[Bug c++/63940] New: C++ constexpr errors on AIX

[Bug c++/63940] New: C++ constexpr errors on AIX

[dje at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63940

            Bug ID: 63940
           Summary: C++ constexpr errors on AIX
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dje at gcc dot gnu.org

Created attachment 34029
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=34029&action=edit
Full compiler error output

/tmp/20141118/powerpc-ibm-aix7.1.0.0/libstdc++-v3/include/bits/regex_compiler.h:
424:31:   in constexpr expansion of 'std::__detail::_BracketMatcher<
<template-parameter-1-1>, <anonymous>, <anonymous>
>::_S_cache_size<std::regex_traits<char>, false, false>()'
/tmp/20141118/powerpc-ibm-aix7.1.0.0/libstdc++-v3/include/bits/regex_compiler.h:
425:20: error: '(int)(1ul * 8ul)' is not a constant expression
/tmp/20141118/powerpc-ibm-aix7.1.0.0/libstdc++-v3/include/bits/regex_compiler.h:
424:31:   in constexpr expansion of 'std::__detail::_BracketMatcher<
<template-parameter-1-1>, <anonymous>, <anonymous>
>::_S_cache_size<std::regex_traits<char>, false, true>()'
/tmp/20141118/powerpc-ibm-aix7.1.0.0/libstdc++-v3/include/bits/regex_compiler.h:
425:20: error: '(int)(1ul * 8ul)' is not a constant expression
/tmp/20141118/powerpc-ibm-aix7.1.0.0/libstdc++-v3/include/bits/regex_compiler.h:
424:31:   in constexpr expansion of 'std::__detail::_BracketMatcher<
<template-parameter-1-1>, <anonymous>, <anonymous>
>::_S_cache_size<std::regex_traits<char>, true, false>()'
/tmp/20141118/powerpc-ibm-aix7.1.0.0/libstdc++-v3/include/bits/regex_compiler.h:
425:20: error: '(int)(1ul * 8ul)' is not a constant expression
/tmp/20141118/powerpc-ibm-aix7.1.0.0/libstdc++-v3/include/bits/regex_compiler.h:
424:31:   in constexpr expansion of 'std::__detail::_BracketMatcher<
<template-parameter-1-1>, <anonymous>, <anonymous>
>::_S_cache_size<std::regex_traits<char>, true, true>()'
/tmp/20141118/powerpc-ibm-aix7.1.0.0/libstdc++-v3/include/bits/regex_compiler.h:
425:20: error: '(int)(1ul * 8ul)' is not a constant expression

[Bug c++/63940] C++ constexpr errors on AIX

[dje at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63940

--- Comment #1 from David Edelsohn <dje at gcc dot gnu.org> ---
Created attachment 34030
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=34030&action=edit
preprocessed source for testcase

[Bug c++/63940] C++ constexpr errors on AIX

[schwab@linux-m68k.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63940

--- Comment #2 from Andreas Schwab <schwab@linux-m68k.org> ---
This looks bogus (from cxx_eval_binary_expression):

  if (lhs == orig_lhs && rhs == orig_rhs)
    return t;

Just because both operands are self-evaluating doesn't mean we can return the
unfolded expression.

[Bug c++/63940] C++ constexpr errors on AIX

[jason at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63940

--- Comment #3 from Jason Merrill <jason at gcc dot gnu.org> ---
Author: jason
Date: Tue Nov 18 21:14:49 2014
New Revision: 217738

URL: https://gcc.gnu.org/viewcvs?rev=217738&root=gcc&view=rev
Log:
    PR c++/63940
    * constexpr.c (cxx_eval_binary_expression): Don't assume the
    expression was already folded.
    (cxx_eval_unary_expression): Likewise.

Modified:
    trunk/gcc/cp/ChangeLog
    trunk/gcc/cp/constexpr.c

[Bug c++/63940] C++ constexpr errors on AIX

[jason at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63940

Jason Merrill <jason at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |FIXED
           Assignee|unassigned at gcc dot gnu.org      |jason at gcc dot gnu.org

--- Comment #4 from Jason Merrill <jason at gcc dot gnu.org> ---
Fixed.

[Bug c++/63940] C++ constexpr errors on AIX

[dje at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63940

David Edelsohn <dje at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
   Last reconfirmed|                            |2014-11-23
         Resolution|FIXED                       |---
     Ever confirmed|0                           |1

--- Comment #5 from David Edelsohn <dje at gcc dot gnu.org> ---
The failures continue to appear on AIX.  AIX defaults to 32 bit.  Maybe the
tests on x86-64 need -m32 to see additional causes of the problem?

[Bug c++/63940] C++ constexpr errors on AIX

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63940

--- Comment #6 from H.J. Lu <hjl.tools at gmail dot com> ---
32-bit x86 looks OK:

https://gcc.gnu.org/ml/gcc-testresults/2014-11/msg02534.html
https://gcc.gnu.org/ml/gcc-testresults/2014-11/msg02537.html

[Bug c++/63940] C++ constexpr errors on AIX

[dje at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63940

--- Comment #7 from David Edelsohn <dje at gcc dot gnu.org> ---
HJ,

The test was not failing on Linux x86-64 nor x86-32. I sent pre-processed
testcase from AIX that Jason was able to reproduce on Linux. But the fix did
not solve the failure on AIX.

[Bug c++/63940] C++ constexpr errors on AIX

[jason at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63940

--- Comment #8 from Jason Merrill <jason at gcc dot gnu.org> ---
(In reply to David Edelsohn from comment #7)
> The test was not failing on Linux x86-64 nor x86-32. I sent pre-processed
> testcase from AIX that Jason was able to reproduce on Linux. But the fix did
> not solve the failure on AIX.

I just checked in an additional fix for bug 63942.  Does that resolve it?

[Bug c++/63940] C++ constexpr errors on AIX

[dje at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63940

David Edelsohn <dje at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|---                         |FIXED

--- Comment #9 from David Edelsohn <dje at gcc dot gnu.org> ---
Yea!  Yes, the most recent patch fixed it.