[Bug 73383] [IVB/HSW ppgtt bisected] When run nightly piglit testing causes X no responsive

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=73383

--- Comment #16 from Ville Syrjala <ville.syrjala at linux.intel.com> ---
If just follow the code around I see plenty of codepaths that look like they
might lead to such recursion.

i915_vma_unbind
 i915_gem_object_finish_gpu
  i915_gem_object_wait_rendering
   i915_gem_object_wait_rendering__tail
   i915_gem_retire_requests_ring
    i915_gem_object_move_to_inactive
     drm_gem_object_unreference
      i915_gem_free_object
       i915_vma_unbind
        ...
    i915_gem_free_request
     i915_gem_context_unreference
      drm_gem_object_unreference
       i915_gem_free_object
        i915_vma_unbind
         ...
      ppgtt_release
       i915_gem_evict_vm
        i915_gem_retire_requests_ring
         ...
       i915_gem_retire_requests
        i915_gem_retire_requests_ring
         ...

I'm not really sure which paths actually are possible. I suppose I could try to
trace it in action and see where it goes.

This is the backtrace from my last oops BTW. Seems it gets a signal and then
tries to clean up, and ends up exploding. Eero showed me an apitrace from the
test in question and it doesn't seem to be doing anything special. Just a big
draw call w/ 6000000 vertices. I quickly tried to come up with an i-g-t test
that would create a context and then do a big rendercopy and kill itself mid
operation, but sadly I wasn't able to trigger the bug.

[   83.860988] Modules linked in: x86_pkg_temp_thermal coretemp kvm_intel
snd_hda_codec_hdmi kvm snd_hda_codec_realtek hid_generic iTCO_wdt aesni_intel
snd_hda_intel aes_x86_64 i915 glue_helper lrw snd_hda_codec gf128mul snd_hwdep
ablk_helper snd_pcm cryptd psmouse pcspkr usbhid hid snd_page_alloc i2c_i801
snd_timer lpc_ich mfd_core snd cfbfillrect cfbimgblt soundcore cfbcopyarea
drm_kms_helper wmi evdev
[   83.861640] CPU: 0 PID: 2807 Comm: GpuTest Not tainted 3.13.0-recurse+ #1
[   83.861667] Hardware name:                  /DZ77BH-55K, BIOS
BHZ7710H.86A.0100.2013.0517.0942 05/17/2013
[   83.861701] task: ffff8800cf3b61e0 ti: ffff8800d014a000 task.ti:
ffff8800d014a000
[   83.861728] RIP: 0010:[<ffffffffa015906d>]  [<ffffffffa015906d>]
i915_vma_unbind+0x7d/0x220 [i915]
[   83.861780] RSP: 0018:ffff8800d014bd48  EFLAGS: 00010246
[   83.861803] RAX: 6b6b6b6b6b6b6b6b RBX: ffff8800cf11ce38 RCX:
0000000000000006
[   83.861830] RDX: 0000000000000006 RSI: 0000000000260013 RDI:
ffff8800cf3b61e0
[   83.861856] RBP: ffff8800d014bd78 R08: 0000000000000002 R09:
0000000000000000
[   83.861882] R10: 0000000000000000 R11: 0000000000000001 R12:
ffff8800d01400f0
[   83.861908] R13: ffff8800d30fc900 R14: 0000000000000000 R15:
ffff8800d4fd8000
[   83.862251] FS:  00007f90e0e4a700(0000) GS:ffff88011f200000(0000)
knlGS:0000000000000000
[   83.862279] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   83.862304] CR2: 00007f90e4afe380 CR3: 00000000cf1de000 CR4:
00000000001407f0
[   83.862332] Stack:
[   83.862349]  ffff8800d014bd98 ffff8800cf11e080 ffff8800d01400f0
0000000000000000
[   83.862420]  ffff8800cd0d9ce0 ffff88011a46d6a0 ffff8800d014bda8
ffffffffa015e375
[   83.862491]  ffff8800d27836a8 ffff8800d27836a8 ffff8800d0140000
ffff8800d44a22e8
[   83.862560] Call Trace:
[   83.862585]  [<ffffffffa015e375>] i915_gem_evict_vm+0x75/0x120 [i915]
[   83.862616]  [<ffffffffa015c8cb>] i915_gem_context_free+0x14b/0x150 [i915]
[   83.862647]  [<ffffffffa015dbc5>] i915_gem_context_close+0xb5/0xc0 [i915]
[   83.862678]  [<ffffffffa0146a38>] i915_driver_preclose+0x38/0x60 [i915]
[   83.862708]  [<ffffffff81354398>] drm_release+0x78/0x5c0
[   83.862734]  [<ffffffff81150026>] __fput+0xd6/0x230
[   83.862757]  [<ffffffff811501ce>] ____fput+0xe/0x10
[   83.862782]  [<ffffffff81066327>] task_work_run+0xa7/0xe0
[   83.862807]  [<ffffffff810028a9>] do_notify_resume+0x59/0x80
[   83.862833]  [<ffffffff8129bd0e>] ? trace_hardirqs_on_thunk+0x3a/0x3f
[   83.862861]  [<ffffffff815150da>] int_signal+0x12/0x17
[   83.862884] Code: 0f 85 30 01 00 00 49 83 bd 40 01 00 00 00 0f 84 54 01 00
00 4c 89 ef e8 c2 d5 ff ff 85 c0 41 89 c6 0f 85 a9 00 00 00 48 8b 43 50 <48> 8b
90 a8 00 00 00 48 8b 92 b8 03 00 00 48 81 c2 a0 36 00 00
[   83.864309] RIP  [<ffffffffa015906d>] i915_vma_unbind+0x7d/0x220 [i915]

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20140207/026c8a87/attachment.html>