Bug 83484 - [BYT/BDW/BSW]igt/drv_module_reload was killed and system can't reboot
Summary: [BYT/BDW/BSW]igt/drv_module_reload was killed and system can't reboot
Status: CLOSED FIXED
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: unspecified
Hardware: Other All
: high critical
Assignee: Intel GFX Bugs mailing list
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2014-09-04 08:41 UTC by Guo Jinxian
Modified: 2017-10-06 14:35 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments
dmesg (86.26 KB, text/plain)
2014-09-04 08:41 UTC, Guo Jinxian
no flags Details
kernel config file (103.13 KB, text/plain)
2014-09-05 07:00 UTC, Guo Jinxian
no flags Details
Possible fix (3.45 KB, patch)
2014-10-14 21:18 UTC, Paulo Zanoni
no flags Details | Splinter Review
dmesg (81.70 KB, text/plain)
2014-10-16 06:15 UTC, Guo Jinxian
no flags Details

Description Guo Jinxian 2014-09-04 08:41:14 UTC
Created attachment 105725 [details]
dmesg

==System Environment==
--------------------------
Regression: Not sure.
There has another bug 77511 for system hung

Non-working platforms: BDW

==kernel==
--------------------------
origin/drm-intel-nightly: 4144c90b76dfe6eaa2205ac947090786b5091cff(fails)
    drm-intel-nightly: 2014y-09m-03d-15h-44m-26s integration manifest
origin/drm-intel-next-queued: 8fe8a3feeb5aedd17f7b09e51b60b38a6ba7005f(fails)
    drm/i915: init sprites with univeral plane init function
origin/drm-intel-fixes: 77520c8fccc1a599ff7a7b786dacdfdad38b96e4(fails)
    drm/i915: Prevent recursive deadlock on releasing a busy userptr

==Bug detailed description==
-----------------------------
igt/drv_module_reload was killed

Output:
root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./drv_module_reload
unbinding /sys/class/vtconsole/vtcon1/: (M) frame buffer device
module successfully unloaded
libkmod: ERROR ../libkmod/libkmod.c:554 kmod_search_moddep: could not open moddep file '/lib/modules/3.17.0-rc2_drm-intel-nightly_4144c9_20140904+/modules.dep.bin'
Killed
root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# echo $?
137



==Reproduce steps==
---------------------------- 
1. ./drv_module_reload
Comment 1 Chris Wilson 2014-09-04 09:12:33 UTC
[  851.291557] Console: switching to mono frame buffer device 80x25

Otherwise known as wtf. Where did that come from and why didn't it switch back to the dummy device?
Comment 2 Paulo Zanoni 2014-09-04 13:09:41 UTC
Can you please attach your Kernel's .config?
Comment 3 Guo Jinxian 2014-09-05 07:00:55 UTC
Created attachment 105782 [details]
kernel config file

(In reply to comment #2)
> Can you please attach your Kernel's .config?

Please check the attachment.
Comment 4 Guo Jinxian 2014-09-11 07:35:08 UTC
This bug is able to reproduce on BSW on latest -nightly(72faa6a63351879b381bd9e4013f38bc94f11999)

[root@x-bsw01 tests]# ./drv_module_reload
unbinding /sys/class/vtconsole/vtcon1/: (M) frame buffer device
module successfully unloaded
Killed
[root@x-bsw01 tests]# echo $?
137
Comment 5 Guo Jinxian 2014-09-22 07:31:14 UTC
The failure is able to reproduce on BYT on latest -nightly(c5660b4ad395f1e34eacc22cf81c687edfc9c83c)


root@x-byt06:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./drv_module_reload
unbinding /sys/class/vtconsole/vtcon1/: (M) frame buffer device
module successfully unloaded
libkmod: ERROR ../libkmod/libkmod.c:554 kmod_search_moddep: could not open moddep file '/lib/modules/3.17.0-rc5_drm-intel-nightly_c5660b_20140922+/modules.dep.bin'
Killed
Comment 6 lu hua 2014-10-11 06:24:02 UTC
After run this case, the screen is black, connect it successfully via ssh, then run another igt cases, shows: No intel gpu found, run reboot, system is no response. It blocks testing.
Comment 7 Paulo Zanoni 2014-10-14 21:18:58 UTC
Created attachment 107840 [details] [review]
Possible fix

Can you please apply this patch, retest and post the dmesg here?

