|Summary:||BUG: kernel NULL pointer dereference - supervisor read access in kernel mode|
|Product:||DRI||Reporter:||Tom Seewald <tseewald>|
|Component:||DRM/AMDgpu||Assignee:||Default DRI bug account <dri-devel>|
|Status:||RESOLVED FIXED||QA Contact:|
|i915 platform:||i915 features:|
Description Tom Seewald 2019-08-18 04:36:39 UTC
Created attachment 145088 [details] dmesg of freeze Distribution: Fedora 30 Kernel: 5.2.8 Mesa: 19.1.4 GPU: RX 560 Problem: While browsing the internet using Chromium my entire machine froze, it was completely unresponsive to input but audio continued to play. Snippet of BUG: [173497.820810] BUG: kernel NULL pointer dereference, address: 00000000000002b4 [173497.820815] #PF: supervisor read access in kernel mode [173497.820817] #PF: error_code(0x0000) - not-present page [173497.820818] PGD 0 P4D 0 [173497.820822] Oops: 0000 [#1] SMP PTI [173497.820825] CPU: 2 PID: 21197 Comm: kworker/u8:2 Not tainted 5.2.8-200.fc30.x86_64 #1 [173497.820826] Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./Z77 Extreme4, BIOS P2.90 07/11/2013 [173497.820839] Workqueue: events_unbound commit_work [drm_kms_helper] [173497.820953] RIP: 0010:dc_stream_log+0x6/0xb0 [amdgpu] See attached dmesg for full message. Let me know if you need any additional information.
Comment 1 Tom Seewald 2019-08-18 20:53:50 UTC
This bug report on the kernel bugtracker appears to be the same or at least a very similar problem: https://bugzilla.kernel.org/show_bug.cgi?id=204181 In that report, Nicholas Kazlauskas posted a possible fix: https://patchwork.freedesktop.org/series/64505/ I have yet to test to see if this patch resolves the crashes I am experiencing, but I don't see this as being in the 5.3-rc series, and certainly not in the 5.2 stable releases. Was this fix perhaps overlooked/missed? So far I just see that it is in amd-staging-drm-next as of August 1st: https://cgit.freedesktop.org/~agd5f/linux/commit/?h=amd-staging-drm-next&id=a4f6a4c7de97335f3452229e67521eae338af10e
Comment 2 Tom Seewald 2019-10-03 04:14:58 UTC
The patches mentioned earlier have landed in 5.4 and 5.3.2, and after running 5.4-rc1 for several days now I have been unable to reproduce the crash.