RFC: separate 3dfx_linux module

[Scott Long]

Yar Tikhiy wrote:
> On Sat, Feb 25, 2006 at 03:11:02PM -0500, Coleman Kane wrote:
> 
>>I have unfortunately lost all of my voodoo hardware.
>>
>>On Sat, Feb 25, 2006 at 09:17:24AM -0700, Scott Long wrote:
>>
>>>Yar Tikhiy wrote:
>>>
>>>>Hi there,
>>>>
>>>>In the course of reviewing and cleaning up the default configuration
>>>>of kernel options, it was suggested by the Release Engineers that
>>>>we had a separate module for TDFX_LINUX instead of placing the
>>>>burden on the device tdfx and module 3dfx.  Could anybody interested
>>>>test this change?  I've made sure it builds, but I have no 3dfx hw
>>>>to really test it.  The testing is as simple as building the new
>>>>3dfx and 3dfx_linux modules, loading them, and verifying that the
>>>>linux apps work with the device as before.  Thanks in advance!
>>>>
>>
>>Sounds goo to me. I am all for further modularization of the codebase.
> 
> 
> Thanks!
> 
> 
>>>Why keep the TDFX_LINUX option defined in sys/conf/options?
>>
>>Sounds good to me. In the event that you want to build this statically
>>into the kernel, doesn't the option still need to be available,
>>or are we talking about a device tdfxlinux ?
> 
> 
> It was exactly my point, too: the TDFX_LINUX option has to be there
> so that people still can compile device tdfx with Linux support into
> the main kernel file.
> 

It's a little confusing though since it's now treated as a separate driver.

Scott