tests: Check for both quiet and signaling NaN on mipsel/hppa (#6029)
MIPSEL, MIPS64EL and HP-PA architectures have quiet and
signaling NaNs swapped. The test assumes correct values
for i386 / amd64 but obviously fails on these architectures.
It is documented in many places, like
https://software.llnl.gov/yorick-doc/stdlib/ieee.html :
Warning-- apparently there is no universal standard for what
constitutes signalling versus quiet NaN
on MIPS and HPPA architectures, qNaN and sNaN are reversed
https://gcc.gnu.org/onlinedocs/gcc/MIPS-Options.html :
-mnan=2008
-mnan=legacy
These options control the encoding of the special
not-a-number (NaN) IEEE 754 floating-point data.
The -mnan=legacy option selects the legacy encoding.
In this case quiet NaNs (qNaNs) are denoted by the first
bit of their trailing significand field being 0, whereas
signaling NaNs (sNaNs) are denoted by the first bit of
their trailing significand field being 1.
The -mnan=2008 option selects the IEEE 754-2008 encoding.
In this case qNaNs are denoted by the first bit of their
trailing significand field being 1, whereas sNaNs are
denoted by the first bit of their trailing significand
field being 0.
The default is -mnan=legacy unless GCC has been configured
with --with-nan=2008.
The behavior mentioned above causes FTBFS in Debian starting
with 1.12.0:
https://buildd.debian.org/status/fetch.php?pkg=flatbuffers&arch=mips64el&ver=1.12.1%7Egit20200711.33e2d80%2Bdfsg1-0.1&stamp=
1594830681&raw=0
https://buildd.debian.org/status/fetch.php?pkg=flatbuffers&arch=mipsel&ver=1.12.1%7Egit20200711.33e2d80%2Bdfsg1-0.1&stamp=
1594832647&raw=0
https://buildd.debian.org/status/fetch.php?pkg=flatbuffers&arch=hppa&ver=1.12.1%7Egit20200711.33e2d80%2Bdfsg1-0.1&stamp=
1594830726&raw=0
For HP-PA (PA-RISC) there is no way to override the reversal,
and for MIPSEL/MIPS64EL there is no macro defining the '-mnan'
choice, so both variants were added with OR boolean operator.
Signed-off-by: Vasyl Gello <vasek.gello@gmail.com>