1 |
=head1 NAME |
2 |
|
3 |
Async::Interrupt - allow C/XS libraries to interrupt perl asynchronously |
4 |
|
5 |
=head1 SYNOPSIS |
6 |
|
7 |
use Async::Interrupt; |
8 |
|
9 |
=head1 DESCRIPTION |
10 |
|
11 |
This module implements a single feature only of interest to advanced perl |
12 |
modules, namely asynchronous interruptions (think "UNIX signals", which |
13 |
are very similar). |
14 |
|
15 |
Sometimes, modules wish to run code asynchronously (in another thread, |
16 |
or from a signal handler), and then signal the perl interpreter on |
17 |
certain events. One common way is to write some data to a pipe and use an |
18 |
event handling toolkit to watch for I/O events. Another way is to send |
19 |
a signal. Those methods are slow, and in the case of a pipe, also not |
20 |
asynchronous - it won't interrupt a running perl interpreter. |
21 |
|
22 |
This module implements asynchronous notifications that enable you to |
23 |
signal running perl code from another thread, asynchronously, and |
24 |
sometimes even without using a single syscall. |
25 |
|
26 |
=head2 USAGE SCENARIOS |
27 |
|
28 |
=over 4 |
29 |
|
30 |
=item Race-free signal handling |
31 |
|
32 |
There seems to be no way to do race-free signal handling in perl: to |
33 |
catch a signal, you have to execute Perl code, and between entering the |
34 |
interpreter C<select> function (or other blocking functions) and executing |
35 |
the select syscall is a small but relevant timespan during which signals |
36 |
will be queued, but perl signal handlers will not be executed and the |
37 |
blocking syscall will not be interrupted. |
38 |
|
39 |
You can use this module to bind a signal to a callback while at the same |
40 |
time activating an event pipe that you can C<select> on, fixing the race |
41 |
completely. |
42 |
|
43 |
This can be used to implement the signal hadling in event loops, |
44 |
e.g. L<AnyEvent>, L<POE>, L<IO::Async::Loop> and so on. |
45 |
|
46 |
=item Background threads want speedy reporting |
47 |
|
48 |
Assume you want very exact timing, and you can spare an extra cpu core |
49 |
for that. Then you can run an extra thread that signals your perl |
50 |
interpreter. This means you can get a very exact timing source while your |
51 |
perl code is number crunching, without even using a syscall to communicate |
52 |
between your threads. |
53 |
|
54 |
For example the deliantra game server uses a variant of this technique |
55 |
to interrupt background processes regularly to send map updates to game |
56 |
clients. |
57 |
|
58 |
L<IO::AIO> and L<BDB> could also use this to speed up result reporting. |
59 |
|
60 |
=item Speedy event loop invocation |
61 |
|
62 |
One could use this module e.g. in L<Coro> to interrupt a running coro-thread |
63 |
and cause it to enter the event loop. |
64 |
|
65 |
Or one could bind to C<SIGIO> and tell some important sockets to send this |
66 |
signal, causing the event loop to be entered to reduce network latency. |
67 |
|
68 |
=back |
69 |
|
70 |
=head2 HOW TO USE |
71 |
|
72 |
You can use this module by creating an C<Async::Interrupt> object for each |
73 |
such event source. This object stores a perl and/or a C-level callback |
74 |
that is invoked when the C<Async::Interrupt> object gets signalled. It is |
75 |
executed at the next time the perl interpreter is running (i.e. it will |
76 |
interrupt a computation, but not an XS function or a syscall). |
77 |
|
78 |
You can signal the C<Async::Interrupt> object either by calling it's C<< |
79 |
->signal >> method, or, more commonly, by calling a C function. There is |
80 |
also the built-in (POSIX) signal source. |
81 |
|
82 |
The C<< ->signal_func >> returns the address of the C function that is to |
83 |
be called (plus an argument to be used during the call). The signalling |
84 |
function also takes an integer argument in the range SIG_ATOMIC_MIN to |
85 |
SIG_ATOMIC_MAX (guaranteed to allow at least 0..127). |
86 |
|
87 |
Since this kind of interruption is fast, but can only interrupt a |
88 |
I<running> interpreter, there is optional support for signalling a pipe |
89 |
- that means you can also wait for the pipe to become readable (e.g. via |
90 |
L<EV> or L<AnyEvent>). This, of course, incurs the overhead of a C<read> |
91 |
and C<write> syscall. |
92 |
|
93 |
=head1 THE Async::Interrupt CLASS |
94 |
|
95 |
