Event queue services for periodic asynchronous callbacks
#include <pcp/pmapi.h>
#include <pcp/impl.h>
int __pmAFregister(const struct timeval *delta, void *data, void (*func)(int, void *));
int __pmAFunregister(int afid);
void __pmAFblock(void);
void __pmAFunblock(void);
int __pmAFisempty(void);
cc ... -lpcp
The routines implement an event queue and callback framework that supports periodic evaluation of a series of events with varying frequencies for Performance Co-Pilot (PCP) applications.
The pmlogger(1) application, the pmdatrace(1) PMDA and the pmdahotproc(1) PMDA are the principal users of these services.
An event is registered by calling __pmAFregister, and on success the return value is an event number greater than zero. The event has associated event data identified by the opaque pointer data. The event will occur with frequency delta (the first instance will be delta after the current time when the event is registered), and each time the event occurs the function func will be called with the event number and the event data as arguments.
Once the event occurs and the callback has been executed, the event will be rescheduled for delta into the future, except if all the fields of delta are zero, in which case the event will not be rescheduled (a ``one trip'' event).
Internally, events are processed serially so there is no possibility of nested callbacks or re-entrant callbacks from the event management routines.
Given an event number afid, __pmAFunregister will permanently remove the corresponding entry from the event queue.
To control the event queue processing, __pmAFblock and __pmAFunblock may be used to explicitly block and unblock the dispatch of events. This is most useful when the caller wishes to set up a number of events via __pmAFregister and complete the registration phase before the first event callback occurs.
A call to __pmAFisempty returns 1 or 0 depending on whether the event queue is empty or not.
__pmAFregister and __pmAFunregister return values less than zero in the case of an error. These values are PCP error codes, and may be used to produce error messages via pmErrStr(3).
The routines support the standard PCP debug tracing, and the value DBG_TRACE_AF (or -D af on the command line) will produce diagnostics on standard error that trace the enqueueing and execution of events.
These routines rely on setitimer(2) and manipulate the handling of SIGALRM signals, and hence are probably ill-suited for applications that require direct and concurrent access to these services and resources.
If the callback functions are slow, or delayed, it is possible that the event scheduling could fall behind and never catchup. When this begins to happen, events are silently skipped and rescheduled at the earliest possible time on the future according to the fixed schedule defined by the time of the call to __pmAFregister and the value of the delta argument to __pmAFregister.
In addition, the semantics of the interval timer(s) and the global state needed to support these services demand that applications calling these routines must do so from a single thread. This restriction is enforced at the PMAPI(3), where routines may return the error code PM_ERR_THREAD if the library detects calls from more than one thread.