| Summary: | [ALL] race in tty register/deregister | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | DRI | Reporter: | fangxun <xunx.fang> | ||||||||||
| Component: | DRM/Intel | Assignee: | Imre Deak <imre.deak> | ||||||||||
| Status: | CLOSED FIXED | QA Contact: | Intel GFX Bugs mailing list <intel-gfx-bugs> | ||||||||||
| Severity: | major | ||||||||||||
| Priority: | high | CC: | christophe.prigent, intel-gfx-bugs | ||||||||||
| Version: | unspecified | ||||||||||||
| Hardware: | All | ||||||||||||
| OS: | Linux (All) | ||||||||||||
| Whiteboard: | |||||||||||||
| i915 platform: | i915 features: | ||||||||||||
| Attachments: |
|
||||||||||||
The sysfs stuff should be fixed. Please retest. Created attachment 92135 [details] [review] fix tty unregistering lockdep I just looked at this yesterday and could still reproduce it. I saw two issues, a tty vt unregistering lockdep warning (also in the reporter's dmesg) and a second warning of a i915 i2c cleanup ordering issue. Attaching two patches to fix those. Created attachment 92136 [details] [review] fix i915 i2c cleanup order Could you retest if the two attached patch applied on -nightly fixes the problem? Verify the patches applied on -nightly fixes the problem. Presumed fixed with Imre's commits. Please reopen if the problem persists with recent kernels. It still happens with recent kernels. (In reply to fangxun from comment #7) > It still happens with recent kernels. This is not very helpful. Please specify kernel version and attach dmesg with drm.debug=14 module parameter set on the latest kernel version tried. Created attachment 107602 [details]
dmesg wiht drm.debug=14
I test it on ivybridge with kernel drm-intel-fixes(91e564).
Imre, any update on your fixes? (In reply to Jesse Barnes from comment #10) > Imre, any update on your fixes? I posted an updated patchset, it's waiting for review: https://lkml.org/lkml/2014/12/15/847 Patches are merged, even with all the discussion about how to fix the "let's deprecate it instead" fbdev etc subsystem. Yay! commit 2cf30f752c8f73713f58796eb218cafa242d66fe Author: Imre Deak <imre.deak@intel.com> Date: Tue Dec 16 00:15:59 2014 +0200 vt: fix check for system/busy console drivers when unregistering them Oh, the last patch still isn't merged. Will follow up on lkml. All three merged, closing. commit 2cf30f752c8f73713f58796eb218cafa242d66fe Author: Imre Deak <imre.deak@intel.com> Date: Tue Dec 16 00:15:59 2014 +0200 vt: fix check for system/busy console drivers when unregistering them commit 4c215fe8e288c87b34c3d5bc910c77af586859d2 Author: Imre Deak <imre.deak@intel.com> Date: Tue Dec 16 00:16:00 2014 +0200 vt: fix locking around vt_bind/vt_unbind commit d364b5c3e446bbf9892edbb000bf4c358f86bd8b Author: Imre Deak <imre.deak@intel.com> Date: Wed Apr 1 21:06:16 2015 +0300 vt: fix console lock vs. kernfs s_active lock order |
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.
Created attachment 87716 [details] dmesg System Environment: -------------------------- Platform: Ivybridge Kernel: (drm-intel-fixes) cd66407810869709e0d6764409137d48c6812cfc Bug detailed description: ----------------------------- Start system with next-queued or next-fixed debug_build kernel, then run module_reload, you will see "[ INFO: possible circular locking dependency detected ]" in dmesg. It happens on next-queued and next-fixed branch. I try to bisect, but find it also exists in kernel 3.9.0-rc6. Reproduce steps: --------------------------- 1. start system with debug kernel 2 ./module_reload 2. see dmesg