Re: [Bug report] AMD Radeon R9 380 (Tonga) suspend resume regression: black screen / no EDID on kernel 6.13+ (ref: Ubuntu #2142389)

2026-03-06 Thread Thorsten Leemhuis 
On 3/3/26 15:26, Alex Huang wrote:
> On Tue, Mar 3, 2026 at 7:30 AM Danilo Machado
> mailto:[email protected]>>
> wrote:
> 
> To: [email protected]
> 
> Subject: [Bug report] AMD Radeon R9 380 (Tonga) suspend resume
> regression: black screen / no EDID on kernel 6.13+ (ref: Ubuntu
> #2142389)
> 
> Dear AMDGPU/DRM maintainers,
> 
> I'm reporting a suspend/resume regression affecting the Radeon R9
> 380 (Tonga / GCN 3rd gen) on recent kernels (6.13+), observed on
> Ubuntu 24.10 derivatives (Zorin OS 18) with kernel 6.17.0-14-generic.

That is a vendor kernel that contains quite a few modifications that
might contain the bug you are seeing; hence please either report the
problem to your vendor or check with the latest vanilla mainline kernel
(e.g. 7.0-rc2; latest 6.19.y might also suffice, but better test
7.0-rc). The latter is needed anyway, as the bug might already be fixed.
If the problem still happens there, you as mentioned by Alex most likely
need to do a bisection to get some tracking on this.

Ciao, Thorsten

> A nontrivial amount of pm/EDID changes occurred between 6.12 and 6.13,
> are you able to bisect the kernel between those two revisions? Even
> narrowing it down to a release candidate should be useful.
> 
> #regzbot introduced: v5.12..v5.13
> #regzbot from: Danilo Machado  >
> #regzbot monitor: https://bugs.launchpad.net/ubuntu/+source/linux/
> +bug/2142389 
> 
> Best,
> Alex H
> 
> 
> **Hardware:**
> - GPU: AMD Radeon R9 380 Series (Tonga XT, Tonga)
> - Motherboard: Gigabyte B450 AORUS PRO WIFI (BIOS F66)
> - CPU: AMD Ryzen 5 5500
> - Connection: Direct HDMI
> - Display server: X11
> - Power state: Deep sleep (S3)
> 
> **Symptoms:**
> - After suspend (first cycle often works; subsequent cycles fail
> consistently):
>   - System wakes (fans/LEDs active), but display shows black
> screen / no signal ("HDMI Out of Range" on monitor).
>   - EDID read/handshake fails → no video output recovered.
>   - Keyboard/mouse may respond briefly, but desktop freezes → hard
> reboot required.
> - Logs show:
>   [drm] *ERROR* HDMI-A-1: probed a monitor but no|invalid EDID
>   [drm] *ERROR* No EDID read.
> 
> 
> **Kernels tested:**
> - Broken: 6.13.12 → 6.17.0-14 (regression starts between 6.12 and 6.13)
> - Stable: 6.12.74 (LTS) – suspend/resume fully reliable, no EDID errors
> 
> **GRUB parameters used:**
> quiet splash amdgpu.si_support=1 radeon.si_support=0 amdgpu.runpm=0
> amdgpu.dc=0
> 
> **Reference:**
> This matches Ubuntu bug #2142389: https://bugs.launchpad.net/ubuntu/
> +source/linux/+bug/2142389  +source/linux/+bug/2142389>
> (Reported by me on Launchpad; includes full dmesg logs and attachments)
> 
> The issue appears to be in the DRM/DC layer (HDMI EDID re-detection
> during multi-cycle suspend/resume on Tonga). Downgrading to 6.12.74
> resolves it completely.
> 
> Is this a known regression? Any patches or additional debug info
> needed from my side?
> 
> Happy to test patches, provide more logs (dmesg, journalctl -b -u
> systemd-suspend, full amdgpu debug), or try kernel builds.
> 
> Thanks for your attention and work on amdgpu!
> 
> 
> Tested kernels:
> 
> 6.12.74 — stable, HDMI resume works
> 6.13.12 — first resume OK, second suspend fails
> 6.14.x — HDMI issues
> 6.17.0-14 — frequent resume black screen
> 
> Best regards,  
> Danilo Machado  
> Minas Gerais, Brazil  
> (Original reporter of Ubuntu bug #2142389)
>

