[Devel] [PATCH rh7 v2 1/3] ve/kstat/alloc_lat: Don't separate GFP_HIGHMEM and !GFP_HIGHMEM allocation latencies

Andrey Ryabinin aryabinin at virtuozzo.com
Thu Feb 15 17:48:13 MSK 2018


We use mostly 64-bit systems this days. Since they don't have higmem
it's better to not segregate GFP_HIGHMEM and !GFP_HIGHMEM latencies.
For backward compatibility we still output alochigh/alochighmp fields
in /proc/vz/latency but show only zeroes.

https://jira.sw.ru/browse/PSBM-81395
Signed-off-by: Andrey Ryabinin <aryabinin at virtuozzo.com>
---
 mm/page_alloc.c | 7 +------
 1 file changed, 1 insertion(+), 6 deletions(-)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index cd8ed1f5543e..f08ddeb77b82 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -3147,16 +3147,11 @@ static void __alloc_collect_stats(gfp_t gfp_mask, unsigned int order,
 	time = jiffies_to_usecs(jiffies - time) * 1000;
 	if (!(gfp_mask & __GFP_WAIT))
 		ind = KSTAT_ALLOCSTAT_ATOMIC;
-	else if (!(gfp_mask & __GFP_HIGHMEM))
+	else
 		if (order > 0)
 			ind = KSTAT_ALLOCSTAT_LOW_MP;
 		else
 			ind = KSTAT_ALLOCSTAT_LOW;
-	else
-		if (order > 0)
-			ind = KSTAT_ALLOCSTAT_HIGH_MP;
-		else
-			ind = KSTAT_ALLOCSTAT_HIGH;
 
 	local_irq_save(flags);
 	cpu = smp_processor_id();
-- 
2.16.1



More information about the Devel mailing list