Re:Re: [PATCH] blk: use REQ_OP_WRITE instead of hard code

2020-10-19 Thread 苏辉 
Yeah, you are right, thanks for your explanationMaybe we should 
define a MASK to do this?
At 2020-10-19 22:09:40, "Johannes Thumshirn"  wrote:
>On 19/10/2020 16:06, Hui Su wrote:
>> use REQ_OP_WRITE instead of hard code in
>> op_is_write().
>> 
>> Signed-off-by: Hui Su 
>> ---
>>  include/linux/blk_types.h | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>> 
>> diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
>> index 7d7c13238fdb..7b9b02378c24 100644
>> --- a/include/linux/blk_types.h
>> +++ b/include/linux/blk_types.h
>> @@ -440,7 +440,7 @@ static inline void bio_set_op_attrs(struct bio *bio, 
>> unsigned op,
>>  
>>  static inline bool op_is_write(unsigned int op)
>>  {
>> -return (op & 1);
>> +return (op & REQ_OP_WRITE);
>
>op_is_write() returns true for every req_op that writes to a device (they all 
>have 
>the lowest bit set), while REQ_OP_WRITE means a WRITE. You'll change the 
>semantics
>with this patch.

Re: [PATCH] blk: use REQ_OP_WRITE instead of hard code

2020-10-19 Thread Johannes Thumshirn 
On 19/10/2020 16:30, 苏辉 wrote:
> Yeah, you are right, thanks for your explanationMaybe we should 
> define a MASK to do this?

Why? I personally find a '& 1' way more understandable than a 
REQ_OP_IS_WRITE_MASK or sth like that. The former I can just read,
for the latter I would need to look up the definition to be able to
understand the code.

Re: [PATCH] blk: use REQ_OP_WRITE instead of hard code

2020-10-19 Thread Johannes Thumshirn 
On 19/10/2020 16:06, Hui Su wrote:
> use REQ_OP_WRITE instead of hard code in
> op_is_write().
> 
> Signed-off-by: Hui Su 
> ---
>  include/linux/blk_types.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
> index 7d7c13238fdb..7b9b02378c24 100644
> --- a/include/linux/blk_types.h
> +++ b/include/linux/blk_types.h
> @@ -440,7 +440,7 @@ static inline void bio_set_op_attrs(struct bio *bio, 
> unsigned op,
>  
>  static inline bool op_is_write(unsigned int op)
>  {
> - return (op & 1);
> + return (op & REQ_OP_WRITE);

op_is_write() returns true for every req_op that writes to a device (they all 
have 
the lowest bit set), while REQ_OP_WRITE means a WRITE. You'll change the 
semantics
with this patch.

[PATCH] blk: use REQ_OP_WRITE instead of hard code

2020-10-19 Thread Hui Su 
use REQ_OP_WRITE instead of hard code in
op_is_write().

Signed-off-by: Hui Su 
---
 include/linux/blk_types.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index 7d7c13238fdb..7b9b02378c24 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -440,7 +440,7 @@ static inline void bio_set_op_attrs(struct bio *bio, 
unsigned op,
 
 static inline bool op_is_write(unsigned int op)
 {
-   return (op & 1);
+   return (op & REQ_OP_WRITE);
 }
 
 /*
-- 
2.25.1