[PATCH] Add device file bindings for MAPLE

[PATCH] Add device file bindings for MAPLE

[Shaveta Leekha]

Signed-off-by: Shaveta Leekha <shaveta@freescale.com>
---
 .../devicetree/bindings/powerpc/fsl/maple.txt      |   50 ++++++++++++++++++++
 1 files changed, 50 insertions(+), 0 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/powerpc/fsl/maple.txt

diff --git a/Documentation/devicetree/bindings/powerpc/fsl/maple.txt b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
new file mode 100644
index 0000000..23b80a7
--- /dev/null
+++ b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
@@ -0,0 +1,50 @@
+* Freescale MAPLE Multi Accelerator Platform Engine Baseband 3
+  (MAPLE-B3)device nodes
+
+Supported chips:
+Example: B4860
+
+Required properties:
+
+- compatible:	Should contain "fsl,maple-b3" as the value
+		This identifies Multi Accelerator Platform Engine
+		Baseband 3 block.
+
+- reg:		offset and length of the register set for the device
+
+- interrupts
+	Usage: required
+	Value type: <prop_encoded-array>
+	Definition:  Specifies the interrupts generated by this device.  The
+			value of the interrupts property consists of one interrupt
+			specifier. The format of the specifier is defined by the
+			binding document describing the node's interrupt parent.
+
+		A single IRQ that handles error conditions is specified by
+		this property.  (Typically shared with port-write).
+
+Devices that have LIODNs need to specify links to the parent PAMU controller
+(the actual PAMU controller that this device is connected to) and a pointer to
+the LIODN register, if applicable.
+
+- fsl,iommu-parent
+		: <phandle>
+		This property should be present
+
+- status = "disabled"
+		In this example, status is set "disabled",
+		As Maple device wouldn't be used by PPC Linux. This representation is required
+		for doing the PAMU programming on the Linux side.
+
+Example:
+	/* B4860 */
+
+	maple@800000 {
+		#address-cells = <0>;
+		#size-cells = <0>;
+		status = "disabled";
+		compatible = "fsl,maple-b3";
+		reg = <0x8000000 0x10000>;
+		interrupts = <16 2 1 18>;
+		fsl,iommu-parent = <&pamu1>;
+	};
-- 
1.7.6.GIT

Re: [PATCH] Add device file bindings for MAPLE

[Scott Wood]

On 07/27/2013 08:03:42 AM, Shaveta Leekha wrote:
> Signed-off-by: Shaveta Leekha <shaveta@freescale.com>
> ---
>  .../devicetree/bindings/powerpc/fsl/maple.txt      |   50 =20
> ++++++++++++++++++++
>  1 files changed, 50 insertions(+), 0 deletions(-)
>  create mode 100644 =20
> Documentation/devicetree/bindings/powerpc/fsl/maple.txt
>=20
> diff --git a/Documentation/devicetree/bindings/powerpc/fsl/maple.txt =20
> b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> new file mode 100644
> index 0000000..23b80a7
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> @@ -0,0 +1,50 @@
> +* Freescale MAPLE Multi Accelerator Platform Engine Baseband 3
> +  (MAPLE-B3)device nodes
> +
> +Supported chips:
> +Example: B4860
> +
> +Required properties:
> +
> +- compatible:	Should contain "fsl,maple-b3" as the value
> +		This identifies Multi Accelerator Platform Engine
> +		Baseband 3 block.
> +
> +- reg:		offset and length of the register set for the =20
> device
> +
> +- interrupts
> +	Usage: required
> +	Value type: <prop_encoded-array>
> +	Definition:  Specifies the interrupts generated by this =20
> device.  The
> +			value of the interrupts property consists of =20
> one interrupt
> +			specifier. The format of the specifier is =20
> defined by the
> +			binding document describing the node's =20
> interrupt parent.

Please clean up whitespace so the text aligns.

