On Thu, 04 Feb 2021 11:42:20 +0100, Kai Vehmanen wrote: > > Hey, > > On Thu, 4 Feb 2021, Takashi Iwai wrote: > > > On Thu, 04 Feb 2021 09:07:09 +0100, Kai Vehmanen wrote: > > > Takashi and others, this patch is a follow-up to the earlier thread about > > > hdac link management: > > > https://mailman.alsa-project.org/pipermail/alsa-devel/2021-January/179715.html > > > > > > We have absolutely no calls from sound/pci/hda/ to hdaudio_ext.h before > > > this patch, so I'm wondering if this is the right approach to take. > > > > An obvious drawback by this patch is that it'll make the hda-ext code > > always loaded whenever it's enabled, no matter whether it's really > > used or not. > > > > Maybe adding a new callback (link) in hdac_bus_ops can help? > > hmm, that's true. Let me try adding a separate callback and resend for > review. That will look cleaner in hda_codec.c. Thanks. BTW, one more thing: the code link up/down isn't always paired. It's a bit map and only holding states. For example, in hda_codec.c, the link power down may happen conditionally marked in hda_codec_runtime_suspend() while the resume always marks the link power up. So you'd need to check the bitmap change to perform the refcount up/down in the hda-ext side. Takashi