[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <nycvar.YSQ.7.76.1710031638450.5407@knanqh.ubzr>
Date: Tue, 3 Oct 2017 16:57:44 -0400 (EDT)
From: Nicolas Pitre <nicolas.pitre@...aro.org>
To: Tejun Heo <tj@...nel.org>, Christoph Lameter <cl@...ux.com>
cc: linux-mm@...ck.org, linux-kernel@...r.kernel.org
Subject: [PATCH] mm/percpu.c: use smarter memory allocation for struct
pcpu_alloc_info
This can be much smaller than a page on very small memory systems.
Always rounding up the size to a page is wasteful in that case, and
required alignment is smaller than the memblock default. Let's round
things up to a page size only when the actual size is >= page size, and
then it makes sense to page-align for a nicer allocation pattern.
Signed-off-by: Nicolas Pitre <nico@...aro.org>
diff --git a/mm/percpu.c b/mm/percpu.c
index 434844415d..fe37f85cc2 100644
--- a/mm/percpu.c
+++ b/mm/percpu.c
@@ -1410,13 +1410,17 @@ struct pcpu_alloc_info * __init pcpu_alloc_alloc_info(int nr_groups,
struct pcpu_alloc_info *ai;
size_t base_size, ai_size;
void *ptr;
- int unit;
+ int unit, align;
- base_size = ALIGN(sizeof(*ai) + nr_groups * sizeof(ai->groups[0]),
- __alignof__(ai->groups[0].cpu_map[0]));
+ align = __alignof__(ai->groups[0].cpu_map[0]);
+ base_size = ALIGN(sizeof(*ai) + nr_groups * sizeof(ai->groups[0]), align);
ai_size = base_size + nr_units * sizeof(ai->groups[0].cpu_map[0]);
+ if (ai_size >= PAGE_SIZE) {
+ ai_size = PFN_ALIGN(ai_size);
+ align = PAGE_SIZE;
+ }
- ptr = memblock_virt_alloc_nopanic(PFN_ALIGN(ai_size), 0);
+ ptr = memblock_virt_alloc_nopanic(ai_size, align);
if (!ptr)
return NULL;
ai = ptr;
@@ -1428,7 +1432,7 @@ struct pcpu_alloc_info * __init pcpu_alloc_alloc_info(int nr_groups,
ai->groups[0].cpu_map[unit] = NR_CPUS;
ai->nr_groups = nr_groups;
- ai->__ai_size = PFN_ALIGN(ai_size);
+ ai->__ai_size = ai_size;
return ai;
}
Powered by blists - more mailing lists