On Sat, May 23, 2009 at 07:13:07PM -0400, Jon Smirl wrote:
AC97 codec for STAC9766 used on the Efika. Datasheet: http://www.idt.com/products/getDoc.cfm?docID=13134007
Signed-off-by: Jon Smirl jonsmirl@gmail.com
This looks mostly good, a couple of issues below. I'll apply the patch but please submit a followup patch for resume as discussed below.
+static int ac97_digital_trigger(struct snd_pcm_substream *substream,
int cmd, struct snd_soc_dai *dai)
Very strange indentation here...
+static int stac9766_codec_resume(struct platform_device *pdev) +{
- /* give the codec an AC97 warm reset to start the link */
+reset:
- if (reset > 5) {
printk(KERN_ERR "stac9766 failed to resume");return -EIO;- }
- codec->ac97->bus->ops->warm_reset(codec->ac97);
- id = soc_ac97_ops.read(codec->ac97, AC97_VENDOR_ID2);
- if (id != 0x4c13) {
stac9766_reset(codec, 0);reset++;goto reset;- }
As I said last time I'd this looks like it should be using the reset function that you've provided?
The loop is new since last time and seems very suspicious - are you sure that this is a CODEC problem that's being worked around and not an issue with the AC97 controller or with the specific system starting up the master clock for the CODEC after resume? If it's an issue with the controler then it'd be better to fix it there so that all CODEC drivers get the benefit. If it's a CODEC issue a comment explaining the problem would be helpful since it's not the sort of issue I'd expect to see in a CODEC. The fact that it's not needed on initial boot suggests something controller or external clock related.
Either way, it'd be much clearer to rewrite it using a for or while loop rather than a goto. Please submit a followup patch fixing at least this issue.
+static int __init stac9766_modinit(void) +{
- return snd_soc_register_dais(stac9766_dai, ARRAY_SIZE(stac9766_dai));
+} +module_init(stac9766_modinit);
+static void __exit stac9766_exit(void) +{
- snd_soc_unregister_dais(stac9766_dai, ARRAY_SIZE(stac9766_dai));
+} +module_exit(stac9766_exit);
These are not needed for AC97 CODECs, ASoC doesn't wait for them to probe. I'll apply a patch removing these.