For SMP, Is platform specific secondary startup function really needed?

[Catalin Marinas]

On Wed, 2011-04-06 at 14:48 +0100, John Linn wrote:
> I'm doing some cleanup of SMP patches based on Russell and Catalin's
> input which was real helpful. It seems to me that there is not always a
> need for a platform specific secondary startup.
> 
> I have tested with just letting CPU1 go straight to secondary_startup()
> in head.S and it works fine.
> 
> I'm probably missing something obvious, if so can you share it?

We use the pen release mechanism serialise the secondary CPUs booting.

-- 
Catalin