On Wed, 13 Jul 2016, Michal Hocko wrote:
> On Wed 13-07-16 10:18:35, Mikulas Patocka wrote:
> >
> >
> > On Wed, 13 Jul 2016, Michal Hocko wrote:
> >
> > > [CC David]
> > >
> > > > > It is caused by the commit f9054c70d28bc214b2857cf8db8269f4f45a5e23.
> > > > > Prior to this commit, mempool allocations set __GFP_NOMEMALLOC, so
> > > > > they never exhausted reserved memory. With this commit, mempool
> > > > > allocations drop __GFP_NOMEMALLOC, so they can dig deeper (if the
> > > > > process has PF_MEMALLOC, they can bypass all limits).
> > > >
> > > > I wonder whether commit f9054c70d28bc214 ("mm, mempool: only set
> > > > __GFP_NOMEMALLOC if there are free elements") is doing correct thing.
> > > > It says
> > > >
> > > > If an oom killed thread calls mempool_alloc(), it is possible that
> > > > it'll
> > > > loop forever if there are no elements on the freelist since
> > > > __GFP_NOMEMALLOC prevents it from accessing needed memory reserves in
> > > > oom conditions.
> > >
> > > I haven't studied the patch very deeply so I might be missing something
> > > but from a quick look the patch does exactly what the above says.
> > >
> > > mempool_alloc used to inhibit ALLOC_NO_WATERMARKS by default. David has
> > > only changed that to allow ALLOC_NO_WATERMARKS if there are no objects
> > > in the pool and so we have no fallback for the default __GFP_NORETRY
> > > request.
> >
> > The swapper core sets the flag PF_MEMALLOC and calls generic_make_request
> > to submit the swapping bio to the block driver. The device mapper driver
> > uses mempools for all its I/O processing.
>
> 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.
What are the real problems that f9054c70d28bc214b2857cf8db8269f4f45a5e23
tries to fix?
Do you have a stacktrace where it deadlocked, or was just a theoretical
consideration?
Mempool users generally (except for some flawed cases like fs_bio_set) do
not require memory to proceed. So if you just loop in mempool_alloc, the
processes that exhasted the mempool reserve will eventually return objects
to the mempool and you should proceed.
If you can't proceed, it is a bug in the code that uses the mempool.
Mikulas
--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/dm-devel
