[PATCH 1/1] block/bio.c: use GFP_NOFS

[PATCH 1/1] block/bio.c: use GFP_NOFS

[Fabian Frederick]

Use GFP_NOFS instead of its definition.

Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Fabian Frederick <fabf@skynet.be>
---
 block/bio.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/block/bio.c b/block/bio.c
index 8c2e55e..ec5d172 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -211,7 +211,7 @@ fallback:
 		bvl = mempool_alloc(pool, gfp_mask);
 	} else {
 		struct biovec_slab *bvs = bvec_slabs + *idx;
-		gfp_t __gfp_mask = gfp_mask & ~(__GFP_WAIT | __GFP_IO);
+		gfp_t __gfp_mask = gfp_mask & ~GFP_NOFS;
 
 		/*
 		 * Make this allocation restricted and don't dump info on
-- 
1.8.4.5

Re: [PATCH 1/1] block/bio.c: use GFP_NOFS

[Hugh Dickins]

On Sat, 14 Jun 2014, Fabian Frederick wrote:

> Use GFP_NOFS instead of its definition.
> 
> Cc: Jens Axboe <axboe@kernel.dk>
> Signed-off-by: Fabian Frederick <fabf@skynet.be>
> ---
>  block/bio.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/block/bio.c b/block/bio.c
> index 8c2e55e..ec5d172 100644
> --- a/block/bio.c
> +++ b/block/bio.c
> @@ -211,7 +211,7 @@ fallback:
>  		bvl = mempool_alloc(pool, gfp_mask);
>  	} else {
>  		struct biovec_slab *bvs = bvec_slabs + *idx;
> -		gfp_t __gfp_mask = gfp_mask & ~(__GFP_WAIT | __GFP_IO);
> +		gfp_t __gfp_mask = gfp_mask & ~GFP_NOFS;
>  
>  		/*
>  		 * Make this allocation restricted and don't dump info on
> -- 
> 1.8.4.5

Please no.  The original was fine, this just makes it harder to read.

GFP_NOFS is a useful and widely used combination of flags for passing
down to the page allocator; but it is merely obscure when used as you
have for masking bits off.

(But I confess to hypocrisy in finding "& ~GFP_KERNEL" useful myself
in the past somewhere.)

Hugh

Re: [PATCH 1/1] block/bio.c: use GFP_NOFS

[Jens Axboe]

On 2014-06-15 18:28, Hugh Dickins wrote:
> On Sat, 14 Jun 2014, Fabian Frederick wrote:
>
>> Use GFP_NOFS instead of its definition.
>>
>> Cc: Jens Axboe <axboe@kernel.dk>
>> Signed-off-by: Fabian Frederick <fabf@skynet.be>
>> ---
>>   block/bio.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/block/bio.c b/block/bio.c
>> index 8c2e55e..ec5d172 100644
>> --- a/block/bio.c
>> +++ b/block/bio.c
>> @@ -211,7 +211,7 @@ fallback:
>>   		bvl = mempool_alloc(pool, gfp_mask);
>>   	} else {
>>   		struct biovec_slab *bvs = bvec_slabs + *idx;
>> -		gfp_t __gfp_mask = gfp_mask & ~(__GFP_WAIT | __GFP_IO);
>> +		gfp_t __gfp_mask = gfp_mask & ~GFP_NOFS;
>>
>>   		/*
>>   		 * Make this allocation restricted and don't dump info on
>> --
>> 1.8.4.5
>
> Please no.  The original was fine, this just makes it harder to read.

I agree. GFP_NOFS is useful when it's used as a whole, but for this use 
case, explicitly saying that we want to wait for mem and we allow IO is 
much clearer. The fact that this just happens to be GFP_NOFS doesn't 
really matter.

> (But I confess to hypocrisy in finding "& ~GFP_KERNEL" useful myself
> in the past somewhere.)

That's just sick, Hugh :-)

-- 
Jens Axboe