#
✕ nick:jannau
#asahi-dev · 2026-11-17 18:48
jannau"Couldn't sanitize GL_RENDERER AGX G13/G14" is expected, "dmabuf image neither requires nor prefers a dedicated allocation" doesn't sound fatal eitherand I can't find references to it in web search
#asahi-dev · 2026-11-17 16:22
jannauthe main point of using the VM on physical storage is that you can work on the same source when booted Linux and macOS. It's probably the easiest way to install Linux on M4 devices right now and verify that it boots
#asahi-dev · 2026-11-17 15:52
jannaunot really possible for other devices
#asahi-dev · 2026-11-17 15:52
jannaufor usb the addresses haven't changed (much) so we could probably just hardcode
#asahi-dev · 2026-11-17 15:50
jannaumatching addresses by iterating over compatible FDT node was my plan as well
#asahi-dev · 2026-11-17 15:50
jannauThis is a little complicated. If you're confident that you'll manage to set the Linux VM up, we can discuss how to build and boot Linux kernel via m1n1 on the secondary macos
#asahi-dev · 2026-11-17 15:48
jannauOvertone[m]: my idea would be to create a secondary macos install and partition your ssd to create a linux VM in macos which uses SSD partitions as storage. The linux partition when could be used from the VM and from booting linux via m1n1 on the secondary macos.
#asahi-dev · 2026-11-17 15:40
jannauwe don't even have matching addresses? I've factored so far the ADT address collection from dt_disable_missing_devs() and written fdt_node_offset_by_compatible_suffix() so we can iterate over nodes which are compatible to "-atcphy$"
#asahi-dev · 2026-11-17 15:36
jannauatcphy is annoying
#asahi-dev · 2026-11-17 15:36
jannaufixing touchbar and atcphy atm. touchbar was ok. already kinda hardcoded to the only 2 devices
#asahi-dev · 2026-11-17 15:34
jannauthe aliases were a mayor help matching ADT devices
#asahi-dev · 2026-11-17 15:32
jannauOvertone[m]: do you have access second non-appple silicon computer? without it development will be annoying. I believe M4 support is at a state where it is possible do something but it'll be hard to learn from when something does not work
#asahi-dev · 2026-11-17 13:38
jannauyou're probably right. I didn't see his email to remove the aliases
#asahi-dev · 2026-11-17 13:05
jannau"'usbdrd_phy' nodes should have numbered alias in the aliases node, in the form of usbdrdphyN, N = 0, 1... (depending on number of controllers).:"
#asahi-dev · 2026-11-17 13:05
jannausee Documentation/devicetree/bindings/phy/samsung,usb3-drd-phy.yaml for example
#asahi-dev · 2026-11-17 13:04
jannauI think we just have to mention the aliases in the dt-bindings
#asahi-dev · 2026-11-17 13:03
jannaumaybe "Do not add instance index (IDs) properties or custom OF aliases" in writing-bindings.rst but it's not clear from the context
#asahi-dev · 2026-11-17 12:57
jannauhuh, I thought aliases were a free form. neither the documentation or the schema yaml mentions any restrictions on entries in /aliases
#asahi-dev · 2026-11-17 12:38
jannauso we're just missing to mark the usb-c ports with "wakeup-source" and fix the irq driver?
