2018-04-09 14:15:06

by Jia-Ju Bai

[permalink] [raw]
Subject: [PATCH] net: dccp: Replace GFP_ATOMIC with GFP_KERNEL in dccp_init

dccp_init() is never called in atomic context.
This function is only set as a parameter of module_init().

Despite never getting called from atomic context,
dccp_init() calls __get_free_pages() with GFP_ATOMIC,
which waits busily for allocation.
GFP_ATOMIC is not necessary and can be replaced with GFP_KERNEL,
to avoid busy waiting and improve the possibility of sucessful allocation.

This is found by a static analysis tool named DCNS written by myself.
And I also manually check it.

Signed-off-by: Jia-Ju Bai <[email protected]>
---
net/dccp/proto.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/dccp/proto.c b/net/dccp/proto.c
index b68168f..f63ba93 100644
--- a/net/dccp/proto.c
+++ b/net/dccp/proto.c
@@ -1159,7 +1159,7 @@ static int __init dccp_init(void)
hash_size--;
dccp_hashinfo.ehash_mask = hash_size - 1;
dccp_hashinfo.ehash = (struct inet_ehash_bucket *)
- __get_free_pages(GFP_ATOMIC|__GFP_NOWARN, ehash_order);
+ __get_free_pages(GFP_KERNEL|__GFP_NOWARN, ehash_order);
} while (!dccp_hashinfo.ehash && --ehash_order > 0);

if (!dccp_hashinfo.ehash) {
@@ -1182,7 +1182,7 @@ static int __init dccp_init(void)
bhash_order > 0)
continue;
dccp_hashinfo.bhash = (struct inet_bind_hashbucket *)
- __get_free_pages(GFP_ATOMIC|__GFP_NOWARN, bhash_order);
+ __get_free_pages(GFP_KERNEL|__GFP_NOWARN, bhash_order);
} while (!dccp_hashinfo.bhash && --bhash_order >= 0);

if (!dccp_hashinfo.bhash) {
--
1.9.1



2018-04-09 14:46:44

by Eric Dumazet

[permalink] [raw]
Subject: Re: [PATCH] net: dccp: Replace GFP_ATOMIC with GFP_KERNEL in dccp_init



On 04/09/2018 07:10 AM, Jia-Ju Bai wrote:
> dccp_init() is never called in atomic context.
> This function is only set as a parameter of module_init().
>
> Despite never getting called from atomic context,
> dccp_init() calls __get_free_pages() with GFP_ATOMIC,
> which waits busily for allocation.

What do you mean by "waits busily" ?

GFP_ATOMIC does not sleep, does not wait.

> GFP_ATOMIC is not necessary and can be replaced with GFP_KERNEL,
> to avoid busy waiting and improve the possibility of sucessful allocation.
>
> This is found by a static analysis tool named DCNS written by myself.
> And I also manually check it.
>
> Signed-off-by: Jia-Ju Bai <[email protected]>
> ---



2018-04-09 14:54:07

by Jia-Ju Bai

[permalink] [raw]
Subject: Re: [PATCH] net: dccp: Replace GFP_ATOMIC with GFP_KERNEL in dccp_init



On 2018/4/9 22:42, Eric Dumazet wrote:
>
> On 04/09/2018 07:10 AM, Jia-Ju Bai wrote:
>> dccp_init() is never called in atomic context.
>> This function is only set as a parameter of module_init().
>>
>> Despite never getting called from atomic context,
>> dccp_init() calls __get_free_pages() with GFP_ATOMIC,
>> which waits busily for allocation.
> What do you mean by "waits busily" ?
>
> GFP_ATOMIC does not sleep, does not wait.

Sorry, I should modify it to "does not sleep".
Do you think it is okay?


Best wishes,
Jia-Ju Bai