.\" $OpenBSD: pool.9,v 1.42 2011/07/05 23:42:44 tedu Exp $ .\" $NetBSD: pool.9,v 1.18 2001/06/21 11:59:01 wiz Exp $ .\" .\" Copyright (c) 1997, 1998 The NetBSD Foundation, Inc. .\" All rights reserved. .\" .\" This code is derived from software contributed to The NetBSD Foundation .\" by Paul Kranenburg. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS .\" ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED .\" TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR .\" PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS .\" BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" .Dd $Mdocdate: July 5 2011 $ .Dt POOL 9 .Os .Sh NAME .Nm pool_init , .Nm pool_destroy , .Nm pool_get , .Nm pool_put , .Nm pool_prime , .Nm pool_setipl , .Nm pool_sethiwat , .Nm pool_setlowat , .Nm pool_sethardlimit .Nd resource-pool manager .Sh SYNOPSIS .Fd #include <sys/types.h> .Fd #include <sys/pool.h> .Ft void .Fo pool_init .Fa "struct pool *pool" .Fa "size_t size" .Fa "u_int align" .Fa "u_int align_offset" .Fa "int flags" .Fa "const char *wmesg" .Fa "struct pool_allocator *palloc" .Fc .Ft void .Fo pool_destroy .Fa "struct pool *pp" .Fc .Ft void * .Fn pool_get "struct pool *pp" "int flags" .Ft void .Fn pool_put "struct pool *pp" "void *item" .Ft int .Fn pool_prime "struct pool *pp" "int nitems" .Ft void .Fn pool_setipl "struct pool *pp" "int ipl" .Ft void .Fn pool_sethiwat "struct pool *pp" "int n" .Ft void .Fn pool_setlowat "struct pool *pp" "int n" .Ft int .Fo pool_sethardlimit .Fa "struct pool *pp" .Fa "unsigned n" .Fa "const char *warnmess" .Fa "int ratecap" .Fc .Sh DESCRIPTION These utility routines provide management of pools of fixed-sized areas of memory. Resource pools set aside an amount of memory for exclusive use by the resource pool owner. This can be used by applications to guarantee the availability of a minimum amount of memory needed to continue operation independent of the memory resources currently available from the system-wide memory allocator .Pq Xr malloc 9 . The pool manager obtains memory by using the special-purpose memory allocator .Fa palloc passed to .Fn pool_init , for extra pool items in case the number of allocations exceeds the nominal number of pool items managed by a pool resource. This temporary memory will be automatically returned to the system at a later time. .Ss CREATING A POOL The function .Fn pool_init initializes a resource pool. The arguments are: .Bl -tag -offset indent -width "align_offset" .It Fa pool Specifies the pool storage to be initialized. .It Fa size Specifies the size of the memory items managed by the pool. .It Fa align Specifies the memory address alignment of the items returned by .Fn pool_get . This argument must be a power of two. If zero, the alignment defaults to an architecture-specific natural alignment. .It Fa align_offset The offset within an item to which the .Fa align parameter applies. .It Fa flags Specifies various flags set on the pool at creation time. .It Fa wmesg The message passed on to .Xr tsleep 9 if .Fn pool_get must wait for items to be returned to the pool. .It Fa palloc The back-end allocator used to manage the memory for the pool. .Fa palloc may be .Dv NULL , in which case the pool manager uses an interrupt safe allocator. It is recommended that this be set to .Em pool_allocator_nointr if the pool will never be accessed in an interrupt context. .El .Ss DESTROYING A POOL The .Fn pool_destroy function destroys a resource pool. It takes a single argument .Fa pp identifying the pool resource instance. .Ss ALLOCATING ITEMS FROM A POOL .Fn pool_get allocates an item from the pool and returns a pointer to it. .Bl -tag -offset indent -width "flags" .It Fa pp The handle identifying the pool resource instance. .It Fa flags One or more of .Dv PR_URGENT , .Dv PR_WAITOK , .Dv PR_NOWAIT or .Dv PR_LIMITFAIL , that define behaviour in case the pooled resources are depleted. If no resources are available and .Dv PR_WAITOK is given, this function will wait until items are returned to the pool. Otherwise .Fn pool_get returns .Dv NULL . .Dv PR_NOWAIT should be provided in cases where .Dv PR_WAITOK is not. If .Dv PR_URGENT is specified and no items are available and .Fa palloc cannot allocate a new page, the system will panic .Pq XXX . .\"Undefined behaviour results if .\".Dv PR_MALLOCOK .\"is specified on a pool handle that was created using client-provided .\"storage. .