| Summary: | Assembler Error: symbol `_ZT' is already defined | ||
|---|---|---|---|
| Product: | gcc | Reporter: | Ludovic Brenta <ludovic> |
| Component: | bootstrap | Assignee: | Not yet assigned to anyone <unassigned> |
| Status: | RESOLVED INVALID | ||
| Severity: | normal | CC: | ebotcazou, gcc-bugs |
| Priority: | P3 | ||
| Version: | 4.1.0 | ||
| Target Milestone: | --- | ||
| Host: | sparc-sun-solaris2.8 | Target: | sparc-sun-solaris2.8 |
| Build: | sparc-sun-solaris2.8 | Known to work: | |
| Known to fail: | Last reconfirmed: | ||
|
Description
Ludovic Brenta
2006-04-05 13:02:39 UTC
> As can be seen from the configure options, this is with GNU > binutils 2.16.1. What's the configure shell? What's the version of GNU make? What's the bootstrap compiler? > I also tried the Sun assembler and linker, and get similar errors > but thousands of times instead of just 7. Complete logs available > on request (1.2 Mb for GNU as/ld, 91 Mb for Sun as/ld; respectively > 54k and 2553k compressed with bzip2). Both sets of tools are known to work with GCC 4.1.0 on SPARC/Solaris 8. > Am I doing something wrong? What happens if you bootstrap the compiler (make bootstrap instead of make)? $ make --version GNU Make 3.80 Copyright (C) 2002 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. $ $SHELL --version GNU bash, version 2.03.0(1)-release (sparc-sun-solaris) Copyright 1998 Free Software Foundation, Inc. $ gcc --version 2.8.1 (that's from GNAT 3.15p, binary distribution from AdaCore; but the compiler used to build strstream.cc is ./xgcc, i.e. GCC 4.1.0 built in stage1. And, as I said, the same errors occur with --enable-bootstrap=yes, so I can confidently say that the bootstrap compiler is irrelevant). I've started the "make bootstrap" and will report later. To speed it up I've only passed --enable-languages=c,c++. -- Ludovic Brenta. > (that's from GNAT 3.15p, binary distribution from AdaCore; but the
> compiler used to build strstream.cc is ./xgcc, i.e. GCC 4.1.0
> built in stage1. And, as I said, the same errors occur with
> --enable-bootstrap=yes, so I can confidently say that the
> bootstrap compiler is irrelevant).
--enable-bootstrap=yes has not been tested with GCC 4.1.x so I'd avoid it.
--enable-bootstrap=no + "make bootstrap" cause a successful build: ... /export/home/lbre/src/gcc-4.1.0.bootstrap/./gcc/xgcc -shared-libgcc -B/export/home/lbre/src/gcc-4.1.0.bootstrap/./gcc -nostdinc++ -L/export/home/lbre/src/gcc-4.1.0.bootstrap/sparc-sun-solaris2.8/libstdc++-v3/src -L/export/home/lbre/src/gcc-4.1.0.bootstrap/sparc-sun-solaris2.8/libstdc++-v3/src/.libs -B/export/home/lbre/opt/gcc-4.1.0-bootstrap/sparc-sun-solaris2.8/bin/ -B/export/home/lbre/opt/gcc-4.1.0-bootstrap/sparc-sun-solaris2.8/lib/ -isystem /export/home/lbre/opt/gcc-4.1.0-bootstrap/sparc-sun-solaris2.8/include -isystem /export/home/lbre/opt/gcc-4.1.0-bootstrap/sparc-sun-solaris2.8/sys-include -I/export/home/lbre/src/gcc-4.1.0.bootstrap/sparc-sun-solaris2.8/libstdc++-v3/include/sparc-sun-solaris2.8 -I/export/home/lbre/src/gcc-4.1.0.bootstrap/sparc-sun-solaris2.8/libstdc++-v3/include -I/export/home/lbre/src/gcc-4.1.0/libstdc++-v3/libsupc++ -g -O2 -fno-implicit-templates -Wall -Wextra -Wwrite-strings -Wcast-qual -fdiagnostics-show-location=once -ffunction-sections -fdata-sections -I/export/home/lbre/src/gcc-4.1.0.bootstrap/sparc-sun-solaris2.8/libstdc++-v3/include/backward -Wno-deprecated -c ../../../../gcc-4.1.0/libstdc++-v3/src/strstream.cc -fPIC -DPIC -o .libs/strstream.o /export/home/lbre/src/gcc-4.1.0.bootstrap/./gcc/xgcc -shared-libgcc -B/export/home/lbre/src/gcc-4.1.0.bootstrap/./gcc -nostdinc++ -L/export/home/lbre/src/gcc-4.1.0.bootstrap/sparc-sun-solaris2.8/libstdc++-v3/src -L/export/home/lbre/src/gcc-4.1.0.bootstrap/sparc-sun-solaris2.8/libstdc++-v3/src/.libs -B/export/home/lbre/opt/gcc-4.1.0-bootstrap/sparc-sun-solaris2.8/bin/ -B/export/home/lbre/opt/gcc-4.1.0-bootstrap/sparc-sun-solaris2.8/lib/ -isystem /export/home/lbre/opt/gcc-4.1.0-bootstrap/sparc-sun-solaris2.8/include -isystem /export/home/lbre/opt/gcc-4.1.0-bootstrap/sparc-sun-solaris2.8/sys-include -I/export/home/lbre/src/gcc-4.1.0.bootstrap/sparc-sun-solaris2.8/libstdc++-v3/include/sparc-sun-solaris2.8 -I/export/home/lbre/src/gcc-4.1.0.bootstrap/sparc-sun-solaris2.8/libstdc++-v3/include -I/export/home/lbre/src/gcc-4.1.0/libstdc++-v3/libsupc++ -g -O2 -fno-implicit-templates -Wall -Wextra -Wwrite-strings -Wcast-qual -fdiagnostics-show-location=once -ffunction-sections -fdata-sections -I/export/home/lbre/src/gcc-4.1.0.bootstrap/sparc-sun-solaris2.8/libstdc++-v3/include/backward -Wno-deprecated -c ../../../../gcc-4.1.0/libstdc++-v3/src/strstream.cc -o strstream.o >/dev/null 2>&1 (no output) The bootstrap then completes successfully. So: "make" fails, "make bootstrap" works, but the commands invoked are identical. Could it be that gcc 2.8.1 silently miscompiled cc1plus in the case of "make", and introduced an obscure bug which only strstream.cc triggers? -- Ludovic Brenta. > The bootstrap then completes successfully. Wunderbar. :-) > So: "make" fails, "make bootstrap" works, but the commands > invoked are identical. Could it be that gcc 2.8.1 silently > miscompiled cc1plus in the case of "make", and introduced an > obscure bug which only strstream.cc triggers? Most likely, yes; --enable-bootstrap=yes should have worked around the problem, but it may have its own problems in 4.1.x. |