On Thu, Feb 06, 2003 at 11:29:35AM +0100, Magnus Pfeffer wrote:
> On Tuesday 04 February 2003 17:03, Branden Robinson wrote:
>
> > Upstream doesn't build this module for Linux/PowerPC by default. I
> > suppose I could turn it on, but given your other bad experiences I don't
> > feel at all confident that it would work.
>
> Do accelerated drivers work on other ppc computers at all?
Yes. The various ati subdrivers (misc, r128, radeon) work fairly well,
though with occasional bugs. The imstt driver should work well given
that PowerPC is the native platform for such things.
> If so, it might be worth a try.
I'll mull it over. It would be helpful to know what criteria upstream
uses for deciding which drivers to turn on for various architectures.
Presumably, it's more complex than "if it builds, ship it", but perhaps
not. Perhaps i386-centric people update xfree86.cf and often forget to
enable drivers for other architectures.
I don't honestly know.
> > SuSE patches the hell out of their XFree86 (to be fair, all the major
> > distributions do). That could have something to do with it.
>
> I got the config file from a suse user and noticed that suse cheated: They
> simply used the "fbdev" driver. I did not know about this driver (always
> thought you should use "vga" with the UseFBDev switch).
Oh, yuck no. Use UseFBDev with an accelerated driver when possible, and
with fbdev otherwise.
> Using the "fbdev" driver the powerstack no longer crashes hard when starting
> X. The framebuffer device is queried for its capabilities, and X chooses the
> highest possible resolution. But with the Ati Mach64 as well as the GLD5446
> the outout is garbled, and stopping X leaves the framebuffer console
> unusable. No problems show up in the log files.
Hmph. Driver bugs. Please consider filing a bug for each of these.
Normal serverity, against xserver-xfree86.
> The Matrox Millenium still crashes the system hard even when used with the
> "fbdev" driver.
Hmph. A serious driver bug ("important" as far as the Debian BTS is
concerned). Please consider filing a bug on this. Important severity,
against xserver-xfree86.
It is possible that kernel bugs are also responsible for all 3 instances
of the above bad behavior. It takes two to tango when you're using
X with fbcon; X has to tell the kernel the right stuff, and the kernel
has to do the right stuff.
--
G. Branden Robinson | It's not a matter of alienating
Debian GNU/Linux | authors. They have every right to
branden@debian.org | license their software however we
http://people.debian.org/~branden/ | like. -- Craig Sanders
Attachment:
pgpkfCjIuU1Pz.pgp
Description: PGP signature