arm: work-around register pressure limitations

arm: work-around register pressure limitations

The recent move to percpu allocator breaks the 32-bit ARM build due to
register pressure. Fix this by using explicit register clobbers and
loading input values from "m" input operands.

Note that old gcc does not support output operands for asm goto, so we
cannot simply re-use an input register as output.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I1d546e427c7f6b62035433815806c8ec275a55f8