Hang, then recovery with garbled screen, reboot on X restart

[abs0]

OS: NetBSD-6_BETA amd64 from netbsd-daily 2012-03-21, plus ftp://ftp.netbsd.org/pub/NetBSD/misc/gsutre/drmgem-NetBSD-6.99.3-201203091940Z-amd64/

System: Thinkpad T500 with "Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller (rev 07)"

After running fine for several hours system appeared to hang while page loading on Firefox.

Appeared non responsive with Ctrl+Alt+F1.

After around a minute display switched to black screen with white cursor at top left, system then was able to switch back to X session but with garbled graphics, and then switch back to constty with correct display.

Restarting xdm rebooted machine

[abs0]

During the garbled screen an xterm at 0 0 appears to render scrolled text offset at around +770+242.

Moving the xterm right and down produces a matching offset in rendering.

Moving a firefox window around exhibits the same offset rendering, but any updates to content (typing into thie form for example) render correctly

[abs0]

relevant section from /var/log/messages below, including error messages from hang and panic string

Apr 4 15:09:10 forsaken /netbsd: sd0: detached
Apr 4 15:09:10 forsaken /netbsd: scsibus0: detached
Apr 4 15:09:10 forsaken /netbsd: umass0: detached
Apr 4 15:09:10 forsaken /netbsd: umass0: at uhub7 port 2 (addr 2) disconnected

v---- system appears hung
Apr 4 17:43:48 forsaken /netbsd: sg_dmamap_load_raw: unable to allocate extent
subregion: 35
Apr 4 17:44:40 forsaken syslogd[465]: last message repeated 2 times
Apr 4 17:44:40 forsaken /netbsd: e to allocate extent subregion: 35
Apr 4 17:44:40 forsaken /netbsd: sg_dmamap_load_raw: unable to allocate extent
subregion: 35
Apr 4 17:44:53 forsaken syslogd[465]: last message repeated 544 times
Apr 4 17:44:53 forsaken /netbsd: usl_detachtimeout
Apr 4 17:44:53 forsaken /netbsd: usl_detachack: not detaching
Apr 4 17:45:10 forsaken console-kit-daemon[926]: WARNING: Error waiting for nat
ive console 6 activation: Device not configured
Apr 4 17:45:10 forsaken console-kit-daemon[926]: WARNING: Error waiting for nat
ive console 7 activation: Device not configured
Apr 4 17:45:13 forsaken console-kit-daemon[926]: WARNING: Error waiting for nat
ive console 6 activation: Device not configured
Apr 4 17:45:13 forsaken console-kit-daemon[926]: WARNING: Error waiting for nat
ive console 7 activation: Device not configured
Apr 4 17:46:38 forsaken console-kit-daemon[926]: WARNING: Error waiting for nat
ive console 6 activation: Device not configured
Apr 4 17:46:38 forsaken console-kit-daemon[926]: WARNING: Error waiting for nat
ive console 7 activation: Device not configured
Apr 4 17:46:39 forsaken console-kit-daemon[926]: WARNING: Error waiting for nat
ive console 6 activation: Device not configured
Apr 4 17:46:39 forsaken console-kit-daemon[926]: WARNING: Error waiting for nat
ive console 7 activation: Device not configured
Apr 4 17:46:43 forsaken /netbsd: inteldrm0: gpu hung!
Apr 4 17:46:46 forsaken /netbsd: error: [drm:pid664:i915_gem_cleanup_ringbuffer
] ERROR failed to quiesce ring buffer whilst cleaning up: 16
Apr 4 17:46:46 forsaken console-kit-daemon[926]: WARNING: Error waiting for nat
ive console 7 activation: Device not configured
Apr 4 17:46:46 forsaken console-kit-daemon[926]: WARNING: Error waiting for nat
ive console 6 activation: Device not configured

v---- /etc/rc.d/xdm restart ran (and system rebooted)
Apr 4 17:50:23 forsaken syslogd[398]: restart
Apr 4 17:50:23 forsaken /netbsd: error: [drm:pid2539:i915_gem_entervt_ioctl] E
RROR Reenabling wedged hardware, good luck
Apr 4 17:50:23 forsaken /netbsd: panic: kernel diagnostic assertion "TAILQ_EMPT
Y(&dev_priv->mm.request_list)" failed: file "../../../../dev/pci/drm/i915_drv.c"
, line 4581
Apr 4 17:50:23 forsaken /netbsd: cpu0: Begin traceback...
Apr 4 17:50:23 forsaken /netbsd: kern_assert() at netbsd:kern_assert+0x48

[abs0]

Upon reboot system has re-entered offset rendering state, but without any apparent hang during X startup and login, presumably some state is persistent across warm boot.

[gsutre]

Thanks for your detailed report. The first error message (unable to allocate extent subregion: 35) suggests that this might be related to issue #9. I've already looked at it for some time, but couldn't find the cause for this extent allocation error.

I've also experienced the offset rendering state, and, in my experiments, this state is transient: the X server hangs shortly afterwards. But the persistence (across warm boot) of the offset rendering state is something I've never experienced, and I don't see where this could come from.

[abs0]

Issue just reoccurred - hang as clicking on link in firefox.

Waited a minute or so, hit ALT+F1, then after a moment it switched to console but only showing cursor, switching back to X had offset issue.

Manually rebooted at that point.

[abs0]

Also has happened twice overnight - while screen was running xmatrix screensaver from kde4.

Similar symptoms - apparent hang but will switch to blank with cursor VT after some time, can switch back but with xterm rendering offset & now corrupted glyph.

will try restarting X to see if it reboots :)

[abs0]

Hmm - that time a /etc/rc.d/xdm restart seems to have been happy