[PATCH 4/6] arm64: Add DTS support for FSL's LS2085A SoC

[Catalin Marinas]

On Fri, Aug 15, 2014 at 04:57:13PM +0100, Stuart Yoder wrote:
> > On Fri, Aug 15, 2014 at 03:31:48PM +0100, bhupesh.sharma at freescale.com wrote:
> > > > On Fri, Aug 15, 2014 at 01:53:13PM +0100, Stuart Yoder wrote:
> > > > > > On Fri, Aug 15, 2014 at 10:49:13AM +0100, Bhupesh Sharma wrote:
> > > > > > > +	cpus {
> > > > > > > +		#address-cells = <2>;
> > > > > > > +		#size-cells = <0>;
> > > > > > > +
> > > > > > > +		/* We have 4 clusters having 2 Cortex-A57 cores each */
> > > > > > > +		cpu at 0 {
> > > > > > > +			device_type = "cpu";
> > > > > > > +			compatible = "arm,cortex-a57";
> > > > > > > +			reg = <0x0 0x0>;
> > > > > > > +			enable-method = "spin-table";
> > > > > > > +			cpu-release-addr = <0x0 0x8000fff8>;
> > > > > > > +		};
> > > > > >
> > > > > > Why not PSCI?
> > > > >
> > > > > It simply is where we are today-- we don't have functioning PSCI yet
> > > > > but plan to get there over time.
> > > >
> > > > Thanks for clarification.
> > > >
> > > > > All the existing device trees
> > > > > in arch/arm64 use "spin-table", so it seems that other platforms are
> > > > > in the same situation:
> > > > >    apm-storm.dtsi
> > > >
> > > > Not possible because there is no EL3 mode on the CPU implementation.
> > > >
> > > > >    foundation-v8.dts
> > > > >    rtsm_ve-aemv8a.dts
> > > >
> > > > These work with the latest boot wrapper (which overrides the DT nodes and
> > > > passes the PSCI information).
> > >
> > > Right. We are working on our PSCI handler implementation in EL3 and the spin-
> > table
> > > approach it something we are currently supporting on the simulator model
> > using
> > > u-boot bootloader setting up the spin-tables.
> > >
> > > Our patches for implementing spin-table approach in armv8 u-boot are already
> > under
> > > review and corresponding rework [1].
> > 
> > It would be better if you helped with this effort to bring PSCI to
> > U-Boot (ARMv7 currently but easy to port to AArch64):
> > 
> > https://git.kernel.org/cgit/linux/kernel/git/maz/u-boot.git/log/?h=wip/psci-v4
> > https://git.kernel.org/cgit/linux/kernel/git/maz/u-boot.git/log/?h=wip/psci-v4-
> > a20
> 
> Is the intent to eventually purge the device trees of enable-method/cpu-release-addr
> and have that set by the boot firmware?  Or, keep the spin-table method 
> around as a least common denominator mechanism and override it when
> necessary?   Wondering what the longer term thinking is...

There are only two dts files that would support PSCI but because of
older boot wrappers, we can't change the default. However, the aim for
new platforms is to start with PSCI by default rather than spin-table
and looking at changing them later.

-- 
Catalin