[fglrx] fglrx freezes randomly : Soft lockup

[JM Barbier]

Using fresh Hardy on a Toshiba Satellite A210-1BH with fglrx, while on opengl screensaver or compiz, my system freezes randomly. I have to reboot using Ctrl+Alt+SysRq+B. The problem seems to come from fglrx : syslog shows these lines (lspci, xorg.conf and xorg.log attached)

Apr 4 02:04:18 portjm kernel: [ 7046.526632] BUG: soft lockup - CPU#1 stuck for 11s! [Xorg:5363]
Apr 4 02:04:18 portjm kernel: [ 7046.526639]
Apr 4 02:04:18 portjm kernel: [ 7046.526642] Pid: 5363, comm: Xorg Tainted: P (2.6.24-14-generic #1)
Apr 4 02:04:18 portjm kernel: [ 7046.526646] EIP: 0060:[<f8d4eeda>] EFLAGS: 00003246 CPU: 1
Apr 4 02:04:18 portjm kernel: [ 7046.526750] EIP is at _ZN8AsicR60020readRetiredTimestampEv+0x1a/0x70 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.526755] EAX: 00000000 EBX: 001e0747 ECX: f94b2020 EDX: f94b2620
Apr 4 02:04:18 portjm kernel: [ 7046.526758] ESI: 00000000 EDI: f94b2020 EBP: f71b1cf0 ESP: f71b1cf0
Apr 4 02:04:18 portjm kernel: [ 7046.526761] DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
Apr 4 02:04:18 portjm kernel: [ 7046.526764] CR0: 80050033 CR2: a6530000 CR3: 374f5000 CR4: 00000690
Apr 4 02:04:18 portjm kernel: [ 7046.526767] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
Apr 4 02:04:18 portjm kernel: [ 7046.526770] DR6: ffff0ff0 DR7: 00000400
Apr 4 02:04:18 portjm kernel: [ 7046.526787] [<f8d454dd>] _ZN4Asic18isTimeStampExpiredE14_LARGE_INTEGERj+0x4d/0x150 [fglrx
]
Apr 4 02:04:18 portjm kernel: [ 7046.526884] [common_interrupt+0x23/0x30] common_interrupt+0x23/0x30
Apr 4 02:04:18 portjm kernel: [ 7046.526897] [<f8d45de0>] _ZN4Asic22ElapsedTS_PollingUntil19ConditionSuccessfulEv+0x30/0x8
0 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.527001] [<f8d456e8>] _ZN4Asic9WaitUntil15WaitForCompleteEv+0x38/0xf0 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.527105] [<f8d44776>] _ZN4Asic19PM4ElapsedTimeStampERK23PM4_TS_INTERRUPT_PARAMSj14_LAR
GE_INTEGER+0x166/0x200 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.527229] [<f8d274b6>] QSSubmitList+0x146/0x150 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.527321] [unix_stream_recvmsg+0x25d/0x550] unix_stream_recvmsg+0x25d/0x550
Apr 4 02:04:18 portjm kernel: [ 7046.527341] [<f8d39030>] _Z19uQSTimeStampRetiredjjj14_LARGE_INTEGER+0xf0/0x100 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.527447] [<f8d3642f>] _Z8uCWDDEQCjjjPvjS_+0x2cf/0x11a0 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.527551] [<f8d26df4>] CMMQS_uCWDDEQC+0x34/0x40 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.527647] [<f8cede48>] firegl_cmmqs_CWDDE_32+0x238/0x340 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.527735] [pid_getattr+0x26/0x70] pid_getattr+0x26/0x70
Apr 4 02:04:18 portjm kernel: [ 7046.527742] [<f8cece06>] firegl_cmmqs_CWDDE32+0x76/0x110 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.527828] [<f8ce1746>] firegl_lock+0x96/0x220 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.527911] [<f8cecd90>] firegl_cmmqs_CWDDE32+0x0/0x110 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.527989] [<f8cdf4be>] firegl_ioctl+0x19e/0x220 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.528065] [pid_getattr+0x26/0x70] pid_getattr+0x26/0x70
Apr 4 02:04:18 portjm kernel: [ 7046.528073] [__rcu_process_callbacks+0x69/0x1d0] __rcu_process_callbacks+0x69/0x1d0
Apr 4 02:04:18 portjm kernel: [ 7046.528082] [__rcu_process_callbacks+0x69/0x1d0] __rcu_process_callbacks+0x69/0x1d0
Apr 4 02:04:18 portjm kernel: [ 7046.528092] [pid_getattr+0x26/0x70] pid_getattr+0x26/0x70
Apr 4 02:04:18 portjm kernel: [ 7046.528097] [<f8cd485c>] ip_firegl_ioctl+0x1c/0x30 [fglrx]
Apr 4 02:04:18 portjm kernel: [ 7046.528171] [pid_getattr+0x26/0x70] pid_getattr+0x26/0x70
Apr 4 02:04:18 portjm kernel: [ 7046.528180] [do_ioctl+0x78/0x90] do_ioctl+0x78/0x90
Apr 4 02:04:18 portjm kernel: [ 7046.528190] [vfs_ioctl+0x22e/0x2b0] vfs_ioctl+0x22e/0x2b0
Apr 4 02:04:18 portjm kernel: [ 7046.528201] [sys_ioctl+0x56/0x70] sys_ioctl+0x56/0x70
Apr 4 02:04:18 portjm kernel: [ 7046.528209] [sysenter_past_esp+0x6b/0xa9] sysenter_past_esp+0x6b/0xa9
Apr 4 02:04:18 portjm kernel: [ 7046.528215] [pid_getattr+0x26/0x70] pid_getattr+0x26/0x70
Apr 4 02:04:18 portjm kernel: [ 7046.528236] =======================

