ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/BDB/BDB.pm
(Generate patch)

Comparing BDB/BDB.pm (file contents):
Revision 1.4 by root, Mon Feb 5 23:46:15 2007 UTC vs.
Revision 1.22 by root, Mon Dec 10 03:57:27 2007 UTC

4 4
5=head1 SYNOPSIS 5=head1 SYNOPSIS
6 6
7 use BDB; 7 use BDB;
8 8
9 my $env = db_env_create;
10
11 mkdir "bdtest", 0700;
12 db_env_open
13 $env,
14 "bdtest",
15 BDB::INIT_LOCK | BDB::INIT_LOG | BDB::INIT_MPOOL
16 | BDB::INIT_TXN | BDB::RECOVER | BDB::USE_ENVIRON | BDB::CREATE,
17 0600;
18
19 $env->set_flags (BDB::AUTO_COMMIT | BDB::TXN_NOSYNC, 1);
20
21 my $db = db_create $env;
22 db_open $db, undef, "table", undef, BDB::BTREE, BDB::AUTO_COMMIT | BDB::CREATE
23 | BDB::READ_UNCOMMITTED, 0600;
24 db_put $db, undef, "key", "data", 0, sub {
25 db_del $db, undef, "key";
26 };
27 db_sync $db;
28
29 # automatic result processing with AnyEvent:
30 our $FH; open $FH, "<&=" . BDB::poll_fileno;
31 our $WATCHER = AnyEvent->io (fh => $FH, poll => 'r', cb => \&BDB::poll_cb);
32
33 # automatic result processing with EV:
34 my $WATCHER = EV::io BDB::poll_fileno, EV::READ, \&BDB::poll_cb;
35
36 # with Glib:
37 add_watch Glib::IO BDB::poll_fileno,
38 in => sub { BDB::poll_cb; 1 };
39
40 # or simply flush manually
41 BDB::flush;
42
43
9=head1 DESCRIPTION 44=head1 DESCRIPTION
10 45
11=head2 EXAMPLE 46See the BerkeleyDB documentation (L<http://www.oracle.com/technology/documentation/berkeley-db/db/index.html>).
47The BDB API is very similar to the C API (the translation has been very faithful).
48
49See also the example sections in the document below and possibly the eg/
50subdirectory of the BDB distribution. Last not least see the IO::AIO
51documentation, as that module uses almost the same asynchronous request
52model as this module.
53
54I know this is woefully inadequate documentation. Send a patch!
55
12 56
13=head1 REQUEST ANATOMY AND LIFETIME 57=head1 REQUEST ANATOMY AND LIFETIME
14 58
15Every request method creates a request. which is a C data structure not 59Every request method creates a request. which is a C data structure not
16directly visible to Perl. 60directly visible to Perl.
63use strict 'vars'; 107use strict 'vars';
64 108
65use base 'Exporter'; 109use base 'Exporter';
66 110
67BEGIN { 111BEGIN {
68 our $VERSION = '0.1'; 112 our $VERSION = '1.3';
69 113
70 our @BDB_REQ = qw( 114 our @BDB_REQ = qw(
71 db_env_create db_env_open db_env_close 115 db_env_open db_env_close db_env_txn_checkpoint db_env_lock_detect
116 db_env_memp_sync db_env_memp_trickle
72 db_create db_open db_close db_compact db_sync db_put db_get db_pget 117 db_open db_close db_compact db_sync db_put db_get db_pget db_del db_key_range
73 db_txn_commit db_txn_abort 118 db_txn_commit db_txn_abort db_txn_finish
119 db_c_close db_c_count db_c_put db_c_get db_c_pget db_c_del
120 db_sequence_open db_sequence_close
121 db_sequence_get db_sequence_remove
74 ); 122 );
75 our @EXPORT = (@BDB_REQ, qw(dbreq_pri dbreq_nice)); 123 our @EXPORT = (@BDB_REQ, qw(dbreq_pri dbreq_nice db_env_create db_create));
124 our @EXPORT_OK = qw(
76 our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush 125 poll_fileno poll_cb poll_wait flush
77 min_parallel max_parallel max_idle 126 min_parallel max_parallel max_idle
78 nreqs nready npending nthreads 127 nreqs nready npending nthreads
79 max_poll_time max_poll_reqs); 128 max_poll_time max_poll_reqs
129 );
80 130
81 require XSLoader; 131 require XSLoader;
82 XSLoader::load ("BDB", $VERSION); 132 XSLoader::load ("BDB", $VERSION);
83} 133}
134
135=head2 BERKELEYDB FUNCTIONS
136
137All of these are functions. The create functions simply return a new
138object and never block. All the remaining functions all take an optional
139callback as last argument that gets passed the return value. If it is
140missing, then the function will be executed synchronously, and the return
141value is returned as normally.
