1Stdlib.Gc(3) OCaml library Stdlib.Gc(3)
2
3
4
6 Stdlib.Gc - no description
7
9 Module Stdlib.Gc
10
12 Module Gc
13 : (module Stdlib__Gc)
14
15
16
17
18
19
20
21 type stat = {
22 minor_words : float ; (* Number of words allocated in the minor heap
23 since the program was started.
24 *)
25 promoted_words : float ; (* Number of words allocated in the minor
26 heap that survived a minor collection and were moved to the major heap
27 since the program was started.
28 *)
29 major_words : float ; (* Number of words allocated in the major heap,
30 including the promoted words, since the program was started.
31 *)
32 minor_collections : int ; (* Number of minor collections since the
33 program was started.
34 *)
35 major_collections : int ; (* Number of major collection cycles com‐
36 pleted since the program was started.
37 *)
38 heap_words : int ; (* Total size of the major heap, in words.
39 *)
40 heap_chunks : int ; (* Number of contiguous pieces of memory that
41 make up the major heap.
42 *)
43 live_words : int ; (* Number of words of live data in the major heap,
44 including the header words.
45
46 Note that "live" words refers to every word in the major heap that
47 isn't currently known to be collectable, which includes words that have
48 become unreachable by the program after the start of the previous gc
49 cycle. It is typically much simpler and more predictable to call
50 Gc.full_major (or Gc.compact ) then computing gc stats, as then "live"
51 words has the simple meaning of "reachable by the program". One caveat
52 is that a single call to Gc.full_major will not reclaim values that
53 have a finaliser from Gc.finalise (this does not apply to Gc.fi‐
54 nalise_last ). If this caveat matters, simply call Gc.full_major twice
55 instead of once.
56 *)
57 live_blocks : int ; (* Number of live blocks in the major heap.
58
59 See live_words for a caveat about what "live" means.
60 *)
61 free_words : int ; (* Number of words in the free list.
62 *)
63 free_blocks : int ; (* Number of blocks in the free list.
64 *)
65 largest_free : int ; (* Size (in words) of the largest block in the
66 free list.
67 *)
68 fragments : int ; (* Number of wasted words due to fragmentation.
69 These are 1-words free blocks placed between two live blocks. They are
70 not available for allocation.
71 *)
72 compactions : int ; (* Number of heap compactions since the program
73 was started.
74 *)
75 top_heap_words : int ; (* Maximum size reached by the major heap, in
76 words.
77 *)
78 stack_size : int ; (* Current size of the stack, in words.
79
80
81 Since 3.12.0
82 *)
83 forced_major_collections : int ; (* Number of forced full major col‐
84 lections completed since the program was started.
85
86
87 Since 4.12.0
88 *)
89 }
90
91
92 The memory management counters are returned in a stat record.
93
94 The total amount of memory allocated by the program since it was
95 started is (in words) minor_words + major_words - promoted_words .
96 Multiply by the word size (4 on a 32-bit machine, 8 on a 64-bit ma‐
97 chine) to get the number of bytes.
98
99
100 type control = {
101
102 mutable minor_heap_size : int ; (* The size (in words) of the minor
103 heap. Changing this parameter will trigger a minor collection. De‐
104 fault: 256k.
105 *)
106
107 mutable major_heap_increment : int ; (* How much to add to the major
108 heap when increasing it. If this number is less than or equal to 1000,
109 it is a percentage of the current heap size (i.e. setting it to 100
110 will double the heap size at each increase). If it is more than 1000,
111 it is a fixed number of words that will be added to the heap. Default:
112 15.
113 *)
114
115 mutable space_overhead : int ; (* The major GC speed is computed from
116 this parameter. This is the memory that will be "wasted" because the
117 GC does not immediately collect unreachable blocks. It is expressed as
118 a percentage of the memory used for live data. The GC will work more
119 (use more CPU time and collect blocks more eagerly) if space_overhead
120 is smaller. Default: 120.
121 *)
122
123 mutable verbose : int ; (* This value controls the GC messages on
124 standard error output. It is a sum of some of the following flags, to
125 print messages on the corresponding events:
126
127 - 0x001 Start and end of major GC cycle.
128
129 - 0x002 Minor collection and major GC slice.
130
131 - 0x004 Growing and shrinking of the heap.
132
133 - 0x008 Resizing of stacks and memory manager tables.
134
135 - 0x010 Heap compaction.
136
137 - 0x020 Change of GC parameters.
138
139 - 0x040 Computation of major GC slice size.
