|
Bugzilla – Full Text Bug Listing |
| Summary: | sax2 crashes on Intel i810 graphics | ||
|---|---|---|---|
| Product: | [openSUSE] SUSE Linux 10.1 | Reporter: | Michael McCarthy <sysop> |
| Component: | SaX2 | Assignee: | Steffen Winterfeldt <snwint> |
| Status: | RESOLVED FIXED | QA Contact: | E-mail List <qa-bugs> |
| Severity: | Major | ||
| Priority: | P5 - None | CC: | eich, mrmazda, ms, pgunreben, sndirsch, snwint, suse-beta |
| Version: | Final | ||
| Target Milestone: | --- | ||
| Hardware: | i586 | ||
| OS: | SuSE Linux 10.1 | ||
| Whiteboard: | |||
| Found By: | Customer | Services Priority: | |
| Business Priority: | Blocker: | --- | |
| Marketing QA Status: | --- | IT Deployment: | --- |
|
Description
Michael McCarthy
2006-07-28 13:49:49 UTC
this is a known bug while calling X -probeonly while another X-Server is running. It's a bug in the intel driver and happens on i830 mostly I suggest to use "sax2 -a" instead to write an initial xorg.conf. Later you can start sax2 again *on* your desktop to prevent that a second Xserver is started and configure your monitor, etc. if required. Currently we have no other "fix" for this issue. :-( OK. I did the "sax2 -a" and restarted. Then I go into sax2 via yast. After setting monitor and resolution (either Dell E772p or VESA 1024x768@85Hz), I do a "Test" then a "Save". After exiting the save, I get no GUI. The monitor reports "Frequency Out of Range 87K/109Hz" which is what I am getting in bug # 195623 as well. It looks like the i810 driver is very broken. Do *not* test the configuration. This will start a second Xserver again. :-( Here we go again. I set the monitor to a Dell E772p, hit save with no test. After restarting, I get invalid frequency (72K/89Hz). After running sax2 -a and running it again, I forced the maximum refresh frequency to 85Hz then saved it again with no test. This time it seems to function. It also appears to be OK with VESA 1024x768@85Hz for a monitor. Could you please test with latest X.Org and i810 driver again with SUSE 10.2 Alpha3? BTW, this looks related to Bug #195623. Is there a way to test this without having to load Alpha3? Yes, it would be possbile by using the X.Org 7 RPMs for SUSE 10.1 from our openSUSE buildservice, Here is the URL for installation. http://software.opensuse.org/download/xorg7/Java_Sun-Java-1.5_SUSE_Linux_10.1/ I think you can add it as additional installation source to YaST and select all the packages from it. As long as you cannot test the newer driver setting to LATER. reopen to set it to NEEDINFO. NEEDINFO openSUSE 10.2 Beta1 has just been released. beta 1 with i815 sax2 and sax2 -a and sax2 -i -r in runlevel 3 all hang with xmode using 100% CPU. Quite a number of processes require kill -9 in order to try again. Oh well. I think I could reproduce this with my i810 machine. xmode hangs? This gfx chip/driver is really very "special". That may be, but it works on 10.0 using the very same system as the comment 13 report. > xmode hangs? This gfx chip/driver is really very "special".
I can reproduce this
# ps aux
[...]
/usr/sbin/xmode -x 800 -y 600 -d -1075647892 -r 75 -s 40
What a dacspeed. Do we need another check for this gfx driver? :-(
Marcus, you can reproduce this on d208.
Even "sax2 -a -r -m 0=vga" hangs ("sysp -s xstuff").
> Even "sax2 -a -r -m 0=vga" hangs ("sysp -s xstuff").
The same for "... -m 0=vesa".
still investigating, by the way "hwinfo --monitor" hangs as well and produces no output. please do not reboot the machine, thanks problem is hwinfo returns a null pointer for the ddc detection.
This shouldn't happen. I fixed sysp to be aware of that. The fix is
installed on d208. The call of
sax2 -r -a
produces a valid config file at 800x600 (because no other information is
available). Does X running with this file ?
anyway it takes a long time. This is because hwinfo --monitor takes
a very long time to finish with no output
X runs fine with this configuration. I assign this bugreport to you, so you can add the fix/workaround to SaX2 for Beta2. Do we need this fix/workaround also for hwinfo? I've added Steffen therefore. I think it might be interesting for Steffen because hwinfo is consuming 100% CPU time for half a minute because it returns with simply nothing :-) I already committed my patch so I will reassign this to Steffen. I'm pretty sure it makes sense to still have d208 open for testing Thanks I presume d208 is some computer belonging to the SUSE team? Yes. *** Bug 217005 has been marked as a duplicate of this bug. *** I read https://bugzilla.novell.com/show_bug.cgi?id=210923#c28 and upgraded the indicated packages by doing rpm -Uvh * from the download directory after fetching them all with wget. Is there an easier way using apt or rug or something just by providing a download source location? On i815 sax2 -a generates a valid generic monitor 800x600 xorg.conf. sax2 and sax2 -i -r both generate identical valid 1600x1200 xorg.conf files for my display. ad comment 23: just looked at d208 and it really does take that long (the bios is not crashed and loops endlessly or so). In fact libhd will cut out bios runs after 20s but the one on d208 finished after 19.5. Double the times for 2 ports and you end up with 40s. Workaround: boot with 'hwprobe=-bios.cpuemu,bios.ddc.ports=1'. *** Bug 218744 has been marked as a duplicate of this bug. *** |