On 2019/5/4 下午3:18, Takashi Iwai wrote:
On Sat, 04 May 2019 04:45:36 +0200, Hui Wang wrote:
On 2019/5/3 下午11:57, Takashi Iwai wrote:
On Fri, 03 May 2019 06:05:09 +0200, Hui Wang wrote:
Right...
And inspired by your change, maybe we just make this change, then it is enough to fix the falsely report issue here.
@@ -1551,8 +1551,11 @@ static bool hdmi_present_sense_via_verbs(struct hdmi_spec_per_pin *per_pin, ret = !repoll || !eld->monitor_present || eld->eld_valid; jack = snd_hda_jack_tbl_get(codec, pin_nid);
- if (jack)
- if (jack) { jack->block_report = !ret;
jack->pin_sense = (eld->monitor_present && eld->eld_valid) ?AC_PINSENSE_PRESENCE : 0;- }
because in the snd_hda_pin_sense(), the jack_dirty is set to 0, then we change the jack->pin_sense, and in the report_sync() it will decide the jack state according to the jack->pin_sense we changed.
OK, that should work, just overriding and correcting the pin_sense.
I guess the only missing code path is the case where jack->dirty is set manually without the callback call via snd_hda_jack_set_dirty_all(). Through a quick glance, it's called from the common resume code, hda_call_codec_resume(). But, again, HDMI codec has its own resume to refresh pin detection (generic_hdmi_resume()), and the changed code-path should be involved.
The rest callers of snd_hda_jack_set_dirty_all() are the polling mode and the unsol event handler for the old HDMI codecs, both of which we don't care much.
That said, your minimal change looks good to me, and I'll happily apply as long as it's tested. Of course, it needs some more careful comments about the behavior.
OK, after testing it, I will send the patch to review.
Thanks.
Thanks!
Takashi _______________________________________________ Alsa-devel mailing list Alsa-devel@alsa-project.org https://mailman.alsa-project.org/mailman/listinfo/alsa-devel