[PATCH] Adding shared code support for ia32 and amd64 -- x86 sub-branch

[John Baldwin]

On Wednesday 24 February 2010 10:50:49 am Attilio Rao wrote:
> 2010/2/24 John Baldwin <jhb at freebsd.org>:
> > On Wednesday 24 February 2010 10:18:34 am Attilio Rao wrote:
> >> 2010/2/16 Kostik Belousov <kostikbel at gmail.com>:
> >> > On Tue, Feb 16, 2010 at 01:10:37AM +0100, Attilio Rao wrote:
> >> >> The following patch:
> >> >> http://www.freebsd.org/~attilio/x86.diff
> >> >>
> >> >> starts the effort for having a shared sub-tree between amd64 and ia32.
> >> >> In this initial pass I putted the low-hanging fruits (bios/cpufreq)
> >> >> and what my customer was more interested in (isa/*) in order to
> >> >> kick-off the effort and, in the future, move gradually the code there.
> >> >> With the machine/isa/* cleanup about 10 files are trimmed and I'm sure
> >> >> more can be achieved easilly.
> >> >> There are few things to discuss. One, that I had not necessity to dig
> >> >> about still, is about how to organize headers (include/). Maybe some
> >> >> replication ala pc98 may be good.
> >> >>
> >> >> The patch is big but it is mostly added and removed files (look at the
> >> >> files.X in order to understand better how files movements happened).
> >> >>
> >> >> Hope to see comments and reviews.
> >> >
> >> > IMO the diff is unreadable. I suggest to do actual svn cp (not svn mv)
> >> > operation now, without a review, and post a diff that should be applied
> >> > to x86/ directory, as well as to build glue.
> >>
> >> I think that this patch juices out all the relevant part without noise:
> >> http://www.freebsd.org/~attilio/x86-2.diff
> >
> > I think this looks good.  We should likely be unifying the approach to
> > suspend/resume for timers across i386 and amd64 btw.  pmtimer should be
> > available for amd64 as well for example.  I'm also not sure if adding a resume
> > method for atrtc means that pmtimer needs to change to not frob the RTC in its
> > suspend and resume methods now as well.
> 
> Yes, I would do this (and other simple, already compelling,
> unifications, like the e/rflags one) into further passes.
> In this case, probabilly, more mealpieces we do the better it is, IMHO.

Yes, I would definitely split this up to move single "entities" (e.g. smbios
or atpic) per commit.

-- 
John Baldwin