2 .\" Copyright (c) 2001 Dag-Erling Coïdan Smørgrav
3 .\" All rights reserved.
5 .\" Redistribution and use in source and binary forms, with or without
6 .\" modification, are permitted provided that the following conditions
8 .\" 1. Redistributions of source code must retain the above copyright
9 .\" notice, this list of conditions and the following disclaimer.
10 .\" 2. Redistributions in binary form must reproduce the above copyright
11 .\" notice, this list of conditions and the following disclaimer in the
12 .\" documentation and/or other materials provided with the distribution.
14 .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15 .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16 .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17 .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18 .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19 .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20 .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21 .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22 .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 .Nm uma_zalloc_domain ,
38 .Nm uma_zfree_domain ,
40 .Nm uma_zone_set_max ,
41 .Nm uma_zone_get_max ,
42 .Nm uma_zone_get_cur ,
43 .Nm uma_zone_set_warning ,
44 .Nm uma_zone_set_maxaction
52 .Fa "char *name" "int size"
53 .Fa "uma_ctor ctor" "uma_dtor dtor" "uma_init uminit" "uma_fini fini"
54 .Fa "int align" "uint16_t flags"
57 .Fn uma_zalloc "uma_zone_t zone" "int flags"
59 .Fn uma_zalloc_arg "uma_zone_t zone" "void *arg" "int flags"
61 .Fn uma_zalloc_domain "uma_zone_t zone" "void *arg" "int domain" "int flags"
63 .Fn uma_zfree "uma_zone_t zone" "void *item"
65 .Fn uma_zfree_arg "uma_zone_t zone" "void *item" "void *arg"
67 .Fn uma_zfree_domain "uma_zone_t zone" "void *item" "void *arg"
69 .Fn uma_zdestroy "uma_zone_t zone"
71 .Fn uma_zone_set_max "uma_zone_t zone" "int nitems"
73 .Fn uma_zone_get_max "uma_zone_t zone"
75 .Fn uma_zone_get_cur "uma_zone_t zone"
77 .Fn uma_zone_set_warning "uma_zone_t zone" "const char *warning"
79 .Fn uma_zone_set_maxaction "uma_zone_t zone" "void (*maxaction)(uma_zone_t)"
81 .Fn SYSCTL_UMA_MAX parent nbr name access zone descr
82 .Fn SYSCTL_ADD_UMA_MAX ctx parent nbr name access zone descr
83 .Fn SYSCTL_UMA_CUR parent nbr name access zone descr
84 .Fn SYSCTL_ADD_UMA_CUR ctx parent nbr name access zone descr
86 The zone allocator provides an efficient interface for managing
87 dynamically-sized collections of items of identical size.
88 The zone allocator can work with preallocated zones as well as with
89 runtime-allocated ones, and is therefore available much earlier in the
90 boot process than other memory management routines. The zone allocator
91 provides per-cpu allocation caches with linear scalability on SMP
92 systems as well as round-robin and first-touch policies for NUMA
95 A zone is an extensible collection of items of identical size.
96 The zone allocator keeps track of which items are in use and which
97 are not, and provides functions for allocating items from the zone and
98 for releasing them back (which makes them available for later use).
100 After the first allocation of an item,
101 it will have been cleared to zeroes, however subsequent allocations
102 will retain the contents as of the last free.
106 function creates a new zone from which items may then be allocated from.
109 argument is a text name of the zone for debugging and stats; this memory
110 should not be freed until the zone has been deallocated.
116 arguments are callback functions that are called by
117 the uma subsystem at the time of the call to
122 Their purpose is to provide hooks for initializing or
123 destroying things that need to be done at the time of the allocation
124 or release of a resource.
130 might be to adjust a global count of the number of objects allocated.
136 arguments are used to optimize the allocation of
137 objects from the zone.
138 They are called by the uma subsystem whenever
139 it needs to allocate or free several items to satisfy requests or memory
145 callbacks might be to
146 initialize and destroy mutexes contained within the object.
148 allow one to re-use already initialized mutexes when an object is returned
149 from the uma subsystem's object cache.
150 They are not called on each call to
154 but rather in a batch mode on several objects.
160 is a subset of the following flags:
161 .Bl -tag -width "foo"
162 .It Dv UMA_ZONE_NOFREE
163 Slabs of the zone are never returned back to VM.
164 .It Dv UMA_ZONE_NODUMP
165 Pages belonging to the zone will not be included into mini-dumps.
167 An allocation from zone would have
169 shadow copies, that are privately assigned to CPUs.
170 A CPU can address its private copy using base allocation address plus
171 multiple of current CPU id and
172 .Fn sizeof "struct pcpu" :
173 .Bd -literal -offset indent
174 foo_zone = uma_zcreate(..., UMA_ZONE_PCPU);
176 foo_base = uma_zalloc(foo_zone, ...);
179 foo_pcpu = (foo_t *)zpcpu_get(foo_base);
180 /* do something with foo_pcpu */
183 .It Dv UMA_ZONE_OFFPAGE
184 By default book-keeping of items within a slab is done in the slab page itself.
185 This flag explicitly tells subsystem that book-keeping structure should be
186 allocated separately from special internal zone.
