X server I830WaitLpRing crash with PyMOL and direct rendering on Intel 965Q

[Lieven Buts]

I have a completely reproducible X server crash running PyMOL on a Dell
Optiplex 745 desktop system with integrated Intel 965Q graphics controller.

Steps to reproduce:
  1) Install Ubuntu Gutsy Gibbon i386 Desktop on Dell Optiplex 745
      with single 1280x1024 LCD monitor.
  2) Install PyMOL (sudo apt-get install pymol)
  3) Reboot cleanly and log in (no suspending, no VC switching, ...)
  4) Start PyMOL (Applications|Education|...)
  5) Start "Representations" demo (Wizard|Demo|Representations)
  6) Wait a few seconds
  7) Open any menu from the menu bar
  8) X server freezes and goes into endless restart cycle

Software: fully up-to-date i386 gutsy, X.org xserver 1.3.0, xf86-video-intel 2.1.0.

Notes: problem also occurs with xf86-video-i810 driver and on x86_64 OpenSuSE 10.2.
Disabling DRI (Option "NoAccel" "true") prevents crash, with obvious loss of rendering speed.
Most likely due to common Intel graphics controller DRI code, and not specific to old i810
or new intel driver. Not a 32/64-bit issue. Possible interaction with kernel framebuffer
driver for text consoles.

Possibly related bugs: launchpad #56501, #117892, #28326; freedesktop.org bugzilla #9415.

Relevant log files at:
http://ultr23.vub.ac.be/~lieven/965/Xorg.0.log-i810-02072007
and
http://ultr23.vub.ac.be/~lieven/965/Xorg.0.log-intel-02072007

[Lieven Buts (lievbuts)]

More information:

uname -a:
Linux iche-org79 2.6.22-7-generic #1 SMP Mon Jun 25 17:33:14 GMT 2007 i686 GNU/Linux

lspci -n:
00:00.0 0600: 8086:2990 (rev 02)
00:02.0 0300: 8086:2992 (rev 02)
00:02.1 0380: 8086:2993 (rev 02)
00:1a.0 0c03: 8086:2834 (rev 02)
00:1a.1 0c03: 8086:2835 (rev 02)
00:1a.7 0c03: 8086:283a (rev 02)
00:1b.0 0403: 8086:284b (rev 02)
00:1c.0 0604: 8086:283f (rev 02)
00:1c.4 0604: 8086:2847 (rev 02)
00:1d.0 0c03: 8086:2830 (rev 02)
00:1d.1 0c03: 8086:2831 (rev 02)
00:1d.2 0c03: 8086:2832 (rev 02)
00:1d.7 0c03: 8086:2836 (rev 02)
00:1e.0 0604: 8086:244e (rev f2)
00:1f.0 0601: 8086:2810 (rev 02)
00:1f.2 0101: 8086:2820 (rev 02)
00:1f.3 0c05: 8086:283e (rev 02)
00:1f.5 0101: 8086:2825 (rev 02)
02:00.0 0200: 14e4:167a (rev 02)

xorg.conf: attached

[Lieven Buts (lievbuts)]

Changing colour depth or EXA/XAA setting has no effect on the symptoms.

[Michael Banck (mbanck)]

As a data point, pymol segfaults on me on Debian testing when DRI is active, while it works fine when rendering is indirect.

[Bryce Harrington (bryce)]

Here's the backtrace portions of the logs:

(II) I810(0): [drm] removed 1 reserved context for kernel
(II) I810(0): [drm] unmapping 8192 bytes of SAREA 0xf8d30000 at 0xb7a94000
Error in I830WaitLpRing(), now is -2033746721, start is -2033748722
pgetbl_ctl: 0x3ff80001 pgetbl_err: 0x0
ipeir: 0 iphdr: 0
LP ring tail: 38 head: 0 len: 1f001 start 0
eir: 0 esr: 1 emr: ffdf
instdone: 0 instpm: 0
memmode: 0 instps: 0
hwstam: dfff ier: 0 imr: dfff iir: 0
space: 131008 wanted 131064

Fatal server error:
lockup

Backtrace:
0: /usr/bin/X(xf86SigHandler+0x81) [0x80c7f11]
1: [0xffffe420]
2: /usr/bin/X(AbortDDX+0xad) [0x80a6c7d]
3: /usr/bin/X(AbortServer+0x23) [0x81c3293]
4: /usr/bin/X(FatalError+0x66) [0x81c3796]
5: /usr/lib/xorg/modules/drivers//i810_drv.so(I830WaitLpRing+0x116) [0xb7c07b46]
6: /usr/lib/xorg/modules/drivers//i810_drv.so(I830Sync+0x104) [0xb7c08334]
7: /usr/lib/xorg/modules/drivers//i810_drv.so [0xb7c0d679]
8: /usr/lib/xorg/modules/drivers//i810_drv.so [0xb7c0e706]
9: /usr/lib/xorg/modules/drivers//i810_drv.so [0xb7c14683]
10: /usr/lib/xorg/modules//libshadow.so [0xb7a9741e]
11: /usr/bin/X [0x8168c37]
12: /usr/bin/X [0x80cd8c9]
13: /usr/bin/X [0x8142658]
14: /usr/bin/X [0x81664c5]
15: /usr/bin/X [0x8105c75]
16: /usr/bin/X(main+0x50c) [0x80767dc]
17: /lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xdc) [0xb7deeebc]
18: /usr/bin/X(FontFileCompleteXLFD+0x1e5) [0x8075a81]

