[PATCH v1] arm64: dts: freescale: imx95-toradex-smarc: Support Cortex M7

[PATCH v1] arm64: dts: freescale: imx95-toradex-smarc: Support Cortex M7

[Emanuele Ghidoli]

From: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>

Enable Cortex M7, the vring nodes, a mailbox and reserve DDR memory for
the M7. The remoteproc framework is so capable to load and run the M7
firmware.

Signed-off-by: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
---
 .../dts/freescale/imx95-toradex-smarc.dtsi    | 50 +++++++++++++++++++
 1 file changed, 50 insertions(+)

diff --git a/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi b/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi
index 5932ba238a8a..77c4a6dee098 100644
--- a/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi
+++ b/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi
@@ -42,6 +42,16 @@ clk_serdes_eth_ref: clock-eth-ref {
 		enable-gpios = <&som_gpio_expander_1 13 GPIO_ACTIVE_HIGH>;
 	};
 
+	cm7: remoteproc-cm7 {
+		compatible = "fsl,imx95-cm7";
+		mbox-names = "tx", "rx", "rxdb";
+		mboxes = <&mu7 0 1
+			  &mu7 1 1
+			  &mu7 3 1>;
+		memory-region = <&vdevbuffer>, <&vdev0vring0>, <&vdev0vring1>,
+				<&vdev1vring0>, <&vdev1vring1>, <&rsc_table>, <&m7_reserved>;
+	};
+
 	connector {
 		compatible = "gpio-usb-b-connector", "usb-b-connector";
 		/* SMARC P64 - USB0_OTG_ID */
@@ -156,6 +166,42 @@ linux_cma: linux,cma {
 			alloc-ranges = <0 0x80000000 0 0x7f000000>;
 			linux,cma-default;
 		};
+
+		m7_reserved: memory@80000000 {
+			reg = <0 0x80000000 0 0x1000000>;
+			no-map;
+		};
+
+		rsc_table: memory@88220000 {
+			reg = <0 0x88220000 0 0x1000>;
+			no-map;
+		};
+
+		vdev0vring0: memory@88000000 {
+			reg = <0 0x88000000 0 0x8000>;
+			no-map;
+		};
+
+		vdev0vring1: memory@88008000 {
+			reg = <0 0x88008000 0 0x8000>;
+			no-map;
+		};
+
+		vdev1vring0: memory@88010000 {
+			reg = <0 0x88010000 0 0x8000>;
+			no-map;
+		};
+
+		vdev1vring1: memory@88018000 {
+			reg = <0 0x88018000 0 0x8000>;
+			no-map;
+		};
+
+		vdevbuffer: memory@88020000 {
+			compatible = "shared-dma-pool";
+			reg = <0 0x88020000 0 0x100000>;
+			no-map;
+		};
 	};
 };
 
@@ -572,6 +618,10 @@ &lpuart3 {
 	pinctrl-0 = <&pinctrl_uart3>;
 };
 
+&mu7 {
+	status = "okay";
+};
+
 /* SMARC MDIO, shared between all ethernet ports */
 &netc_emdio {
 	pinctrl-names = "default";
-- 
2.43.0

Re: [PATCH v1] arm64: dts: freescale: imx95-toradex-smarc: Support Cortex M7

[Frank Li]

