1Gc(3) OCaml library Gc(3)
2
6 Gc - Memory management control and statistics; finalised values.
7
9 Module Gc
10
12 Module Gc
13 : sig end
14 Memory management control and statistics; finalised values.
17 type stat = {
23 minor_words : float ; (* Number of words allocated in the minor heap
24 since the program was started.
25 *)
26 promoted_words : float ; (* Number of words allocated in the minor
27 heap that survived a minor collection and were moved to the major heap
28 since the program was started.
29 *)
30 major_words : float ; (* Number of words allocated in the major heap,
31 including the promoted words, since the program was started.
32 *)
33 minor_collections : int ; (* Number of minor collections since the
34 program was started.
35 *)
36 major_collections : int ; (* Number of major collection cycles com‐
37 pleted since the program was started.
38 *)
39 heap_words : int ; (* Total size of the major heap, in words.
40 *)
41 heap_chunks : int ; (* Number of contiguous pieces of memory that
42 make up the major heap.
43 *)
44 live_words : int ; (* Number of words of live data in the major heap,
45 including the header words.
46 *)
47 live_blocks : int ; (* Number of live blocks in the major heap.
48 *)
49 free_words : int ; (* Number of words in the free list.
50 *)
51 free_blocks : int ; (* Number of blocks in the free list.
52 *)
53 largest_free : int ; (* Size (in words) of the largest block in the
54 free list.
55 *)
56 fragments : int ; (* Number of wasted words due to fragmentation.
57 These are 1-words free blocks placed between two live blocks. They are
58 not available for allocation.
59 *)
60 compactions : int ; (* Number of heap compactions since the program
61 was started.
62 *)
63 top_heap_words : int ; (* Maximum size reached by the major heap, in
64 words.
65 *)
66 stack_size : int ; (* Current size of the stack, in words.
67 Since 3.12.0
70 *)
71 forced_major_collections : int ; (* Number of forced full major col‐
72 lections completed since the program was started.
73 Since 4.12.0
76 *)
77 }
78 The memory management counters are returned in a stat record.
81 The total amount of memory allocated by the program since it was
83 started is (in words) minor_words + major_words - promoted_words .
84 Multiply by the word size (4 on a 32-bit machine, 8 on a 64-bit ma‐
85 chine) to get the number of bytes.
86 type control = {
89 mutable minor_heap_size : int ; (* The size (in words) of the minor
91 heap. Changing this parameter will trigger a minor collection. De‐
92 fault: 256k.
93 *)
94 mutable major_heap_increment : int ; (* How much to add to the major
96 heap when increasing it. If this number is less than or equal to 1000,
97 it is a percentage of the current heap size (i.e. setting it to 100
98 will double the heap size at each increase). If it is more than 1000,
99 it is a fixed number of words that will be added to the heap. Default:
100 15.
101 *)
102 mutable space_overhead : int ; (* The major GC speed is computed from
104 this parameter. This is the memory that will be "wasted" because the
105 GC does not immediately collect unreachable blocks. It is expressed as
106 a percentage of the memory used for live data. The GC will work more
107 (use more CPU time and collect blocks more eagerly) if space_overhead
108 is smaller. Default: 80.
109 *)
110 mutable verbose : int ; (* This value controls the GC messages on
112 standard error output. It is a sum of some of the following flags, to
113 print messages on the corresponding events:
114 - 0x001 Start and end of major GC cycle.
116 - 0x002 Minor collection and major GC slice.
118 - 0x004 Growing and shrinking of the heap.
120 - 0x008 Resizing of stacks and memory manager tables.
122 - 0x010 Heap compaction.
124 - 0x020 Change of GC parameters.
126 - 0x040 Computation of major GC slice size.
128 - 0x080 Calling of finalisation functions.
130 - 0x100 Bytecode executable and shared library search at start-up.
132 - 0x200 Computation of compaction-triggering condition.
134 - 0x400 Output GC statistics at program exit. Default: 0.