140
141 - 0x080 Calling of finalisation functions.
142
143 - 0x100 Bytecode executable and shared library search at start-up.
144
145 - 0x200 Computation of compaction-triggering condition.
146
147 - 0x400 Output GC statistics at program exit. Default: 0.
148
149 *)
150
151 mutable max_overhead : int ; (* Heap compaction is triggered when the
152 estimated amount of "wasted" memory is more than max_overhead percent
153 of the amount of live data. If max_overhead is set to 0, heap com‐
154 paction is triggered at the end of each major GC cycle (this setting is
155 intended for testing purposes only). If max_overhead >= 1000000 , com‐
156 paction is never triggered. If compaction is permanently disabled, it
157 is strongly suggested to set allocation_policy to 2. Default: 500.
158 *)
159
160 mutable stack_limit : int ; (* The maximum size of the stack (in
161 words). This is only relevant to the byte-code runtime, as the native
162 code runtime uses the operating system's stack. Default: 1024k.
163 *)
164
165 mutable allocation_policy : int ; (* The policy used for allocating in
166 the major heap. Possible values are 0, 1 and 2.
167
168
169 -0 is the next-fit policy, which is usually fast but can result in
170 fragmentation, increasing memory consumption.
171
172
173 -1 is the first-fit policy, which avoids fragmentation but has corner
174 cases (in certain realistic workloads) where it is sensibly slower.
175
176
177 -2 is the best-fit policy, which is fast and avoids fragmentation. In
178 our experiments it is faster and uses less memory than both next-fit
179 and first-fit. (since OCaml 4.10)
180
181 The default is best-fit.
182
183 On one example that was known to be bad for next-fit and first-fit,
184 next-fit takes 28s using 855Mio of memory, first-fit takes 47s using
185 566Mio of memory, best-fit takes 27s using 545Mio of memory.
186
187 Note: If you change to next-fit, you may need to reduce the space_over‐
188 head setting, for example using 80 instead of the default 120 which is
189 tuned for best-fit. Otherwise, your program will need more memory.
190
191 Note: changing the allocation policy at run-time forces a heap com‐
192 paction, which is a lengthy operation unless the heap is small (e.g. at
193 the start of the program).
194
195 Default: 2.
196
197
198 Since 3.11.0
199 *)
200 window_size : int ; (* The size of the window used by the major GC
201 for smoothing out variations in its workload. This is an integer be‐
202 tween 1 and 50. Default: 1.
203
204
205 Since 4.03.0
206 *)
207 custom_major_ratio : int ; (* Target ratio of floating garbage to ma‐
208 jor heap size for out-of-heap memory held by custom values located in
209 the major heap. The GC speed is adjusted to try to use this much memory
210 for dead values that are not yet collected. Expressed as a percentage
211 of major heap size. The default value keeps the out-of-heap floating
212 garbage about the same size as the in-heap overhead. Note: this only
213 applies to values allocated with caml_alloc_custom_mem (e.g. bigar‐
214 rays). Default: 44.
215
216
217 Since 4.08.0
218 *)
219 custom_minor_ratio : int ; (* Bound on floating garbage for
220 out-of-heap memory held by custom values in the minor heap. A minor GC
221 is triggered when this much memory is held by custom values located in
222 the minor heap. Expressed as a percentage of minor heap size. Note:
223 this only applies to values allocated with caml_alloc_custom_mem (e.g.
224 bigarrays). Default: 100.
225
226
227 Since 4.08.0
228 *)
229 custom_minor_max_size : int ; (* Maximum amount of out-of-heap memory
230 for each custom value allocated in the minor heap. When a custom value
231 is allocated on the minor heap and holds more than this many bytes,
232 only this value is counted against custom_minor_ratio and the rest is
233 directly counted against custom_major_ratio . Note: this only applies
234 to values allocated with caml_alloc_custom_mem (e.g. bigarrays). De‐
235 fault: 8192 bytes.
236
237
238 Since 4.08.0
239 *)
240 }
241
242
243 The GC parameters are given as a control record. Note that these pa‐
244 rameters can also be initialised by setting the OCAMLRUNPARAM environ‐
245 ment variable. See the documentation of ocamlrun .
246
247
248
249 val stat : unit -> stat
250
251 Return the current values of the memory management counters in a stat
252 record. This function examines every heap block to get the statistics.
253
254
255
256 val quick_stat : unit -> stat
257
258 Same as stat except that live_words , live_blocks , free_words ,
259 free_blocks , largest_free , and fragments are set to 0. This function
260 is much faster than stat because it does not need to go through the
261 heap.