thank you

JMB

[JM Barbier (jm-barbier)]

This bug is also signaled in Ati Unofficial Bugzilla : #1052

http://ati.cchtml.com/show_bug.cgi?id=1052

with same symptoms...

[Dominic Evans (oldmanuk)]

Seeing the exact same problem using ATI MOBILITY FireGL V5200 on a Thinkpad T60p. Frustrating random lockups.

[JM Barbier (jm-barbier)]

New restricted modules (2.6.24-17-generic) does not change anything to this bug...
I also tried to force agp 4x in xorg.conf : Option "AGPv3Mask" "0x00000002". This seems to give a little more stability, but the same freezes are still here :

May 5 00:29:28 portjm kernel: [ 448.803457] BUG: soft lockup - CPU#1 stuck for 11s! [Xorg:7195]
May 5 00:29:28 portjm kernel: [ 448.803461]
May 5 00:29:28 portjm kernel: [ 448.803463] Pid: 7195, comm: Xorg Tainted: P (2.6.24-17-generic #1)
May 5 00:29:28 portjm kernel: [ 448.803468] EIP: 0060:[<f8dffe19>] EFLAGS: 00203202 CPU: 1
May 5 00:29:28 portjm kernel: [ 448.803556] EIP is at _ZN4Asic22ElapsedTS_PollingUntil19ConditionSuccessfulEv+
0x69/0x80 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.803560] EAX: 00000000 EBX: f6d45d88 ECX: 00000001 EDX: 00000000
May 5 00:29:28 portjm kernel: [ 448.803563] ESI: 00000000 EDI: f6d45d88 EBP: f6d45d30 ESP: f6d45d18
May 5 00:29:28 portjm kernel: [ 448.803567] DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
May 5 00:29:28 portjm kernel: [ 448.803570] CR0: 80050033 CR2: a61f0000 CR3: 3715c000 CR4: 00000690
May 5 00:29:28 portjm kernel: [ 448.803574] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
May 5 00:29:28 portjm kernel: [ 448.803577] DR6: ffff0ff0 DR7: 00000400
May 5 00:29:28 portjm kernel: [ 448.803595] [<f8dff6e8>] _ZN4Asic9WaitUntil15WaitForCompleteEv+0x38/0xf0 [fgl
rx]
May 5 00:29:28 portjm kernel: [ 448.803697] [<f8dfe776>] _ZN4Asic19PM4ElapsedTimeStampERK23PM4_TS_INTERRUPT_P
ARAMSj14_LARGE_INTEGER+0x166/0x200 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.803820] [<f8de14b6>] QSSubmitList+0x146/0x150 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.803926] [<f8df3030>] _Z19uQSTimeStampRetiredjjj14_LARGE_INTEGER+0xf0/0x10
0 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.804029] [<f8df042f>] _Z8uCWDDEQCjjjPvjS_+0x2cf/0x11a0 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.804132] [<f8de0df4>] CMMQS_uCWDDEQC+0x34/0x40 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.804225] [<f8da7e48>] firegl_cmmqs_CWDDE_32+0x238/0x340 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.804311] [sys_quotactl+0x5a6/0x680] sys_quotactl+0x5a6/0x680
May 5 00:29:28 portjm kernel: [ 448.804317] [<f8da6e06>] firegl_cmmqs_CWDDE32+0x76/0x110 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.804399] [<f8d9b746>] firegl_lock+0x96/0x220 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.804481] [<f8da6d90>] firegl_cmmqs_CWDDE32+0x0/0x110 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.804560] [<f8d994be>] firegl_ioctl+0x19e/0x220 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.804636] [sys_quotactl+0x5a6/0x680] sys_quotactl+0x5a6/0x680
May 5 00:29:28 portjm kernel: [ 448.804645] [update_curr+0x9f/0x150] update_curr+0x9f/0x150
May 5 00:29:28 portjm kernel: [ 448.804657] [sys_quotactl+0x5a6/0x680] sys_quotactl+0x5a6/0x680
May 5 00:29:28 portjm kernel: [ 448.804662] [<f8d8e85c>] ip_firegl_ioctl+0x1c/0x30 [fglrx]
May 5 00:29:28 portjm kernel: [ 448.804735] [sys_quotactl+0x5a6/0x680] sys_quotactl+0x5a6/0x680
May 5 00:29:28 portjm kernel: [ 448.804745] [do_ioctl+0x78/0x90] do_ioctl+0x78/0x90
May 5 00:29:28 portjm kernel: [ 448.804754] [vfs_ioctl+0x22e/0x...

