Fix remaining CXX cross-compilation builds in Makefile.

Fix remaining CXX cross-compilation builds in Makefile.

r22509 fixed arm and mips (big-endian) cross-compilation builds.

It did not fix mipsel, mips64el, or arm64 builds, since the
toolchain prefix did not match the v8_target_arch.

For mips, the toolchain is typically named mips-linux-gnu-g++,
but it can compile for mipsel or mips64el via flags. (There can
also be mipsel-linux-gnu-g++ toolchains, and this patch will
work for mipsel build in that case).

For arm64, the toolchains seem to be named aarch64-linux-gnu-g++.

BUG=
R=jkummerow@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=22531

[paul.l..., 2014-07-22 04:32:01]

PTAL. Tests fine for me, but please check the arm64 side, I made have made bad
assumption on toolchain name.

[Jakob Kummerow, 2014-07-22 08:53:29]

LGTM, thanks!

One comment for the record: this approach might get confused by "arm" vs.
"arm64". I don't think that's a problem right now; but *if* we find ourselves in
need of yet more logic, I think the next attempt should be an explicit table of
replacements (cf. build/detect_v8_host_arch.py; might even extend that file and
have Make call out to Python).

[rmcilroy, 2014-07-22 09:10:58]

On 2014/07/22 08:53:29, Jakob wrote:
> LGTM, thanks!
> 
> One comment for the record: this approach might get confused by "arm" vs.
> "arm64". I don't think that's a problem right now; but *if* we find ourselves
in
> need of yet more logic, I think the next attempt should be an explicit table
of
> replacements (cf. build/detect_v8_host_arch.py; might even extend that file
and
> have Make call out to Python).

lgtm too (btw, I'm rmcilroy, not mcilroy - not sure who that is :))

[paul.l..., 2014-07-22 15:29:27]

Committed patchset #1 manually as r22531 (presubmit successful).