[ViewVC] Diff of: cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.188 by root, Tue Sep 30 18:35:47 2008 UTC vs.
Revision 1.193 by root, Wed Oct 1 04:25:25 2008 UTC

…		…
2384	=over 4	2384	=over 4
2385		2385
2386	=item queueing from a signal handler context	2386	=item queueing from a signal handler context
2387		2387
2388	To implement race-free queueing, you simply add to the queue in the signal	2388	To implement race-free queueing, you simply add to the queue in the signal
2389	handler but you block the signal handler in the watcher callback. Here is an example that does that for	2389	handler but you block the signal handler in the watcher callback. Here is
2390	some fictitious SIGUSR1 handler:	2390	an example that does that for some fictitious SIGUSR1 handler:
2391		2391
2392	static ev_async mysig;	2392	static ev_async mysig;
2393		2393
2394	static void	2394	static void
2395	sigusr1_handler (void)	2395	sigusr1_handler (void)
…		…
2502	=over 4	2502	=over 4
2503		2503
2504	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)	2504	=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
2505		2505
2506	This function combines a simple timer and an I/O watcher, calls your	2506	This function combines a simple timer and an I/O watcher, calls your
2507	callback on whichever event happens first and automatically stop both	2507	callback on whichever event happens first and automatically stops both
2508	watchers. This is useful if you want to wait for a single event on an fd	2508	watchers. This is useful if you want to wait for a single event on an fd
2509	or timeout without having to allocate/configure/start/stop/free one or	2509	or timeout without having to allocate/configure/start/stop/free one or
2510	more watchers yourself.	2510	more watchers yourself.
2511		2511
2512	If C<fd> is less than 0, then no I/O watcher will be started and events	2512	If C<fd> is less than 0, then no I/O watcher will be started and the
2513	is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and	2513	C<events> argument is being ignored. Otherwise, an C<ev_io> watcher for
2514	C<events> set will be created and started.	2514	the given C<fd> and C<events> set will be created and started.
2515		2515
2516	If C<timeout> is less than 0, then no timeout watcher will be	2516	If C<timeout> is less than 0, then no timeout watcher will be
2517	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and	2517	started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
2518	repeat = 0) will be started. While C<0> is a valid timeout, it is of	2518	repeat = 0) will be started. C<0> is a valid timeout.
2519	dubious value.
2520		2519
2521	The callback has the type C<void (cb)(int revents, void arg)> and gets	2520	The callback has the type C<void (cb)(int revents, void arg)> and gets
2522	passed an C<revents> set like normal event callbacks (a combination of	2521	passed an C<revents> set like normal event callbacks (a combination of
2523	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>	2522	C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>
2524	value passed to C<ev_once>:	2523	value passed to C<ev_once>. Note that it is possible to receive I<both>
		2524	a timeout and an io event at the same time - you probably should give io
		2525	events precedence.
		2526
		2527	Example: wait up to ten seconds for data to appear on STDIN_FILENO.
2525		2528
2526	static void stdin_ready (int revents, void *arg)	2529	static void stdin_ready (int revents, void *arg)
2527	{	2530	{
		2531	if (revents & EV_READ)
		2532	/* stdin might have data for us, joy! */;
2528	if (revents & EV_TIMEOUT)	2533	else if (revents & EV_TIMEOUT)
2529	/* doh, nothing entered */;	2534	/* doh, nothing entered */;
2530	else if (revents & EV_READ)
2531	/* stdin might have data for us, joy! */;
2532	}	2535	}
2533		2536
2534	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);	2537	ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
2535		2538
2536	=item ev_feed_event (ev_loop , watcher , int revents)	2539	=item ev_feed_event (ev_loop , watcher , int revents)
…		…
3306	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:	3309	And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
3307		3310
3308	#include "ev_cpp.h"	3311	#include "ev_cpp.h"
3309	#include "ev.c"	3312	#include "ev.c"
3310		3313
		3314	=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES
3311		3315
3312	=head1 THREADS AND COROUTINES	3316	=head2 THREADS AND COROUTINES
3313		3317
3314	=head2 THREADS	3318	=head3 THREADS
3315		3319
3316	All libev functions are reentrant and thread-safe unless explicitly	3320	All libev functions are reentrant and thread-safe unless explicitly
3317	documented otherwise, but it uses no locking itself. This means that you	3321	documented otherwise, but libev implements no locking itself. This means
3318	can use as many loops as you want in parallel, as long as there are no	3322	that you can use as many loops as you want in parallel, as long as there
3319	concurrent calls into any libev function with the same loop parameter	3323	are no concurrent calls into any libev function with the same loop
3320	(C<ev_default_*> calls have an implicit default loop parameter, of	3324	parameter (C<ev_default_*> calls have an implicit default loop parameter,
3321	course): libev guarantees that different event loops share no data	3325	of course): libev guarantees that different event loops share no data
3322	structures that need any locking.	3326	structures that need any locking.
