[PATCH Resend] ARM: EXYNOS: Map SYSRAM address through DT

Arnd Bergmann arnd at arndb.de
Wed Apr 30 03:52:56 PDT 2014


On Wednesday 30 April 2014 09:39:54 Sachin Kamat wrote:
> On 16 April 2014 22:55, Heiko Stübner <heiko at sntech.de> wrote:
> > Am Mittwoch, 16. April 2014, 16:35:36 schrieb Arnd Bergmann:
> >> On Wednesday 16 April 2014 17:20:51 Sachin Kamat wrote:
> >> > Instead of hardcoding the SYSRAM details for each SoC,
> >> > pass this information through device tree (DT) and make
> >> > the code SoC agnostic.
> >> >
> >> > Signed-off-by: Sachin Kamat <sachin.kamat at linaro.org>
> >> > ---
> >> > Rebased on latest linux-next.
> >>
> >> Thanks for sending this again. I'd like Heiko to have a look
> >> and provide an Ack if he's happy with it.
> >>
> >> It seems similar to what he did with the SRAM for mach-rockchip,
> >> and if it is we should use the same binding that he introduced,
> >> which would be a minor variation of this.
> >
> > The sram binding is derived from the generic reserved-memory bindings to
> > enable the sram in general to be used generically through the sram driver,
> > while still retaining some areas for special purposes, like the smp-trampoline
> > in my case.
> >
> > From my reading of platsmp.c, it looks like offset+0x4 starts the so called
> > boot-registesr, which get the smp-start-address written to.
> >
> > So I guess it all depends on what is contained in the rest of the sysram. If
> > it is all covered with such special registers or other special uses, the code
> > below is fine. But if the most of the area is just general purpose sram, a
> > solution like on rockchip might be nicer - i.e. handling the sysram via the
> > sram driver and declaring a reserved section for the boot registers.
> 
> Thanks for your inputs. In our case, we use sram for secondary boot
> addresses but could not find any other general purpose use.
> 
> > So, depending on the above:
> > Acked-by: Heiko Stuebner <heiko at sntech.de>
> 
> So I believe your ack applies to our case :). Thanks again.

My interpretation is the opposite. :)

For what I can tell, you use only part of the SRAM for this, and the rest
could be used for something else, you just haven't had the need for it.

If this is the case, you really should be using Heiko's binding instead,
to make it future-proof in case some other use for the SRAM comes up.

	Arnd



More information about the linux-arm-kernel mailing list