RKL: soundcard is failed to be initialized for link error

[RanderWang]

[  884.040113] skl_hda_dsp_generic skl_hda_dsp_generic: ASoC: error at snd_soc_card_add_dai_link on sof-hda-dsp: -22
[  884.040121] sof-audio-pci 0000:00:1f.3: ASoC: adding FE link failed
[  884.040125] sof-audio-pci 0000:00:1f.3: ASoC: topology: could not load header: -22

Now RKL GPU driver is not included in our kernel, so there is no HDMI support now. It failed at https://github.com/thesofproject/linux/blob/topic/sof-dev/sound/soc/intel/boards/skl_hda_dsp_generic.c#L83. @kv2019i Do you have any advice ?

[kv2019i]

@RanderWang hmm, this is a new kind of error. Are you using an unmodified topology file and hda machine driver?

Possible things to check:

• idisp codec is reported to be present even if driver is not in kernel (should not happen, but best to check):
  -> try with modprobe option: "options snd_sof_intel_hda_common codec_mask=0x1"

• in skl_hda_hdmi_add_pcm(), only thing that could fail seems to be snd_soc_card_get_codec_dai() ... can you print out what is the "dai_name" when the error happens?

[RanderWang]

@RanderWang hmm, this is a new kind of error. Are you using an unmodified topology file and hda machine driver?

Possible things to check:

• idisp codec is reported to be present even if driver is not in kernel (should not happen, but best to check):
  -> try with modprobe option: "options snd_sof_intel_hda_common codec_mask=0x1"

I checked idsp codec is not enabled

• in skl_hda_hdmi_add_pcm(), only thing that could fail seems to be snd_soc_card_get_codec_dai() ... can you print out what is the "dai_name" when the error happens?

dai_name = intel-hdmi-hifi0 and no such dai in sound card. I checked hdmi in sof_sdw. you added check

sof_sdw_hdmi_card_late_probe  {
......
	if (!ctx->idisp_codec)
		return 0;
......
}

How about apply the idea to hda machine driver to skip creating hdmi pcm in this case ?

[kv2019i]

@RanderWang Ok, interesting. It seems we have a regression in the HDA driver for this. I'm investigating what has happened.

[kv2019i]

Ok, this is a regression in 5.8-rc :( ... I'm now bisecting what happened.

[kv2019i]

Found it, it's this patch in 5.8-rc that breaks current HDA generic driver when i915 init fails:

cbc7a6b
ASoC: soc-card: add snd_soc_card_add_dai_link()

... I'll make a fix on Monday.

[ClarexZhou]

SOF can be loaded with this patch #2270 after workarround thesofproject/sof#3173