142
143BDB functions that cannot block (mostly functions that manipulate
144settings) are method calls on the relevant objects, so the rule of thumb
145is: if its a method, its not blocking, if its a function, it takes a
146callback as last argument.
147
148In the following, C<$int> signifies an integer return value,
149C<octetstring> is a "binary string" (i.e. a perl string with no character
150indices >255), C<U32> is an unsigned 32 bit integer, C<int> is some
151integer, C<NV> is a floating point value.
152
153The C<SV *> types are generic perl scalars (for input and output of data
154values), and the C<SV *callback> is the optional callback function to call
155when the request is completed.
156
157The various C<DB_ENV> etc. arguments are handles return by
158C<db_env_create>, C<db_create>, C<txn_begin> and so on. If they have an
159appended C<_ornull> this means they are optional and you can pass C<undef>
160for them, resulting a NULL pointer on the C level.
161
162=head3 BDB functions
163
164Functions in the BDB namespace, exported by default:
165
166 $env = db_env_create (U32 env_flags = 0)
167 flags: RPCCLIENT
168
169 db_env_open (DB_ENV *env, octetstring db_home, U32 open_flags, int mode, SV *callback = &PL_sv_undef)
170 open_flags: INIT_CDB INIT_LOCK INIT_LOG INIT_MPOOL INIT_REP INIT_TXN RECOVER RECOVER_FATAL USE_ENVIRON USE_ENVIRON_ROOT CREATE LOCKDOWN PRIVATE REGISTER SYSTEM_MEM
171 db_env_close (DB_ENV *env, U32 flags = 0, SV *callback = &PL_sv_undef)
172 db_env_txn_checkpoint (DB_ENV *env, U32 kbyte = 0, U32 min = 0, U32 flags = 0, SV *callback = &PL_sv_undef)
173 flags: FORCE
174 db_env_lock_detect (DB_ENV *env, U32 flags = 0, U32 atype = DB_LOCK_DEFAULT, SV *dummy = 0, SV *callback = &PL_sv_undef)
175 atype: LOCK_DEFAULT LOCK_EXPIRE LOCK_MAXLOCKS LOCK_MAXWRITE LOCK_MINLOCKS LOCK_MINWRITE LOCK_OLDEST LOCK_RANDOM LOCK_YOUNGEST
176 db_env_memp_sync (DB_ENV *env, SV *dummy = 0, SV *callback = &PL_sv_undef)
177 db_env_memp_trickle (DB_ENV *env, int percent, SV *dummy = 0, SV *callback = &PL_sv_undef)
178
179 $db = db_create (DB_ENV *env = 0, U32 flags = 0)
180 flags: XA_CREATE
181
182 db_open (DB *db, DB_TXN_ornull *txnid, octetstring file, octetstring database, int type, U32 flags, int mode, SV *callback = &PL_sv_undef)
183 flags: AUTO_COMMIT CREATE EXCL MULTIVERSION NOMMAP RDONLY READ_UNCOMMITTED THREAD TRUNCATE
184 db_close (DB *db, U32 flags = 0, SV *callback = &PL_sv_undef)
185 flags: DB_NOSYNC
186 db_compact (DB *db, DB_TXN_ornull *txn = 0, SV *start = 0, SV *stop = 0, SV *unused1 = 0, U32 flags = DB_FREE_SPACE, SV *unused2 = 0, SV *callback = &PL_sv_undef)
187 flags: FREELIST_ONLY FREE_SPACE
188 db_sync (DB *db, U32 flags = 0, SV *callback = &PL_sv_undef)
189 db_key_range (DB *db, DB_TXN_ornull *txn, SV *key, SV *key_range, U32 flags = 0, SV *callback = &PL_sv_undef)
