Re: [PATCH] ASoC: Intel: Do not load legacy SST driver on BYT when SND_SOC_SOF_BAYTRAIL is enabled

[Hans de Goede <hdegoede@redhat.com>]

Hi,

On 10/12/20 9:24 AM, Rojewski, Cezary wrote:
> On 2020-10-11 11:53 AM, Hans de Goede wrote:
>> The legacy 80860F28 / sst_acpi_baytrail_desc match in sst_acpi_match
>> is already conditional on the the newer SND_SST_IPC_ACPI driver not
>> being enabled.
>>
>> But now that we have an even newer driver in the form of SOF support
>> for BYT devices, we also need to take this into account, so we also
>> must not include the sst_acpi_baytrail_desc match when
>> SND_SOC_SOF_BAYTRAIL is enabled.
>>
>> This fixes snd-soc-sst-acpi binding to the 80860F28 platform device,
>> blocking snd-sof-acpi from binding, which breaks audio support.
>>
>> Signed-off-by: Hans de Goede <hdegoede@redhat.com>
>> ---
> 
> Hello,
> 
> Series:
> [PATCH v2 00/13] ASoC: Intel: Remove obsolete solutions and components
> https://lore.kernel.org/alsa-devel/20201006064907.16277-1-cezary.rojewski@intel.com/
> 
> removes sst-acpi component along with many others so further changes to
> said component will only cause conflicts -or- require commit reordering.
> I'd advice against that.

As I already mentioned in the private-thread which Pierre-Louis started
with me, Jaroslav Kysela and Liam about this I would advice against applying
that series for now. First we need to put in more work to make sure that
the new drivers are actually ready.

Also I must say that I'm quite disappointed that since I, as the person
who more or less single handedly have made sure that audio works properly o
Bay Trail and Cherry Traul devices (*), has not been Cc-ed on that series,
that seems like a huge oversight.

Anyways I will reply in the thread of the series and ask Mark to revert
the entire series. Since IMHO the new drivers are clearly not ready yet.
Yesterday I ran my first set of tested and I immediately hit a DSP
hang doing just a few very basic tests.

Regards,

Hans



*) And kept it working properly despite other people breaking it with changes
like moving the userspace stuff to UCM2.