On Mon, 26 Jun 2006, Martin Bochnig wrote:
#######################################################################
#######################################################################
##
## I have uploaded two files to
## http://fiesta.cs.tu-berlin.de/~mbeinsx/aperture_amd64_sun4v/
##
#######################################################################
#######################################################################
... it was summer solstice on June 20/21 (UTC Date).
And therefore also half-time between two "Happy Holidays"
seasons, Tempus Fugit!
So Cristmas is six short long months away - in whatever direction
we look. This will, however, not hinder me from releasing
my _few_ added bits to the public, which make XFree86's / Xorg's
aperture driver work on amd64 64 bit (Open)Solaris kernels, where
the un-open /dev/xsvc driver can not be distributed legally, and
where the lack of a working amd64-aperture module has been kind of
a show-stopper for over a year.
So I'm indeed publishing those changes, before I actually have out
marTux for x86/x64 (which I publically announce hereby) and because
of that give Belenix, Nextenda and Schillix the chance to be out
with a release featuring X11 in amd64 mode, before myself's marTux is.
So go, hurry! :-)
You may notice, that interestingly both XFree86 and Xorg still have
100% exactly the same apSolaris.shar inside their current CVS, last
modified in 2002 (though the webcvs entries and even revisions
look different at first):
http://cvsweb.xfree86.org/cvsweb/xc/programs/Xserver/hw/xfree86/etc/apSolaris.shar
http://webcvs.freedesktop.org/xorg/xserver/xorg/hw/xfree86/os-support/solaris/apSolaris.shar?view=log
--->>
-rw-r--r-- 1 bochnig bochnig 16546 Jun 26 00:30 XF86_apSolaris.shar
-rw-r--r-- 1 bochnig bochnig 16546 Jun 26 00:29 Xorg_apSolaris.shar
bash-3.1$ diff -cu XF86_apSolaris.shar Xorg_apSolaris.shar
No differences encountered
Both projects can (or could?) therefore use the same attached diff,
if they decide to incorporate something.
I also chose a new detection mechanism for ISA-dependent
selection of Makefiles: I use "isainfo -k" instead of "uname -m".
The reasons for this are:
#0.) You cannot determine with uname (on Solaris), whether or not
we are running on a plain x86, or on amd64. Especially
can't we determine, wich kernel we're running.
"uname -m" would always and only give i86pc on amd64.
#1.) sun4u is by no means the only implementation of sparcv9 anymore:
Take into account SUNW's throughput computing (sun4v) or - not
to forget - the vendor FJSV, that may become much more wide-
spread in the future, when SUNW/FJSV's APL will be out.
The ISA is important to us, rather then the machine platform.
To summarize this: Integrated support for generic sparcv9 - and therefore also
sun4v aka Niagara servers, later APL, Rock, Rock2 etc. in the mid term future.
All that by means of a rather cosmetical change.
This has now been committed to our repository, modulo a small number of
cosmetic changes.
Thanks for the patch!
Marc.
+----------------------------------+-----------------------------------+
| Marc Aurele La France | work: 1-780-492-9310 |
| Academic Information and | fax: 1-780-492-1729 |
| Communications Technologies | email: tsi@xxxxxxxxxxx |
| 352 General Services Building +-----------------------------------+
| University of Alberta | |
| Edmonton, Alberta | Standard disclaimers apply |
| T6G 2H1 | |
| CANADA | |
+----------------------------------+-----------------------------------+
XFree86 developer and VP. ATI driver and X server internals.
_______________________________________________
Devel@xxxxxxxxxxx
http://XFree86.Org/mailman/listinfo/devel