3323		3327
3324	Or to put it differently: calls with different loop parameters can be done	3328	Or to put it differently: calls with different loop parameters can be done
3325	concurrently from multiple threads, calls with the same loop parameter	3329	concurrently from multiple threads, calls with the same loop parameter
3326	must be done serially (but can be done from different threads, as long as	3330	must be done serially (but can be done from different threads, as long as
…		…
3366	default loop and triggering an C<ev_async> watcher from the default loop	3370	default loop and triggering an C<ev_async> watcher from the default loop
3367	watcher callback into the event loop interested in the signal.	3371	watcher callback into the event loop interested in the signal.
3368		3372
3369	=back	3373	=back
3370		3374
3371	=head2 COROUTINES	3375	=head3 COROUTINES
3372		3376
3373	Libev is much more accommodating to coroutines ("cooperative threads"):	3377	Libev is very accommodating to coroutines ("cooperative threads"):
3374	libev fully supports nesting calls to it's functions from different	3378	libev fully supports nesting calls to its functions from different
3375	coroutines (e.g. you can call C<ev_loop> on the same loop from two	3379	coroutines (e.g. you can call C<ev_loop> on the same loop from two
3376	different coroutines and switch freely between both coroutines running the	3380	different coroutines, and switch freely between both coroutines running the
3377	loop, as long as you don't confuse yourself). The only exception is that	3381	loop, as long as you don't confuse yourself). The only exception is that
3378	you must not do this from C<ev_periodic> reschedule callbacks.	3382	you must not do this from C<ev_periodic> reschedule callbacks.
3379		3383
3380	Care has been taken to ensure that libev does not keep local state inside	3384	Care has been taken to ensure that libev does not keep local state inside
3381	C<ev_loop>, and other calls do not usually allow coroutine switches.	3385	C<ev_loop>, and other calls do not usually allow for coroutine switches as
		3386	they do not clal any callbacks.
3382		3387
		3388	=head2 COMPILER WARNINGS
3383		3389
3384	=head1 COMPLEXITIES	3390	Depending on your compiler and compiler settings, you might get no or a
		3391	lot of warnings when compiling libev code. Some people are apparently
		3392	scared by this.
3385		3393
3386	In this section the complexities of (many of) the algorithms used inside	3394	However, these are unavoidable for many reasons. For one, each compiler
3387	libev will be explained. For complexity discussions about backends see the	3395	has different warnings, and each user has different tastes regarding
3388	documentation for C<ev_default_init>.	3396	warning options. "Warn-free" code therefore cannot be a goal except when
		3397	targeting a specific compiler and compiler-version.
3389		3398
3390	All of the following are about amortised time: If an array needs to be	3399	Another reason is that some compiler warnings require elaborate
3391	extended, libev needs to realloc and move the whole array, but this	3400	workarounds, or other changes to the code that make it less clear and less
3392	happens asymptotically never with higher number of elements, so O(1) might	3401	maintainable.
3393	mean it might do a lengthy realloc operation in rare cases, but on average
3394	it is much faster and asymptotically approaches constant time.
3395		3402
3396	=over 4	3403	And of course, some compiler warnings are just plain stupid, or simply
		3404	wrong (because they don't actually warn about the condition their message
		3405	seems to warn about). For example, certain older gcc versions had some
		3406	warnings that resulted an extreme number of false positives. These have
		3407	been fixed, but some people still insist on making code warn-free with
		3408	such buggy versions.
