FS#61327 - [linux] 4.19+ AMDGPU fails to read EDID for a monitor

Attached to Project: Arch Linux
Opened by Ivan Molodetskikh (YaLTeR) - Wednesday, 09 January 2019, 19:30 GMT
Last edited by Andreas Radke (AndyRTR) - Tuesday, 17 March 2020, 09:44 GMT
Task Type Bug Report
Category Kernel
Status Closed
Assigned To Andreas Radke (AndyRTR)
Jan Alexander Steffens (heftig)
Architecture All
Severity High
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 1
Private No

Details

Description:
Starting from Linux 4.19 (4.20 is affected too, works fine on 4.18) AMDGPU fails to read EDID for one of my monitors. I can see the following message in the boot log:

[drm:dc_link_detect [amdgpu]] *ERROR* No EDID read.

This results in the affected monitor having a single tiny resolution.

Additional info:

Tested linux 4.18.16 (works), linux 4.19.12 (doesn't work), linux-zen 4.20.zen1-1 (doesn't work). The attached kernel logs are from linux 4.19.12 (bad-boot.log) and 4.18.16 (good-boot.log).

The GPU is AMD RX 580. The affected monitor is ASUS VG248QE (G-Sync modded), my other monitor DELL S2716DG (also supports G-Sync) is not affected.

Before this I used NVIDIA GTX 970 (with the proprietary driver), had no issues there.


Steps to reproduce:

1. Install linux 4.19 or above.
2. Reboot and observe that the monitor does not have the correct resolution.
This task depends upon

Closed by  Andreas Radke (AndyRTR)
Tuesday, 17 March 2020, 09:44 GMT
Reason for closing:  Upstream
Comment by Ivan Molodetskikh (YaLTeR) - Sunday, 20 January 2019, 08:24 GMT
I tried extracting EDID from a kernel 4.18 boot and forcing it with drm_kms_helper.edid_firmware=DP-4:edid/ASUS_VG248QE.bin and drm.edid_firmware=DP-4:edid/ASUS_VG248QE.bin but that doesn't seem to do anything.
Comment by Ivan Molodetskikh (YaLTeR) - Thursday, 31 January 2019, 06:48 GMT Comment by Dr. Evil (evilbot) - Monday, 04 February 2019, 04:14 GMT
Confirm identical issue. AMD Radeon RX Vega 64.
Comment by Andreas Radke (AndyRTR) - Tuesday, 10 December 2019, 13:09 GMT
Is this still an issue?
Comment by Sébastien Bernard (sbskas) - Monday, 16 December 2019, 19:11 GMT
It's still an issue running the 5.4.3 kernel all the way back from 4.19.
I managed to workaround the problem by including the modelines in the Xorg.conf.
Absolutely nothing happened in upstream.
Comment by loqs (loqs) - Monday, 16 December 2019, 20:11 GMT
Is 084b3765ec4baf8b7de57e7dc1e1ba1b6905e9d9 the causal commit?

Loading...