Bug 524244

Created attachment 362899 [details]
X log with 20090927.16.iso

With desktop-i386-20090927.16.iso, GDM appears and both displays are active. X server crashes after logging in.

Backtrace:
0: /usr/bin/Xorg (xorg_backtrace+0x3c) [0x80e562c]
1: /usr/bin/Xorg (0x8048000+0x5eae6) [0x80a6ae6]
2: (vdso) (__kernel_rt_sigreturn+0x0) [0x66240c]
3: /usr/lib/libpixman-1.so.0 (0x261000+0x38b0) [0x2648b0]
4: /usr/lib/libpixman-1.so.0 (0x261000+0x3f6d5) [0x2a06d5]
5: /usr/lib/libpixman-1.so.0 (0x261000+0x1472a) [0x27572a]
6: /usr/lib/libpixman-1.so.0 (0x261000+0x39cf0) [0x29acf0]
7: /usr/lib/libpixman-1.so.0 (0x261000+0x41a4a) [0x2a2a4a]
8: /usr/lib/libpixman-1.so.0 (0x261000+0x42c6b) [0x2a3c6b]
9: /usr/lib/libpixman-1.so.0 (0x261000+0x39994) [0x29a994]
10: /usr/lib/libpixman-1.so.0 (0x261000+0x15294) [0x276294]
11: /usr/lib/libpixman-1.so.0 (0x261000+0x3b94f) [0x29c94f]
12: /usr/lib/libpixman-1.so.0 (0x261000+0x15294) [0x276294]
13: /usr/lib/libpixman-1.so.0 (0x261000+0x46dc4) [0x2a7dc4]
14: /usr/lib/libpixman-1.so.0 (0x261000+0x15294) [0x276294]
15: /usr/lib/libpixman-1.so.0 (0x261000+0x4d59a) [0x2ae59a]
16: /usr/lib/libpixman-1.so.0 (0x261000+0x15294) [0x276294]
17: /usr/lib/libpixman-1.so.0 (pixman_image_composite+0x177) [0x29b7e7]
18: /usr/lib/xorg/modules/libfb.so (fbComposite+0x170) [0x4cad70]
19: /usr/lib/xorg/modules/drivers/intel_drv.so (0x80b000+0x5a0ec) [0x8650ec]
20: /usr/lib/xorg/modules/drivers/intel_drv.so (0x80b000+0x5850a) [0x86350a]
21: /usr/bin/Xorg (0x8048000+0xd3647) [0x811b647]
22: /usr/bin/Xorg (CompositePicture+0x290) [0x810eba0]
23: /usr/bin/Xorg (0x8048000+0x16834d) [0x81b034d]
24: /usr/bin/Xorg (0x8048000+0x168669) [0x81b0669]
25: /usr/bin/Xorg (0x8048000+0xa1668) [0x80e9668]
26: /usr/bin/Xorg (0x8048000+0xa1761) [0x80e9761]
27: /usr/bin/Xorg (0x8048000+0x4b623) [0x8093623]
28: /usr/bin/Xorg (MapWindow+0x15e) [0x80978ae]
29: /usr/bin/Xorg (0x8048000+0x25809) [0x806d809]
30: /usr/bin/Xorg (0x8048000+0x26167) [0x806e167]
31: /usr/bin/Xorg (0x8048000+0x1a855) [0x8062855]
32: /lib/libc.so.6 (__libc_start_main+0xe6) [0x9c7b36]
33: /usr/bin/Xorg (0x8048000+0x1a441) [0x8062441]
Segmentation fault at address 0x89f4470

Fatal server error:
Caught signal 11 (Segmentation fault). Server aborting

I have the same problem on I945GMS.
Have noticed there is bug #509712 with similar backtrace