3397		3409
3398	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)	3410	While libev is written to generate as few warnings as possible,
		3411	"warn-free" code is not a goal, and it is recommended not to build libev
		3412	with any compiler warnings enabled unless you are prepared to cope with
		3413	them (e.g. by ignoring them). Remember that warnings are just that:
		3414	warnings, not errors, or proof of bugs.
3399		3415
3400	This means that, when you have a watcher that triggers in one hour and
3401	there are 100 watchers that would trigger before that then inserting will
3402	have to skip roughly seven (C<ld 100>) of these watchers.
3403		3416
3404	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)	3417	=head2 VALGRIND
3405		3418
3406	That means that changing a timer costs less than removing/adding them	3419	Valgrind has a special section here because it is a popular tool that is
3407	as only the relative motion in the event queue has to be paid for.	3420	highly useful. Unfortunately, valgrind reports are very hard to interpret.
3408		3421
3409	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)	3422	If you think you found a bug (memory leak, uninitialised data access etc.)
		3423	in libev, then check twice: If valgrind reports something like:
3410		3424
3411	These just add the watcher into an array or at the head of a list.	3425	==2274== definitely lost: 0 bytes in 0 blocks.
		3426	==2274== possibly lost: 0 bytes in 0 blocks.
		3427	==2274== still reachable: 256 bytes in 1 blocks.
3412		3428
3413	=item Stopping check/prepare/idle/fork/async watchers: O(1)	3429	Then there is no memory leak, just as memory accounted to global variables
		3430	is not a memleak - the memory is still being refernced, and didn't leak.
3414		3431
3415	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))	3432	Similarly, under some circumstances, valgrind might report kernel bugs
		3433	as if it were a bug in libev (e.g. in realloc or in the poll backend,
		3434	although an acceptable workaround has been found here), or it might be
		3435	confused.
3416		3436
3417	These watchers are stored in lists then need to be walked to find the	3437	Keep in mind that valgrind is a very good tool, but only a tool. Don't
3418	correct watcher to remove. The lists are usually short (you don't usually	3438	make it into some kind of religion.
3419	have many watchers waiting for the same fd or signal).
3420		3439
3421	=item Finding the next timer in each loop iteration: O(1)	3440	If you are unsure about something, feel free to contact the mailing list
		3441	with the full valgrind report and an explanation on why you think this
		3442	is a bug in libev (best check the archives, too :). However, don't be
		3443	annoyed when you get a brisk "this is no bug" answer and take the chance
		3444	of learning how to interpret valgrind properly.
3422		3445
3423	By virtue of using a binary or 4-heap, the next timer is always found at a	3446	If you need, for some reason, empty reports from valgrind for your project
3424	fixed position in the storage array.	3447	I suggest using suppression lists.
3425		3448
3426	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
3427		3449
3428	A change means an I/O watcher gets started or stopped, which requires	3450	=head1 PORTABILITY NOTES
3429	libev to recalculate its status (and possibly tell the kernel, depending
3430	on backend and whether C<ev_io_set> was used).
3431		3451
3432	=item Activating one watcher (putting it into the pending state): O(1)
3433
3434	=item Priority handling: O(number_of_priorities)
3435
3436	Priorities are implemented by allocating some space for each
3437	priority. When doing priority-based operations, libev usually has to
3438	linearly search all the priorities, but starting/stopping and activating
3439	watchers becomes O(1) with respect to priority handling.
3440
3441	=item Sending an ev_async: O(1)
3442
3443	=item Processing ev_async_send: O(number_of_async_watchers)
3444
3445	=item Processing signals: O(max_signal_number)
3446
3447	Sending involves a system call I<iff> there were no other C<ev_async_send>
3448	calls in the current loop iteration. Checking for async and signal events
3449	involves iterating over all running async watchers or all signal numbers.
3450
3451	=back
3452
3453
3454	=head1 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS	3452	=head2 WIN32 PLATFORM LIMITATIONS AND WORKAROUNDS
3455		3453
3456	Win32 doesn't support any of the standards (e.g. POSIX) that libev	3454	Win32 doesn't support any of the standards (e.g. POSIX) that libev
3457	requires, and its I/O model is fundamentally incompatible with the POSIX	3455	requires, and its I/O model is fundamentally incompatible with the POSIX
3458	model. Libev still offers limited functionality on this platform in	3456	model. Libev still offers limited functionality on this platform in
3459	the form of the C<EVBACKEND_SELECT> backend, and only supports socket	3457	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
…		…
3546	wrap all I/O functions and provide your own fd management, but the cost of	3544	wrap all I/O functions and provide your own fd management, but the cost of
3547	calling select (O(n²)) will likely make this unworkable.	3545	calling select (O(n²)) will likely make this unworkable.
