https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100896
Fangrui Song <i at maskray dot me> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |i at maskray dot me
--- Comment #4 from Fangrui Song <i at maskray dot me> ---
In gcc/acinclude.m4:285, if cross-compiling, (`if test "x${build}" =
"x${target}" && test "x${build}" = "x${host}"; then `) will not be taken.
else
case "${target}" in
aarch64*-linux-gnu*)
# AArch64 postdates glibc support for .init_array/.fini_array,
# so we don't need the preprocessor test above.
gcc_cv_initfini_array=yes
;;
*)
AC_MSG_CHECKING(cross compile... guessing)
gcc_cv_initfini_array=no
;;
esac
fi])
On non-aarch64, `gcc_cv_initfini_array=no` will run,
`HAVE_INITFINI_ARRAY_SUPPORT` will therefore be 0.
compilers/powerpc64le-linux-gnu/gcc/gcc/auto-host.h
1578:#define HAVE_INITFINI_ARRAY_SUPPORT 0
compilers/powerpc64le-linux-gnu/gcc/gcc/config.status
1257:D["HAVE_INITFINI_ARRAY_SUPPORT"]=" 0"
compilers/powerpc64le-linux-gnu/gcc/gcc/config.log
6900:| #define HAVE_INITFINI_ARRAY_SUPPORT 0
7169:| #define HAVE_INITFINI_ARRAY_SUPPORT 0
7484:| #define HAVE_INITFINI_ARRAY_SUPPORT 0
8557:#define HAVE_INITFINI_ARRAY_SUPPORT 0
The built GCC will use the legacy .ctors
% many=/tmp/glibc-many
% cat a.c
__attribute__ ((constructor)) static int foo (void) { return 42; }
%
/tmp/glibc-many/install/compilers/powerpc64le-linux-gnu/bin/powerpc64le-glibc-linux-gnu-gcc
-c a.c && readelf -WS a.o | egrep 'ctors|init_array'
[ 4] .ctors PROGBITS 0000000000000000 000070 000008 00 WA
0 0 8
[ 5] .rela.ctors RELA 0000000000000000 000218 000018 18 I
10 4 8
---
Noticed the problem when using scripts/build-many-glibcs.py built GCC
(cd ~/Dev/glibc)
scripts/build-many-glibcs.py /tmp/glibc-many checkout --shallow
scripts/build-many-glibcs.py /tmp/glibc-many host-libraries
scripts/build-many-glibcs.py /tmp/glibc-many compilers powerpc64le-linux-gnu
--keep all
