#
✕ nick:jannau
#asahi-dev · 2023-01-03 10:41
jannauthe dart_hw_reset() in runtime resume seems break PCIe. no issues so far after replacing the reset with a restore, i.e. instead of resetting tcr and ttbr it restores the old values and calls invalidate_tlb() only after the values are restored
#asahi-dev · 2023-01-02 23:53
jannauhttps://github.com/jannau/linux/tree/dcp/sleep is dcp/sleep rebased onto asahi-6.1-3_dcp-components, it still has issue (PCIe) but dcp runtime pm seems to work. tested on j274, j375 and j493 but not j314
#asahi-dev · 2023-01-02 23:48
jannaureplaces Thierry's reserved region patches with the latest version. implemented by reverting the original commits and redoing the updated changes
#asahi-dev · 2023-01-02 23:46
jannaumarcan: https://github.com/jannau/linux/tree/asahi-6.1-3_dcp-components has the components based probing and other changes, depends on https://github.com/AsahiLinux/m1n1/pull/272
#asahi-dev · 2023-01-02 23:32
jannaujust means we want to continue to lie about supporting it
#asahi-dev · 2023-01-02 23:09
jannaukwin/wayland uses a2rgb8 and falls back to argb8 if it is not supported
#asahi-dev · 2023-01-02 20:15
jannauah, no it should work with the pmgr reset
#asahi-dev · 2023-01-02 18:52
jannausigh, I can't unload the dcp module if we can not restart dcp
#asahi-dev · 2023-01-02 14:22
jannauyes, a2rgb10 is mostly useless. it would support hole punching for video underlays. that breaks with blending with the video, 2 bits of alpha are not enough to look good
#asahi-dev · 2023-01-02 14:17
jannauyuv overlay on surf 1 over surf 0 was only working with the premultied bit set
#asahi-dev · 2023-01-02 13:34
jannaucurrently ok but a2rgb10 overlays will not work
#asahi-dev · 2023-01-02 13:33
jannauoh, I missed that w30r is without alpha, we probably shouldn't claim to support a2rgb10 then
#asahi-dev · 2023-01-02 11:30
jannauI would expect the actual changes to be rather small on linux sise as well but cleanup and code for backwards compatibility is more complicated
#asahi-dev · 2023-01-02 11:24
jannauas patches in the drm code? the plan on linux side is https://lore.kernel.org/dri-devel/0d188965-d809-81b5-74ce-7d30c49fee2d@redhat.com/ and it is still in the clean-up phase
#asahi-dev · 2023-01-02 11:14
jannaudrm integrated backlight handling is the only upcoming DRM change I want to integrate asap but as far as I'm aware that still needs to be written
#asahi-dev · 2023-01-02 11:11
jannauwe're still based on 6.1 and I wouldn't expect that we lose backwards compatibility with 6.1 drm anytime soon
#asahi-dev · 2023-01-02 10:49
jannaukettenis: please ping me when you start porting the dcp driver. we should avoid squashing changes after that, do you have other suggestions to make porting easier?
