svn commit: r220185 - in head/sys: amd64/conf arm/conf conf dev/ath i386/conf mips/conf modules modules/ath modules/ath_ahb modules/ath_pci pc98/conf sparc64/conf

[Warner Losh]

On Mar 31, 2011, at 5:40 AM, John Baldwin wrote:

> On Thursday, March 31, 2011 4:07:13 am Adrian Chadd wrote:
>> Author: adrian
>> Date: Thu Mar 31 08:07:13 2011
>> New Revision: 220185
>> URL: http://svn.freebsd.org/changeset/base/220185
>> 
>> Log:
>>  Break out the ath PCI logic into a separate device/module.
>> 
>>  Introduce the AHB glue for Atheros embedded systems. Right now it's
>>  hard-coded for the AR9130 chip whose support isn't yet in this HAL;
>>  it'll be added in a subsequent commit.
>> 
>>  Kernel configuration files now need both 'ath' and 'ath_pci' devices; both
>>  modules need to be loaded for the ath device to work.
> 
> Err, there's no need to break the kernel config.  Just have the kernel DTRT if 
> both ath and pci are enabled using something like:
> 
> sys/dev/ath/if_ath_pci.c		ath pci
> 
> in sys/conf/files.
> 
> You can still have an if_ath_pci.ko module with this arrangement.  This is 
> what almost all other drivers due that have multiple bus backends.

Well, most drivers with multiple back ends compile all the relevant ones for the platform in.  There's very few that call out individual ones (ata and its kin being the big exception).

Personally, I'd not bother having a if_ath_pci.ko, and instead bundle it in with if_ath.ko and omitting if_ath_ahb.c from the x86 builds, but I understand why this might be important in the embedded builds.

Warner