#
✕ nick:chaos_princess
#asahi-dev · 2026-03-26 16:59
chaos_princessdisp driver can't re-init from pmp shutdown
#asahi-dev · 2026-03-26 13:24
chaos_princessignore all the pmp_ptd bits, those are v2-only
#asahi-dev · 2026-03-26 13:23
chaos_princesshttps://github.com/AsahiLinux/m1n1/pull/545/files - here is the pmp v2 tracer, it is different, but the basic ideas are the same
#asahi-dev · 2026-03-26 13:22
chaos_princessno, that is arm64 msrs, pmp is attached over mmio
#asahi-dev · 2026-03-26 13:20
chaos_princessmarcan's old streams can also be used as a user guide to m1n1 tracing.
#asahi-dev · 2026-03-26 13:19
chaos_princesslook in the adt for the "reg" properties of the pmp node
#asahi-dev · 2026-03-26 13:15
chaos_princessit runs macos in a hypervisor and traces all memory read/writes to a specified area
#asahi-dev · 2026-03-26 13:12
chaos_princessonce you understand enough, you go and either write a python driver in m1n1 and see if it works, or jump straight to writing a linux driver
#asahi-dev · 2026-03-26 13:12
chaos_princessbased on that you edit the tracer to print out more info based on what you figured out
#asahi-dev · 2026-03-26 13:11
chaos_princessyou run macos using run_guest.py with the tracer active and see what it does
#asahi-dev · 2026-03-26 13:10
chaos_princess*harmless
#asahi-dev · 2026-03-26 13:10
chaos_princessyea, it is harmelss
#asahi-dev · 2026-03-26 13:06
chaos_princesshttps://github.com/AsahiLinux/m1n1/blob/main/proxyclient/m1n1/fw/pmp.py
#asahi-dev · 2026-03-26 13:05
chaos_princessand i am saying that yes, it most likely didn't but i am not sure if what exists as a m1n1 tracer answers all the questions that need answering
#asahi-dev · 2026-03-26 13:03
chaos_princessthe question is not if it has changed, but if it is enough for a driver
#asahi-dev · 2026-03-26 13:01
chaos_princessi will probably knock it out at some point in the future as marcan already did some RE work forever ago, just need to confirm it and write the linux drivers
#asahi-dev · 2026-03-26 12:58
chaos_princessbattery stuff is currently adding support for pmp v2, and vanilla m1 uses pmp v1, so there is some work there
#asahi-dev · 2026-03-26 12:56
chaos_princesswhat hw do you have
#asahi-dev · 2026-03-26 12:54
chaos_princesssep, as a big chunk of battery management is done and coming in not very distant future
#asahi-dev · 2026-03-25 08:04
chaos_princesshold option in boot picker
#asahi-dev · 2026-03-22 17:24
chaos_princessno rush ofc, just when time permits as this is something i can't confirm myself
#asahi-dev · 2026-03-22 17:20
chaos_princesswould also appreciate a confirm/deny that the new afnc and table implementations work
#asahi-dev · 2026-03-22 17:19
chaos_princessyes, please
#asahi-dev · 2026-03-22 15:23
chaos_princessyea, it was a good time back then, but we lost a lot of time due to misreading the M as measurement :P
#asahi-dev · 2026-03-22 15:22
chaos_princessah
#asahi-dev · 2026-03-22 15:16
chaos_princessjannau: i don't think i follow the implication. is it that it was being kept on so it does not write to pmp sram and crash as that was not powered?
#asahi-dev · 2026-03-22 14:25
chaos_princesskind of a problem for what i am doing since macos very much does turn off at least dcpexts via pmp
#asahi-dev · 2026-03-22 14:07
chaos_princesswhat was the reason for not fully powering down dcps when going to sleep? some missing initialization that they are losing?
#asahi-dev · 2026-03-22 09:39
chaos_princessno need
#asahi-dev · 2026-03-22 09:35
chaos_princessbtw, when you tested on m1max with val |= 0x60000000 hack, what kernel version was it?
#asahi-dev · 2026-03-22 09:10
chaos_princessare we using one of the dcpexts to drive the internal display on m1pro/max ?\
#asahi-dev · 2026-03-21 21:33
chaos_princessuse macos 14.8.3 as base
#asahi-dev · 2026-03-21 21:33
chaos_princesshttps://asahilinux.org/docs/sw/m1n1-user-guide/
#asahi-dev · 2026-03-21 21:30
chaos_princessyou do know how to install m1n1 manually, right?
#asahi-dev · 2026-03-21 21:28
chaos_princessthose are pretty similar, so use m1pro/max or m2pro/max as reference
#asahi-dev · 2026-03-21 21:28
chaos_princessand your specific board
#asahi-dev · 2026-03-21 21:28
chaos_princessactually, you will have to adapt it to m3pro first
#asahi-dev · 2026-03-21 21:27
chaos_princesscherry-pick that on top of asahi-wip
#asahi-dev · 2026-03-21 21:24
chaos_princesshttps://github.com/AsahiLinux/linux/pull/446 - device trees from here
#asahi-dev · 2026-03-21 21:24
chaos_princessoh, you only added the bindings there, right
#asahi-dev · 2026-03-21 21:22
chaos_princesscompile the latest m1n1, the kernel from asahi-wip and try
#asahi-dev · 2026-03-21 13:47
chaos_princessshould work with smaller one too
#asahi-dev · 2026-03-21 13:33
chaos_princessnow it crashes without always-on, and with - exiting sleep fails until killed by watchdog
#asahi-dev · 2026-03-21 13:33
chaos_princessthat activates fabric links that are not there on m1pro
#asahi-dev · 2026-03-21 13:22
chaos_princesswill fix that in the next PR. after the rebase i now also get pmp crashes on m1pro and not 100% sure why yet
#asahi-dev · 2026-03-21 13:20
chaos_princesspio-ranges is the iova-table
#asahi-dev · 2026-03-21 13:20
chaos_princessbecause there weren't any
#asahi-dev · 2026-03-21 12:59
chaos_princessanything it is mapping in the darts aside from the the iokit communication bits?
#asahi-dev · 2026-03-21 12:45
chaos_princesssince there are too few reg entries in the adt
#asahi-dev · 2026-03-21 12:44
chaos_princessone thing that is most likely wrong for m2 is the "iova table"

plus ›