https://bugzilla.kernel.org/show_bug.cgi?id=221649
Bug ID: 221649
Summary: amdgpu regression on RX 5700 (Navi10): 6.17.9 works,
6.18.x causes GPU hangs
Product: Drivers
Version: 2.5
Hardware: AMD
OS: Linux
Status: NEW
Severity: high
Priority: P3
Component: Video(DRI - non Intel)
Assignee: [email protected]
Reporter: [email protected]
Regression: No
Created attachment 310318
--> https://bugzilla.kernel.org/attachment.cgi?id=310318&action=edit
GPU Coredump on 6.18.1 this morning.
GPU: RX 5700 (Navi10)
Distro: Arch Linux
Hello,
I've been getting ring gfx timeouts, ring comp timeouts, GPU resets, and the
like (see my attached coredump) on any kernel after 6.17.9 during certain
gaming workloads. In order to rule out userspace graphics regressions, I did
test several Mesa and Proton versions.
I tested kernels:
1. 7.0.1-zen (bad)
2. 6.18.34-lts (bad)
3. 6.18.1-arch (bad)
but rollbacks to 6.12.75-lts and 6.17.9 did not reproduce any of my issues,
hence my suspicion that it is a regression of some sort between 6.17.9 and
6.18.1.
During my testing, I did find a faulty RAM DIMM and removed it from my system,
as well as reinstalling all of my packages and running a fsck to ensure no
corruption was the cause, but I did not alleviate the issues.
My GPU Coredump is attached in a compressed .tar.gz format due to file size.
Let me know if anything else is needed, and I'll update as it is.
Thanks
--
You may reply to this email to add a comment.
You are receiving this mail because:
You are watching the assignee of the bug.