On Tue, Mar 03, 2026 at 01:13:06PM +0100, Emanuele Ghidoli wrote:
> From: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
>
> Enable Cortex M7, the vring nodes, a mailbox and reserve DDR memory for
> the M7. The remoteproc framework is so capable to load and run the M7
> firmware.
>
> Signed-off-by: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
> ---
>  .../dts/freescale/imx95-toradex-smarc.dtsi    | 50 +++++++++++++++++++
>  1 file changed, 50 insertions(+)
>
> diff --git a/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi b/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi
> index 5932ba238a8a..77c4a6dee098 100644
> --- a/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi
> +++ b/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi
> @@ -42,6 +42,16 @@ clk_serdes_eth_ref: clock-eth-ref {
>  		enable-gpios = <&som_gpio_expander_1 13 GPIO_ACTIVE_HIGH>;
>  	};
>
> +	cm7: remoteproc-cm7 {
> +		compatible = "fsl,imx95-cm7";
> +		mbox-names = "tx", "rx", "rxdb";
> +		mboxes = <&mu7 0 1
> +			  &mu7 1 1
> +			  &mu7 3 1>;
> +		memory-region = <&vdevbuffer>, <&vdev0vring0>, <&vdev0vring1>,
> +				<&vdev1vring0>, <&vdev1vring1>, <&rsc_table>, <&m7_reserved>;
> +	};
> +
>  	connector {
>  		compatible = "gpio-usb-b-connector", "usb-b-connector";
>  		/* SMARC P64 - USB0_OTG_ID */
> @@ -156,6 +166,42 @@ linux_cma: linux,cma {
>  			alloc-ranges = <0 0x80000000 0 0x7f000000>;
>  			linux,cma-default;
>  		};
> +
> +		m7_reserved: memory@80000000 {
> +			reg = <0 0x80000000 0 0x1000000>;
> +			no-map;
> +		};
> +
> +		rsc_table: memory@88220000 {

I remember driver use hardcode rsc-table, I suggest fan pan to update
driver to support memory-region-names to avoid use hidden ABI.

Fan peng, what's current status?

Frank

> +			reg = <0 0x88220000 0 0x1000>;
> +			no-map;
> +		};
> +
> +		vdev0vring0: memory@88000000 {
> +			reg = <0 0x88000000 0 0x8000>;
> +			no-map;
> +		};
> +
> +		vdev0vring1: memory@88008000 {
> +			reg = <0 0x88008000 0 0x8000>;
> +			no-map;
> +		};
> +
> +		vdev1vring0: memory@88010000 {
> +			reg = <0 0x88010000 0 0x8000>;
> +			no-map;
> +		};
> +
> +		vdev1vring1: memory@88018000 {
> +			reg = <0 0x88018000 0 0x8000>;
> +			no-map;
> +		};
> +
> +		vdevbuffer: memory@88020000 {
> +			compatible = "shared-dma-pool";
> +			reg = <0 0x88020000 0 0x100000>;
> +			no-map;
> +		};
>  	};
>  };
>
> @@ -572,6 +618,10 @@ &lpuart3 {
>  	pinctrl-0 = <&pinctrl_uart3>;
>  };
>
> +&mu7 {
> +	status = "okay";
> +};
> +
>  /* SMARC MDIO, shared between all ethernet ports */
>  &netc_emdio {
>  	pinctrl-names = "default";
> --
> 2.43.0
>

Re: [PATCH v1] arm64: dts: freescale: imx95-toradex-smarc: Support Cortex M7

[Emanuele Ghidoli]

