#
#asahi-dev · 2026-02-21 03:27
linuxusere
#asahi-dev · 2026-02-21 03:27
curl_alx_shm3 when?
#asahi-dev · 2026-02-21 03:27
jni am now
#asahi-dev · 2026-02-21 03:26
linuxuserjn when
#asahi-dev · 2026-02-21 03:26
jnhuh?
#asahi-dev · 2026-02-21 03:25
linuxuserjn
#asahi-dev · 2026-02-21 03:25
linuxusere
#asahi-dev · 2026-02-21 03:24
linuxuser21
#asahi-dev · 2026-02-21 03:23
tpw_rulesare you bad at irc or trying to waste time?
#asahi-dev · 2026-02-21 03:22
curl_alx_shm3 when?
#asahi-dev · 2026-02-21 03:19
linuxusercurl https://alx.sh | sh
#asahi-dev · 2026-02-21 03:16
curl_alx_shm3 when?
#asahi-dev · 2026-02-21 03:09
curl_alx_shm3 when?
#asahi-dev · 2026-02-21 02:39
chadmedthings only seem to break when trying to swap >2 surfaces
#asahi-dev · 2026-02-21 02:38
chadmedif i prevent dcp from clearing null surfaces, they can stay onscreen just fine while other surfaces are swapped out
#asahi-dev · 2026-02-21 02:38
chadmedok yeah i think i might be right
#asahi-dev · 2026-02-21 01:08
chadmedunless kwin is updating every plane every time
#asahi-dev · 2026-02-21 01:03
chadmeds/3/2
#asahi-dev · 2026-02-21 01:03
chadmedi get the feeling this is supposed to work from looking at traces. when a video is playing the cursor is moved to surface 3, the video is on surface 1 and the desktop on surface 0 is simply not swapped
#asahi-dev · 2026-02-21 01:02
chadmedyeah theres something else going on here. even if we skip enabling swap on surfaces that did not change between swaps, one surface always becomes weird
#asahi-dev · 2026-02-21 00:31
chadmedwhenever a surface glitches out, the plane loses the fb reference
#asahi-dev · 2026-02-21 00:30
chadmedi feel like our multi surface swap woes are not entirely dcp related
#asahi-dev · 2026-02-21 00:06
nicolas17and the string "K59 or J59 detected, setting %s" likely refers to Apple's Cinema Displays
#asahi-dev · 2026-02-21 00:05
chadmedglad theyre dealing with it for us :P
#asahi-dev · 2026-02-21 00:04
nicolas17so I bet there is indeed a table of quirks :D
#asahi-dev · 2026-02-21 00:04
nicolas17chadmed: according to blacktop's string diffs, the DCP firmware contains the string "DELL U3224KB" since macOS 15.4
#asahi-dev · 2026-02-20 23:24
chadmedi wonder if they have a giant table of quirks in the firmware blob
#asahi-dev · 2026-02-20 23:14
chaos_princessso if apple broke your monitor with new dcp firmware - the reason it works is because we are using old one
#asahi-dev · 2026-02-20 23:13
chaos_princessi think the reason it works is because asahi is pretending to be macos 13.5
#asahi-dev · 2026-02-20 23:12
chaos_princesshahaha
#asahi-dev · 2026-02-20 23:12
andi1[m]👍
#asahi-dev · 2026-02-20 23:11
andi1[m]great work
#asahi-dev · 2026-02-20 23:11
andi1[m]chaos_princess: found it. updated to fairydust to notice I can get a display with USB-C type connector to work which does not work with macOS from Ventura I think. It needed some USB-C ->HDMI->USB-C adapter chain
#asahi-dev · 2026-02-20 22:16
jannaumemory map comes mostly from the ranges node in the ADT and limiting sizes for pcie/usb
#asahi-dev · 2026-02-20 22:14
jannauI haven't yet added the new "apple,t8103-*" base compatibles to u-boot so those have to be added as well
#asahi-dev · 2026-02-20 22:14
chaos_princessandi1[m]: the dtbs are inside m1n1.bin on your ESP. so back it up before running update-m1n1 (it also does a backup, but another one can't hurt). And if you completely break that file, macos is capable of mounting fat32
#asahi-dev · 2026-02-20 22:14
fl0_idsiirko[m] it is not using. the DT. but afaik the memory layout gets output during m1n1 boot log
#asahi-dev · 2026-02-20 22:13
jannausiirko[m]: not that demanding. I think it's mostly adding new compatibles and a new mem-layout in arch/arm/mach-apple/board.c since m3's physical mem also starts at 1 << 40
#asahi-dev · 2026-02-20 22:13
siirko[m]* doing things with the device tree.
#asahi-dev · 2026-02-20 22:13
siirko[m]fl0_id: I will try to do that, but I have the impression that it hangs before doing things device tree. I’m trying to reads more about u-boot for understanding how can i toy with it for debugging
#asahi-dev · 2026-02-20 22:12
fl0_id(afaik)
#asahi-dev · 2026-02-20 22:11
fl0_idsiirko[m] you maybe can get the right values from DT or m1n1 boot log
#asahi-dev · 2026-02-20 22:11
fl0_idabrender that seems surprisingly civil and constructive
#asahi-dev · 2026-02-20 22:09
andi1[m]ok, I check if I understand that correctly. I am used to kernel development on arm (upstreamed a dozen drivers) but do not know the special stuff about the boot process in asahi well
#asahi-dev · 2026-02-20 22:08
chaos_princesstry adding the memory map for your device into board.c and see how it goes
#asahi-dev · 2026-02-20 22:05
siirko[m]* demanding work to do ?
#asahi-dev · 2026-02-20 22:04
chaos_princessback up m1n1.bin on the ESP. if the new device trees are so bad that they completely break the old kernel, you can use macos to restore it
#asahi-dev · 2026-02-20 22:04
siirko[m]jannau: Alright, do you think it is a demanding work ?
#asahi-dev · 2026-02-20 22:01
andi1[m]How do I test a newer kernel requiring devicetree changes with still being able to just boot the know working one afterwards without any problems by just selecting the known-working one in grub?
#asahi-dev · 2026-02-20 21:37
abrenderFor anyone interested in the context of the past discussions: https://lore.kernel.org/lkml/5f741a4f-f37d-079b-d464-59045ebef1ce@marcan.st/t/#u

plus ›