#
✕ nick:sven
#asahi-re · 2025-04-03 16:55
sventime to RE iboot then I guess :(
#asahi-re · 2025-02-23 11:37
svenbut that's why I'd start with a python thing that drives it first to understand what exactly we need to do to drive it
#asahi-re · 2025-02-23 11:37
sventhe only part where that may or may not work is full disk encryption since that needs a channel between our nvme driver and SEP and we might have to do that entirely inside the kernel
#asahi-re · 2025-02-23 11:36
sventhen think about how to properly implement this. from what i saw a few years ago when I looked at it a bit what we probably want is a dumb kernel driver that allows to send/recv messages to those SEP endpoints and do most of the heavy lifting in a userspace rust daemon
#asahi-re · 2025-02-23 11:35
svenif I were to reverse SEP I'd probably do it similarly to how lina did agx: first write a python thing that can drive it inside m1n1 to figure out how it actually works
#asahi-re · 2025-02-23 09:50
svenits contiguous and doesn’t change size
#asahi-re · 2025-02-23 09:49
svenyou don’t need a apfs driver for the rest. just figure out the offsets if the gigalocker file on that xart partition and do raw read/writes
#asahi-re · 2024-08-08 08:33
svenmake sure to read https://alx.sh/re "Reverse engineering policy" as well
#asahi-re · 2024-08-08 08:32
svenno idea, i'd be surprised if you actually need to look at the SEP firmware though
#asahi-re · 2024-08-08 08:28
svenand then we can discuss how this entire thing should look like in production
#asahi-re · 2024-08-08 08:28
sventhen prototype the client in python using m1n1
#asahi-re · 2024-08-08 08:28
svenfirst step is probably to improve the sep tracer to be more useful
#asahi-re · 2024-08-08 08:26
svenhrm, no, there should be something more
#asahi-re · 2024-08-08 08:25
svenhttps://github.com/AsahiLinux/m1n1/blob/main/proxyclient/m1n1/hw/sep.py i think
#asahi-re · 2024-08-08 08:25
sventhere's a very basic tracer somewhere in m1n1
#asahi-re · 2024-08-08 08:25
svenyeah, something like that
#asahi-re · 2024-08-08 08:20
svena sep client for things like full disk encryption and touchid support would be great, but signing the kernel only works from recovery mode
#asahi-re · 2024-07-23 18:40
svenstill hope I never have to look at it
#asahi-re · 2024-07-23 18:40
svenhuh, sweet
#asahi-re · 2024-04-07 20:52
svenand yeah, you should be able to see everything with the hv
#asahi-re · 2024-04-07 20:52
svenit’s just the first layer which happens to be fairly simple. still lots of work to be done
#asahi-re · 2024-04-07 20:35
svenvery likely even the same data the Bluetooth keyboards with touchid will send/receive
#asahi-re · 2024-04-07 20:34
svenI’d expect that it’s essentially just passing encrypted data between the sensor and sep
#asahi-re · 2024-04-07 20:34
svenTouch ID enrollment will already require the sep
#asahi-re · 2024-04-07 20:33
svenand some code to boot the sep
#asahi-re · 2024-04-07 20:33
sventhere’s a tracer in m1n1 that should decode the first layer ipc
#asahi-re · 2024-04-07 20:32
svendunno if amarioguy made any progress
#asahi-re · 2024-04-07 20:32
svenI did a bit of work a long time ago
#asahi-re · 2024-02-13 18:49
sventhat's how all good reverse engineering projects start!
#asahi-re · 2024-01-20 15:33
svenroxfan: ^--
#asahi-re · 2024-01-20 15:33
svenmaybe it's that?
#asahi-re · 2024-01-20 15:33
sventhere's "mild_dsb_enabled" in there
#asahi-re · 2024-01-20 15:32
svenit might from that weird XNU source drop where they included llvm bitcode instead of compiled object files
#asahi-re · 2024-01-20 14:58
svenwe probably only know the name
#asahi-re · 2024-01-20 14:58
svenlooks like something that was left over in some xnu or llvm strings or source code
#asahi-re · 2023-12-27 20:15
svenlike the second half of the Mailbox regs for each co processor? :p
#asahi-re · 2023-12-10 14:51
sven(see e.g. all those retro game decompilation projects which create source code that compiles 1:1 to the original game for an extreme example of that)
#asahi-re · 2023-12-10 14:50
sventhe issue with decompilation is that it's very easy to end up implementing apple's logic 1:1 instead of understanding the hardware. that's why it's better to be very careful and the reason for that policy
#asahi-re · 2023-11-10 07:59
svenamarioguy: sure, adding the opcodes to the wiki can’t hurt
#asahi-re · 2023-08-28 12:52
svenI think there's a software logic analyzer for GPIO signals (which also includes many busses like spi or i2c) of the m1/m2 inside m1n1 somewhere
#asahi-re · 2023-08-28 12:51
svenluix: depends on what you want to do with it I guess. I don't think I've used a logic analyzer for anything asahi-related yet, maybe #asahi-offtopic is a better fit for that question
#asahi-re · 2023-08-02 18:27
svensure :)
#asahi-re · 2023-06-13 10:17
svenI guess we can now claim that there are no hidden blobs even when using sep now ;p
#asahi-re · 2023-06-13 10:17
svenhuh, interesting. macOS sep kext are very verbose so I’m not sure if those keys will be useful. certainly good to have them though :)
#asahi-re · 2023-03-01 10:20
svenagain, just ignore it and /msg the admins to deal with them. no need to discuss the craziness any further in here
#asahi-re · 2023-03-01 10:19
svenplease just ignore the trolls
#asahi-re · 2022-12-03 09:39
svenand I think marcan tried turning them on in pmgr and that just didn’t work
#asahi-re · 2022-12-03 09:38
svenIboot removes them from the template device tree iirc
#asahi-re · 2022-12-03 00:01
sventhat one just walks the pagetable to find the physical address and then just reads that
#asahi-re · 2022-12-03 00:00
sventhere’s dart.ioread or something like that

plus ›