Re: ASoC: Intel: sst: Missing IRQ at index 5 on BYT-T device

[Stephan Gerhold <stephan@gerhold.net>]

On Mon, Dec 17, 2018 at 09:32:42PM +0100, Stephan Gerhold wrote:
> On Mon, Dec 17, 2018 at 01:39:13PM -0600, Pierre-Louis Bossart wrote:
> > Thanks for the additional information.
> > > The call to iosf_mbi_read() returns 0x400b0100
> > > 
> > > 	/* bits 26:27 mirror PMIC options */
> > > 	bios_status = (bios_status >> 26) & 3;
> > > 
> > > Results in bios_status = 0x0
> > So that's a fail.
> > > 
> > > The stock kernel printed this on every startup:
> > > 
> > > SPID updated according to ACPI Table:
> > > 	spid customer id : 0000
> > > 	spid vendor id : 0000
> > > 	spid manufacturer id : 00ff
> > > 	spid platform family id : 0007 --> INTEL_BYT_TABLET
> > > 	spid product line id : 0000 --> INTEL_BYT_TABLET_BLK_PRO
> > > 	spid hardware id : 0004 --> BYT_TABLET_BLK_8PR0 /* Bay Lake FFRD-8 PR0 */
> > > 	spid fru[4..0] : 00 00 00 00 00
> > > 	spid fru[9..5] : 00 00 00 00 00
> > > 
> > > Based on spid.h [1] I added the "-->" above. Then I guessed that this is
> > > BYT-T (there is another "BYT T CR V2" value), but to be honest I don't
> > > know for sure.
> > > 
> > > [1]: https://github.com/me176c-dev/me176c-kernel/blob/stock/kernel/arch/x86/include/asm/spid.h
> > 
> > Oh man, Bay Lake...this must be at least 6 years old and 30+ kernel versions
> > behind... Only a couple of years and it'll be a collector item  :-)
> > 
> 
> Yeah, the device was shipped with a 3.10 kernel but I believe that file 
> was just copied from an earlier 3.4 kernel. I have never bothered to even 
> try to compile that thing, I just use it as reference every now and then. :)
> 
> > I can't recall any of the details so we'll have to wing it. it could be that
> > it was baytrail-T but with the software/BIOS for Baytrail-Cr, who knows.
> > 
> > > 
> > > > I don't mean to dismiss your claim, just want to find out if this is a case
> > > > where the PMIC-type-based byt_cr detection fails or if we have a BIOS issue.
> > > > Another smoking gun is if you find in your code traces of SSP0 being used.
> > > > 
> > > The quirks to get sound working with bytcr-rt5640 on that device are:
> > > BYT_RT5640_SSP0_AIF1 | BYT_RT5640_IN1_MAP | BYT_RT5640_MCLK_EN
> > > 
> > > I guess this means that SSP0 is being used?
> > 
> > Yes indeed, and that makes me think we should force this device to look like
> > Baytrail-CR.
> > 
> > You can do this with a DMI-based quirk (preferably in is_byt_cr directly so
> > that I can reuse the code when I move it to a helper at some point).
> 
> Okay - thanks! One last question:
> I was looking at the ACPI DSDT tables of some similar devices and have 
> found two others that look the same (only one IRQ listed). In this case, 
> the BYT-T acpi_ipc_irq_index = 5 will never work, and we will definitely 
> have a better chances with trying Baytrail-CR.
> 
> One of them actually had a similar patch proposed at [1] (although they 
> did it in a weird way and also need an extra machine driver).
> 
> We could also detect this situation in a generic way with something like
> 
>   if (platform_irq_count(pdev) == 1) {
>   	*bytcr = true;
> 	return 0;
>   }
> 
> ... instead of a DMI quirk. What do you think?
> 

To avoid confusion: The existing PMIC-type based is_byt_cr() detection 
would be used in all other cases (i.e. if irq_count != 1), so it won't 
make any difference for the devices that are already working fine.
(Most BYT-CR devices seem to have 5 IRQs listed)

So it's more like

  if (platform_irq_count(pdev) == 1) {
  	*bytcr = true;
  } else {
  	// pmic-type based detection...
  }

with platform_irq_count == 1 as condition for the "quirk".

> [1]: https://patchwork.kernel.org/patch/9764493/#20539549
> 
> > 
> > Also I guess you'd need a second quirk in bytcr_rt5640 since the default is
> > SSP0-AIF2.
> > 
> > -Pierre
> >