[JM Barbier (jm-barbier)]

Problem fixed using last fglrx from ATI...

[vxy (vyasevich)]

Exactly the same problem on ATI FirgGL R520GL (V7200).

The problem is triggered by opening an xterm window in a plain old Gnome session.

I've tried a few different version of flgrx dirvers non of them worked. The driver downloaded
from the ATI web site doesn't lock-up, but the screen goes an white/gray and nothing
else is possible. Killing the X-server at this point with Atl-Ctrl-Backspaced flashed the
background and the then GDM comes up.

Right now, using radeonhd to get around this issue, but there are some artifacts on the
display that are driving me nuts.

All in all, the system is completely unusable under Gnome.

[Alberto Milone (albertomilone)]

vxy: can you enable the hardy-proposed repositories and install the latest driver through envyng?

[vxy (vyasevich)]

Alberto:

Still didn't work. Launched the Xterm and soft-locked-uped.

kernel: linux-image-2.6.24-19-generic 2.6.24-19.34 (2.6.24-19-generic #1 SMP Wed Jun 18 14:43:41 UTC 2008 i686 GNU/Linux)

ati module: module loaded - fglrx 8.47.3 [Mar 29 2008] on minor 0

# dpkg -l | grep fglrx
ii fglrx-amdcccle-envy 2.6.24.501-501.30 Dummy package for easy transition
ii fglrx-control-envy 1:8-4+2.6.24.501-501.30 Control panel for the ATI graphics accelerat
ii fglrx-kernel-source-envy 1:8-4+2.6.24.501-501.30 ATI binary kernel module source
ii xorg-driver-fglrx-dev-envy 1:7.1.0-8-4+2.6.24.501-501.30 Video driver for ATI graphics accelerators (
ii xorg-driver-fglrx-envy 1:7.1.0-8-4+2.6.24.501-501.30 Video driver for ATI graphics accelerators

