Re: System freezes after OOM

[Mikulas Patocka]

On Thu, 14 Jul 2016, Tetsuo Handa wrote:

> Michal Hocko wrote:
> > OK, this is the part I have missed. I didn't realize that the swapout
> > path, which is indeed PF_MEMALLOC, can get down to blk code which uses
> > mempools. A quick code travers shows that at least
> > 	make_request_fn = blk_queue_bio
> > 	blk_queue_bio
> > 	  get_request
> > 	    __get_request
> > 
> > might do that. And in that case I agree that the above mentioned patch
> > has unintentional side effects and should be re-evaluated. David, what
> > do you think? An obvious fixup would be considering TIF_MEMDIE in
> > mempool_alloc explicitly.
> 
> TIF_MEMDIE is racy. Since the OOM killer sets TIF_MEMDIE on only one thread,
> there is no guarantee that TIF_MEMDIE is set to the thread which is looping
> inside mempool_alloc().

If the device mapper subsystem is not returning objects to the mempool, it 
should be investigated as a bug in the device mapper.

There is no need to add workarounds to mempool_alloc to work around that 
bug.

Mikulas

> And since __GFP_NORETRY is used (regardless of
> f9054c70d28bc214), out_of_memory() is not called via __alloc_pages_may_oom().
> This means that the thread which is looping inside mempool_alloc() can't
> get TIF_MEMDIE unless TIF_MEMDIE is set by the OOM killer.
> 
> Maybe set __GFP_NOMEMALLOC by default at mempool_alloc() and remove it
> at mempool_alloc() when fatal_signal_pending() is true? But that behavior
> can OOM-kill somebody else when current was not OOM-killed. Sigh...
> 
> David Rientjes wrote:
> > On Wed, 13 Jul 2016, Mikulas Patocka wrote:
> > 
> > > What are the real problems that f9054c70d28bc214b2857cf8db8269f4f45a5e23 
> > > tries to fix?
> > > 
> > 
> > It prevents the whole system from livelocking due to an oom killed process 
> > stalling forever waiting for mempool_alloc() to return.  No other threads 
> > may be oom killed while waiting for it to exit.
> 
> Is that concern still valid? We have the OOM reaper for CONFIG_MMU=y case.
>