> +
> +		A single IRQ that handles error conditions is specified =20
> by
> +		this property.  (Typically shared with port-write).
> +
> +Devices that have LIODNs need to specify links to the parent PAMU =20
> controller
> +(the actual PAMU controller that this device is connected to) and a =20
> pointer to
> +the LIODN register, if applicable.
> +
> +- fsl,iommu-parent
> +		: <phandle>
> +		This property should be present
> +
> +- status =3D "disabled"
> +		In this example, status is set "disabled",
> +		As Maple device wouldn't be used by PPC Linux. This =20
> representation is required
> +		for doing the PAMU programming on the Linux side.

Is it proper for Linux to even set the LIODN on a device that is =20
"disabled"?  Perhaps the compatible should be different, such as =20
"fsl,maple-b3-liodn" to indicate that that is the only portion of the =20
device that is owned by this partition.

> +Example:
> +	/* B4860 */
> +
> +	maple@800000 {
> +		#address-cells =3D <0>;
> +		#size-cells =3D <0>;
> +		status =3D "disabled";
> +		compatible =3D "fsl,maple-b3";
> +		reg =3D <0x8000000 0x10000>;
> +		interrupts =3D <16 2 1 18>;
> +		fsl,iommu-parent =3D <&pamu1>;
> +	};

This error interrupt is documented as 13, not 18.  b4860 (and maybe =20
other b4?) have an erratum that says that error interrupts are =20
reversed, but this is generally implemented in b4si-post.dtsi, not in =20
block-specific include files.  Will no chip without the erratum ever =20
have maple?

-Scott=

Re: [PATCH] Add device file bindings for MAPLE

[Benjamin Herrenschmidt]

On Mon, 2013-07-29 at 19:00 -0500, Scott Wood wrote:
> On 07/27/2013 08:03:42 AM, Shaveta Leekha wrote:
> > Signed-off-by: Shaveta Leekha <shaveta@freescale.com>
> > ---
> >  .../devicetree/bindings/powerpc/fsl/maple.txt      |   50  
> > ++++++++++++++++++++

BTW, that's confusing, we have a platform called maple ... oh well...

Cheers,
Ben.

RE: [PATCH] Add device file bindings for MAPLE

[Leekha Shaveta-B20052]

-----Original Message-----
From: Wood Scott-B07421=20
Sent: Tuesday, July 30, 2013 5:31 AM
To: Leekha Shaveta-B20052
Cc: devicetree-discuss@lists.ozlabs.org; linuxppc-dev@lists.ozlabs.org; Lee=
kha Shaveta-B20052
Subject: Re: [PATCH] Add device file bindings for MAPLE

On 07/27/2013 08:03:42 AM, Shaveta Leekha wrote:
> Signed-off-by: Shaveta Leekha <shaveta@freescale.com>
> ---
>  .../devicetree/bindings/powerpc/fsl/maple.txt      |   50 =20
> ++++++++++++++++++++
>  1 files changed, 50 insertions(+), 0 deletions(-)  create mode 100644=20
> Documentation/devicetree/bindings/powerpc/fsl/maple.txt
>=20
> diff --git a/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> new file mode 100644
> index 0000000..23b80a7
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> @@ -0,0 +1,50 @@
> +* Freescale MAPLE Multi Accelerator Platform Engine Baseband 3
> +  (MAPLE-B3)device nodes
> +
> +Supported chips:
> +Example: B4860
> +
> +Required properties:
> +
> +- compatible:	Should contain "fsl,maple-b3" as the value
> +		This identifies Multi Accelerator Platform Engine
> +		Baseband 3 block.
> +
> +- reg:		offset and length of the register set for the =20
> device
> +
> +- interrupts
> +	Usage: required
> +	Value type: <prop_encoded-array>
> +	Definition:  Specifies the interrupts generated by this
> device.  The
> +			value of the interrupts property consists of
> one interrupt
> +			specifier. The format of the specifier is
> defined by the
> +			binding document describing the node's
> interrupt parent.

Please clean up whitespace so the text aligns.
[Leekha Shaveta-B20052] ok. Will remove whitespaces.