\" a bunch of other flags aren't documented. If both .Dv PR_LIMITFAIL and .Dv PR_WAITOK are specified, and the pool has reached its hard limit, .Fn pool_get will return .Dv NULL without waiting, allowing the caller to do its own garbage collection; however, it will still wait if the pool is not yet at its hard limit. If .Dv PR_ZERO is passed and an item has been successfully allocated, it is zeroed before being returned to the caller. .El .Ss RETURNING ITEMS TO A POOL .Fn pool_put returns the pool item pointed at by .Fa item to the resource pool identified by the pool handle .Fa pp . If the number of available items in the pool exceeds the maximum pool size set by .Fn pool_sethiwat and there are no outstanding requests for pool items, the excess items will be returned to the system by calling .Fn prelease . .Bl -tag -offset indent -width "item" .It Fa pp The handle identifying the pool resource instance. .It Fa item A pointer to a pool item previously obtained by .Fn pool_get . .El .Ss PRIMING A POOL .Fn pool_prime adds items to the pool. Storage space for the items is allocated by using the page allocation routine specified to .Fn pool_init . .Pp .Fn pool_prime .Bl -tag -offset indent -width "nitems" .It Fa pp The handle identifying the pool resource instance. .It Fa nitems The number of items to add to the pool. .El .Pp This function may return .Dv ENOMEM in case the requested number of items could not be allocated. Otherwise, the return value is 0. .Ss SETTING POOL RESOURCE WATERMARKS A pool will attempt to increase its resource usage to keep up with the demand for its items. Conversely, it will return unused memory to the system should the number of accumulated unused items in the pool exceed a programmable limit. The limits for the minimum and maximum number of items which a pool should keep at hand are known as the high and low .Sy watermarks . The functions .Fn pool_sethiwat and .Fn pool_setlowat set a pool's high and low watermarks, respectively. .Pp .Fn pool_sethiwat .Bl -tag -offset indent -width "flags" .It Fa pp The handle identifying the pool resource instance. .It Fa n The maximum number of items to keep in the pool. As items are returned and the total number of pages in the pool is larger than the maximum set by this function, any completely unused pages are released immediately .Pq by calling Fn prelease . If this function is not used to specify a maximum number of items, the pages will remain associated with the pool until the system runs low on memory, at which point the VM system will try to reclaim unused pages. .El .Pp .Fn pool_setlowat .Bl -tag -offset indent -width "flags" .It Fa pp The handle identifying the pool resource instance. .It Fa n The minimum number of items to keep in the pool. The number of pages in the pool will not decrease below the required value to accommodate the minimum number of items specified by this function. Unlike .Fn pool_prime , this function does not allocate the necessary memory up-front. .El .Ss SETTING THE PROTECTION LEVEL The .Fn pool_setipl function is used to specify the interrupt protection level at which the pool can be safely used. .Pp .Fn pool_setipl .Bl -tag -offset indent -width "flags" .It Fa pp The handle identifying the pool resource instance. .It Fa ipl The interrupt protection level used to protect the pool's internals. See .Xr spl 9 for a list of the IPLs. .El .Ss SETTING HARD LIMITS The function .Fn pool_sethardlimit sets a hard limit on the pool to .Fa n items. If the hard limit is reached .Fa warnmess will be printed to the console, but no more than every .Fa ratecap seconds. Upon successful completion, a value of 0 is returned. The value EINVAL is returned when the current size of the pool already exceeds the requested hard limit. .Ss POTENTIAL PITFALLS Note that undefined behaviour results when mixing the storage providing methods supported by the pool resource routines. .Pp The pool resource code uses a per-pool lock to protect its internal state. If any pool functions are called in an interrupt context, the caller must block all interrupts that might cause the code to be reentered. .Ss DEBUGGING To debug a misbehaving pool, a kernel can be compiled with the .Dv MALLOC_DEBUG option and memory debugging on pools can be enabled with the .Dv PR_DEBUG flag passed in the .Fa flags argument in the call to .Fn pool_init . See .Xr malloc 9 for more information about .Dv MALLOC_DEBUG . Alternatively, the .Dv PR_DEBUGCHK flag can be passed to enable pool internal consistency checks before and after each allocation and free. .Sh CODE REFERENCES The pool manager is implemented in the file .Pa sys/kern/subr_pool.c . .Sh SEE ALSO .Xr free 9 , .Xr malloc 9 , .Xr spl 9 , .Xr uvm 9 .Sh HISTORY The pool manager first appeared in .Nx 1.4 and was ported to .Ox by .An Artur Grabowski Aq art@openbsd.org .