svn commit: r265359 - head/sys/modules/sound/sound

[John-Mark Gurney]

Marius Strobl wrote this message on Tue, May 06, 2014 at 18:01 +0200:
> On Tue, May 06, 2014 at 05:45:05PM +0200, Hans Petter Selasky wrote:
> > On 05/05/14 22:39, Warner Losh wrote:
> > >
> > > On May 5, 2014, at 8:31 AM, Hans Petter Selasky <hselasky at freebsd.org> wrote:
> > >
> > >> Author: hselasky
> > >> Date: Mon May  5 14:31:34 2014
> > >> New Revision: 265359
> > >> URL: http://svnweb.freebsd.org/changeset/base/265359
> > >>
> > >> Log:
> > >>   Build the kernel sound module without ISA DMA support for ARM and MIPS
> > >>   platforms, because these platforms do not implement the ISA DMA
> > >>   API. Else the sound modules cannot be loaded when running these
> > >>   platforms.
> > >>
> > >>   MFC after:	2 weeks
> > >>
> > >> Modified:
> > >>   head/sys/modules/sound/sound/Makefile
> > >>
> > >> Modified: head/sys/modules/sound/sound/Makefile
> > >> ==============================================================================
> > >> --- head/sys/modules/sound/sound/Makefile	Mon May  5 11:50:52 2014	(r265358)
> > >> +++ head/sys/modules/sound/sound/Makefile	Mon May  5 14:31:34 2014	(r265359)
> > >> @@ -44,7 +44,8 @@ CLEANFILES+=	feeder_eq_gen.h feeder_rate
> > >>
> > >> EXPORT_SYMS=	YES	# XXX evaluate
> > >>
> > >> -.if ${MACHINE_CPUARCH} == "sparc64" || ${MACHINE_CPUARCH} == "powerpc"
> > >> +.if ${MACHINE_CPUARCH} == "sparc64" || ${MACHINE_CPUARCH} == "powerpc" || \
> > >> +    ${MACHINE_CPUARCH} == "arm" || ${MACHINE_CPUARCH} == "mips"
> > >> # Create an empty opt_isa.h in order to keep kmod.mk from linking in an
> > >> # existing one from KERNBUILDDIR which possibly has DEV_ISA defined so
> > >> # sound.ko is always built without isadma support.
> > >
> > > Rather than an opt-in approach here, why not lust list the three that have it since
> > > they are the only ones that will ever have it?
> > >
> > > Warner
> > >
> > 
> > Hi,
> > 
> > Maybe the sound DMA file belongs in its own module?
> > 
> 
> It's not just sndbuf_dma.c but also the #ifdef'ed DEV_ISA code in
> channel.c on architectures that have ISA but not ISA DMA support,
> hence the current kludge.

I thought busdma was suppose to abstract all of this away?  Is it
just that this was never complete? or?

-- 
  John-Mark Gurney				Voice: +1 415 225 5579

     "All that I will do, has been done, All that I have, has not."