136 *)
138 mutable max_overhead : int ; (* Heap compaction is triggered when the
140 estimated amount of "wasted" memory is more than max_overhead percent
141 of the amount of live data. If max_overhead is set to 0, heap com‐
142 paction is triggered at the end of each major GC cycle (this setting is
143 intended for testing purposes only). If max_overhead >= 1000000 , com‐
144 paction is never triggered. If compaction is permanently disabled, it
145 is strongly suggested to set allocation_policy to 2. Default: 500.
146 *)
147 mutable stack_limit : int ; (* The maximum size of the stack (in
149 words). This is only relevant to the byte-code runtime, as the native
150 code runtime uses the operating system's stack. Default: 1024k.
151 *)
152 mutable allocation_policy : int ; (* The policy used for allocating in
154 the major heap. Possible values are 0, 1 and 2.
155 -0 is the next-fit policy, which is usually fast but can result in
158 fragmentation, increasing memory consumption.
159 -1 is the first-fit policy, which avoids fragmentation but has corner
162 cases (in certain realistic workloads) where it is sensibly slower.
163 -2 is the best-fit policy, which is fast and avoids fragmentation. In
166 our experiments it is faster and uses less memory than both next-fit
167 and first-fit. (since OCaml 4.10)
168 The current default is next-fit, as the best-fit policy is new and not
170 yet widely tested. We expect best-fit to become the default in the fu‐
171 ture.
172 On one example that was known to be bad for next-fit and first-fit,
174 next-fit takes 28s using 855Mio of memory, first-fit takes 47s using
175 566Mio of memory, best-fit takes 27s using 545Mio of memory.
176 Note: When changing to a low-fragmentation policy, you may need to aug‐
178 ment the space_overhead setting, for example using 100 instead of the
179 default 80 which is tuned for next-fit. Indeed, the difference in frag‐
180 mentation behavior means that different policies will have different
181 proportion of "wasted space" for a given program. Less fragmentation
182 means a smaller heap so, for the same amount of wasted space, a higher
183 proportion of wasted space. This makes the GC work harder, unless you
184 relax it by increasing space_overhead .
185 Note: changing the allocation policy at run-time forces a heap com‐
187 paction, which is a lengthy operation unless the heap is small (e.g. at
188 the start of the program).
189 Default: 0.
191 Since 3.11.0
194 *)
195 window_size : int ; (* The size of the window used by the major GC
196 for smoothing out variations in its workload. This is an integer be‐
197 tween 1 and 50. Default: 1.
198 Since 4.03.0
201 *)
202 custom_major_ratio : int ; (* Target ratio of floating garbage to ma‐
203 jor heap size for out-of-heap memory held by custom values located in
204 the major heap. The GC speed is adjusted to try to use this much memory
205 for dead values that are not yet collected. Expressed as a percentage
206 of major heap size. The default value keeps the out-of-heap floating
207 garbage about the same size as the in-heap overhead. Note: this only
208 applies to values allocated with caml_alloc_custom_mem (e.g. bigar‐
209 rays). Default: 44.
210 Since 4.08.0
213 *)
214 custom_minor_ratio : int ; (* Bound on floating garbage for
215 out-of-heap memory held by custom values in the minor heap. A minor GC
216 is triggered when this much memory is held by custom values located in
217 the minor heap. Expressed as a percentage of minor heap size. Note:
218 this only applies to values allocated with caml_alloc_custom_mem (e.g.
219 bigarrays). Default: 100.
220 Since 4.08.0
223 *)
224 custom_minor_max_size : int ; (* Maximum amount of out-of-heap memory
225 for each custom value allocated in the minor heap. When a custom value
226 is allocated on the minor heap and holds more than this many bytes,
227 only this value is counted against custom_minor_ratio and the rest is
228 directly counted against custom_major_ratio . Note: this only applies
229 to values allocated with caml_alloc_custom_mem (e.g. bigarrays). De‐
230 fault: 8192 bytes.
231 Since 4.08.0
234 *)
235 }
236 The GC parameters are given as a control record. Note that these pa‐
239 rameters can also be initialised by setting the OCAMLRUNPARAM environ‐
240 ment variable. See the documentation of ocamlrun .
241 val stat : unit -> stat
245 Return the current values of the memory management counters in a stat
247 record. This function examines every heap block to get the statistics.