190 db_put (DB *db, DB_TXN_ornull *txn, SV *key, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef)
191 flags: APPEND NODUPDATA NOOVERWRITE
192 db_get (DB *db, DB_TXN_ornull *txn, SV *key, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef)
193 flags: CONSUME CONSUME_WAIT GET_BOTH SET_RECNO MULTIPLE READ_COMMITTED READ_UNCOMMITTED RMW
194 db_pget (DB *db, DB_TXN_ornull *txn, SV *key, SV *pkey, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef)
195 flags: CONSUME CONSUME_WAIT GET_BOTH SET_RECNO MULTIPLE READ_COMMITTED READ_UNCOMMITTED RMW
196 db_del (DB *db, DB_TXN_ornull *txn, SV *key, U32 flags = 0, SV *callback = &PL_sv_undef)
197 db_txn_commit (DB_TXN *txn, U32 flags = 0, SV *callback = &PL_sv_undef)
198 flags: TXN_NOSYNC TXN_SYNC
199 db_txn_abort (DB_TXN *txn, SV *callback = &PL_sv_undef)
200
201 db_c_close (DBC *dbc, SV *callback = &PL_sv_undef)
202 db_c_count (DBC *dbc, SV *count, U32 flags = 0, SV *callback = &PL_sv_undef)
203 db_c_put (DBC *dbc, SV *key, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef)
204 flags: AFTER BEFORE CURRENT KEYFIRST KEYLAST NODUPDATA
205 db_c_get (DBC *dbc, SV *key, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef)
206 flags: CURRENT FIRST GET_BOTH GET_BOTH_RANGE GET_RECNO JOIN_ITEM LAST NEXT NEXT_DUP NEXT_NODUP PREV PREV_DUP PREV_NODUP SET SET_RANGE SET_RECNO READ_UNCOMMITTED MULTIPLE MULTIPLE_KEY RMW
207 db_c_pget (DBC *dbc, SV *key, SV *pkey, SV *data, U32 flags = 0, SV *callback = &PL_sv_undef)
208 db_c_del (DBC *dbc, U32 flags = 0, SV *callback = &PL_sv_undef)
209
210 db_sequence_open (DB_SEQUENCE *seq, DB_TXN_ornull *txnid, SV *key, U32 flags = 0, SV *callback = &PL_sv_undef)
211 flags: CREATE EXCL
212 db_sequence_close (DB_SEQUENCE *seq, U32 flags = 0, SV *callback = &PL_sv_undef)
213 db_sequence_get (DB_SEQUENCE *seq, DB_TXN_ornull *txnid, int delta, SV *seq_value, U32 flags = DB_TXN_NOSYNC, SV *callback = &PL_sv_undef)
214 flags: TXN_NOSYNC
215 db_sequence_remove (DB_SEQUENCE *seq, DB_TXN_ornull *txnid = 0, U32 flags = 0, SV *callback = &PL_sv_undef)
216 flags: TXN_NOSYNC
217
218=head4 db_txn_finish (DB_TXN *txn, U32 flags = 0, SV *callback = &PL_sv_undef)
219
220This is not actually a Berkeley DB function but a BDB module
221extension. The background for this exytension is: It is very annoying to
222have to check every single BDB function for error returns and provide a
223codepath out of your transaction. While the BDB module still makes this
224possible, it contains the following extensions:
225
226When a transaction-protected function returns any operating system
227error (errno > 0), BDB will set the C<TXN_DEADLOCK> flag on the
228transaction. This flag is also set by Berkeley DB functions themselves
229when an operation fails with LOCK_DEADLOCK, and it causes all further
230operations on that transaction (including C<db_txn_commit>) to fail.
