Re: [PATCH 20/24] ASoC: codecs: rt1015p: Update definition of device_id tables

[Cezary Rojewski <cezary.rojewski@xxxxxxxxx>]

On 2025-02-27 6:11 AM, Oder Chiou wrote:
> > -----Original Message-----
> > From: Cezary Rojewski <cezary.rojewski@xxxxxxxxx>
> > Sent: Wednesday, February 26, 2025 5:42 PM
> > To: Oder Chiou <oder_chiou@xxxxxxxxxxx>; broonie@xxxxxxxxxx
> > Cc: tiwai@xxxxxxxx; perex@xxxxxxxx; amadeuszx.slawinski@xxxxxxxxxxxxxxx;
> > linux-sound@xxxxxxxxxxxxxxx; Flove(HsinFu) <flove@xxxxxxxxxxx>;
> > andriy.shevchenko@xxxxxxxxxxxxxxx; Shuming [范書銘]
> > <shumingf@xxxxxxxxxxx>; Jack Yu <jack.yu@xxxxxxxxxxx>; Derek [方德義]
> > <derek.fang@xxxxxxxxxxx>
> > Subject: Re: [PATCH 20/24] ASoC: codecs: rt1015p: Update definition of
> > device_id tables

...

> > > > diff --git a/sound/soc/codecs/rt1015p.c b/sound/soc/codecs/rt1015p.c
> > > > index 44e7fe3c32da..33917438196f 100644
> > > > --- a/sound/soc/codecs/rt1015p.c
> > > > +++ b/sound/soc/codecs/rt1015p.c
> > > > @@ -133,9 +133,11 @@ MODULE_DEVICE_TABLE(of, rt1015p_device_id);
> > > >
> > > >    #ifdef CONFIG_ACPI
> > > >    static const struct acpi_device_id rt1015p_acpi_match[] = {
> > > > -       { "RTL1015", 0},
> > > > -       { "RTL1019", 0},
> > > > -       { },
> > > > +       { "RTL1015" },
> > > > +       { "RTL1019" },
> > > > +       { "10EC1015" },
> > > > +       { "10EC1019" },
> > > > +       {}
> > >
> > > There are different settings for RTL1015, RTL1019, 10EC1015, and 10EC1019.
> > > RTL1015 and RTL1019 are used for powering up and down via the GPIO.
> > > 10EC1015 and 10EC1019 are used for the I2C-controlled driver.
> > > Therefore 10EC1015 and 10EC1019 should be removed.
> >
> > Sorry for the delay in response and thank for you the feedback.
> >
> > I'm surprised this hasn't come up earlier in the discussion. I'll
> > probably cut off problematic patches so that majority can be merged.
> > It's important to have some follow up though, on the subject you
> > mentioned. Correct me if I'm wrong but from the .c files it seems:
> >
> > for DT/ARM rt1015 & rt1015p are represented by:
> >          .compatible = "realtek,rt1015"
> >          .compatible = "realtek,rt1015p"
> >
> > for ACPI/x86 rt1015 & rt1015p are represented by:
> >          "10EC1015"
> >          "RTL1015"
> >
> > The first pair is nice and clean, the second is confusing and error
> > prone. Do you know where the second pair originated from?
> 10EC is defined by the PCI domain, and RTL is defined by the PNP domain.
> To distinguish these, 1015, 1019, and 5682 use different drivers,
> with the RTL prefix being used to instead of the p postfix in the DT.

Thank you for the reminder. I'm aware of the origins for PCI/PNP 
prefixes, though. In fact, there is more to it - OBDA [1] and ALG [2]. 
The idea is to choose one _recommended path_ and follow it. If the 
desire is to name future chips using suffixes such as -p or -s, then 
perhaps PNP-based naming should have been selected instead. Another 
option would be to change the hardware naming scheme e.g: '1016' instead 
of '1015p'.

If the latter is not an option, why not:
	RTL1015
	RTL1015P

on ACPI/x86 side? Aligns with ARM equivalent nicely, no?

[1]: https://uefi.org/ACPI_ID_List?acpi_search=realtek
[2]: https://uefi.org/PNP_ID_List?pnp_search=realtek

Kind regards,
Czarek