Re: [PATCH v2 2/9] block: introduce blk_init_flush and its pair

[Ming Lei]

On Fri, Sep 12, 2014 at 11:45 PM, Jens Axboe <axboe@xxxxxxxxx> wrote:
> On 2014-09-12 09:41, Ming Lei wrote:
>>
>> On Fri, Sep 12, 2014 at 11:18 PM, Jens Axboe <axboe@xxxxxxxxx> wrote:
>>>
>>> On 2014-09-12 08:47, Ming Lei wrote:
>>>>
>>>>
>>>> These two functions are introduced to initialize and de-initialize
>>>> flush stuff centrally.
>>>
>>>
>>>
>>> I know you said these change later to more proper naming, but that only
>>> happens further down. Lets get rid of these wrappers and just call
>>> blk_alloc_flush_queue() directly.
>>
>>
>> It is too early to call blk_alloc_flush_queue() because flush queue
>> doesn't
>> come until patch 5 appears, :-)  And I think it is cleaner to put
>> flush initialization
>> stuff together first, then introduce flush_queue.
>
>
> Then do it later. Fact is, final result still looks like this in
> blk_mq_init_queue():
>
> if (blk_init_flush(q))
>     ...
>
> where it would be cleaner as just assigning q->fq. My previous (and
> existing) point is that you have no idea what this init_flush() function
> does without jumping in and reading it.

That is a good point, and we can change to call blk_alloc_flush_queue()
directly in patch 5.

Thanks,
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/