left lots of ugly and imho misleading messages in the kernel log.
> > > > > here's a minor improvement
> > >
> > > As a followup this was with swap over dm crypt. I've recently heard
> > > about other people having trouble with this too so this setup seems to
> > > trigger something bad in the VM.
> >
> > Where's the backtrace and show_mem() output? :)
>
> I don't have it anymore. You want me to reproduce it? I don't think
> I saw messages from the other people either; just heard complaints.

May as well - it doesn't sound like it'll fix itself...

> > > > That information is useful for working out why a userspace allocation
> > > > attempt failed. If we don't print it, and the application gets killed
> > > > and thus frees a lot of memory, we will just never know why the
> > > > allocation failed.
> > >
> > > But it's basically only either page fault (direct or indirect) or write
> > > et.al. who do these page cache allocations. Do you really think it is
> > > that important to distingush these cases individually? In 95+% of all
> > > cases it should be a standard user page fault which always has the same
> > > backtrace.
> >
> > Sure, the backtrace isn't very important. The show_mem() output is vital.
>
> I see. So would the patch be acceptable if it only disabled the backtrace?

Spose so. The show_mem() spew is probably larger than the backtrace
though.

Are you sure we aren't doing dump_stack()/show_mem() mutiple times for a
single process? If we are, that would mena the TIF_MEMDIE thing broke.

It must have been one heck of an oomkilling slaughter.

> > Plus an additional function call. On the already-deep page allocation
> > path, I might add.
>
> The function call is already there if the kernel has CPUSETs enabled.

s/CPUSETS/NUMA/, which makes rather a difference.

> And that is what distribution kernels usually do. And most users
> use distribution kernels or distribution .config.




\
 
 \ /
  Last update: 2008-01-28 10:29    [W:0.843 / U:0.012 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site