| 1 |
=head1 NAME |
| 2 |
|
| 3 |
Coro::State - first class continuations |
| 4 |
|
| 5 |
=head1 SYNOPSIS |
| 6 |
|
| 7 |
use Coro::State; |
| 8 |
|
| 9 |
$new = new Coro::State sub { |
| 10 |
print "in coro (called with @_), switching back\n"; |
| 11 |
$new->transfer ($main); |
| 12 |
print "in coro again, switching back\n"; |
| 13 |
$new->transfer ($main); |
| 14 |
}, 5; |
| 15 |
|
| 16 |
$main = new Coro::State; |
| 17 |
|
| 18 |
print "in main, switching to coro\n"; |
| 19 |
$main->transfer ($new); |
| 20 |
print "back in main, switch to coro again\n"; |
| 21 |
$main->transfer ($new); |
| 22 |
print "back in main\n"; |
| 23 |
|
| 24 |
=head1 DESCRIPTION |
| 25 |
|
| 26 |
This module implements coro. Coros, similar to threads and continuations, |
| 27 |
allow you to run more than one "thread of execution" in parallel. Unlike |
| 28 |
so-called "kernel" threads, there is no parallelism and only voluntary |
| 29 |
switching is used so locking problems are greatly reduced. The latter is |
| 30 |
called "cooperative" threading as opposed to "preemptive" threading. |
| 31 |
|
| 32 |
This can be used to implement non-local jumps, exception handling, |
| 33 |
continuation objects and more. |
| 34 |
|
| 35 |
This module provides only low-level functionality. See L<Coro> and related |
| 36 |
modules for a higher level threads abstraction including a scheduler. |
| 37 |
|
| 38 |
=head2 MODEL |
| 39 |
|
| 40 |
Coro::State implements two different thread models: Perl and C. The C |
| 41 |
threads (called cctx's) are basically simplified perl interpreters |
| 42 |
running/interpreting the Perl threads. A single interpreter can run any |
| 43 |
number of Perl threads, so usually there are very few C threads. |
| 44 |
|
| 45 |
When Perl code calls a C function (e.g. in an extension module) and that C |
| 46 |
function then calls back into Perl or transfers control to another thread, |
| 47 |
the C thread can no longer execute other Perl threads, so it stays tied to |
| 48 |
the specific thread until it returns to the original Perl caller, after |
| 49 |
which it is again available to run other Perl threads. |
| 50 |
|
| 51 |
The main program always has its own "C thread" (which really is |
| 52 |
*the* Perl interpreter running the whole program), so there will always |
| 53 |
be at least one additional C thread. You can use the debugger (see |
| 54 |
L<Coro::Debug>) to find out which threads are tied to their cctx and |
| 55 |
which aren't. |
| 56 |
|
| 57 |
=head2 MEMORY CONSUMPTION |
| 58 |
|
| 59 |
A newly created Coro::State that has not been used only allocates a |
| 60 |
relatively small (a hundred bytes) structure. Only on the first |
| 61 |
C<transfer> will perl allocate stacks (a few kb, 64 bit architetcures |
| 62 |
use twice as much, i.e. a few kb :) and optionally a C stack/thread |
| 63 |
(cctx) for threads that recurse through C functions. All this is very |
| 64 |
system-dependent. On my x86-pc-linux-gnu system this amounts to about 2k |
| 65 |
per (non-trivial but simple) Coro::State. |
| 66 |
|
| 67 |
You can view the actual memory consumption using Coro::Debug. Keep in mind |
| 68 |
that a for loop or other block constructs can easily consume 100-200 bytes |
| 69 |
per nesting level. |
| 70 |
|
| 71 |
=cut |
| 72 |
|
| 73 |
package Coro::State; |
| 74 |
|
| 75 |
use common::sense; |
| 76 |
|
| 77 |
use Carp; |
| 78 |
|
| 79 |
use Time::HiRes (); # currently only used for PerlIO::cede |
| 80 |
|
| 81 |
our $DIEHOOK; |
| 82 |
our $WARNHOOK; |
| 83 |
|
| 84 |
BEGIN { |
| 85 |
$DIEHOOK = sub { }; |
| 86 |
$WARNHOOK = sub { warn $_[0] }; |
| 87 |
} |
| 88 |
|
| 89 |
sub diehook { &$DIEHOOK } |
| 90 |
sub warnhook { &$WARNHOOK } |
| 91 |
