Comment by OutOfHere

4 days ago

An agentic LLM should be pretty good at Arm64 assembly generation, but maintainability of large code could become an issue. Why would it not run on Linux?

13 comments

OutOfHere

Reply
imtomt  4 days ago

I wrote it for MacOS because I don't have a Linux machine right now :( Once I get one up and running again, I'll probably work on porting this.

As for why it wouldn't run on Linux, there are some pretty big differences in the actual assembly. One pretty superficial difference is calling conventions -- MacOS uses the x16 register for syscall numbers, Linux uses x8. Calling the kernel in Mac uses "svc #0x80", in Linux it's "svc #0". That's ~120 lines that need to be replaced, but easy enough to just use sed. Syscall numbers are all different, as are the struct layouts for sigaction(), MacOS has an "sa_tramp" field that Linux doesn't have. Enforcing max processes is done here using the MacOS-specific proc_info() syscall, which can be used to get the number of children any given process has. Linux doesn't have an equivalent, so process tracking would need to be done differently. Finally, Linux has the getdents64() syscall, rather than getdirentries64(), which uses a different struct and is called differently.

I'm sure an LLM could make all those changes, but it's a pretty large codebase, so it would probably make some mistakes or miss things.

  • OutOfHere  4 days ago

    Is it not possible to maintain a single codebase for both OS with appropriate .ifdef commands?

    • imtomt  4 days ago

      It would be possible, but impractical for where this project is now. In retrospect, I could have written it with more portability from the start, but since it was just a personal learning project I didn’t think too hard about it. Aside from syscall numbers all being different (which would just require changing the consts in defs.S), syscall numbers are stored in the x16 register on MacOS vs x8 on Linux, and the kernel is called with svc #0x80 on MacOS vs svc #0 on Linux. The bigger issue at this point is MacOS-specific syscalls. proc_info() doesn’t exist on Linux, so that would need to be rewritten. The sigaction struct also has different members (Linux doesn’t have sa_tramp), as well as every other struct having different offsets and sizes. getdirentries64() also doesn’t exist on Linux, it has getdents64() which is called differently and has different output as well I believe, so directory.S would need to be reworked too.