2 .\" Copyright (c) 2005, Sun Microsystems, Inc. All Right Reserved.
3 .\" The contents of this file are subject to the terms of the Common Development and Distribution License (the "License"). You may not use this file except in compliance with the License.
4 .\" You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing. See the License for the specific language governing permissions and limitations under the License.
5 .\" When distributing Covered Code, include this CDDL HEADER in each file and include the License file at usr/src/OPENSOLARIS.LICENSE. If applicable, add the following below this CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your own identifying information: Portions Copyright [yyyy] [name of copyright owner]
6 .TH MADVISE 3C "Mar 28, 2016"
8 madvise \- provide advice to VM system
12 #include <sys/types.h>
15 \fBint\fR \fBmadvise\fR(\fBcaddr_t\fR \fIaddr\fR, \fBsize_t\fR \fIlen\fR, \fBint\fR \fIadvice\fR);
20 The \fBmadvise()\fR function advises the kernel that a region of user mapped
21 memory in the range [\fIaddr\fR, \fIaddr\fR + \fIlen\fR) will be accessed
22 following a type of pattern. The kernel uses this information to optimize the
23 procedure for manipulating and maintaining the resources associated with the
24 specified mapping range. In general (and true to the name of the function),
25 the advice is merely advisory, and the only user-visible ramifications
26 are in terms of performance, not semantics. Note that
27 \fBMADV_PURGE\fR is an exception to this; see below for details.
30 Values for \fIadvice\fR are defined in <\fBsys/mman.h\fR> as:
34 #define MADV_NORMAL 0x0 /* No further special treatment */
35 #define MADV_RANDOM 0x1 /* Expect random page references */
36 #define MADV_SEQUENTIAL 0x2 /* Expect sequential page references */
37 #define MADV_WILLNEED 0x3 /* Will need these pages */
38 #define MADV_DONTNEED 0x4 /* Don't need these pages */
39 #define MADV_FREE 0x5 /* Contents can be freed */
40 #define MADV_ACCESS_DEFAULT 0x6 /* default access */
41 #define MADV_ACCESS_LWP 0x7 /* next LWP to access heavily */
42 #define MADV_ACCESS_MANY 0x8 /* many processes to access heavily */
43 #define MADV_PURGE 0x9 /* contents will be purged */
50 \fB\fBMADV_NORMAL\fR\fR
53 This is the default system characteristic where accessing memory within the
54 address range causes the system to read data from the mapped file. The kernel
55 reads all data from files into pages which are retained for a period of time as
56 a "cache." System pages can be a scarce resource, so the kernel steals pages
57 from other mappings when needed. This is a likely occurrence, but adversely
58 affects system performance only if a large amount of memory is accessed.
64 \fB\fBMADV_RANDOM\fR\fR
67 Tell the kernel to read in a minimum amount of data from a mapped file on any
68 single particular access. If \fBMADV_NORMAL\fR is in effect when an address of
69 a mapped file is accessed, the system tries to read in as much data from the
70 file as reasonable, in anticipation of other accesses within a certain
77 \fB\fBMADV_SEQUENTIAL\fR\fR
80 Tell the system that addresses in this range are likely to be accessed only
81 once, so the system will free the resources mapping the address range as
88 \fB\fBMADV_WILLNEED\fR\fR
91 Tell the system that a certain address range is definitely needed so the kernel
92 will start reading the specified range into memory. This can benefit programs
93 wanting to minimize the time needed to access memory the first time, as the
94 kernel would need to read in from the file.
100 \fB\fBMADV_DONTNEED\fR\fR
103 Tell the kernel that the specified address range is no longer needed, so the
104 system starts to free the resources associated with the address range.
105 While the semantics of \fBMADV_DONTNEED\fR are similar to other systems,
106 they differ significantly from the semantics on Linux, where
107 \fBMADV_DONTNEED\fR will actually synchronously purge the address range,
108 and subsequent faults will load from either backing store or be
109 zero-filled on demand. If the peculiar Linux semantics are
110 desired, \fBMADV_PURGE\fR should be used in lieu of \fBMADV_DONTNEED\fR.
