| Summary: | 7870 radeon crashes after kernel msg drm:radeon_mn_invalidate_range_start | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | DRI | Reporter: | JATothrim <jarmo.tiitto> | ||||||||
| Component: | DRM/Radeon | Assignee: | Default DRI bug account <dri-devel> | ||||||||
| Status: | RESOLVED FIXED | QA Contact: | |||||||||
| Severity: | major | ||||||||||
| Priority: | medium | ||||||||||
| Version: | unspecified | ||||||||||
| Hardware: | x86-64 (AMD64) | ||||||||||
| OS: | Linux (All) | ||||||||||
| Whiteboard: | |||||||||||
| i915 platform: | i915 features: | ||||||||||
| Attachments: |
|
||||||||||
Please attach your xorg log and dmesg output. Created attachment 117497 [details]
dmesg log
Created attachment 117498 [details]
Xorg.0.log
It was hard to catch the dmesg and Xorg log because soon as I close the Dolphin emu the system hard freezes --> no data synced.
Note that after the kernel prints those last lines the the bug is just triggered - after I close the Dolphin main window does the crash occurs.
If some one is interested I have kdump data of the system. Thanks to my kexec dump kernel. :)
Created attachment 117555 [details] [review] Maybe fix? I searched Vanilla kernel git logs/patches and found this patch. I hope it could fix the kernel crash on kernel 3.18.x series. I'm testing the patch now. I did more research and I was able to capture full kernel crash dump. Debugging the crashdump I found that The kernel dies at kernel version v3.18.19 in src lines: drivers/gpu/drm/radeon/radeon_mn.c:75 or drivers/gpu/drm/radeon/radeon_mn.c:76 on to NULL ptr dereference. At both code lines the kernel tries to lock a mutex apparently. I don't know exactly which one fails as I'm no expert debugging x86-64 assembly. Well have you tested this with the current upstream kernel? E.g. 4.1 or a recent rc of 4.2? That's usually the first thing you should to to figure out if newer kernels are already fixed. (In reply to Christian König from comment #6) > Well have you tested this with the current upstream kernel? E.g. 4.1 or a > recent rc of 4.2? > > That's usually the first thing you should to to figure out if newer kernels > are already fixed. Phew. I finally managed to compile/test newer kernel v4.2 and this bug is fixed! :) Yay! My tip for radeon 7870 owners: if your screen freezes and turns into a weird rainbow mess try kernel v4.2. |
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.
On my case is is a severe bug preventing me from playing games on dolphin. System: Linux Mint 17.3 using kernel 3.18.19 and oibaf graphics drivers ppa 100% reproducible in my system, happens always: -Start dolphin 4.+ GCN emulator and play a minute e.g. Super Mario Sunshine. -Close emulator window: kernel crashes soon after outputting 1-5 lines of: Jul 31 22:43:47 kernel: [ 6008.091188] [drm:radeon_mn_invalidate_range_start] *ERROR* (-1) failed to wait for user bo Jul 31 22:43:47 kernel: [ 6008.092895] [drm:radeon_mn_invalidate_range_start] *ERROR* (-1) failed to wait for user bo inxi -Gxx output: Graphics: Card: Advanced Micro Devices [AMD/ATI] Pitcairn XT [Radeon HD 7870 GHz Edition] bus-ID: 01:00.0 chip-ID: 1002:6818 X.Org: 1.15.1 driver: radeon Resolution: 1920x1080@60.0hz, 1680x1050@60.0hz GLX Renderer: Gallium 0.4 on AMD PITCAIRN GLX Version: 3.0 Mesa 11.0.0-devel (git-4d7e0fa 2015-08-01 trusty-oibaf-ppa) Direct Rendering: Yes PS: plz. be nice - this is my first bug report on here.