[Bug target/39501] -O -ffinite-math-only gets min(x,y) optimization wrong for soft-float on arm-*-gnueabi

[martinwguy at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39501

--- Comment #17 from Martin Guy <martinwguy at gmail dot com> 2010-12-03 14:46:28 UTC ---
Sort of. The cause of the bug was never found, and the workaround is to disable
an instruction.  It might be worth trying enabling movsfcc in current GCC and
re-running the tests, since the conditional execution stuff in the middle end
was rewritten between 4.3 and 4.4 if I remember correctly, so the actual bug in
the middle end may have gone away with that rewrite.