#asahi-dev · 2023-01-02 10:42
jannaumarcan: unrelated, do you plan to submit t8110 dart for 6.3? if not I'll prepare a base t8112 device tree / bindings series
#asahi-dev · 2023-01-02 10:40
jannaubut as long as we use just a single plane blending as premutiplied alpha with the black background will work
#asahi-dev · 2023-01-02 10:39
jannauI don't see anything for 8-bit XRGB though
#asahi-dev · 2023-01-02 10:38
jannaufor 10-bit xrgb and many yuv formats we can workaround the missing support by using the corresponding pixelformat with a separate alpha plane and use driver managed buffer of 0xFFs
#asahi-dev · 2023-01-02 10:35
jannaumarcan: makes sense in general
#asahi-dev · 2023-01-01 17:09
jannauand premultiplied is set for the yuv surface
#asahi-dev · 2023-01-01 17:03
jannauenabling the yuv overlay looks a little strange. the UI switches from surface 0 to surface 1, surface 0 is cleared and unk_10c/bg_color in swap rec is set to 0xff000000 (it is usually 0x0)
#asahi-dev · 2023-01-01 16:54
jannaumarcan: either I tested only with the media player or I misremember things. safari with vp9 video from youtube uses '&8v0' tiled 8-bit nv12
#asahi-dev · 2023-01-01 16:45
jannauI haven't seen yuv surfaces with windowed video playback
#asahi-dev · 2023-01-01 16:42
jannauvia dcp
#asahi-dev · 2023-01-01 16:42
jannauthey output yuv surfaces for fullscreen video
#asahi-dev · 2023-01-01 16:40
jannaus/I/A/
#asahi-dev · 2023-01-01 16:40
jannauI pity that macos' window sytem doesn't support overlays. not even the native media player or safari are using underlays or overlays for video
#asahi-dev · 2023-01-01 16:38
jannauthere's A006/A007 set_csc_mode / set_op_mode, A044 set_blendout_CSC_mode, A111 vftg_set_color_channels which are tangentially related
#asahi-dev · 2023-01-01 16:32
jannauI didn't spot anything obvious in the unused Calls. color_remap_mode has 2 bits for invert and luminance only
#asahi-dev · 2023-01-01 15:57
jannauany idea what the '[' in the fourcc means? '&' signifies tiled formats
#asahi-dev · 2023-01-01 15:55
jannaumarcan: thanks. we have two u32 arrays which seem to contain flags for each surface in the swap struct
#asahi-dev · 2023-01-01 13:14
jannauyes, from a m1 max. is suspend with both p clusters in deep WFI?
#asahi-dev · 2023-01-01 12:58
jannaudid you test that on m1 pro / max as well? those have currently an unreasonable idle power use of ~5W, m2 is more reasonable with less than 1W assuming PHPC measures more or less the same things
#asahi-dev · 2023-01-01 10:20
jannaumarcan: possibly although it appears little strange that it is required for overlaying YUV formats. I put it on my todo list for testing. I think it's also worth checking if there's a global alpha field in the surface struct
#asahi-dev · 2022-12-31 10:19
jannauhttps://gist.github.com/jannau/05622a7c59c3517c3c25aaa9cecfdbf3
#asahi-dev · 2022-12-31 10:17
jannaumarcan: dart runtime PM seems (not sure what else in dcp/sleep) to "break" ethernet on the mini
#asahi-dev · 2022-12-30 15:03
jannaucorruption was only visible with xrgb8888 and not xrgb2101010 though
#asahi-dev · 2022-12-30 15:02
jannauclearing surface 0 fixes the apple logo. opaque logic is still correct so opaque is not the correct meaning for that field
#asahi-dev · 2022-12-30 13:06
jannauthe .opaque logic might be broken when starting at surface[1]
#asahi-dev · 2022-12-30 13:05
jannaulol, on the mbp 13" m2 I see a apple logo shaped corruption with kernel boot messages, I think we need to clear surface[0]
#asahi-dev · 2022-12-30 12:50
jannausigh, minimal display backlight after dpms is still there with dcp/sleep
#asahi-dev · 2022-12-30 00:33
jannauI'll test tomorrow on t600x/t8112
#asahi-dev · 2022-12-30 00:32
jannaumarcan: dcp/sleep rebased onto my probing changes + 'power-domains = <&ps_pmp>;' for ps_gfx works on j274
#asahi-dev · 2022-12-28 12:56
jannausee "Battery charge control" in Documentation/admin-guide/laptops/thinkpad-acpi.rst for example
#asahi-dev · 2022-12-26 10:01
jannaufor dcp/disp0 on t8103/t600x as well. vm-size is iirc 0xFC000000 but we start allocating at the top of the 32-bit address space
#asahi-dev · 2022-12-26 09:57
jannauI suspect vm-size in the ADT is just a configured limit. I haven't looked at vm-base. the only device with an non zero vm-base is dcp on m2 and it's probbaly fine to ignore for now since the managed iommu domain allocates iova top down
#asahi-dev · 2022-12-25 23:23
jannauthe component change is good nevertheless and I'll keep it

plus ›