My usual solution is much like of ' Bartosz Wucke'.

But since my timer interrupts are likely used for other important jobs, I add in the least important one a line like the following one:

JNB FLAG_wait, TMRcont

So as long as the 'FLAG_wait' is not set, the other functions in the timer interrupt routine are normally executed. Otherwise the following lines will be executed instead (might need more instructions like CLR TR?... etc):

CLR FLAG_wait ; since it is set for one time job.
POP tempByte ; 'tempByte' is any redundant register
POP tempByte ; to remove the unwanted actual address
PUSH DPL ; DPTR has to be already loaded to
PUSH DPH ; point the new destination address
RETI ; Return from timer interrupt

But before the check-pin instruction the following need to be added too:

SETB FLAG_wait ; important
MOV DPTR, #exit_label ; the new destination, important
CLR TR0 ; or TR1, to change TLx and/or THx
MOV TL0, #-limitL ; limitL= low byte of maxDelayCycles
MOV TH0, #-limitH ; limitH= high byte of maxDelayCycles
SETB TR0 ; or TR1, to activate the timer

JNB P1.0,$ ; or alike, now has a time limit
....
....

(by the way, why does the post here look different than the text I copied from?)