=over 4 |
96 |
|
97 |
=cut |
98 |
|
99 |
package Async::Interrupt; |
100 |
|
101 |
use common::sense; |
102 |
|
103 |
BEGIN { |
104 |
# the next line forces initialisation of internal |
105 |
# signal handling # variables |
106 |
$SIG{KILL} = sub { }; |
107 |
|
108 |
our $VERSION = '0.6'; |
109 |
|
110 |
require XSLoader; |
111 |
XSLoader::load ("Async::Interrupt", $VERSION); |
112 |
} |
113 |
|
114 |
our $DIED = sub { warn "$@" }; |
115 |
|
116 |
=item $async = new Async::Interrupt key => value... |
117 |
|
118 |
Creates a new Async::Interrupt object. You may only use async |
119 |
notifications on this object while it exists, so you need to keep a |
120 |
reference to it at all times while it is used. |
121 |
|
122 |
Optional constructor arguments include (normally you would specify at |
123 |
least one of C<cb> or C<c_cb>). |
124 |
|
125 |
=over 4 |
126 |
|
127 |
=item cb => $coderef->($value) |
128 |
|
129 |
Registers a perl callback to be invoked whenever the async interrupt is |
130 |
signalled. |
131 |
|
132 |
Note that, since this callback can be invoked at basically any time, it |
133 |
must not modify any well-known global variables such as C<$/> without |
134 |
restoring them again before returning. |
135 |
|
136 |
The exceptions are C<$!> and C<$@>, which are saved and restored by |
137 |
Async::Interrupt. |
138 |
|
139 |
If the callback should throw an exception, then it will be caught, |
140 |
and C<$Async::Interrupt::DIED> will be called with C<$@> containing |
141 |
the exception. The default will simply C<warn> about the message and |
142 |
continue. |
143 |
|
144 |
=item c_cb => [$c_func, $c_arg] |
145 |
|
146 |
Registers a C callback the be invoked whenever the async interrupt is |
147 |
signalled. |
148 |
|
149 |
The C callback must have the following prototype: |
150 |
|
151 |
void c_func (pTHX_ void *c_arg, int value); |
152 |
|
153 |
Both C<$c_func> and C<$c_arg> must be specified as integers/IVs, and |
154 |
C<$value> is the C<value> passed to some earlier call to either C<$signal> |
155 |
or the C<signal_func> function. |
156 |
|
157 |
Note that, because the callback can be invoked at almost any time, you |
158 |
have to be careful at saving and restoring global variables that Perl |
159 |
might use (the exception is C<errno>, which is saved and restored by |
160 |
Async::Interrupt). The callback itself runs as part of the perl context, |
161 |
so you can call any perl functions and modify any perl data structures (in |
162 |
which case the requirements set out for C<cb> apply as well). |
163 |
|
164 |
=item var => $scalar_ref |
165 |
|
166 |
When specified, then the given argument must be a reference to a |
167 |
scalar. The scalar will be set to C<0> intiially. Signalling the interrupt |
168 |
object will set it to the passed value, handling the interrupt will reset |
169 |
it to C<0> again. |
170 |
|
171 |
Note that the only thing you are legally allowed to do is to is to check |
172 |
the variable in a boolean or integer context (e.g. comparing it with a |
173 |
string, or printing it, will I<destroy> it and might cause your program to |
174 |
crash or worse). |
175 |
|
176 |
=item signal => $signame_or_value |
177 |
|
178 |
When this parameter is specified, then the Async::Interrupt will hook the |
179 |
given signal, that is, it will effectively call C<< ->signal (0) >> each time |
180 |
the given signal is caught by the process. |
181 |
|
182 |
Only one async can hook a given signal, and the signal will be restored to |
183 |
defaults when the Async::Interrupt object gets destroyed. |
184 |
|
185 |
=item pipe => [$fileno_or_fh_for_reading, $fileno_or_fh_for_writing] |
186 |
|
187 |
Specifies two file descriptors (or file handles) that should be signalled |
188 |
whenever the async interrupt is signalled. This means a single octet will |
189 |
be written to it, and before the callback is being invoked, it will be |
190 |
read again. Due to races, it is unlikely but possible that multiple octets |
191 |
are written. It is required that the file handles are both in nonblocking |
192 |
mode. |
193 |
|
194 |
The object will keep a reference to the file handles. |
195 |
|
196 |
This can be used to ensure that async notifications will interrupt event |
197 |
frameworks as well. |
198 |
|
199 |
Note that C<Async::Interrupt> will create a suitable signal fd |
