Re: [PATCH 03/24] arm: mm: add proc info for ScorpionMP

[Catalin Marinas <catalin.marinas@arm.com>]

On Fri, 2010-08-27 at 16:29 +0100, Daniel Walker wrote:
> On Fri, 2010-08-27 at 14:54 +0100, Catalin Marinas wrote:
> > On Wed, 2010-08-25 at 05:57 +0100, Jeff Ohlstein wrote:
> > > From: Daniel Walker <dwalker@codeaurora.org>
> > >
> > > ScorpionMP does not have the SMP/nAMP and TLB ops broadcasting bits in
> > > ACTLR.
> > >
> > > Signed-off-by: Daniel Walker <dwalker@codeaurora.org>
> > > Signed-off-by: Jeff Ohlstein <johlstei@codeaurora.org>
> > > ---
> > >  arch/arm/mm/proc-v7.S |   24 ++++++++++++++++++++++++
> > >  1 files changed, 24 insertions(+), 0 deletions(-)
> > >
> > > diff --git a/arch/arm/mm/proc-v7.S b/arch/arm/mm/proc-v7.S
> > > index 7aaf88a..98fd7e5 100644
> > > --- a/arch/arm/mm/proc-v7.S
> > > +++ b/arch/arm/mm/proc-v7.S
> > > @@ -196,6 +196,7 @@ __v7_setup:
> > >         orreq   r0, r0, #(1 << 6) | (1 << 0)    @ Enable SMP/nAMP mode and
> > >         mcreq   p15, 0, r0, c1, c0, 1           @ TLB ops broadcasting
> > >  #endif
> > > +__v7_msm_setup:
> > >         adr     r12, __v7_setup_stack           @ the local stack
> > >         stmia   r12, {r0-r5, r7, r9, r11, lr}
> > >         bl      v7_flush_dcache_all
> >
> > It may be better to move the generic __v7_setup here and have a specific
> > entry for ARM Ltd cores (mask 0xff0f0000 and value 0x410f0000).
> 
> I'm not sure I follow you.. Are you saying there are other cpu's that
> don't need the SMP stuff at the top of __v7_setup ?

The SMP/nAMP bit is something specific to the ARM11MPCore and Cortex-A9
processors produced by ARM. If you core doesn't need this, it should use
a default __v7_setup without any additional bits being set.

I'm not aware of other MP ARM processors produced by other companies
that would require exactly the same bit being set.

> > In general, I would have liked a different way (still ID-based) of
> > handling such issues without duplicating the whole structure. But I see
> > this becoming the standard as proc-v6.S follows it as well.
> 
> This whole thing was something Russell specifically asked for (and
> practically wrote for me).

Yes, I know and I'm not saying that you should change it in this patch.
It's just that I would have liked to only duplicate the setup function
rather than the whole procinfo structure.

-- 
Catalin