1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
|
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<!--
Documentation for Lua Lanes
-->
<html>
<head>
<meta name="description" content="Lua Lanes - multithreading in Lua" />
<meta name="keywords" content="Lua, Library, Multithreading, Threads, Rocks" />
<title>Lua Lanes - multithreading in Lua</title>
</head>
<body>
<div class="header">
<hr />
<center>
<table summary="Lua logo">
<tbody>
<tr>
<td align="center">
<a href="http://www.lua.org">
<img src="multi.png" alt="Lua" align="middle" border="0" height="120" width="128" />
<img src="multi.png" alt="Lua" align="middle" border="0" height="120" width="128" />
<img src="multi.png" alt="Lua" align="middle" border="0" height="120" width="128" />
<img src="multi.png" alt="Lua" align="middle" border="0" height="120" width="128" />
<img src="multi.png" alt="Lua" align="middle" border="0" height="120" width="128" />
</a></td>
</tr>
<tr>
<td align="center" valign="top"><h1>Lua Lanes - multithreading in Lua</h1>
</td>
</tr>
</tbody>
</table>
<p class="bar">
<a href="#description">Description</a> ·
<a href="#systems">Supported systems</a> ·
<a href="#installing">Building and Installing</a>
</p><p class="bar">
<a href="#creation">Creation</a> ·
<a href="#status">Status</a> ·
<a href="#results">Results and errors</a>
</p><p class="bar">
<a href="#cancelling">Cancelling</a> ·
<a href="#finalizers">Finalizers</a> ·
<a href="#lindas">Lindas</a> ·
<a href="#timers">Timers</a> ·
<a href="#locks">Locks etc.</a>
</p><p class="bar">
<a href="#other">Other issues</a> ·
<a href="#changes">Change log</a>
<!-- ... -->
<p><br/><font size="-1"><i>Copyright © 2007-12 Asko Kauppi, Benoit Germain. All rights reserved.</i>
<br>Lua Lanes is published under the same <A HREF="http://en.wikipedia.org/wiki/MIT_License">MIT license</A> as Lua 5.1 and 5.2.
</p><p>This document was revised on 13-Aug-12, and applies to version 3.1.6
</font></p>
</center>
</div>
<!-- description +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="description">Description</h2>
<p>Lua Lanes is a Lua extension library providing
the possibility to run multiple Lua states in parallel. It is intended to
be used for optimizing performance on multicore CPU's and to study ways to make Lua programs naturally parallel to begin with.
</p><p>
Lanes is included into your software by the regular
<tt>require "lanes"</tt> method. No C side programming is needed; all APIs are Lua side, and most existing extension modules should
work seamlessly together with the multiple lanes.
</p><p>
Starting with version 3.1.6, Lanes should build and run identically with either Lua 5.1 or Lua 5.2.
</p><p>
See <A HREF="comparison.html">comparison</A> of Lua Lanes with other Lua multithreading solutions.
</p><p>
<h3>Features:</h3>
<ul>
<li>Lanes have separated data, by default. Shared data is possible with Linda objects.
</li>
<li>Communications is separate of threads, using Linda objects.
</li>
<li>Data passing uses fast inter-state copies (no serialization required)</li>
</li>
<li>"Deep userdata" concept, for sharing userdata over multiple lanes
</li>
<li>Millisecond level timers, integrated with the Linda system.
</li>
<li>Threads can be given priorities -2..+2 (default is 0).
</li>
<li>Lanes are cancellable, with proper cleanup.
</li>
<li>No application level locking - ever!
</li>
</ul>
<h3>Limitations:</h3>
<ul><li>coroutines are not passed between states
</li>
<li>sharing full userdata between states needs special C side
preparations (-> <A HREF="#deep_userdata">deep userdata</A>)
</li>
<li>network level parallelism not included
</li>
</ul>
</p>
<!-- systems +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="systems">Supported systems</h2>
<p>Lua Lanes supports the following operating systems:
<ul>
<li>Mac OS X PowerPC / Intel (10.4 and later)</li>
<li>Linux x86</li>
<li>Windows 2000/XP and later <font size="-1">(MinGW or Visual C++ 2005/2008)</font></li>
<!--
Other OS'es here once people help test them. (and the tester's name)
Win64, BSD, Linux x64, Linux embedded, QNX, Solaris, ...
-->
</ul>
<p>The underlying threading code can be compiled either towards Win32 API
or <a TARGET="_blank" HREF="http://en.wikipedia.org/wiki/POSIX_Threads">Pthreads</a>. Unfortunately, thread prioritization under Pthreads is a JOKE,
requiring OS specific tweaks and guessing undocumented behaviour. Other features should be portable to any modern platform.
</p>
</p>
<!-- installing +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="installing">Building and Installing</h2>
<p>Lua Lanes is built simply by <tt>make</tt> on the supported platforms
(<tt>make-vc</tt> for Visual C++). See <tt>README</tt> for system specific
details and limitations.
