Comment # 17 on bug 1206843 from Patrik Jakobsson

(In reply to Takashi Iwai from comment #11)
> There seems still some issue with MST.
> Now we received a Lenovo dock and I've tested T14s Gen 3 with the Lenovo
> dock.
> 
> Both S3 and S4 work fine as long as no dock is connected or the dock is kept
> connected while suspend/resume (although we see some spurious kernel
> WARNING).
> 
> But when a dock is connected at S3/S4 and gets removed before the resume:
> the machine can be resumed on the laptop screen, but the monitor on the dock
> can't be enabled after re-plugging the dock.
> 
> (Actually, the first replug doesn't detect the monitor at all.  When you
> replug twice, it'll probe the monitor, but the mode switching fails.)
> 
> On 6.1.12 and 6.2.9 kernels, the situation is better, the monitor works
> again after the second replug.
> 
> Patrik, on which kernel version is the DRM code of our SP5 based?
> I'm suspecting we're scratching the bug in the past upstream:
>   https://gitlab.freedesktop.org/drm/amd/-/issues/2171

This sounds familiar. I updated DRM to v6.0 but AMDGPU to v6.1. While doing so
I had to exclude a series of DP-MST refactoring from Lyude Paul [1]. Perhaps
this is a series that is needed? IIRC these patches also affected the extra DG2
patches I had to add to i915.

[1] https://patchwork.freedesktop.org/series/104844/#rev5