Move the workaround for the VT8251 up a bit, and check for STAT_EOL
rather than STAT_ACTIVE. This resolves issues some people were having
with certain ALSA clients (and allows the STAT_ACTIVE check to do what
it was intended to do).
This change was suggested by Andrew Daviel.
Signed-off-by: Bastiaan Jacques <b.jacques@planet.nl>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
        if (!status)
                status = inb(VIADEV_REG(viadev, OFFSET_STATUS));
 
+       /* An apparent bug in the 8251 is worked around by sending a 
+        * REG_CTRL_START. */
+       if (chip->revision == VIA_REV_8251 && (status & VIA_REG_STAT_EOL))
+               snd_via82xx_pcm_trigger(substream, SNDRV_PCM_TRIGGER_START);
+
        if (!(status & VIA_REG_STAT_ACTIVE)) {
-               /* An apparent bug in the 8251 is worked around by sending
-                * a REG_CTRL_START. */
-               if (chip->revision == VIA_REV_8251)
-                       snd_via82xx_pcm_trigger(substream,
-                                               SNDRV_PCM_TRIGGER_START);
-               else {
-                       res = 0;
-                       goto unlock;
-               }
+               res = 0;
+               goto unlock;
        }
        if (count & 0xffffff) {
                idx = count >> 24;