Re: xserver segfaults

2002-01-17 Thread Arlie Glenn Capps
Hello,

I am one of the several trying unsuccessfully to use the accelerated driver for
Voodoo3 video cards.  I have a 6500, booting out of BootX, running a 2.4.16
kernel that I recompiled to have Voodoo framebuffer support.

> http://v.favrenicolin.free.fr/linuxppc/ is where you want to go to get some 
> information about voodoo3 and linux...
> 

I am trying to follow the procedure on this page, but I haven't been able to get
fbset to work properly: when I run fbset it says that I am running at
1024x768-116 (?!).  When I try to set the mode (say, to 1024x768-75 or
800x600-75 or whatever) nothing appears to happen, and the next time I run it
fbset says I am still at 1024x768-116.  I am including in the kernel arguments
video=tdfx:vmode=16:cmode=2,noaccel.
What might I be doing wrong? :)

Thanks very much.
Arlie Capps

I am trying to _find_ the fine manual!



Re: xserver segfaults

2002-01-17 Thread Arlie Glenn Capps
>You're probably running OFfb, which can't even know what the real mode
>is, let alone change it. You need to get tdfxfb going.
> 
> 

Yes, when I "cat /proc/fb" it says something like "OFfb Voodoo3" (I'm not using
that computer right now, but something like that).
Anyway, when I recompiled the kernel I chose Yes to 3Dfx Banshee/Voodoo3 display
support (CONFIG_FB_3DFX) as well as Open Firmware frame buffer support
(CONFIG_FB_OF).  Should I recompile with those as modules, or does tdfxfb depend
on other config options?

Thanks very much,
Arlie Capps



X now working with a Voodoo 3

2002-01-18 Thread Arlie Glenn Capps
Hooray!  Thank-you all very much for the help!  I realized what I was doing
wrong.  I was booting using BootX, with the "no video driver" check-box checked,
which ensures that the framebuffer is used.  Once I unchecked that and put the
right "video=tdfx:" kernel arg in, things started working beautifully.  :)

Thank-you,
Arlie Capps



A troubling plethora of OOPSs

2002-01-21 Thread Arlie Glenn Capps
Hello,

I have been having a troubling problem: kernel Oops have been occuring quite
often.  It may be that I am running out of memory; my PowerPC 6500 has 32MB RAM
and 80 MB swap.  This problem only seems to happen in X, running several
different apps, but quite inconsistently.  However, when I looked at the log
files, most of the time the first thing that causes an Oops is kswapd; other
times it has been dpkg and ps.  I understand that this is a very vague
description.  I'm running woody with packages from main, non-free, contrib and
non-US.

Can anyone tell me how I might check this out a little better?  or how I can
make sure that it isn't a configuration problem?  Also, is this the right
mailing list for this problem?

Thanks very much,
Arlie Capps



Re: A troubling plethora of OOPSs

2002-01-21 Thread Arlie Glenn Capps
> Hello,
[snip]
> I'm running woody with packages from main, non-free, contrib and
> non-US.
Oh yes, and kernel 2.4.16.

And now INIT is panicking because of segfaults and sleeping for 30 seconds.(!!) 
I don't think it's hardware, because right now Mac OS is running just fine.  I
looked back at old logs and the very first thing that gave an oops was telnet
client.

What can I DO?  How can I diagnose this better?

Thanks very much,
Arlie Capps



2.4 kernel on a 6500

2002-02-13 Thread Arlie Glenn Capps
Hello,

I will soon compile the 2.4 kernel on my 6500.  I have two questions:
1. A while ago (the last time I did this) I had to configure for USB 
support in order to use the new input layer, even though the 6500 
doesn't have USB.  Is this still the case?
2. I'm pretty sure I must configure support for via-cuda, but do I need
to also choose support for pmu? 

Thanks very much for the help.  Also, I looked all over for doc's
on the new input layer before writing this, but every link to "Franzo's 
Blurb on the New Input Layer" gets me 404/Not Found.  Are there other 
good places to look?

Thanks,
Arlie Capps



Re: 2.4 kernel on a 6500

2002-02-13 Thread Arlie Glenn Capps
Thanks very much, vinai, Nathan Ingersoll.

I will try that and see what happens.

Arlie Capps



X with 2.4 kernel on a 6500

2002-02-18 Thread Arlie Glenn Capps
Hello!

I have a strange problem with my 6500/275, with 96MB of RAM and about twice that
of swap.  I installed Woody and compiled a 2.4.16 kernel (thanks, vinai, Nathan
Ingersoll), with the tdfx framebuffer driver.  It is booting fine with BootX,
and so I installed the X window system.  Now it has severe problems, which show
up in usually two different ways:

1.  When I run startx, many startup messages scroll past, the screen goes black
for four or five seconds, then the text screen comes back with the last message
shown being something about XKB not working or failing to load (I am writing
this in the Mac side; I don't have the exact message).

2.  X comes up successfully, I can use it, but when I exit strange things
happen.

At this point, after either 1. or 2., every command I type that is not a shell
builtin gets either "Segmentation fault" or "Invalid instruction", and I have to
reboot with Control-Option-Power.  I think when I'm root it gives "Invalid
instruction" and as a normal user "segfault", but nothing at all shows up in the
log files.

I decided to be tricky, to avoid the 3-finger salute:  as root I did "shutdown
-h 5", switched to a different virtual terminal, got X running, quit out of X,
verified that any command segfaulted, then sat back and waited for the impending
reboot: no such luck.  At the end of five minutes the shutdown command
segfaulted and I got the root command prompt back, and every command I typed
segfaulted.  

Once it gave an Oops instead of the above problem (the oopses did get logged); I
ran the result through ksymoops.  I got these warnings:
Warning (read_lsmod): no symbols in lsmod, is /proc/modules a valid lsmod file?
Warning (compare_maps): mismatch on symbol xchg_u32  , ksyms_base says c000ba08,
vmlinux says c0006728.  Ignoring ksyms_base entry
Warning (Oops_read): Code line not seen, dumping what data is available
All three oopses started with this line:
>>NIP; c003b168<=

I am feeling a bit glum.  Has anyone experienced or fixed problems like these? 
or do I just have wierd and defective hardware?  Many thanks in advance for
help, ideas or pointers to documentation.

Arlie Capps



Re: X with 2.4 kernel on a 6500

2002-02-18 Thread Arlie Glenn Capps
I forgot one important bit:  I have a Voodoo 3 installed, and these problems
only happen when I use the tdfx driver.  No problems with the framebuffer
driver, but it insists on running at 800x600 at a very low refresh rate.

Thanks much,
Arlie Capps



2.4.16 on a 6500

2002-02-20 Thread Arlie Glenn Capps
It looks like no-one else has had these problems with segfaulting/illegal
instructions after using X with the tdfx driver.

Hmm.  I also a few days ago had a problem during the middle of using dselect,
without running X first.  The console font, specifically capital T and capital
P, was corrupted and segfaults started happening all over the place.  I have a
6500 with 96 MB of RAM and a Voodoo 3.  If anyone has overcome such problems, or
even heard of them happening, I'd appreciate an e-mail (then again, it may well
be that my hardware is fried...)

Thanks in advance,
Arlie Capps