</p>
<p>To install Lanes, all you need are the <tt>lanes.lua</tt> <tt>lanes-keeper.lua</tt> and <tt>lanes/core.so|dll</tt>
files to be reachable by Lua (see LUA_PATH, LUA_CPATH).
Or use <A HREF="http://www.luarocks.org" TARGET="_blank">Lua Rocks</A> package management.
</p>
<pre>
> luarocks search lanes
... output listing Lua Lanes is there ...
> luarocks install lanes
... output ...
</pre>
<!-- launching +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="initialization">Initialization</h2>
<p>
The following sample shows how to initialize the Lanes module.
</p>
<table border="1" bgcolor="#FFFFE0" width="500">
<tr>
<td>
<pre>
local lanes = require "lanes".configure()
</pre>
</table>
<p>
Starting with version 3.0-beta, requiring the module follows Lua 5.2 rules:
the module is not available under the global name "lanes", but has to be accessed
through require's return value.
After lanes is required, it is necessary to call <tt>lanes.configure()</tt>, which is the
only function exposed by the module at this point. Calling <tt>configure()</tt> will
perform one-time initializations and make the rest of the API available.
At the same time, <tt>configure()</tt> itself will be replaced by another function that
raises an error if called again with differing arguments.
</p>
<p>
<table border="1" bgcolor="#E0E0FF" cellpadding="10">
<tr>
<td>
<code>
lanes.configure( [opt_tbl])
</code>
</table>
</p>
<p>
<tt>lanes.configure</tt> accepts an optional options table as sole argument.
<table>
<tr valign=top><td width=40></td><td>
<code>.nb_keepers</code> <br/><nobr>N</nobr></td><td width=40></td>
<td>
Controls the number of keeper states used internally by lindas to transfer data between lanes. (see below). Default is 1.
</td></tr>
<tr valign=top><td/><td>
<code>.with_timers</code> <br/>nil/false/anything</td><td/>
<td>
If equal to <tt>false</tt>, Lanes doesn't start the timer service,
and the associated API will be absent from the interface (see below).
Any other value (including <tt>nil</tt>), starts the timer service.
</td></tr>
<tr valign="top">
<td/>
<td>
<code>.on_create_state</code> <br/>C function/nil
</td><td/>
<td>
If provided, will be called in every created Lua state (keepers and lanes) right after it is created, and *before* any library is loaded.
That way, all C functions it loads in the state can be added to the function lookup database.
</td>
</tr>
</table>
<h2 id="creation">Creation</h2>
<p>The following sample shows preparing a function for parallel calling, and
calling it with varying arguments. Each of the two results is calculated in
a separate OS thread, parallel to the calling one. Reading the results
joins the threads, waiting for any results not already there.
</p>
<table border=1 bgcolor="#FFFFE0" width=500><tr><td>
<pre>
local lanes = require "lanes"
lanes.configure()
f= lanes.gen( function(n) return 2*n end )
a= f(1)
b= f(2)
print( a[1], b[1] ) -- 2 4
</pre>
</table>
<p>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td>
<code>func= lanes.gen( [libs_str | opt_tbl [, ...],] lane_func )
<br/><br/>
lane_h= func( ... )</code>
</table>
</p>
<p>
The function returned by <tt>lanes.gen()</tt> is a "generator" for
launching any number of lanes. They will share code, options, initial globals,
but the particular arguments may vary. Only calling the generator function
actually launches a lane, and provides a handle for controlling it.
Alternatively, <tt>lane_func</tt> may be a string, in which case it will be compiled
in the lane. This was to be able to launch lanes with older versions of LuaJIT,
which didn't not support lua_dump, used internally to transfer functions to the lane.
<!--
</p>
<p>This prepares <tt>lane_func</tt> to be called in parallel. It does not yet start
anything, but merely returns a <i>generator function</i> that can be called
any number of times, with varying parameters. Each call will spawn a new lane.
-->
</p><p>
Lanes automatically copies upvalues over to the new lanes, so you
need not wrap all the required elements into one 'wrapper' function. If
<tt>lane_func</tt> uses some local values, or local functions, they will be there
also in the new lanes.
</p><p>
<code>libs_str</code> defines the standard libraries made available to the
new Lua state:
<table>
<tr><td width=40></td><td>(nothing)</td><td width=40></td><td>no standard libraries (default)</td></tr>
<tr><td/>
<td><tt>"base"</tt> or <tt>""</tt></td><td/>
<td>root level names, <tt>print</tt>, <tt>assert</tt>, <tt>unpack</tt> etc.</td>
</tr>
<tr><td/><td><tt>"coroutine"</tt></td><td/><td><tt>coroutine.*</tt> namespace <font size="-1">(part of base in Lua 5.1 and 5.2)</font></td></tr>
<tr><td/><td><tt>"debug"</tt></td><td/><td><tt>debug.*</tt> namespace</td></tr>
<tr><td/><td><tt>"io"</tt></td><td/><td><tt>io.*</tt> namespace</td></tr>
<tr><td/><td><tt>"math"</tt></td><td/><td><tt>math.*</tt> namespace</td></tr>
<tr><td/><td><tt>"os"</tt></td><td/><td><tt>os.*</tt> namespace</td></tr>
<tr><td/><td><tt>"package"</tt></td><td/><td><tt>package.*</tt> namespace and <tt>require</tt></td></tr>
<tr><td/><td><tt>"string"</tt></td><td/><td><tt>string.*</tt> namespace</td></tr>
<tr><td/><td><tt>"table"</tt></td><td/><td><tt>table.*</tt> namespace</td></tr>
<br/>
<tr><td/><td><tt>"*"</tt></td><td/><td>all standard libraries</td></tr>
</table>
</p><p>
Initializing the standard libs takes a bit of time at each lane invocation.
