Re: switching to <sys/bus.h>: the non-trivial patches

[Masao Uebayashi <uebayasi%gmail.com@localhost>]

On Sat, Jul 9, 2011 at 12:45 PM, David Young <dyoung%pobox.com@localhost> wrote:
> On Fri, Jul 08, 2011 at 08:22:25PM -0700, Matt Thomas wrote:
>>
>> On Jul 8, 2011, at 12:21 PM, David Young wrote:
>>
>> > For review, here are the remainder of the patches that switch
>> > the architectures with PCI support from <machine/bus.h> to a
>> > common <sys/bus.h> that includes MD <machine/bus_defs.h> and
>> > <machine/bus_funcs.h>.
>> >
>> > Please review, test, comment.  Thanks.
>> >
>> > The bulk of this patch moves tons of inline bus_space(9) routines to
>> > appropriate .c files where they are not any longer inline.  The changes
>> > to PowerPC involving BUS_DMA_DONTCACHE need to be reviewed.
>> >
>> > I have not included in the patch the changes to umpteen
>> > <machine/types.h>'s that #define __HAVE_NEW_STYLE_BUS_H, or the deletion
>> > of umpteen <machine/bus.h>'s.
>>
>> We aren't installing <sys/buf.h> anymore so why are installing
>> bus_defs.h?
>
> I painstakingly got each PCI release to build, and that's what it took
> for alpha. *shrug*
>
> I can just let alpha break until somebody figures out why it wanted
> <machine/bus.h> in userland to begin with.

Kind of experimental userspace bus_space(9) implementation which past
architects envisioned?