MCLR from sleep...?

[ktallevi]

I am using a PIC18LF8722, once the PIC has started up I put it immediately to sleep. I have a push button which can wake up the PIC to execute the code following the sleep(); command. ( I can post code if need be).

Anyways upon coming out of sleep, the PIC resets and the reset cause is MCLR from sleep.

Any idea what could be causing this?

thanks

[PCM programmer]

[ktallevi]

the push button (which is hardware debounced) is connected to RB0/INT0.

I have used push button interrupts in previous projects with no issues, with this particular one I am getting those odd MCLR from sleep resets.

[PCM programmer]

Post a short test program that demonstrates the problem. Be sure to
post all #include, #fuses, and #use statements. The program should be
compilable. Also post your compiler version.

[Ttelmah]

It is worth looking at the data sheet.
MCLR from sleep (CCS definition), is code '11'.
Now, the corresponds to a watchdog timeout, _not_ occurring, a Power on reset _not_ occurring, and a brownout reset _not_ occuring. There is no actual detection bit for 'MCLR'. So what it suggests, is that you are arriving at the start of the code, for no normal reason.
The obvious thought, is that you have got the GIE bit set, and no interrupt handler present for this, which could give this sort of odd behaviour.
To use an interrupt to 'wake up', but not cause an interrupt handler to be called, you need to have the interrupt itself enabled, but the GIE (global interrupt enable) disabled. The sequence is:

disable_interrupts(GLOBAL); //If other interrupts are being used
clear_interrupts(INT_EXT); //To ensure any _past_ triggers are ignored
enable_interupts(INT_EXT);
sleep();
delay_cycles(1);
disable_interrupts(INT_EXT); //disable this _before_ re-enabling the GIE
enable_interrupts(GLOBAL); //Re-enable is other interrupts are being used

The 'delay', is optional, but often worth having. The instruction _after_ the sleep, is 'prefetched', when the sleep executes, and in some cases, this can give unexpected results (if this is a test for example). Hence a single 'NOP' (which iswhat a single cycle delay generates), avoids 'oddities' from this.

Best Wishes