Re: [PATCH 1/2] nfit: Use init_completion() in acpi_nfit_flush_probe()

[Arnd Bergmann]

On Thu, Aug 24, 2017 at 3:28 PM, Boqun Feng <boqun.feng@xxxxxxxxx> wrote:
> On Thu, Aug 24, 2017 at 03:07:42PM +0200, Thomas Gleixner wrote:
>> On Wed, 23 Aug 2017, Boqun Feng wrote:
>>
>> > There is no need to use COMPLETION_INITIALIZER_ONSTACK() in
>> > acpi_nfit_flush_probe(), replace it with init_completion().
>>
>> You completely fail to explain WHY.
>>
>
> I thought COMPLETION_INITIALIZER_ONSTACK() should only use in assigment
> or compound literals, so the usage here is obviously wrong, but seems
> I was wrong?
>
> Ingo,
>
> Is the usage of COMPLETION_INITIALIZER_ONSTACK() correct? If not,
> I could rephrase my commit log saying this is a fix for wrong usage of
> COMPLETION_INITIALIZER_ONSTACK(), otherwise, I will rewrite the commit
> indicating this patch is a necessary dependency for patch #2. Thanks!

I think your patch is correct, but your changelog text is useless, as
Thomas mentioned: you should instead explain that it breaks with the
other fix in the series, and what the difference between init_completion()
and COMPLETION_INITIALIZER_ONSTACK() is.

      Arnd