[alsa-devel] [alsa-cvslog] alsa-kernel: Official ALSA project GIT repository for Linux 2.6 branch, master now at v2.6.27-rc5-122-g1328f04

[Takashi Iwai]

At Fri, 29 Aug 2008 11:35:25 +0200 (CEST),
noreply-git at alsa-project.org wrote:
> 
> commit 1328f04260822e45a2717b65a0e2bc0c9c1255be
> Author:     Jaroslav Kysela <perex at perex.cz>
> AuthorDate: Fri Aug 29 11:29:39 2008 +0200
> Commit:     Jaroslav Kysela <perex at perex.cz>
> CommitDate: Fri Aug 29 11:29:39 2008 +0200
> 
>     ALSA: intel8x0: implement ac97_clock whitelist
>     
>     The AC97 clock detection is not accurate in some cases. This patch adds
>     an initial whitelist for audio devices gathered from RedHat's bugzilla:
>     
>     https://bugzilla.redhat.com/show_bug.cgi?id=441087
>     
>     As a side effect, white-listing might speedup kernel booting (AC97 clock
>     measuring code is not activated).
>     
>     Signed-off-by: Jaroslav Kysela <perex at perex.cz>
(snip)
> +struct intel8x0_clock_list {
> +	unsigned short subvendor;
> +	unsigned short subdevice;
> +	unsigned int rate;
> +};
> +
> +static struct intel8x0_clock_list intel8x0_clock_list[] __devinitdata = {
> +	{ 0x1028, 0x00be, 44100 },	/* Analog Devices AD1885 */
> +	{ 0x1028, 0x0177, 48000 },	/* Analog Devices AD1980 */
> +	{ 0x1043, 0x80f3, 48000 },	/* Analog Devices AD1985 */
> +	{ 0x0000, 0x0000, 00000 }	/* terminator */
> +};
> +
> +static int __devinit intel8x0_in_clock_list(struct intel8x0 *chip)
> +{
> +	struct pci_dev *pci = chip->pci;
> +	struct intel8x0_clock_list *wl;
> +
> +	for (wl = intel8x0_clock_list; wl->subvendor; wl++) {
> +		if (wl->subvendor == pci->subsystem_vendor &&
> +		    wl->subdevice == pci->subsystem_device) {
> +		    	printk(KERN_INFO "intel8x0: white list rate for %04x:%04x is %i\n",
> +		    		pci->subsystem_vendor,
> +		    		pci->subsystem_device, wl->rate);
> +			chip->ac97_bus->clock = wl->rate;
> +			return 1;
> +		}
> +	}
> +	return 0;
> +}

Use snd_pci_quirk_lookup() please.

But, a better fix would be to fix the clock measurement.
AFAIK, this seems to happen when the driver is built-in.  Thus I guess
it's rather a problem of time-keeping core side.


Takashi