231
232The C<db_txn_finish> request will look at this flag, and, if it is set,
233will automatically call C<db_txn_abort> (setting errno to C<LOCK_DEADLOCK>
234if it isn't set to something else yet). If it isn't set, it will call
235C<db_txn_commit> and return the error normally.
236
237How to use this? Easy: just write your transaction normally:
238
239 my $txn = $db_env->txn_begin;
240 db_get $db, $txn, "key", my $data;
241 db_put $db, $txn, "key", $data + 1 unless $! == BDB::NOTFOUND;
242 db_txn_finish $txn;
243 die "transaction failed" if $!;
244
245That is, handle only the expected errors. If something unexpected happens
246(EIO, LOCK_NOTGRANTED or a deadlock in either db_get or db_put), then the remaining
247requests (db_put in this case) will simply be skipped (they will fail with
248LOCK_DEADLOCK) and the transaction will be aborted.
249
250You can use the C<< $txn->failed >> method to check wether a transaction
251has failed in this way and abort further processing (excluding
252C<db_txn_finish>).
253
254=head3 DB_ENV/database environment methods
255
256Methods available on DB_ENV/$env handles:
257
258 DESTROY (DB_ENV_ornull *env)
259 CODE:
260 if (env)
261 env->close (env, 0);
262
263 $int = $env->set_data_dir (const char *dir)
264 $int = $env->set_tmp_dir (const char *dir)
265 $int = $env->set_lg_dir (const char *dir)
266 $int = $env->set_shm_key (long shm_key)
267 $int = $env->set_cachesize (U32 gbytes, U32 bytes, int ncache = 0)
268 $int = $env->set_flags (U32 flags, int onoff)
269 $env->set_errfile (FILE *errfile = 0)
270 $env->set_msgfile (FILE *msgfile = 0)
271 $int = $env->set_verbose (U32 which, int onoff = 1)
272 $int = $env->set_encrypt (const char *password, U32 flags = 0)
273 $int = $env->set_timeout (NV timeout_seconds, U32 flags = SET_TXN_TIMEOUT)
274 $int = $env->set_mp_max_openfd (int maxopenfd);
275 $int = $env->set_mp_max_write (int maxwrite, int maxwrite_sleep);
276 $int = $env->set_mp_mmapsize (int mmapsize_mb)
277 $int = $env->set_lk_detect (U32 detect = DB_LOCK_DEFAULT)
278 $int = $env->set_lk_max_lockers (U32 max)
279 $int = $env->set_lk_max_locks (U32 max)
280 $int = $env->set_lk_max_objects (U32 max)
281 $int = $env->set_lg_bsize (U32 max)
282 $int = $env->set_lg_max (U32 max)
283 $int = $env->mutex_set_increment (U32 increment)
284 $int = $env->mutex_set_tas_spins (U32 tas_spins)
285 $int = $env->mutex_set_max (U32 max)
286 $int = $env->mutex_set_align (U32 align)
287
288 $txn = $env->txn_begin (DB_TXN_ornull *parent = 0, U32 flags = 0)
289 flags: READ_COMMITTED READ_UNCOMMITTED TXN_NOSYNC TXN_NOWAIT TXN_SNAPSHOT TXN_SYNC TXN_WAIT TXN_WRITE_NOSYNC
290
291=head4 Example:
292
293 use AnyEvent;
294 use BDB;
295
296 our $FH; open $FH, "<&=" . BDB::poll_fileno;
297 our $WATCHER = AnyEvent->io (fh => $FH, poll => 'r', cb => \&BDB::poll_cb);
298
299 BDB::min_parallel 8;
300
301 my $env = db_env_create;
302
303 mkdir "bdtest", 0700;
304 db_env_open
305 $env,
306 "bdtest",
307 BDB::INIT_LOCK | BDB::INIT_LOG | BDB::INIT_MPOOL | BDB::INIT_TXN | BDB::RECOVER | BDB::USE_ENVIRON | BDB::CREATE,
308 0600;
309
310 $env->set_flags (BDB::AUTO_COMMIT | BDB::TXN_NOSYNC, 1);
311
312
313=head3 DB/database methods
