[RFC PATCH 2/7] arm/dts: Add initial device tree support for OMAP4 SoC

Rob Herring robherring2 at gmail.com
Thu Aug 25 23:37:08 EST 2011


Benoit,

On 08/24/2011 05:17 AM, Benoit Cousson wrote:
> Add initial device-tree support for OMAP4 SoC.
> 
> This is based on the original panda board patch done by Manju:
> http://permalink.gmane.org/gmane.linux.ports.arm.omap/60393
> 
> Add the generic GIC interrupt-controller from ARM.
> 
> Add an empty "soc" node to contain non memory mapped IPs
> (DSP, MPU, IPU...).
> 
> Note: Since reg, irq and dma are provided by hwmod for the
> moment, these attributes will not be present at all in DTS
> to highlight the gap. They will be added as soon as dma bindings
> will be there and drivers will be adapted.
> 
> Signed-off-by: Benoit Cousson <b-cousson at ti.com>
> Cc: Grant Likely <grant.likely at secretlab.ca>
> Cc: G, Manjunath Kondaiah <manjugk at ti.com>
> ---
>  arch/arm/boot/dts/omap4.dtsi |   80 ++++++++++++++++++++++++++++++++++++++++++
>  1 files changed, 80 insertions(+), 0 deletions(-)
>  create mode 100644 arch/arm/boot/dts/omap4.dtsi
> 
> diff --git a/arch/arm/boot/dts/omap4.dtsi b/arch/arm/boot/dts/omap4.dtsi
> new file mode 100644
> index 0000000..97a3ea7
> --- /dev/null
> +++ b/arch/arm/boot/dts/omap4.dtsi
> @@ -0,0 +1,80 @@
> +/*
> + * Copyright (C) 2011 Texas Instruments Incorporated - http://www.ti.com/
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + */
> +
> +/*
> + * Carveout for multimedia usecases
> + * It should be the last 48MB of the first 512MB memory part
> + * In theory, it should not even exist. That zone should be reserved
> + * dynamically during the .reserve callback.
> + */
> +/memreserve/ 0x9d000000 0x03000000;
> +
> +/include/ "skeleton.dtsi"
> +
> +/ {
> +	model = "TI OMAP4430";
> +	compatible = "ti,omap4430", "ti,omap4";
> +	interrupt-parent = <&gic>;
> +
> +	aliases {
> +	};
> +
> +	/*
> +	 * The soc node represents the soc top level view. It is uses for IPs
> +	 * that are not memory mapped in the MPU view or for the MPU itself.
> +	 */
> +	soc {
> +		compatible = "ti,omap-infra";
> +	};
> +
> +	/*
> +	 * XXX: Use a flat representation of the OMAP4 interconnect.
> +	 * The real OMAP interconnect network is quite complex.
> +	 *
> +	 * MPU -+-- MPU_PRIVATE - GIC, L2
> +	 *      |
> +	 *      +----------------+----------+
> +	 *      |                |          |
> +	 *      +            +- EMIF - DDR  |
> +	 *      |            |              |
> +	 *      |            +     +--------+
> +	 *      |            |     |
> +	 *      |            +- L4_ABE - AESS, MCBSP, TIMERs...
> +	 *      |            |
> +	 *      +- L3_MAIN --+- L4_CORE - IPs...
> +	 *                   |
> +	 *                   +- L4_PER - IPs...
> +	 *                   |
> +	 *                   +- L4_CFG -+- L4_WKUP - IPs...
> +	 *                   |          |
> +	 *                   |          +- IPs...
> +	 *                   +- IPU ----+
> +	 *                   |          |
> +	 *                   +- DSP ----+
> +	 *                   |          |
> +	 *                   +- DSS ----+
> +	 *
> +	 * Since that will not bring real advantage to represent that in DT for
> +	 * the moment, just use a fake OCP bus entry to represent the whole bus
> +	 * hierarchy.
> +	 */
> +	ocp {
> +		compatible = "simple-bus";
> +		#address-cells = <1>;
> +		#size-cells = <1>;
> +		ranges;
> +
> +		gic: interrupt-controller at 48241000 {
> +			compatible = "ti,omap4-gic", "arm,gic";

The gic binding is still being hashed out. This needs binding
documentation and handling of PPIs. I'm planning on posting an updated
series today with this.

"arm,gic" should be dropped or replaced with "arm,cortex-a9-gic".
Non-specific DT bindings are not well received. Is OMAP4 gic different
from standard Cortex-A9?

> +			interrupt-controller;
> +			#interrupt-cells = <1>;
> +			reg = <0x48241000 0x1000>,
> +			      <0x48240100 0x0200>;

Isn't the cpu interface register space 0x100 bytes long?

Rob

> +		};
> +	};
> +};



More information about the devicetree-discuss mailing list