https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61237
Bug ID: 61237
Summary: gcc puts line number 11 ahead of line number 10 in
function call in debug info
Product: gcc
Version: 4.8.1
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: debug
Assignee: unassigned at gcc dot gnu.org
Reporter: dje at google dot com
I can imagine gcc's trying to be helpful here, but it's not working in this
case (at least in IMHO).
I want to set a breakpoint on line 10, but when I get there line 11 has already
been executed, and by that time it's too late.
[This example is stripped down from the original.]
1 int
2 foo (int a, int b)
3 {
4 return a + b;
5 }
6
7 int
8 main ()
9 {
10 foo (1,
11 -foo (2, 3));
12 return 0;
13 }
$ gdb a.out
(gdb) disas /m main
Dump of assembler code for function main:
9 {
0x0000000000400484 <+0>: push %rbp
0x0000000000400485 <+1>: mov %rsp,%rbp
10 foo (1,
0x0000000000400497 <+19>: neg %eax
0x0000000000400499 <+21>: mov %eax,%esi
0x000000000040049b <+23>: mov $0x1,%edi
0x00000000004004a0 <+28>: callq 0x400470 <foo>
11 -foo (2, 3));
0x0000000000400488 <+4>: mov $0x3,%esi
0x000000000040048d <+9>: mov $0x2,%edi
0x0000000000400492 <+14>: callq 0x400470 <foo>
12 return 0;
0x00000000004004a5 <+33>: mov $0x0,%eax
13 }
0x00000000004004aa <+38>: pop %rbp
0x00000000004004ab <+39>: retq
End of assembler dump.
gdb's attempt at making mixed disassembly output more readable isn't helpful
either. The thing to note here is that line 11 appears before line 10 in the
code. Technically, one can argue that that is correct I suppose. And if that
is the answer, fine. I'm just filing this so that the answer is written down
somewhere. I did spend 10 minutes trying to find an existing bug in the
database (which feels like enough - I have to time out at some point), and I
grepped for O0 in the "not a bug" page
https://gcc.gnu.org/bugs/#nonbugs
and didn't find anything.
OTOH, if you want to agree this is a bug, fine by me. :-)