http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54373
Bug #: 54373
Summary: [4.7/4.8 Regression]: build fails for
mmix-knuth-mmixware libobjc GCC_NO_EXECUTABLES
Classification: Unclassified
Product: gcc
Version: 4.7.0
Status: UNCONFIRMED
Keywords: build, ice-on-valid-code
Severity: normal
Priority: P3
Component: target
AssignedTo: unassig...@gcc.gnu.org
ReportedBy: h...@gcc.gnu.org
Host: x86_64-unknown-linux-gnu
Target: mmix-knuth-mmixware
Some time after the 4.6 branch building binutils separately (not in a combined
tree) broke. A build of binutils, installing that, then building gcc with PATH
including the binutils installation now results in:
(building gcc, then most other things including newlib, libstdc++, then:)
...
checking whether make sets $(MAKE)... yes
checking stdio.h usability... yes
checking stdio.h presence... yes
checking for stdio.h... yes
checking for ANSI C header files... (cached) yes
checking sched.h usability... yes
checking sched.h presence... yes
checking for sched.h... yes
checking whether the target supports thread-local storage... configure: error:
Link tests are not allowed after GCC_NO_\
EXECUTABLES.
make[1]: *** [configure-target-libobjc] Error 1
make[1]: Leaving directory `/expvol/pp_slask/hp/cstmmbtmp/cstmmb_47/gccobj'
(gcc-4_7-branch revision 190562)
Similarly for trunk, for example revision 190649.
For failing versions, I see in gccobj/mmix/libobjc/config.log traces of basic
compilation problems:
configure:2869: /expvol/pp_slask/hp/cstmmbtmp/cstmmb3/gccobj/./gcc/xgcc
-B/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/gccobj/\
./gcc/ -nostdinc -B/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/gccobj/mmix/newlib/
-isystem /expvol/pp_slask/hp/cstmmbtmp/cst\
mmb3/gccobj/mmix/newlib/targ-include -isystem
/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/gcc/newlib/libc/include -B/expvol/p\
p_slask/hp/cstmmbtmp/cstmmb3/gccobj/mmix/libgloss/mmix
-L/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/gccobj/mmix/libgloss/lib\
nosys -L/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/gcc/libgloss/mmix
-B/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/pre/mmix/bin/ -\
B/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/pre/mmix/lib/ -isystem
/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/pre/mmix/include -i\
system /expvol/pp_slask/hp/cstmmbtmp/cstmmb3/pre/mmix/sys-include -o
conftest -g -O2 conftest.c >&5
/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/pre/mmix/bin/ld:
/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/gccobj/./gcc/crtbegin.o: b\
ase-plus-offset relocation against register symbol: __TMC_END__ in
.MMIX.reg_contents
/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/gccobj/./gcc/crtbegin.o: In function
`deregister_tm_clones':
(.text+0xa): relocation truncated to fit: R_MMIX_BASE_PLUS_OFFSET against
`__TMC_END__'+7
/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/pre/mmix/bin/ld:
/expvol/pp_slask/hp/cstmmbtmp/cstmmb3/gccobj/./gcc/crtbegin.o: I\
nternal inconsistency error for value for
linker-allocated global register: linked: 0x2000000020000000000007e8 !=
relaxed: 0x2000000020000000000007ef
thus I label this a target problem (maybe even a binutils and not gcc problem),
though the libobjc failure is far from gracious.
