lkml.org 
[lkml]   [2015]   [Apr]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
Subject[PATCH] thp: cleanup how khugepaged enters freezer
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>
---

Stumbled upon this when debugging something completely unrelated.

mm/huge_memory.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 078832c..b3d8cd8 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -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;
--
Jiri Kosina
SUSE Labs


\
 
 \ /
  Last update: 2015-04-17 21:41    [W:0.028 / U:5.072 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site