https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125129
--- Comment #10 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> --- > --- Comment #9 from Gaius Mulley <gaius at gcc dot gnu.org> --- > Created attachment 64976 > --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=64976&action=edit > Proposed fix includes documentation and python3 unavailability fallback > > Bootstrapped on Debian and Freebsd15 when python3 is unavailable. Maybe you could use https://github.com/autoconf-archive/autoconf-archive/blob/master/m4/ax_python_devel.m4 for cases where python3 is available, just not by that generic name? > All regression tests pass. Thanks. The revision arrived just in time when I'd started fresh builds: they passed on x86_64-pc-linux-gnu (-j128) and i386-pc-solaris2.11. (-j28). However, on sparc-sun-solaris2.11 (-j96), I got the same hang as before. Of the three instances of the generated SYSTEM.def, only two are non-empty: -rw-r--r-- 1 ro gcc 7352 Jul 9 10:51 m2/gm2-libs-coroutines/SYSTEM.def -rw-r--r-- 1 ro gcc 6590 Jul 9 10:51 m2/gm2-libs-iso/SYSTEM.def -rw-r--r-- 1 ro gcc 0 Jul 9 10:51 m2/gm2-libs/SYSTEM.def When I manually remove the empty one and run > make -n m2/gm2-libs/SYSTEM.def make: Nothing to be done for '/vol/gcc/src/hg/master/local/gcc/m2/gm2-libs/SYSTEM.def'. This seems odd: it tries to remake the source file, not the generated one. Seems like a missing dependency or a race condition to me. I've yet to try a build on amd64-pc-netbsd10.1 which lacks python3 by that name, although python3.13 is present. I might cheat by placing a python3 symlink to that somewhere in PATH. There's an issue that may well benefit from this patch. Several links failed with libgm2/libm2pim/../../gcc/m2/gm2-libs/Builtins.mod:343:(.text+0x5f1): undefined reference to `cabsl' and indeed <complex.h> has long double cabsl(long double complex) __RENAME(__c99_cabsl); so __c99_cabsl would need to be used instead. AFAICS this is already handled in gcc/config/netbsd.cc.
