Re: [PATCH 0/9] mm: improve OOM mechanism v2

[Michal Hocko]

On Thu 30-04-15 18:44:25, Tetsuo Handa wrote:
> Michal Hocko wrote:
> > I mean we should eventually fail all the allocation types but GFP_NOFS
> > is coming from _carefully_ handled code paths which is an easier starting
> > point than a random code path in the kernel/drivers. So can we finally
> > move at least in this direction?
> 
> I agree that all the allocation types can fail unless GFP_NOFAIL is given.
> But I also expect that all the allocation types should not fail unless
> order > PAGE_ALLOC_COSTLY_ORDER or GFP_NORETRY is given or chosen as an OOM
> victim.

Yeah, let's keep shooting our feet and then look for workarounds to deal
with it...
 
> We already experienced at Linux 3.19 what happens if !__GFP_FS allocations
> fails. out_of_memory() is called by pagefault_out_of_memory() when 0x2015a
> (!__GFP_FS) allocation failed.

I have posted a patch to deal with this
(http://marc.info/?l=linux-mm&m=142770374521952&w=2). There is no real
reason to do the GFP_NOFS from the page fault context just because the
mapping _always_ insists on it. Page fault simply _has_ to be GFP_FS
safe, we are badly broken otherwise. That patch should go in hand with
GFP_NOFS might fail one. I haven't posted it yet because I was waiting
for the merge window to close.

> This looks to me that !__GFP_FS allocations
> are effectively OOM killer context. It is not fair to kill the thread which
> triggered a page fault, for that thread may not be using so much memory
> (unfair from memory usage point of view) or that thread may be global init
> (unfair because killing the entire system than survive by killing somebody).

Why would we kill the faulting process?

> Also, failing the GFP_NOFS/GFP_NOIO allocations which are not triggered by
> a page fault generally causes more damage (e.g. taking filesystem error
> action) than survive by killing somebody. Therefore, I think we should not
> hesitate invoking the OOM killer for !__GFP_FS allocation.

No, we should fix those places and use proper gfp flags rather than
pretend that the problem doesn't exist and deal with all the side
effectes.
-- 
Michal Hocko
SUSE Labs
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/