IRC channel logs
2022-11-14.log
back to list of logs
<stikonas>even M2-Planet code with blood-elf footers is a bit hard to debug... I'll probably have to extract relavant bits of malloc/free code to a separate file buildable with gcc to debug remaining issues... <stikonas>I guess one can improve M2/blood-elf to spit out more debug info but that's a much larger project <stikonas>building with gcc was a useful exercise... Just compililng caught some bugs and things seems to be working better <oriansj>stikonas: well blood-elf was orginally a quick hack to create the bare minimal amount of debug information and that is where the improvements would be (unless they needed information from the C code itself in which case we would need to figure an information passing method between M2-Planet and Blood-elf that doesn't impact M1 or hex2) <stikonas>oriansj: anyway, I think I now have a working free implementation <stikonas>it's fairly simple and goes only part-way (for now it never reduced brk) but at least it reuses freed space <stikonas>I'm not fully convinced yet that it is bug free <stikonas>I've used that very simple first fit free list, it migt not be as good as that buddy or slab allocator but very simple <stikonas>even then I had to adapt quite a bit to M2 dialecct <oriansj>well if you are worried about the free logic having bugs, we can always set a -D variable to disable that bit of logic <stikonas>or in case of bugs we can debug them later. Still stage0-posix working means it's not fundamentally broken <oriansj>I'd rather have your possibly broken code stikonas, than some of my coworker's "working code" as yours tends to have less bugs and problems <oriansj>I'll give it some testing later today to see if I can find some bugs in it <oriansj>thank you for driving M2libc forward ^_^