--Signature_Tue__10_May_2005_07_22_47_-0400_TW5zBZ2BZ5gXCBRi
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable

Hello,

> > > (To work around LP64 related GCC internal bugs in the HP-PA
> > > backend I need to use a 32 bit machine for building hp700...)
> > Ouch...=20
> Yes. Unfortunately there is still a lot of not-LP64 aware software out
> there. What was it with Mozilla / Firefox on LP64 big endian
> machines...

Indeed. I keep sending out bug reports about things like passing an
(int *) to functions that want (size_t *) and other trivialities that go
undetected on x86 but bite on sparc64...

> > > BTW: I am building -current at the moment to test the recent ATI
> > > PCI VGA support. I have a U10 ("Darwin") with "ATI Technologies 3D
> > > Rage I/II" and a U5 ("Darwin+") with an ATI Rage Pro. Will this
> > > work on Darwin and Darwin+ or Darwin+ only?
> > Both will work as console, but only the Rage Pro will work properly
> > with XFree.
> Sounds promising.

I got your PCI dump - looks like one main difference is that the Rage II
has only two BARs - the aperture and IO registers with the latter set to
0. The Pro has another memory BAR containing the register block. So
there are 3 ways to get the registers - BAR 0x14 (IO), BAR 0x18 (memory
mapped) or BAR (0x10) - end of the aperture. machfb always uses the end
of the aperture if I remember correctly. Maybe XFree tries to use the
other memory mapped block and doesn't realize that this can't work on
the Rage II because the BAR is unused. Or it falls back to IO which - as
far as I can tell - won't work on sparc64 either, at least not right now
( although that would be easy to fix ). The last possibility would be
that it gets the aperture size wrong, but that seems unlikely as long as
machfb gets it right.

> > For some reason it doesn't like the Rage II - feel free to poke
> > around=20
> I can run some testes if you instruct me, but I have no time left for
> just an other NetBSD hacking project. Sorry.

That's all I want - someone who tries a patch now and then.

> > - I have no way to do that myself right now since I don't have the
> > hardware ( my U10 has a Rage Pro ). Please mail me the dmesg output
> > of the machine and a PCI device dump=20
> =3D> PM.

Got it.

> > > Maybe this should be added to the FAQ on the port-sparc64 page at
> > > the NetBSD web site. ;-)
> > This stuff is brand new and needs testing :)
> I am a willing victim. ;-)

Thanks :)

have fun
Michael

--Signature_Tue__10_May_2005_07_22_47_-0400_TW5zBZ2BZ5gXCBRi
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (NetBSD)

iQEVAwUBQoCZh8pnzkX8Yg2nAQIiUQf/TijgYzCid0GK93LZG+0YgJNFJTK8BCFG
RhACrtSGx1SsNUTJfXfWIOmFGAehO21AXzt3qwHVjxKm8VJNVQseghp/hVp7m6w1
jE2QiRQ5wDB8OZTdSF9qn/UMrNFlcGjrpCcs+dmvegYCkh2cmcRKJjRU415jDXp9
dTKZkCtf8pq2ESewjg5qnsSfiQ6JnaeFhDtPIFgksxaPFs7rWw9bDWut21RS0icm
ivQ4AZiiY8MAbCMHgeG0r82+t38z7eRSlMeRqwkISi55XssbYKsCLAhJBOM7RKNW
EkRnhqH0ZYRuXPld5VkwZfBXqLpPEnELTJqxnUHGI9fjZoEebrBcTQ==
=DNZl
-----END PGP SIGNATURE-----

--Signature_Tue__10_May_2005_07_22_47_-0400_TW5zBZ2BZ5gXCBRi--