248 val quick_stat : unit -> stat
252 Same as stat except that live_words , live_blocks , free_words ,
254 free_blocks , largest_free , and fragments are set to 0. This function
255 is much faster than stat because it does not need to go through the
256 heap.
257 val counters : unit -> float * float * float
261 Return (minor_words, promoted_words, major_words) . This function is
263 as fast as quick_stat .
264 val minor_words : unit -> float
268 Number of words allocated in the minor heap since the program was
270 started. This number is accurate in byte-code programs, but only an ap‐
271 proximation in programs compiled to native code.
272 In native code this function does not allocate.
274 Since 4.04
277 val get : unit -> control
281 Return the current values of the GC parameters in a control record.
283 val set : control -> unit
287 set r changes the GC parameters according to the control record r .
290 The normal usage is: Gc.set { (Gc.get()) with Gc.verbose = 0x00d }
291 val minor : unit -> unit
296 Trigger a minor collection.
298 val major_slice : int -> int
302 major_slice n Do a minor collection and a slice of major collection. n
305 is the size of the slice: the GC will do enough work to free (on aver‐
306 age) n words of memory. If n = 0, the GC will try to do enough work to
307 ensure that the next automatic slice has no work to do. This function
308 returns an unspecified integer (currently: 0).
309 val major : unit -> unit
313 Do a minor collection and finish the current major collection cycle.
315 val full_major : unit -> unit
319 Do a minor collection, finish the current major collection cycle, and
321 perform a complete new cycle. This will collect all currently unreach‐
322 able blocks.
323 val compact : unit -> unit
327 Perform a full major collection and compact the heap. Note that heap
329 compaction is a lengthy operation.
330 val print_stat : out_channel -> unit
334 Print the current values of the memory management counters (in hu‐
336 man-readable form) into the channel argument.
337 val allocated_bytes : unit -> float
341 Return the total number of bytes allocated since the program was
343 started. It is returned as a float to avoid overflow problems with int
344 on 32-bit machines.
345 val get_minor_free : unit -> int
349 Return the current size of the free space inside the minor heap.
351 Since 4.03.0
354 val get_bucket : int -> int
358 get_bucket n returns the current size of the n -th future bucket of the
361 GC smoothing system. The unit is one millionth of a full GC.
362 Since 4.03.0
365 Raises Invalid_argument if n is negative, return 0 if n is larger than
368 the smoothing window.
369 val get_credit : unit -> int
373 get_credit () returns the current size of the "work done in advance"
376 counter of the GC smoothing system. The unit is one millionth of a full
377 GC.
378 Since 4.03.0
381 val huge_fallback_count : unit -> int
385 Return the number of times we tried to map huge pages and had to fall
387 back to small pages. This is always 0 if OCAMLRUNPARAM contains H=1 .
388 Since 4.03.0
391 val finalise : ('a -> unit) -> 'a -> unit
395 finalise f v registers f as a finalisation function for v . v must be
398 heap-allocated. f will be called with v as argument at some point be‐
399 tween the first time v becomes unreachable (including through weak
400 pointers) and the time v is collected by the GC. Several functions can
401 be registered for the same value, or even several instances of the same
402 function. Each instance will be called once (or never, if the program
403 terminates before v becomes unreachable).
404 The GC will call the finalisation functions in the order of dealloca‐
406 tion. When several values become unreachable at the same time (i.e.
407 during the same GC cycle), the finalisation functions will be called in
408 the reverse order of the corresponding calls to finalise . If finalise
409 is called in the same order as the values are allocated, that means
410 each value is finalised before the values it depends upon. Of course,
411 this becomes false if additional dependencies are introduced by assign‐
412 ments.
413 In the presence of multiple OCaml threads it should be assumed that any
415 particular finaliser may be executed in any of the threads.
416 Anything reachable from the closure of finalisation functions is con‐
418 sidered reachable, so the following code will not work as expected:
419 - let v = ... in Gc.finalise (fun _ -> ...v...) v
421 Instead you should make sure that v is not in the closure of the final‐
423 isation function by writing:
424 - let f = fun x -> ... let v = ... in Gc.finalise f v
426 The f function can use all features of OCaml, including assignments
428 that make the value reachable again. It can also loop forever (in this
429 case, the other finalisation functions will not be called during the
430 execution of f, unless it calls finalise_release ). It can call fi‐
431 nalise on v or other values to register other functions or even itself.