|
| 92 |
use XSLoader; |
| 93 |
|
| 94 |
BEGIN { |
| 95 |
our $VERSION = 5.21; |
| 96 |
|
| 97 |
# must be done here because the xs part expects it to exist |
| 98 |
# it might exist already because Coro::Specific created it. |
| 99 |
$Coro::current ||= { }; |
| 100 |
|
| 101 |
{ |
| 102 |
# save/restore the handlers before/after overwriting %SIG magic |
| 103 |
local $SIG{__DIE__}; |
| 104 |
local $SIG{__WARN__}; |
| 105 |
|
| 106 |
XSLoader::load __PACKAGE__, $VERSION; |
| 107 |
} |
| 108 |
|
| 109 |
# need to do it after overwriting the %SIG magic |
| 110 |
$SIG{__DIE__} ||= \&diehook; |
| 111 |
$SIG{__WARN__} ||= \&warnhook; |
| 112 |
} |
| 113 |
|
| 114 |
use Exporter; |
| 115 |
use base Exporter::; |
| 116 |
|
| 117 |
=head2 GLOBAL VARIABLES |
| 118 |
|
| 119 |
=over 4 |
| 120 |
|
| 121 |
=item $Coro::State::DIEHOOK |
| 122 |
|
| 123 |
This works similarly to C<$SIG{__DIE__}> and is used as the default die |
| 124 |
hook for newly created Coro::States. This is useful if you want some generic |
| 125 |
logging function that works for all threads that don't set their own |
| 126 |
hook. |
| 127 |
|
| 128 |
When Coro::State is first loaded it will install these handlers for the |
| 129 |
main program, too, unless they have been overwritten already. |
| 130 |
|
| 131 |
The default handlers provided will behave like the built-in ones (as if |
| 132 |
they weren't there). |
| 133 |
|
| 134 |
If you don't want to exit your program on uncaught exceptions, you must |
| 135 |
not return from your die hook - call C<Coro::terminate> instead. |
| 136 |
|
| 137 |
Note 1: You I<must> store a valid code reference in these variables, |
| 138 |
C<undef> will I<not> do. |
| 139 |
|
| 140 |
Note 2: The value of this variable will be shared among all threads, so |
| 141 |
changing its value will change it in all threads that don't have their |
| 142 |
own die handler. |
| 143 |
|
| 144 |
=item $Coro::State::WARNHOOK |
| 145 |
|
| 146 |
Similar to above die hook, but augments C<$SIG{__WARN__}>. |
| 147 |
|
| 148 |
=back |
| 149 |
|
| 150 |
=head2 FUNCTIONS |
| 151 |
|
| 152 |
=over 4 |
| 153 |
|
| 154 |
=item $coro = new Coro::State [$coderef[, @args...]] |
| 155 |
|
| 156 |
Create a new Coro::State thread object and return it. The first |
| 157 |
C<transfer> call to this thread will start execution at the given |
| 158 |
coderef, with the given arguments. |
| 159 |
|
| 160 |
Note that the arguments will not be copied. Instead, as with normal |
| 161 |
function calls, the thread receives passed arguments by reference, so |
| 162 |
make sure you don't change them in unexpected ways. |
| 163 |
|
| 164 |
Returning from such a thread is I<NOT> supported. Neither is calling |
| 165 |
C<exit> or throwing an uncaught exception. The following paragraphs |
| 166 |
describe what happens in current versions of Coro. |
| 167 |
|
| 168 |
If the subroutine returns the program will be terminated as if execution |
| 169 |
of the main program ended. |
| 170 |
|
| 171 |
If it throws an exception the program will terminate unless the exception |
| 172 |
is caught, exactly like in the main program. |
| 173 |
|
| 174 |
Calling C<exit> in a thread does the same as calling it in the main |
| 175 |
program, but due to libc bugs on many BSDs, this doesn't work reliable |
| 176 |
everywhere. |
| 177 |
|
| 178 |
If the coderef is omitted this function will create a new "empty" |
| 179 |
thread, i.e. a thread that cannot be transfered to but can be used |
| 180 |
to save the current thread state in (note that this is dangerous, as no |
| 181 |
reference is taken to ensure that the "current thread state" survives, |
| 182 |
the caller is responsible to ensure that the cloned state does not go |
