Hi Takashi, Nikil and others,
On Fri, 29 Nov 2019, Kai Vehmanen wrote:
This difference leads to some subtle differences in hdmi_find_pcm_slot() with regards to how non-MST monitors are assigned to PCMs. This patch restores the original behaviour on Intel platforms while keeping the new allocation policy on other platforms.
hmm, there seems a couple of more issues. The first patch is a clear bug that leads to segfault with SOF+patch_hdmi on some platforms (pipe B used for single monitor HDMI case -> dev_id=1 -> non-existant pcm selected and eventual kernel oops).
This second patch is however trickier. Nikhil your patch changed the default allocation a bit, so the routing might be difference also with snd-hda-intel (i.e. not SOF) for existing platforms and this may surprise users.
Digging deeper, we seem to have a slight semantics difference in how intel_pin_eld_notify() and generic_acomp_pin_eld_notify() handle the third pipe/dev_id parameter.
Any thoughts how to solve? I first I thought making separate functions for hdmi_find_pcm_slot() and allow platforms to define an alternative implementation, but in this RFC patch I opted for a simpler quirk in the function. This is becoming fairly messy I must say -- the amount of code commentary needed is a good indication some simplifaction would be in order.
PS I did not have time to fully test the RFC patch, so this is just for discussion now...
Br, Kai