#asahi-dev · 2023-09-03 16:46
jannausternez: do not use that branch, it's know broken and requires different m1n1. it is just old test branch which was used for GPU tests which "required" a display#asahi-dev · 2023-09-03 16:40
jannaut6002 and t8112 appear ok. no noticable perf change on kcbench#asahi-dev · 2023-09-03 16:31
jannaukcbench is still at the value I remember from m1 ultra bringup which should predate the the bug#asahi-dev · 2023-09-03 14:22
jannauyou should be able to boot linux now, probably needs pd_ignore_unused to avoid turning the display out off#asahi-dev · 2023-09-03 14:20
jannauthanks for testing, pushed the power on change to the PR#asahi-dev · 2023-09-03 14:05
jannauthat looks truncated, which hdmi capture device is this? a ms2109/ms2130 one?#asahi-dev · 2023-09-03 14:00
jannauthat sounds like m1n1 is unlucky in selecting a color mode #asahi-dev · 2023-09-03 13:57
jannaudoes the hdmi capture work with macos?#asahi-dev · 2023-09-03 13:55
jannauthat looks like it worked. did it?#asahi-dev · 2023-09-03 13:54
jannauedit hv/trace_dcp.py and set dcp_name to "dcpext4"#asahi-dev · 2023-09-03 13:53
jannaucan you run m1n1 under the hypervisor `tools/run_guest.py -m hv/trace_dcp.py -m hv/trace_dptx.py -l dcp_dptx.log -r ../build/m1n1.bin`#asahi-dev · 2023-09-03 13:50
jannaubut no picture?#asahi-dev · 2023-09-03 13:31
jannausternez: please try to add 'pmgr_adt_power_enable(cfg->dptx_phy);' next to the other one in dcp_init()#asahi-dev · 2023-09-03 06:46
jannauplease test https://github.com/AsahiLinux/m1n1/pull/329 - HDMI output support for m2+ desktops, requires proxy mode to power the dp2hdmi converter on#asahi-dev · 2023-09-02 21:01
jannauso the actual mapped memory in dart-isp is from the top of the memory#asahi-dev · 2023-09-02 20:29
jannauno, when m1n1 is running in proxy mode#asahi-dev · 2023-09-02 20:27
jannaudo you see it in the output of `experiments/dart_dump.py dart-isp0`?#asahi-dev · 2023-09-02 20:25
jannau0x10000a58000 is low considering that the memory is mapped to 0x100_0000_0000. I would expect that there's memory-region with the same address, size in the adt fetched from the device#asahi-dev · 2023-09-02 20:22
jannauyes, for the iommu/iop 0x0 is ok#asahi-dev · 2023-09-02 20:20
jannauah-: which address seems low to you? "iova: 0x0"?#asahi-dev · 2023-09-02 14:22
jannaua possible solution for nodes which exists only once could be an adt_path_offset that matches the prefix and returns the device if the match is unambigguous#asahi-dev · 2023-09-02 14:18
jannauwhat do you mean by issue? yes, adt nodes have number suffixes on on t600x/t602x. yes, it is annoying but I don't think that patching the adt will be an improvement#asahi-dev · 2023-09-02 13:46
jannausternez: in the case you read this, you will have to change the write to "0x39c000010 (lpdptx-phy0[1]" at the beginning from 0x0 to 0x1#asahi-dev · 2023-09-02 13:44
jannauand m2 pro mac mini has a picture on hdmi as well#asahi-dev · 2023-09-02 05:47
jannausternez: indeed, they use die1 on the m2 ultra mac studio, sigh. you need to use dcpext4/dispext4#asahi-dev · 2023-09-01 18:37
jannauthe 0x1_860E_8000 from the adt is offsetted by 0x2_0000_0000 through the ranges property#asahi-dev · 2023-09-01 18:36
jannaudart-isp0 should be at 0x3_860E_8000 for both t6000 and t6001#asahi-dev · 2023-09-01 18:20
jannaua full kernel (re-)build takes a few minutes#asahi-dev · 2023-09-01 18:16
jannauwhat does file report on it?#asahi-dev · 2023-09-01 18:15
jannau55k zstd compressed / 250k uncompressed here#asahi-dev · 2023-09-01 18:14
jannauand than report how large that it#asahi-dev · 2023-09-01 18:14
jannaufind /lib/modules/$KRNEL_VER -name 'apple_dcp.ko*'#asahi-dev · 2023-09-01 18:13
jannauare that maybe leftovers because kernel module compression method was changed#asahi-dev · 2023-09-01 18:01
jannauthat is a build with debug symbols#asahi-dev · 2023-09-01 17:38
jannaumy vmlinuz (uncompressed) and initrd (gzipped) are each less than 25M#asahi-dev · 2023-09-01 17:35
jannaumy modules (`du -sch /lib/modules/6.4.0-asahi-9-arm64`) are 34M, less modules than a distro built but not really disabled everything I don't need#asahi-dev · 2023-09-01 17:34
jannauhow large are you installed modules (or vmlinuz / Image.gz in the case you've built everything in)#asahi-dev · 2023-09-01 17:33
jannauno, you can't. the bootloader has no idea that it has to mount a second partition nor how to mount it so that your symlink resolves#asahi-dev · 2023-09-01 16:31
jannauset that to DEBUG_INFO_NONE#asahi-dev · 2023-09-01 16:28
jannauknedlik: build with out (full) debug symbols#asahi-dev · 2023-09-01 07:04
jannauor you have to modify the installer_data.json and boot first via m1n1 proxy#asahi-dev · 2023-09-01 07:03
jannauI think you'll have to use the fedora installer to get a minimal system for bootstrap installed#asahi-dev · 2023-09-01 06:51
jannaumarcan: dottedmag but it didn't work out and the yak got unclaimed: https://github.com/AsahiLinux/docs/wiki/Yaks-in-need-of-shaving/_compare/c9a916b3fded7be3e4a39e22e35ecd20503e4293...15b5c09d55aa2ac07a913ca5e9ddb6a5b2f88771#asahi-dev · 2023-08-31 06:19
jannauMac Studio (2023) device trees are up as well#asahi-dev · 2023-08-31 04:47
jannauchadmed: no, the 15" air dts is already tested, merged and released#asahi-dev · 2023-08-30 16:16
jannaufedora release dead line got extended. there will be no release this month#asahi-dev · 2023-08-30 16:14
jannauthat sounds outlandish. if used for video would be over 3 seconds of video#asahi-dev · 2023-08-30 08:46
jannaumaybe similar to https://bugs.chromium.org/p/chromium/issues/detail?id=1145581#c60#asahi-dev · 2023-08-30 08:45
jannauhow would chromium on fedora end up to with sigill (iff bti is enabled) in __libc_start_main@plt ? backtrace is not not useful, only frame below is _start with 'bti c' but the function call is a bl, unfortunately it branches to different address than I end up at#asahi-dev · 2023-08-29 12:02
jannauthere's a work-around. I think it was `bootefi bootmgr` but it could be something else