Re: [PATCH 5/6] tty: serial: Add linflexuart driver for S32V234

From: Will Deacon
Date: Fri Aug 09 2019 - 05:53:47 EST


On Fri, Aug 09, 2019 at 09:45:40AM +0000, Stefan-gabriel Mirea wrote:
> On 8/8/2019 8:17 PM, Will Deacon wrote:
> > On Thu, Aug 08, 2019 at 12:47:00PM +0000, Stefan-gabriel Mirea wrote:
> >> On 8/8/2019 11:08 AM, Will Deacon wrote:
> >>> On Fri, Aug 02, 2019 at 07:47:23PM +0000, Stefan-gabriel Mirea wrote:
> >>>> + linflex,<addr>
> >>>> + Use early console provided by Freescale LinFlex UART
> >>>> + serial driver for NXP S32V234 SoCs. A valid base
> >>>> + address must be provided, and the serial port must
> >>>> + already be setup and configured.
> >>>
> >>> Why isn't earlycon= sufficient for this?
> >>
> >> "earlycon=" is not actually supported. I will fix this in the next
> >> version by adding a /chosen/stdout-path to the dts. The compatible
> >> string provided to OF_EARLYCON_DECLARE will also be changed from
> >> "fsl,s32v234-linflexuart" to "fsl,s32-linflexuart" to match the one in
> >> the device tree nodes. I missed this after importing a rename from our
> >> codebase.
> >>
> >> Should I remove this addition from kernel-parameters.txt after that?
> >
> > Yes, if you can use earlycon instead, then you can drop your custom option
> > entirely and therefore there's no need to document it either.
>
> Can you please clarify what you mean by "drop your custom option"? The
> "linflex" option documented in the paragraph is the name of the
> earlycon_id declared via "OF_EARLYCON_DECLARE(linflex, ...)". We have
> not done anything particular to accept it in the parameter value.

My apologies, I completely misread your patch and thought your were adding
a *new* kernel parameter. Having looked at the context, I now see you're
adding a new earlycon option, which is what I was suggesting :)

By all means, please update the stdout-path, because then you can just pass
"earlycon" and things will work as expected.

Will