314
315Methods available on DB/$db handles:
316
317 DESTROY (DB_ornull *db)
318 CODE:
319 if (db)
320 {
321 SV *env = (SV *)db->app_private;
322 db->close (db, 0);
323 SvREFCNT_dec (env);
324 }
325
326 $int = $db->set_cachesize (U32 gbytes, U32 bytes, int ncache = 0)
327 $int = $db->set_flags (U32 flags)
328 flags: CHKSUM ENCRYPT TXN_NOT_DURABLE
329 Btree: DUP DUPSORT RECNUM REVSPLITOFF
330 Hash: DUP DUPSORT
331 Queue: INORDER
332 Recno: RENUMBER SNAPSHOT
333
334 $int = $db->set_encrypt (const char *password, U32 flags)
335 $int = $db->set_lorder (int lorder)
336 $int = $db->set_bt_minkey (U32 minkey)
337 $int = $db->set_re_delim (int delim)
338 $int = $db->set_re_pad (int re_pad)
339 $int = $db->set_re_source (char *source)
340 $int = $db->set_re_len (U32 re_len)
341 $int = $db->set_h_ffactor (U32 h_ffactor)
342 $int = $db->set_h_nelem (U32 h_nelem)
343 $int = $db->set_q_extentsize (U32 extentsize)
344
345 $dbc = $db->cursor (DB_TXN_ornull *txn = 0, U32 flags = 0)
346 flags: READ_COMMITTED READ_UNCOMMITTED WRITECURSOR TXN_SNAPSHOT
347 $seq = $db->sequence (U32 flags = 0)
348
349=head4 Example:
350
351 my $db = db_create $env;
352 db_open $db, undef, "table", undef, BDB::BTREE, BDB::AUTO_COMMIT | BDB::CREATE | BDB::READ_UNCOMMITTED, 0600;
353
354 for (1..1000) {
355 db_put $db, undef, "key $_", "data $_";
356
357 db_key_range $db, undef, "key $_", my $keyrange;
358 my ($lt, $eq, $gt) = @$keyrange;
359 }
360
361 db_del $db, undef, "key $_" for 1..1000;
362
363 db_sync $db;
364
365
366=head3 DB_TXN/transaction methods
367
368Methods available on DB_TXN/$txn handles:
369
370 DESTROY (DB_TXN_ornull *txn)
371 CODE:
372 if (txn)
373 txn->abort (txn);
374
375 $int = $txn->set_timeout (NV timeout_seconds, U32 flags = SET_TXN_TIMEOUT)
376 flags: SET_LOCK_TIMEOUT SET_TXN_TIMEOUT
377
378 $bool = $txn->failed
379 # see db_txn_finish documentation, above
380
381
382=head3 DBC/cursor methods
383
384Methods available on DBC/$dbc handles:
385
386 DESTROY (DBC_ornull *dbc)
387 CODE:
388 if (dbc)
389 dbc->c_close (dbc);
390
391=head4 Example:
392
393 my $c = $db->cursor;
394
395 for (;;) {
396 db_c_get $c, my $key, my $data, BDB::NEXT;
397 warn "<$!,$key,$data>";
398 last if $!;
399 }
400
401 db_c_close $c;
402
403
404=head3 DB_SEQUENCE/sequence methods
405
406Methods available on DB_SEQUENCE/$seq handles:
407
408 DESTROY (DB_SEQUENCE_ornull *seq)
409 CODE:
410 if (seq)
411 seq->close (seq, 0);
412
413 $int = $seq->initial_value (db_seq_t value)
414 $int = $seq->set_cachesize (U32 size)
415 $int = $seq->set_flags (U32 flags)
416 flags: SEQ_DEC SEQ_INC SEQ_WRAP
417 $int = $seq->set_range (db_seq_t min, db_seq_t max)
418
419=head4 Example:
420
421 my $seq = $db->sequence;
422
423 db_sequence_open $seq, undef, "seq", BDB::CREATE;
424 db_sequence_get $seq, undef, 1, my $value;
425
84 426
85=head2 SUPPORT FUNCTIONS 427=head2 SUPPORT FUNCTIONS
86 428
87=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION 429=head3 EVENT PROCESSING AND EVENT LOOP INTEGRATION
88 430
133interactiveness when perl is not fast enough to process all requests in 475interactiveness when perl is not fast enough to process all requests in
134time. 476time.
