public inbox for libstdc++-cvs@sourceware.org
help / color / mirror / Atom feed
From: Matthias Kretz <mkretz@gcc.gnu.org>
To: gcc-cvs@gcc.gnu.org, libstdc++-cvs@gcc.gnu.org
Subject: [gcc r12-6734] libstdc++: Fix for non-constexpr math_errhandling
Date: Wed, 19 Jan 2022 12:46:09 +0000 (GMT)	[thread overview]
Message-ID: <20220119124609.225593858D37@sourceware.org> (raw)

https://gcc.gnu.org/g:39f581028c76ebfc94a5c2714f43c9e56089a9b0

commit r12-6734-g39f581028c76ebfc94a5c2714f43c9e56089a9b0
Author: Matthias Kretz <m.kretz@gsi.de>
Date:   Tue Jan 18 17:04:06 2022 +0100

    libstdc++: Fix for non-constexpr math_errhandling
    
    Use SFINAE magic to support: "It is unspecified whether math_errhandling
    is a macro or an identifier with external linkage." [C Standard]
    
    Signed-off-by: Matthias Kretz <m.kretz@gsi.de>
    
    libstdc++-v3/ChangeLog:
    
            * include/experimental/bits/simd.h (__floating_point_flags): Do
            not rely on math_errhandling to expand to a constant expression.

Diff:
---
 libstdc++-v3/include/experimental/bits/simd.h | 26 ++++++++++++++++++++------
 1 file changed, 20 insertions(+), 6 deletions(-)

diff --git a/libstdc++-v3/include/experimental/bits/simd.h b/libstdc++-v3/include/experimental/bits/simd.h
index c991e3f223e..82e9841195e 100644
--- a/libstdc++-v3/include/experimental/bits/simd.h
+++ b/libstdc++-v3/include/experimental/bits/simd.h
@@ -283,20 +283,34 @@ constexpr inline bool __have_power_vmx = __have_power_vsx;
 
 namespace __detail
 {
-  constexpr bool __handle_fpexcept =
 #ifdef math_errhandling
-    math_errhandling & MATH_ERREXCEPT;
-#elif defined __FAST_MATH__
-    false;
+  // Determines _S_handle_fpexcept from math_errhandling if it is defined and expands to a constant
+  // expression. math_errhandling may expand to an extern symbol, in which case a constexpr value
+  // must be guessed.
+  template <int = math_errhandling>
+    constexpr bool __handle_fpexcept_impl(int)
+    { return math_errhandling & MATH_ERREXCEPT; }
+#endif
+
+  // Fallback if math_errhandling doesn't work: with fast-math assume floating-point exceptions are
+  // ignored, otherwise implement correct exception behavior.
+  constexpr bool __handle_fpexcept_impl(float)
+  {
+#if defined __FAST_MATH__
+    return false;
 #else
-    true;
+    return true;
 #endif
+  }
+
+  /// True if math functions must raise floating-point exceptions as specified by C17.
+  static constexpr bool _S_handle_fpexcept = __handle_fpexcept_impl(0);
 
   constexpr std::uint_least64_t
   __floating_point_flags()
   {
     std::uint_least64_t __flags = 0;
-    if constexpr (__handle_fpexcept)
+    if constexpr (_S_handle_fpexcept)
       __flags |= 1;
 #ifdef __FAST_MATH__
     __flags |= 1 << 1;


                 reply	other threads:[~2022-01-19 12:46 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20220119124609.225593858D37@sourceware.org \
    --to=mkretz@gcc.gnu.org \
    --cc=gcc-cvs@gcc.gnu.org \
    --cc=libstdc++-cvs@gcc.gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).