On 3/3/26 16:38, Frank Li wrote:
> On Tue, Mar 03, 2026 at 01:13:06PM +0100, Emanuele Ghidoli wrote:
>> From: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
>>
>> Enable Cortex M7, the vring nodes, a mailbox and reserve DDR memory for
>> the M7. The remoteproc framework is so capable to load and run the M7
>> firmware.
>>
>> Signed-off-by: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
>> ---
>>  .../dts/freescale/imx95-toradex-smarc.dtsi    | 50 +++++++++++++++++++
>>  1 file changed, 50 insertions(+)
>>
>> diff --git a/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi b/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi
>> index 5932ba238a8a..77c4a6dee098 100644
>> --- a/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi
>> +++ b/arch/arm64/boot/dts/freescale/imx95-toradex-smarc.dtsi
>> @@ -42,6 +42,16 @@ clk_serdes_eth_ref: clock-eth-ref {
>>  		enable-gpios = <&som_gpio_expander_1 13 GPIO_ACTIVE_HIGH>;
>>  	};
>>
>> +	cm7: remoteproc-cm7 {
>> +		compatible = "fsl,imx95-cm7";
>> +		mbox-names = "tx", "rx", "rxdb";
>> +		mboxes = <&mu7 0 1
>> +			  &mu7 1 1
>> +			  &mu7 3 1>;
>> +		memory-region = <&vdevbuffer>, <&vdev0vring0>, <&vdev0vring1>,
>> +				<&vdev1vring0>, <&vdev1vring1>, <&rsc_table>, <&m7_reserved>;
>> +	};
>> +
>>  	connector {
>>  		compatible = "gpio-usb-b-connector", "usb-b-connector";
>>  		/* SMARC P64 - USB0_OTG_ID */
>> @@ -156,6 +166,42 @@ linux_cma: linux,cma {
>>  			alloc-ranges = <0 0x80000000 0 0x7f000000>;
>>  			linux,cma-default;
>>  		};
>> +
>> +		m7_reserved: memory@80000000 {
>> +			reg = <0 0x80000000 0 0x1000000>;
>> +			no-map;
>> +		};
>> +
>> +		rsc_table: memory@88220000 {
> 
> I remember driver use hardcode rsc-table, I suggest fan pan to update
> driver to support memory-region-names to avoid use hidden ABI.
> 
> Fan peng, what's current status?
> 
> Frank
> 
>> +			reg = <0 0x88220000 0 0x1000>;
>> +			no-map;
>> +		};
>> +
>> +		vdev0vring0: memory@88000000 {
>> +			reg = <0 0x88000000 0 0x8000>;
>> +			no-map;
>> +		};
>> +
>> +		vdev0vring1: memory@88008000 {
>> +			reg = <0 0x88008000 0 0x8000>;
>> +			no-map;
>> +		};
>> +
>> +		vdev1vring0: memory@88010000 {
>> +			reg = <0 0x88010000 0 0x8000>;
>> +			no-map;
>> +		};
>> +
>> +		vdev1vring1: memory@88018000 {
>> +			reg = <0 0x88018000 0 0x8000>;
>> +			no-map;
>> +		};
>> +
>> +		vdevbuffer: memory@88020000 {
>> +			compatible = "shared-dma-pool";
>> +			reg = <0 0x88020000 0 0x100000>;
>> +			no-map;
>> +		};
>>  	};
>>  };
>>
>> @@ -572,6 +618,10 @@ &lpuart3 {
>>  	pinctrl-0 = <&pinctrl_uart3>;
>>  };
>>
>> +&mu7 {
>> +	status = "okay";
>> +};
>> +
>>  /* SMARC MDIO, shared between all ethernet ports */
>>  &netc_emdio {
>>  	pinctrl-names = "default";
>> --
>> 2.43.0
>>

Thanks Frank. You’re right.
I spotted that my patch is not working as expected, due to a last minute
renaming on my side.
I’ll rework the patch accordingly and send a v2 to align to current expected
naming.

Emanuele

Re: [PATCH v1] arm64: dts: freescale: imx95-toradex-smarc: Support Cortex M7

[Frank Li]

From: Frank Li (AI-BOT) <frank.li@nxp.com>

AI bot review and may be useless.

---

> +	cm7: remoteproc-cm7 {
> +		compatible = "fsl,imx95-cm7";
> +		mbox-names = "tx", "rx", "rxdb";
> +		mboxes = <&mu7 0 1
> +			  &mu7 1 1
> +			  &mu7 3 1>;
> +		memory-region = <&vdevbuffer>, <&vdev0vring0>, <&vdev0vring1>,
> +				<&vdev1vring0>, <&vdev1vring1>, <&rsc_table>, <&m7_reserved>;
> +	};

Line 51 exceeds 80 columns; consider breaking memory-region into
multiple lines for readability.

---

> +	m7_reserved: memory@80000000 {
> +		reg = <0 0x80000000 0 0x1000000>;
> +		no-map;
> +	};

Consider adding a comment explaining the 16 MB reservation size and
purpose (firmware load address, etc.).

---

> +	rsc_table: memory@88220000 {
> +		reg = <0 0x88220000 0 0x1000>;
> +		no-map;
> +	};

Consider documenting why rsc_table is placed at this specific offset
relative to m7_reserved.

---

> +	vdevbuffer: memory@88020000 {
> +		compatible = "shared-dma-pool";
> +		reg = <0 0x88020000 0 0x100000>;
> +		no-map;
> +	};

Verify that the 1 MB vdevbuffer size is sufficient for the expected
virtio traffic between A55 and M7.

---

> +&mu7 {
> +	status = "okay";
> +};

Ensure mu7 node is defined elsewhere in the base device tree; this
patch only enables it without defining its properties.

