Hello everybody,
I have successfully built gcc-4.0.0 out of the release
tar-file (complete distribution) on an athlon system
targeting i386-scientific-linux.
The C and C++ parts work like a charm so far, compiling
substantial amounts of C/C++ code without noticeable glitches.
Congratulations and thank you for doing that good job!
There seems, however, to be an issue with the Fortran 90
part: Compiling the Fortran interface of the HDF5 libraries
(a hierarchical data format library for storing and managing
scientific data) triggers an internal compiler error.
Isolating the cause I find the following source code
to trigger the error:
--- test.f90 ------------------------------------------------------
MODULE H5GLOBAL
INTEGER, PARAMETER :: H5F_FLAGS_LEN = 16
INTEGER H5F_flags(H5F_FLAGS_LEN)
COMMON /H5F_FLAGS/ H5F_flags
INTEGER :: H5F_CLOSE_WEAK_F
EQUIVALENCE(H5F_flags(1), H5F_CLOSE_WEAK_F)
END MODULE H5GLOBAL
-------------------------------------------------------------------
[EMAIL PROTECTED] ~]$ gfortran -v -c test.f90
Using built-in specs.
Target: i386-scientific-linux
Configured with: ../gcc-4.0.0/configure --target=i386-scientific-linux
--host=i386-scientific-linux --prefix=/usr --bindir=/usr/bin --
libdir=/usr/lib --libexecdir=/usr/lib --infodir=/usr/share/info --
datadir=/usr/share --mandir=/usr/share/man --enable-static --enable-
shared --enable-threads=posix --enable-multilib --enable-build-warnings
--enable-install-libiberty --enable-c-mbchar --enable-__cxa_atexit --
disable-checking --disable-coverage --with-gnu-ld --with-gnu-as --with-
ld=/usr/bin/ld --with-as=/usr/bin/as --with-elf --with-system-zlib --
enable-languages=c,c++,f95,java,objc
Thread model: posix
gcc version 4.0.0
/usr/lib/gcc/i386-scientific-linux/4.0.0/f951 test.f90 -quiet -dumpbase
test.f90 -auxbase test -version -o /tmp/ccoCpasL.s
GNU F95 version 4.0.0 (i386-scientific-linux)
compiled by GNU C version 4.0.0.
GGC heuristics: --param ggc-min-expand=99 --param ggc-min-
heapsize=129482
test.f90:0: internal compiler error: backend decl for module variable
h5f_close_weak_f already exists
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.
----------------------------------------------------------------------
I am not very familiar with Fortran 90, so I cannot state
whether the Fortran90 source code itself is correct.
Given that HDF5 is quite popular I guess so...
However, I hope this helps to resolve the issue. If you have
any further questions, let me know.
Thanks again for all your efforts,
Steffen.
--
Steffen Seeger <[EMAIL PROTECTED]>
TU Chemnitz
