EVMK2H: How to confgurate IPC 3.x MessageQ to support communicate both ARM <-> DSP and DSP <-> DSP?

Tony,

There are some IPC_3.x documents and lab manuals mentioned here. This will be useful.

http://processors.wiki.ti.com/images/6/69/IPC_Training_2_21.pdf

http://processors.wiki.ti.com/images/a/a4/IPC_Lab_1_Hello.pdf
http://processors.wiki.ti.com/images/5/5e/IPC_Lab_2_MessageQ.pdf
http://processors.wiki.ti.com/images/1/19/IPC_Lab_3_Scalability.pdf

Hi pubesh,

Thanks for your info.

I have reviewed these documents, they doen't suitable for Keystone2. 

For example, in IPC_Training_2_21.pdf, MessageQ module uses the below transport method for actual message delivery.

•TransportShm – slowest, largest data footprint, most robust (default)
•TransportShmCirc – medium, fixed length transport buffer
•TransportShmNotify – fastest, may cause sender to busy wait

But in fact, for EVMK2H, we use VirtioSetup for MessageQ transport between ARM and DSP.

Another example in  IPC_Training_2_21.pdf, we must call Ipc_start() to use IPC.

But in fact, there in NO Ipc_start() in image_processing demo codes.

Is there other example to show how to do IPC MessageQ communication between ARM and DSP , and between DSP and DSP at  the time?

Tony

 

Tony,

The IPC in this context is MessageQ based IPC between ARM to DSP . This transport is based on rpmsg for communication between ARM and DSPs, and a proprietary shared memory protocol for communication between
DSPs. MessageQ is a reliable control path for inter-processor communication. See reference to details at
Control_path_IPC.

The demo application uses a IPC benchmark application MessageQBench. Please see details at IPC_BenchMarking.

http://processors.wiki.ti.com/index.php/IPC_BenchMarking

Tony,

I hope, you have refered the MessageQBench and Image processing example? Is this helpful for your implementation like ARM to DSP and DSP to DSP communication using IPC. You can comeback if you need further discussion on this. Otherwise, try to close this thread. This discussion of thread can helpful for future readers, If you verified this thread.

Hi Pubesh,

I build MessageQBench and test it.
There is a dual_transports.c in the example, demonstrating transportRpmsg and transportShm coexistence. So I use dual_transports.xe66 to and MessageQBench to verify.
But unfortunately, the example doesn't work. I am not sure what's wrong.

#mpmcl reset dsp0
#mpmcl load dsp0 dual_transports.xe66
#mpmcl run dsp0
#MessageQBench
Using numLoops: 1000; payloadSize: 8, procId : 1
Entered MessageQApp_execute
Local MessageQId: 0x1
Error in MessageQ_open [-6]
Leaving MessageQApp_execute
# cat /sys/kernel/debug/remoteproc/remoteproc0/trace0
3 Resource entries at 0x800000
main: MultiProc id: 1
registering rpmsg-proto service on 61 with HOST

It looks like DSP doesn't create IPC messageQ. Why?

Tony

Tony,

What the status on you side, is there any update? I cann't setup to test this in my side. I will try to give technical suggestion based on your status now.

Hi Pubsh,

There is still no progress now.

I find the below descriptions from IPC_Users_Guide.pdf.  I need the example or test case to know how to implement it exactly.

In a multi-processor system, MessageQ communicates with other processors via
ti.sdo.ipc.interfaces.IMessageQTransport instances. There can be up to two IMessageQTransport
instances for each processor to which communication is desired. One can be a normal-priority transport
and the other for handling high-priority messages. This is done via the priority parameter in the transport
create() function. If there is only one register to a remote processor (either normal or high), all messages
go via that transport.

BTW, you said that you cann't setup to test in your side. What's wrong? Maybe I can help.

 

Tony