This is the main reason why "no libraries" is the default.
</p><p>
<code>opt_tbl</code> is a collection of named options to control the way
lanes are run:
</p><p>
<table>
<tr valign=top><td width=40></td><td>
<code>.cancelstep</code> <br/><nobr>N / true</nobr></td>
<td width=40></td><td>
By default, lanes are only cancellable when they <u>enter</u> a pending
<tt>:receive()</tt> or <tt>:send()</tt> call.
With this option, one can set cancellation check to occur every <tt>N</tt>
Lua statements. The value <tt>true</tt> uses a default value (100).
It is also possible to manually test for cancel requests with <tt>cancel_test()</tt>.
</td></tr>
<tr valign=top><td/><td>
<code>.globals</code> <br/>globals_tbl</td><td/>
<td>
Sets the globals table for the launched threads. This can be used for giving
them constants.
<br>
The global values of different lanes are in no manner connected;
modifying one will only affect the particular lane.
</td></tr>
<tr valign="top">
<td/>
<td>
<code>.required</code> <br/>modules_tbl
</td><td/>
<td>
Lists modules that have to be required in order to be able to trasnfer
functions they exposed. Starting with Lanes 3.0-beta, non-Lua functions are
no longer copied by recreating a C closure from a C pointer, but are searched
in lookup tables. These tables are built from the modules listed here. <tt>required</tt>
must be a list of strings, each one being the name of a module to be required.
</td>
</tr>
<tr valign=top><td width=40><td>
<code>.priority</code> <br/><nobr>-2..+2</nobr></td><td/>
<td>The priority of lanes generated. -2 is lowest, +2 is highest.
<br>
Implementation and dependability of priorities varies
by platform. Especially Linux kernel 2.6 is not supporting priorities in user mode.
</td></tr>
<tr valign="top">
<td width="40">
<td>
<code>.package</code><br/>
</td>
<td><td/>
<code>package</code> contents overrides, if needed.
Specifying it when <code>libs_str</code> doesn't cause the <code>package</code> library to be loaded will generate an error.
If not specified, the created lane will receive the current values of <tt>package</tt>. Only path, cpath, preload and loaders are transfered.
<br>
</td>
</tr>
</table>
<p>Each lane also gets a function <tt>set_debug_threadname()</tt> that it can use anytime to do as the name says.
Supported debuggers are Microsoft Visual Studio (for the C side) and Decoda (for the Lua side).
</p>
<p>
If a lane body pulls a C function imported by a module required before Lanes itself (thus not through a hooked <tt>require</tt>), the lane generator creation will raise an error.
The function name it shows is a path where it was found by scanning _G. As a utility, the name guessing functionality is exposed as such:
<table border="1" bgcolor="#E0E0FF" cellpadding="10">
<tr>
<td>
<code>
"type", "name" = lanes.nameof( o)
</code>
</table>
</p>
<h3>Free running lanes</h3>
<p>
The lane handles are allowed to be 'let loose'; in other words you may execute
a lane simply by:
<pre>
lanes.gen( function() ... end ) ()
</pre>
Normally, this kind of lanes will be in an eternal loop handling messages.
Since the lane handle is gone,
there is no way to control such a lane from the outside, nor read its potential
return values. Then again, such a lane does not even normally return.
</p>
<!-- status +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="status">Status</h2>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td>
<code>str= lane_h.status</code>
</table>
<p>The current execution state of a lane can be read via its <tt>status</tt>
member, providing one of these values: <sup>(<a href="#2">2</a></sup>
<table>
<tr><td width=40><td><tt>"pending"</tt></td><td>not started, yet</td></tr>
<tr><td/><td><tt>"running"</tt></td><td>running</td></tr>
<tr><td/><td><tt>"waiting"</tt></td><td>waiting at a Linda <tt>:receive()</tt> or <tt>:send()</tt></td></tr>
<tr><td/><td><tt>"done"</tt></td><td>finished executing (results are ready)</td></tr>
<tr><td/><td><tt>"error"</tt></td><td>met an error (reading results will propagate it)</td></tr>
<tr><td/><td><tt>"cancelled"</tt></td><td>received cancellation and finished itself</td></tr>
</table>
</p><p>
This is similar to <tt>coroutine.status</tt>, which has: <tt>"running"</tt> /
<tt>"suspended"</tt> / <tt>"normal"</tt> / <tt>"dead"</tt>. Not using the
exact same names is intentional.