> +
> +		A single IRQ that handles error conditions is specified
> by
> +		this property.  (Typically shared with port-write).
> +
> +Devices that have LIODNs need to specify links to the parent PAMU
> controller
> +(the actual PAMU controller that this device is connected to) and a
> pointer to
> +the LIODN register, if applicable.
> +
> +- fsl,iommu-parent
> +		: <phandle>
> +		This property should be present
> +
> +- status =3D "disabled"
> +		In this example, status is set "disabled",
> +		As Maple device wouldn't be used by PPC Linux. This
> representation is required
> +		for doing the PAMU programming on the Linux side.

Is it proper for Linux to even set the LIODN on a device that is "disabled"=
?  Perhaps the compatible should be different, such as "fsl,maple-b3-liodn"=
 to indicate that that is the only portion of the device that is owned by t=
his partition.

> +Example:
> +	/* B4860 */
> +
> +	maple@800000 {
> +		#address-cells =3D <0>;
> +		#size-cells =3D <0>;
> +		status =3D "disabled";
> +		compatible =3D "fsl,maple-b3";
> +		reg =3D <0x8000000 0x10000>;
> +		interrupts =3D <16 2 1 18>;
> +		fsl,iommu-parent =3D <&pamu1>;
> +	};

This error interrupt is documented as 13, not 18.  b4860 (and maybe other b=
4?) have an erratum that says that error interrupts are reversed, but this =
is generally implemented in b4si-post.dtsi, not in block-specific include f=
iles.  Will no chip without the erratum ever have maple?
[Leekha Shaveta-B20052] You are right here. Will add the interrupts in b4si=
 or b4860si file, as other chips may have maple.
Regards,
Shaveta

-Scott

[PATCH] Add device file bindings for MAPLE

[Shaveta Leekha]

Signed-off-by: Shaveta Leekha <shaveta@freescale.com>
---
 .../devicetree/bindings/powerpc/fsl/maple.txt      |   30 ++++++++++++++++++++
 1 files changed, 30 insertions(+), 0 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/powerpc/fsl/maple.txt

diff --git a/Documentation/devicetree/bindings/powerpc/fsl/maple.txt b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
new file mode 100644
index 0000000..da51c5f
--- /dev/null
+++ b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
@@ -0,0 +1,30 @@
+* Freescale MAPLE Multi Accelerator Platform Engine Baseband 3
+  (MAPLE-B3)device nodes
+
+Supported chips:
+Example: B4860
+
+Required properties:
+
+- compatible:	Should contain "fsl,maple-b3-liodn" as the value
+		This identifies Multi Accelerator Platform Engine
+		Baseband 3 block.This representation is required
+		for doing the PAMU/LIODN programming on the Linux side.
+
+- reg:		offset and length of the register set for the device
+
+Devices that have LIODNs need to specify links to the parent PAMU controller
+(the actual PAMU controller that this device is connected to) and a pointer to
+the LIODN register, if applicable.
+
+- fsl,iommu-parent
+		: <phandle>
+		This property should be present
+
+Example:
+	/* B4860 */
+	maple@800000 {
+		compatible = "fsl,maple-b3-liodn";
+		reg = <0x8000000 0x10000>;
+		fsl,iommu-parent = <&pamu1>;
+	};
-- 
1.7.6.GIT

Re: [PATCH] Add device file bindings for MAPLE

[Kumar Gala]

On Aug 1, 2013, at 6:02 AM, Shaveta Leekha wrote:

> Signed-off-by: Shaveta Leekha <shaveta@freescale.com>
> ---
> .../devicetree/bindings/powerpc/fsl/maple.txt      |   30 =
++++++++++++++++++++
> 1 files changed, 30 insertions(+), 0 deletions(-)
> create mode 100644 =
Documentation/devicetree/bindings/powerpc/fsl/maple.txt
>=20
> diff --git a/Documentation/devicetree/bindings/powerpc/fsl/maple.txt =
b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> new file mode 100644
> index 0000000..da51c5f
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> @@ -0,0 +1,30 @@
> +* Freescale MAPLE Multi Accelerator Platform Engine Baseband 3
> +  (MAPLE-B3)device nodes
> +
> +Supported chips:
> +Example: B4860
> +
> +Required properties:
> +
> +- compatible:	Should contain "fsl,maple-b3-liodn" as the value
> +		This identifies Multi Accelerator Platform Engine
> +		Baseband 3 block.This representation is required
> +		for doing the PAMU/LIODN programming on the Linux side.

