On Tue, Jan 01, 2019 at 04:02:01PM +0000, Robert Swindells wrote: > > m...@netbsd.org wrote: > >On Tue, Jan 01, 2019 at 03:34:47PM +0100, Martin Husemann wrote: > >> On Tue, Jan 01, 2019 at 02:22:54PM +0000, maya%netbsd.org@localhost wrote: > >> > I'm stuck between the hammer and the anvil here. > >> > I'd like NetBSD to use a recent MesaLib, the kind that uses LLVM. > >> > >> Can you explain how the MesaLib build uses llvm? > >> > >> This kinda sounds like you need a special build time binary that would > >> live in $TOOLDIR and it is not exactly clear how installing some headers > >> helps building this tool (as you can not assume a NetBSD build host). > >> > >> Martin > > > >I haven't dug up too much into it in the past, so this was fun to > >discover: > > > >MesaLib is a compiler for turning agnostic representations like GLSL and > >SPIR-V into hardware appropriate ones. It has passes and intermediate > >formats (plural!) and everything. > > > >AMD drivers use LLVM as a step in the process, they defined an AMDGPU as > >a target architecture. > > > >I'm not sure how a JIT works, but swrast uses LLVM for its JIT. > > > >at least on the pkgsrc build, swrast and radeonsi are dynamically linked > >with libLLVM. > > How does the usage of libLLVM from MesaLib make use of the LLVM headers > at *runtime* ? > > If they are not needed at runtime then why do they need to be installed > on a target machine ?
it doesn't need the headers at runtime. joerg is currently objecting to installing the llvm libraries too.
