[PATCH 4/4] RealView: Add sparsemem support for the RealView PBX platform

Catalin Marinas catalin.marinas at arm.com
Mon Nov 9 05:34:06 EST 2009


On Mon, 2009-11-09 at 10:18 +0000, Russell King - ARM Linux wrote:
> On Mon, Nov 09, 2009 at 09:51:40AM +0000, Catalin Marinas wrote:
> > On Sun, 2009-11-08 at 17:45 +0000, Russell King - ARM Linux wrote:
> > > On Thu, Oct 15, 2009 at 01:22:09PM +0100, Catalin Marinas wrote:
> > > > +config ARCH_FLATMEM_ENABLE
> > > > +	bool
> > > > +	default y if ARCH_SPARSEMEM_ENABLE
> > > > +
> > > 
> > > I seem to have missed this.  Why do we want to enable flatmem support for
> > > _everything_ which enables sparsemem?
> > 
> > Just to have the option of selecting either flatmem or sparsemem with
> > PBX (though I don't have a strong opinion about this). This can also be
> > done transparently via REALVIEW_HIGH_PHYS_OFFSET.
> 
> If you want to do this, put ARCH_FLATMEM_ENABLE into arch/arm/Kconfig,
> and arrange for the platform(s) where this is optional _and_ not the
> default to select this symbol, in precisely the same way that
> ARCH_SPARSEMEM_ENABLE is handled today.

I don't think it's worth the hassle anyway. If you have
REALVIEW_HIGH_PHYS_OFFSET disabled and FLATMEM, you end up with only
256MB of RAM available, I don't really see a point in this
configuration. With the new patch, if you don't like SPARSEMEM (maybe
for benchmarking reasons), just enable REALVIEW_HIGH_PHYS_OFFSET.

So, are you OK with the latest patch?

-- 
Catalin




More information about the linux-arm-kernel mailing list