Re: FreeBSD's Google Summer of Code 2016

[Rich Felker <dalias@libc.org>]

On Sat, Mar 05, 2016 at 03:11:28PM -0500, Pedro Giffuni wrote:
> Hello;
> 
> Just thought I'd point this out in case there are students looking
> for a summer project (paid by Google):
> 
> In FreeBSD we are always open to alternatives. For this GSoC we
> would consider someone willing to work on a musl port.
> 
> https://wiki.freebsd.org/SummerOfCodeIdeas#Port_musl_libc
> 
> For such a port you obviously have to have love for the C language
> and specific skills (version control, knowledge of the build system,
> UNIX/syscall knowledge), but I think it may very well be a fun
> project for the right person.
> 
> We generally ask for very detailed proposals and it certainly
> would help if you get to try FreeBSD before considering a
> proposal, The process is competitive so we are interested
> in your background.
> 
> We are aware musl is centered around linux but just as someone
> managed to port glibc to FreeBSD we think it should be possible
> to make a port clean port of musl. This project would only make
> sense if the code is upstreamed, so at some stage we would also
> appreciate support from musl developers.
> 
> For applying you should follow the standard procedure with Google.
> https://summerofcode.withgoogle.com/how-it-works/

Nice!

From our side (musl), the biggest obstacle when people have looked at
doing such a port in the past seems to be missing syscalls, especially
futex and related functionality. For that reason, up til now,
FreeBSD's Linux syscall layer has seemed more functional, and a better
basis for musl-on-FreeBSD, than the native syscall layer. I think a
viable proposal needs a solution to this problem on the kernel side --
perhaps exposing some of these syscalls that are Linux-emulation-only
now in the native API?

Rich