135 477
136For interactive programs, values such as C<0.01> to C<0.1> should be fine. 478For interactive programs, values such as C<0.01> to C<0.1> should be fine.
137 479
138Example: Install an Event watcher that automatically calls 480Example: Install an EV watcher that automatically calls
139BDB::poll_cb with low priority, to ensure that other parts of the 481BDB::poll_cb with low priority, to ensure that other parts of the
140program get the CPU sometimes even under high AIO load. 482program get the CPU sometimes even under high load.
141 483
142 # try not to spend much more than 0.1s in poll_cb 484 # try not to spend much more than 0.1s in poll_cb
143 BDB::max_poll_time 0.1; 485 BDB::max_poll_time 0.1;
144 486
145 # use a low priority so other tasks have priority 487 my $bdb_poll = EV::io BDB::poll_fileno, EV::READ, \&BDB::poll_cb);
146 Event->io (fd => BDB::poll_fileno,
147 poll => 'r', nice => 1,
148 cb => &BDB::poll_cb);
149 488
150=item BDB::poll_wait 489=item BDB::poll_wait
151 490
152If there are any outstanding requests and none of them in the result 491If there are any outstanding requests and none of them in the result
153phase, wait till the result filehandle becomes ready for reading (simply 492phase, wait till the result filehandle becomes ready for reading (simply
165 504
166 BDB::poll_wait, BDB::poll_cb 505 BDB::poll_wait, BDB::poll_cb
167 506
168=item BDB::flush 507=item BDB::flush
169 508
170Wait till all outstanding AIO requests have been handled. 509Wait till all outstanding BDB requests have been handled.
171 510
172Strictly equivalent to: 511Strictly equivalent to:
173 512
174 BDB::poll_wait, BDB::poll_cb 513 BDB::poll_wait, BDB::poll_cb
175 while BDB::nreqs; 514 while BDB::nreqs;
176 515
516=back
517
177=head3 CONTROLLING THE NUMBER OF THREADS 518=head3 CONTROLLING THE NUMBER OF THREADS
178 519
520=over 4
521
179=item BDB::min_parallel $nthreads 522=item BDB::min_parallel $nthreads
180 523
181Set the minimum number of AIO threads to C<$nthreads>. The current 524Set the minimum number of BDB threads to C<$nthreads>. The current
182default is C<8>, which means eight asynchronous operations can execute 525default is C<8>, which means eight asynchronous operations can execute
183concurrently at any one time (the number of outstanding requests, 526concurrently at any one time (the number of outstanding requests,
184however, is unlimited). 527however, is unlimited).
185 528
186BDB starts threads only on demand, when an AIO request is queued and 529BDB starts threads only on demand, when an BDB request is queued and
187no free thread exists. Please note that queueing up a hundred requests can 530no free thread exists. Please note that queueing up a hundred requests can
188create demand for a hundred threads, even if it turns out that everything 531create demand for a hundred threads, even if it turns out that everything
189is in the cache and could have been processed faster by a single thread. 532is in the cache and could have been processed faster by a single thread.
190 533
191It is recommended to keep the number of threads relatively low, as some 534It is recommended to keep the number of threads relatively low, as some
196Under most circumstances you don't need to call this function, as the 539Under most circumstances you don't need to call this function, as the
197module selects a default that is suitable for low to moderate load. 540module selects a default that is suitable for low to moderate load.