FatalError re-entered, aborting
Caught signal 11. Server aborting

---

Error in I830WaitLpRing(), now is -2032605365, start is -2032607366
pgetbl_ctl: 0x3ff80001 pgetbl_err: 0x0
ipeir: 0 iphdr: 0
LP ring tail: d88 head: 0 len: 1f001 start 0
eir: 0 esr: 1 emr: ffdf
instdone: 0 instpm: 0
memmode: 0 instps: 0
hwstam: dfff ier: 0 imr: dfff iir: 0
space: 127600 wanted 131064

Fatal server error:
lockup

Backtrace:
0: /usr/bin/X(xf86SigHandler+0x81) [0x80c7f11]
1: [0xffffe420]
2: /usr/bin/X(AbortDDX+0xad) [0x80a6c7d]
3: /usr/bin/X(AbortServer+0x23) [0x81c3293]
4: /usr/bin/X(FatalError+0x66) [0x81c3796]
5: /usr/lib/xorg/modules/drivers//intel_drv.so(I830WaitLpRing+0x123) [0xb7bb5ed3]
6: /usr/lib/xorg/modules/drivers//intel_drv.so(I830Sync+0xfd) [0xb7bb60fd]
7: /usr/lib/xorg/modules/drivers//intel_drv.so [0xb7bbceba]
8: /usr/lib/xorg/modules/drivers//intel_drv.so [0xb7bbcfc0]
9: /usr/lib/xorg/modules/drivers//intel_drv.so [0xb7bbd3c3]
10: /usr/bin/X [0x8168c37]
11: /usr/bin/X [0x80cd8c9]
12: /usr/bin/X [0x8142658]
13: /usr/bin/X [0x81664c5]
14: /usr/bin/X [0x8105c75]
15: /usr/bin/X(main+0x50c) [0x80767dc]
16: /lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xdc) [0xb7d9aebc]
17: /usr/bin/X(FontFileCompleteXLFD+0x1e5) [0x8075a81]

FatalError re-entered, aborting
Caught signal 11. Server aborting

[Greg Edwards (gedwards)]

I see the same thing on my Dell 745, but it happens when one of the screensavers kicks in (haven't
been around to see which one).

[Russell Sears (sears)]

I'm running ubuntu gutsy with an intel 965gm (Lenovo x61 tablet), but with the newer intel video drivers from hardy (2:2.2.0-1ubuntu1). I haven't tested it heavily, but pymol doesn't seem to crash X on my system. Instead, if you move the window, pymol's rendering window (the 3d one) is no longer updated. The menus are still responsive.

[Bryce Harrington (bryce)]

Thanks for confirming this issue is now fixed. It sounds like it was the same as bug 119977.

The other issue, where 3d windows don't get updated properly when dragged, that sounds like bug 116793 so I'm going to assume that's covered. Please feel free to add additional information to that bug for your case, or open a new bug if you suspect it's a separate issue.

[Anders Norgaard (anders-norgaard)]

Hi Bryce,

How is it fixed? Will there be an update to the driver? (or pymol?). And are you sure this issue is the same as "bug 119977"? the descriptions don't look similar at all.

Thanks
Anders

PS Here is an extract from my log after a pymol induced crash

Fatal server error:
lockup

(II) AIGLX: Suspending AIGLX clients for VT switch
Error in I830WaitLpRing(), timeout for 2 seconds
pgetbl_ctl: 0x7ff80001 pgetbl_err: 0x0
ipeir: 0 iphdr: bf149d99
LP ring tail: a48 head: 0 len: 1f001 start 0
Err ID (eir): 0 Err Status (esr): 1 Err Mask (emr): ffffffdf
instdone: ffe5fafd instdone_1: fffff
instpm: 0
memmode: 0 instps: 8001e022
HW Status mask (hwstam): fff8dffe
IRQ enable (ier): 2 imr: fff80000 iir: 20
acthd: 3e93f10 dma_fadd_p: 3e94000
ecoskpd: 307 excc: 0
cache_mode: 6800/180
mi_arb_state: 44
IA_VERTICES_COUNT_QW 0/0
IA_PRIMITIVES_COUNT_QW 0/0
VS_INVOCATION_COUNT_QW 0/0
GS_INVOCATION_COUNT_QW 0/0
GS_PRIMITIVES_COUNT_QW 0/0
CL_INVOCATION_COUNT_QW 0/0
CL_PRIMITIVES_COUNT_QW 0/0
PS_INVOCATION_COUNT_QW 0/0
PS_DEPTH_COUNT_QW 0/0
WIZ_CTL 0
TS_CTL 0 TS_DEBUG_DATA 223725fe
TD_CTL 0 / 0
space: 128432 wanted 131064

FatalError re-entered, aborting
lockup