| Issue |
144061
|
| Summary |
[lldb] Step over can step too far
|
| Labels |
lldb
|
| Assignees |
|
| Reporter |
labath
|
```
$ cat a.cc
void f1() {}
void f2() {}
void __attribute__((always_inline)) f3() { f2(); }
int main() {
f1();
f3();
return 0;
}
$ clang -g a.cc
$ ~/ll/build/opt/bin/lldb ./a.out
(lldb) target create "./a.out"
Current executable set to '/tmp/R/a.out' (x86_64).
(lldb) b main
Breakpoint 1: where = a.out`main + 19 at a.cc:6:3, address = 0x00000000000011a3
(lldb) r
Process 8667 launched: '/tmp/R/a.out' (x86_64)
Process 8667 stopped
* thread #1, name = 'a.out', stop reason = breakpoint 1.1
frame #0: 0x00005555555551a3 a.out`main at a.cc:6:3
3 void __attribute__((always_inline)) f3() { f2(); }
4
5 int main() {
-> 6 f1();
7 f3();
8 return 0;
9 }
(lldb) disassemble
a.out`main:
0x555555555190 <+0>: endbr64
0x555555555194 <+4>: pushq %rbp
0x555555555195 <+5>: movq %rsp, %rbp
0x555555555198 <+8>: subq $0x10, %rsp
0x55555555519c <+12>: movl $0x0, -0x4(%rbp)
-> 0x5555555551a3 <+19>: callq 0x555555555160 ; f1 at a.cc:1
0x5555555551a8 <+24>: callq 0x555555555170 ; f2 at a.cc:2
0x5555555551ad <+29>: xorl %eax, %eax
0x5555555551af <+31>: addq $0x10, %rsp
0x5555555551b3 <+35>: popq %rbp
0x5555555551b4 <+36>: retq
(lldb) next
Process 8667 stopped
* thread #1, name = 'a.out', stop reason = step over
frame #0: 0x00005555555551ad a.out`main at a.cc:8:3
5 int main() {
6 f1();
7 f3();
-> 8 return 0;
9 }
```
The step operation should stop at 0x5555555551a8, and not 0x5555555551ad. I expect this happens because, after stepping out of `f1`, the code finds itself thinking it's in `f3` and decides it should step out of that.
This is a relatively recent regression as lldb-19 (at least) doesn't exhibit this behavior (it correctly stops at 0x5555555551a8).
![]()
_______________________________________________
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
