Bug#630532: xorg hangs in screensaver (radeon chipset)

[Julien Cristau]

On Tue, Jun 14, 2011 at 17:04:28 -0400, Daniel Dickinson wrote:

> Xorg gets the kernel hung_task_timeout message while in screensaver at random times.  This usually happens when I've left the computer to a long job, and means that a forced shutdown is necessary (SysRq works) and causes (so far recoverable) filesystem issues.
> 
Is this reproducible with a newer kernel?  With a newer
xserver-xorg-video-radeon?  With a newer libgl1-mesa-dri?  Do you know
what screensaver triggers it exactly?

> Jun 14 16:41:30 daniloth kernel: [312480.864110] INFO: task Xorg:3828 blocked for more than 120 seconds.
> Jun 14 16:41:30 daniloth kernel: [312480.864119] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables
>  this message.
> Jun 14 16:41:30 daniloth kernel: [312480.864126] Xorg            D ffff8801185f21c0     0  3828   3826 0x004
> 00004
> Jun 14 16:41:30 daniloth kernel: [312480.864139]  ffff8801185f21c0 0000000000000086 0000000000000296 ffff880
> 009494380
> Jun 14 16:41:30 daniloth kernel: [312480.864151]  0000000000013700 ffff88011997bfd8 ffff88011997bfd8 0000000
> 000013700
> Jun 14 16:41:30 daniloth kernel: [312480.864162]  ffff8801185f21c0 ffff88011997a010 ffffffff00000001 ffff880
> 0bb9cb448
> Jun 14 16:41:30 daniloth kernel: [312480.864173] Call Trace:
> Jun 14 16:41:30 daniloth kernel: [312480.864191]  [<ffffffff813262f2>] ? __mutex_lock_common+0x127/0x193
> Jun 14 16:41:30 daniloth kernel: [312480.864239]  [<ffffffff8132642a>] ? mutex_lock+0x1a/0x33
> Jun 14 16:41:30 daniloth kernel: [312480.864295]  [<ffffffffa045156b>] ? radeon_cs_ioctl+0x30/0x191 [radeon]
> Jun 14 16:41:30 daniloth kernel: [312480.864304]  [<ffffffff8132641d>] ? mutex_lock+0xd/0x33
> Jun 14 16:41:30 daniloth kernel: [312480.864350]  [<ffffffffa044f740>] ? drm_gem_object_unreference_unlocked
> +0x2d/0x3c [radeon]
> Jun 14 16:41:30 daniloth kernel: [312480.864395]  [<ffffffffa044fc1c>] ? radeon_gem_create_ioctl+0x66/0x7c [
> radeon]
> Jun 14 16:41:30 daniloth kernel: [312480.864424]  [<ffffffffa03d97c0>] ? drm_ioctl+0x24e/0x321 [drm]
> Jun 14 16:41:30 daniloth kernel: [312480.864467]  [<ffffffffa045153b>] ? radeon_cs_ioctl+0x0/0x191 [radeon]
> Jun 14 16:41:30 daniloth kernel: [312480.864479]  [<ffffffff811a132d>] ? timerqueue_add+0x80/0x9b
> Jun 14 16:41:30 daniloth kernel: [312480.864491]  [<ffffffff8100f502>] ? read_tsc+0x5/0x16
> Jun 14 16:41:30 daniloth kernel: [312480.864501]  [<ffffffff81062d25>] ? enqueue_hrtimer+0x3f/0x53
> Jun 14 16:41:30 daniloth kernel: [312480.864509]  [<ffffffff8106379a>] ? __hrtimer_start_range_ns+0x2ba/0x2c
> d
> Jun 14 16:41:30 daniloth kernel: [312480.864521]  [<ffffffff81104b0b>] ? do_vfs_ioctl+0x467/0x4b4
> Jun 14 16:41:30 daniloth kernel: [312480.864531]  [<ffffffff8104ae06>] ? do_setitimer+0x15a/0x1dc
> Jun 14 16:41:30 daniloth kernel: [312480.864540]  [<ffffffff81104ba3>] ? sys_ioctl+0x4b/0x70
> Jun 14 16:41:30 daniloth kernel: [312480.864548]  [<ffffffff81009952>] ? system_call_fastpath+0x16/0x1b
> 
> This happens with older version than I am using as well.  The experimental version didn't seem to fix be any better when I tried it.
> 
Cheers,
Julien