262
263
264
265 val counters : unit -> float * float * float
266
267 Return (minor_words, promoted_words, major_words) . This function is
268 as fast as quick_stat .
269
270
271
272 val minor_words : unit -> float
273
274 Number of words allocated in the minor heap since the program was
275 started. This number is accurate in byte-code programs, but only an ap‐
276 proximation in programs compiled to native code.
277
278 In native code this function does not allocate.
279
280
281 Since 4.04
282
283
284
285 val get : unit -> control
286
287 Return the current values of the GC parameters in a control record.
288
289
290
291 val set : control -> unit
292
293
294 set r changes the GC parameters according to the control record r .
295 The normal usage is: Gc.set { (Gc.get()) with Gc.verbose = 0x00d }
296
297
298
299
300 val minor : unit -> unit
301
302 Trigger a minor collection.
303
304
305
306 val major_slice : int -> int
307
308
309 major_slice n Do a minor collection and a slice of major collection. n
310 is the size of the slice: the GC will do enough work to free (on aver‐
311 age) n words of memory. If n = 0, the GC will try to do enough work to
312 ensure that the next automatic slice has no work to do. This function
313 returns an unspecified integer (currently: 0).
314
315
316
317 val major : unit -> unit
318
319 Do a minor collection and finish the current major collection cycle.
320
321
322
323 val full_major : unit -> unit
324
325 Do a minor collection, finish the current major collection cycle, and
326 perform a complete new cycle. This will collect all currently unreach‐
327 able blocks.
328
329
330
331 val compact : unit -> unit
332
333 Perform a full major collection and compact the heap. Note that heap
334 compaction is a lengthy operation.
335
336
337
338 val print_stat : out_channel -> unit
339
340 Print the current values of the memory management counters (in hu‐
341 man-readable form) into the channel argument.
342
343
344
345 val allocated_bytes : unit -> float
346
347 Return the total number of bytes allocated since the program was
348 started. It is returned as a float to avoid overflow problems with int
349 on 32-bit machines.
350
351
352
353 val get_minor_free : unit -> int
354
355 Return the current size of the free space inside the minor heap.
356
357
358 Since 4.03.0
359
360
361
362 val get_bucket : int -> int
363
364
365 get_bucket n returns the current size of the n -th future bucket of the
366 GC smoothing system. The unit is one millionth of a full GC.
367
368
369 Since 4.03.0
370
371
372 Raises Invalid_argument if n is negative, return 0 if n is larger than
373 the smoothing window.
374
375
376
377 val get_credit : unit -> int
378
379
380 get_credit () returns the current size of the "work done in advance"
381 counter of the GC smoothing system. The unit is one millionth of a full
382 GC.
383
384
385 Since 4.03.0
386
387
388
389 val huge_fallback_count : unit -> int
390
391 Return the number of times we tried to map huge pages and had to fall
392 back to small pages. This is always 0 if OCAMLRUNPARAM contains H=1 .
393
394
395 Since 4.03.0
396
397
398
399 val finalise : ('a -> unit) -> 'a -> unit
400
401
402 finalise f v registers f as a finalisation function for v . v must be
403 heap-allocated. f will be called with v as argument at some point be‐
404 tween the first time v becomes unreachable (including through weak
405 pointers) and the time v is collected by the GC. Several functions can
406 be registered for the same value, or even several instances of the same
407 function. Each instance will be called once (or never, if the program
408 terminates before v becomes unreachable).
409
410 The GC will call the finalisation functions in the order of dealloca‐
411 tion. When several values become unreachable at the same time (i.e.
412 during the same GC cycle), the finalisation functions will be called in
413 the reverse order of the corresponding calls to finalise . If finalise
414 is called in the same order as the values are allocated, that means
415 each value is finalised before the values it depends upon. Of course,
416 this becomes false if additional dependencies are introduced by assign‐
417 ments.
418
419 In the presence of multiple OCaml threads it should be assumed that any
420 particular finaliser may be executed in any of the threads.
421
422 Anything reachable from the closure of finalisation functions is con‐
423 sidered reachable, so the following code will not work as expected:
424
425 - let v = ... in Gc.finalise (fun _ -> ...v...) v
426
427 Instead you should make sure that v is not in the closure of the final‐
428 isation function by writing:
429
430 - let f = fun x -> ... let v = ... in Gc.finalise f v
431
432 The f function can use all features of OCaml, including assignments
433 that make the value reachable again. It can also loop forever (in this
434 case, the other finalisation functions will not be called during the
435 execution of f, unless it calls finalise_release ). It can call fi‐
436 nalise on v or other values to register other functions or even itself.
