Rather than making it safe to call get_online_cpus() from the page
allocator, this patch simply removes the page allocator call to
drain_all_pages(). To avoid impacting high-order allocation success
rates, it still drains the local per-cpu lists for high-order
allocations that failed. As a side effect, this reduces the number
of IPIs sent during low memory situations.

Signed-off-by: Mel Gorman <mgorman@suse.de>
---
 mm/page_alloc.c |   16 ++++++++++++----
 1 files changed, 12 insertions(+), 4 deletions(-)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 2b8ba3a..b6df6fc 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c

@@ -1119,7 +1119,9 @@ void drain_local_pages(void *arg)

 */
 void drain_all_pages(void)
 {
+       get_online_cpus();
       on_each_cpu(drain_local_pages, NULL, 1);
+       put_online_cpus();
 }

 #ifdef CONFIG_HIBERNATION

@@ -1982,11 +1984,17 @@ retry:

                                       migratetype);

       /*
-        * If an allocation failed after direct reclaim, it could be because
-        * pages are pinned on the per-cpu lists. Drain them and try again
+        * If a high-order allocation failed after direct reclaim, there is a
+        * possibility that it is because the necessary buddies have been
+        * freed to the per-cpu list. Drain the local list and try again.
+        * drain_all_pages is not used because it is unsafe to call
+        * get_online_cpus from this context as it is possible that kthreadd
+        * would block during thread creation and the cost of sending storms
+        * of IPIs in low memory conditions is quite high.
        */
-       if (!page && !drained) {
-               drain_all_pages();
+       if (!page && order && !drained) {
+               drain_pages(get_cpu());
+               put_cpu();
               drained = true;
               goto retry;
       }
--
1.7.3.4