This compatible makes no sense, we shouldn't be marking a full HW block =
with some name that is just intended for LIODN convenance.

Is this version 3 of the block?  If so a name like "fsl,maple-v3" or =
"fsl,maple-v3.0" would be more appropriate.

- k

> +
> +- reg:		offset and length of the register set for the =
device
> +
> +Devices that have LIODNs need to specify links to the parent PAMU =
controller
> +(the actual PAMU controller that this device is connected to) and a =
pointer to
> +the LIODN register, if applicable.


Does Maple not have any IRQs associated with it?

If you are going to have a minimal binding, at least cover the basics =
associated with a device.

> +
> +- fsl,iommu-parent
> +		: <phandle>
> +		This property should be present
> +
> +Example:
> +	/* B4860 */
> +	maple@800000 {
> +		compatible =3D "fsl,maple-b3-liodn";
> +		reg =3D <0x8000000 0x10000>;
> +		fsl,iommu-parent =3D <&pamu1>;
> +	};
> --=20
> 1.7.6.GIT
>=20
>=20
> _______________________________________________
> Linuxppc-dev mailing list
> Linuxppc-dev@lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/linuxppc-dev

Re: [PATCH] Add device file bindings for MAPLE

[Scott Wood]

On Thu, 2013-08-01 at 11:05 -0500, Kumar Gala wrote:
> On Aug 1, 2013, at 6:02 AM, Shaveta Leekha wrote:
> 
> > Signed-off-by: Shaveta Leekha <shaveta@freescale.com>
> > ---
> > .../devicetree/bindings/powerpc/fsl/maple.txt      |   30 ++++++++++++++++++++
> > 1 files changed, 30 insertions(+), 0 deletions(-)
> > create mode 100644 Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> > 
> > diff --git a/Documentation/devicetree/bindings/powerpc/fsl/maple.txt b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> > new file mode 100644
> > index 0000000..da51c5f
> > --- /dev/null
> > +++ b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> > @@ -0,0 +1,30 @@
> > +* Freescale MAPLE Multi Accelerator Platform Engine Baseband 3
> > +  (MAPLE-B3)device nodes
> > +
> > +Supported chips:
> > +Example: B4860
> > +
> > +Required properties:
> > +
> > +- compatible:	Should contain "fsl,maple-b3-liodn" as the value
> > +		This identifies Multi Accelerator Platform Engine
> > +		Baseband 3 block.This representation is required
> > +		for doing the PAMU/LIODN programming on the Linux side.
> 
> This compatible makes no sense, we shouldn't be marking a full HW block with some name that is just intended for LIODN convenance.

The point is we're not describing the whole block here, because the rest
of the block is owned by an external entity (the DSP cores).

> Is this version 3 of the block?  If so a name like "fsl,maple-v3" or "fsl,maple-v3.0" would be more appropriate.
> 
> - k
> 
> > +
> > +- reg:		offset and length of the register set for the device
> > +
> > +Devices that have LIODNs need to specify links to the parent PAMU controller
> > +(the actual PAMU controller that this device is connected to) and a pointer to
> > +the LIODN register, if applicable.
> 
> 
> Does Maple not have any IRQs associated with it?

maple-liodn doesn't.

