int
kfilter_register(
const char *name
, struct filterops *filtops
, int *retfilter
)
int
kfilter_unregister(
const char *name
)
)
function adds a new kernel event filter (kfilter) to the system,
for use by callers of
kqueue(2)
and
kevent(2).
name
is the name of the new filter (which must not already exist), and
filtops
is a pointer to a
filterops
structure which describes the filter operations.
Both
name
and
filtops
will be copied to an internal data structure, and a new filter number
will be allocated.
If
retfilter
is not
NULL
,
then the new filter number will be returned in the address pointed at by
retfilter
.
The
kfilter_unregister()
function removes a kfilter named
name
that was previously registered with
kfilter_register().
If a filter with the same
name
is later reregistered with
kfilter_register(),
it will get a different filter number
(i.e., filter numbers are not recycled).
It is not possible to unregister the system filters
(i.e., those that start with
``EVFILT_''
and are documented in
kqueue(2)).
The filterops structure is defined as follows:
struct filterops {
int f_isfd; /* true if ident == filedescriptor */
int (*f_attach)(struct knote *kn);
/* called when knote is ADDed */
void (*f_detach)(struct knote *kn);
/* called when knote is DELETEd */
int (*f_event)(struct knote *kn, long hint);
/* called when event is triggered */
};
If the filter operation is for a file descriptor, f_isfd should be non-zero, otherwise it should be zero. This controls where the kqueue(2) system stores the knotes for an object.
)
returns 0 on success,
EINVAL
if there's an invalid argument, or
EEXIST
if the filter already exists,
kfilter_unregister()
returns 0 on success,
EINVAL
if there's an invalid argument, or
ENOENT
if the filter doesn't exist.
)
and
kfilter_unregister(
)
functions first appeared in
NetBSD2.0.
)
and
kfilter_unregister(
)
functions were implemented by
Luke Mewburn