esb handler crash

RE: esb handler crash

ovrebekk — Wed, 29 Nov 2017 08:35:54 GMT

Could you try to increase the length of the timeslots on the TX side?
That seemed to help for me. It has been running for a while now without any resets.

I simply changed the TS_LEN_US and TX_LEN_EXTENSION_US defines at the top of esb_timeslot.c to 25000.

This also allowed me to increase the MAX_TX_ATTEMPTS define, so I changed it to 16 to get less failed packets.

Best regards

RE: esb handler crash

Motti Ng — Mon, 27 Nov 2017 02:04:09 GMT

Thank you for your helping.

The issue seems related to the re-transmission in nrf_esb.c.

NRF_ESB_EVENT_TX_FAILED is happened only when re-transmission is done and fail.

However, when the timeslot ends before all re-transmissions are done, the re-transmission count is reset and thus re-transmission never reach the end within one timeslot.

It seems I can avoid the reset issue by changing the re-transmission count to 0 (but still do re-attempts in esb_timeslot.c).

Unfortunately, a new problem appear after that. The time it takes to do one tx attempt is somehow lengthened gradually. It takes like 10 seconds to do one tx attempt which is quite unfeasible.

RE: esb handler crash

ovrebekk — Fri, 24 Nov 2017 14:44:37 GMT

I was able to reproduce the issue on my end, and it seems to happen pretty consistently as a result of the UART and ESB interrupts happening simultaneously.

I don't think changing the S130 connection parameters will make a big difference. I tried to increase the advertising interval from 40ms to 400ms, and somehow it just made the issue worse...

I will have to do some more digging next week to try to get to the bottom of this.

Best regards

RE: esb handler crash

Motti Ng — Thu, 23 Nov 2017 02:13:37 GMT

What should I do to check the S130 stack?

I am running one to one connection and the interval is 1second.

I have directly copied the the example from github link text

Yesterday, I tried to change the connection interval to 5seconds or to make it driven by button (pressed every 5~10s).

It took longer time but the same error would still occur.

When I add debug message in the event handler, it is found that nrf_esb_event_handler is NOT called such that it endlessly retry since m_tx_attempt-- is not called.

The error appear when a new payload is added to fifo before the endlessly retrying message is transmitted.

I will add this finding to the post.

RE: esb handler crash

ovrebekk — Wed, 22 Nov 2017 11:35:54 GMT

What is the S130 stack doing when the problem occurs?
How many links do you have running, and what are the connection intervals?

Have you made any changes to the ESB event handler, or is it copied directly from the example?

Best regards