Hi,
My software bali-phy is having build failures on amd64 with hardware
floating-point (armhf) that don't occur on x86-64, and seemingly do not
occur with software floating-point (armel).
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=918047
The bug seems to come from NaN's getting generated on arm64/armhf that
do not occur on x86-64. Further inspection of the build log shows that
the same bug occurs on hppa, but no other architectures.
I will investigate further (for example, turning off -ffast-math on
x86-64) but the obvious route would be to run a debugger on hardware
where the bug can be reproduced. Any suggestions on how to deal with this?
-BenRI