static void *a0ra, *a1ra;
void __attribute__((noinline)) a0() { a0ra = __builtin_return_address(0); }
void __attribute__((noinline)) a1() { a1ra = __builtin_return_address(0); }
int foo() { a0(); a1(); return a1ra - a0ra; }
int main()
{
printf("pd=%d\n", foo());
return 0;
}
$ gcc -O3 --save-temps call.c && ./a.out
in theroy this code should return on vary platforms
the sizeof(call/bl [mem] opcode).
it works fine ix86 (pd=5), ppc (pd=4) but doesn't work on amd64 (pd=7 !!!)
in the assemler dump i see redundant `xor eax,eax`.
a1: movq (%rsp), %rax
movq %rax, a1ra(%rip)
ret
a0: movq (%rsp), %rax
movq %rax, a0ra(%rip)
ret
foo: xorl %eax, %eax <== what for?
call a0
xorl %eax, %eax <== what for?
call a1
movq a1ra(%rip), %rax
subl a0ra(%rip), %eax
ret
--
Summary: redundant opcodes before function call.
Product: gcc
Version: 4.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: other
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: pluto at agmk dot net
GCC build triplet: x86-64
GCC host triplet: x86-64
GCC target triplet: x86-64
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25345
