[alsa-devel] A Weird issue on Gigabyte board

[Takashi Iwai]

At Thu, 17 Jan 2008 16:44:07 +0800,
Jiang zhe wrote:
> 
> Hi,
> 
> I meet with a weird issue.
> Please refer to the bug 0003674 on bugtrack.alsa-project.org.
> 
> It use alc889 codec but the codec vendor id is 0885.
> 
> With the alsa-driver-hg20080110.tar.bz2, it works well with option
> "model=6stack-dig".
> 
> But after I simply added the this line into the alc882_cfg_tbl[]:
> SND_PCI_QUIRK(0x1458, 0xa002, "Gigabyte P35-DS3R", ALC882_6ST_DIG),
> 
> The dmesg still show "hda_codec: Unknown model for ALC882, trying
> auto-probe from BIOS...". and driver will crash.
> 
> With this patch, even loading it with option "model=6stack-dig" doesn't
> help, driver will still crash.
> 
> I found a line in the dmesg:
> "hda_intel: azx_get_response timeout, switching to polling mode: last
> cmd=0xbe0f0009."
> The Cad is 0xb.

Very weird.

> The codec_mask can change from 0x4 to 0x1:
> alsa-kernel/pci/hda/hda_intel.c:732: codec_mask = 0x4
> alsa-kernel/pci/hda/hda_intel.c:732: codec_mask = 0x1

This implies the hardware is somehow unstable.


Anyway, try the patch below.  It cleans up a bit and avoids possible
invalid commands, plus puts stack traces at a fatal communication
error.


Takashi

diff -r f94766e99413 pci/hda/hda_codec.c
--- a/pci/hda/hda_codec.c	Thu Jan 17 11:28:47 2008 +0100
+++ b/pci/hda/hda_codec.c	Thu Jan 17 12:13:30 2008 +0100
@@ -2739,8 +2739,7 @@ int snd_hda_parse_pin_def_config(struct 
 				 struct auto_pin_cfg *cfg,
 				 hda_nid_t *ignore_nids)
 {
-	hda_nid_t nid, nid_start;
-	int nodes;
+	hda_nid_t nid, end_nid;
 	short seq, assoc_line_out, assoc_speaker;
 	short sequences_line_out[ARRAY_SIZE(cfg->line_out_pins)];
 	short sequences_speaker[ARRAY_SIZE(cfg->speaker_pins)];
@@ -2753,8 +2752,8 @@ int snd_hda_parse_pin_def_config(struct 
 	memset(sequences_hp, 0, sizeof(sequences_hp));
 	assoc_line_out = assoc_speaker = 0;
 
-	nodes = snd_hda_get_sub_nodes(codec, codec->afg, &nid_start);
-	for (nid = nid_start; nid < nodes + nid_start; nid++) {
+	end_nid = codec->start_nid + codec->num_nodes;
+	for (nid = codec->start_nid; nid < end_nid; nid++) {
 		unsigned int wid_caps = get_wcaps(codec, nid);
 		unsigned int wid_type =
 			(wid_caps & AC_WCAP_TYPE) >> AC_WCAP_TYPE_SHIFT;
diff -r f94766e99413 pci/hda/hda_intel.c
--- a/pci/hda/hda_intel.c	Thu Jan 17 11:28:47 2008 +0100
+++ b/pci/hda/hda_intel.c	Thu Jan 17 12:13:30 2008 +0100
@@ -587,6 +587,7 @@ static unsigned int azx_rirb_get_respons
 	snd_printk(KERN_ERR "hda_intel: azx_get_response timeout, "
 		   "switching to single_cmd mode: last cmd=0x%08x\n",
 		   chip->last_cmd);
+	dump_stack(); /* It's an error - show stack trace */
 	chip->rirb.rp = azx_readb(chip, RIRBWP);
 	chip->rirb.cmds = 0;
 	/* switch to single_cmd mode */
@@ -1738,6 +1739,7 @@ static int __devinit azx_create(struct s
 
 	err = pci_request_regions(pci, "ICH HD audio");
 	if (err < 0) {
+		printk(KERN_ERR SFX "Fail to request regions\n");
 		kfree(chip);
 		pci_disable_device(pci);
 		return err;
diff -r f94766e99413 pci/hda/hda_local.h
--- a/pci/hda/hda_local.h	Thu Jan 17 11:28:47 2008 +0100
+++ b/pci/hda/hda_local.h	Thu Jan 17 12:13:30 2008 +0100
@@ -375,7 +375,7 @@ static inline u32 get_wcaps(struct hda_c
 {
 	if (nid < codec->start_nid ||
 	    nid >= codec->start_nid + codec->num_nodes)
-		return snd_hda_param_read(codec, nid, AC_PAR_AUDIO_WIDGET_CAP);
+		return 0;
 	return codec->wcaps[nid - codec->start_nid];
 }