- in august of last year i was working on memory analysis in yaxpeax-core and then i decided part of that was hard, so instead i would go
- set up yaxpeax-x86 to codegen all the uarch-specific decoder logic
and then i got stuck on how to deal with x86-64 having more than 64 feature flags (this then requires more than a u64 of bitmasks to decide which instructions reject. so it would make the likely path of "just decode what you think looks like an instruction" slower. unacceptable.)
after getting that most of the way i decided i wanted to use the library more to get a feel for what changes would or wouldn't be okay so
- i started working on zvm again and put a jit in it so i could use yaxpeax-x86 to disassemble the jitted code
but then i got distracted by https://github.com/athre0z/disas-bench/pull/9, wherein i had to pull together random yaxpeax-x86 stuff that has been kicking around for months but i never actually released and
- found that i still had to remember to
make testto check that yaxpeax-x86 builds in fact build and passed tests in all the configurations i check against
so i wanted an actual testing server for this stuff but still have an objection to relying on github for anything, because github.
- went and built a CI system for myself. it works now, and demonstrates my fantastic css skills
now i can get back to working on yaxpeax-x86 so i can get back to jitting through dynamic dispatch in zvm so i can get back to a substantial yaxpeax-x86 refactor and maybe one day i'll get back to yaxpeax-core and memory ssa stuff
god