Another ULP timer question

[wevets]

My experiments show that there is an initial sleep on starting the ULP based on the the value put in SENS_ULP_CP_SLEEP_CYC0_REG by calling ulp_set_wakeup_period(), but that's the only delay available from any of these registers in deep sleep mode. I can use that. I haven't done the experiments in light sleep mode. That may work, but light sleep is not what I need.

I tried the experiment you suggested in my own test bed program, that of setting up a loop such that halt was called once to try to set up a sleep, and then bypassing the halt the next time through and returning to the C code. However, the ULP never returned indicating that the ULP never work up on expiry of the sleep period. I don't even know if the sleep period count was actually started. I just know that she never returned. While I could easily be wrong, not knowing the ESP32 all that well yet, this feels like a hardware bug. Do you have any insight?

Another way to test this would be to set up a blinky in the ULP, and then set up, say, a 5 second delay in SENS_ULP_CP_SLEEP_CYC0_REG and then call halt right after returning the C code without ever explicitly starting the ULP again. The blinky will stop when the ULP hits that halt instruction. If "ULP wake after halt-sleep" works, the LED will start blinking again after the halt-sleep expires as long as the ULP jumps around the halt instruction forever. If the LED state remains constant beyond the halt-sleep timeout, wake after halt-sleep doesn't work.

[WiFive]

That's exactly what the blink edit shown does. ON, wait 2s, OFF, sleep 2s. Led keeps blinking.

[wevets]

Right, but I was wondering about a "halt-sleep". The docs say that the halt instruction stats a timer indicated n the last "sleep n" instruction. That doesn't work. But there is that value still in SENS_ULP_CP_SLEEP_CYC0_REG, so I was thinking of an experiment to determine if that halt started a timer based on the "...CYC0..." register, which ran after the ULP awakened the SoC, allowing the ULP to awaken while the SoC never started the ULP again but went on to do other stuff. I'm guessing not, but I've learned not to rely on my guesses much, especially on something I know so little about as this chip.

[WiFive]

Yes you can disable the timer so the ulp does not start again, it is in some of the examples