Re: [RFC-PATCH 1/2] mm: Add __GFP_NO_LOCKS flag

[Uladzislau Rezki <urezki@gmail.com>]

On Tue, Aug 11, 2020 at 12:28:18PM +0200, Michal Hocko wrote:
> On Tue 11-08-20 11:42:51, Uladzislau Rezki wrote:
> > On Tue, Aug 11, 2020 at 11:37:13AM +0200, Uladzislau Rezki wrote:
> > > On Tue, Aug 11, 2020 at 10:19:17AM +0200, Michal Hocko wrote:
> [...]
> > > > Anyway, if the zone->lock is not a good fit for raw_spin_lock then the
> > > > only way I can see forward is to detect real (RT) atomic contexts and
> > > > bail out early before taking the lock in the allocator for NOWAIT/ATOMIC
> > > > requests.
> > > >
> > This is similar what i have done with mm: Add __GFP_NO_LOCKS flag. I just
> > did it for order-0 pages(other paths are impossible) and made it common for
> > any kernel.
> > 
> > Because when you say "bail out early" i suspect that we would like to check
> > the per-cpu-list cache.
> 
> Bail out early means to do as much as possible until a raw non-compliant
> lock has to be taken.
> 

<snip>
struct page *rmqueue(struct zone *preferred_zone,
   struct zone *zone, unsigned int order,
   gfp_t gfp_flags, unsigned int alloc_flags,
   int migratetype)
{
 unsigned long flags;
 struct page *page;

 if (likely(order == 0)) {
  page = rmqueue_pcplist(preferred_zone, zone, gfp_flags,
     migratetype, alloc_flags);
  goto out;
 }

 /*
  * We most definitely don't want callers attempting to
  * allocate greater than order-1 page units with __GFP_NOFAIL.
  */
 WARN_ON_ONCE((gfp_flags & __GFP_NOFAIL) && (order > 1));
 spin_lock_irqsave(&zone->lock, flags);
<snip>

only order-0 allocations can be checked if CPUs pcp-list-cache has something.
I mean without taking any locks, i.e. it is lockless. "Pre-fetching" is not
possible since it takes zone->lock in order to do transfer pages from the buddy
to the per-cpu-lists. It is done in the rmqueue_bulk() function.

--
Vlad Rezki