#asahi-dev · 2026-06-26 07:14
jannauthe cross file != cross toolchain. the file tells meson what's build#asahi-dev · 2026-06-26 07:10
chadmedclang/llvm shouldnt need a cross toolchain, we just pass -target and -mcpu#asahi-dev · 2026-06-26 07:08
jannaure meson, I think we can get rid of the llvm cross file and just add one for gcc / binutils builds#asahi-dev · 2026-06-26 07:06
jannauthe faults happen during the reset and dart's irq handler tries to read from mmio which fails due to the reset#asahi-dev · 2026-06-26 07:04
integralpiloti can't manage to reproduce your error sorry, if i put a completetly bogus reg then i get a SError during probe, not only when "cat" called#asahi-dev · 2026-06-26 07:03
jannausofus: yes, I tested and it works as expected. the 2 avd instances are completly separate#asahi-dev · 2026-06-26 07:03
sofusthats true, not sure why i thought they would be connected somehow#asahi-dev · 2026-06-26 06:57
chadmedsofus: i would be concerned if they couldnt run simultaneously, they should be completely independent of each other #asahi-dev · 2026-06-26 06:56
chaos_princessyuka: iova table turned out to only be a thing on t600x, everything else sets it to null address#asahi-dev · 2026-06-26 06:56
integralpilotit might be something wrong with the driver then, i did only code it up very quickly just to test echo/cat didn't error out#asahi-dev · 2026-06-26 06:55
integralpilotyuka: that seems reasonable, strange then its not working#asahi-dev · 2026-06-26 06:53
sofusand thanks for the pull, that makes it a lot easier #asahi-dev · 2026-06-26 06:50
sofusi wonder if both avd instances can be run at the same time, did you test that (on m2 ultra)?#asahi-dev · 2026-06-26 06:47
sofusjannau: i have only seen faults on stream 0, this only happens during resets?#asahi-dev · 2026-06-26 06:12
yukaI autocompleted this to 0x38810c000, 0x388128000, 0x38812c000#asahi-dev · 2026-06-26 06:11
yukayour device tree has 0x2e410c000, 0x2e4128000, 0x2e412c000#asahi-dev · 2026-06-26 06:10
yukaintegralpilot: from my adt I had the following regs to go from (0x388128000, 0x10000), (0x38810c000, 0x1000)#asahi-dev · 2026-06-26 06:07
yukadon't have time to debug it now, just FYI or if anyone has had the same and has a solution :)#asahi-dev · 2026-06-26 06:05
yukaTried PMPTracer on t6034, it's crashing somewhere in ASCTracer: https://x0.at/3gyR.txt#asahi-dev · 2026-06-26 04:39
diesller[m]i own an m4 air#asahi-dev · 2026-06-26 04:39
diesller[m]hello, i really want to help but i don't know where to start from. i don't have much coding experience but can i still try to contribute?#asahi-dev · 2026-06-26 01:34
integralpilothow did you get the regs?#asahi-dev · 2026-06-26 01:30
integralpilotit looks to me like the regs might be wrong or for some weird reason on m4 it might need some power domains?#asahi-dev · 2026-06-26 01:27
integralpilotin the meantime ill try and see if i can reproduce it#asahi-dev · 2026-06-26 01:27
integralpilotyuka: huh, can you try writing (echo "hello" > /dev/ttydc0) and see if that crashes as well? would be interesting to see if its just an issue with reads on happens on both#asahi-dev · 2026-06-25 22:51
chadmedi had a shell script with a bunch of for tier in {...} nested in a case statement #asahi-dev · 2026-06-25 22:51
chadmedah thats a better solution than what i was doing :P #asahi-dev · 2026-06-25 22:50
jannauchadmed: https://github.com/sofus13/avd-fw/pull/1 to simplify avd testing on multiple targets#asahi-dev · 2026-06-25 21:25
enverbalalicok yeah i can't test that, if i linux my gf's air i might end up in a ditch somewhere#asahi-dev · 2026-06-25 21:25
enverbalalictuxvdmtool also requires M series on both ends, just booting asahi itself ?#asahi-dev · 2026-06-25 21:17
yuka(possible that I got the regs wrong)#asahi-dev · 2026-06-25 21:17
yukahttps://x0.at/Mj7e.txt#asahi-dev · 2026-06-25 21:16
yukaAnd crash#asahi-dev · 2026-06-25 21:13
yukafor... i don't know what?#asahi-dev · 2026-06-25 21:13
yukaoh, it just takes 20 seconds#asahi-dev · 2026-06-25 21:13
yukaintegralpilot: unfortunately my j773g just gets stuck in boot (with maxcpus=1) shortly after "apple-dockchannel-tty soc:serial: registered ttydc0"#asahi-dev · 2026-06-25 21:08
yukaYes#asahi-dev · 2026-06-25 21:07
mazhopefully the dockchannel uart can replace all that shortly, and we can retire the CSs.#asahi-dev · 2026-06-25 21:07
flokliunfortunately it's buried in months of matrix history that's unsearchable.#asahi-dev · 2026-06-25 21:07
flokliyes, there were various error messages in macvdmtool / tuxvdmtool.#asahi-dev · 2026-06-25 21:03
yuka*debuguarts#asahi-dev · 2026-06-25 21:02
yukaThat it actually gived a vdm error reply,... yes, odd. Did flokli also have that? There is definitely weirdness with the uart hardware itself, like not all macs can do 1,5M baudrate between their debugusbs.#asahi-dev · 2026-06-25 21:02
mazI nicked it almost verbatim from the Arduino script, which itself is a transcription of macvdmtool.#asahi-dev · 2026-06-25 21:01
mazit is rather odd, given that this is mostly the same code...#asahi-dev · 2026-06-25 21:01
yukaAha!#asahi-dev · 2026-06-25 21:00
flokliI never got it to work with another mac and non-debugusb, only with the CS.#asahi-dev · 2026-06-25 20:59
yukaI remember he had lots of issues with the serial too#asahi-dev · 2026-06-25 20:59
yukaOr only using the central scrutinizer#asahi-dev · 2026-06-25 20:59
yukaDid flokli get the M4 Pro Mac Mini serial working with another mac / macvdmtool on the other side?#asahi-dev · 2026-06-25 20:52
yukaI remember doing a lot of roudtrips through 1TR since no way to chainload until I fixed the pmgr and had working usb, so I guess serial was not working