Re: [PATCH v3] media: rc: pwm-ir-tx: Switch to atomic PWM API

[Sean Young <sean@mess.org>]

Hi Maíra,

On Wed, Oct 27, 2021 at 09:43:47AM -0300, Maíra Canal wrote:
> [resend it in Plain Text]
> Thank you for the feedback! I appreciate that! I'm new at the kernel
> and I got a little confused about how to send the new patch. Should I
> send a v4 of this patch or just send a new patch fixing this issue?
> I'm sorry about the question and thank you for your attention.

Please send out a v4 with the problem fixed.

Also top-posting is deprecated on linux mailing lists.

Thanks,

Sean

> > Em qua., 27 de out. de 2021 às 04:32, Sean Young <sean@mess.org> escreveu:
> >>
> >> On Wed, Oct 27, 2021 at 08:15:52AM +0200, Uwe Kleine-König wrote:
> >> > On Wed, Oct 27, 2021 at 02:07:19PM +0800, kernel test robot wrote:
> >> > > If you fix the issue, kindly add following tag as appropriate
> >> > > Reported-by: kernel test robot <lkp@intel.com>
> >> > >
> >> > > All errors (new ones prefixed by >>, old ones prefixed by <<):
> >> > >
> >> > > >> ERROR: modpost: "__udivdi3" [drivers/media/rc/pwm-ir-tx.ko] undefined!
> >> >
> >> > This comes from the line:
> >> >
> >> >       state.duty_cycle = DIV_ROUND_CLOSEST(pwm_ir->duty_cycle * state.period, 100);
> >> >
> >> > where DIV_ROUND_CLOSEST expands to a normal division but state.period is
> >> > a u64. So this should use DIV64_U64_ROUND_CLOSEST I guess.
> >>
> >> DIV64_U64_ROUND_CLOSEST is for dividing a u64 with a u64. We're dividing
> >> by 100 here so this is not necessary.
> >>
> >> It should use DIV_ROUND_CLOSEST_ULL, however it might be nicer to use:
> >>
> >>         pwm_set_relative_duty_cycle(&state, pwm_ir->duty_cycle, 100);
> >>
> >> Thanks
> >>
> >> Sean