man2/posix_fadvise.2

   1 .\" Copyright 2003 Abhijit Menon-Sen <ams@wiw.org>
   2 .\"
   3 .\" %%%LICENSE_START(VERBATIM)
   4 .\" Permission is granted to make and distribute verbatim copies of this
   5 .\" manual provided the copyright notice and this permission notice are
   6 .\" preserved on all copies.
   7 .\"
   8 .\" Permission is granted to copy and distribute modified versions of this
   9 .\" manual under the conditions for verbatim copying, provided that the
  10 .\" entire resulting derived work is distributed under the terms of a
  11 .\" permission notice identical to this one.
  12 .\"
  13 .\" Since the Linux kernel and libraries are constantly changing, this
  14 .\" manual page may be incorrect or out-of-date.  The author(s) assume no
  15 .\" responsibility for errors or omissions, or for damages resulting from
  16 .\" the use of the information contained herein.  The author(s) may not
  17 .\" have taken the same level of care in the production of this manual,
  18 .\" which is licensed free of charge, as they might when working
  19 .\" professionally.
  20 .\"
  21 .\" Formatted or processed versions of this manual, if unaccompanied by
  22 .\" the source, must acknowledge the copyright and authors of this work.
  23 .\" %%%LICENSE_END
  24 .\"
  25 .\" 2005-04-08 mtk, noted kernel version and added BUGS
  26 .\" 2010-10-09, mtk, document arm_fadvise64_64()
  27 .\"
  28 .TH POSIX_FADVISE 2 2015-04-19 "Linux" "Linux Programmer's Manual"
  29 .SH NAME
  30 posix_fadvise \- predeclare an access pattern for file data
  31 .SH SYNOPSIS
  32 .nf
  33 .B #include <fcntl.h>
  34 .sp
  35 .BI "int posix_fadvise(int " fd ", off_t " offset ", off_t " len \
  36 ", int " advice ");"
  37 .fi
  38 .sp
  39 .ad l
  40 .in -4n
  41 Feature Test Macro Requirements for glibc (see
  42 .BR feature_test_macros (7)):
  43 .in
  44 .sp
  45 .BR posix_fadvise ():
  46 .RS 4
  47 _XOPEN_SOURCE\ >=\ 600 || _POSIX_C_SOURCE\ >=\ 200112L
  48 .RE
  49 .ad
  50 .SH DESCRIPTION
  51 Programs can use
  52 .BR posix_fadvise ()
  53 to announce an intention to access
  54 file data in a specific pattern in the future, thus allowing the kernel
  55 to perform appropriate optimizations.
  56
  57 The \fIadvice\fP applies to a (not necessarily existent) region starting
  58 at \fIoffset\fP and extending for \fIlen\fP bytes (or until the end of
  59 the file if \fIlen\fP is 0) within the file referred to by \fIfd\fP.
  60 The \fIadvice\fP is not binding;
  61 it merely constitutes an expectation on behalf of
  62 the application.
  63
  64 Permissible values for \fIadvice\fP include:
  65 .TP
  66 .B POSIX_FADV_NORMAL
  67 Indicates that the application has no advice to give about its access
  68 pattern for the specified data.
  69 If no advice is given for an open file,
  70 this is the default assumption.
  71 .TP
  72 .B POSIX_FADV_SEQUENTIAL
  73 The application expects to access the specified data sequentially (with
  74 lower offsets read before higher ones).
  75 .TP
  76 .B POSIX_FADV_RANDOM
  77 The specified data will be accessed in random order.
  78 .TP
  79 .B POSIX_FADV_NOREUSE
  80 The specified data will be accessed only once.
  81 .TP
  82 .B POSIX_FADV_WILLNEED
  83 The specified data will be accessed in the near future.
  84 .TP
  85 .B POSIX_FADV_DONTNEED
  86 The specified data will not be accessed in the near future.
  87 .SH RETURN VALUE
  88 On success, zero is returned.
  89 On error, an error number is returned.
  90 .SH ERRORS
  91 .TP
  92 .B EBADF
  93 The \fIfd\fP argument was not a valid file descriptor.
  94 .TP
  95 .B EINVAL
  96 An invalid value was specified for \fIadvice\fP.
  97 .TP
  98 .B ESPIPE
  99 The specified file descriptor refers to a pipe or FIFO.
 100 .RB ( ESPIPE
 101 is the error specified by POSIX,
 102 but before kernel version 2.16,
 103 .\" commit 87ba81dba431232548ce29d5d224115d0c2355ac
 104 Linux returned
 105 .B EINVAL
 106 in this case.)
