mantriyogesh
commented
1 week ago Thank you @naguirre , for detailed debug and following the documentation. Much appreciated..
I still see that you had to run the different SPI modes at slave and host. I highly suspect the timing issues. How to confirm?
Enable checksum at slave and try to run raw throughput test in TX and RX using transport test.
Additionally, you can change the logs to info and verify the tx from esp matches to rx at host and vice versa. esp logs you already know. host side logs:
host tx: https://github.com/espressif/esp-hosted/blob/b1422afc4fe0a2ea1ffd6b5459edddc4d968ee71/esp_hosted_ng/host/spi/esp_spi.c#L322 You can optionally add log in 'else' case (As spi is full duplex, when host doesn't have any valid data to tx while doing rx from slave, it would send dummy data, which would be ignored at receiver end)
host rx:
After spi_sync_transfer() att rx print at: https://github.com/espressif/esp-hosted/blob/b1422afc4fe0a2ea1ffd6b5459edddc4d968ee71/esp_hosted_ng/host/spi/esp_spi.c#L351
add
esp_hex_dump_verbose("Rx: ", trans.rx_buf, 32);This would help you analyse any mismatch between the tx and rx message either side.
As you are enabling checksum (using idf.py menuconfig at slave), if any message silently getting ignored would be printing error, tto understand any underlying transport timing issue.
naguirre
kapilkedawat
Hello. I'm trying to set up ESP Hosted NG. My hardware is a custom board I developed based on an ESP32-C3-Mini and an Allwinner f1c200s. I built the latest version of esp-hosted-ng (
b1422af) firmware, and the Linux kernel version on the F1c200s is6.9.2. I followed the porting guide and started by validating the SPI communication between ESP32-C3 and F1c200s withspidev. So far so good, I was able to validate the communication at 30Mhz between the two devices.I noticed, though, that when the SPI slave uses
MODE2, the data received on F1c200s are invalid. Whether I'm usingMODE0orMODE1, the data are correct. So, for the next tests, I'm usingMODE1for esp32 andMODE2for the host.With
spidevI didn't notice any issue with the communication after those changes. I was able to read and write data from the ESP32-C3 to the F1c200s and vice versa. I used the ESP32 SPI slave example and developed a small app for the F1c200s.I then flashed the ESP Hosted NG firmware, and built
esp32_spi.The reset GPIO line is fine, and I can see the ESP32 resetting when the line is asserted even if I had to raise the pin to 1 and increase the reset time a bit instead of just getting it back to input.
The Handhsake and data-ready pins look good as well, and I can see the interrupts being triggered on the F1c200s side.
The main problem is that the command
iw espsta0 scandoes not work. It hangs, and I never get a result.The logs from ESP32 :
The logs from the host :
The command I ran from the host to get these logs :
I'm wondering if you have any ideas :D I running out of idea now.