[llvm-bugs] [Bug 47400] New: constexpr variables in Lambda not known in Debug Info

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=47400

            Bug ID: 47400
           Summary: constexpr variables in Lambda not known in Debug Info
           Product: clang
           Version: trunk
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: C++
          Assignee: unassignedclangbugs at nondot.org
          Reporter: melanie.blower at intel.com
                CC: blitzrakete at gmail.com, dgregor at apple.com,
                    erik.pilkington at gmail.com, llvm-bugs at lists.llvm.org,
                    richard-llvm at metafoo.co.uk

Within the lambda body, nothing is known about the constexpr identifier

Small example:

#define MAX_N_STEPS 8192
int main() {
  constexpr int n_steps = MAX_N_STEPS + 2;
  auto F = [=](unsigned int param) {
    for (short i = n_steps; i >= 0; --i)
      param += i;
  };
  F(42);
  return 0;
}

gcc does produce information about the identifier, tho' the value is not
available (optimized out)
(gdb) b 6
Breakpoint 1 at 0x40111a: file test.cpp, line 6.
(gdb) r
Starting program: a.out.g++
Breakpoint 1, <lambda(unsigned int)>::operator()(unsigned int) const (
    __closure=0x7fffffff96eb, param=42) at test.cpp:6
6             param += i;
(gdb) whatis n_steps
type = const int

However, if you build with clang the identifier isn't available
(gdb) whatis n_steps
No symbol "n_steps" in current context. 

Debugger engineer at Intel recommended that the debug info could be emitted as
if the test program were written this way:
#define MAX_N_STEPS 8192
class lambda {
private:
  static constexpr int n_steps = MAX_N_STEPS + 2;
public:
  void operator() (unsigned int param) {
    for (short i = n_steps; i >= 0; --i)
      param += i;
  }
};
int main() {
  constexpr int n_steps = MAX_N_STEPS + 2;
  lambda F;
  F(42);
  return 0;
}

My colleague at Intel Mike Rice wrote up an exploratory patch, I will link it
to this bug report.  Hoping to get some feedback from debug info experts.

Thanks and regards --Melanie Blower

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20200902/8efa38d0/attachment.html>