cvs commit: src/sys/dev/sound/pci ich.c

[Orion Hodson]

/-- Wilko Bulte wrote:
| On Mon, Aug 18, 2003 at 02:28:16PM -0700, Orion Hodson wrote:
| > /-- Wilko Bulte wrote:
| > | wkb at freebie ~: dmesg | grep pcm
| > | pcm0: <ESS Solo-1E> port
| > | 0xdf4c-0xdf4f,0xdf7c-0xdf7f,0xdf50-0xdf5f,0xdf60-0xdf6f,0xde80-0xdebf irq
|  11
| > | at device 13.0 on pci2
| > | pcm1: <Intel ICH5 (82801EB)> port 0xee80-0xeebf,0xe800-0xe8ff mem
| > | 0xfebff400-0xfebff4ff,0xfebff800-0xfebff9ff irq 10 at device 31.5 on pci0
| > | pcm1: <unknown ac97 codec> (id=0x41445375)
| > |
| > | I'm wondering about the ac97 codec message?
| >
| > The ac97 code has a table of codecs that provide the name and optionally a
| > function pointer to a vendor specific initialization routine.  On
| > the AD1985 does not have an entry, so the codec is indicated as "unknown"
| > with the relevant codec id.  The latter is useful for diagnostics.
|
| OK. To what extent needs a codec specific code to work, or is it sufficient
| to recognise it as a ac97?

The vast majority of codecs work without any vendor/codec specific
hooks.  We only have one codec hook at present (ad1885), with a second
one due to go into -CURRENT shortly (ad1980).  Other projects make
more extensive use of initialization hooks, but we're able to leverage
less custom AC97 functionality because of the OSS mixer interface :-(

The "unknown" string seems to cause an undue amount of concern from
users, the intent was a string that would get people to report id's,
but this hasn't proved a very effective way of getting this info from
users.

| Will the MFC be in time for 4.9R ? Murray just posted an update
| that the code freeze for 4.9R will start coming Monday.

Until your last message, there was no plan to, but after looking at
the effort involved, I'm resolved to do it later today.

- Orion