[RFC] DT-based pinmux configuration for Renesas platforms (was: [PATCH v2 01/15] pinctrl: sh-pfc: Ad

[RFC] DT-based pinmux configuration for Renesas platforms (was: [PATCH v2 01/15] pinctrl: sh-pfc: Ad

[Laurent Pinchart]

Hi Geert,

On Tuesday 05 May 2015 11:07:20 Geert Uytterhoeven wrote:
> Hi Ulrich,
> 
> On Thu, Apr 30, 2015 at 4:18 PM, Ulrich Hecht
> 
> <ulrich.hecht+renesas@gmail.com> wrote:
> > Part of PFC support for R-Car M2-N (r8a7793) that defines the pinmux
> > data.
> 
> As this patch is difficult to quote, I'm just summarizing my comments:
>   - Other R-Car Gen2 variants use "MLB_CLK" instead of "MLB_CK",
>   - Compared to M2-W, CAN0/1, HSCIF2_D, and MLB+ are missing,
>     but they do exist on M2-N,
>   - Compared to M2-W, SSP is added, but it does exist on M2-W.
> 
> After compensating for that, pfc-r8a7793.c is identical to pfc-r8a7791.c.
> So I think we should use a single file for both, or at least share (most of)
> the data structures (sh_pfc_soc_info contains the SoC name, so it can't
> be shared).

I agree with that, and I'd even go further : I think we won't be able to scale 
much further with our large in-kernel tables. We currently have around 150kB 
or PFC .rodata when enabling all Renesas platforms, which is far from being 
negligible in a multiplatform build. I could be time to reconsider the 
approach and specify data in DT.

> BTW, it's my understanding M2-W and M2-N are identical, except for the
> wide/double vs. narrow/single memory channels, which doesn't impact
> configurable pins. Please correct me if I'm wrong.
> Are there other differences?

-- 
Regards,

Laurent Pinchart

Re: [RFC] DT-based pinmux configuration for Renesas platforms (was: [PATCH v2 01/15] pinctrl: sh-pfc

[Geert Uytterhoeven]

Hi Laurent,

On Wed, May 6, 2015 at 6:15 AM, Laurent Pinchart
<laurent.pinchart@ideasonboard.com> wrote:
> On Tuesday 05 May 2015 11:07:20 Geert Uytterhoeven wrote:
>> On Thu, Apr 30, 2015 at 4:18 PM, Ulrich Hecht
>>
>> <ulrich.hecht+renesas@gmail.com> wrote:
>> > Part of PFC support for R-Car M2-N (r8a7793) that defines the pinmux
>> > data.
>>
>> As this patch is difficult to quote, I'm just summarizing my comments:
>>   - Other R-Car Gen2 variants use "MLB_CLK" instead of "MLB_CK",
>>   - Compared to M2-W, CAN0/1, HSCIF2_D, and MLB+ are missing,
>>     but they do exist on M2-N,
>>   - Compared to M2-W, SSP is added, but it does exist on M2-W.
>>
>> After compensating for that, pfc-r8a7793.c is identical to pfc-r8a7791.c.
>> So I think we should use a single file for both, or at least share (most of)
>> the data structures (sh_pfc_soc_info contains the SoC name, so it can't
>> be shared).
>
> I agree with that, and I'd even go further : I think we won't be able to scale
> much further with our large in-kernel tables. We currently have around 150kB
> or PFC .rodata when enabling all Renesas platforms, which is far from being
> negligible in a multiplatform build. I could be time to reconsider the

I've been thinking about compressing it using zlib/lzo/lz4, and only
decompressing the data for the platform you're running on.

> approach and specify data in DT.

Ohoh... ;-)

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds