On Wed, 13 Mar 2019 21:55:52 +0100, Pierre-Louis Bossart wrote:
#define is_geminilake(codec) (((codec)->core.vendor_id == 0x8086280d) || \ ((codec)->core.vendor_id == 0x80862800))
+HDA_CODEC_ENTRY(0x80862800, "Geminilake HDMI", patch_i915_glk_hdmi),
This 0x2800 ID does not seem to be legit. No one at Intel, including the hardware IP owner, was able to confirm why this value was provided upstream, it is not part of any internal hardware documentation, is suspiciously out-of-sequence and is not used in the hdac_hdmi code. The contributor who released this value is no longer with Intel so I would really suggest removing this ID from the list supported by the legacy driver. I can send a patch next week for this.
Thanks, that'll be helpful.
Takashi