#asahi-dev · 2026-08-05 01:07
Eighth_Doctorthere is a large amount of Android apps built against the old fixed 4k size#asahi-dev · 2026-08-05 01:06
Eighth_DoctorI don't know how broken Android userspace will be with that#asahi-dev · 2026-08-05 01:06
AreedElahi[m]Not that new#asahi-dev · 2026-08-05 01:06
Eighth_Doctorthe biggest challenge is that 16k page support in Android is very new#asahi-dev · 2026-08-05 01:05
AreedElahi[m]Is it worth it as a project?#asahi-dev · 2026-08-05 01:05
AreedElahi[m]Guys, I lowkey wanna boot Android on mac os (m1), I know its possible, will be annoying, but not as hard cause I can compile against Asahi Device tree, then painfully fix and patch crashes yk. #asahi-dev · 2026-08-04 19:54
jannaupushed to asahi-wip. please test on m3 mba (and imac) with https://github.com/AsahiLinux/m1n1/pull/650#asahi-dev · 2026-08-04 19:39
jannauit uses isp-imx364.dtsi which is cfg0#asahi-dev · 2026-08-04 19:38
chaos_princessah. But the point about different isp stands, it will need a cfg0 dtsi#asahi-dev · 2026-08-04 19:37
jannausensor as in to which of the sensor-presets (isp-*.dtsi) to include, not sensor-type#asahi-dev · 2026-08-04 19:33
chaos_princessand also, "sensor-type" is kind of a misleading name, as the j514c and j514s have different sensor-type, despite obviously being the same chassis#asahi-dev · 2026-08-04 19:33
chaos_princessm3 one should have different isp#asahi-dev · 2026-08-04 19:29
jannaum1 and m3 imacs have the same camera spare part number so should use the same sensor#asahi-dev · 2026-08-04 19:24
chaos_princessi think so#asahi-dev · 2026-08-04 19:22
jannaudo I keep 'apple,temporal-filter = <1>;' for the m3 mba?#asahi-dev · 2026-08-04 19:13
jannauyes, I was looking at the j514 adt#asahi-dev · 2026-08-04 19:12
chaos_princessand 16 is j514#asahi-dev · 2026-08-04 19:11
chaos_princess sensor-type = 4 - this is what in my template adt file#asahi-dev · 2026-08-04 19:11
chaos_princessare you sure#asahi-dev · 2026-08-04 19:08
jannauthat doesn't match. j504 has sensor-type = 16 but the macos trace uses platform-id 4 in the boot args#asahi-dev · 2026-08-04 19:06
chaos_princessno, it is sensor-type under the isp node#asahi-dev · 2026-08-04 19:06
chaos_princess10 for imacs, 18, for airs#asahi-dev · 2026-08-04 19:05
jannauand platform-id is board-id in the adt?#asahi-dev · 2026-08-04 19:04
chaos_princesssensor-type field#asahi-dev · 2026-08-04 19:04
chaos_princessits in adt#asahi-dev · 2026-08-04 19:04
jannaudo we know the platform-id and sensors used in m3 imacs and macbook airs?#asahi-dev · 2026-08-04 19:03
jannaucursory comparisson of macos/linux traces looked similar. macos had additional cmds, face detection, motion history and MLVNR#asahi-dev · 2026-08-04 18:57
chaos_princessthere is the rather annoying approach of diffing basically every single command that is being sent#asahi-dev · 2026-08-04 18:56
jannauyes, no complaint and CISP_CMD_CH_SET_FILE_LOAD has at least the same size as the macos trace. one of the ISP_IMX558 files#asahi-dev · 2026-08-04 18:52
chaos_princessand the calibration files are loaded, right?#asahi-dev · 2026-08-04 18:52
jannauit's clearly worse than m2 pro 14-inch mbp. I assume that's the better comparisson due to the same or similar sensor#asahi-dev · 2026-08-04 18:45
chaos_princessso, m2 is the interesting comparison#asahi-dev · 2026-08-04 18:45
chaos_princessalso, i missed it, but yes, we expect to be worse than macos, since we are using cfg0, and that does not do whatever "MLVNR" does (it requires the ane)#asahi-dev · 2026-08-04 18:43
jannauchanging CISP_CMD_CH_LPDP_HS_RECEIVER_TUNING_SET makes no difference#asahi-dev · 2026-08-04 18:19
jannausomething is still not 100% correct: "apple-isp 2a4000000.isp: [isp_stop_command_processor] never received magic number from firmware"#asahi-dev · 2026-08-04 18:17
chaos_princessi guess unk1 is value count, since it is now 2, and there are two values following#asahi-dev · 2026-08-04 18:17
chaos_princessCISP_CMD_CH_LPDP_HS_RECEIVER_TUNING_SET also changed#asahi-dev · 2026-08-04 18:16
jannauinstead of 3#asahi-dev · 2026-08-04 18:16
jannauI have atm in isp-fw.c 'args.unk9 = 0x1;' from the macos trace#asahi-dev · 2026-08-04 18:14
chaos_princesss/pr/branch/#asahi-dev · 2026-08-04 18:14
chaos_princesshave you made any other changes beyond what is in the pr above#asahi-dev · 2026-08-04 18:14
jannauworse than macos. not sure if as bad as without the "firmware". let me check if that's present#asahi-dev · 2026-08-04 18:08
chaos_princessmuch worse than m2?#asahi-dev · 2026-08-04 18:08
chaos_princesshow noisy?#asahi-dev · 2026-08-04 18:08
jannaupicture looks noisy though#asahi-dev · 2026-08-04 18:07
jannauworks, was a power-state ordering issue leading me to drop some of the power-states#asahi-dev · 2026-08-04 17:22
jannauno difference#asahi-dev · 2026-08-04 17:22
jannauI just noticed that t6030 uses ps_ispsens0, let me try that#asahi-dev · 2026-08-04 17:21
jannauthere is only isp_sys, and ispsens[0-3]#asahi-dev · 2026-08-04 17:16
chaos_princessand you may want to monitor the entire unknown pmgr area, as there is stuff below the 0x4000 mark sometimes