[RFC PATCH 0/8] Introduce logical CPU mapping

Will Deacon will.deacon at arm.com
Wed Aug 10 04:58:24 EDT 2011


On Wed, Aug 10, 2011 at 09:50:08AM +0100, Santosh wrote:
> Will,

Hi Santosh,

> On Wednesday 10 August 2011 03:12 AM, Will Deacon wrote:
> > Hello,
> >
> > This patch series is an RFC following a discussion I had with Russell
> > the other day:
> >
> >   http://lists.infradead.org/pipermail/linux-arm-kernel/2011-August/060370.html
> >
> > The approach I've taken is fairly simple but is enough for booting on a
> > CPU with physical ID != 0 on platforms which can handle it.
> >
> > Note that I haven't touched the OMAP4 or Tegra platforms since it's not
> > clear from the code how they would be adapted to make use of the
> > physical CPU number.
> >
> On OMAP, we use hardware register hand-shake for boot between a boot CPU
> and non-boot CPU. On OMAP4, boot CPU has to be always CPU0 to ensure
> that security and firewall initialisation is taken care. Other CPU can't
> do those things.

Understood. We can leave OMAP like it is then and assume that bootloaders
will always ensure physical CPU0 is used for booting.

Cheers,

Will



More information about the linux-arm-kernel mailing list