thp: cleanup how khugepaged enters freezer
authorJiri Kosina <jkosina@suse.cz>
Wed, 24 Jun 2015 23:56:07 +0000 (16:56 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Thu, 25 Jun 2015 00:49:41 +0000 (17:49 -0700)
khugepaged_do_scan() checks in every iteration whether freezing(current)
is true, and in such case breaks out of the loop, which causes
try_to_freeze() to be called immediately afterwards in
khugepaged_wait_work().

If nothing else, this causes unnecessary freezing(current) test, and also
makes the way khugepaged enters freezer a bit less obvious than necessary.

Let's just try to freeze directly, instead of splitting it into two
(directly adjacent) phases.

Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/huge_memory.c

index 078832cf3636579e1d9cc1c396b037319e52a6bd..b3d8cd8d6968e752676598ade11cb996fc84d48e 100644 (file)
@@ -2799,7 +2799,7 @@ static void khugepaged_do_scan(void)
 
                cond_resched();
 
-               if (unlikely(kthread_should_stop() || freezing(current)))
+               if (unlikely(kthread_should_stop() || try_to_freeze()))
                        break;
 
                spin_lock(&khugepaged_mm_lock);
@@ -2820,8 +2820,6 @@ static void khugepaged_do_scan(void)
 
 static void khugepaged_wait_work(void)
 {
-       try_to_freeze();
-
        if (khugepaged_has_work()) {
                if (!khugepaged_scan_sleep_millisecs)
                        return;