I tested this on BDW, and module_reload works for me on this machine, with this patch on top of drm-intel-nightly.
Comment 8 Guo Jinxian 2014-10-16 06:15:55 UTC
Created attachment 107913 [details]
dmesg

(In reply to Paulo Zanoni from comment #7)
> Created attachment 107840 [details] [review] [review]
> Possible fix
> 
> Can you please apply this patch, retest and post the dmesg here?
> 
> I tested this on BDW, and module_reload works for me on this machine, with
> this patch on top of drm-intel-nightly.

The failure still able to reproduce with this patch on latest -nightly()


root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./drv_module_reload
unbinding /sys/class/vtconsole/vtcon1/: (M) frame buffer device
module successfully unloaded
libkmod: ERROR ../libkmod/libkmod.c:554 kmod_search_moddep: could not open moddep file '/lib/modules/3.17.0_kcloud_43c5c7_20141016+/modules.dep.bin'
Killed
root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# echo $?
137
Comment 9 Paulo Zanoni 2014-10-17 21:47:03 UTC
(In reply to Guo Jinxian from comment #8)
> Created attachment 107913 [details]
> dmesg
> 
> (In reply to Paulo Zanoni from comment #7)
> > Created attachment 107840 [details] [review] [review] [review]
> > Possible fix
> > 
> > Can you please apply this patch, retest and post the dmesg here?
> > 
> > I tested this on BDW, and module_reload works for me on this machine, with
> > this patch on top of drm-intel-nightly.
> 
> The failure still able to reproduce with this patch on latest -nightly()
> 
> 
> root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests#
> ./drv_module_reload
> unbinding /sys/class/vtconsole/vtcon1/: (M) frame buffer device
> module successfully unloaded
> libkmod: ERROR ../libkmod/libkmod.c:554 kmod_search_moddep: could not open
> moddep file '/lib/modules/3.17.0_kcloud_43c5c7_20141016+/modules.dep.bin'

Does this file exist on your machine? Maybe your system setup is wrong? This really looks like a bug in your machine setup, not our driver.
Comment 10 Guo Jinxian 2014-10-21 06:32:47 UTC
(In reply to Paulo Zanoni from comment #9)
> (In reply to Guo Jinxian from comment #8)
> > Created attachment 107913 [details]
> > dmesg
> > 
> > (In reply to Paulo Zanoni from comment #7)
> > > Created attachment 107840 [details] [review] [review] [review] [review]
> > > Possible fix
> > > 
> > > Can you please apply this patch, retest and post the dmesg here?
> > > 
> > > I tested this on BDW, and module_reload works for me on this machine, with
> > > this patch on top of drm-intel-nightly.
> > 
> > The failure still able to reproduce with this patch on latest -nightly()
> > 
> > 
> > root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests#
> > ./drv_module_reload
> > unbinding /sys/class/vtconsole/vtcon1/: (M) frame buffer device
> > module successfully unloaded
> > libkmod: ERROR ../libkmod/libkmod.c:554 kmod_search_moddep: could not open
> > moddep file '/lib/modules/3.17.0_kcloud_43c5c7_20141016+/modules.dep.bin'
> 
> Does this file exist on your machine? Maybe your system setup is wrong? This
> really looks like a bug in your machine setup, not our driver.

Yes, some modules don't install by default on our Ubuntu devices. Execute commands below then run the tests, the result passes.

apt-get install nfs-common nfs-kernel-server
/etc/init.d/client_module
mount -a 


root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./drv_module_reload
unbinding /sys/class/vtconsole/vtcon1/: (M) frame buffer device
module successfully unloaded
module successfully loaded again
root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# echo $?
0


[root@x-bsw01 tests]#./drv_module_reload
unbinding /sys/class/vtconsole/vtcon1/: (M) frame buffer device
module successfully unloaded
module successfully loaded again

[root@x-bsw01 tests]# echo $?
0
Comment 11 Elizabeth 2017-10-06 14:35:57 UTC
Closing old verified.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.