</p>
<!-- results +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="results">Results and errors</h2>
<p>A lane can be waited upon by simply reading its results. This can be done
in two ways.
</p><p>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td>
<code>[val]= lane_h[1]</code>
</table>
<p>
Makes sure lane has finished, and gives its first (maybe only) return value.
Other return values will be available in other <tt>lane_h</tt> indices.
</p><p>
If the lane ended in an error, it is propagated to master state at this place.
</p>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td>
<code>[...]|[nil,err,stack_tbl]= lane_h:join( [timeout_secs] )</code>
</table>
<p>
Waits until the lane finishes, or <tt>timeout</tt> seconds have passed.
Returns <tt>nil</tt> on timeout, <tt>nil,err,stack_tbl</tt> if the lane hit an error,
or the return values of the lane. Unlike in reading the results in table
fashion, errors are not propagated.
</p><p>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td>
<code>set_error_reporting("basic"|"extended")</code>
</table>
</p><p>
Sets the error reporting mode. "basic" is selected by default.
</p><p>
<tt>stack_tbl</tt> is a table describing where the error was thrown.<br/>
In extended mode, <tt>stack_tbl</tt> is an array of tables containing info gathered with <tt>lua_getinfo()</tt> (<tt>"source"</tt>,<tt>"currentline"</tt>,<tt>"name"</tt>,<tt>"namewhat"</tt>,<tt>"what"</tt>).<br/>
In "basic mode", <tt>stack_tbl</tt> is an array of "<filename>:<line>" strings. Use <tt>table.concat()</tt> to format it to your liking (or just ignore it).
</p><p>
If you use <tt>:join</tt>, make sure your lane main function returns
a non-nil value so you can tell timeout and error cases apart from succesful
return (using the <tt>.status</tt> property may be risky, since it might change
between a timed out join and the moment you read it).
</p><p>
<table border=1 bgcolor="#FFFFE0" width=500><tr><td>
<pre>
require "lanes"
f= lanes.gen( function() error "!!!" end )
a= f(1)
--print( a[1] ) -- propagates error
v,err= a:join() -- no propagation
if v==nil then
error( "'a' faced error"..tostring(err) ) -- manual propagation
end
</pre>
</table>
<p>
If you want to wait for multiple lanes to finish (any of a set of lanes), use
a <a href="#lindas">Linda</a> object. Give each lane a specific id, and send
that id over a Linda once that thread is done (as the last thing you do).
</p>
<!-- cancelling +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="cancelling">Cancelling</h2>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td>
<code>bool= lane_h:cancel( [timeout_secs=0.0,] [force_kill_bool=false] )
<br/>
<br/>
<code>
bool= cancel_test()
</code>
</table>
<p><tt>cancel()</tt>sends a cancellation request to the lane. If <tt>timeout_secs</tt> is non-zero, waits
for the request to be processed, or a timeout to occur.
Returns <tt>true</tt> if the lane was already done (in <tt>"done"</tt>, <tt>"error"</tt> or <tt>"cancelled"</tt> status)
or if the cancellation was fruitful within timeout period.
</p><p>
If the lane is still running and <tt>force_kill</tt> is <tt>true</tt>, the
OS thread running the lane is forcefully killed. This means no GC, and should
generally be the last resort.
</p>
<p>Cancellation is tested <u>before</u> going to sleep in <tt>receive()</tt> or <tt>send()</tt> calls
and after executing <tt>cancelstep</tt> Lua statements. Starting with version 3.0-beta, a pending <tt>receive()</tt>
or <tt>send()</tt> call is awakened. This means the execution of the lane will resume although the operation has
not completed, to give the lane a chance to detect cancellation. The code should be able to handle this situation appropriately if required.
It is also possible to manually test for cancel requests with <tt>cancel_test()</tt>.
</p>
<!-- finalizers +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="finalizers">Finalizers</h2>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td>
<code>set_finalizer( finalizer_func )</code>
<br/><br/>
<code>void= finalizer_func( [err, stack_tbl] )</code>
</table>
<p>The <tt>error</tt> call is used for throwing exceptions in Lua. What Lua
does not offer, however, is scoped <a href="http://en.wikipedia.org/wiki/Finalizer">finalizers</a>
that would get called when a certain block of instructions gets exited, whether
through peaceful return or abrupt <tt>error</tt>.
</p>
<p>Since 2.0.3, Lanes prepares a function <tt>set_finalizer</tt> for doing this.
Any functions given to it will be called in the lane Lua state, just prior to
closing it. They are not called in any particular order.
</p>
<p>An error in a finalizer itself overrides the state of the regular chunk
(in practise, it would be highly preferable <i>not</i> to have errors in finalizers).
