Bug 184072
Summary: | OGL call freezes X11 | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Michael Godfrey <godfrey> | ||||||||
Component: | xorg-x11-server | Assignee: | Adam Jackson <ajax> | ||||||||
Status: | CLOSED CURRENTRELEASE | QA Contact: | David Lawrence <dkl> | ||||||||
Severity: | urgent | Docs Contact: | |||||||||
Priority: | medium | ||||||||||
Version: | 5 | CC: | gresko, j, mcepl, ncunning, xgl-maint | ||||||||
Target Milestone: | --- | Keywords: | Reopened | ||||||||
Target Release: | --- | ||||||||||
Hardware: | All | ||||||||||
OS: | Linux | ||||||||||
Whiteboard: | |||||||||||
Fixed In Version: | FC7 | Doc Type: | Bug Fix | ||||||||
Doc Text: | Story Points: | --- | |||||||||
Clone Of: | Environment: | ||||||||||
Last Closed: | 2007-08-27 14:05:00 UTC | Type: | --- | ||||||||
Regression: | --- | Mount Type: | --- | ||||||||
Documentation: | --- | CRM: | |||||||||
Verified Versions: | Category: | --- | |||||||||
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||||
Cloudforms Team: | --- | Target Upstream Version: | |||||||||
Embargoed: | |||||||||||
Attachments: |
|
Description
Michael Godfrey
2006-03-05 21:59:48 UTC
This seg fault seems to happen on more FC5 test machines. Here is the tail of the same crash on another machine (also ATI video card): Backtrace: 0: /usr/bin/Xorg(xf86SigHandler+0x87) [0x80b85e7] 1: [0xf56420] 2: /usr/lib/xorg/modules/extensions/libGLcore.so(_mesa_make_current+0xf2) [0x74abf2] 3: /usr/lib/xorg/modules/extensions/libGLcore.so(XMesaMakeCurrent2+0x55) [0x8a40f5] 4: /usr/lib/xorg/modules/extensions/libGLcore.so(__MESA_makeCurrent+0x55) [0x88c2f5] 5: /usr/lib/xorg/modules/extensions/libglx.so(DoMakeCurrent+0x3f8) [0x1ccce8] 6: /usr/lib/xorg/modules/extensions/libglx.so(__glXMakeCurrent+0x39) [0x1cceb9] 7: /usr/lib/xorg/modules/extensions/libglx.so [0x1cf9c7] 8: /usr/bin/Xorg(Dispatch+0x19b) [0x80888eb] 9: /usr/bin/Xorg(main+0x487) [0x80701d7] 10: /lib/libc.so.6(__libc_start_main+0xdc) [0x9947e4] 11: /usr/bin/Xorg(FontFileCompleteXLFD+0xb9) [0x806f511] Fatal server error: Caught signal 11. Server aborting 2 additional notes: 1. This works correctly under FC4 with all latest updates. 2. I can supply URL to download (source), and provide install instructions for the application which causes this failure. Note for the future: When adding files to bugzilla, please do not cut and paste huge log files/configs into the bug comment box. Use the bugzilla file attachment feature below, so that the bug remains readable. Sorry about that. Will do bettewr next time. This error report seems to be static in the NEEDINFO_ENG state. Does this mean that I should do something? This error will prevent us from using FC5, so I would like to get it fixed. NEEDINFO_ENG means "need information from the assigned engineer", which would be me. i definitely need to know the application that triggers this bug in order to reproduce it. Created attachment 126057 [details]
Source of program which seg faults X11
BUG: 184072
Here are the instructions for recreating the OGL X11 seg
fault:
Compiling and running the attached Magic source:
First, make sure you have tcl/tk installed. Of course, you
need the OGL stuff, too.
Create a new directory for the attached file: magic-7.4.6.tgz
tar -xzf magic-7.4.6.tgz
cd magic
./configure
make
su root
make install
exit
magic -d OGL & (needs /usr/local/bin in your path)
At this point 2 windows will appear: a drawing window and a console
window. The drawing window should have a row of menus at the top.
Click on options, and select toolbar. If your experience is like mine,
X11 will crash at this point (instead of drawing a toolbar along
the right hand side of the window).
The compile/installation will put the installed stuff in /usr/local/bin
and usr/local/lib. If you do not want this, it is possible to name
the target directories on the ./configure command. Check the included
documentation about this if needed.
Michael
godfrey.edu
I just installed the FC5 release and tried Magic -d OGL. Now it works without segment fault. Therefore, this error report should be closed. Thanks, Michael Sorry, I spoke too soon. I just tried the same test as before on an FC5 system with an ATI Radeon 9200PRO, which provides "direct rendering: on" It crashed X11 as before. All FC5 updates as of today (25 Mar) had been applied. Michael Created attachment 126866 [details]
Xorg.log.old (from FC5 release, updated to 27 March)
This is similar the the previous Xorg.log, but is from the
current FC5 release Xorg level.
