> I installed Xfree 2.1 about two days ago ( XF86_Mach32 ) but I always get
>a message stating that "none of the configured devices were found". I have
>tried with different settings in the Xconfig file with no luck. I have also
>tried different settings for the ATI Graphics Ultra Pro ( VLB 1 MB ) using
>the DOS install utility with no luck..
> Superprobe recognizes the card and Xfree 2.0 was running flawlessly
>before. I tried the XF86_8514 server and it works except for the fact that
>when I exit to text mode the screens garbled and text mode is not entered.
>Any help with the Mach32 server is appreciated...
>Ohh, btw, this is Linux 1.0, lib?.4.5.21 and ld.so 1.4.3
>There is one other person with the same problem in comp.windows.x.i386 but
>his post has not been answered.
>Appreciate your help! Thanks.
Well, I guess this question should actually make it to the Xfree FAQ,
but the new server has problems giving the ATI card enough time to
answer that it exists. (I'm not a XFree developer, so take this with a
grain of salt, but its mostly accurate I believe.) Anyway from what I
understand the 2.0 server had problems* up machines that didn't
have an ATI if they were accidentally run on them, so the code was
changed so that the server would exit after a certain period of time.
There are several solutions to be tried to fix this. I have experience
with a couple of them.
1) I have just reinstalled Linux on my computer from Slackware 1.2, and
the mach32 server from slackware seems to have the problem fixed because
it worked on my machine where the original server did not. I think I
saw a message in one of the Slackware files saying that he had relinked
the server, and he hoped it fixed the problem. So try getting the x
distrib from Slackware, and seeing if that kernel works.
2) Try relinking the server yourself. I haven't done this myself, so I
don't know where to look for all the required stuff, but I'm sure there
is a help file somewhere.
3) Get the source for the server, and recompile it without
optimizations. This seems to slow down the code enough to give the card
time to respond. I read that several people have done this, and it
worked, but have not done this myself.
4) Recompile the server, but instead of turning off optimization change
every occurence of ProbeWaitIdleEmpty to WaitIdleEmpty in the file
mit/server/ddx/x386/accel/mach32/mach32.c . This is what I did
initially to make the server work. I don't know whether this is an
official way to fix the problem (in fact I haven't seen an official way
to fix the problem, and I have been looking for one), but David Dawes
told me to do this to see if it would work when the mach server problem
first reared its head. Try this one at your own risk, because I don't
know if it does anything bad or dangerous to monitor or card, but I used
a server fixed this way on my machine for about two months (I think)
I think that's all the solutions I have seen for the problem. I saw
that someone had relinked the kernel and uploaded it somwhere, but I
can't remember where, and when I looked for it, I couldn't find it.
Hope this helps