man2/alloc_hugepages.2

   1 .\" Hey Emacs! This file is -*- nroff -*- source.
   2 .\"
   3 .\" Copyright 2003 Andries E. Brouwer (aeb@cwi.nl)
   4 .\"
   5 .\" Permission is granted to make and distribute verbatim copies of this
   6 .\" manual provided the copyright notice and this permission notice are
   7 .\" preserved on all copies.
   8 .\"
   9 .\" Permission is granted to copy and distribute modified versions of this
  10 .\" manual under the conditions for verbatim copying, provided that the
  11 .\" entire resulting derived work is distributed under the terms of a
  12 .\" permission notice identical to this one.
  13 .\"
  14 .\" Since the Linux kernel and libraries are constantly changing, this
  15 .\" manual page may be incorrect or out-of-date.  The author(s) assume no
  16 .\" responsibility for errors or omissions, or for damages resulting from
  17 .\" the use of the information contained herein.  The author(s) may not
  18 .\" have taken the same level of care in the production of this manual,
  19 .\" which is licensed free of charge, as they might when working
  20 .\" professionally.
  21 .\"
  22 .\" Formatted or processed versions of this manual, if unaccompanied by
  23 .\" the source, must acknowledge the copyright and authors of this work.
  24 .\"
  25 .TH ALLOC_HUGEPAGES 2 2007-05-31 "Linux" "Linux Programmer's Manual"
  26 .SH NAME
  27 alloc_hugepages, free_hugepages \- allocate or free huge pages
  28 .SH SYNOPSIS
  29 .nf
  30 .BI "void *alloc_hugepages(int " key ", void *" addr ", size_t " len ,
  31 .BI "                      int " prot ", int " flag );
  32 .\" asmlinkage unsigned long sys_alloc_hugepages(int key, unsigned long addr,
  33 .\" unsigned long len, int prot, int flag);
  34 .sp
  35 .BI "int free_hugepages(void *" addr );
  36 .\" asmlinkage int sys_free_hugepages(unsigned long addr);
  37 .fi
  38 .SH DESCRIPTION
  39 The system calls
  40 .BR alloc_hugepages ()
  41 and
  42 .BR free_hugepages ()
  43 were introduced in Linux 2.5.36 and removed again in 2.5.54.
  44 They existed only on i386 and ia64 (when built with CONFIG_HUGETLB_PAGE).
  45 In Linux 2.4.20 the syscall numbers exist,
  46 but the calls fail with the error
  47 .BR ENOSYS .
  48 .LP
  49 On i386 the memory management hardware knows about ordinary pages (4 KiB)
  50 and huge pages (2 or 4 MiB).
  51 Similarly ia64 knows about huge pages of
  52 several sizes.
  53 These system calls serve to map huge pages into the
  54 process's memory or to free them again.
  55 Huge pages are locked into memory, and are not swapped.
  56 .LP
  57 The
  58 .I key
  59 parameter is an identifier.
  60 When zero the pages are private, and
  61 not inherited by children.
  62 When positive the pages are shared with other applications using the same
  63 .IR key ,
  64 and inherited by child processes.
  65 .LP
  66 The
  67 .I addr
  68 parameter of
  69 .BR free_hugepages ()
  70 tells which page is being freed: it was the return value of a
  71 call to
  72 .BR alloc_hugepages ().
  73 (The memory is first actually freed when all users have released it.)
  74 The
  75 .I addr
  76 parameter of
  77 .BR alloc_hugepages ()
  78 is a hint, that the kernel may or may not follow.
  79 Addresses must be properly aligned.
  80 .LP
  81 The
  82 .I len
  83 parameter is the length of the required segment.
  84 It must be a multiple of the huge page size.
  85 .LP
  86 The
  87 .I prot
  88 parameter specifies the memory protection of the segment.
  89 It is one of PROT_READ, PROT_WRITE, PROT_EXEC.
  90 .LP
  91 The
  92 .I flag
  93 parameter is ignored, unless
  94 .I key
  95 is positive.
  96 In that case, if
  97 .I flag
  98 is IPC_CREAT, then a new huge page segment is created when none
  99 with the given key existed.
 100 If this flag is not set, then ENOENT
 101 is returned when no segment with the given key exists.
 102 .IR
 103 .SH "RETURN VALUE"
 104 On success,
 105 .BR alloc_hugepages ()
 106 returns the allocated virtual address, and
 107 .BR free_hugepages ()
 108 returns zero.
 109 On error, \-1 is returned, and
 110 .I errno
 111 is set appropriately.
 112 .SH ERRORS
 113 .TP
 114 .B ENOSYS
 115 The system call is not supported on this kernel.
 116 .SH FILES
 117 .I /proc/sys/vm/nr_hugepages
 118 Number of configured hugetlb pages.
 119 This can be read and written.
 120 .LP
 121 .I /proc/meminfo
 122 Gives info on the number of configured hugetlb pages and on their size
 123 in the three variables HugePages_Total, HugePages_Free, Hugepagesize.
 124 .SH "CONFORMING TO"
 125 These calls are specific to Linux on Intel processors, and should not be
 126 used in programs intended to be portable.
 127 .SH NOTES
 128 These system calls are gone;
 129 they existed only in Linux 2.5.36 through to 2.5.54.
 130 Now the hugetlbfs filesystem can be used instead.
 131 Memory backed by huge pages (if the CPU supports them) is obtained by
 132 using
 133 .BR mmap (2)
 134 to map files in this virtual filesystem.
 135 .LP
 136 The maximal number of huge pages can be specified using the
 137 .B hugepages=
 138 boot parameter.
 139
 140 .\" requires CONFIG_HUGETLB_PAGE (under "Processor type and features")
 141 .\" and CONFIG_HUGETLBFS (under "Filesystems").
 142 .\" mount -t hugetlbfs hugetlbfs /huge
 143 .\" SHM_HUGETLB