On Sun, Aug 29, 2021 at 06:46:38PM +0200, Marek Vasut wrote:
> On 8/15/21 9:50 PM, Tom Rini wrote:
> > On Sun, Aug 15, 2021 at 08:13:13PM +0200, Marek Vasut wrote:
> > 
> > > Mark arch_lmb_reserve() weak on architecture level, so it can be
> > > overridden if necessary. This might be necessary if there is some
> > > sort of architectural exception where e.g. more LMB areas have to
> > > be reserved.
> > > 
> > > Note that sandbox now needs an empty implementation of
> > > arch_lmb_reserve().
> > 
> > I'm not sure we need this.  We have the arch and board call-outs, and
> > I've already addressed the imx example.  I really want to see how this
> > works on the I believe you've said rcar example.
> 
> Since they disable relocation in the bsp too, gd->ram_top would be replaced
> by fixed value there. I can skip this patch for now.

OK.

-- 
Tom