| 183 |
away). |
| 184 |
|
| 185 |
The returned object is an empty hash which can be used for any purpose |
| 186 |
whatsoever, for example when subclassing Coro::State. |
| 187 |
|
| 188 |
Certain variables are "localised" to each thread, that is, certain |
| 189 |
"global" variables are actually per thread. Not everything that would |
| 190 |
sensibly be localised currently is, and not everything that is localised |
| 191 |
makes sense for every application, and the future might bring changes. |
| 192 |
|
| 193 |
The following global variables can have different values per thread, |
| 194 |
and have the stated initial values: |
| 195 |
|
| 196 |
Variable Initial Value |
| 197 |
@_ whatever arguments were passed to the Coro |
| 198 |
$_ undef |
| 199 |
$@ undef |
| 200 |
$/ "\n" |
| 201 |
$SIG{__DIE__} aliased to $Coro::State::DIEHOOK(*) |
| 202 |
$SIG{__WARN__} aliased to $Coro::State::WARNHOOK(*) |
| 203 |
(default fh) *STDOUT |
| 204 |
$^H, %^H zero/empty. |
| 205 |
$1, $2... all regex results are initially undefined |
| 206 |
|
| 207 |
(*) reading the value from %SIG is not supported, but local'ising is. |
| 208 |
|
| 209 |
If you feel that something important is missing then tell me. Also |
| 210 |
remember that every function call that might call C<transfer> (such |
| 211 |
as C<Coro::Channel::put>) might clobber any global and/or special |
| 212 |
variables. Yes, this is by design ;) You can always create your own |
| 213 |
process abstraction model that saves these variables. |
| 214 |
|
| 215 |
The easiest way to do this is to create your own scheduling primitive like |
| 216 |
in the code below, and use it in your threads: |
| 217 |
|
| 218 |
sub my_cede { |
| 219 |
local ($;, ...); |
| 220 |
Coro::cede; |
| 221 |
} |
| 222 |
|
| 223 |
Another way is to use dynamic winders, see C<Coro::on_enter> and |
| 224 |
C<Coro::on_leave> for this. |
| 225 |
|
| 226 |
=item $prev->transfer ($next) |
| 227 |
|
| 228 |
Save the state of the current subroutine in C<$prev> and switch to the |
| 229 |
thread saved in C<$next>. |
| 230 |
|
| 231 |
The "state" of a subroutine includes the scope, i.e. lexical variables and |
| 232 |
the current execution state (subroutine, stack). |
| 233 |
|
| 234 |
=item $state->is_new |
| 235 |
|
| 236 |
Returns true iff this Coro::State object is "new", i.e. has never been run |
| 237 |
yet. Those states basically consist of only the code reference to call and |
| 238 |
the arguments, but consumes very little other resources. New states will |
| 239 |
automatically get assigned a perl interpreter when they are transfered to. |
| 240 |
|
| 241 |
=item $state->is_destroyed |
| 242 |
|
| 243 |
Returns true iff the Coro::State object has been destroyed (by |
| 244 |
C<cancel>), i.e. it's resources freed because they were C<cancel>'d (or |
| 245 |
C<terminate>'d). |
| 246 |
|
| 247 |
=item $state->cancel |
| 248 |
|
| 249 |
Forcefully destructs the given Coro::State. While you can keep the |
| 250 |
reference, and some memory is still allocated, the Coro::State object is |
| 251 |
effecticely dead, destructors have been freed, it cannot be transfered to |
| 252 |
anymore. |
| 253 |
|
| 254 |
=item $state->throw ([$scalar]) |
| 255 |
|
| 256 |
See L<< Coro->throw >>. |
| 257 |
|
| 258 |
=item $state->call ($coderef) |
| 259 |
|
| 260 |
Try to call the given C<$coderef> in the context of the given state. This |
| 261 |
works even when the state is currently within an XS function, and can |
| 262 |
be very dangerous. You can use it to acquire stack traces etc. (see the |
| 263 |
Coro::Debug module for more details). The coderef MUST NOT EVER transfer |
| 264 |
to another state. |
| 265 |
|
| 266 |
