Tool/software: Linux
Received this question from an ecosystem partner:
"our goal is improving of the jitter and performance of sending()/receiving() ethernet frames.
We expect to get better results on PRU comparing to CPSW
As I read here http://software-dl.ti.com/processor-sdk-linux/esd/docs/latest/linux/Foundational_Components_PRU-ICSS_PRU_ICSSG.html#pru-icss-ethernet
“With the PRU-ICSS Ethernet support, you get two (or four, on the AM571x IDK and K2G ICE) additional 100 Mbps Ethernet interfaces that show up just like any other Ethernet interface in Linux. These interfaces are there ‘out-of-the-box’ in the latest Linux Processor SDK when ran on any of the boards shown above”
It means, that we can check our traditional approach (used already with CPSW) to access PRU, using raw sockets in Linux and compare the results with CPSW.
I also read the following:
“Linux needs memory mapping of memory regions to user space, and this is done using the UIO driver with linux file-I/O like API such as open(), close(), read(), write(), and etc”
The more promising approach is bypassing TCP/IP stack in Linux and access PRU directly in order to send()/receive() a frame.
I have installed the latest SDK ti-processor-sdk-linux-rt-am57xx-evm-05.02.00.10, but unfortunately couldn’t find any appropriate example for that (accessing of PRU in Linux userspace in order to send/receive frame).
Could you please point to such example in the SDK.