Re: [Bug report] AMD Radeon R9 380 (Tonga) suspend resume regression: black screen / no EDID on kernel 6.13+ (ref: Ubuntu #2142389)

2026-03-03 Thread Alex Huang 
Hi,
+cc amd-gfx
+cc regressions
-cc freedesktop

On Tue, Mar 3, 2026 at 7:30 AM Danilo Machado 
wrote:

> To: [email protected]
> Subject: [Bug report] AMD Radeon R9 380 (Tonga) suspend resume regression:
> black screen / no EDID on kernel 6.13+ (ref: Ubuntu #2142389)
>
> Dear AMDGPU/DRM maintainers,
>
> I'm reporting a suspend/resume regression affecting the Radeon R9 380
> (Tonga / GCN 3rd gen) on recent kernels (6.13+), observed on Ubuntu 24.10
> derivatives (Zorin OS 18) with kernel 6.17.0-14-generic.
>
A nontrivial amount of pm/EDID changes occurred between 6.12 and 6.13, are
you able to bisect the kernel between those two revisions? Even narrowing
it down to a release candidate should be useful.

#regzbot introduced: v5.12..v5.13
#regzbot from: Danilo Machado 
#regzbot monitor:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2142389

Best,
Alex H

>
> **Hardware:**
> - GPU: AMD Radeon R9 380 Series (Tonga XT, Tonga)
> - Motherboard: Gigabyte B450 AORUS PRO WIFI (BIOS F66)
> - CPU: AMD Ryzen 5 5500
> - Connection: Direct HDMI
> - Display server: X11
> - Power state: Deep sleep (S3)
>
> **Symptoms:**
> - After suspend (first cycle often works; subsequent cycles fail
> consistently):
>   - System wakes (fans/LEDs active), but display shows black screen / no
> signal ("HDMI Out of Range" on monitor).
>   - EDID read/handshake fails → no video output recovered.
>   - Keyboard/mouse may respond briefly, but desktop freezes → hard reboot
> required.
> - Logs show:
>   [drm] *ERROR* HDMI-A-1: probed a monitor but no|invalid EDID
>   [drm] *ERROR* No EDID read.
>

> **Kernels tested:**
> - Broken: 6.13.12 → 6.17.0-14 (regression starts between 6.12 and 6.13)
> - Stable: 6.12.74 (LTS) – suspend/resume fully reliable, no EDID errors
>
> **GRUB parameters used:**
> quiet splash amdgpu.si_support=1 radeon.si_support=0 amdgpu.runpm=0
> amdgpu.dc=0
>
> **Reference:**
> This matches Ubuntu bug #2142389:
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2142389
> (Reported by me on Launchpad; includes full dmesg logs and attachments)
>
> The issue appears to be in the DRM/DC layer (HDMI EDID re-detection during
> multi-cycle suspend/resume on Tonga). Downgrading to 6.12.74 resolves it
> completely.
>
> Is this a known regression? Any patches or additional debug info needed
> from my side?
>
> Happy to test patches, provide more logs (dmesg, journalctl -b -u
> systemd-suspend, full amdgpu debug), or try kernel builds.
>
> Thanks for your attention and work on amdgpu!
>
>
> Tested kernels:
>
> 6.12.74 — stable, HDMI resume works
> 6.13.12 — first resume OK, second suspend fails
> 6.14.x — HDMI issues
> 6.17.0-14 — frequent resume black screen
>
> Best regards,
> Danilo Machado
> Minas Gerais, Brazil
> (Original reporter of Ubuntu bug #2142389)
>