odd SNTP behavior that sometimes happens when synchronizing

[jkent]

My project uses the SNTP code -- most of the time it behaves as expected, but occasionally it will act up and time will go backwards a few hours a second, starting from the synchronized time. Here is my code I've tested it with:

https://gitlab.jkent.net/wtf/wifilcd/blob/master/main.c#L53 https://gitlab.jkent.net/wtf/wifilcd/blob/master/clock.c#L245

and here is the observed behavior:

https://www.youtube.com/watch?v=eYdcxB7kLnQ

[kanflo]

That's odd. Have you verified the SNPT replies with printf's or eg. wireshark?

[jkent]

Not yet, I should have time this week to dig in deeper.

[Rutger798]

I have the same issue, and enabled some extra SNTP logging with: -DSNTP_LOGD -DSNTP_LOGD_WITH_PRINTF after some day's of logging it looks like "RTC counter wrapped" occurred

And that only happens in: // Check if a timer wrap has occurred. Compensate sntp_base reference // if affirmative. // TODO: think about multitasking and race conditions static inline void sntp_check_timer_wrap(uint32_t current_value) { if (current_value < tim_ref) { // Timer wrap has occurred, compensate by subtracting 2^32 to ref. sntp_base -= 1LLU<<32; // DEBUG SNTP_LOGD("SNTP RTC counter wrapped"); } }

and that also shows the issue in the comment, "multitasking"...

[Rutger798]

Soo that brings me to the open pull requests: But both are to old to merge.....

https://github.com/SuperHouse/esp-open-rtos/pull/157 https://github.com/SuperHouse/esp-open-rtos/pull/148