198 541
199=item BDB::max_parallel $nthreads 542=item BDB::max_parallel $nthreads
200 543
201Sets the maximum number of AIO threads to C<$nthreads>. If more than the 544Sets the maximum number of BDB threads to C<$nthreads>. If more than the
202specified number of threads are currently running, this function kills 545specified number of threads are currently running, this function kills
203them. This function blocks until the limit is reached. 546them. This function blocks until the limit is reached.
204 547
205While C<$nthreads> are zero, aio requests get queued but not executed 548While C<$nthreads> are zero, aio requests get queued but not executed
206until the number of threads has been increased again. 549until the number of threads has been increased again.
309 652
310=head2 FORK BEHAVIOUR 653=head2 FORK BEHAVIOUR
311 654
312This module should do "the right thing" when the process using it forks: 655This module should do "the right thing" when the process using it forks:
313 656
314Before the fork, IO::AIO enters a quiescent state where no requests 657Before the fork, BDB enters a quiescent state where no requests
315can be added in other threads and no results will be processed. After 658can be added in other threads and no results will be processed. After
316the fork the parent simply leaves the quiescent state and continues 659the fork the parent simply leaves the quiescent state and continues
317request/result processing, while the child frees the request/result queue 660request/result processing, while the child frees the request/result queue
318(so that the requests started before the fork will only be handled in the 661(so that the requests started before the fork will only be handled in the
319parent). Threads will be started on demand until the limit set in the 662parent). Threads will be started on demand until the limit set in the
320parent process has been reached again. 663parent process has been reached again.
321 664
322In short: the parent will, after a short pause, continue as if fork had 665In short: the parent will, after a short pause, continue as if fork had
323not been called, while the child will act as if IO::AIO has not been used 666not been called, while the child will act as if BDB has not been used
324yet. 667yet.
325 668
326=head2 MEMORY USAGE 669=head2 MEMORY USAGE
327 670
328Per-request usage: 671Per-request usage:
331bytes of memory. In addition, stat requests need a stat buffer (possibly 674bytes of memory. In addition, stat requests need a stat buffer (possibly
332a few hundred bytes), readdir requires a result buffer and so on. Perl 675a few hundred bytes), readdir requires a result buffer and so on. Perl
333scalars and other data passed into aio requests will also be locked and 676scalars and other data passed into aio requests will also be locked and
334will consume memory till the request has entered the done state. 677will consume memory till the request has entered the done state.
335 678
336This is now awfully much, so queuing lots of requests is not usually a 679This is not awfully much, so queuing lots of requests is not usually a
337problem. 680problem.
338 681
339Per-thread usage: 682Per-thread usage:
340 683
341In the execution phase, some aio requests require more memory for 684In the execution phase, some aio requests require more memory for
342temporary buffers, and each thread requires a stack and other data 685temporary buffers, and each thread requires a stack and other data
343structures (usually around 16k-128k, depending on the OS). 686structures (usually around 16k-128k, depending on the OS).
344 687
345=head1 KNOWN BUGS 688=head1 KNOWN BUGS
346 689
347Known bugs will be fixed in the next release. 690Known bugs will be fixed in the next release, except:
691
692 If you use a transaction in any request, and the request returns
693 with an operating system error or DB_LOCK_NOTGRANTED, the internal
694 TXN_DEADLOCK flag will be set on the transaction. See C<db_txn_finish>,
695 above.
348 696
349=head1 SEE ALSO 697=head1 SEE ALSO
350 698
351L<Coro::AIO>. 699L<Coro::BDB>, L<IO::AIO>.
352 700
353=head1 AUTHOR 701=head1 AUTHOR
354 702
355 Marc Lehmann <schmorp@schmorp.de> 703 Marc Lehmann <schmorp@schmorp.de>
356 http://home.schmorp.de/ 704 http://home.schmorp.de/

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines