STM32MP1 A7 and M4 communication

[ASega.1]

Hi,

I'm using STM32MP153C and I would like to communicate between the A7 and the M4 using the Virtual Uart (OPENAMP and IPCC).

I'm using a custom distribution created using Yocto.

When I initialize the virtual uart in the M4, /dev/ttyRPMSG0 is instantiated in the A7.

[ 142.543903] remoteproc remoteproc0: powering up m4

[ 142.649727] remoteproc remoteproc0: Booting fw image DHCOR_MBM_GEN_3_v4.1_CM4.elf, size 3518900

[ 142.661217] remoteproc0#vdev0buffer: assigned reserved memory node vdev0buffer@10042000

[ 142.669330] remoteproc0#vdev0buffer: registered virtio0 (type 7)

[ 142.674309] remoteproc remoteproc0: remote processor m4 is now up

[ 142.722658] virtio_rpmsg_bus virtio0: rpmsg host is online

[ 148.575929] virtio_rpmsg_bus virtio0: creating channel rpmsg-tty-channel addr 0x400

[ 149.074414] virtio_rpmsg_bus virtio0: creating channel rpmsg-sdb-channel addr 0x401

But when I use this command:

stty -onlcr -echo -F /dev/ttyRPMSG0

linux returns:

[ 366.375280] rpmsg_tty virtio0.rpmsg-tty-channel.-1.1024: Invalid message: size 34, type 67

[ 366.384280] rpmsg_tty virtio0.rpmsg-tty-channel.-1.1024: Invalid message: size 34, type 67

I saw that this error is returned by the function rpmsg_tty_cb in the if condition (line 109 of the driver rpmsg_tty.c):

static int rpmsg_tty_cb(struct rpmsg_device *rpdev, void *data, int len,
void *priv, u32 src)
{
struct rpmsg_tty_payload *rbuf = data;
 
if (len <= sizeof(*rbuf) || rbuf->cmd >= NUM_RPMSG_TTY_TYPE) 
{
dev_err(&rpdev->dev, "Invalid message: size %d, type %d 
len, rbuf->cmd);
return -EINVAL;
}
 
rpmsg_tty_handler[rbuf->cmd](rpdev, &rbuf->data,
len - sizeof(rbuf->cmd), priv, src);
return 0;
 
}

The error is caused because rbuf->cmd is greater than NUM_RPMSG_TTY_TYPE. In this case:

rbuf->cmd = 67

NUM_RPMSG_TTY_TYPE = 2

What is the meaning of this parameter NUM_RPMSG_TTY_TYPE?

It is declared in the driver at lines 21 without initialization (I can't find it):

enum rpmsg_tty_type_t {
RPMSG_DATA,
RPMSG_CTRL,
NUM_RPMSG_TTY_TYPE
};

The problem is caused by the A7 or M4 code? Because even if i'm not sending anything with the command "stty -onlcr -echo -F /dev/ttyRPMSG0", the M4 seems to receive something and (I think) it tries to send it back (M4 code send back what received).

I try also to send directly a simple "hello" to M4 from A7:

echo "Hello" > /dev/ttyRPMSG0

[ 105.752908] rpmsg_tty virtio0.rpmsg-tty-channel.-1.1024: Invalid message: size 34, type 67

[ 105.782157] rpmsg_tty virtio0.rpmsg-tty-channel.-1.1024: Invalid message: size 33, type 67

[ 105.810142] rpmsg_tty virtio0.rpmsg-tty-channel.-1.1024: Invalid message: size 34, type 67

Any suggestion to fix my communication?

Thanks in advance,

Andrea Segalla

[ASega.1]

Hi,

Small update of my issue.

I find the problem but I don't know how fix it.

When I use in linux the following command:

stty -onlcr -echo -F /dev/ttyRPMSG0

in M4 code is called the callback associated to the Virtual UART 0:

sprintf(m_uart_buff_tx, "CM4 : VIRT_UART_RxCpltCallback: %s\n", virt_uart_channel_buff_rx);
 
VIRT_UART_Transmit(hdr_descr.virt_huart, (uint8_t*)m_uart_buff_tx, strlen(m_uart_buff_tx));

And It sends back what received, in this case:

"CM4 : VIRT_UART_RxCpltCallback: \n"

This string has length 34 so I can understand the value size:

[ 366.375280] rpmsg_tty virtio0.rpmsg-tty-channel.-1.1024: Invalid message: size 34, type 67

However I don't understand why the callback function is called with this command:

stty -onlcr -echo -F /dev/ttyRPMSG0

and what is the value tpye 67?

Thanks in advance,

Andrea Segalla

[ArnaudP · Accepted solution]

Hello,

Seems that you are using a "customized" version of the rpmsg_tty driver.

The enum rpmsg_tty_type_t as been introduced in a version of the rpmsg_tty proposed for upstream:

https://www.mail-archive.com/linux-kernel@vger.kernel.org/msg2005980.html. But this temporary version should not have been neither delivered nor upstreamed.

This version is not compatible with the Cortex-M4 "virtual rpmsg"

So I would suggest you to come back to rpmsg_tty Linux driver delivered by ST.

Regards,

Arnaud

[ASega.1]

Hi,

Thanks for the reply.

What commit/release should I refer to?

Best Regards,

Andrea

[ArnaudP]

You should find releases information on the ST Microelectronics wiki: stm32mpu

[ASega.1]

Hi Arnaud,

thank you so much for the help and your time. I fixed my problem changing the driver as you told me.

Now it works correctly.

Best Regards,

Andrea

[mpeton]

Hi,
I'am also using  STM32MP157C DK2  to communicate between the A7 and the M4 using the Virtual Uart (OPENAMP and IPCC).

I was trying to use STM32MXCube to configure the peripherals but the STM32MP157C DK2 board is not available instead it has STM32MP157D DK1.

I can create a project with MCU selector but that doesn't allow me enable OpenAMP.

As both boards have different MCU CAC3 and DAC1 respectively can I still run MXCube project created with D-DK1 board selected on C-DK2 board