To reset watchdog counter I need do some steps: disable all interrupts, write 0x1E then 0xE1 into watchdog SFR then re-enable interrupts.
These specific steps makes it virtually impossible that runaway code can pet the puppy.

; somewhere inside ISR of PCA (counter overflow condition):
INT_PCA_OVER:
; check if watchdog counter reset is required somewhere
	JNB	FLG_WATCHDOG,INT_PCA_WATCHDOG_END
	CLR	EA
	MOV	WDTRST,#0x1E	; reset watchdog counter
	MOV	WDTRST,#0xE1
	CLR	FLG_WATCHDOG    ; clear request flag
	SETB	EA
INT_PCA_WATCHDOG_END:
; ...

Now you have a situation where all runaway code has to do is set a bit. That is much more likely to happen than the steps required to pet the puppy. You still need to do some action (in your case set a flag as opposed to invoking a macro) in a timely manner, in that respect there is no difference.

Please explain, where is the gain using
setb FLG_WATCHDOG

over
PET_THE_PUPPY ; a macro

Erik