If one finalizer errors, the others may not get called.
</p>
<!-- lindas +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="lindas">Lindas</h2>
<p>Communications between lanes is completely detached from the lane handles
themselves. By itself, a lane can only provide return values once it's finished,
or throw an error. Needs to communicate during runtime are handled by <A HREF="http://en.wikipedia.org/wiki/Linda_%28coordination_language%29" TARGET="_blank">Linda objects</A>, which are
<A HREF="#deep_userdata">deep userdata</A> instances. They can be provided to a lane
as startup parameters, upvalues or in some other Linda's message.
</p><p>
Access to a Linda object means a lane can read or write to any of its data
slots. Multiple lanes can be accessing the same Linda in parallel. No application
level locking is required; each Linda operation is atomic.
</p><p>
<table border=1 bgcolor="#FFFFE0" width=500><tr><td>
<pre>
require "lanes"
local linda= lanes.linda()
local function loop( max )
for i=1,max do
print( "sending: "..i )
linda:send( "x", i ) -- linda as upvalue
end
end
a= lanes.gen("",loop)( 10000 )
while true do
local val= linda:receive( 3.0, "x" ) -- timeout in seconds
if val==nil then
print( "timed out" )
break
end
print( "received: "..val )
end
</pre>
</table>
</p>
<p>Characteristics of the Lanes implementation of Lindas are:
<ul>
<li>keys can be of boolean, number, string or light userdata type
</li>
<li>values can be any type supported by inter-state copying (same limits
as for function parameters and upvalues)
</li>
<li>consuming method is <tt>:receive</tt> (not in)
</li>
<li>non-consuming method is <tt>:get</tt> (not rd)
</li>
<li>two producer-side methods: <tt>:send</tt> and <tt>:set</tt> (not out)
</li>
<li><tt>send</tt> allows for sending multiple values -atomically- to a
given key
</li>
<li><tt>receive</tt> can wait for multiple keys at once
</li>
<li>
<tt>receive</tt> has a batched mode to consume more than one value from a single key, as in <tt>linda:receive( 1.0, linda.batched, "key", 3, 6)</tt>
</li>
<li>individual keys' queue length can be limited, balancing speed differences
in a producer/consumer scenario (making <tt>:send</tt> wait)
</li>
</ul>
</p>
<p>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td>
<code>h= lanes.linda( [opt_name])</code>
<br/><br/>
<code>bool= h:send( [timeout_secs,] key, ... )</code>
<br/>
<code>[val, key]= h:receive( [timeout_secs,] key [, ...] )</code>
<br/><br/>
<code>= h:limit( key, n_uint )</code>
</table>
<p>The <tt>send</tt> and <tt>receive</tt> methods use Linda keys as FIFO stacks
(first in, first out). Timeouts are given in seconds (millisecond accuracy).
If using numbers as the first Linda key, one must explicitly give <tt>nil</tt>
as the timeout parameter to avoid ambiguities.
</p><p>
By default, stack sizes are unlimited but limits can be
enforced using the <tt>limit</tt> method. This can be useful to balance execution
speeds in a producer/consumer scenario.
</p><p>
Note that any number of lanes can be reading or writing a Linda. There can be
many producers, and many consumers. It's up to you.
</p>
<p><tt>send</tt> returns <tt>true</tt> if the sending succeeded, and <tt>false</tt>
if the queue limit was met, and the queue did not empty enough during the given
timeout.
</p><p>
Equally, <tt>receive</tt> returns a value and the key that provided the value,
or nothing for timeout. Note that <tt>nil</tt>s can be sent and received;
the <tt>key</tt> value will tell it apart from a timeout.
</p><p>
Multiple values can be sent to a given key at once, atomically (the send will
fail unless all the values fit within the queue limit). This can be useful for
multiple producer scenarios, if the protocols used are giving data in streams
of multiple units. Atomicity avoids the producers from garbling each others
messages, which could happen if the units were sent individually.
</p><p>
When receiving from multiple slots, the keys are checked in order, which can
be used for making priority queues.
</p><p>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td>
<code>linda_h:set( key, [val] )</code>
<br/>
<code>[val]= linda_h:get( key )</code>
</table>
</p><p>
The table access methods are for accessing a slot without queuing or consuming.
They can be used for making shared tables of storage among the lanes.
</p><p>
Writing to a slot overwrites existing value, and clears any possible queued
entries. Table access and <tt>send</tt>/<tt>receive</tt> can be used together;
reading a slot essentially peeks the next outcoming value of a queue.
</p>
<p>
<table border="1" bgcolor="#E0E0FF" cellpadding="10">
<tr>
<td>
<code>[val]= linda_h:count( [key[,...]])</code>
</table>
</p>
<p>
Returns some information about the contents of the linda.
<br/>If no key is specified, and the linda is empty, returns nothing.
<br/>If no key is specified, and the linda is not empty, returns a table of key/count pairs that counts the number of items in each
of the exiting keys of the linda. This count can be 0 if the key has been used but is empty.
