Catalyst 9.8 system hang (trace included)

[RealNC]

This is the first time fglrx hangs my system after quite some while (used to get hangs like this around Catalyst 9.1/9.2). It happened while watching a movie in mplayer (using gl2 back-end), while at the same time I did some alt+tabbing.

GPU: GeCube Radeon HD4870 512MB
Kernel 2.6.30.5
Distro: Gentoo Linux 64-bit
CPU: Core2 Duo E6600
RAM: 6GB DDR2 800
Mainboard: Asus P5E
X.Org: 1.6.3
glibc: 2.10.1

I was able to save the kernel messages. Here they are in case they're useful:


Aug 23 23:29:44 gentoo divide error: 0000 [#1] SMP
Aug 23 23:29:44 gentoo last sysfs file: /sys/devices/pci0000:00/0000:00:1c.4/0000:02:00.0/host6/target6:0:0/6:0:0:0/block/sdb/uevent
Aug 23 23:29:44 gentoo CPU 1
Aug 23 23:29:44 gentoo Modules linked in: oss_audigyls osscore fglrx(P) [last unloaded: microcode]
Aug 23 23:29:44 gentoo Pid: 2936, comm: mplayer Tainted: P 2.6.30-gentoo-r5 #1 P5E
Aug 23 23:29:44 gentoo RIP: 0010:[<ffffffffa008ad84>] [<ffffffffa008ad84>] _ZN15CMMSurface_CORE20setSurfaceAttributesEP16MSF_ SURF_ATTRIB
Sj+0x64/0xe0 [fglrx]
Aug 23 23:29:44 gentoo RSP: 0018:ffff880117853a50 EFLAGS: 00010246
Aug 23 23:29:44 gentoo RAX: 0000000000000100 RBX: 0000000000003308 RCX: 0000000000000000
Aug 23 23:29:44 gentoo RDX: 0000000000000000 RSI: ffff880117853c08 RDI: ffffc20011b7d220
Aug 23 23:29:44 gentoo RBP: ffffc20011b7d220 R08: 0000000000000001 R09: 0000000000000000
Aug 23 23:29:44 gentoo R10: 0000000000000000 R11: 0000000000000001 R12: ffff880117853c08
Aug 23 23:29:44 gentoo R13: ffff880117853b18 R14: ffffc200110bcca0 R15: ffffc20011b9dbc0
Aug 23 23:29:44 gentoo FS: 00007ffe067f4770(0000) GS:ffff88002804d000(0000) knlGS:0000000000000000
Aug 23 23:29:44 gentoo CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Aug 23 23:29:44 gentoo CR2: 0000003003505008 CR3: 000000011ce78000 CR4: 00000000000006e0
Aug 23 23:29:44 gentoo DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
Aug 23 23:29:44 gentoo DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Aug 23 23:29:44 gentoo Process mplayer (pid: 2936, threadinfo ffff880117852000, task ffff88018fcebd30)
Aug 23 23:29:44 gentoo Stack:
Aug 23 23:29:44 gentoo ffffffffa0079617 0000000000000100 ffffc200110bcca0 ffffffffa0108590
