08:44DPA: I'm on mesa 20.1.10.
08:56Vanfanel: emersion: Is it expected that some drivers don't like the FB_ID & CRTC_ID props set to 0 for the primary plane? Seems to happen still in AMDGPU
08:56Vanfanel: the atomic commit returns -22 when trying to set FB_ID & CRTC_ID to 0
08:57Vanfanel: I seemed to understand it was "legal" to set them to 0
09:43Vanfanel: emersion: I am doing all this in a single commit: set the CONNECTOR prop CRTC_ID to 0, set CRTC prop ACTIVE to 0, set the PRIMARY PLANE prop CRTC_ID props FB_ID and CRTC_ID to 0. But I still get -22 on the commit.
09:43emersion: can you try setting the CRTC MODE_ID to zero as well?
09:43Vanfanel: emersion: right, I will try
09:43emersion: if you're still getting EINVAL, you might find it useful to enable verbose DRM logs
10:00Vanfanel: emersion: I'm still getting EINVAL, I have enabled verbose DRM logs, and I get "Atomic check failed with err: -22". Complete result log of the commit is: https://pastebin.ubuntu.com/p/WW8dmTyrGR/
10:00Vanfanel: emersion: any ideas on what's failing here? It does not fail on VC4, just on AMDGPU
10:04emersion: ah, how convenient, amdgpu doesn't print any error message
10:05emersion: it's definitely amdgpu checks that are failing
10:05emersion: do you have ALLOW_MODESET enabled?
10:05emersion: apparently amdgpu isn't happy about the primary plane configuration, since it's failing in drm_atomic_helper_check_planes
10:07Vanfanel: emersion: Yes, I have the ALLOW_MODESET flag enabled
18:59DrNick: so I discovered this years ago but forgot about it until an hackernews reminded me yesterday, there are DDC MCCS commands to put your monitor in power saving mode
18:59DrNick: which is interesting because they execute instantly
18:59DrNick: as opposed to the current situation with DPMS where I get like 10 seconds of a big ugly NO SIGNAL error before the monitor powers off
19:00DrNick: is there are reason they DDC commands aren't used in conjunction with DPMS?