[Bug 100964] RX-480 [drm:gfx_v8_0_ring_test_ring [amdgpu]] *ERROR* amdgpu: ring 0 test failed (scratch(0xC040)=0xCAFEDEAD)

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=100964

            Bug ID: 100964
           Summary: RX-480 [drm:gfx_v8_0_ring_test_ring [amdgpu]] *ERROR*
                    amdgpu: ring 0 test failed
                    (scratch(0xC040)=0xCAFEDEAD)
           Product: DRI
           Version: unspecified
          Hardware: x86-64 (AMD64)
                OS: Linux (All)
            Status: NEW
          Severity: major
          Priority: medium
         Component: DRM/AMDgpu
          Assignee: dri-devel at lists.freedesktop.org
          Reporter: suzaku.29a at gmail.com

Created attachment 131248
  --> https://bugs.freedesktop.org/attachment.cgi?id=131248&action=edit
dmesg output on amdgpu load failure

Inconsistent amdgpu driver loading for an RX-480:

Most of the time, the driver will fail to load starting with the error:
    [drm:gfx_v8_0_ring_test_ring [amdgpu]] *ERROR* amdgpu: ring 0 test failed
(scratch(0xC040)=0xCAFEDEAD)
The screen goes into stand-by and I get no display output. The rest of
the system still loads normally and I can ssh and look around.

------------
But it will, very seldom, load the driver normally (KMS enabled, display
still active).
------------

I can use nomodeset to prevent amdgpu from being initialized and the
display continues to work, but without 3D accel and just one monitor
working (out of two).

To confirm that the GPU actually works (both in general and in linux
with amdgpu), I tested it on a friend's PC with the same distro I'm
using (Debian Testing/amd64). The driver loaded normally on the first
try and worked consistently for all the reboots we tried.

Since then I've read anything I could find on the issue and I couldn't
find any solution. There are similar reports on other AMD videocards,
but none of them give me a solution for the inconsistency.

Things tried so far:
    Debian testing and unstable with most of the kernels released since
        4.7.0-1 up to their latest kernel 4.9.25-1 (4.9.0-3 in their
        versioning system).
    Gentoo with the genkernel 4.9.16.
    Gentoo custom kernel 4.9.16
    Gentoo with kernel 4.11 from https://cgit.freedesktop.org/~agd5f/linux/
        (drm-fixes-4.11).
    Alternating ACPI and most, if not all, amdgpu parameters in all
        tested kernels.
    Using different outputs (DVI, HDMI and both. I cannot test
        DisplayPort).

The Debian and Gentoo installs are fresh in different drives. I can
test things on either one.

I'd be ok with my MB or CPU somehow being too old or incompatible. But
the few successful boots tell me that it can and *does* work on my pc.

I'm attaching two dmesg logs, one for the working boot and one for the
failing one. They were run shortly after each other and nothing was
changed in between.

I'm attaching lspci output too.

Please let me know if you need any other info or clarification on the
error.

Regards.

HV

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/dri-devel/attachments/20170508/1229ef2a/attachment.html>