💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › kisslinux-2023-08-10.txt captured on 2024-05-12 at 16:00:02.
⬅️ Previous capture (2023-09-08)
-=-=-=-=-=-=-
[2023-08-10T04:31:39Z] <sad_plan> hi [2023-08-10T05:05:06Z] <sad_plan> rfaa: llvm should already find /usr/include, as its a default path for include. anyway, did you fork llvm, and bootstrap it? am assuming youre trying to replace gcc here initially [2023-08-10T05:05:55Z] <sad_plan> theres a bootstrap script here https://github.com/konimex/kiss-llvm if thats what youre looking for anyway [2023-08-10T07:15:52Z] <rfaa> copied the llvm from the kiss community repo and added the LLVM_BINUTILS_INCDIR in the build, haven't compiled yet.. :D [2023-08-10T07:16:40Z] <rfaa> i'm thinking i need to add that parameter for it to generate the LLVMgold.so? [2023-08-10T08:10:01Z] <sad_plan> hm. strange. could you perhaps disable it though? [2023-08-10T08:10:45Z] <sad_plan> I dont even have llvmgold.so on my system. although I also dont have binutils gold. not sure if thats related [2023-08-10T08:16:49Z] <sad_plan> it seems you were on to something; https://releases.llvm.org/3.0/docs/GoldPlugin.html [2023-08-10T08:17:25Z] <sad_plan> atleast for enabling gold for llvm. [2023-08-10T08:20:41Z] <rfaa> yeah it just finished compiling, installing it now, will see if the gold file is included [2023-08-10T08:23:07Z] <rfaa> yep, there it is [2023-08-10T08:25:01Z] <sad_plan> nice [2023-08-10T08:25:40Z] <rfaa> trying to build mold again with -flto=thin now, seems to be working so far [2023-08-10T08:28:44Z] <sad_plan> I dont really see how mold would error out on llvm stuff though. are you using llvm instead of gcc? [2023-08-10T08:29:39Z] <rfaa> binutils in the repo is configured with --enable-gold but it seems like llvm needs the LLVMgold.so to be able to use it, perhaps the LLVM_BINUTILS_INCDIR should be added to the build for llvm? don't know the procedure here but i could look into creating a pull request if it seems like a good idea [2023-08-10T08:29:43Z] <rfaa> yeah llvm [2023-08-10T08:32:13Z] <rfaa> oops, [100%] Linking CXX executable mold [2023-08-10T08:32:14Z] <rfaa> clang-16: error: unable to execute command: Segmentation fault [2023-08-10T08:32:14Z] <rfaa> clang-16: error: linker command failed due to signal (use -v to see invocation) [2023-08-10T08:32:14Z] <rfaa> make[2]: *** [CMakeFiles/mold.dir/build.make:660: mold] Error 1 [2023-08-10T08:32:14Z] <rfaa> make[1]: *** [CMakeFiles/Makefile2:240: CMakeFiles/mold.dir/all] Error 2 [2023-08-10T08:32:15Z] <rfaa> make: *** [Makefile:166: all] Error 2 [2023-08-10T08:33:33Z] <sad_plan> if youre using llvm instead of gcc, I dont really think thats applicable for the upstream. but you could always ping one of the maintainers. or make a pr for it on codeberg [2023-08-10T08:37:53Z] <rfaa> thanks, i'll look into it after i get this sorted :) i'm basically just trying out llvm/clang/mold to see if it speeds up this old machine [2023-08-10T08:40:15Z] <sad_plan> lld is faster than binutils. and I belive mold is even faster, so yeah, one would assume that would be generally faster. lld is also packaged in community, if you rather wanna use lld instead of binutils ld or gold