187 This flag requires either
191 since subsystem requires a mechanism to find a book-keeping structure
192 to an item being freed.
193 The subsystem may choose to prefer offpage book-keeping for certain zones
195 .It Dv UMA_ZONE_ZINIT
196 The zone will have its
198 method set to internal method that initializes a new allocated slab
206 flag would not return zeroed memory on every
209 The zone should use an internal hash table to find slab book-keeping
210 structure where an allocation being freed belongs to.
211 .It Dv UMA_ZONE_VTOSLAB
212 The zone should use special field of
214 to find slab book-keeping structure where an allocation being freed belongs to.
215 .It Dv UMA_ZONE_MALLOC
220 The zone is for the VM subsystem.
222 The zone should use a first-touch NUMA policy rather than the round-robin
223 default. Callers that do not free memory on the same domain it is allocated
224 from will cause mixing in per-cpu caches. See
225 .Xr numa 9 for more details.
228 To allocate an item from a zone, simply call
230 with a pointer to that zone
233 argument to selected flags as documented in
235 It will return a pointer to an item if successful,
238 in the rare case where all items in the zone are in use and the
239 allocator is unable to grow the zone
244 Items are released back to the zone from which they were allocated by
247 with a pointer to the zone and a pointer to the item.
261 specify an argument for the
265 functions, respectively.
267 .Fn uma_zalloc_domain
268 function allows callers to specify a fixed
269 .Xr numa 9 domain to allocate from. This uses a guaranteed but slow path in
270 the allocator which reduces concurrency. The
272 function should be used to return memory allocated in this fashion. This
273 function infers the domain from the pointer and does not require it as an
278 can be destroyed using
280 freeing all memory that was allocated for the zone.
281 All items allocated from the zone with
283 must have been freed with
289 function limits the number of items
290 .Pq and therefore memory
291 that can be allocated to
295 argument specifies the requested upper limit number of items.
296 The effective limit is returned to the caller, as it may end up being higher
297 than requested due to the implementation rounding up to ensure all memory pages
298 allocated to the zone are utilised to capacity.
299 The limit applies to the total number of items in the zone, which includes
300 allocated items, free items and free items in the per-cpu caches.
301 On systems with more than one CPU it may not be possible to allocate
302 the specified number of items even when there is no shortage of memory,
303 because all of the remaining free items may be in the caches of the
304 other CPUs when the limit is hit.
308 function returns the effective upper limit number of items for a zone.
312 function returns the approximate current occupancy of the zone.
313 The returned value is approximate because appropriate synchronisation to
314 determine an exact value is not performed by the implementation.
315 This ensures low overhead at the expense of potentially stale data being used
319 .Fn uma_zone_set_warning
320 function sets a warning that will be printed on the system console when the
321 given zone becomes full and fails to allocate an item.
322 The warning will be printed no more often than every five minutes.
323 Warnings can be turned off globally by setting the
329 .Fn uma_zone_set_maxaction
330 function sets a function that will be called when the given zone becomes full
331 and fails to allocate an item.
332 The function will be called with the zone locked.
334 that called the allocation function may have held additional locks.
336 this function should do very little work (similar to a signal handler).
339 .Fn SYSCTL_UMA_MAX parent nbr name access zone descr
340 macro declares a static
342 oid that exports the effective upper limit number of items for a zone.
345 argument should be a pointer to
347 A read of the oid returns value obtained through
348 .Fn uma_zone_get_max .
349 A write to the oid sets new value via
350 .Fn uma_zone_set_max .
352 .Fn SYSCTL_ADD_UMA_MAX ctx parent nbr name access zone descr
353 macro is provided to create this type of oid dynamically.
356 .Fn SYSCTL_UMA_CUR parent nbr name access zone descr
357 macro declares a static read-only
359 oid that exports the approximate current occupancy of the zone.
362 argument should be a pointer to
364 A read of the oid returns value obtained through
365 .Fn uma_zone_get_cur .
367 .Fn SYSCTL_ADD_UMA_CUR ctx parent nbr name zone descr
368 macro is provided to create this type of oid dynamically.
372 function returns a pointer to an item, or
374 if the zone ran out of unused items
381 The zone allocator first appeared in
383 It was radically changed in
385 to function as a slab allocator.
388 The zone allocator was written by
390 The zone allocator was rewritten in large parts by
391 .An Jeff Roberson Aq Mt jeff@FreeBSD.org
392 to function as a slab allocator.
394 This manual page was written by
395 .An Dag-Erling Sm\(/orgrav Aq Mt des@FreeBSD.org .
397 .An Jeroen Ruigrok van der Werven Aq Mt asmodai@FreeBSD.org .