From: Michal Hocko <mhocko@xxxxxxxx>
should_reclaim_retry() should be a natural reschedule point. PF_WQ_WORKER
is a special case which needs a stronger rescheduling policy. However,
since schedule_timeout_uninterruptible(1) for PF_WQ_WORKER depends on
__zone_watermark_ok() == true, PF_WQ_WORKER is currently counting on
mutex_trylock(&oom_lock) == 0 in __alloc_pages_may_oom() which is a bad
expectation.
Doing schedule_timeout_uninterruptible(1) at should_reclaim_retry()
unconditionally seems more straightforward than depending on a zone being
a good candidate for a further reclaim.
Signed-off-by: Michal Hocko <mhocko@xxxxxxxx>
Cc: Hillf Danton <hillf.zj@xxxxxxxxxxxxxxx>
Cc: David Rientjes <rientjes@xxxxxxxxxx>
Cc: Johannes Weiner <hannes@xxxxxxxxxxx>
Cc: Joonsoo Kim <js1304@xxxxxxxxx>
Cc: Mel Gorman <mgorman@xxxxxxx>
Cc: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
Cc: Vladimir Davydov <vdavydov@xxxxxxxxxxxxx>
Cc: Vlastimil Babka <vbabka@xxxxxxx>
---
mm/page_alloc.c | 34 ++++++++++++++++++----------------
1 file changed, 18 insertions(+), 16 deletions(-)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 1521100..f56cc09 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -3922,6 +3922,7 @@ bool gfp_pfmemalloc_allowed(gfp_t gfp_mask)
{
struct zone *zone;
struct zoneref *z;
+ bool ret = false;
/*
* Costly allocations might have made a progress but this doesn't mean
@@ -3985,25 +3986,26 @@ bool gfp_pfmemalloc_allowed(gfp_t gfp_mask)
}
}
- /*
- * Memory allocation/reclaim might be called from a WQ
- * context and the current implementation of the WQ
- * concurrency control doesn't recognize that
- * a particular WQ is congested if the worker thread is
- * looping without ever sleeping. Therefore we have to
- * do a short sleep here rather than calling
- * cond_resched().
- */
- if (current->flags & PF_WQ_WORKER)
- schedule_timeout_uninterruptible(1);
- else
- cond_resched();
-
- return true;
+ ret = true;
+ goto out;
}
}
- return false;
+out:
+ /*
+ * Memory allocation/reclaim might be called from a WQ
+ * context and the current implementation of the WQ
+ * concurrency control doesn't recognize that
+ * a particular WQ is congested if the worker thread is
+ * looping without ever sleeping. Therefore we have to
+ * do a short sleep here rather than calling
+ * cond_resched().
+ */
+ if (current->flags & PF_WQ_WORKER)
+ schedule_timeout_uninterruptible(1);
+ else
+ cond_resched();
+ return ret;
}
static inline bool
--
1.8.3.1
