On Mon, 21 Aug 2023 11:06:20 +0200,
Andy Shevchenko wrote:
>
> On Sun, Aug 20, 2023 at 11:16:08AM +0200, Takashi Iwai wrote:
> > On Fri, 18 Aug 2023 19:01:16 +0200,
> > Andy Shevchenko wrote:
> > > On Fri, Aug 18, 2023 at 11:00:34AM -0500, Pierre-Louis Bossart wrote:
>
> ...
>
> > > > > +static void tas2781_fixup_i2c(struct hda_codec *cdc,
> > > > > + const struct hda_fixup *fix, int action)
> > > > > +{
> > > > > + tas2781_generic_fixup(cdc, action, "i2c", "TIAS2781");
> > > >
> > > > TI ACPI ID is TXNW
> > > >
> > > > https://uefi.org/ACPI_ID_List?search=TEXAS
> > > >
> > > > There's also a PNP ID PXN
> > > >
> > > > https://uefi.org/PNP_ID_List?search=TEXAS
> > > >
> > > > "TIAS" looks like an invented identifier. It's not uncommon but should
> > > > be recorded with a comment if I am not mistaken.
> > > >
> > > > > +}
> > >
> > > Thank you, but actually it's a strong NAK to this even with the comment.
> > > We have to teach people to follow the specification (may be even hard way).
> > >
> > > So where did you get the ill-formed ACPI ID?
> > > Is Texas Instrument aware of this?
> > > Can we have a confirmation letter from TI for this ID, please?
> >
> > This is used already for products that have been long in the market,
> > so it's way too late to correct it, I'm afraid.
> >
> > What we can do is to get the confirmation from TI, complain it, and
> > some verbose comment in the code, indeed.
>
> Oh, no! Who made that ID, I really want to point that at their faces.
> Look at the Coreboot (successful) case, they created something, but
> in time asked and then actually fixed the ill-formed ID (that was for
> one of RTC chips).
>
> For this, please make sure that commit message has that summary, explaining that
> - states that ID is ill-formed
> - states that there are products with it (DSDT excerpt is a must)
> - lists (a few?) products where that ID is used
> - ideally explains who invented that and Cc them to the patch, so they will
> know they made a big mistake
Sure, we should complain further and ask them that such a problem
won't happen again. I'm 100% for it.
But the fact is that lots of machines have been already shipped with
this ID since long time ago, and 99.99% of them have been running on
Windows. Hence I expect that the chance to get a corrected ID is very
very low, and waiting for the support on Linux until the correction of
ID actually happens makes little sense; that's my point.
Takashi
