#asahi-dev · 2025-12-30 11:16
chadmedmy plan at some point is to see what happens when we just fill in the mode properly and modeset to the vrr timings#asahi-dev · 2025-12-30 11:10
chadmedthere seems to be a lot of slack and no rhyme/reason to the values. some swaps dont even populate them at all#asahi-dev · 2025-12-30 11:09
chadmedthough on the topic i am less sure about those timestamps now than i was a week ago#asahi-dev · 2025-12-30 11:08
chadmedeliminating the lack of edid/timingelements as a factor to consider#asahi-dev · 2025-12-30 11:08
chadmedive been avoiding using that as a base because i want to see what dcp does for a display that behaves properly first#asahi-dev · 2025-12-30 11:06
chadmedyeah i mightve spent like 4 hours yesterday trying to figure out why HDR wasnt coming up in plasma#asahi-dev · 2025-12-30 11:05
chadmedso i cant even arm the setting#asahi-dev · 2025-12-30 11:05
chadmedi dont get that far because macos doesnt recognise the monitors as vrr capable#asahi-dev · 2025-12-30 11:04
chadmedon macos? plug in my monitor then look for the setting in system settings#asahi-dev · 2025-12-30 11:03
chadmedmacos doesnt want to do vrr with any of my monitors for some reason even though TimingElements brings back a bunch of supported VRR modes#asahi-dev · 2025-12-30 11:03
chadmednot sure what we need to do for VRR. hopefully its just a matter of filling out the drm_mode with the right info from TimingElements, setting a VRR mode then letting iomfb do the needful #asahi-dev · 2025-12-30 11:01
chadmedthis is doubly compounded when using multiple displays (especially across hdmi and dp) as the colour spaces can be different and we cant deal with that gracefully#asahi-dev · 2025-12-30 11:00
chadmedtrying to grab the hdr metadata off the connector at the point we need it for dcp to properly deal with mixed hdr/sdr surfaces is flaky, and compositors cant yet deal with it anyway#asahi-dev · 2025-12-30 10:59
chadmedhdr is a no-go until the colour op stuff is merged so dont even bother with that (cc oliver)#asahi-dev · 2025-12-30 10:59
chadmednow that the cat's out of the bag ive pushed a rebased dcp/scanout branch #asahi-dev · 2025-12-30 10:53
chadmedi would submit it as part of a "make rtkit on m3+ work" series #asahi-dev · 2025-12-30 10:52
chadmedi dont think theres any point in submitting the oslog thing by itself #asahi-dev · 2025-12-30 10:47
chadmedanything thats already upstream please just submit upstream #asahi-dev · 2025-12-29 15:19
chadmedor yknow at least generating the right edids for the right displays instead of using random blobs#asahi-dev · 2025-12-29 15:19
chadmedyeah but we can do it properly with edid blobs in /lib/firmware/vendor#asahi-dev · 2025-12-29 15:16
chadmedso having a fake edid doesnt help us tell dcp to drive the monitor correctly#asahi-dev · 2025-12-29 15:16
chadmedbut the timing and colour elements dcp detects are wrong #asahi-dev · 2025-12-29 15:16
chadmedyes, i know#asahi-dev · 2025-12-29 15:15
chadmeddcp gives us {a, b, c} timing modes and {x, y, z} colour modes to pick from#asahi-dev · 2025-12-29 15:13
chadmedwe cant just say to it "try to drive this monitor with 30bpp BT.2020 PQ wide gamut"#asahi-dev · 2025-12-29 15:13
chadmedDCP does that by itself#asahi-dev · 2025-12-29 15:12
chadmedwe dont feed the EDID back into DCP for it to parse into modes#asahi-dev · 2025-12-29 15:11
chadmedwe get real EDIDs from DCP#asahi-dev · 2025-12-29 15:11
chadmedno its the other way around#asahi-dev · 2025-12-29 15:10
chadmedthe problem with faking the EDIDs is that we cant just tell IOMFB to modeset into an arbitrary colour/timing mode#asahi-dev · 2025-12-29 15:10
chadmedyeah i want to do that Properly(tm) #asahi-dev · 2025-12-29 15:06
chadmedive tested that the driver itself works using gstreamer but i wanted to stress test mutliple planes all going crazy #asahi-dev · 2025-12-29 15:06
chadmedmutter can too#asahi-dev · 2025-12-29 15:05
chadmedyes, its a recent feature#asahi-dev · 2025-12-29 15:05
chadmed(and enabling direct scanout of non-rgb framebuffers but kwin refuses to cooperate on that front)#asahi-dev · 2025-12-29 15:04
chadmedthe main thing in that branch is the reworking of how we set and commit planes #asahi-dev · 2025-12-29 15:04
chadmedor you can rebase on top of my branch and keep doing whatever#asahi-dev · 2025-12-29 15:03
chadmedif you want :P#asahi-dev · 2025-12-29 14:51
chadmedwe have issues with modsetting the correct colourimetry and i havent found a non-revolting way to correctly set the per-plane colourspace and eotf yet but i have some ideas ill play with tomorrow#asahi-dev · 2025-12-29 14:50
chadmedoliver: my dcp/scanout branch has hdr support for hdmi provided the edid has all the required bits#asahi-dev · 2025-12-29 00:19
chadmedhm aiui dcp wont let us modeset into a mode that it doesnt have a timing/colour element for#asahi-dev · 2025-12-28 23:09
chadmedeach install gets its own m1n1#asahi-dev · 2025-12-28 23:07
chadmedleio: vlc does not support hdr on macos at all and never has, likewise oliver's dcp hacks will only work on the builtin screens on 14/16" machines#asahi-dev · 2025-12-28 11:49
chadmedbut moving the page flip event handling to any other iomfb callback causes page flip timeouts#asahi-dev · 2025-12-28 11:48
chadmedi suspect we may get better and more accurate frame timing if we emit the page flip done event after iomfb has verified the swap request, since thats the point that the page flip actually begins#asahi-dev · 2025-12-28 00:32
chadmedive commented them all out and the only one that seems to do anything important is the one where we fake a swap with no surfaces#asahi-dev · 2025-12-28 00:31
chadmedshould we be faking vblanks if crtc_state->no_vblank is armed? i kinda feel like the fact that we "have" to when kms already knows how to deal with vblankless hardware is an us problem #asahi-dev · 2025-12-27 23:56
chadmedit uses the format fourccs for more than just the framebuffer format, e.g. r03w is argb2101010 but so is r01l. only the latter works properly for sdr argb2101010 framebuffers#asahi-dev · 2025-12-27 23:55
chadmedif its xrgb2101010/argb2101010 we will need to rework how we map drm formats to dcp formats#asahi-dev · 2025-12-27 23:51
chadmedoliver: what framebuffer format does gnome use for hdr