2020-09-15 10:50:44

by Xu, Yanfei

Subject: [PATCH] mm/page_alloc.c: variable type of 'progress' should be 'unsigned long'

From: Yanfei Xu <[email protected]>

try_to_free_pages returns the number of pages reclaimed, and the type of
returns is 'unsigned long'. So we should use a matched type for storing
it.

Signed-off-by: Yanfei Xu <[email protected]>
---
mm/page_alloc.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index fab5e97dc9ca..5f1016c70b94 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -4245,9 +4245,8 @@ static int
__perform_reclaim(gfp_t gfp_mask, unsigned int order,
const struct alloc_context *ac)
{
- int progress;
unsigned int noreclaim_flag;
- unsigned long pflags;
+ unsigned long pflags, progress;

cond_resched();

--
2.18.2

2020-09-16 00:49:59

by Andrew Morton

Subject: Re: [PATCH] mm/page_alloc.c: variable type of 'progress' should be 'unsigned long'

On Tue, 15 Sep 2020 18:46:20 +0800 <[email protected]> wrote:

> From: Yanfei Xu <[email protected]>
>
> try_to_free_pages returns the number of pages reclaimed, and the type of
> returns is 'unsigned long'. So we should use a matched type for storing
> it.
>

__perform_reclaim() returns an int, so this change is fairly pointless.

However __perform_reclaim()'s single caller expects it to return
unsigned long, so including that change in this patch would make more
sense.

2020-09-16 01:58:04

by Xu, Yanfei

Subject: Re: [PATCH] mm/page_alloc.c: variable type of 'progress' should be 'unsigned long'

On 9/16/20 8:48 AM, Andrew Morton wrote:
> On Tue, 15 Sep 2020 18:46:20 +0800 <[email protected]> wrote:
>
>> From: Yanfei Xu <[email protected]>
>>
>> try_to_free_pages returns the number of pages reclaimed, and the type of
>> returns is 'unsigned long'. So we should use a matched type for storing
>> it.
>>
>
> __perform_reclaim() returns an int, so this change is fairly pointless.
>
> However __perform_reclaim()'s single caller expects it to return
> unsigned long, so including that change in this patch would make more
> sense.
>
Yeah, thanks for reminding. I will add that and send a v2. :)

Regards,
Yanfei