116 \fB\fBMADV_FREE\fR\fR
119 Tell the kernel that contents in the specified address range are no longer
120 important and the range will be overwritten. When there is demand for memory,
121 the system will free pages associated with the specified address range. In this
122 instance, the next time a page in the address range is referenced, it will
123 contain all zeroes. Otherwise, it will contain the data that was there prior
124 to the \fBMADV_FREE\fR call. References made to the address range will not make
125 the system read from backing store (swap space) until the page is modified
128 This value cannot be used on mappings that have underlying file objects.
134 \fB\fBMADV_PURGE\fR\fR
137 Tell the kernel to purge the specified address range. The mapping will
138 be retained, but the pages themselves will be destroyed; subsequent
139 faults on the range will result in the page being read from backing
140 store (if file-backed) or being zero-filled on demand (if anonymous). Note
141 that these semantics are generally inferior to \fBMADV_FREE\fR, which gives the
142 system more flexibility and results in better performance
143 when pages are, in fact, reused by the caller. Indeed, \fBMADV_PURGE\fR only
144 exists to provide an equivalent to the unfortunate
145 \fBMADV_DONTNEED\fR semantics found in Linux, upon which some programs
146 have (regrettably) come to depend. In de novo applications,
147 \fBMADV_PURGE\fR should be avoided; \fBMADV_FREE\fR should always be
154 \fB\fBMADV_ACCESS_LWP\fR\fR
157 Tell the kernel that the next LWP to touch the specified address range will
158 access it most heavily, so the kernel should try to allocate the memory and
159 other resources for this range and the LWP accordingly.
165 \fB\fBMADV_ACCESS_MANY\fR\fR
168 Tell the kernel that many processes and/or LWPs will access the specified
169 address range randomly across the machine, so the kernel should try to allocate
170 the memory and other resources for this range accordingly.
176 \fB\fBMADV_ACCESS_DEFAULT\fR\fR
179 Reset the kernel's expectation for how the specified range will be accessed to
185 The \fBmadvise()\fR function should be used by applications with specific
186 knowledge of their access patterns over a memory object, such as a mapped file,
187 to increase system performance.
190 Upon successful completion, \fBmadvise()\fR returns \fB0\fR; otherwise, it
191 returns \fB\(mi1\fR and sets \fBerrno\fR to indicate the error.
198 Some or all mappings in the address range [\fIaddr\fR, \fIaddr\fR +
199 \fIlen\fR) are locked for I/O.
208 Some or all of the addresses in the range [\fIaddr\fR, \fIaddr\fR + \fIlen\fR)
209 are locked and \fBMS_SYNC\fR with the \fBMS_INVALIDATE\fR option is specified.
218 Some or all of the addresses in the specified range could not be read into
219 memory from the underlying object when performing \fBMADV_WILLNEED\fR. The
220 \fBmadvise()\fR function could return prior to this condition being detected,
221 in which case \fBerrno\fR will not be set to \fBEFAULT\fR.
230 The \fIaddr\fR argument is not a multiple of the page size as returned by
231 \fBsysconf\fR(3C), the length of the specified address range is equal to 0, or
232 the \fIadvice\fR argument was invalid.
241 An I/O error occurred while reading from or writing to the file system.
250 Addresses in the range [\fIaddr\fR, \fIaddr\fR + \fIlen\fR) are outside the
251 valid range for the address space of a process, or specify one or more pages
261 Stale \fBNFS\fR file handle.
266 See \fBattributes\fR(5) for descriptions of the following attributes:
274 ATTRIBUTE TYPE ATTRIBUTE VALUE
276 Interface Stability Stable
283 \fBmeminfo\fR(2), \fBmmap\fR(2), \fBsysconf\fR(3C), \fBattributes\fR(5)