#
✕ nick:marcan
#asahi · 2021-12-17 12:52
marcanbut the payload thing always relied on abusing the format and having a section hang "off the end"
#asahi · 2021-12-17 12:52
marcanit seems they parse the macho in kmutil now and iBoot just does something simpler
#asahi · 2021-12-17 12:52
marcanbut also I wouldn't be surprised if the new way of doing things broke the payload concatenation trick
#asahi · 2021-12-17 12:52
marcanthe macho file needs some extra stuff we don't know about
#asahi · 2021-12-17 12:51
marcannot that big a change
#asahi · 2021-12-17 12:51
marcan-rw-r--r-- 1 marcan marcan 589824 Dec 17 21:43 m1n1.macho
#asahi · 2021-12-17 12:51
marcan-rw-r--r-- 1 marcan marcan 753664 Dec 17 21:43 m1n1.bin
#asahi · 2021-12-17 12:51
marcanthat should work with payloads; it includes all .bss, but I removed a lot of bloat
#asahi · 2021-12-17 12:50
marcanhttps://github.com/AsahiLinux/m1n1#usage there, added a m1n1.bin target
#asahi · 2021-12-17 12:12
marcan... just need to do something about video on the Mac Mini
#asahi · 2021-12-17 12:12
marcaneither way, I'm likely going to make 12.1 the minimum version soon enough, and this new feature is a good excuse to do that
#asahi · 2021-12-17 12:12
marcanI wonder if this broke kmutil for 11.x; for those iBoot wouldn't understand a non-macho, but you'd be using SFR kmutil
#asahi · 2021-12-17 12:03
marcanI'm sure some are in here :p
#asahi · 2021-12-17 12:03
marcanthey show up to my youtube streams sometimes
#asahi · 2021-12-17 12:03
marcanthe engineers working on stuff? no
#asahi · 2021-12-17 12:03
marcanmanagement? maybe
#asahi · 2021-12-17 12:02
marcanthey opened up the platform, why wouldn't they have the tiniest bit of courtesy and add a couple features that cost them ~nothing on top and make our lives easier?
#asahi · 2021-12-17 12:02
marcanI don't know why people have this idea of Apple as the big evil behemoth we have to fight
#asahi · 2021-12-17 12:01
marcanwhich is required for our installer to work
#asahi · 2021-12-17 12:01
marcanthey also enabled self-signed binaries in recovery mode after launch
#asahi · 2021-12-17 12:01
marcanit really isn't
#asahi · 2021-12-17 11:59
marcanit having --raw mode seems squarely aimed at us
#asahi · 2021-12-17 11:59
marcankmutil has always been a developer tool
#asahi · 2021-12-17 11:58
marcanchadmed: this is a developer tool. none of this would be used for a hypothetical official windows port.
#asahi · 2021-12-17 11:45
marcanthat means iBoot learned to boot raw images
#asahi · 2021-12-17 11:44
marcanheh, interesting, it doesn't actually build a mach-o, instead it sticks the load addr and entrypoint into img4 tags
#asahi · 2021-12-17 11:28
marcanthis is a developer tool
#asahi · 2021-12-17 11:27
marcanwhich makes sense anyway given the variable CPU count
#asahi · 2021-12-17 11:27
marcanit's mostly secondary_stacks; we can probably move that to dynamic allocation
#asahi · 2021-12-17 11:26
marcanbut it's not terrible either, given the compression. might just roll with the padding.
#asahi · 2021-12-17 11:13
marcanwe have 4MB or so of BSS right now, which isn't great
#asahi · 2021-12-17 11:12
marcanah, but then that'd clobber some stuff that gets placed after that... hmm
#asahi · 2021-12-17 11:11
marcanbut I'm more tempted to just memmove the payload off of bss in the startup code
#asahi · 2021-12-17 11:11
marcanthere's a compression mode, so we can use that if we must
#asahi · 2021-12-17 11:04
marcanwell the raw thing works, though you need to pass the addresses in decimal for some inexplicable reason
#asahi · 2021-12-17 10:51
marcanok, I think the __info section is what we need
#asahi · 2021-12-17 10:45
marcanwill it just make a mach-o for us?
#asahi · 2021-12-17 10:44
marcanI wonder how the --raw option works though
#asahi · 2021-12-17 10:44
marcanif I were to use this, I'd just add some bootstrap code to move the payload over to skip bss, though that might slow things down a bit with large payloads
#asahi · 2021-12-17 10:43
marcanyeah
#asahi · 2021-12-17 10:43
marcanonly problem is the mach-o isn't contiguous for us; the payload skips bss, so if we do it this way we'll have to add a huge chunk of bss padding which isn't great
#asahi · 2021-12-17 10:42
marcan(and people said apple wouldn't help... I can't think of how this could possibly *not* be intended for us)
#asahi · 2021-12-17 10:42
marcanbut I'm going to take this as a "use this, this will never break" message...
#asahi · 2021-12-17 10:41
marcanplus I already assume all that for chainloading anyway, heh
#asahi · 2021-12-17 10:41
marcan:)
#asahi · 2021-12-17 10:41
marcan... maybe we should switch to that
#asahi · 2021-12-17 10:40
marcanwhich along with --entry-point and --lowest-virtual-address, lets you load a raw blob, not a mach-o
#asahi · 2021-12-17 10:40
marcanbut they *also* added a --raw option
#asahi · 2021-12-17 10:40
marcanso apple broke kmutil in 12.1 for us, by requiring $something from the mach-o we don't know yet
#asahi · 2021-12-17 10:40
marcanHA

plus ›