https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98773
--- Comment #5 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> --- > --- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> --- > So I wasn't able to reproduce on a x86_64 host bootstrapping a 32bit > compiler configured as > > --enable-languages=c++,ada --disable-multilib --target=i686-pc-linux-gnu > --host=i686-pc-linux-gnu --build=i686-pc-linux-gnu --without-zstd > > with a host 64bit GCC 7 (using -m32). That's almost the same I've been using. My configure options boil down to configure \ --target=i686-pc-linux-gnu --build=i686-pc-linux-gnu --host=i686-pc-linux-gnu --enable-targets=all \ --with-as=/vol/gcc/bin/i686/gas-2.35 --with-gnu-as \ --with-ld=/vol/gcc/bin/i686/gld-2.35 --with-gnu-ld \ --enable-languages=all,ada,obj-c++,go \ --with-zstd-include=/vol/gcc/include --with-zstd-lib=/vol/gcc/lib/i686 Maybe the use of binutils 2.35 makes a difference? Those are locally-built vanilla versions, btw., nothing delivered by a distro. However, I've seen the same failure with native Solaris assembler and linker, making it unlikely that the binutils version is a factor. > Since the rev. didn't fix a regression reverting is a good option but of > course > it would be nice to understand why it's broken ... :/ Absolutely, otherwise this is going to raise its ugly head again. > Any hint as to why the above isn't enough to reproduce? Nothing obvious. However, a sparc-sun-solaris2.11 bootstrap with just the tree-data-ref.c snippet just finished, failing just as with your full patch.
