[Bug 110848] Everything using GPU gets stuck after running+killing parallel Media loads (after running 3D benchmarks)

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=110848

--- Comment #6 from Eero Tamminen <eero.t.tamminen at intel.com> ---
>From the full ffmpeg stack it's clearer what is the trigger.

SIGINT is sent for the transcoding process group, kernel reacts to that and on
do_exit() for the FFmpeg it calls task_work_run(), which I assume to be adding
stuff for the kworker that then calls __i915_gem_free_work()?

Name:   kworker/u8:0+i915
State:  D (disk sleep)
Pid:    7274
[<0>] __i915_gem_free_work+0x5f/0x90 [i915]
[<0>] process_one_work+0x1e9/0x410
[<0>] worker_thread+0x2d/0x3d0
[<0>] kthread+0x113/0x130
[<0>] ret_from_fork+0x35/0x40
Name:   ffmpeg
State:  Z (zombie)
Pid:    7285
Name:   ffmpeg
State:  Z (zombie)
Pid:    7292
Name:   ffmpeg
State:  D (disk sleep)
Pid:    7293
[<0>] __fput+0xae/0x200
[<0>] task_work_run+0x84/0xa0
[<0>] do_exit+0x308/0xba0
[<0>] do_group_exit+0x33/0xa0
[<0>] get_signal+0x121/0x910
[<0>] do_signal+0x30/0x6b0
[<0>] exit_to_usermode_loop+0x96/0xb0
[<0>] prepare_exit_to_usermode+0x8c/0xb0
[<0>] retint_user+0x8/0x8


(In reply to Eero Tamminen from comment #3)
> PS: while set of 3D tests run before media tests in setup 1 and setup 2
> differ, some of the tests in setup 1 are Wayland/EGL variants of the X/GLX
> tests in setup 2.  Those are GfxBench 5.0 and SynMark2 v7.  X based GPU
> memory bandwidth tests are same in both setups.  Rest of the tests differ,
> but are also simpler, so most likely they don't matter for this bug.

Running just gfxbench, synmark & membw 3D tests before media tests wasn't
enough either, so for now I can can test this only after full 3-4h test round
=> can't really bisect this to steps that would take only minutes, iterating
from full set down to something like that would take way more time than I'm
allowed for this. :-/

-- 
You are receiving this mail because:
You are the assignee for the bug.
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: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20190607/734478f2/attachment.html>