On 3/9/20 8:47 AM, Cezary Rojewski wrote:
On 2020-03-06 21:57, Pierre-Louis Bossart wrote:
On 3/5/20 8:53 AM, Cezary Rojewski wrote:
With _reset_link removed from the probe sequence, codec_mask at the time skl_find_hda_machine() is invoked will always be 0, so hda machine will never be chosen. Rather than reorganizing boot flow, be permissive about invalid mask. codec_mask will be set to proper value during probe_work - before skl_codec_create() ever gets called.
humm, what would happen e.g. if you have select the SKL driver but there is no ACPI information to select an I2S-based machine driver, and HDaudio/iDISP are disabled? You would have no error checks then?
Laptops I've been testing this with have had Realtek + iDisp present onboard. Now, if you disable Realtek + HDMI/DP modules within legacy HDaudio Kconfig and HD audio support within Intel Skylake tree then you end up with no required modules for said configuration at all. Nothing will happen really: no warnings, no sound card either.
I meant enable the HDaudio controller but disable HDaudio codecs/HDMI at the BIOS level. In that case the codec_mask will never be set.
Just run such kernel on my setup and results are quite obvious:
- skl boots
- no machines present
- drv stays dormant
Czarek