Bug 107793

Summary: Black screen on boot for Fedora 28 with 4.17 kernel (i.e. with amdgpu.dc defaulted)
Product: DRI Reporter: Simon Geard <simon.geard>
Component: DRM/AMDgpuAssignee: Default DRI bug account <dri-devel>
Status: RESOLVED MOVED QA Contact:
Severity: normal    
Priority: medium CC: harry.wentland, nicholas.kazlauskas, sunpeng.li
Version: unspecified   
Hardware: x86-64 (AMD64)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
Contents of /proc/cpuinfo on successful boot
none
Output of lspci on successful boot
none
Output of dmesg on successful boot
none
Output of journalctl -k on failed boot
none
Output of journalctl -k on successful boot
none
output of journalctl on failed boot with amdgpu.dc=1 amdgpu.dc_log=1 none

Description Simon Geard 2018-09-03 08:32:07 UTC
Created attachment 141422 [details]
Contents of /proc/cpuinfo on successful boot

My Fedora 28 desktop stopped booting up cleanly after switching to the 4.17 kernel where amdgpu.dc became the default... the system seems to be working (e.g when installing updates, it automatically reboots once done), but the screen goes black immediately after grub loads the kernel, and it remains that way indefinitely. Passing "amdgpu.dc=0" on the kernel command line resolves the problem.

The active graphics card is an AMD RX-470, the CPU is an AMD A10-7850k (with unused onboard graphics).


This was logged on the RedHat bugzilla, but after receiving no response in over a month, I've opted to log it here instead. Here's the link to the downstream bug:

https://bugzilla.redhat.com/show_bug.cgi?id=1601812

I've re-attached the same files provided there – the output of lspci, dmesg, and /proc/cpuinfo for a successful boot – but if you need anything else to track this down, let me know.
Comment 1 Simon Geard 2018-09-03 08:32:32 UTC
Created attachment 141423 [details]
Output of lspci on successful boot
Comment 2 Simon Geard 2018-09-03 08:33:09 UTC
Created attachment 141424 [details]
Output of dmesg on successful boot
Comment 3 Nicholas Kazlauskas 2018-09-04 12:41:13 UTC
If you can provide a dmesg log for your black screen boot that would help.
Comment 4 Simon Geard 2018-09-05 08:26:14 UTC
Created attachment 141459 [details]
Output of journalctl -k on failed boot
Comment 5 Simon Geard 2018-09-05 08:26:33 UTC
Created attachment 141460 [details]
Output of journalctl -k on successful boot
Comment 6 Simon Geard 2018-09-05 08:38:23 UTC
Ok, I've added the output of "journalctl -k -b" (the current, successful boot) and "journalctl -k -b -1" (a failed boot a few minutes previously).
Comment 7 Nicholas Kazlauskas 2018-09-05 15:12:37 UTC
Can you try booting with amdgpu.dc=0 in your kernel boot parameters?
Comment 8 Simon Geard 2018-09-06 08:33:33 UTC
(In reply to Nicholas Kazlauskas from comment #7)
> Can you try booting with amdgpu.dc=0 in your kernel boot parameters?

As noted in the original bug report, "passing "amdgpu.dc=0" on the kernel command line resolves the problem." The "successful boot" logs I've provided were all captured under that setting, since the system is unusable without it.

I *suspect* the system is booting fine apart from the graphics, because when Fedora reboots to apply updates, it does so successfully before powering off as expected – I just can't see anything on the screen.
Comment 9 ZLD 2018-09-08 12:38:44 UTC
Just to add - Same happens to me with an R9 380 graphics card since amdgpu.dc was enabled. I am using my monitor through the vga port (with an dvi to vga adapter on my gpu). Passing "amdgpu.dc=0" also resolves the problem for me. Is there any debug that I should provide that may help?
Comment 10 ZLD 2018-09-08 12:41:48 UTC
Created attachment 141483 [details]
output of journalctl on failed boot with amdgpu.dc=1 amdgpu.dc_log=1
Comment 11 Simon Geard 2018-12-03 07:51:13 UTC
Is there any update on this, any more information I can provide to help get this fixed? I can confirm it's still occurring on the 4.19.5-200 kernel currently included in Fedora 28...
Comment 12 Martin Peres 2019-11-19 08:50:26 UTC
-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/drm/amd/issues/503.

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.