Re: [PATCH 02/11] blk: make the bioset rescue_workqueue optional.

[NeilBrown <neilb@suse.com>]

[-- Attachment #1: Type: text/plain, Size: 1638 bytes --]

On Fri, Apr 21 2017, Christoph Hellwig wrote:

> On Thu, Apr 20, 2017 at 03:38:48PM +1000, NeilBrown wrote:
>> This patch converts bioset_create() and
>> bioset_create_nobvec() to not create a workqueue so
>> alloctions will never trigger punt_bios_to_rescuer().  It
>> also introduces bioset_create_rescued() and
>> bioset_create_nobvec_rescued() which preserve the old
>> behaviour.
>
> Why these super-early line breaks in the committ message?  They make
> the text much more awkware compared to say:

I usually set a smaller wrap-margin for git comments because of the
extra space that git inserts on the left.  Maybe I over-do it.

>
> This patch converts bioset_create() and bioset_create_nobvec() to not
> create a workqueue so alloctions will never trigger
> punt_bios_to_rescuer(). It also introduces bioset_create_rescued() and
> bioset_create_nobvec_rescued() which preserve the old behaviour.
>
>>  static struct bio_set *__bioset_create(unsigned int pool_size,
>>  				       unsigned int front_pad,
>> -				       bool create_bvec_pool)
>> +				       bool create_bvec_pool,
>> +				       bool create_rescue_workqueue)
>
> I'd much prefer a single new bioset_create with a bunch of flags
> arguments over the number of new functions and these bool arguments.

I was following the existing practice exemplified by
bioset_create_nobvec().
By not changing the signature of the function, I can avoid touching
quite a few places where it is called.
I hope to get rid of the _rescued() versions eventually.  That is easier
if there are a separate function rather than an extra arg that needs
to be removed everywhere.

NeilBrown

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]