[Bug 253313] lib/msun: hypotl(3) mishandles subnormal numbers

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=253313

--- Comment #1 from Dimitry Andric <dim at FreeBSD.org> ---
Another interesting detail, if I recompile libm's e_hypotl.c with gcc 10.2 and
link it into libm.so, the result from the earlier z-amd64 executable becomes:

% LD_LIBRARY_PATH=/usr/obj/usr/src/amd64.amd64/lib/msun
/share/dim/bugs/bug253313/z-amd64
Via scaling and sqrtl
x=2.853684e-4932 y=1.444012e-4922 a=1.444012e-4922
x=0x1.b2933cafa0bb7p-16383 y=0x1.fffffffffffffp-16351
a=0x1.000000006ca4c72cp-16350

Via hypotl
x=2.853684e-4932 y=1.444012e-4922 z=0.000000e+00
x=0x1.b2933cafa0bb7p-16383 y=0x1.fffffffffffffp-16351 z=0x0p+0

so now hypotl doesn't result in nan, but zero isn't right either, I guess...

Interestingly the glibc and msun hypotl code seem to have the same Sun
Microsystems origin, but have diverged quite a bit. Hard to tell from a simple
viewing what causes this difference.

-- 
You are receiving this mail because:
You are the assignee for the bug.