Copy head (r256279) to stable/10 as part of the 10.0-RELEASE cycle.

Approved by: re (implicit)
Sponsored by: The FreeBSD Foundation

Forced commit to denote that the claim in r235388, that libcompiler_rt when
used as a replacement for libgcc on sparc64 would cause a 60% decrease in
performance is wrong. Actually, the performance loss is only marginal and
the huge difference seen was caused by a pilot error comparing libgcc builds
using MALLOC_PRODUCTION with libcompiler_rt builds that did not.

- Get rid of debugging support in order to get rid of the V8-specific C
compiler frame size used there so this whole thing is V8/V9-agnostic.
- Use 32-bit function alignment as GCC does when using UltraSPARC I or
higher optimizations.
- Don't waste delay slots when possible.

Unfortunately, this still doesn't make libcompiler_rt a viable replacement
for libgcc on sparc64 though as once installed instead, buildworld times
increase by nearly 60% (which isn't related to these assembler functions).

Add SPARC64 version of div/mod written in assembly.

This version is similar to the code shipped with libgcc. It is based on
the code from the SPARC64 architecture manual, provided without any
restrictions.

Tested by: flo@