3548		3546
3549	=back	3547	=back
3550		3548
3551
3552	=head1 PORTABILITY REQUIREMENTS	3549	=head2 PORTABILITY REQUIREMENTS
3553		3550
3554	In addition to a working ISO-C implementation, libev relies on a few	3551	In addition to a working ISO-C implementation and of course the
3555	additional extensions:	3552	backend-specific APIs, libev relies on a few additional extensions:
3556		3553
3557	=over 4	3554	=over 4
3558		3555
3559	=item C<void ()(ev_watcher_type , int revents)> must have compatible	3556	=item C<void ()(ev_watcher_type , int revents)> must have compatible
3560	calling conventions regardless of C<ev_watcher_type *>.	3557	calling conventions regardless of C<ev_watcher_type *>.
…		…
3585	except the initial one, and run the default loop in the initial thread as	3582	except the initial one, and run the default loop in the initial thread as
3586	well.	3583	well.
3587		3584
3588	=item C<long> must be large enough for common memory allocation sizes	3585	=item C<long> must be large enough for common memory allocation sizes
3589		3586
3590	To improve portability and simplify using libev, libev uses C<long>	3587	To improve portability and simplify its API, libev uses C<long> internally
3591	internally instead of C<size_t> when allocating its data structures. On	3588	instead of C<size_t> when allocating its data structures. On non-POSIX
3592	non-POSIX systems (Microsoft...) this might be unexpectedly low, but	3589	systems (Microsoft...) this might be unexpectedly low, but is still at
3593	is still at least 31 bits everywhere, which is enough for hundreds of	3590	least 31 bits everywhere, which is enough for hundreds of millions of
3594	millions of watchers.	3591	watchers.
3595		3592
3596	=item C<double> must hold a time value in seconds with enough accuracy	3593	=item C<double> must hold a time value in seconds with enough accuracy
3597		3594
3598	The type C<double> is used to represent timestamps. It is required to	3595	The type C<double> is used to represent timestamps. It is required to
3599	have at least 51 bits of mantissa (and 9 bits of exponent), which is good	3596	have at least 51 bits of mantissa (and 9 bits of exponent), which is good
…		…
3603	=back	3600	=back
3604		3601
3605	If you know of other additional requirements drop me a note.	3602	If you know of other additional requirements drop me a note.
3606		3603
3607		3604
3608	=head1 COMPILER WARNINGS	3605	=head1 ALGORITHMIC COMPLEXITIES
3609		3606
3610	Depending on your compiler and compiler settings, you might get no or a	3607	In this section the complexities of (many of) the algorithms used inside
3611	lot of warnings when compiling libev code. Some people are apparently	3608	libev will be documented. For complexity discussions about backends see
3612	scared by this.	3609	the documentation for C<ev_default_init>.
3613		3610
3614	However, these are unavoidable for many reasons. For one, each compiler	3611	All of the following are about amortised time: If an array needs to be
3615	has different warnings, and each user has different tastes regarding	3612	extended, libev needs to realloc and move the whole array, but this
3616	warning options. "Warn-free" code therefore cannot be a goal except when	3613	happens asymptotically rarer with higher number of elements, so O(1) might
3617	targeting a specific compiler and compiler-version.	3614	mean that libev does a lengthy realloc operation in rare cases, but on
		3615	average it is much faster and asymptotically approaches constant time.
3618		3616
3619	Another reason is that some compiler warnings require elaborate	3617	=over 4
3620	workarounds, or other changes to the code that make it less clear and less
3621	maintainable.
3622		3618
3623	And of course, some compiler warnings are just plain stupid, or simply	3619	=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
3624	wrong (because they don't actually warn about the condition their message
3625	seems to warn about).