<br/>If a single key is specified, returns the number of pending items, or nothing if the key is unknown.
<br/>If more than one key is specified, return a table of key/count pairs for the known keys.
</p>
<!--
<p>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td>
<code>lightuserdata= linda_h:deep()</code>
</table>
<p>There is one more method that is not required in applications, but
discussing it is good for a preview of how deep userdata works.
</p><p>
Because proxy objects (<tt>linda_h</tt>) are just pointers to the real, deep
userdata, they cannot be used to identify a certain Linda from the others.
The internal timer system needs to do this, and the <tt>:deep()</tt> method
has been added for its use. It returns a light userdata pointing to the
<i>actual</i> deep object, and thus can be used for seeing, which proxies actually
mean the same underlying object. You might or might not need a similar system
with your own deep userdata.
</p>
-->
<h3>Granularity of using Lindas</h3>
<p>A single Linda object provides an infinite number of slots, so why would
you want to use several?
</p><p>There are some important reasons:
<ul>
<li>Access control. If you don't trust certain code completely, or just
to modularize your design, use one Linda for one usage and another one
for the other. This keeps your code clear and readable. You can pass
multiple Linda handles to a lane with practically no added cost.
</li>
<li>Namespace control. Linda keys have a "flat" namespace, so collisions
are possible if you try to use the same Linda for too many separate uses.
</li>
<li>Performance. Changing any slot in a Linda causes all pending threads
for that Linda to be momentarily awakened (at least in the C level).
This can degrade performance due to unnecessary OS level context switches.
The more keeper states you declared with <tt>lanes.configure()</tt> the less
this should be a problem.
</li>
</ul>
On the other side, you need to use a common Linda for waiting for multiple
keys. You cannot wait for keys from two separate Linda objects at the same
time.
</p><p>
<font size="-1">Actually, you can. Make separate lanes to wait each, and then multiplex those
events to a common Linda, but... :).</font>
</p>
<!-- timers +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="timers">Timers</h2>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td>
<code>= lanes.timer( linda_h, key, date_tbl|first_secs [,period_secs] )</code>
</table>
<p>
Timers are implemented as a lane. They can be disabled by setting <tt>"with_timers"</tt> to <tt>nil</tt> or <tt>false</tt>
to <tt>lanes.configure()</tt>.
</p>
<p>
Timers can be run once, or in a reoccurring fashion (<tt>period_secs > 0</tt>).
The first occurrence can be given either as a date or as a relative delay in seconds.
The <tt>date</tt> table is like what <tt>os.date("*t")</tt> returns, in the
local time zone.
</p><p>
Once a timer expires, the <tt>key</tt> is set with the current time
(in seconds, same offset as <tt>os.time()</tt> but with millisecond accuracy).
The key can be waited upon using the regular Linda <tt>:receive()</tt>
method.
</p><p>
A timer can be stopped simply with <tt>first_secs=0|nil</tt> and no period.
</p><p>
<table border=1 bgcolor="#FFFFE0" width=500><tr><td>
<pre>
local lanes = require "lanes"
lanes.configure( 1)
local linda= lanes.linda()
-- First timer once a second, not synchronized to wall clock
--
lanes.timer( linda, "sec", 1, 1 )
-- Timer to a future event (next even minute); wall clock synchronized
--
local t= os.date( "*t", os.time()+60 ) -- now + 1min
t.sec= 0
lanes.timer( linda, "min", t, 60 ) -- reoccur every minute (sharp)
while true do
local v,key= linda:receive( "sec", "min" )
print( "Timer "..key..": "..v )
end
</pre>
</table>
</p><p>
NOTE: Timer keys are set, not queued, so missing a beat is possible especially
if the timer cycle is extremely small. The key value can be used to know the
actual time passed.
</p><p>
<table>
<tr><td valign=top><nobr><i>Design note:</i></nobr> </td>
<td>
<font size="-1">
Having the API as <tt>lanes.timer()</tt> is intentional. Another
alternative would be <tt>linda_h:timer()</tt> but timers are not traditionally
seen to be part of Lindas. Also, it would mean any lane getting a Linda handle
would be able to modify timers on it. A third choice could
be abstracting the timers out of Linda realm altogether (<tt>timer_h= lanes.timer( date|first_secs, period_secs )</tt>)
but that would mean separate waiting functions for timers, and lindas. Even if
a linda object and key was returned, that key couldn't be waited upon simultaneously
with one's general linda events.
The current system gives maximum capabilities with minimum API, and any smoothenings
can easily be crafted in Lua at the application level.
</font>
</td>
</tr>
</table>
</p>
<!-- locks +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="locks">Locks etc.</h2>
<p>
Lanes does not generally require locks or critical sections to be used, at all.
