Re: Situation of CONFIG_BLK_WBT_MQ after commit b11d31ae01e6 ("blk-wbt: remove unnecessary check in wbt_enable_default()")

[Lukas Bulwahn]

On Wed, May 10, 2023 at 5:24 AM Yu Kuai <yukuai1@xxxxxxxxxxxxxxx> wrote:
>
> Hi,
>
> 在 2023/05/08 12:37, Lukas Bulwahn 写道:
> > Dear Yu Kuai, dear Christoph, dear Jens,
> >
> >
> > The commit b11d31ae01e6 ("blk-wbt: remove unnecessary check in
> > wbt_enable_default()") removes the only reference to the config
> > BLK_WBT_MQ in the kernel tree.
> >
> > The commit comes with the statement "If CONFIG_BLK_WBT_MQ is disabled,
> > wbt_init() won't do anything.". The statement "If CONFIG_BLK_WBT is
> > disabled, wbt_init() won't do anything." (note: CONFIG_BLK_WBT vs.
> > CONFIG_BLK_WBT_MQ) is certainly true, but I do not see that "If
> > CONFIG_BLK_WBT_MQ is disabled, wbt_init() won't do anything.", but I
> > believe it would simply do what wbt_init() does with CONFIG_BLK_WBT
> > being enabled.
> >
> > Now, it seems that with this commit applied, the intended switch of
> > the config BLK_WBT_MQ is gone, and the config really now has no effect
> > at all.
> >
> > So, I am a bit puzzled:
> >
> > 1. Either the config BLK_WBT_MQ does have an effect somewhere, but one
> > cannot find its reference with 'git grep -i "BLK_WBT_MQ" .' --- so, my
> > investigation is just incomplete or wrong, or
> >
> > 2. We really do not need this further build config BLK_WBT_MQ beyond
> > the other configs already there --- then this config should just be
> > removed, or
> >
> > 3. the commit unintentionally broke the purpose of the config
> > BLK_WBT_MQ --- then this commit above should be reverted.
>
> Thanks for the report, it's the above case and it's my mistake.
> I will fix this.
>

Kuai, thanks for letting me know. Feel free to add a "Reported-by:
Lukas Bulwahn <lukas.bulwahn@xxxxxxxxx>" tag, and to include in the
list of recipients when you send out the fix to the mailing list.

I am glad to see that my investigation of kernel code is finding some
issues that we can fix.

Lukas