hi,

now I say a thing which may look terrible but please read all before claim my carma (=

Indeed, some my projects have deep subroutines which take a lot of time to be done. Fortunatelly, they are called from main cycle time to time; but when one of them is entered, it takes long time before program returns to main().
From other side I like watchdog and I always use it. To reset watchdog counter I need do some steps: disable all interrupts, write 0x1E then 0xE1 into watchdog SFR then re-enable interrupts.
Well, first time I have creating a macro and then inserted it in main() and inside of those "long-time" subroutines.

After some time I switched to another way (and now the time you kick me =)
I moved this piece of code inside a timer ISR which executes often enough to reset watchdog counter before it resets program (here I use PCA counter overflow interrupt).
I know, I know, you say:

     !!! DO NOT PUT WATCHDOG RESET CODE INTO TIMER ISR !!!

.I have read it in many books, yes. But look here:

; flag allocation
FLAGS_MISC	DATA	0x2D
FLG_WATCHDOG	BIT	FLAGS_MISC.7	; reset counter request flag
; ...

; 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:
; ...

; somewhere inside main() and inside each "long-time" subroutines:
;
	SETB	FLG_WATCHDOG    ; pending request
;

As you can see, PCA timer ISR does not reset watchdog timer at each execution undoubtedly. Instead, it checks is a part of program requests for such task. And if some part of code has set request flag then watchdog timer been reset only.

Regards,
Oleg