Re: [PATCH v1] arm64: dts: freescale: imx95-toradex-smarc: Support Cortex M7

[Krzysztof Kozlowski]

On 03/03/2026 22:01, Frank Li wrote:
> From: Frank Li (AI-BOT) <frank.li@nxp.com>
> 
> AI bot review and may be useless.
> 

..

> ---
> 
>> +&mu7 {
>> +	status = "okay";
>> +};
> 
> Ensure mu7 node is defined elsewhere in the base device tree; this
> patch only enables it without defining its properties.

What? That's useless comment. Stop sending such feedback. It is just WRONG.

Best regards,
Krzysztof

Re: [PATCH v1] arm64: dts: freescale: imx95-toradex-smarc: Support Cortex M7

[Daniel Baluta]

On 3/3/26 23:01, Frank Li wrote:
> From: Frank Li (AI-BOT) <frank.li@nxp.com>
>
> AI bot review and may be useless.
>
> ---
>
>> +	cm7: remoteproc-cm7 {
>> +		compatible = "fsl,imx95-cm7";
>> +		mbox-names = "tx", "rx", "rxdb";
>> +		mboxes = <&mu7 0 1
>> +			  &mu7 1 1
>> +			  &mu7 3 1>;
>> +		memory-region = <&vdevbuffer>, <&vdev0vring0>, <&vdev0vring1>,
>> +				<&vdev1vring0>, <&vdev1vring1>, <&rsc_table>, <&m7_reserved>;
>> +	};
> Line 51 exceeds 80 columns; consider breaking memory-region into
> multiple lines for readability.

Frank please relax your bot!

As of commit bdc48fa11e46f867ea ("checkpatch/coding-style: deprecate 80-column warning")

there is no such requirement for wrapping at 80 chars anymore.

<snip>

>> +&mu7 {
>> +	status = "okay";
>> +};
> Ensure mu7 node is defined elsewhere in the base device tree; this
> patch only enables it without defining its properties.
>
Well it should fail compilation if mu7 was not defined? Sometimes this comments can be confusing so please

review them carefully before replying.

Re: [PATCH v1] arm64: dts: freescale: imx95-toradex-smarc: Support Cortex M7

[Peng Fan]

On Tue, Mar 03, 2026 at 10:38:09AM -0500, Frank Li wrote:
>On Tue, Mar 03, 2026 at 01:13:06PM +0100, Emanuele Ghidoli wrote:
>> From: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
>>
>> +
>> +		rsc_table: memory@88220000 {
>
>I remember driver use hardcode rsc-table, I suggest fan pan to update
>driver to support memory-region-names to avoid use hidden ABI.
>
>Fan peng, what's current status?

It is delayed because of the i.MX95 remoteproc patchset.
And i.MX94/952 is still on my list as higher priority, so the moving
to memory-region-names will be after that.

Regards
Peng

>
>Frank
>
>> +			reg = <0 0x88220000 0 0x1000>;
>> +			no-map;
>> +		};
>> +
>> +		vdev0vring0: memory@88000000 {
>> +			reg = <0 0x88000000 0 0x8000>;
>> +			no-map;
>> +		};
>> +
>> +		vdev0vring1: memory@88008000 {
>> +			reg = <0 0x88008000 0 0x8000>;
>> +			no-map;
>> +		};
>> +
>> +		vdev1vring0: memory@88010000 {
>> +			reg = <0 0x88010000 0 0x8000>;
>> +			no-map;
>> +		};
>> +
>> +		vdev1vring1: memory@88018000 {
>> +			reg = <0 0x88018000 0 0x8000>;
>> +			no-map;
>> +		};
>> +
>> +		vdevbuffer: memory@88020000 {
>> +			compatible = "shared-dma-pool";
>> +			reg = <0 0x88020000 0 0x100000>;
>> +			no-map;
>> +		};
>>  	};
>>  };
>>
>> @@ -572,6 +618,10 @@ &lpuart3 {
>>  	pinctrl-0 = <&pinctrl_uart3>;
>>  };
>>
>> +&mu7 {
>> +	status = "okay";
>> +};
>> +
>>  /* SMARC MDIO, shared between all ethernet ports */
>>  &netc_emdio {
>>  	pinctrl-names = "default";
>> --
>> 2.43.0
>>