Backtrace:
0: /usr/bin/X (xorg_backtrace+0x28) [0x49e758]
1: /usr/bin/X (0x400000+0x619a9) [0x4619a9]
2: /lib64/libpthread.so.0 (0x7ffa0cb5a000+0xf320) [0x7ffa0cb69320]
3: /usr/lib64/libpixman-1.so.0 (0x7ffa0d1c8000+0x6ba0) [0x7ffa0d1ceba0]
4: /usr/lib64/libpixman-1.so.0 (0x7ffa0d1c8000+0x35289) [0x7ffa0d1fd289]
5: /usr/lib64/libpixman-1.so.0 (0x7ffa0d1c8000+0x2fa5b) [0x7ffa0d1f7a5b]
6: /usr/lib64/libpixman-1.so.0 (0x7ffa0d1c8000+0x376e2) [0x7ffa0d1ff6e2]
7: /usr/lib64/libpixman-1.so.0 (0x7ffa0d1c8000+0x38a50) [0x7ffa0d200a50]
8: /usr/lib64/libpixman-1.so.0 (0x7ffa0d1c8000+0x2f75a) [0x7ffa0d1f775a]
9: /usr/lib64/libpixman-1.so.0 (pixman_image_composite+0x17c) [0x7ffa0d1f84ec]
10: /usr/lib64/xorg/modules/libfb.so (fbComposite+0x160) [0x7ffa07d95cd0]
11: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7ffa089c7000+0x5fa23) [0x7ffa08a26a23]
12: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7ffa089c7000+0x5e0b3) [0x7ffa08a250b3]
13: /usr/bin/X (0x400000+0xd1b70) [0x4d1b70]
14: /usr/bin/X (0x400000+0x15e2f9) [0x55e2f9]
15: /usr/bin/X (0x400000+0x15e450) [0x55e450]
16: /usr/bin/X (0x400000+0xa1d33) [0x4a1d33]
17: /usr/bin/X (ConfigureWindow+0xa9f) [0x453c3f]
18: /usr/bin/X (0x400000+0x2bbe7) [0x42bbe7]
19: /usr/bin/X (0x400000+0x2c60c) [0x42c60c]
20: /usr/bin/X (0x400000+0x21c9a) [0x421c9a]
21: /lib64/libc.so.6 (__libc_start_main+0xfd) [0x7ffa0b32bb4d]
22: /usr/bin/X (0x400000+0x21849) [0x421849]
Segmentation fault at address 0x2f514b0

*** Bug 531395 has been marked as a duplicate of this bug. ***

I'm unable to reproduce this on a 965GM, so I'm fearful it's a gen3-specific problem.  I'll try to find gen3 machine.

So, it kind of is gen3-specific.  It relies on hitting software fallbacks on the Render path, which is a lot easier to trigger on gen3 because the hardware coordinate limit is only 2048, so two monitors side by side are very likely to exceed that.

This should be fixed (badly, but) in xorg-x11-server 1.7.1-3.fc12.  It'll misrender in some cases, but it won't crash.

The build is running here:

http://koji.fedoraproject.org/koji/buildinfo?buildID=139659

please test, once it's done. I can't provide a new live build for you unfortunately (I can only do x86-64 live builds), but you should be able to test without an installed F12; boot from a recent nightly live image (with no external monitor attached, obviously), update all installed xorg-x11-server packages from the above URL, stop X, attach monitor, restart X. there's a couple of ways to do it - go to a console and do 'telinit 3' then 'telinit 5', or boot at runlevel 3 initially, startx to do the package update, then log out, attach the monitor, and startx again.

feedback would be very valuable...thanks!

-- 
Fedora Bugzappers volunteer triage team
https://fedoraproject.org/wiki/BugZappers

1.7.1-3.fc12 works for me on a box that was reliably reproducing this before.

This bug appears to cause problems with the behaviour of qemu/KVM-based virtual machines (i.e. the Fedora virt stack). Using an X server with this patch applied - either 1.7.1-3 or a 1.7.0-6 scratch build I did which is just 1.7.0-5 with this patch added - causes problems: the first attempt to log in to the desktop from GDM fails and loops back to GDM, the second hangs at just the background and a cursor for a long time before finally showing a dbus error dialog and logging into a somewhat screwed desktop (any app takes an age to load, the bottom panel is MIA). Both I and Jesse Keating confirm this problem.

1.7.0-5 and 1.7.1-2 both work fine (according to Jesse, I didn't test yet) so we're fairly confident it's the patch at fault, that's the only common factor.

The patched servers appear to work OK on bare metal (Peter Hutterer and I both have 1.7.1-3 working apparently fine on bare metal machines; the one I'm testing is a Radeon HD 4770).

We can't accept this patch as-is.

-- 
Fedora Bugzappers volunteer triage team
https://fedoraproject.org/wiki/BugZappers

jesse filed the problem we saw as a separate bug: https://bugzilla.redhat.com/show_bug.cgi?id=532831

-- 
Fedora Bugzappers volunteer triage team
https://fedoraproject.org/wiki/BugZappers

Yes, xorg-x11-server-1.7.1-3.fc12 fixes the problem.

Please also test 1.7.1-5.fc12, if you can.

1.7.1-5.fc12 works for me

1.7.1-5.fc12 works for me too.

MODIFIED based on comments 14 and 15.

actually it should stay in ASSIGNED, MODIFIED implies (at this point in the cycle) that we've tagged the build that ought to fix it but we need to confirm that it does; in this case we know what fixes it but it's not tagged yet...we can close as soon as the tag is done, no need to go through modified then.

tag request is: https://fedorahosted.org/rel-eng/ticket/3070

-- 
Fedora Bugzappers volunteer triage team
https://fedoraproject.org/wiki/BugZappers

*** Bug 522875 has been marked as a duplicate of this bug. ***

1.7.1-5 was tagged, closing this.

-- 
Fedora Bugzappers volunteer triage team
https://fedoraproject.org/wiki/BugZappers