If necessary, a limited queue can be used to emulate them. <tt>lanes.lua</tt>
offers some sugar to make it easy:
</p><p>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td><pre>
lock_func= lanes.genlock( linda_h, key [,N_uint=1] )
lock_func( M_uint ) -- acquire
..
lock_func( -M_uint ) -- release
</table>
</p><p>
The generated function acquires M entries from the N available, or releases
them if the value is negative. The acquiring call will suspend the lane, if necessary.
Use <tt>M=N=1</tt> for a critical section lock (only one lane allowed to enter).
</p><p>
Note: The locks generated are <u>not recursive</u>. That would need another
kind of generator, which is currently not implemented.
</p><p>
Similar sugar exists for atomic counters:
</p><p>
<table border=1 bgcolor="#E0E0FF" cellpadding=10><tr><td><pre>
atomic_func= lanes.genatomic( linda_h, key [,initial_num=0.0] )
new_num= atomic_func( [diff_num=+1.0] )
</table>
</p><p>
Each time called, the generated function will change <tt>linda[key]</tt>
atomically, without other lanes being able to interfere. The new value is
returned. You can use either <tt>diff 0.0</tt> or <tt>get</tt> to just read the current
value.
</p><p>
Note that the generated functions can be passed on to other lanes.
</p>
<!-- others +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="other">Other issues</h2>
<h3>Limitations on data passing</h3>
<p>Data passed between lanes (either as starting parameters, return values, upvalues or via Lindas) must conform to the following:
</p>
<p><ul>
<li>Booleans, numbers, strings, light userdata, Lua functions and tables of such can always be passed.
</li>
<li>Cyclic tables and/or duplicate references are allowed and reproduced appropriately,
but only <u>within the same transmission</u>.
<ul>
<li>using the same source table in multiple Linda messages keeps no ties between the tables
</li>
</ul>
</li>
<li>Objects (tables with a metatable) are copyable between lanes.
<ul>
<li>metatables are assumed to be immutable; they are internally indexed and only copied once
per each type of objects per lane
</li>
</ul>
</li>
<li>C functions (<tt>lua_CFunction</tt>) referring to <tt>LUA_ENVIRONINDEX</tt> or <tt>LUA_REGISTRYINDEX</tt> might not
work right in the target
<ul>
<li>rather completely re-initialize a module with <tt>require</tt> in the target lane
</li>
</ul>
</li>
<li>Full userdata can be passed only if it's prepared using the <A HREF="#deep_userdata">deep userdata</A>
system, which handles its lifespan management
<ul>
<li>in particular, lane handles cannot be passed between lanes
</li>
</ul>
</li>
<li>coroutines cannot be passed
</li>
</ul>
</p>
<h3>Required of module makers</h3>
<p>
Most Lua extension modules should work unaltered with Lanes.
If the module simply ties C side features to Lua, everything is fine without
alterations. The <tt>luaopen_...()</tt> entry point will be called separately for each
lane, where the module is <tt>require</tt>'d from.
</p><p>
If it, however, also does one-time C side initializations, these
should be covered into a one-time-only construct such as below.
</p><p>
<table><tr><td width=40>
<td bgcolor="#ffffe0">
<pre>
int luaopen_module( lua_State *L )
{
static char been_here; /* 0 by ANSI C */
/* Calls to 'require' serialized by Lanes; this is safe.
*/
if (!been_here) {
been_here= 1;
... one time initializations ...
}
... binding to Lua ...
}
</pre>
</td></tr></table>
</p>
<h3 id="shared_userdata">Deep userdata in your own apps</h3>
<p>
The mechanism Lanes uses for sharing Linda handles between separate Lua states
can be used for custom userdata as well. Here's what to do.
</p>
<ol>
<li>Provide an <i>identity function</i> for your userdata, in C. This function is
used for creation and deletion of your deep userdata (the shared resource),
and for making metatables for the state-specific proxies for accessing it. The
prototype is
<table border="1" bgcolor="#E0E0FF" cellpadding="10">
<tr>
<td>
<code>void idfunc( lua_State *L, char const * const which);</code>
</table>
<tt>which</tt> can be one of:
<ul>
<li>
<tt>"new"</tt>: requests the creation of a new object, whose pointer is pushed on the stack
as a light userdata.
</li>
<li>
<tt>"delete"</tt>: receives this same pointer on the stack, and should cleanup the object.</li>
<li>
<tt>"metatable"</tt>: should build a metatable for the object. Don't cache the metatable
yourself, Lanes takes care of it ("metatable" should only be invoked once).</li>
<li>
<tt>"module"</tt>: is the name of the module that exports the idfunc,
to be pushed on the stack as a string. It is necessary so that Lanes can require it in
any Lane and keeper state that receives a userdata. This is to prevent crashes in situations
where the module could be unloaded while the idfunc pointer is still held.</li>
</ul>
Take a look at <tt>linda_id</tt> in <tt>lanes.c</tt>.
</li>
<li>Instanciate your userdata using <tt>luaG_deep_userdata()</tt>,
instead of the regular <tt>lua_newuserdata()</tt>.
