> In your test case as written there should not be any such section in
> file.s. file.h only includes header.h, and does not actually define
> any variables at all.
>
> The section group symbol looks right to me--the guard variable should
> be associated with the variable it is guarding, it shouldn't be in its
> own group.
>
This is the content of file.s
.file "file.cpp"
.weak _ZGVZN1A4InitEvE1i
.section .bss._ZGVZN1A4InitEvE1i,"awG",@nobits,_ZZN1A4InitEvE1i,comdat
.align 8
.type _ZGVZN1A4InitEvE1i, @object
.size _ZGVZN1A4InitEvE1i, 8
_ZGVZN1A4InitEvE1i:
.zero 8
.ident "GCC: (GNU) 4.5.0 20091006 (experimental)"
.section .note.GNU-stack,"",@progbits
when compiled with older gcc, the only (apart from .ident) line changed is
.section .bss._ZGVZN1A4InitEvE1i,"awG",@nobits,_ZGVZN1A4InitEvE1i,comdat
so I thought _ZGVZN1A4InitEvE1i is correct there.
> If that is writing to NULL, it suggests that _ZZN1A4InitEvE1i
> (A::Init()::i) was somehow never defined. What do you see for that
> symbol using readelf -s on the executable and on the input object
> files?
When compiled with gcc-4.5, command line "g++ file.o main.o -o a.out"
$ readelf -s file.o | grep ZZN1A4InitEvE1i
$ readelf -s main.o | grep ZZN1A4InitEvE1i
25: 0000000000000000 8 OBJECT WEAK DEFAULT 16 _ZZN1A4InitEvE1i
$ readelf -s a.out | grep ZZN1A4InitEvE1i
62: 0000000000000000 0 OBJECT WEAK DEFAULT UND _ZZN1A4InitEvE1i
(this one crashes)
gcc-4.5, command line "g++ main.o -o a.out" or "g++ main.o file.o -o a.out"
$ readelf -s a.out | grep ZZN1A4InitEvE1i
49: 0000000000601068 8 OBJECT WEAK DEFAULT 26 _ZZN1A4InitEvE1i
When compiled with gcc-4.3:
$ readelf -s file.o | grep ZZN1A4InitEvE1i
$ readelf -s main.o | grep ZZN1A4InitEvE1i
27: 0000000000000000 8 OBJECT WEAK DEFAULT 17 _ZZN1A4InitEvE1i
$ readelf -s a.out | grep ZZN1A4InitEvE1i
49: 0000000000601068 8 OBJECT WEAK DEFAULT 26 _ZZN1A4InitEvE1i
Zdenek
