Re: [PATCH 1/2] mm, oom: introduce oom reaper

[Michal Hocko <mhocko@kernel.org>]

On Thu 24-12-15 20:06:50, Tetsuo Handa wrote:
> Michal Hocko wrote:
> > This is VM_BUG_ON_PAGE(page_mapped(page), page), right? Could you attach
> > the full kernel log? It all smells like a race when OOM reaper tears
> > down the mapping and there is a truncate still in progress. But hitting
> > the BUG_ON just because of that doesn't make much sense to me. OOM
> > reaper is essentially MADV_DONTNEED. I have to think about this some
> > more, though, but I am in a holiday mode until early next year so please
> > bear with me.
> 
> I don't know whether the OOM killer was invoked just before this
> VM_BUG_ON_PAGE().
> 
> > Is this somehow DAX related?
> 
> 4.4.0-rc6-next-20151223_new_fsync_v6+ suggests that this kernel
> has "[PATCH v6 0/7] DAX fsync/msync support" applied. But I think
> http://marc.info/?l=linux-mm&m=145068666428057 should be applied
> when retesting. (20151223 does not have this fix.)

Hmm, I think you are right! Very well spotted! If ignore_dirty ends up
being true then we would simply skip over dirty page and wouldn't end up
doing page_remove_rmap. I can see that the truncation code can later trip
over this page.

Thanks!
-- 
Michal Hocko
SUSE Labs