problem managed ipcf s32k3

[michelet1]

Good evening,

i am analyzing ipcf for s32k324 in the managed case as shown in the attached image.

I instantiated a buffer of 10 values on core0 and tried to send it to core1. By varying the size parameter i did not notice any changes, in the sense that with any size value set other than zero you send the entire buffer to core1.

I wanted to ask what is the meaning of the size parameter in the shm_tx function because seems to don’t have any control on parameter “size”.

The current SW we are developing is based on RTD 3.0.0,S32K3xx_AMMCLIB_v1.1.34, IPCF S32K3xx 4.0.0

thanks in advance for any help and information.

[danielmartynek]

Hi @michelet1,

It is a zero copy mechanism.

When we send a message, we basically provide the other core with a pointer to the buffer and the size of it.

The size is then an argument in the callback.

When CM7_0 sends data it performs the following:

• acquires a buffer
• populates the buffer
• sends a notification to CM7_1

Once CM7_1 receives the callback, it does the following:

• calculates the pointer address
• reads and use the data (no need to copy as the buffer is locked until it is released by the user)
• releases the buffer (in order to be reused)

BR, Daniel

[michelet1]

hi,

thanks for your reply.

i just have a doubt.

if core0 acquire a buffer of 10 items and send it but specifies a size=2,

on core1 should receive only 2 items and not all 10 items, am i right?

why do i find the entire buffer on core1 for any size value i specify?

thanks again for the availability

Michele

[danielmartynek]

Hi Michele,

Because it does not send anything.

The other core is notified that there are data in a buffer of a certain size defined by ipc_shm_tx(size).

BR, Daniel

[michelet1]

Hi,

thanks for the answer.

i'm still not clear about the meaning of the size parameter? shouldn't it indicate how much buffer data i want to send from one core to another?

let's make a practical example:

if i have a buff[10] and specify size=2 as the ipc_shm_tx() parameter on core0, on core1 the notify should go off and only read buf[0] and buf[1], is this reasoning correct?

thanks again for your availability

Michele

[danielmartynek]

Yes, core_1 is notified to read just buf[0] and buf[1].

Regards,

Daniel