[alsa-devel] [PATCH 2/3] swarm_cs4297: Rename AC97 registers to use sound/ac97_codec.h definitions

[Takashi Iwai]

At Tue, 19 Jun 2012 09:41:30 -0300,
Ezequiel Garcia wrote:
> 
> On Tue, Jun 19, 2012 at 2:38 AM, Takashi Iwai <tiwai at suse.de> wrote:
> > At Mon, 18 Jun 2012 14:50:53 -0300,
> > Ezequiel Garcia wrote:
> >>
> >> Hi Takashi,
> >>
> >> On Mon, Jun 18, 2012 at 11:04 AM, Takashi Iwai <tiwai at suse.de> wrote:
> >> > At Mon, 11 Jun 2012 16:58:50 -0300,
> >> > Ezequiel Garcia wrote:
> >> >>
> >> >> This patch removes the last usage of linux/ac97_codec.h
> >> >> by renaming ac97 registers to use sound/ac97_codec.h definitions.
> >> >> This will enable us to remove linux/ac97_codec.h.
> >> >>
> >> >> Not even compilation tested.
> >> >>
> >> >> Cc: Ralf Baechle <ralf at linux-mips.org>
> >> >> Cc: Jaroslav Kysela <perex at perex.cz>
> >> >> Cc: Takashi Iwai <tiwai at suse.de>
> >> >> Cc: Clemens Ladisch <clemens at ladisch.de>
> >> >> Signed-off-by: Ezequiel Garcia <elezegarcia at gmail.com>
> >> >> ---
> >> >> Hi all,
> >> >>
> >> >> This patch is important so we can remove linux/ac97_codec.h usage.
> >> >> Since this driver is mips related, I can't test it until I install
> >> >> a mips toolchain.
> >> >> If someone can compile this for me, or even test it with real
> >> >> hardware I think it would be better.
> >> >> If not then I can install a mips toolchain and compile it myself,
> >> >> but I won't be able to test it on real hardware.
> >> >>
> >> >> This patch should be treated with carefully and be applied only
> >> >> if someone manages to test it.
> >> >
> >> > A slight concern by this change is that the driver includes
> >> > sound/ac97_codec.h although it's based on OSS framework.
> >> > sound/ac97_codec.h is the header for ALSA ac97 structs, and this can't
> >> > be mixed up with OSS.
> >> >
> >> > If the intention is only about AC97 register definition, we may split
> >> > ac97_codec.h into ac97_regs.h and ac97_codecs.h where the former
> >> > contains only the register definitions (thus framework-neutral) and
> >> > the latter includes the former.
> >> >
> >> >
> >>
> >> Yes, splitting sounds good to me. It could be useful for other ac97
> >> drivers (e.g. em28xx).
> 
> On a second thought, I'm not sure splitting the header is the best way
> to proceed. Since swarm just uses some AC97 register definition
> maybe we could just duplicate those (less than ten) macros in swarm
> c file.
> 
> It's a less intrusive aproach and it allows us to remove the unused
> linux/ac97_codec.h.

I'm OK in both ways.


Takashi