200 |
automatically when your program requests one, so you don't have to specify |
201 |
this argument when all you want is an extra file descriptor to watch. |
202 |
|
203 |
If you want to share a single event pipe between multiple Async::Interrupt |
204 |
objects, you can use the C<Async::Interrupt::EventPipe> class to manage |
205 |
those. |
206 |
|
207 |
=back |
208 |
|
209 |
=cut |
210 |
|
211 |
sub new { |
212 |
my ($class, %arg) = @_; |
213 |
|
214 |
bless \(_alloc $arg{cb}, @{$arg{c_cb}}[0,1], @{$arg{pipe}}[0,1], $arg{signal}, $arg{var}), $class |
215 |
} |
216 |
|
217 |
=item ($signal_func, $signal_arg) = $async->signal_func |
218 |
|
219 |
Returns the address of a function to call asynchronously. The function has |
220 |
the following prototype and needs to be passed the specified C<$c_arg>, |
221 |
which is a C<void *> cast to C<IV>: |
222 |
|
223 |
void (*signal_func) (void *signal_arg, int value) |
224 |
|
225 |
An example call would look like: |
226 |
|
227 |
signal_func (signal_arg, 0); |
228 |
|
229 |
The function is safe to call from within signal and thread contexts, at |
230 |
any time. The specified C<value> is passed to both C and Perl callback. |
231 |
|
232 |
C<$value> must be in the valid range for a C<sig_atomic_t>, except C<0> |
233 |
(1..127 is portable). |
234 |
|
235 |
If the function is called while the Async::Interrupt object is already |
236 |
signaled but before the callbacks are being executed, then the stored |
237 |
C<value> is either the old or the new one. Due to the asynchronous |
238 |
nature of the code, the C<value> can even be passed to two consecutive |
239 |
invocations of the callback. |
240 |
|
241 |
=item $address = $async->c_var |
242 |
|
243 |
Returns the address (cast to IV) of an C<IV> variable. The variable is set |
244 |
to C<0> initially and gets set to the passed value whenever the object |
245 |
gets signalled, and reset to C<0> once the interrupt has been handled. |
246 |
|
247 |
Note that it is often beneficial to just call C<PERL_ASYNC_CHECK ()> to |
248 |
handle any interrupts. |
249 |
|
250 |
Example: call some XS function to store the address, then show C code |
251 |
waiting for it. |
252 |
|
253 |
my_xs_func $async->c_var; |
254 |
|
255 |
static IV *valuep; |
256 |
|
257 |
void |
258 |
my_xs_func (void *addr) |
259 |
CODE: |
260 |
valuep = (IV *)addr; |
261 |
|
262 |
// code in a loop, waiting |
263 |
while (!*valuep) |
264 |
; // do something |
265 |
|
266 |
=item $async->signal ($value=1) |
267 |
|
268 |
This signals the given async object from Perl code. Semi-obviously, this |
269 |
will instantly trigger the callback invocation. |
270 |
|
271 |
C<$value> must be in the valid range for a C<sig_atomic_t>, except C<0> |
272 |
(1..127 is portable). |
273 |
|
274 |
=item $async->block |
275 |
|
276 |
=item $async->unblock |
277 |
|
278 |
Sometimes you need a "critical section" of code that will not be |
279 |
interrupted by an Async::Interrupt. This can be implemented by calling C<< |
280 |
$async->block >> before the critical section, and C<< $async->unblock >> |
281 |
afterwards. |
282 |
|
283 |
Note that there must be exactly one call of C<unblock> for every previous |
284 |
call to C<block> (i.e. calls can nest). |
285 |
|
286 |
Since ensuring this in the presence of exceptions and threads is |
287 |
usually more difficult than you imagine, I recommend using C<< |
288 |
$async->scoped_block >> instead. |
289 |
|
290 |
=item $async->scope_block |
291 |
|
292 |
This call C<< $async->block >> and installs a handler that is called when |
293 |
the current scope is exited (via an exception, by canceling the Coro |
294 |
thread, by calling last/goto etc.). |
295 |
|
296 |
This is the recommended (and fastest) way to implement critical sections. |
297 |
|
298 |
=item $async->pipe_enable |
299 |
|
300 |
=item $async->pipe_disable |
301 |
|
302 |
Enable/disable signalling the pipe when the interrupt occurs (default is |
303 |
enabled). Writing to a pipe is relatively expensive, so it can be disabled |
304 |
when you know you are not waiting for it (for example, with L<EV> you |
305 |
could disable the pipe in a check watcher, and enable it in a prepare |
306 |
watcher). |
307 |
|
308 |
Note that currently, while C<pipe_disable> is in effect, no attempt to |
309 |
read from the pipe will be done when handling events. This might change as |
