Tonga PRO [Radeon R9 285/380] [1002:6939] blank screen

Submitted by Steve Falco

Assigned to Default DRI bug account

Link to original bug (#103111)

Description

Created attachment 134681
dmesg file showing a successful boot

Attachment 134681, "dmesg file showing a successful boot":
working.dmesg

added AMDgpu bugzilla labels

Steve Falco uploaded an attachment:

Attachment 134682, "good xorg log":
working.Xorg.0.log

Steve Falco uploaded an attachment:

Attachment 134683, "bad xorg log":
bad.Xorg.0.log

Steve Falco uploaded an attachment:

Attachment 134684, "bad dmesg log":
bad.dmesg

Steve Falco uploaded an attachment:

Attachment 134685, "second bad xorg log":
Xorg.0.log

Steve Falco said:

I have a Radeon R9 285/380 video card in a machine that is running Fedora 26. When I run with Fedora kernel version 4.11.11 the video card works properly, using the amdgpu kernel driver.

However, once I upgrade to a newer kernel in the 4.12 or 4.13 series, my display goes blank during boot and stays that way for a long time. The monitor appears to be getting sync, because it doesn't go to sleep, however there is nothing visible on the screen.

Eventually, error messages are displayed and the system never boots fully.

I've attached logs showing both a good and a bad startup. For some reason, in the "bad case", I got two xorg log files, so I've attached both.

Please let me know if there is more information that I can collect to help in debug.

Steve Falco uploaded an attachment:

Attachment 134713, "dmesg from agd5f 4.13.0 kernel":
dmesg

Steve Falco uploaded an attachment:

Attachment 134714, "xorg log from agd5f 4.13.0 kernel":
Xorg.0.log

Steve Falco said:

As per the mailing list suggestions, I've built a kernel from the amd-staging-drm-next branch of the agd5f kernel tree.

While it gives me a usable gui, I also get "corrupted mapping in tail page" errors. Please see the dmesg attachment at https://bugs.freedesktop.org/attachment.cgi?id=134713 for the details.

Steve Falco uploaded an attachment:

Previous log file was truncated. This one is better.

Attachment 134717, "dmesg with increased log buffer parameter":
dmesg

Steve Falco uploaded an attachment:

Attachment 134718, "xorg log corresponding to the increased log buffer size":
Xorg.0.log

Steve Falco said:

I re-ran the tests with increased log buffer size, since the previous run was truncated.

The first error message that I see is [ 56.199374] BUG: Bad page state in process Xorg pfn:817601, followed by [ 56.199396] page dumped because: nonzero compound_mapcount.

This issue hasn't had any activity since 2019-11-19. The AMD driver stack changes rapidly and contains lots of shared code across products so it's possible that it has already been fixed. Please upgrade to a current stable kernel and userspace stack and try again. If you still experience this issue with the latest driver stack, please capture relevant logging and open a new issue referring back to this one.

closed