=item $state->eval ($string) |
| 267 |
|
| 268 |
Like C<call>, but eval's the string. Dangerous. |
| 269 |
|
| 270 |
=item $state->swap_defsv |
| 271 |
|
| 272 |
=item $state->swap_defav |
| 273 |
|
| 274 |
Swap the current C<$_> (swap_defsv) or C<@_> (swap_defav) with the |
| 275 |
equivalent in the saved state of C<$state>. This can be used to give the |
| 276 |
coro a defined content for C<@_> and C<$_> before transfer'ing to it. |
| 277 |
|
| 278 |
=item $state->swap_sv (\$sv, \$swap_sv) |
| 279 |
|
| 280 |
This (very advanced) function can be used to make I<any> variable local to |
| 281 |
a thread. |
| 282 |
|
| 283 |
It works by swapping the contents of C<$sv> and C<$swap_sv> each time the |
| 284 |
thread is entered and left again, i.e. it is similarly to: |
| 285 |
|
| 286 |
$tmp = $sv; $sv = $swap_sv; $swap_sv = $tmp; |
| 287 |
|
| 288 |
Except that it doesn't make an copies and works on hashes and even more |
| 289 |
exotic values (code references!). |
| 290 |
|
| 291 |
Needless to say, this function can be very very dangerous: you can easily |
| 292 |
swap a hash with a reference (i.e. C<%hash> I<becomes> a reference), and perl |
| 293 |
will not like this at all. |
| 294 |
|
| 295 |
It will also swap "magicalness" - so when swapping a builtin perl variable |
| 296 |
(such as C<$.>), it will lose it's magicalness, which, again, perl will |
| 297 |
not like, so don't do it. |
| 298 |
|
| 299 |
Lastly, the C<$swap_sv> itself will be used, not a copy, so make sure you |
| 300 |
give each thread it's own C<$swap_sv> instance. |
| 301 |
|
| 302 |
It is, however, quite safe to swap some normal variable with |
| 303 |
another. For example, L<PApp::SQL> stores the default database handle in |
| 304 |
C<$PApp::SQL::DBH>. To make this a per-thread variable, use this: |
| 305 |
|
| 306 |
my $private_dbh = ...; |
| 307 |
$coro->swap_sv (\$PApp::SQL::DBH, \$private_dbh); |
| 308 |
|
| 309 |
This results in C<$PApp::SQL::DBH> having the value of C<$private_dbh> |
| 310 |
while it executes, and whatever other value it had when it doesn't |
| 311 |
execute. |
| 312 |
|
| 313 |
You can also swap hashes and other values: |
| 314 |
|
| 315 |
my %private_hash; |
| 316 |
$coro->swap_sv (\%some_hash, \%private_hash); |
| 317 |
|
| 318 |
=item $state->trace ($flags) |
| 319 |
|
| 320 |
Internal function to control tracing. I just mention this so you can stay |
| 321 |
away from abusing it. |
| 322 |
|
| 323 |
=item $bytes = $state->rss |
| 324 |
|
| 325 |
Returns the memory allocated by the coro (which includes static |
| 326 |
structures, various perl stacks but NOT local variables, arguments or any |
| 327 |
C context data). This is a rough indication of how much memory it might |
| 328 |
use. |
| 329 |
|
| 330 |
=item $state->has_cctx |
| 331 |
|
| 332 |
Returns whether the state currently uses a cctx/C context. An active |
| 333 |
state always has a cctx, as well as the main program. Other states only |
| 334 |
use a cctxts when needed. |
| 335 |
|
| 336 |
=item Coro::State::force_cctx |
| 337 |
|
| 338 |
Forces the allocation of a C context for the currently running coro |
| 339 |
(if not already done). Apart from benchmarking there is little point |
| 340 |
in doing so, however. |
| 341 |
|
| 342 |
=item $ncctx = Coro::State::cctx_count |
| 343 |
|
| 344 |
Returns the number of C-level coro allocated. If this number is |
| 345 |
very high (more than a dozen) it might help to identify points of C-level |
| 346 |
recursion in your code and moving this into a separate coro. |
| 347 |
|
| 348 |
=item $nidle = Coro::State::cctx_idle |
| 349 |
|
| 350 |
Returns the number of allocated but idle (free for reuse) C level |
| 351 |
