[llvm-bugs] [Bug 39372] New: Does not complains about loop bound variable not being shared

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=39372

            Bug ID: 39372
           Summary: Does not complains about loop bound variable not being
                    shared
           Product: OpenMP
           Version: unspecified
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Clang Compiler Support
          Assignee: unassignedclangbugs at nondot.org
          Reporter: lebedev.ri at gmail.com
                CC: llvm-bugs at lists.llvm.org

Originally reported as https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87668

https://godbolt.org/z/05Zgdd

#include <vector>
struct b {};
void c() {
  std::vector<b> d;
#ifdef _OPENMP
#pragma omp parallel default(none)
#pragma omp for
#endif
  for (auto f = d.begin(); f < d.end(); ++f)
    ;
}

#ifndef _OPENMP
Yes, there is OMP.
#endif


GCC output:

<source>: In function 'void c()':
<source>:9:25: error: 'd' not specified in enclosing 'parallel'
   for (auto f = d.begin(); f < d.end(); ++f)
                         ^
<source>:6:9: error: enclosing 'parallel'
 #pragma omp parallel default(none)
         ^~~
Compiler returned: 1

Clang is perfectly happy with that code.


Jakub Jelinek 2018-10-21 10:22:10 UTC
GCC is correct here.
Even with the recent 5.0 clarifications, those expressions are to be evaluated
before the worksharing-loop construct, but still inside of the parallel region,
so with default(none) d needs to be specified on the parallel (e.g. shared or
firstprivate).

-- 
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/20181021/6ec2ff34/attachment.html>