cvs commit: src/sys/dev/sound/pci emu10k1.c

Scott Long scottl at freebsd.org
Tue Jun 15 21:19:42 GMT 2004


Gordon Tetlow wrote:

> On Sun, Jun 13, 2004 at 07:18:23PM -0600, Scott Long wrote:
> 
>>>Possibly.  Still doesn't mean we shouldn't add the PCI ID to help people
>>>farther along in determining if it really works or not.  If we get
>>>reliable failure reports in 5-CURRENT (which has a much updated driver
>>
>>>from 5.2.1) we can print out an error message in the probe.
>>
>>Well, this is kinda like finding a new Adaptec ID and adding it to the
>>ahc driver just because it's there, regardless of whether the driver can
>>talk to the chip.
> 
> 
> I would also argue that storage subsystems and sound cards are two very
> different class of devices. We can afford to be a little overzealous with
> sound hardware. If it doesn't work, it's not the end of the world. If a
> storage subsystem that is probed doesn't work (or worse, silently fails
> in subtle ways), that's much worse.
> 
> -gordon

It's fine to put an unknown ID into a driver to do local testing and 
discovery.  It's less acceptable to commit that change to FreeBSD CVS
without knowing what it does and what impact it has.  Anyways, this
has been kicked around enough and I believe that David already said
that he will back out the change.

Scott


More information about the cvs-src mailing list