Any mention of pgcc in a Mesa bug report spells trouble. We've had very
little luck with that compiler over the years, and there isn't much
willingness to put work into tracking down problems that occur on systems
which in any way rely upon it.
Sorry to be so blunt...
Keith
> BlueSky wrote:
>
> Dear Mesa Maintainers,
>
> I've never previously had any problems using Mesa, be it on Windows or Linux
> (normally with the RedHat distribution). I recently installed the Enoch
> distribution, and upon compiling and installing every version of Mesa I have
> been able to find, I get consistent problems (detailed further below). I
> have tried version 3.0, 3.1beta1 and 3.1 full release and 3.1 full release +
> vbrender.c patch.
>
> The significant characteristics of Enoch are that it is pentiumpro optimised
> - the main compiler is pgcc, and every program used on the distribution has
> been compiled with this. For compilation options on each program, the flags
> "-O6 -march=pentiumpro -mpentiumpro" have been used - this sets the compiler
> to optimise for the pentiumpro instruction set... but it also makes the
> compiled programs incompatible with previous intel chipsets.
>
> Whilst compiling Mesa I have tried compiling with and without the extra
> optimisation flags.
>
> What actually occurs is that, for instance, when I run the 'bounce' program
> in the Mesa-3.x/demos directory, I get a window with a ball bouncing, but
> the ball is made of alternately drawn triangles, rather than a completely
> drawn geosphere. Thus, you can see some of the triangles drawn that make up
> the geosphere, but for each of those, there is a gap where it is obvious one
> should have been drawn (that ratio is a guess of course). The background on
> this demo is made up of lines forming a kind of red grid on a black
> background. I can't remember what was supposed to be there, so I don't know
> if that is correct or not.
>
> For many of the demo programs, nothing is displayed at all. But for the
> program which displays Cindy Crawford, nothing is wrong - it seems to
> operate perfectly.
>
> I'm guessing it might be related to my X libraries, but I see no graphical
> artifacts in the operation of any other program. And I don't *think* X was
> compiled with pgcc... in fact, the X executable was not since I'm using a
> particular one for my Voodoo Banshee graphics card - but the X libraries may
> have been.
>
> I have GGI, and I've compiled the an early version of the new windowing
> system 'Berlin', which uses ggimesa. This produces a black window and the
> widgets of the windowing system aren't displayed except for the 'highlights'
> that give GUI components that 3d look. I could try the GGI framebuffer
> device and see if Mesa works okay with that... but given that GGI works fine
> in X (I've tried the demo programs and the show no sign of odd graphical
> behaviour) I'm inclined to believe the problem does not lie in GGI or X.
>
> If I have not provided enough information, please let me know.
>
> And thank you very much for your time - it would be fantastic if I were able
> to get Mesa working on my system again.
>
> Rajit
> Hardware Details:
>
> Pentium III 450
> 64Mb RAM
> 8Gb HD
> Voodoo Banshee Graphics Card (Diamond Monster Fusion)
> Most recent Enoch distribution of Linux (I think the version is 0.7)
>
> (incidentally, when I try to use my 3dfx card, programs segfault - but I
> figure, one thing at a time. the demos in 3dfx/demos used to run fine with
> 3dfx - whilst all the other demos segfaulted if I set MESA_GLX_FX=f - but
> now, after installing Mesa-3.1, they all segfault - and the test3Dfx app
> works fine. Anyway, just thought I'd mention it).
_______________________________________________
Mesa-bug maillist - [EMAIL PROTECTED]
http://lists.mesa3d.org/mailman/listinfo/mesa-bug
_______________________________________________
Mesa-dev maillist - [EMAIL PROTECTED]
http://lists.mesa3d.org/mailman/listinfo/mesa-dev
