It's time to "re-align your thinking."
Your piece of software ... whatever it is ... is
not the
only bit of software (on this planet...) that "is interested in the fact that an interrupt has just occurred on this parallel I/O port ..."
... it
is, rather, 'just 'an interested party.'" Just one of however-many...
Your driver, therefore, is obliged to
register itself upon the appropriate interrupt-handling
chain, in order to be duly
called by the Linux kernel whenever such an interrupt arrives. It will be obliged to indicate to the kernel whether it has "handled" the interrupt, or not. But, in any case, it will
not be "in control of" the situation.
It may well be that part of the process of "handling the interrupt" consists of a
hardware action, such as "setting a status-bit." It may well be that, in your suitably-specialized hardware situation, "no other interrupt-handler need apply." Your
second-level interrupt handler actually has that prerogative.