432 It can raise an exception; in this case the exception will interrupt
433 whatever the program was doing when the function was called.
434 finalise will raise Invalid_argument if v is not guaranteed to be
437 heap-allocated. Some examples of values that are not heap-allocated
438 are integers, constant constructors, booleans, the empty array, the
439 empty list, the unit value. The exact list of what is heap-allocated
440 or not is implementation-dependent. Some constant values can be
441 heap-allocated but never deallocated during the lifetime of the pro‐
442 gram, for example a list of integer constants; this is also implementa‐
443 tion-dependent. Note that values of types float are sometimes allo‐
444 cated and sometimes not, so finalising them is unsafe, and finalise
445 will also raise Invalid_argument for them. Values of type 'a Lazy.t
446 (for any 'a ) are like float in this respect, except that the compiler
447 sometimes optimizes them in a way that prevents finalise from detecting
448 them. In this case, it will not raise Invalid_argument , but you should
449 still avoid calling finalise on lazy values.
450 The results of calling String.make , Bytes.make , Bytes.create , Ar‐
452 ray.make , and ref are guaranteed to be heap-allocated and non-constant
453 except when the length argument is 0 .
454 val finalise_last : (unit -> unit) -> 'a -> unit
458 same as Gc.finalise except the value is not given as argument. So you
460 can't use the given value for the computation of the finalisation func‐
461 tion. The benefit is that the function is called after the value is un‐
462 reachable for the last time instead of the first time. So contrary to
463 Gc.finalise the value will never be reachable again or used again. In
464 particular every weak pointer and ephemeron that contained this value
465 as key or data is unset before running the finalisation function. More‐
466 over the finalisation functions attached with Gc.finalise are always
467 called before the finalisation functions attached with Gc.finalise_last
468 .
469 Since 4.04
472 val finalise_release : unit -> unit
476 A finalisation function may call finalise_release to tell the GC that
478 it can launch the next finalisation function without waiting for the
479 current one to return.
480 type alarm
483 An alarm is a piece of data that calls a user function at the end of
486 each major GC cycle. The following functions are provided to create
487 and delete alarms.
488 val create_alarm : (unit -> unit) -> alarm
492 create_alarm f will arrange for f to be called at the end of each major
495 GC cycle, starting with the current cycle or the next one. A value of
496 type alarm is returned that you can use to call delete_alarm .
497 val delete_alarm : alarm -> unit
501 delete_alarm a will stop the calls to the function associated to a .
504 Calling delete_alarm a again has no effect.
505 val eventlog_pause : unit -> unit
509 eventlog_pause () will pause the collection of traces in the runtime.
512 Traces are collected if the program is linked to the instrumented run‐
513 time and started with the environment variable OCAML_EVENTLOG_ENABLED.
514 Events are flushed to disk after pausing, and no new events will be
515 recorded until eventlog_resume is called.
516 val eventlog_resume : unit -> unit
520 eventlog_resume () will resume the collection of traces in the runtime.
523 Traces are collected if the program is linked to the instrumented run‐
524 time and started with the environment variable OCAML_EVENTLOG_ENABLED.
525 This call can be used after calling eventlog_pause , or if the program
526 was started with OCAML_EVENTLOG_ENABLED=p. (which pauses the collection
527 of traces before the first event.)
528 module Memprof : sig end
531 Memprof is a sampling engine for allocated memory words. Every allo‐
535 cated word has a probability of being sampled equal to a configurable
536 sampling rate. Once a block is sampled, it becomes tracked. A tracked
537 block triggers a user-defined callback as soon as it is allocated, pro‐
538 moted or deallocated.
539 Since blocks are composed of several words, a block can potentially be
541 sampled several times. If a block is sampled several times, then each
542 of the callback is called once for each event of this block: the multi‐
543 plicity is given in the n_samples field of the allocation structure.
544 This engine makes it possible to implement a low-overhead memory pro‐
546 filer as an OCaml library.
547 Note: this API is EXPERIMENTAL. It may change without prior notice.
549OCamldoc 2021-07-22 Gc(3)