Aug 23 23:29:44 gentoo ffff880117853c08 ffffc20011b9e720 ffff880117853b18 ffff880117853c3c
Aug 23 23:29:44 gentoo ffffffffa0079c69 0000000000000001 ffffffffa009d307 0100000000000000
Aug 23 23:29:44 gentoo Call Trace:
Aug 23 23:29:44 gentoo [<ffffffffa0079617>] ? _ZN3MSF11create_surfEP9CMMClientP9CMMDriverPvRA4_K 14CMM_ALLOCATIONP16MSF_SURF_ATTRIBS+0x97/
0x1c0 [fglrx]
Aug 23 23:29:44 gentoo [<ffffffffa0079c69>] ? _ZN3MSF10alloc_surfEP9CMMClientP9CMMDriverP21MEMHE AP_ADDR_RESTRICTjP16MSF_SURF_ATTRIBSP15_C
MM_RETURNCODE+0x529/0x570 [fglrx]
Aug 23 23:29:44 gentoo [<ffffffffa009d307>] ? _ZN8AsicR60025getOffsetAlignmentAndSizeEP16MSF_SUR F_ATTRIBSjj+0x197/0x340 [fglrx]
Aug 23 23:29:44 gentoo [<ffffffffa0074d2a>] ? CMMAllocSurface_WA+0x7fa/0x980 [fglrx]
Aug 23 23:29:44 gentoo [<ffffffffa002f4c2>] ? firegl_trace+0x72/0x1e0 [fglrx]
Aug 23 23:29:44 gentoo [<ffffffffa0085680>] ? _Z8uCWDDEQCmjjPvjS_+0xa80/0x10c0 [fglrx]
Aug 23 23:29:44 gentoo [<ffffffffa00318c8>] ? firegl_cmmqs_CWDDE_32+0x348/0x410 [fglrx]
Aug 23 23:29:44 gentoo [<ffffffffa0030360>] ? firegl_cmmqs_CWDDE32+0x70/0x100 [fglrx]
Aug 23 23:29:44 gentoo [<ffffffffa00302f0>] ? firegl_cmmqs_CWDDE32+0x0/0x100 [fglrx]
Aug 23 23:29:44 gentoo [<ffffffffa000fc1a>] ? firegl_ioctl+0x1ea/0xf40 [fglrx]
Aug 23 23:29:44 gentoo [<ffffffff802ae0b2>] ? vfs_ioctl+0x82/0xb0
Aug 23 23:29:44 gentoo [<ffffffff802ae3c3>] ? do_vfs_ioctl+0x2e3/0x570
Aug 23 23:29:44 gentoo [<ffffffff802ae699>] ? sys_ioctl+0x49/0x80
Aug 23 23:29:44 gentoo [<ffffffff802a1380>] ? sys_read+0x80/0x90
Aug 23 23:29:44 gentoo [<ffffffff8020b36b>] ? system_call_fastpath+0x16/0x1b
Aug 23 23:29:44 gentoo Code: 4c 89 97 e0 00 00 00 44 8b 4e 20 44 89 8f e8 00 00 00 48 8b 56 28 48 89 97 f0 00 00 00 31 d2 8b 46 24 c1 e8
03 89 c1 48 8b 46 28 <48> f7 f1 48 89 87 d0 00 00 00 44 8b 5e 30 44 89 9f f8 00 00 00
Aug 23 23:29:44 gentoo RIP [<ffffffffa008ad84>] _ZN15CMMSurface_CORE20setSurfaceAttributesEP16MSF_ SURF_ATTRIBSj+0x64/0xe0 [fglrx]
Aug 23 23:29:44 gentoo RSP <ffff880117853a50>
Aug 23 23:29:44 gentoo ---[ end trace 5ae134781b9b3181 ]---
Aug 23 23:30:24 gentoo SysRq : Keyboard mode set to system default
Aug 23 23:31:30 gentoo SysRq : Terminate All Tasks