One more bit of information: the Magic maintainer said: First, it tends to bring out all the errors in OpenGL implementations (especially the toolbar, which uses off-screen rendering, which has proven to be the most likely source of error for OpenGL drivers). This was in response to my question about turning on the OGL capability in Magic. Installed new xorg-x11-server (xorg-x11-server-Xorg-1.0.1-9.fc5), but it crashes as before. I also get crashes on fc5 + OpenGL with radeon driver and also with AT fglrx driver. Does this problem happen with Fedora development X? You'll need the X server, mesa, and all their dependencies to test. We'll be releasing X11R7.1 for FC5 as an update, so it would be nice to know if the problem is already fixed there. TIA Mike, I will give it a try soon. I assume the 7.1 stuff is in the development tree somewhere. I looked in .../updates/testing/5/i386 but found nothing that looked like a 7.1 update of xorg. We are hoping for some other corrections to xorg as well, so please tell me when 7.1 is ready for test. Thanks, Michael Created attachment 132874 [details]
zip file containg xorg.conf, Xorg.0.log, and rpm.xorg
Configuration and log files
It appears that one of my comments got lost. The attachement referred to in comment #18 comes from an FC6 development system. I followed the suggestion of Mike Harris in comment #15 and built an FC6 development system. It fails in the same manner as FC5. I have retested this error when xorg updates have appeared. The failure still happens as before. Michael I tried the Magic -d OGL test again on 18 August, using the current development FC6. The problem still exists as before. It is getting closer to FC6 release, so it might be good to work on this. Thanks, Michael I tried the Magic -d OGL test again on about end of August, using current development FC6. The problem still exists as before. It is getting closer to FC6 release, so it might be good to work on this. Probably too late for FC6 release. Too bad. Is OGL being given up on, or what? Thanks, Michael I tried the Magic -d OGL test again on 4 October, using current development FC6. The problem still exists as before. It is getting closer to FC6 release, so it might be good to work on this. Probably too late for FC6 release. Too bad. Michael I'm seeing an essentially identical backtrace on a Radeon 9100 IGP (the machine is an Asus Pundit-R). The application that triggers it is Matlab 2006a. The user seems to be able to trigger it with regularity, although I do not yet have a complete set of instructions for doing so. Just a wild guess here: does setting the environtment variable XLIB_SKIP_ARGB_VISUALS=1 make a difference? does setting the environtment variable XLIB_SKIP_ARGB_VISUALS=1 make a difference? Using the current FC6 development system, the answer is NO. My users have now started reporting this same crash on machines with Savage video: 01:00.0 VGA compatible controller: S3 Inc. VT8375 [ProSavage8 KM266/KL266] so this seems to be a Mesa problem instead of an ATI problem. This is now really starting to kill me, as we use Matlab extensively here. I'll first try disabling DRI and if it doesn't work I'm not sure what I can do. It is now close to 10 months since this error was reported, and the problems it causes seem to be accumulating. Since this is easily reproduced with the program supplied it should not take more than about 10 minutes to recreate the seg. fault in a debig enviroment. Anyone have a spare 15 minutes for us users? I cannot reproduce it here with Fedora Core 6, Intel 915GM (laptop) and "intel" driver in Xorg. I have no clue what the program is supposed to do, but I can certainly compile it, install, and run via menus (at least to the point of clicking on File/Quit). Could you please try to reproduce this bug with FC6? From Comment #28: >>Could you please try to reproduce this bug with FC6? Yes, I did that and it crashes as it did before under FC5 and all test releases of FC6. Could you review the test instructions? It is essential that OGL is used. The instructions are: magic -d OGL & At this point 2 windows will appear: a drawing window and a console window. The drawing window should have a row of menus at the top. Click on options, and select toolbar. If your experience is like mine, X11 will crash at this point (instead of drawing a toolbar along the right hand side of the window). ============================================================= Are you sure your machine has OGL installed an it is supported? Michael Also note that so far I have only seen this problem on Radeon and Savage chips; it is possible that Intel is somehow immune. I have compiled program per your instructions, then run according to your instructions (of course, I have included -d OGL to the command line) and it worked. To make it short, my experience was not like yours :-). As of the X11 updates to FC6 of 21 November, magic -dOGL no longer causes a seg fault in X11. Instead it freezes the console screen. Reset, or login from another machine is the only way out, it seems. I tried 2 machines and got the same results. So, I guess that the summary should be changed from "crash" to "freeze." Michael Concerning comment 30 -- yes, I was testing this on i915GM, so you may be right. Fedora Core 5 is no longer supported, please, could you reproduce this bug with the updated version of the currently supported distribution (Fedora Core 6, or Fedora 7, or Rawhide)? If this issue turns out to still be reproducible, please let us know in this bug report. If after a month's time we have not heard back from you, we will have to close this bug as CANTFIX/INSUFFICIENT_DATA. Setting status to NEEDINFO, and awaiting information from the reporter. Thanks in advance. I just tried the test on current F7. Now it works just fine. Thanks very much for correcting this problem. Michael Godfrey |