Soft Lockup:Jul 14 15:28:36 galen kernel: [ 46.114096] fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies, Starnberg, GERMANY' taints kernel.
Jul 14 15:28:36 galen kernel: [ 46.134345] [fglrx] Maximum main memory to use for locked dma buffers: 2888 MBytes.
Jul 14 15:28:36 galen kernel: [ 46.134374] [fglrx] ASYNCIO init succeed!
Jul 14 15:28:36 galen kernel: [ 46.136902] [fglrx] PAT is enabled successfully!
Jul 14 15:28:36 galen kernel: [ 46.136923] [fglrx] module loaded - fglrx 8.47.3 [Mar 29 2008] on minor 0
Jul 14 15:28:36 galen kernel: [ 46.560074] ACPI: PCI Interrupt 0000:00:1b.0[A] -> GSI 21 (level, low) -> IRQ 21
Jul 14 15:28:36 galen kernel: [ 46.560094] PCI: Setting latency timer of device 0000:00:1b.0 to 64
Jul 14 15:28:36 galen kernel: [ 46.732336] lp: driver loaded but no devices found
Jul 14 15:28:36 galen kernel: [ 46.849793] Adding 7815612k swap on /dev/sda3. Priority:-1 extents:1 across:7815612k
Jul 14 15:28:36 galen kernel: [ 46.853821] Adding 7815612k swap on /dev/sdb3. Priority:-2 extents:1 across:7815612k
Jul 14 15:28:36 galen kernel: [ 47.134383] EXT3 FS on md1, internal journal
Jul 14 15:28:36 galen kernel: [ 47.239662] device-mapper: uevent: version 1.0.3
Jul 14 15:28:36 galen kernel: [ 47.239695] device-mapper: ioctl: 4.12.0-ioctl (2007-10-02) initialised: <email address hidden>
Jul 14 15:28:36 galen kernel: [ 48.273598] NET: Registered protocol family 17
Jul 14 15:28:36 galen kernel: [ 48.639953] kjournald starting. Commit interval 5 seconds
Jul 14 15:28:36 galen kernel: [ 48.676250] EXT3 FS on md0, internal journal
Jul 14 15:28:36 galen kernel: [ 48.676254] EXT3-fs: mounted filesystem with ordered data mode.
Jul 14 15:28:36 galen kernel: [ 48.708817] kjournald starting. Commit interval 5 seconds
Jul 14 15:28:36 galen kernel: [ 48.709021] EXT3 FS on md2, internal journal
Jul 14 15:28:36 galen kernel: [ 48.709028] EXT3-fs: mounted filesystem with ordered data mode.
Jul 14 15:28:36 galen kernel: [ 49.304894] tg3: eth0: Link is up at 1000 Mbps, full duplex.
Jul 14 15:28:36 galen kernel: [ 49.304898] tg3: eth0: Flow control is on for TX and on for RX.
Jul ...

[Stefano Maioli (smaioli)]

Same here, with a Radeon 9500pro and 8.9 drivers from ati. The problem has always been present since hardy release. My dmesg is a little different though. The open-source driver works fine.

It is known to affect _many_ ATI agp cards. It seems related to big texture agp transfers, maybe because of some chipset/card incompatibilities. However, I tried tweaking agp settings in every possible way with no success.

Maybe Canonical can work with ATI on a solution, afaik they have very good communication channels.

Also, bug 254763 might be a dupe of this one.

[Bryce Harrington (bryce)]

Good day JM,

I've just uploaded a new 8.543 version of -fglrx to Intrepid which now should work with xserver 1.5. Would you mind testing this new version and reporting back whether this issue is still present? If it is, it would be helpful if you could (re-)post your Xorg.0.log from running with this version. Thanks ahead of time.

If you don't have Intrepid installed on your system, you can test this by booting an Intrepid LiveCD (available from cdimage.ubuntu.com), using either the -vesa or -ati driver, then update to the latest version of Ubuntu, install fglrx, and then logout and back in. Your /var/log/Xorg.0.log will confirm whether you've loaded FGLRX successfully.

If you find any new issues, please report them as separate bugs. You can use the tool `ubuntu-bug fglrx-installer` which will gather the necessary files and create the launchpad report for you to fill in more easily.

[Bryce Harrington (bryce)]

We're closing this bug since it is has been some time with no response from the original reporter. However, if the issue still exists please feel free to reopen with the requested information. Also, if you could, please test against the latest development version of Ubuntu, since this confirms the bug is one we may be able to pass upstream for help.

[Stefano Maioli (smaioli)]

Bryce:
There has been no response from me because I've been unable to test the new driver as you requested: 8.543 does not work on the r300 (Bug #284408).

But I (and others I think) am still _very_ interested in this long-standing bug. I will test and report back as soon as a compatible driver comes out.

Thanks

[Bryce Harrington (bryce)]

We're closing this bug since it is has been some time with no response from the original reporter. However, if the issue still exists please feel free to reopen with the requested information. Also, if you could, please test against the latest development version of Ubuntu, since this confirms the bug is one we may be able to pass upstream for help.

[Bryce Harrington (bryce)]

Sorry, a script misfired and closed this bug incorrectly. I've reopened it.

[Bryce Harrington (bryce)]

We're closing this bug since it is has been some time with no response from the original reporter. However, if the issue still exists please feel free to reopen with the requested information. Also, if you could, please test against the latest development version of Ubuntu, since this confirms the bug is one we may be able to pass upstream for help.