In message <advt6jl_ovzny...@kib.kiev.ua>, Konstantin Belousov writes: > On Sat, May 31, 2025 at 10:05:16PM -0700, Cy Schubert wrote: > > In message <20250601041116.0a564...@slippy.cwsent.com>, Cy Schubert writes: > > > In message <adoawun7jgs7y...@kib.kiev.ua>, Konstantin Belousov writes: > > > > On Fri, May 30, 2025 at 04:14:44PM +0200, Herbert J. Skuhra wrote: > > > > > On Fri, 30 May 2025 13:32:27 +0200, Konstantin Belousov wrote: > > > > > > > > > > > > On Fri, May 30, 2025 at 09:36:39AM +0200, Herbert J. Skuhra wrote: > > > > > > > Hi, > > > > > > > > > > > > > > building lang/go (e.g. 1.24) and go ports (e.g. aerc, netbird) on > mai > > > n > > > > > > > (amd64) fails with: fatal: bad g in signal handler. > > > > > > > Arm64 seems to be OK. > > > > > > > > > > > > I might have a guess. Try the following untested patch, you need t > o re > > > bu > > > > ild > > > > > > at least kernel, but ideally both kernel and userspace. Also it is > amd > > > 64 > > > > -only. > > > > > > > > > > Thanks a lot! With your patch I could build go 1.24 and netbird again > . > > > > > > > > Is there any go code that uses cgo, and which you could test with the p > atch > > > > as well? > > > > > > > > > > It's a bad system call. Yesterday's buildworld/installworld made the syst > em > > > unable to run today's installworld. > > > > > > --- installworld --- > > > make[1]: /export/obj/opt/src/git-src/amd64.amd64/toolchain-metadata.mk:1: > > > > Using cached toolchain metadata from build at stinky on Sat May 31 20:16: > 05 > > > PDT 2025 > > > --- __installcheck_UGID --- > > > --- __installcheck_sh_check --- > > > Bad system call (core dumped) > > > rescue/sh check failed, installation aborted > > > > > > make[1]: stopped making "installworld" in /opt/src/git-src > > > > > > make: stopped making "installworld installkernel" in /opt/src/git-src > > > > > > The go problem is likely related to this. > > > > The resolution is: > > > > git pull the latest 15-CURRENT. Buildkernel and installkernel. Then > > buildworld and installworld. Remember to rebuild drm-66-kmod or it will > > panic. Any Go software will build correctly. > > In fact, I am not completely sure about Go. > Pure Go binaries should work. > > The interesting case is cgo binaries, which start libpthread threads. > I am interested if they work, and if no, I have some ugly trick to apply.
My thinking at the time was that the problem might have been related to the syscall table. But looking at the commits, I'm not convinced of that now. It must be something else. -- Cheers, Cy Schubert <cy.schub...@cschubert.com> FreeBSD UNIX: <c...@freebsd.org> Web: https://FreeBSD.org NTP: <c...@nwtime.org> Web: https://nwtime.org e^(i*pi)+1=0