3626		3620
3627	While libev is written to generate as few warnings as possible,	3621	This means that, when you have a watcher that triggers in one hour and
3628	"warn-free" code is not a goal, and it is recommended not to build libev	3622	there are 100 watchers that would trigger before that, then inserting will
3629	with any compiler warnings enabled unless you are prepared to cope with	3623	have to skip roughly seven (C<ld 100>) of these watchers.
3630	them (e.g. by ignoring them). Remember that warnings are just that:
3631	warnings, not errors, or proof of bugs.
3632		3624
		3625	=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
3633		3626
3634	=head1 VALGRIND	3627	That means that changing a timer costs less than removing/adding them,
		3628	as only the relative motion in the event queue has to be paid for.
3635		3629
3636	Valgrind has a special section here because it is a popular tool that is	3630	=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
3637	highly useful, but valgrind reports are very hard to interpret.
3638		3631
3639	If you think you found a bug (memory leak, uninitialised data access etc.)	3632	These just add the watcher into an array or at the head of a list.
3640	in libev, then check twice: If valgrind reports something like:
3641		3633
3642	==2274== definitely lost: 0 bytes in 0 blocks.	3634	=item Stopping check/prepare/idle/fork/async watchers: O(1)
3643	==2274== possibly lost: 0 bytes in 0 blocks.
3644	==2274== still reachable: 256 bytes in 1 blocks.
3645		3635
3646	Then there is no memory leak. Similarly, under some circumstances,	3636	=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
3647	valgrind might report kernel bugs as if it were a bug in libev, or it
3648	might be confused (it is a very good tool, but only a tool).
3649		3637
3650	If you are unsure about something, feel free to contact the mailing list	3638	These watchers are stored in lists, so they need to be walked to find the
3651	with the full valgrind report and an explanation on why you think this is	3639	correct watcher to remove. The lists are usually short (you don't usually
3652	a bug in libev. However, don't be annoyed when you get a brisk "this is	3640	have many watchers waiting for the same fd or signal: one is typical, two
3653	no bug" answer and take the chance of learning how to interpret valgrind	3641	is rare).
3654	properly.
3655		3642
3656	If you need, for some reason, empty reports from valgrind for your project	3643	=item Finding the next timer in each loop iteration: O(1)
3657	I suggest using suppression lists.	3644
		3645	By virtue of using a binary or 4-heap, the next timer is always found at a
		3646	fixed position in the storage array.
		3647
		3648	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
		3649
		3650	A change means an I/O watcher gets started or stopped, which requires
		3651	libev to recalculate its status (and possibly tell the kernel, depending
		3652	on backend and whether C<ev_io_set> was used).
		3653
		3654	=item Activating one watcher (putting it into the pending state): O(1)
		3655
		3656	=item Priority handling: O(number_of_priorities)
		3657
		3658	Priorities are implemented by allocating some space for each
		3659	priority. When doing priority-based operations, libev usually has to
		3660	linearly search all the priorities, but starting/stopping and activating
		3661	watchers becomes O(1) with respect to priority handling.
		3662
		3663	=item Sending an ev_async: O(1)
		3664
		3665	=item Processing ev_async_send: O(number_of_async_watchers)
		3666
		3667	=item Processing signals: O(max_signal_number)
		3668
		3669	Sending involves a system call I<iff> there were no other C<ev_async_send>
		3670	calls in the current loop iteration. Checking for async and signal events
		3671	involves iterating over all running async watchers or all signal numbers.
		3672
		3673	=back
3658		3674
3659		3675
3660	=head1 AUTHOR	3676	=head1 AUTHOR
3661		3677
3662	Marc Lehmann <libev@schmorp.de>.	3678	Marc Lehmann <libev@schmorp.de>.

Diff Legend

-–
+Removed lines
-+
+Added lines
-<
+Changed lines
->
+Changed lines

Comparing libev/ev.pod (file contents): Revision 1.188 by root, Tue Sep 30 18:35:47 2008 UTC vs. Revision 1.193 by root, Wed Oct 1 04:25:25 2008 UTC

Diff Legend

Comparing libev/ev.pod (file contents):
Revision 1.188 by root, Tue Sep 30 18:35:47 2008 UTC vs.
Revision 1.193 by root, Wed Oct 1 04:25:25 2008 UTC