[ALSA] ac97 - Add a workaround for broken quirk for VT1617A codec

On boards with VT1617A codec, the sound disappears suddenly.
This looks like a problem with HPE-bit control that is supposed to be
set in patch_vt1617a().  However, on such problematic hardwares, the
bit is actually reset mysteriously.

The patch adds a workaround for the wrong quirk.

Signed-off-by: Takashi Iwai <tiwai@suse.de>

diff --git a/sound/pci/ac97/ac97_patch.c b/sound/pci/ac97/ac97_patch.c
index 39198e505b12aabacd6842389647da8dff2c7c68..2da89810ca10fbab1e440aa4a0ae282e61799c42 100644 (file)
--- a/sound/pci/ac97/ac97_patch.c
+++ b/sound/pci/ac97/ac97_patch.c
@@ -3446,6 +3446,7 @@ static const struct snd_kcontrol_new snd_ac97_controls_vt1617a[] = {
 int patch_vt1617a(struct snd_ac97 * ac97)
 {
        int err = 0;
+       int val;
 
        /* we choose to not fail out at this point, but we tell the
           caller when we return */
@@ -3456,7 +3457,13 @@ int patch_vt1617a(struct snd_ac97 * ac97)
        /* bring analog power consumption to normal by turning off the
         * headphone amplifier, like WinXP driver for EPIA SP
         */
-       snd_ac97_write_cache(ac97, 0x5c, 0x20);
+       /* We need to check the bit before writing it.
+        * On some (many?) hardwares, setting bit actually clears it!
+        */
+       val = snd_ac97_read(ac97, 0x5c);
+       if (!(val & 0x20))
+               snd_ac97_write_cache(ac97, 0x5c, 0x20);
+
        ac97->ext_id |= AC97_EI_SPDIF;  /* force the detection of spdif */
        ac97->rates[AC97_RATES_SPDIF] = SNDRV_PCM_RATE_44100 | SNDRV_PCM_RATE_48000;
        ac97->build_ops = &patch_vt1616_ops;