#asahi-dev · 2026-11-17 12:04
jannauI think you can only use stacked PRs if the branches are in the target repo
#asahi-dev · 2026-11-17 09:33
jannauI was looking for something which verifies the results by comparing it with the result of the same operation in software
#asahi-dev · 2026-11-17 09:11
jannausofus: I haven't tested yet. it looked obvious on msr side so I assumed using 3 MSR_STORAGE_1C8B8 planes would work and setting the 3rd plane up would work
#asahi-dev · 2026-11-16 21:49
jannausofus: have you found a test suite to tests if scaler/msr does roughly the right thing? mostly interested in pixel format conversion. I'm adding 3 plane yuv formats
#asahi-dev · 2026-11-16 18:34
jannauyes, that will need pcie tunnelling support for usb3
#asahi-dev · 2026-11-16 18:33
jannaudanield75: I think you want at least change dcpext1's `mux-controls = <&atcphy1_xbar 0>;` to `<&atcphy1_xbar 1>`
#asahi-dev · 2026-11-16 18:29
jannaudoesn't look like that's included in the branch
#asahi-dev · 2026-11-16 18:25
jannaudanield75: what dock do you have? a thunderbolt3 will not have usb3 tunneling. I suspect new designs after usb4 was available are more likely to use usb3 tunnelling (saves a part)
#asahi-dev · 2026-11-16 13:08
jannauand now I can finally tag m1n1 with M3* dcp support
#asahi-dev · 2026-11-16 13:07
jannauI've tagged asahi-7.3.4-1 and pushed it to github
#asahi-dev · 2026-11-15 22:11
jannausofus: is there a reason why you use VIDEOBUF2_DMA_CONTIG in msr (and avd)? We have an iommu so VIDEOBUF2_DMA_SG should work and is preferable
#asahi-dev · 2026-11-15 22:09
jannausame way
#asahi-dev · 2026-11-15 21:52
jannaupremultiplied alpha is busted with overlays as well (on M3/14.8.3)
#asahi-dev · 2026-11-15 21:51
jannaualso please add a discoverable flag to mark layers explictly as coverage which ignores the pixel's alpha value
#asahi-dev · 2026-11-15 19:59
jannauyes, seems wrong. I don't think it causes issue but it'll miss perf otimization
#asahi-dev · 2026-11-15 19:48
jannauno, you need scatterlist
#asahi-dev · 2026-11-15 19:38
jannauchaos_princess: https://lore.kernel.org/rust-for-linux/20261112-b4-iommu-domain-v1-0-c29e86788dda@collabora.com/T/#me6fcc92b80c8a4302052af2ef785ef154d7d6daf should solve the iommu domain bindings issue
#asahi-dev · 2026-11-15 19:36
jannauaffects PMPv2 as well and is one of the reasons why it's not yet enabled
#asahi-dev · 2026-11-15 19:35
jannaunew DTS breaks kernel without PMP since the power-domains can't be turned on
#asahi-dev · 2026-11-15 19:33
jannauit's probably both for non-upstream nodes
#asahi-dev · 2026-11-15 19:32
jannauDART's dt-bindings allow only "apple,t8112-dart"
#asahi-dev · 2026-11-15 19:32
jannautryhard[m]: not recently. main issue I have with PMP is the devicetree compatibility
#asahi-dev · 2026-11-15 19:30
jannau'compatible = "apple,t8110-dart";' is correct for M2 t8112
#asahi-dev · 2026-11-15 18:26
jannaudoesn't seem to be available anymore
#asahi-dev · 2026-11-15 18:25
jannauleio: hopefully for you amusement: "Dell U3224KB HDMI 6144x3456 clock reduction override" in the 14.8.3 DCP firmware
#asahi-dev · 2026-11-15 18:22
jannausee https://github.com/AsahiLinux/m1n1/blob/main/proxyclient/m1n1/hw/scaler.py and https://github.com/AsahiLinux/m1n1/blob/main/proxyclient/experiments/scaler.py
#asahi-dev · 2026-11-15 15:13
jannausingle full framebuffer layer doesn't appear to do blending. so that explains why XRGB works show_notch=1
#asahi-dev · 2026-11-15 14:27
jannauyes, it's identical if I use surface 1,2 instead of 0,1
#asahi-dev · 2026-11-15 13:55
jannauit is not, surface 0 has a noticeable different color but iirc we already arrived at the conclusion that surface 0 has special color space conversion properties (maybe just something set by iboot)
#asahi-dev · 2026-11-15 13:08
jannau14.x or M3 specific
#asahi-dev · 2026-11-15 12:59
jannauthis is j514s/t6030 with 14.8.3, I don't see how this could be display related. dcp produces everything the display gets. I'll test on t6020 but I'd be surprised if it shows the same behavior. at least the single premultiplied surface over opaque bg works as expected

plus ›