Bug 17572

What likely is important here is that libdrm and the driver were built against the last stable x.org release, 7.3. I have no idea if this is supposed to work at all, but if not, then the driver should do some version checking. As it compiled ok along with libdrm (whose purpose eludes me, I don't need nor want any DRM on my system) I expect it to also run ok without crashing.

This seems dup with bug#17540 (the xf86-video-intel + libdrm tip is broken). Could you try the patch there to see if that helps? Or roll back xf86-video-intel + libdrm to about 1 week ago?

Created attachment 18880 [details]
Xorg.0.log after patch

An updated log file after applying the patch.

The patch applied cleanly to my git version as of yesterday (Sunday). The change is that now the driver does not crash, but it also does nothing useful at all. The screen turns black, the mouse pointer is shown and can be moved, but nothing is ever started (KDM should pop up). I see that /usr/bin/X sits in an endless loop doing:

ioctl(10, 0x40046445, 0x7fff1fa1d620)   = -1 EINTR (Interrupted system call)
--- SIGALRM (Alarm clock) @ 0 (0) ---
rt_sigreturn(0xe)                       = -1 EINTR (Interrupted system call)
ioctl(10, 0x40046445, 0x7fff1fa1d620)   = -1 EINTR (Interrupted system call)
--- SIGALRM (Alarm clock) @ 0 (0) ---
rt_sigreturn(0xe)                       = -1 EINTR (Interrupted system call)

etc. The X server is frozen solid and can not be killed with Ctrl-Alt-Backspace.

Added a new attachment which is the Xorg.0.log file after the patch.

I run a 1920x1200 display, perhaps that can have something to do with it? My monitor says that the solid black screen is using 1920x1200, so it at least managed to initialize something "ok".

Created attachment 18884 [details]
Xorg.0.log after AccelMethod=XAA and DRI=False

From extra kicks I tried disabling DRI and using XAA as the acceleration method (as was suggested to me on IRC the other day) in case that could work. No. Crashes again. The log is at http://bugs.freedesktop.org/attachment.cgi?id=18884

Backtrace:
0: /usr/bin/X(xf86SigHandler+0x65) [0x481005]
1: /lib/libc.so.6 [0x7f0df0ddd120]
2: /usr/lib/xorg/modules/drivers//intel_drv.so [0x7f0def7e6ad3]
3: /usr/local/drm-git/lib/libdrm_intel.so.1 [0x7f0def5c51b4]
4: /usr/lib/xorg/modules/drivers//intel_drv.so(intel_batch_flush+0x68) [0x7f0def7e6948]
5: /usr/lib/xorg/modules/drivers//intel_drv.so(I830Sync+0x5a) [0x7f0def7e624a]
6: /usr/lib/xorg/modules/drivers//intel_drv.so(i830WaitSync+0x47) [0x7f0def7f33b7]
7: /usr/lib/xorg/modules/drivers//intel_drv.so [0x7f0def7eb8fc]
8: /usr/bin/X(xf86CrtcSetMode+0x14c) [0x4a888c]
9: /usr/bin/X(xf86SetDesiredModes+0x1c5) [0x4a8fd5]
10: /usr/lib/xorg/modules/drivers//intel_drv.so [0x7f0def7f3986]
11: /usr/lib/xorg/modules/drivers//intel_drv.so [0x7f0def7f52da]
12: /usr/bin/X(AddScreen+0x1c9) [0x432e19]
13: /usr/bin/X(InitOutput+0x241) [0x469b31]
14: /usr/bin/X(main+0x286) [0x4335b6]
15: /lib/libc.so.6(__libc_start_main+0xe6) [0x7f0df0dc8466]
16: /usr/bin/X [0x432b69]
Saw signal 11.  Server aborting.

(In reply to comment #4)
> The patch applied cleanly to my git version as of yesterday (Sunday). The
> change is that now the driver does not crash, but it also does nothing useful
> at all. The screen turns black, the mouse pointer is shown and can be moved,
> but nothing is ever started (KDM should pop up). I see that /usr/bin/X sits in
> an endless loop doing:

The patch does help so you hit next bug.
So I'm marking this original reported issue dup with bug#17540.

I expect 17540 will be resolved soon. After that, if you see any other issues, please file a new bug. Thanks.

Also I'd mention there're G45 specific issues which may impact you even after 17540 resolved: bug#17235,bug#17460.


*** This bug has been marked as a duplicate of bug 17540 ***