(System became totally locked-up. I had to use SysRq to sync/umount/reboot so I could get the above trace after the reboot.)

[CNCFarraday]

Kernel 2.6.30.5
Distro: Gentoo Linux 64-bit
CPU: Core2 Duo E6600
RAM: 6GB DDR2 800
Mainboard: Asus P5E
X.Org: 1.6.3
glibc: 2.10.1

Erm... what video card?

[mancunian]

I am having similar problems, not sure, though, if related to the first post:

Some programs, primarily firefox, mplayer, and openoffice, first slow down and finally freeze the system. The only way out is a hard-reset. Revertimg to Catalyst 8.5 brings back a stable system.

AMD 64 (Sempron 3200+), 2 GB RAM, ATI Radeon HD 2400 PRO (RV610, I believe), OpenSuse 11.1, KDE 4.3, Kernel 2.6.27.29-0.1-default x86_64

[RealNC]

Erm... what video card?

Oops, there I went and didn't specify the most important spec :p

It's a GeCube Radeon HD4870 512MB. Fixed 1st post.

[RealNC]

Happened again. I was switching mplayer between fullscreen and windowed a few times and then boom. Trace was pretty much the same as the first one I posted.

(Is there actually some place where I can post this stuff and someone from AMD/ATI will notice?)

divide error: 0000 [#1] SMP
last sysfs file: /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors
CPU 1
Modules linked in: oss_audigyls osscore fglrx(P) [last unloaded: microcode]
Pid: 2966, comm: mplayer Tainted: P 2.6.30-gentoo-r5 #1 P5E
RIP: 0010:[<ffffffffa008ad84>] [<ffffffffa008ad84>] _ZN15CMMSurface_CORE20setSurfaceAttributesEP16MSF_ SURF_ATTRIBSj+0x64/0xe0 [fglrx]
RSP: 0018:ffff880191d7da50 EFLAGS: 00010246
RAX: 0000000000000100 RBX: 0000000000003308 RCX: 0000000000000000
RDX: 0000000000000000 RSI: ffff880191d7dc08 RDI: ffffc20011b72420
RBP: ffffc20011b72420 R08: 0000000000000001 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000001 R12: ffff880191d7dc08
R13: ffff880191d7db18 R14: ffffc200110bcca0 R15: ffffc20011b9dbc0
FS: 00007f71533b9770(0000) GS:ffff88002804d000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000006d0c70 CR3: 00000001905b2000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process mplayer (pid: 2966, threadinfo ffff880191d7c000, task ffff8801abe1ac80)
Stack:
ffffffffa0079617 0000000000000100 ffffc200110bcca0 ffffffffa0108590
ffff880191d7dc08 ffffc20011b9e720 ffff880191d7db18 ffff880191d7dc3c
ffffffffa0079c69 0000000000000001 ffffffffa009d307 0100000000000000
Call Trace:
[<ffffffffa0079617>] ? _ZN3MSF11create_surfEP9CMMClientP9CMMDriverPvRA4_K 14CMM_ALLOCATIONP16MSF_SURF_ATTRIBS+0x97/0x1c0 [fglrx]
[<ffffffffa0079c69>] ? _ZN3MSF10alloc_surfEP9CMMClientP9CMMDriverP21MEMHE AP_ADDR_RESTRICTjP16MSF_SURF_ATTRIBSP15_CMM_RETURN CODE+0x529/0x570 [fglrx]
[<ffffffffa009d307>] ? _ZN8AsicR60025getOffsetAlignmentAndSizeEP16MSF_SUR F_ATTRIBSjj+0x197/0x340 [fglrx]
[<ffffffffa0074d2a>] ? CMMAllocSurface_WA+0x7fa/0x980 [fglrx]
[<ffffffffa008aae9>] ? _ZN15CMMSurface_CORE17defineSurfInfo_WAEP18_CMM_SU RF_INFO_WA_+0x289/0x430 [fglrx]
[<ffffffffa002f4c2>] ? firegl_trace+0x72/0x1e0 [fglrx]
[<ffffffffa0085680>] ? _Z8uCWDDEQCmjjPvjS_+0xa80/0x10c0 [fglrx]
[<ffffffffa00318c8>] ? firegl_cmmqs_CWDDE_32+0x348/0x410 [fglrx]
[<ffffffffa0030360>] ? firegl_cmmqs_CWDDE32+0x70/0x100 [fglrx]
[<ffffffffa00302f0>] ? firegl_cmmqs_CWDDE32+0x0/0x100 [fglrx]
[<ffffffffa000fc1a>] ? firegl_ioctl+0x1ea/0xf40 [fglrx]
[<ffffffff802ae0b2>] ? vfs_ioctl+0x82/0xb0
[<ffffffff802ae3c3>] ? do_vfs_ioctl+0x2e3/0x570
[<ffffffff802ae699>] ? sys_ioctl+0x49/0x80
[<ffffffff8020b36b>] ? system_call_fastpath+0x16/0x1b
Code: 4c 89 97 e0 00 00 00 44 8b 4e 20 44 89 8f e8 00 00 00 48 8b 56 28 48 89 97 f0 00 00 00 31 d2 8b 46 24 c1 e8 03 89 c1 48 8b 46 28 <48> f7 f1 48 89 87 d0 00 00 00 44 8b 5e 30 44 89 9f f8 00 00 00
RIP [<ffffffffa008ad84>] _ZN15CMMSurface_CORE20setSurfaceAttributesEP16MSF_ SURF_ATTRIBSj+0x64/0xe0 [fglrx]
RSP <ffff880191d7da50>
---[ end trace 922e2c78c6d37147 ]---

