On Tue, 28 Nov 2023 10:10:48 +0100, Péter Ujfalusi wrote: > > > > On 27/11/2023 17:43, Takashi Iwai wrote: > > On Mon, 27 Nov 2023 16:40:57 +0100, > >>> --- a/include/sound/hdaudio.h > >>> +++ b/include/sound/hdaudio.h > >>> @@ -95,6 +95,7 @@ struct hdac_device { > >>> bool lazy_cache:1; /* don't wake up for writes */ > >>> bool caps_overwriting:1; /* caps overwrite being in process */ > >>> bool cache_coef:1; /* cache COEF read/write too */ > >>> + bool is_hdmi:1; /* a HDMI/DP codec */ > >>> unsigned int registered:1; /* codec was registered */ > >>> }; > >>> > >>> --- a/sound/pci/hda/patch_hdmi.c > >>> +++ b/sound/pci/hda/patch_hdmi.c > >>> @@ -2597,6 +2597,7 @@ static int patch_generic_hdmi(struct hda_codec *codec) > >>> } > >>> > >>> generic_hdmi_init_per_pins(codec); > >>> + codec->core.is_hdmi = true; > >>> return 0; > >>> } > >>> > >>> @@ -3472,6 +3473,7 @@ static int patch_simple_hdmi(struct hda_codec *codec, > >>> spec->pcm_playback = simple_pcm_playback; > >>> > >>> codec->patch_ops = simple_hdmi_patch_ops; > >>> + codec->core.is_hdmi = true; > >>> > >>> return 0; > >>> } > >> > >> I see, so this is what I was not sure how to do ;) > >> I will rework the series and resend tomorrow. > >> > >> Thanks for the code snippet, I will make you as author of it, if it is > >> OK for you. > > > > Sure, no problem. > > The flag does not work with SOF stack which uses the hdac_hda codec driver: > > patch_generic_hdmi() and patch_simple_hdmi() is not entered at all, so > the flag is not set. > > The codec driver have is_hdmi_codec() helper to check the struct > hda_pcm.pcm_type, but that is not set early enough either. > > The is HDMI or not needs to be known in hdac_hda_dev_probe(), I think > this was one of the reason I have opted to have the exported function. > We just don't have other information at the dev probe time. > > # dmesg | grep peter > [ 3.810841] [peter] hdac_hda_dev_probe: is_hdmi_codec(): 0 > [ 3.810846] [peter] hdac_hda_dev_probe: hdev->is_hdmi: 0 > [ 3.810848] [peter] hdac_hda_dev_probe: snd_hda_device_is_hdmi(): 0 > ... > [ 3.814497] [peter] hdac_hda_dev_probe: is_hdmi_codec(): 0 > [ 3.814499] [peter] hdac_hda_dev_probe: hdev->is_hdmi: 0 > [ 3.814500] [peter] hdac_hda_dev_probe: snd_hda_device_is_hdmi(): 1 > ... > [ 3.986610] [peter] generic_hdmi_build_pcms: ENTER > [ 3.986627] [peter] hdac_hda_codec_probe: is_hdmi_codec(): 1 > ... > [ 3.996383] [peter] snd_hda_parse_pin_defcfg: ENTER > [ 4.001562] [peter] hdac_hda_codec_probe: is_hdmi_codec(): 0 Hm... I still find it's a bad move to use an exported symbol from another codec driver. And, I wonder what if you have a system that has only one HDMI codec without analog one? Would it still work with your change? Takashi