[SPARC64]: Refine code sequences to get the cpu id.
authorDavid S. Miller <davem@davemloft.net>
Mon, 27 Feb 2006 07:27:19 +0000 (23:27 -0800)
committerDavid S. Miller <davem@sunset.davemloft.net>
Mon, 20 Mar 2006 09:11:35 +0000 (01:11 -0800)
commit92704a1c63c3b481870d02636d0b5a70c7e21cd1
tree098f96da6ab50a1d878425e2b91a9cf22f78ac80
parentf4e841da30b4bcbb8f1cc20a01157a788ff58b21
[SPARC64]: Refine code sequences to get the cpu id.

On uniprocessor, it's always zero for optimize that.

On SMP, the jmpl to the stub kills the return address stack in the cpu
branch prediction logic, so expand the code sequence inline and use a
code patching section to fix things up.  This also always better and
explicit register selection, which will be taken advantage of in a
future changeset.

The hard_smp_processor_id() function is big, so do not inline it.

Fix up tests for Jalapeno to also test for Serrano chips too.  These
tests want "jbus Ultra-IIIi" cases to match, so that is what we should
test for.

Signed-off-by: David S. Miller <davem@davemloft.net>
arch/sparc64/kernel/entry.S
arch/sparc64/kernel/irq.c
arch/sparc64/kernel/setup.c
arch/sparc64/kernel/smp.c
arch/sparc64/kernel/traps.c
arch/sparc64/kernel/vmlinux.lds.S
include/asm-sparc64/cpudata.h
include/asm-sparc64/head.h
include/asm-sparc64/smp.h