coro. Currently, Coro will limit the number of idle/unused cctxs to |
| 352 |
8. |
| 353 |
|
| 354 |
=item $old = Coro::State::cctx_stacksize [$new_stacksize] |
| 355 |
|
| 356 |
Returns the current C stack size and optionally sets the new I<minimum> |
| 357 |
stack size to C<$new_stacksize> I<long>s. Existing stacks will not |
| 358 |
be changed, but Coro will try to replace smaller stacks as soon as |
| 359 |
possible. Any Coro::State that starts to use a stack after this call is |
| 360 |
guaranteed this minimum stack size. |
| 361 |
|
| 362 |
Please note that coros will only need to use a C-level stack if the |
| 363 |
interpreter recurses or calls a function in a module that calls back into |
| 364 |
the interpreter, so use of this feature is usually never needed. |
| 365 |
|
| 366 |
=item $old = Coro::State::cctx_max_idle [$new_count] |
| 367 |
|
| 368 |
Coro caches C contexts that are not in use currently, as creating them |
| 369 |
from scratch has some overhead. |
| 370 |
|
| 371 |
This function returns the current maximum number of idle C contexts and |
| 372 |
optionally sets the new amount. The count must be at least C<1>, with the |
| 373 |
default being C<4>. |
| 374 |
|
| 375 |
=item @states = Coro::State::list |
| 376 |
|
| 377 |
Returns a list of all states currently allocated. |
| 378 |
|
| 379 |
=item $was_enabled = Coro::State::enable_times [$enable] |
| 380 |
|
| 381 |
Enables/disables/queries the current state of per-thread real and |
| 382 |
cpu-time gathering. |
| 383 |
|
| 384 |
When enabled, the real time and the cpu time (user + system time) |
| 385 |
spent in each thread is accumulated. If disabled, then the accumulated |
| 386 |
times will stay as they are (they start at 0). |
| 387 |
|
| 388 |
Currently, cpu time is only measured on GNU/Linux systems, all other |
| 389 |
systems only gather real time. |
| 390 |
|
| 391 |
Enabling time profiling slows down thread switching by a factor of 2 to |
| 392 |
10, depending on platform on hardware. |
| 393 |
|
| 394 |
The times will be displayed when running C<Coro::Debug::command "ps">, and |
| 395 |
cna be queried by calling C<< $state->times >>. |
| 396 |
|
| 397 |
=item ($real, $cpu) = $state->times |
| 398 |
|
| 399 |
Returns the real time and cpu times spent in the given C<$state>. See |
| 400 |
C<Coro::State::enable_times> for more info. |
| 401 |
|
| 402 |
=item $clone = $state->clone |
| 403 |
|
| 404 |
This exciting method takes a Coro::State object and clones it, i.e., it |
| 405 |
creates a copy. This makes it possible to restore a state more than once, |
| 406 |
and even return to states that have returned or have been terminated. |
| 407 |
|
| 408 |
Since its only known purpose is for intellectual self-gratification, and |
| 409 |
because it is a difficult piece of code, it is not enabled by default, and |
| 410 |
not supported. |
| 411 |
|
| 412 |
Here are a few little-known facts: First, coros *are* full/true/real |
| 413 |
continuations. Secondly Coro::State objects (without clone) *are* first |
| 414 |
class continuations. Thirdly, nobody has ever found a use for the full |
| 415 |
power of call/cc that isn't better (faster, easier, more efficiently) |
| 416 |
implemented differently, and nobody has yet found a useful control |
| 417 |
construct that can't be implemented without it already, just much faster |
| 418 |
and with fewer resources. And lastly, Scheme's call/cc doesn't support |
| 419 |
using call/cc to implement threads. |
| 420 |
|
| 421 |
Among the games you can play with this is implementing a scheme-like |
| 422 |
call-with-current-continuation, as the following code does (well, with |
| 423 |
small differences). |
| 424 |
|
| 425 |
