da-viper wrote:
> @da-viper, thanks for sharing the idea! I will add the test.
>
> My only concern is that the source line after the `bar()` call may not be the
> instruction after the call instruction, for example, compiler may emit extra
> code after call instruction.
I do not think in this test case the compiler will emit any code after the
`bar()` call instruction that does not have a line entry.
The tests are run without optimisation by default. the `#line` directive forces
all lines after it to be relative for example
```cpp
bool foo() { // line 1
bar(); // line 2
#line 0 "test.cpp"
return true; // line 0
} // line 1
// line 2
int main () { // line 3
return 0; // line 4
} // line 5
```
that's why I forward declared `foo` and `foo's` definition is the last function
in the test case.
> So this test depends on the assumption that there is no extra code between
> call instruction and next source line.
No, it makes sure that any instruction after the `bar()` call does not have a
valid line entry. That way we can test the frames in the stack trace will have
a valid source line entry.
https://github.com/llvm/llvm-project/pull/165944
_______________________________________________
lldb-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
