On Wed, 05 Feb 2014 13:05:14 +0100, Heiko St��bner [off-list ref] wrote:

Am Mittwoch, 5. Februar 2014, 11:12:47 schrieb Grant Likely:

quoted

On Mon, 20 Jan 2014 16:42:58 +0100, Heiko St����bner [off-list ref] wrote:

quoted

Some SoCs need parts of their sram for special purposes. So while being
part of the peripheral, it should not be part of the genpool controlling
the sram.

Therefore add an option mmio-sram-reserved to keep arbitrary portions of
the sram from general usage.

Suggested-by: Rob Herring <redacted>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Tested-by: Ulrich Prinz <redacted>
Acked-by: Rob Herring <robh@kernel.org>
---

 Documentation/devicetree/bindings/misc/sram.txt |    8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/Documentation/devicetree/bindings/misc/sram.txt

b/Documentation/devicetree/bindings/misc/sram.txt index 4d0a00e..09ee7a3
100644

--- a/Documentation/devicetree/bindings/misc/sram.txt
+++ b/Documentation/devicetree/bindings/misc/sram.txt

@@ -8,9 +8,17 @@ Required properties:
 - reg : SRAM iomem address range

+Optional properties:
+
+- mmio-sram-reserved: ordered list of reserved chunks inside the sram

that
+  should not be used by the operating system.
+  Format is <base size>, <base size>, ...; with base being relative to
the
+  reg property base.
+

We've now got a draft binding for reserved memory. Can you use the format
here? Basically each reserved region is a sub node with either a reg
property or a size property.

This is specifically for sram, so I won't make a big deal about it, but
it would be good to have some commonality.

I guess you're talking about "[PATCH v2 0/5] reserved-memory regions/CMA in 
devicetree, again", right?

In general I'm all for commonality :-). So I guess you mean it to look 
something like the following:

	sram: sram@10080000 {
		compatible = "rockchip,rk3066-sram", "mmio-sram";
		reg = <0x10080000 0x8000>;

		reserved-memory {
			#address-cells = <1>;
			#size-cells = <1>;

			smp@200 {
				/* hmm, relative or absolute, aka 0x200 or 0x10080200? */
				reg = <0x200 0x50>;
			};
		};
	};

As it looks like only a slight modification of my "parsing" code this should be 
doable. Do you suggest more changes to the example above?

I think the reserved-memory node can actually be dropped in this case.
The only reason we had it for main memory is to deal with regions that
traverse multiple memory nodes. You don't have that case so you can make
the smp@200 node a direct child of SRAM. All the semantics would be the
same otherwise.

It would be useful to allow compatible values or phandle references to
the reserved regions, and the SRAM driver would be responsible to avoid
any region defined unless told explicitly that the region can be used.
Not using a child region is a sensible default for any SRAM driver when
it doesn't have any information otherwise.

g.