# perl disassociates from local lexicals on frame exit, |
| 426 |
# so use a global variable for return values. |
| 427 |
my @ret; |
| 428 |
|
| 429 |
sub callcc($@) { |
| 430 |
my ($func, @arg) = @_; |
| 431 |
|
| 432 |
my $continuation = new Coro::State; |
| 433 |
$continuation->transfer (new Coro::State sub { |
| 434 |
my $escape = sub { |
| 435 |
@ret = @_; |
| 436 |
Coro::State->new->transfer ($continuation->clone); |
| 437 |
}; |
| 438 |
$escape->($func->($escape, @arg)); |
| 439 |
}); |
| 440 |
|
| 441 |
my @ret_ = @ret; @ret = (); |
| 442 |
wantarray ? @ret_ : pop @ret_ |
| 443 |
} |
| 444 |
|
| 445 |
Which could be used to implement a loop like this: |
| 446 |
|
| 447 |
async { |
| 448 |
my $n; |
| 449 |
my $l = callcc sub { $_[0] }; |
| 450 |
|
| 451 |
$n++; |
| 452 |
print "iteration $n\n"; |
| 453 |
|
| 454 |
$l->($l) unless $n == 10; |
| 455 |
}; |
| 456 |
|
| 457 |
If you find this confusing, then you already understand the coolness of |
| 458 |
call/cc: It can turn anything into spaghetti code real fast. |
| 459 |
|
| 460 |
Besides, call/cc is much less useful in a Perl-like dynamic language (with |
| 461 |
references, and its scoping rules) then in, say, scheme. |
| 462 |
|
| 463 |
Now, the known limitations of C<clone>: |
| 464 |
|
| 465 |
It probably only works on perl 5.10; it cannot clone a coro inside |
| 466 |
the substition operator (but windows perl can't fork from there either) |
| 467 |
and some other contexts, and C<abort ()> is the preferred mechanism to |
| 468 |
signal errors. It cannot clone a state that has a c context attached |
| 469 |
(implementing clone on the C level is too hard for me to even try), |
| 470 |
which rules out calling call/cc from the main coro. It cannot |
| 471 |
clone a context that hasn't even been started yet. It doesn't work with |
| 472 |
C<-DDEBUGGING> (but what does). It probably also leaks, and sometimes |
| 473 |
triggers a few assertions inside Coro. Most of these limitations *are* |
| 474 |
fixable with some effort, but that's pointless just to make a point that |
| 475 |
it could be done. |
| 476 |
|
| 477 |
The current implementation could without doubt be optimised to be a |
| 478 |
constant-time operation by doing lazy stack copying, if somebody were |
| 479 |
insane enough to invest the time. |
| 480 |
|
| 481 |
=cut |
| 482 |
|
| 483 |
# used by Coro::Debug only atm. |
| 484 |
sub debug_desc { |
| 485 |
$_[0]{desc} |
| 486 |
} |
| 487 |
|
| 488 |
# for very deep reasons, we must initialise $Coro::main here. |
| 489 |
|
| 490 |
{ |
| 491 |
package Coro; |
| 492 |
|
| 493 |
our $main; # main coro |
| 494 |
our $current; # current coro |
| 495 |
|
| 496 |
$main = Coro::new Coro::; |
| 497 |
|
| 498 |
$main->{desc} = "[main::]"; |
| 499 |
|
| 500 |
# maybe some other module used Coro::Specific before... |
| 501 |
$main->{_specific} = $current->{_specific} |
| 502 |
if $current; |
| 503 |
|
| 504 |
_set_current $main; |
| 505 |
} |
| 506 |
|
| 507 |
# we also make sure we have Coro::AnyEvent when AnyEvent is used, |
| 508 |
# without loading or initialising AnyEvent |
| 509 |
if (defined $AnyEvent::MODEL) { |
| 510 |
require Coro::AnyEvent; |
| 511 |
} else { |
| 512 |
push @AnyEvent::post_detect, sub { require Coro::AnyEvent }; |
| 513 |
} |
| 514 |
|
| 515 |
1; |
| 516 |
|
| 517 |
=back |
| 518 |
|
| 519 |
=head1 BUGS |
| 520 |
|
| 521 |
This module is not thread-safe. You must only ever use this module from |
| 522 |
the same thread (this requirement might be removed in the future). |
| 523 |
|
| 524 |
=head1 SEE ALSO |
| 525 |
|
| 526 |
L<Coro>. |
| 527 |
|
| 528 |
=head1 AUTHOR |
| 529 |
|
| 530 |
Marc Lehmann <schmorp@schmorp.de> |
| 531 |
http://home.schmorp.de/ |
| 532 |
|
| 533 |
=cut |
| 534 |
|
