soc vs localbus for flash in device tree

soc vs localbus for flash in device tree

[Dushara Jayasinghe]

Hi,

I'm working on the kernel version Linux-2.6.29-rc5 with U-Boot 1.2.0-g88e21=
e7b-dirty.

Does the flash node HAVE to be within a localbus node or can it reside with=
in the soc node? I've been basing my work on mpc834x_mds.dts
And hence haven't defined a localbus node.

BTW is there a way to search the ML archive? Currently I browse each month =
to search for prior postings.

Thanks
D

Re: soc vs localbus for flash in device tree

[Ira Snyder]

On Tue, Feb 24, 2009 at 10:31:37AM +1100, Dushara Jayasinghe wrote:
> Hi,
> 
> I'm working on the kernel version Linux-2.6.29-rc5 with U-Boot 1.2.0-g88e21e7b-dirty.
> 
> Does the flash node HAVE to be within a localbus node or can it reside within the soc node? I've been basing my work on mpc834x_mds.dts
> And hence haven't defined a localbus node.
> 

I'm also working on a custom board based on MPC8349EMDS, and have based
my dts off of mpc834x_mds.dts. I've added the localbus node with my
chipselects and flash. It works, but there are no guarantees of
correctness.

I've appended the important portion below, just in case it helps.
Ira

	localbus@e0005000 {
		#address-cells = <2>;
		#size-cells = <1>;
		compatible = "fsl,mpc8349-elbc", "fsl,elbc", "simple-bus";
		reg = <0xe0005000 0x1000>;
		interrupts = <77 0x8>;
		interrupt-parent = <&ipic>;

		// CS0 and CS1 -- probably wrong...
		ranges = <0x0 0x0 0xfc000000 0x04000000
			  0x1 0x0 0xf0000000 0x00004000>;

		flash@0,0 {
			#address-cells = <1>;
			#size-cells = <1>;
			compatible = "cfi-flash";
			reg = <0x0 0x0 0x04000000>;
			bank-width = <2>;
			device-width = <1>;

			u-boot@0 {
				reg = <0x0 0x40000>;
			};

			u-boot-env@60000 {
				reg = <0x60000 0x20000>;
			};

			sys-fpga@3de0000 {
				reg = <0x3de0000 0x220000>;
			};
		};
	};

RE: soc vs localbus for flash in device tree

[Dushara Jayasinghe]

> > Does the flash node HAVE to be within a localbus node or can it
> reside within the soc node? I've been basing my work on mpc834x_mds.dts
> > And hence haven't defined a localbus node.
> >
>=20
> I'm also working on a custom board based on MPC8349EMDS, and have based
> my dts off of mpc834x_mds.dts. I've added the localbus node with my
> chipselects and flash. It works, but there are no guarantees of
> correctness.
>=20

This worked for me. Thanks.

D

Re: soc vs localbus for flash in device tree

[Michael Bergandi]

[-- Attachment #1: Type: text/plain, Size: 626 bytes --]

>
>
> I'm working on the kernel version Linux-2.6.29-rc5 with U-Boot
> 1.2.0-g88e21e7b-dirty.


I recommend you use a more current version of U-Boot. That one is quite old.


Does the flash node HAVE to be within a localbus node or can it reside
> within the soc node? I've been basing my work on mpc834x_mds.dts
> And hence haven't defined a localbus node.


I think it can be in either, but the most common place I've seen is in the
soc node.


> BTW is there a way to search the ML archive? Currently I browse each month
> to search for prior postings.


Yes, Google "site:ozlabs.org <what ever you are looking for>"

Mike

[-- Attachment #2: Type: text/html, Size: 1220 bytes --]

Re: soc vs localbus for flash in device tree

[Grant Likely]

On Mon, Feb 23, 2009 at 5:11 PM, Michael Bergandi <mbergandi@gmail.com> wrote:
>>
>> I'm working on the kernel version Linux-2.6.29-rc5 with U-Boot
>> 1.2.0-g88e21e7b-dirty.
>
> I recommend you use a more current version of U-Boot. That one is quite old.
>
>> Does the flash node HAVE to be within a localbus node or can it reside
>> within the soc node? I've been basing my work on mpc834x_mds.dts
>> And hence haven't defined a localbus node.
>
> I think it can be in either, but the most common place I've seen is in the
> soc node.

If flash hangs off the local bus, then it belongs in the local bus
node.  It doesn't make any sense to put it in the SoC node because the
SoC node describes the internal memory mapped devices.  The local bus
node should have a ranges property that describes the translation
between physical address and individual chip selects.

g.

-- 
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.