Bug#750834: gcc-4.8: [arm64] float to double conversion does not silence sNaN

[Aurelien Jarno <aurel32@debian.org>]

Package: gcc-4.8
Version: 4.8.3-2
Severity: normal
Forwarded: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61441

Consider the following code:

#define _GNU_SOURCE
#include <stdio.h>
#include <math.h>

int main (void)
{
  float sNaN = __builtin_nansf ("");
  double x = (double) sNaN;
  return issignaling(x);
}

It correctly returns 0 at -O0 optimisation level, but returns 1 at -O1
and -O2 optimisation levels.

Here is the generated assembly code at -O0:
0000000000400630 <main>:
  400630:       a9be7bfd        stp     x29, x30, [sp,#-32]!
  400634:       910003fd        mov     x29, sp
  400638:       18000160        ldr     w0, 400664 <main+0x34>
  40063c:       b9001fa0        str     w0, [x29,#28]
  400640:       b9401fa0        ldr     w0, [x29,#28]
  400644:       1e270000        fmov    s0, w0
  400648:       1e22c000        fcvt    d0, s0
  40064c:       9e660000        fmov    x0, d0
  400650:       f9000ba0        str     x0, [x29,#16]
  400654:       fd400ba0        ldr     d0, [x29,#16]
  400658:       97ffff8e        bl      400490 <__issignaling@plt>
  40065c:       a8c27bfd        ldp     x29, x30, [sp],#32
  400660:       d65f03c0        ret
  400664:       7fa00000        .word   0x7fa00000

Here is the generated assembly code at -O1:
0000000000400630 <main>:
  400630:       a9bf7bfd        stp     x29, x30, [sp,#-16]!
  400634:       910003fd        mov     x29, sp
  400638:       5c000080        ldr     d0, 400648 <main+0x18>
  40063c:       97ffff95        bl      400490 <__issignaling@plt>
  400640:       a8c17bfd        ldp     x29, x30, [sp],#16
  400644:       d65f03c0        ret
  400648:       00000000        .word   0x00000000
  40064c:       7ff40000        .word   0x7ff40000

As you can see at -O1, the sNaN constant is propagated, and the
propagated value is loaded and passed directly to issignaling. 
Quoting the IEEE Std 754 standard:
"Under default exception handling, any operation signaling an invalid
operation exception and for which a floating-point result is to be
delivered shall deliver a quiet NaN."

So it looks like the copy propagation is not done correctly, the
sNaN should be silenced in the process.

Note: gcc-4.9 is also affected.

-- System Information:
Debian Release: jessie/sid
  APT prefers unreleased
  APT policy: (500, 'unreleased'), (500, 'unstable')
Architecture: arm64 (aarch64)

Kernel: Linux 3.15.0-5-generic (SMP w/1 CPU core)
Locale: LANG=fr_FR.UTF-8, LC_CTYPE=fr_FR.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash

Versions of packages gcc-4.8 depends on:
ii  binutils        2.24.51.20140604-2
ii  cpp-4.8         4.8.3-2
ii  gcc-4.8-base    4.8.3-2
ii  libc6           2.18-7
ii  libcloog-isl4   0.18.2-1
ii  libgcc-4.8-dev  4.8.3-2
ii  libgmp10        2:6.0.0+dfsg-4
ii  libisl10        0.12.2-1
ii  libmpc3         1.0.1-1
ii  libmpfr4        3.1.2-1
ii  zlib1g          1:1.2.8.dfsg-1

Versions of packages gcc-4.8 recommends:
ii  libc6-dev  2.18-7

Versions of packages gcc-4.8 suggests:
pn  gcc-4.8-doc        <none>
pn  gcc-4.8-locales    <none>
pn  libasan0-dbg       <none>
pn  libatomic1-dbg     <none>
pn  libbacktrace1-dbg  <none>
pn  libgcc1-dbg        <none>
pn  libgomp1-dbg       <none>
pn  libitm1-dbg        <none>
pn  libquadmath-dbg    <none>
pn  libtsan0-dbg       <none>

-- no debconf information