ALSA: hda - Don't trigger pin-sense for STAC/IDT codecs

STAC/IDT codecs seem to behave weird when SET_PIN_SENSE verb is issued
before reading the jack-detection although the TRIG_REQ pin capability
is given by the hardware.

Since snd_hda_jack_detect() issues the SET_PIN_SENSE verb simply judging
from the pincap, we have to revert the change in the commit
  d56757abc11a21996d9839c0d4e3b2c3666cd318
    ALSA: hda - Replace the rest of jack-detections with snd_hda_jack_detect()
to plain GET_PIN_SENSE verb without triggering.

Reported-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
diff --git a/sound/pci/hda/patch_sigmatel.c b/sound/pci/hda/patch_sigmatel.c
index 2a45375..6b0bc04 100644
--- a/sound/pci/hda/patch_sigmatel.c
+++ b/sound/pci/hda/patch_sigmatel.c
@@ -4440,7 +4440,14 @@
 {
 	if (!nid)
 		return 0;
-	return snd_hda_jack_detect(codec, nid);
+	/* NOTE: we can't use snd_hda_jack_detect() here because STAC/IDT
+	 * codecs behave wrongly when SET_PIN_SENSE is triggered, although
+	 * the pincap gives TRIG_REQ bit.
+	 */
+	if (snd_hda_codec_read(codec, nid, 0, AC_VERB_GET_PIN_SENSE, 0) &
+	    AC_PINSENSE_PRESENCE)
+		return 1;
+	return 0;
 }
 
 static void stac92xx_line_out_detect(struct hda_codec *codec,