[Sound-open-firmware] MinnowBoard FW fails to boot

[Guennadi Liakhovetski]

On Mon, Mar 25, 2019 at 10:55:20AM -0400, Pierre-Louis Bossart wrote:
> 
> > > 
> > > did you enable the DSP is the BIOS south cluster? it's not listed in teh
> > > documentation. You need to select the enable but not report case,
> > 
> > Are you sure this was on Minnowboard? Maybe in older BIOS versions? I see an
> > "enable audio DSP" option on Up^2, but not on Minnow. On Minnow I only see
> > "enable audio," which of course is on. But even on "Up^2" I don't see any
> > "not report" choice?
> 
> I am positive that on the 0.96 BIOS there was an ACPI or PCI LPE report. I
> don't recall exactly that this meant, but by default the DSP is *not*
> enabled after you flash the BIOS.
> 
> Actually it's all documented here:
> https://github.com/plbossart/MinnowBoardMaxFirmware
> 
> To enable the audio DSP on MinnowMax, go to the BIOS menu in System Setup ->
> SouthCluster Configuration -> Audio Configuration
> 
> 'LPE Audio Support' should be changed from Disabled to LPE Audio ACPI mode
> 
> 'LPE Audio Reported by DSDT' should be changed to disabled (to remove the
> default DSDT settings and enable the SSDT override)

Ok, yes, that's exactly what I've also done. There's also more audio configuration
in the BIOS below those two entries, for some Azalia device. I tried leaving it in
its default enabled configuration and also disabling it - it didn't change
anything.

I've created https://github.com/thesofproject/linux/issues/751 as you requested.

Thanks
Guennadi

> > > and use LPEX.aml for nocodec mode.
> > 
> > Would this lead to the observed failure - ENODEV? I thought the codec would
> > only come into play later. I was actually running a with-codev kernel and
> > BIOS set up. First I was running that on a MinnowBoard Max, where a codec
> > was indeed connected. Then I switched over to a Torbot without replacing
> > any of the software.
> 
> All .aml files add the relevant ACPI information for LPE, and the acpi
> update ssdt application sets the SSP pins (not enabled by default). the
> point is that if you don't have a codec attached you still need to use the
> LPEX.aml file.
> 
> > 
> > > If this doesn't work, let's continue with a github issue where it's easier
> > > to add logs and keep track of the problem.
> > 
> > It didn't. I've rebuilt the kernel with FORCE_NOCODEC_MODE=y
> > 
> > [    7.951333] sof-audio-acpi 80860F28:01: ACPI DSP detected
> > [    7.951342] sof-audio-acpi 80860F28:01: BYT-CR not detected
> > [    7.951345] sof-audio-acpi 80860F28:01: Force to use nocodec mode
> > [    7.951354] sof-audio-acpi 80860F28:01: probing SOF DSP device....
> > [    7.951358] sof-audio-acpi 80860F28:01: LPE PHY base at 0x90a00000 size 0x200000
> > [    7.951410] sof-audio-acpi 80860F28:01: LPE VADDR 00000000a246845d
> > [    7.951413] sof-audio-acpi 80860F28:01: IMR base at 0x20000000 size 0x100000
> > [    7.951428] sof-audio-acpi 80860F28:01: IMR VADDR 000000005bbccd5b
> > [    7.951432] sof-audio-acpi 80860F28:01: using IRQ 29
> > ...
> > 
> > and I'm still getting the same:
> > 
> > [    8.107441] sof-audio-acpi 80860F28:01: booting DSP firmware
> > [    9.188571] sof-audio-acpi 80860F28:01: error:  unable to run DSP firmware
> > [    9.188926] sof-audio-acpi 80860F28:01: error: unexpected fault 0xffffffff trace 0xffffffff
> > [    9.189014] sof-audio-acpi 80860F28:01: error: failed to reset DSP
> > [    9.189080] sof-audio-acpi 80860F28:01: error: failed to boot DSP firmware -19
> > [    9.189184] sof-audio-acpi 80860F28:01: error: failed to probe DSP hardware!
> > 
> > I'll create a GH issue...
> > 
> > Thanks
> > Guennadi
> > 
>