From: Petr V. <VAN...@vc...> - 2004-08-03 11:26:14
|
On 3 Aug 04 at 13:09, Jonas Meurer wrote: > On 03/08/2004 Petr Vandrovec wrote: > > > i'll try to get an oops trace and some further debugging information. > > > > It would help a lot. > > sorry, it's harder than i even thought. as keyboard and screen are both > frozen, i cannot lurk for console messages etc. and as i've no second > machine, serial console is not possible. > > the logs are silent - they don't give any information: > syslog has the boot logs, and directly afterwards the APCI button power > off message, so nothing in the time where the freeze happened. > XFree86.0.log.old ends with the successful startup of xdm, what finally > causes the freeze. so no log information here too. > > and even kern.log doesn't give any information. > > and i don't understand ksymoops, when i execute it, it gives only: > Error (regular_file): read_ksyms stat /proc/ksyms failed > ksymoops: No such file or directory > No modules in ksyms, skipping objects > No ksyms, skipping lsmod ksymoops do not work on 2.6.x kernel. When you build 2.6.x kernel, you have option (Kernel hacking -> Compile the kernel with debug info) to store all symbols into kernel image, and then 2.6.x kernel will translate oops from numbers to symbols for you, without any need for help from ksymoops (and in this case your klogd should be started with option '-x' or '-2'). > > In 2.4.x it was marked as EXPERIMENTAL only because nobody cared. > > As driver is more than 5 years old, I think that I can say that it > > is well tested now. > > full ack, but in my case the driver doesn't seem to be very stable. > btw. i didn't have any problems with matroxfb until the athlon64 system > with 2.6 kernel (all previous machines ran 2.4 kernels). Do you run 32bit or 64bit kernel? Strange thing is that while you are in the X matroxfb is completely off. Did you tried booting with 'video=matroxfb:disabled' to find whether it does not crash without matroxfb too? Or try 'video=matroxfb:init' so matroxfb initializes your video hardware from scratch, without relying on BIOS doing its job correctly. Or try vesafb instead (if x86-64 supports it). > system boots, xdm starts, and therefore starts xserver, login box pops > up, you're able to input some chars, and then keyboard and screen > are frozen. > only APCI power off button still works to power off the system. Does it power off your box immediately, or after you hold it down for 4 seconds? > another scenario that happens sometimes: system boots, xdm starts, > immediately after login box popped up, i change to tty1, move between > all vc's (tty1-6 and tty8-12), do what i want, and sometimes, nevermind > if 5minutes or 3hours later, switch back to tty7 (with xdm running), and > during this switch, directly after showing a grey screen the first time, > the freeze happens. Can you recheck that you have 'Option "UseFBDev" "true"' and 'Option "HWcursor" "off"' in the XFree config? > > Neither of these bugs are in the matroxfb - they are features of fbdev/fbcon > > layer. > > so no plans to fix them? First three are fixed in my matroxfb patches (which you can use with other drivers too - I use it with radeon on my notebook and with vesafb on Parhelia on my other system). But as yesterday even full removal of VT from kernel was proposed on LKML, chance to get these changes in is pretty low. And for DRI I do not understand X code well enough to find where tests which should prevent X from hitting hardware should be added. Petr Vandrovec |