[bridgman]

If there's not already an up to date ticket at ati.cchtml.com that's the best place.

[RealNC]

If there's not already an up to date ticket at ati.cchtml.com that's the best place.

I just went there and Catalyst 9.8 isn't even supported. I can only file bugs for 9.7 and lower?

Edit:
Posted it anyway under "9.7".

[RealNC]

The issue persists with 9.9 and the Ubuntu 9.10 release. I usually need to reboot 3 or 4 times per day due to this bug.

Can someone try to reproduce? Please open some video in SMPlayer, use gl2(yuv) as renderer and then press "F" repeatedly to switch to full screen and back. It hangs here once about 20 switches.

[Kano]

Well U 9.10 does not allow fglrx 9.9 to run - until you use your own kernel. When you install fglrx from the repository you get a 9.10 RC driver. I have issues with opengl output too, especially with real hd content it pauses 1-2s even from time to time. xv would be much faster but this is without vsync. Maybe my card is too slow for opengl output, only a simple ATI 3450. I really hope that xv will be fixed and maybe something similar to vdpau will work soon. Btw, did you try without OSS4 audio driver?

[RealNC]

With "Ubuntu 9.10" I meant the Catalyst 9.10 Ubuntu-only release. I don't use Ubuntu at all.

No, I didn't try without OSS4 yet. Didn't think it has anything to do with it since there's no sign of it in the backtrace, unless you mean the "Modules linked in" line; those modules are a *complete* list of modules loaded on my system. Everything else is compiled-in in the kernel.

[RealNC]

Tried without OSS. Same result.

Weird thing: It seems I'm not able to reproduce this with anything other than H.264 *.mkv videos. *Very* weird. In any case, mplayer bug or not, a driver should not hang no matter what an unprivileged application throws at it.

[Darksurf]

Lol, you don't need ubuntu to get the FGLRX 9.10beta catalyst drivers. I use sabayon linux (a gentoo derivative) and am using the 9.10beta drivers just fine. I've noticed an issue myself with XV (using mplayer based players smplayer and gnome-player) using an HD2600 512vram in my lappy. 3 out of 4 times when I open an mkv file (matroska video) i get no video, only audio. but if I close and open it a couple of times, the video eventually comes up...

Its really odd, I cannot find any logs specifying any issues.. Xorg.0.log, kdm.log, dmesg, mplayer.log, etc. nothing seems out of the ordinary. And it only happens on my lappy using the fglrx drivers. using the radeon driver with X200m no issues, using fglrx in laptop hd2600 and tower MSI HD3870 OC edition strange XV issues. Hope this is fixed by the 9.10 release.

[Ren Höek]

With "Ubuntu 9.10" I meant the Catalyst 9.10 Ubuntu-only release. I don't use Ubuntu at all. [..]

Lol, you don't need ubuntu to get the FGLRX 9.10beta catalyst drivers. [..]

I guess he doesn't use it... :rolleyes:.