-Scott

Re: [PATCH] Add device file bindings for MAPLE

[Kumar Gala]

On Aug 5, 2013, at 4:11 PM, Scott Wood wrote:

> On Thu, 2013-08-01 at 11:05 -0500, Kumar Gala wrote:
>> On Aug 1, 2013, at 6:02 AM, Shaveta Leekha wrote:
>>=20
>>> Signed-off-by: Shaveta Leekha <shaveta@freescale.com>
>>> ---
>>> .../devicetree/bindings/powerpc/fsl/maple.txt      |   30 =
++++++++++++++++++++
>>> 1 files changed, 30 insertions(+), 0 deletions(-)
>>> create mode 100644 =
Documentation/devicetree/bindings/powerpc/fsl/maple.txt
>>>=20
>>> diff --git a/Documentation/devicetree/bindings/powerpc/fsl/maple.txt =
b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
>>> new file mode 100644
>>> index 0000000..da51c5f
>>> --- /dev/null
>>> +++ b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
>>> @@ -0,0 +1,30 @@
>>> +* Freescale MAPLE Multi Accelerator Platform Engine Baseband 3
>>> +  (MAPLE-B3)device nodes
>>> +
>>> +Supported chips:
>>> +Example: B4860
>>> +
>>> +Required properties:
>>> +
>>> +- compatible:	Should contain "fsl,maple-b3-liodn" as the value
>>> +		This identifies Multi Accelerator Platform Engine
>>> +		Baseband 3 block.This representation is required
>>> +		for doing the PAMU/LIODN programming on the Linux side.
>>=20
>> This compatible makes no sense, we shouldn't be marking a full HW =
block with some name that is just intended for LIODN convenance.
>=20
> The point is we're not describing the whole block here, because the =
rest
> of the block is owned by an external entity (the DSP cores).

So the binding should make that far more clear that the MAPLE would be =
controlled by DSP SW.

>=20
>> Is this version 3 of the block?  If so a name like "fsl,maple-v3" or =
"fsl,maple-v3.0" would be more appropriate.
>>=20
>> - k
>>=20
>>> +
>>> +- reg:		offset and length of the register set for the =
device
>>> +

So the reg should limit itself to the LIODN registers only.

>>> +Devices that have LIODNs need to specify links to the parent PAMU =
controller
>>> +(the actual PAMU controller that this device is connected to) and a =
pointer to
>>> +the LIODN register, if applicable.
>>=20
>>=20
>> Does Maple not have any IRQs associated with it?
>=20
> maple-liodn doesn't.
>=20
> -Scott
>=20
>=20

Re: [PATCH] Add device file bindings for MAPLE

[Scott Wood]

On Thu, 2013-08-01 at 16:32 +0530, Shaveta Leekha wrote:
> Signed-off-by: Shaveta Leekha <shaveta@freescale.com>
> ---
>  .../devicetree/bindings/powerpc/fsl/maple.txt      |   30 ++++++++++++++++++++
>  1 files changed, 30 insertions(+), 0 deletions(-)
>  create mode 100644 Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> 
> diff --git a/Documentation/devicetree/bindings/powerpc/fsl/maple.txt b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> new file mode 100644
> index 0000000..da51c5f
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/powerpc/fsl/maple.txt
> @@ -0,0 +1,30 @@
> +* Freescale MAPLE Multi Accelerator Platform Engine Baseband 3
> +  (MAPLE-B3)device nodes
> +
> +Supported chips:
> +Example: B4860
> +
> +Required properties:
> +
> +- compatible:	Should contain "fsl,maple-b3-liodn" as the value
> +		This identifies Multi Accelerator Platform Engine
> +		Baseband 3 block.This representation is required
> +		for doing the PAMU/LIODN programming on the Linux side.

Device tree bindings should not talk about particular OSes or use cases
(except for use case information that is part of the AMP partitioning,
and thus not at the discretion of the OS).  Just say that this
identifies the LIODN register of the block.

> +
> +- reg:		offset and length of the register set for the device
> +
> +Devices that have LIODNs need to specify links to the parent PAMU controller
> +(the actual PAMU controller that this device is connected to) and a pointer to
> +the LIODN register, if applicable.
> +
> +- fsl,iommu-parent
> +		: <phandle>
> +		This property should be present
> +
> +Example:
> +	/* B4860 */
> +	maple@800000 {
> +		compatible = "fsl,maple-b3-liodn";
> +		reg = <0x8000000 0x10000>;

Unit address doesn't match reg (missing/extra zero).

-Scott