Bug 51677

Summary: [IVB] Running module_reload after calling function drm_open_any() will cause GPU hang
Product: DRI Reporter: Du Yan <yanx.du>
Component: DRM/IntelAssignee: Daniel Vetter <daniel>
Status: CLOSED FIXED QA Contact:
Severity: normal    
Priority: medium CC: ben, chris, daniel, jbarnes
Version: unspecified   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
dmesg
none
dmesg file
none
errorstate none

Description Du Yan 2012-07-02 18:34:45 UTC 
Created attachment 63744 [details]
dmesg

System Environment:
--------------------------
Platform: Ivybridge
Kernel: (drm-intel-fixes)7aa1e7f06d6ea1bce3b27630d50769d13da28b1a
Some additional commit info:
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Wed Jun 27 16:43:36 2012 +0200

Revert "drm/i915: allow PCH PWM override on IVB"

Bug detailed description:
-------------------------
Call the function drm_open_any() first, and then exit without closing the calling function. After that, running the test case ./module_reload could cause GPU hang.

Reproduce steps:
1. Build the following C code:

int main(int argc, char *argv[])
{
int fd;
fd = drm_open_any();
exit(77);
}

2. run it.
3. run case module_reload
4. check the dmesg


Actually Result: GPU hang during the module_reload test. (dmesg log has attached.)
Comment 1 Chris Wilson 2012-07-14 09:19:46 UTC 
Being lazy and not having reproduced this... Is this still an issue?
Comment 2 Du Yan 2012-07-16 05:05:00 UTC 
Issue still can reproduce. Both dmesg log and error states are attached.

Kernel: (drm-intel-fixes)c4aed353b1b079eb4843e6a708fc68b4b28f72aa
Some additional commit info:
Merge: 6bc5154 e15ebe0
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Thu Jul 5 13:20:02 2012 -0700
Comment 3 Du Yan 2012-07-16 05:06:32 UTC 
Created attachment 64252 [details]
dmesg file
Comment 4 Du Yan 2012-07-16 05:07:33 UTC 
Created attachment 64253 [details]
errorstate
Comment 5 Daniel Vetter 2012-08-22 10:39:18 UTC 
Oops, we've lost a bit track of this one. Is this still an issue with latest -fixes/-queued?
Comment 6 shui yangwei 2012-08-29 07:31:01 UTC 
Kernel: (drm-intel-fixes)a51d4ed01e5bb39d2cf36a12f9976ab08872c192
Some additional commit info:
Author: Calvin Walton <calvin.walton@kepstin.ca>
Date:   Fri Aug 24 07:56:31 2012 -0400

    i915: Quirk no_lvds on Gigabyte GA-D525TUD ITX motherboard


This issue not appear in this version.
Comment 7 Daniel Vetter 2012-08-29 07:49:50 UTC 
So can we close this, or does it still happen somewhere? I somewhat suspect it's a timing thing, does few seconds of sleep between drm_open_any and the module reload "fix" the issue?
Comment 8 shui yangwei 2012-09-10 09:01:21 UTC 
Kernel: (drm-intel-testing)c3c3d4e9c2daeca01c42040cda0e5e0579c5c80b



Now this issue can't reproduced. so I think it's possible to close this bug.
Comment 9 Daniel Vetter 2012-09-10 15:27:27 UTC 
Ok, two kernels with quite a few things changed in between worked, I guess we can close this one. Please reopen if you see this issue again, thanks.

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.