437 It can raise an exception; in this case the exception will interrupt
438 whatever the program was doing when the function was called.
439
440
441 finalise will raise Invalid_argument if v is not guaranteed to be
442 heap-allocated. Some examples of values that are not heap-allocated
443 are integers, constant constructors, booleans, the empty array, the
444 empty list, the unit value. The exact list of what is heap-allocated
445 or not is implementation-dependent. Some constant values can be
446 heap-allocated but never deallocated during the lifetime of the pro‐
447 gram, for example a list of integer constants; this is also implementa‐
448 tion-dependent. Note that values of types float are sometimes allo‐
449 cated and sometimes not, so finalising them is unsafe, and finalise
450 will also raise Invalid_argument for them. Values of type 'a Lazy.t
451 (for any 'a ) are like float in this respect, except that the compiler
452 sometimes optimizes them in a way that prevents finalise from detecting
453 them. In this case, it will not raise Invalid_argument , but you should
454 still avoid calling finalise on lazy values.
455
456 The results of calling String.make , Bytes.make , Bytes.create , Ar‐
457 ray.make , and ref are guaranteed to be heap-allocated and non-constant
458 except when the length argument is 0 .
459
460
461
462 val finalise_last : (unit -> unit) -> 'a -> unit
463
464 same as Gc.finalise except the value is not given as argument. So you
465 can't use the given value for the computation of the finalisation func‐
466 tion. The benefit is that the function is called after the value is un‐
467 reachable for the last time instead of the first time. So contrary to
468 Gc.finalise the value will never be reachable again or used again. In
469 particular every weak pointer and ephemeron that contained this value
470 as key or data is unset before running the finalisation function. More‐
471 over the finalisation functions attached with Gc.finalise are always
472 called before the finalisation functions attached with Gc.finalise_last
473 .
474
475
476 Since 4.04
477
478
479
480 val finalise_release : unit -> unit
481
482 A finalisation function may call finalise_release to tell the GC that
483 it can launch the next finalisation function without waiting for the
484 current one to return.
485
486
487 type alarm
488
489
490 An alarm is a piece of data that calls a user function at the end of
491 each major GC cycle. The following functions are provided to create
492 and delete alarms.
493
494
495
496 val create_alarm : (unit -> unit) -> alarm
497
498
499 create_alarm f will arrange for f to be called at the end of each major
500 GC cycle, starting with the current cycle or the next one. A value of
501 type alarm is returned that you can use to call delete_alarm .
502
503
504
505 val delete_alarm : alarm -> unit
506
507
508 delete_alarm a will stop the calls to the function associated to a .
509 Calling delete_alarm a again has no effect.
510
511
512
513 val eventlog_pause : unit -> unit
514
515
516 eventlog_pause () will pause the collection of traces in the runtime.
517 Traces are collected if the program is linked to the instrumented run‐
518 time and started with the environment variable OCAML_EVENTLOG_ENABLED.
519 Events are flushed to disk after pausing, and no new events will be
520 recorded until eventlog_resume is called.
521
522
523 Since 4.11
524
525
526
527 val eventlog_resume : unit -> unit
528
529
530 eventlog_resume () will resume the collection of traces in the runtime.
531 Traces are collected if the program is linked to the instrumented run‐
532 time and started with the environment variable OCAML_EVENTLOG_ENABLED.
533 This call can be used after calling eventlog_pause , or if the program
534 was started with OCAML_EVENTLOG_ENABLED=p. (which pauses the collection
535 of traces before the first event.)
536
537
538 Since 4.11
539
540
541 module Memprof : sig end
542
543
544
545 Memprof is a sampling engine for allocated memory words. Every allo‐
546 cated word has a probability of being sampled equal to a configurable
547 sampling rate. Once a block is sampled, it becomes tracked. A tracked
548 block triggers a user-defined callback as soon as it is allocated, pro‐
549 moted or deallocated.
550
551 Since blocks are composed of several words, a block can potentially be
552 sampled several times. If a block is sampled several times, then each
553 of the callback is called once for each event of this block: the multi‐
554 plicity is given in the n_samples field of the allocation structure.
555
556 This engine makes it possible to implement a low-overhead memory pro‐
557 filer as an OCaml library.
558
559 Note: this API is EXPERIMENTAL. It may change without prior notice.
560
561
562
563
564
565OCamldoc 2022-07-22 Stdlib.Gc(3)