Re: [PATCH -next] blk-mq: Use BUG_ON() instead of BUG()

[Jens Axboe]

On 4/29/20 1:26 AM, Christoph Hellwig wrote:
> On Wed, Apr 29, 2020 at 10:10:24AM +0800, Zou Wei wrote:
>> Fixes coccicheck warning:
>>
>>  block/blk-mq.c:546:2-5: WARNING: Use BUG_ON instead of if condition followed by BUG.
>>
>> Fixes: 63151a449eba ("blk-mq: allow drivers to hook into I/O completion")
>> Reported-by: Hulk Robot <hulkci@xxxxxxxxxx>
>> Signed-off-by: Zou Wei <zou_wei@xxxxxxxxxx>
>> ---
>>  block/blk-mq.c | 3 +--
>>  1 file changed, 1 insertion(+), 2 deletions(-)
>>
>> diff --git a/block/blk-mq.c b/block/blk-mq.c
>> index bcc3a23..49a227e 100644
>> --- a/block/blk-mq.c
>> +++ b/block/blk-mq.c
>> @@ -542,8 +542,7 @@ EXPORT_SYMBOL(__blk_mq_end_request);
>>  
>>  void blk_mq_end_request(struct request *rq, blk_status_t error)
>>  {
>> -	if (blk_update_request(rq, error, blk_rq_bytes(rq)))
>> -		BUG();
>> +	BUG_ON(blk_update_request(rq, error, blk_rq_bytes(rq)));
> 
> I don't think hiding something that actually does do the work in a
> BUG_ON ever is a good style.

Agree, it's a lot less readable. And, not that we've ever done that, but
also fragile in a lot of code bases where a non-debug build would turn
off the BUG_ON() equivalent, and hence never call blk_update_request().
So not a good practice anywhere for statements that have side effects.

-- 
Jens Axboe