Re: [PATCH 11/23] media: atomisp: Remove test pattern generator (TPG) support

[Andy Shevchenko <andy.shevchenko@gmail.com>]

On Tue, Apr 16, 2024 at 03:40:43PM +0100, Kieran Bingham wrote:
> Quoting Kieran Bingham (2024-04-16 15:37:39)
> > Quoting Andy Shevchenko (2024-04-16 14:34:45)
> > > On Tue, Apr 16, 2024 at 12:25 PM Hans de Goede <hdegoede@redhat.com> wrote:
> > > > On 4/15/24 4:40 PM, Andy Shevchenko wrote:
> > > > > On Mon, Apr 15, 2024 at 02:02:08PM +0200, Hans de Goede wrote:

...

> > > > >>              unsigned int hblank_cycles = 100,
> > > > >>              vblank_lines = 6,
> > > > >>              width,
> > > > >
> > > > > Urgh... These comma operators probably is subject to replace with separate
> > > > > definitions or being grouped on a single line (as it suppose to be in this
> > > > > case).
> > > >
> > > > That indeed is ugly, but fixing this is out of scope for this patch,
> > > > so I've added an extra patch to the set to address this:
> > > >
> > > > https://git.kernel.org/pub/scm/linux/kernel/git/hansg/linux.git/commit/?h=media-atomisp&id=48d93af9d9b0fd40a21a656cb8cd8e25700bfed5
> > > 
> > > WFM, thanks!
> > > 
> > > Btw (yes, I see you tagged that already, so just FYI then)
> > > (1 + (stream->config.pixels_per_clock == 2));
> > > is an interesting way of using
> > > 
> > >  (stream->config.pixels_per_clock == 2) ? 2 : 1);
> > > 
> > > which likely can produce slightly better code (due to use of constant
> > > 2 twice), although it is a pure speculation by me.
> 
> Ooops, I didn't mean to send that. Well I wrote something but then
> closed it when I wasn't going to bother and instead some how sent an
> empty mail ..
> 
> So now I guess I should actually say what I wasn't going to end up
> saying? haha.
> 
> This was too intriguing (to me) so I threw it in godbolt, and for
> various levels of compiler optimsations, both options generate the same
> code, even with -O0. Of course the different optimsation levels produce
> different code, but the two options above always match at the given
> level.

Thanks for playing with this.

The issue with the original code is readability and potential oddity about
boolean to integer conversion. That said, I prefer my variant for the sake
of clarity.

> Fun with godbolt ;-)
> 
> https://godbolt.org/z/hYjrvK6hn

-- 
With Best Regards,
Andy Shevchenko