Hi everyone, @Middleware & OS I’m struggling to find sufficient material on handling non-atomic operations, and I’d appreciate some guidance.

I need to maintain a microsecond count in a 64-bit variable, with a 32-bit processor system that uses a non-preemptive scheduler. An ISR will update this variable every microsecond. My system provides functions to clear the variable and read its value.
Since the processor is 32-bit, accessing the entire 64-bit variable atomically might not be possible. Concerned how I can ensure that the read function doesn’t retrieve a partially updated value as in reading half from the old value and half from the new value during an interrupt?

In non-preemptive scheduling context now, are there standard approaches to guarantee consistent reads of the 64-bit variable? If so, could anyone help me on some of these techniques?