http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46518
--- Comment #6 from Jack Howarth <howarth at nitro dot med.uc.edu> 2010-11-17 15:44:38 UTC --- (In reply to comment #4) > (In reply to comment #2) > > Only occurs for -fPIC on x86_64-apple-darwin10. Adding -fno-PIC on darwin > > suppresses the ICE. > > Is that in relation to bootstrap flags? There's no PIC flags in the original > report. > > The breakage is rather curious. Build-with-CXX should not cause code > generation differences. Can you please build normally-bootstrapped compiler > from the same source, build the testcase with the two compilers appending > -fdump-tree-all -fdump-rtl-all and attach the two series of dumps? Under x86_64 Fedora 10 using... ../gcc/configure --with-gmp=/usr --with-mpfr=/usr --with-mpc=/usr --prefix=/home/howarth/dist --enable-languages=c,c++,fortran --enable-build-with-cxx --disable-lto --disable-bootstrap I get a slightly different crash (which might not be surprising since the compiler isn't built entirely with -fPIC like on darwin). Executing on host: /home/howarth/work/gcc/testsuite/gfortran/../../gfortran -B/home/howarth/work/gcc/testsuite/gfortran/../../ -L/home/howarth/work/x86_64-unknown-linux-gnu/32/libgfor tran /home/howarth/gcc/gcc/testsuite/gfortran.dg/pr42294.f -O -O2 -fselective-scheduling2 -fsel-sched-pipelining -funroll-all-loops -S -m32 -o pr42294.s (timeout = 300) /home/howarth/gcc/gcc/testsuite/gfortran.dg/pr42294.f: In function 'orien': /home/howarth/gcc/gcc/testsuite/gfortran.dg/pr42294.f:39:0: internal compiler error: Segmentation fault
