OSDN Git Service

cbf2ff4fb61cc04e20150d1717534f75a3e794a7
[linuxjm/LDP_man-pages.git] / release / man5 / proc.5
1 .\" Copyright (C) 1994, 1995 by Daniel Quinlan (quinlan@yggdrasil.com)
2 .\" and Copyright (C) 2002-2008 Michael Kerrisk <mtk.manpages@gmail.com>
3 .\" with networking additions from Alan Cox (A.Cox@swansea.ac.uk)
4 .\" and scsi additions from Michael Neuffer (neuffer@mail.uni-mainz.de)
5 .\" and sysctl additions from Andries Brouwer (aeb@cwi.nl)
6 .\" and System V IPC (as well as various other) additions from
7 .\" Michael Kerrisk <mtk.manpages@gmail.com>
8 .\"
9 .\" %%%LICENSE_START(GPLv2+_DOC_FULL)
10 .\" This is free documentation; you can redistribute it and/or
11 .\" modify it under the terms of the GNU General Public License as
12 .\" published by the Free Software Foundation; either version 2 of
13 .\" the License, or (at your option) any later version.
14 .\"
15 .\" The GNU General Public License's references to "object code"
16 .\" and "executables" are to be interpreted as the output of any
17 .\" document formatting or typesetting system, including
18 .\" intermediate and printed output.
19 .\"
20 .\" This manual is distributed in the hope that it will be useful,
21 .\" but WITHOUT ANY WARRANTY; without even the implied warranty of
22 .\" MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
23 .\" GNU General Public License for more details.
24 .\"
25 .\" You should have received a copy of the GNU General Public
26 .\" License along with this manual; if not, see
27 .\" <http://www.gnu.org/licenses/>.
28 .\" %%%LICENSE_END
29 .\"
30 .\" Modified 1995-05-17 by faith@cs.unc.edu
31 .\" Minor changes by aeb and Marty Leisner (leisner@sdsp.mc.xerox.com).
32 .\" Modified 1996-04-13, 1996-07-22 by aeb@cwi.nl
33 .\" Modified 2001-12-16 by rwhron@earthlink.net
34 .\" Modified 2002-07-13 by jbelton@shaw.ca
35 .\" Modified 2002-07-22, 2003-05-27, 2004-04-06, 2004-05-25
36 .\"    by Michael Kerrisk <mtk.manpages@gmail.com>
37 .\" 2004-11-17, mtk -- updated notes on /proc/loadavg
38 .\" 2004-12-01, mtk, rtsig-max and rtsig-nr went away in 2.6.8
39 .\" 2004-12-14, mtk, updated 'statm', and fixed error in order of list
40 .\" 2005-05-12, mtk, updated 'stat'
41 .\" 2005-07-13, mtk, added /proc/sys/fs/mqueue/*
42 .\" 2005-09-16, mtk, Added /proc/sys/fs/suid_dumpable
43 .\" 2005-09-19, mtk, added /proc/zoneinfo
44 .\" 2005-03-01, mtk, moved /proc/sys/fs/mqueue/* material to mq_overview.7.
45 .\" 2008-06-05, mtk, Added /proc/[pid]/oom_score, /proc/[pid]/oom_adj,
46 .\"     /proc/[pid]/limits, /proc/[pid]/mountinfo, /proc/[pid]/mountstats,
47 .\"     and /proc/[pid]/fdinfo/*.
48 .\" 2008-06-19, mtk, Documented /proc/[pid]/status.
49 .\" 2008-07-15, mtk, added /proc/config.gz
50 .\"
51 .\" FIXME 2.6.13 seems to have /proc/vmcore implemented
52 .\" in the source code, but there is no option available under
53 .\" 'make xconfig'; eventually this should be fixed, and then info
54 .\" from the patch-2.6.13 and change log could be used to write an
55 .\" entry in this man page.
56 .\" Needs CONFIG_VMCORE
57 .\"
58 .\" FIXME cross check against Documentation/filesystems/proc.txt
59 .\" to see what information could be imported from that file
60 .\" into this file.
61 .\"
62 .\"*******************************************************************
63 .\"
64 .\" This file was generated with po4a. Translate the source file.
65 .\"
66 .\"*******************************************************************
67 .TH PROC 5 2013\-06\-27 Linux "Linux Programmer's Manual"
68 .SH 名前
69 proc \- プロセスの情報を含む擬似ファイルシステム
70 .SH 説明
71 \fIproc\fP ファイルシステムは擬似的なファイルシステムであり、 カーネル内のデータへのインターフェースとして使用される。 一般的には
72 \fI/proc\fP にマウントされる。 大部分のファイルは読み出し専用 (read\-only) であるが、 いくつかのファイルは書き込み可能であり、
73 そのファイルに書き込めばカーネルの内部変数を変更できる。
74 .LP
75 以下は \fI/proc\fP 階層の簡単なあらましである。
76 .PD 1
77 .TP 
78 \fI/proc/[pid]\fP
79 .\" FIXME Describe /proc/[pid]/attr and
80 .\"       /proc/[pid]/task/[tid]/attr
81 .\"       This is a directory
82 .\"       Added in ???
83 .\"       CONFIG_SECURITY
84 .\"
85 実行中のプロセスについてのサブディレクトリ。 サブディレクトリ名は (そのプロセスの) プロセス ID である。
86 各サブディレクトリは、以下の擬似ファイルとディレクトリを含む。
87 .TP 
88 \fI/proc/[number]/auxv\fP (カーネル 2.6.0\-test7 以降)
89 実行時にプロセスに渡された ELF インタプリタ情報が格納されている。 個々のエントリは、\fIunsigned long\fP 型の ID 1 個に
90 \fIunsigned long\fP 型の値 1 個が続くフォーマットである。 最後のエントリには 0 が 2 個入っている。
91 .TP 
92 \fI/proc/[pid]/cgroup\fP (Linux 2.6.24 以降)
93 .\"       Info in Documentation/cgroups/cgroups.txt
94 このファイルは、プロセスやタスクが所属するコントロールグループを示す。
95 cgroup の各階層についてエントリーが 1 つあり、
96 各エントリーは以下の形式のコロン区切りのフィールドで構成される。
97 .nf
98
99 \f(CW    5:cpuacct,cpu,cpuset:/daemons\fP
100 .fi
101 .IP
102 コロン区切りの各フィールドは、左から右の順で、以下の意味を持つ。
103 .RS 11
104 .IP 1. 3
105 階層 ID 番号
106 .IP 2.
107 その階層に関連付けたサブシステムの集合
108 .IP 3.
109 プロセスが所属する階層内のコントロールグループ
110 .RE
111 .IP
112 .\"
113 .\" FIXME Describe /proc/[pid]/clear_refs
114 .\"       Added in 2.6.22
115 .\"       "Clears page referenced bits shown in smaps output"
116 .\"       write-only
117 .\"       CONFIG_PROC_PAGE_MONITOR
118 このファイルが存在するのは、カーネルのコンフィギュレーション・オプショ
119 ン \fBCONFIG_CGROUPS\fP を有効にした場合だけである。
120 .TP 
121 \fI/proc/[pid]/cmdline\fP
122 .\" In 2.3.26, this also used to be true if the process was swapped out.
123 プロセスの完全なコマンド行を保持する。ただし、そのプロセスがