With LAST_UPDATED "Thu Jun 30 14:21:26 UTC 2005", gfortran.fortran-torture/execute/common.f90 fails: FAIL: gfortran.fortran-torture/execute/common.f90 execution, -O1 FAIL: gfortran.fortran-torture/execute/common.f90 execution, -O2 FAIL: gfortran.fortran-torture/execute/common.f90 execution, -O2 -fomit-frame-pointer -finline-functions FAIL: gfortran.fortran-torture/execute/common.f90 execution, -O2 -fno-repack-arrays FAIL: gfortran.fortran-torture/execute/common.f90 execution, -Os
whereas cross from i686-pc-linux-gnu with "Mon Jul 4 10:24:52 UTC 2005" shows no fail for that case. (Results from previous tests has been flushed.) In constrast, the "nearby" in-pack.f90 fails with -Os on both hosts, regardless of -fpic:y options. Lots of apples and oranges regarding source dates, but this is just a public note that I need to investigate, particularly since (1) it does not fail *at all* when adding (just) -fpic, (2) a target patch will cause it to fail when -O2 -fpic. It also fails baseline with -mno-gotplt -fPIC and -mno-gotplt *without* -fpic/-fPIC! Also marked "target" so nobody thinks I'm implying there's a bug elsewhere, like in libgfortran. I suspect a "64-bit bug" lurking somewhere. For the record, random people wanting to repeat this, need to install the glibc from cris-dist-1.62 (developer.axis.com, use the rpm, the dpkg, or follow the instructions and gcc in that distribution to compile it) and use --prefix=/usr/local/cris with the current tree to be able to repeat the observations noted above. -- Summary: cris-axis-linux-gnu cross: gfortran.fortran- torture/execute/common.f90 -O2 host difference? Product: gcc Version: 4.1.0 Status: UNCONFIRMED Severity: normal Priority: P2 Component: target AssignedTo: hp at gcc dot gnu dot org ReportedBy: hp at gcc dot gnu dot org CC: gcc-bugs at gcc dot gnu dot org GCC host triplet: x86_64-unknown-linux-gnu GCC target triplet: cris-axis-linux-gnu http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22382