310 |
soon as I realize why this is a mistake. |
311 |
|
312 |
=item $fileno = $async->pipe_fileno |
313 |
|
314 |
Returns the reading side of the signalling pipe. If no signalling pipe is |
315 |
currently attached to the object, it will dynamically create one. |
316 |
|
317 |
Note that the only valid oepration on this file descriptor is to wait |
318 |
until it is readable. The fd might belong currently to a pipe, a tcp |
319 |
socket, or an eventfd, depending on the platform, and is guaranteed to be |
320 |
C<select>able. |
321 |
|
322 |
=item $async->pipe_autodrain ($enable) |
323 |
|
324 |
Enables (C<1>) or disables (C<0>) automatic draining of the pipe (default: |
325 |
enabled). When automatic draining is enabled, then Async::Interrupt will |
326 |
automatically clear the pipe. Otherwise the user is responsible for this |
327 |
draining. |
328 |
|
329 |
This is useful when you want to share one pipe among many Async::Interrupt |
330 |
objects. |
331 |
|
332 |
=item $async->post_fork |
333 |
|
334 |
The object will not normally be usable after a fork (as the pipe fd is |
335 |
shared between processes). Calling this method after a fork in the child |
336 |
ensures that the object will work as expected again. It only needs to be |
337 |
called when the async object is used in the child. |
338 |
|
339 |
This only works when the pipe was created by Async::Interrupt. |
340 |
|
341 |
Async::Interrupt ensures that the reading file descriptor does not change |
342 |
it's value. |
343 |
|
344 |
=back |
345 |
|
346 |
=head1 THE Async::Interrupt::EventPipe CLASS |
347 |
|
348 |
Pipes are the predominent utility to make asynchronous signals |
349 |
synchronous. However, pipes are hard to come by: they don't exist on the |
350 |
broken windows platform, and on GNU/Linux systems, you might want to use |
351 |
an C<eventfd> instead. |
352 |
|
353 |
This class creates selectable event pipes in a portable fashion: on |
354 |
windows, it will try to create a tcp socket pair, on GNU/Linux, it will |
355 |
try to create an eventfd and everywhere else it will try to use a normal |
356 |
pipe. |
357 |
|
358 |
=over 4 |
359 |
|
360 |
=item $epipe = new Async::Interrupt::EventPipe |
361 |
|
362 |
This creates and returns an eventpipe object. This object is simply a |
363 |
blessed array reference: |
364 |
|
365 |
=item ($r_fd, $w_fd) = $epipe->filenos |
366 |
|
367 |
Returns the read-side file descriptor and the write-side file descriptor. |
368 |
|
369 |
Example: pass an eventpipe object as pipe to the Async::Interrupt |
370 |
constructor, and create an AnyEvent watcher for the read side. |
371 |
|
372 |
my $epipe = new Async::Interrupt::EventPipe; |
373 |
my $asy = new Async::Interrupt pipe => [$epipe->filenos]; |
374 |
my $iow = AnyEvent->io (fh => $epipe->fileno, poll => 'r', cb => sub { }); |
375 |
|
376 |
=item $r_fd = $epipe->fileno |
377 |
|
378 |
Return only the reading/listening side. |
379 |
|
380 |
=item $epipe->signal |
381 |
|
382 |
Write something to the pipe, in a portable fashion. |
383 |
|
384 |
=item $epipe->drain |
385 |
|
386 |
Drain (empty) the pipe. |
387 |
|
388 |
=item $epipe->renew |
389 |
|
390 |
Recreates the pipe (useful after a fork). The reading side will not change |
391 |
it's file descriptor number, but the writing side might. |
392 |
|
393 |
=back |
394 |
|
395 |
=cut |
396 |
|
397 |
1; |
398 |
|
399 |
=head1 EXAMPLE |
400 |
|
401 |
There really should be a complete C/XS example. Bug me about it. Better |
402 |
yet, create one. |
403 |
|
404 |
=head1 IMPLEMENTATION DETAILS AND LIMITATIONS |
405 |
|
406 |
This module works by "hijacking" SIGKILL, which is guaranteed to always |
407 |
exist, but also cannot be caught, so is always available. |
408 |
|
409 |
Basically, this module fakes the occurance of a SIGKILL signal and |
410 |
then intercepts the interpreter handling it. This makes normal signal |
411 |
handling slower (probably unmeasurably, though), but has the advantage |
412 |
of not requiring a special runops function, nor slowing down normal perl |
413 |
execution a bit. |
414 |
|
415 |
It assumes that C<sig_atomic_t>, C<int> and C<IV> are all async-safe to |
416 |
modify. |
417 |
|
418 |
=head1 AUTHOR |
419 |
|
420 |
Marc Lehmann <schmorp@schmorp.de> |
421 |
http://home.schmorp.de/ |
422 |
|
423 |
=cut |
424 |
|