 107 .SH VERSIONS
 108 Kernel support first appeared in Linux 2.5.60;
 109 the underlying system call is called
 110 .BR fadvise64 ().
 111 .\" of fadvise64_64()
 112 Library support has been provided since glibc version 2.2,
 113 via the wrapper function
 114 .BR posix_fadvise ().
 115
 116 Since Linux 3.18,
 117 .\" commit d3ac21cacc24790eb45d735769f35753f5b56ceb
 118 support for the underlying system call is optional,
 119 depending on the setting of the
 120 .B CONFIG_ADVISE_SYSCALLS
 121 configuration option.
 122 .SH CONFORMING TO
 123 POSIX.1-2001.
 124 Note that the type of the
 125 .I len
 126 argument was changed from
 127 .I size_t
 128 to
 129 .I off_t
 130 in POSIX.1-2003 TC1.
 131 .SH NOTES
 132 Under Linux, \fBPOSIX_FADV_NORMAL\fP sets the readahead window to the
 133 default size for the backing device; \fBPOSIX_FADV_SEQUENTIAL\fP doubles
 134 this size, and \fBPOSIX_FADV_RANDOM\fP disables file readahead entirely.
 135 These changes affect the entire file, not just the specified region
 136 (but other open file handles to the same file are unaffected).
 137
 138 \fBPOSIX_FADV_WILLNEED\fP initiates a
 139 nonblocking read of the specified region into the page cache.
 140 The amount of data read may be decreased by the kernel depending
 141 on virtual memory load.
 142 (A few megabytes will usually be fully satisfied,
 143 and more is rarely useful.)
 144
 145 In kernels before 2.6.18, \fBPOSIX_FADV_NOREUSE\fP had the
 146 same semantics as \fBPOSIX_FADV_WILLNEED\fP.
 147 This was probably a bug; since kernel 2.6.18, this flag is a no-op.
 148
 149 \fBPOSIX_FADV_DONTNEED\fP attempts to free cached pages associated with
 150 the specified region.
 151 This is useful, for example, while streaming large
 152 files.
 153 A program may periodically request the kernel to free cached data
 154 that has already been used, so that more useful cached pages are not
 155 discarded instead.
 156
 157 Requests to discard partial pages are ignored.
 158 It is preferable to preserve needed data than discard unneeded data.
 159 If the application requires that data be considered for discarding then
 160 .I offset
 161 and
 162 .I len
 163 must be page-aligned.
 164
 165 Pages that have not yet been written out will be unaffected, so if the
 166 application wishes to guarantee that pages will be released, it should
 167 call
 168 .BR fsync (2)
 169 or
 170 .BR fdatasync (2)
 171 first.
 172 .SS C library/kernel differences
 173 The name of the wrapper function in the C library is
 174 .BR posix_fadvise ().
 175 The underlying system call is called
 176 .BR fadvise64 ()
 177 (or, on some architectures,
 178 .BR fadvise64_64 ()).
 179 .SS Architecture-specific variants
 180 Some architectures require
 181 64-bit arguments to be aligned in a suitable pair of registers (see
 182 .BR syscall (2)
 183 for further detail).
 184 On such architectures, the call signature of
 185 .BR posix_fadvise ()
 186 shown in the SYNOPSIS would force
 187 a register to be wasted as padding between the
 188 .I fd
 189 and
 190 .I offset
 191 arguments.
 192 Therefore, these architectures define a version of the
 193 system call that orders the arguments suitably,
 194 but is otherwise exactly the same as
 195 .BR posix_fadvise ().
 196
 197 For example, since Linux 2.6.14, ARM has the following system call:
 198 .PP
 199 .in +4n
 200 .nf
 201 .BI "long arm_fadvise64_64(int " fd ", int " advice ,
 202 .BI "                      loff_t " offset ", loff_t " len );
 203 .fi
 204 .in
 205 .PP
 206 These architecture-specific details are generally
 207 hidden from applications by the glibc
 208 .BR posix_fadvise ()
 209 wrapper function,
 210 which invokes the appropriate architecture-specific system call.
 211 .SH BUGS
 212 In kernels before 2.6.6, if
 213 .I len
 214 was specified as 0, then this was interpreted literally as "zero bytes",
 215 rather than as meaning "all bytes through to the end of the file".
 216 .SH SEE ALSO
 217 .BR readahead (2),
 218 .BR sync_file_range (2),
 219 .BR posix_fallocate (3),
 220 .BR posix_madvise (3)