]> git.openfabrics.org - ~shefty/rdma-dev.git/commitdiff
ALSA: hda_intel: more strict alc880_parse_auto_config dig_nid checking
authorJaroslav Kysela <perex@perex.cz>
Fri, 10 Jul 2009 10:27:31 +0000 (12:27 +0200)
committerTakashi Iwai <tiwai@suse.de>
Fri, 10 Jul 2009 10:55:49 +0000 (12:55 +0200)
On some IbexPeak systems with ALC889A errors like "azx_get_response
timeout, switching to polling mode: last cmd=0xaf9f000b" are produced,
because non-existent codec #10 is wrongly accessed.

The problem is that snd_hda_get_connections() returns out-of-range result
for NID 0x1c (something like 0xf8f9 or 0xffff).

This patch adds a check to alc880_parse_auto_config() to avoid using
of this out-of-range NIDs. A better fix maybe to improve
snd_hda_get_connections() routine to check for valid NID ranges if
NIDs are expected as result.

Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
sound/pci/hda/patch_realtek.c

index bbb9b42e2604c87944dd1567375b375fd3352430..7e99763ca527209ad3243e3b0f75228aeb440fce 100644 (file)
@@ -4505,6 +4505,12 @@ static int alc880_parse_auto_config(struct hda_codec *codec)
                                              &dig_nid, 1);
                if (err < 0)
                        continue;
+               if (dig_nid > 0x7f) {
+                       printk(KERN_ERR "alc880_auto: invalid dig_nid "
+                               "connection 0x%x for NID 0x%x\n", dig_nid,
+                               spec->autocfg.dig_out_pins[i]);
+                       continue;
+               }
                if (!i)
                        spec->multiout.dig_out_nid = dig_nid;
                else {