Re: [RFC PATCH 0/8] block: fix bio_add_XXX_page() return type

[Chaitanya Kulkarni <Chaitanya.Kulkarni@wdc.com>]

On 5/21/21 15:37, Omar Sandoval wrote:
> On Fri, May 21, 2021 at 09:37:43PM +0000, Chaitanya Kulkarni wrote:
>> On 5/21/21 03:25, Johannes Thumshirn wrote:
>>> I couldn't spot any errors, but I'm not sure it's worth the effort.
>>>
>>> If Jens decides to take it:
>>> Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
>>>
>> It does create confusion on the code level which can result in
>> invalid error checks.
> Do you have any examples of bugs caused by this confusion (whether they
> were fixed in the past, currently exist, or were caught in code review)?
> That would be good justification for doing this.
>

Yes, while implementing the ZBD over NVMeOF I accidentally ended up checking
the error value < 0 based on the function following declaration :-

  1 F   f    bio_add_zone_append_page  block/bio.c
  int bio_add_zone_append_page(struct bio *bio, struct page *page,

I did fix it later when doing the next code review myself, maybe that is
just
me :P.

Also, new functions inherit the same style such as
bio_add_zone_append_page() from bio_add_page() and bio_add_pc_page() we
can prevent that.

What you do you think about the approach suggested by Matthew  size_t ?


If it is too much trouble we can drop it.