Given an <tt>idfunc</tt>, it sets up the support
structures and returns a state-specific proxy userdata for accessing your
data. This proxy can also be copied over to other lanes.
</li>
<li>Accessing the deep userdata from your C code, use <tt>luaG_todeep()</tt>
instead of the regular <tt>lua_touserdata()</tt>.
</li>
</ol>
<p>Deep userdata management will take care of tying to <tt>__gc</tt> methods,
and doing reference counting to see how many proxies are still there for
accessing the data. Once there are none, the data will be freed through a call
to the <tt>idfunc</tt> you provided.
</p>
<p><b>NOTE</b>: The lifespan of deep userdata may exceed that of the Lua state
that created it. The allocation of the data storage should not be tied to
the Lua state used. In other words, use <tt>malloc</tt>/<tt>free</tt> or
similar memory handling mechanism.
</p>
<h3>Lane handles don't travel</h3>
<p>
Lane handles are not implemented as deep userdata, and cannot thus be
copied across lanes. This is intentional; problems would occur at least when
multiple lanes were to wait upon one to get ready. Also, it is a matter of
design simplicity.
</p><p>
The same benefits can be achieved by having a single worker lane spawn all
the sublanes, and keep track of them. Communications to and from this lane
can be handled via a Linda.
</p>
<h3>Beware with print and file output</h3>
<p>
In multithreaded scenarios, giving multiple parameters to <tt>print()</tt>
or <tt>file:write()</tt> may cause them to be overlapped in the output,
something like this:
<pre>
A: print( 1, 2, 3, 4 )
B: print( 'a', 'b', 'c', 'd' )
1 a b 2 3 c d 4
</pre>
Lanes does not protect you from this behaviour. The thing to do is either to
concentrate your output to a certain lane per stream, or to concatenate output
into a single string before you call the output function.
</p>
<h3 id="performance">Performance considerations</h3>
<p>
Lanes is about making multithreading easy, and natural in the Lua state of mind.
Expect performance not to be an issue, if your program is logically built.
Here are some things one should consider, if best performance is vital:
</p><p>
<ul>
<li>Data passing (parameters, upvalues, Linda messages) is generally fast,
doing two binary state-to-state copies (from source state to hidden state,
hidden state to target state). Remember that not only the function you
specify but also its upvalues, their upvalues, etc. etc. will get copied.
</li>
<li>Lane startup is fast (1000's of lanes a second), depending on the
number of standard libraries initialized. Initializing all standard libraries
is about 3-4 times slower than having no standard libraries at all. If you
throw in a lot of lanes per second, make sure you give them minimal necessary
set of libraries.
</li>
<li>Waiting Lindas are woken up (and execute some hidden Lua code) each
time <u>any</u> key in the Lindas they are waiting for are changed. This
may give essential slow-down (not measured, just a gut feeling) if a lot
of Linda keys are used. Using separate Linda objects for logically separate
issues will help (which is good practise anyhow).
</li>
<li>Linda objects are light. The memory footprint is two OS-level signalling
objects (<tt>HANDLE</tt> or <tt>pthread_cond_t</tt>) for each, plus one
C pointer for the proxies per each Lua state using the Linda. Barely nothing.
</li>
<li>Timers are light. You can probably expect timers up to 0.01 second
resolution to be useful, but that is very system specific. All timers are
merged into one main timer state (see <tt>timer.lua</tt>); no OS side
timers are utilized.
</li>
<li>Lindas are hashed to a fixed number of "keeper states", which are a locking entity.
If you are using a lot of Linda objects,
it may be useful to try having more of these keeper states. By default,
only one is used (see <tt>KEEPER_STATES_N</tt>), but this is an implementation detail.
</li>
</ul>
</p>
<h3 id="cancelling_cancel">Cancelling cancel</h3>
<p>
Cancellation of lanes uses the Lua error mechanism with a special lightuserdata
error sentinel.
If you use <tt>pcall</tt> in code that needs to be cancellable
from the outside, the special error might not get through to Lanes, thus
preventing the Lane from being cleanly cancelled. You should throw any
lightuserdata error further.
</p><p>
This system can actually be used by application to detect cancel, do your own
cancellation duties, and pass on the error so Lanes will get it. If it does
not get a clean cancellation from a lane in due time,
it may forcefully kill the lane.
</p><p>
The sentinel is exposed as <tt>lanes.cancel_error</tt>, if you wish to use
its actual value.
</p>
<!-- change log +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<h2 id="changes">Change log</h2>
<p>
See CHANGES.
</p>
<!-- footnotes +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ -->
<hr/>
<p>For feedback, questions and suggestions:
<UL>
<li><A HREF="http://github.com/LuaLanes/lanes">Lanes @ GitHub</A></li>
<li><A HREF="mailto:bnt.germain@gmail.com">the maintainer</A></li>
<li><A HREF="http://www.lua.org/lua-l.html">the lua mailing list</A></li>
</UL>
</p>
<p><br/></p>
</body>
</html>
|
