There is no weakness in address randomization I relied on for exploitation.
It relies on two distinct bugs, an info-leak to obtain a pointer to an allocation in the kalloc.1024 zone and a memory corruption primitive (deriving from a NULL pointer dfr. in IOKit) allowing me to OR 0x10 anywhere in kernel memory.
To break kASLR I corrupt the size of a vm_map_copy struct, which allows me to read the adjacent allocation to the struct, which is a C++ object. First 8 bytes of said C++ object is a pointer to the vtable, which resides in __TEXT of some kernel extension. Since I can calculate the unslid address from userland without any issue, by subtracting what gets leaked with what gets calculated you get to know the kASLR slide.
Just to clarify: The code execution part has 100% reliability rate. The kASLR leaking part does have some chance in it, however empirical evidence indicates that the failure rate is extremely low.
If the heap info leak fails, I bail out cleanly.
If the kASLR leak fails, it is usually because instead of hitting a vm_map_copy (the intended structure I need to corrupt), something completely unrelated is hit instead. If it happens to hit something different than expected, that's undefined behaviour usually ending up in a panic.
If osx were s/xnu/minix 3-style, full microkernel/, sploiting Iokit as a least priv'd process, only it would get pwned and be limited to iokit's acls. Still bad, but it likely wouldnt have rights to exec a root shell.
XNU kexts have way too much authority, and all the syscalls they each tack on compounds the attack surface to the total codebases of all Apple and third-party kexts. Because once you've found and symbolicated the not-really-hidden call table, you're pretty much able to do whatever. And with a mutating mem kext bug ...
while I agree with you on the security benefits of a full microkernel, to be entirely honest, if you had access to just IOKit you could easily use a network card or an hard drive controller to get a physical memory write-what-where, which in turn would allow you to gain access to anything, plus the microkernel performance issues of e.g. having to context switch on interrupts.
That things may be broken is no argument against defense-in-depth and least privilege. By having smaller codebases and smaller system components, the attack surface is far, far smaller than say Linux. IOKit is shit as is, and a full microkernel would break it up into processes based on areas of responsibility. Also, that shows the hardware needs better bus- and command-level security to prevent such attacks. (Don't even get me started on closed firmware blobs or unverifiability of commercial cores.)
Expecting a finished product of a new project would be unreasonable. Minix 3 is early on and not the only full microkernel out there. They will probably find an approach to reduce context switches if it's a mature optimization to make. An Android-like mobile/embedded platform would make a sensible research -> real use-case, minus Java.
It relies on two distinct bugs, an info-leak to obtain a pointer to an allocation in the kalloc.1024 zone and a memory corruption primitive (deriving from a NULL pointer dfr. in IOKit) allowing me to OR 0x10 anywhere in kernel memory.
To break kASLR I corrupt the size of a vm_map_copy struct, which allows me to read the adjacent allocation to the struct, which is a C++ object. First 8 bytes of said C++ object is a pointer to the vtable, which resides in __TEXT of some kernel extension. Since I can calculate the unslid address from userland without any issue, by subtracting what gets leaked with what gets calculated you get to know the kASLR slide.
Just to clarify: The code execution part has 100% reliability rate. The kASLR leaking part does have some chance in it, however empirical evidence indicates that the failure rate is extremely low.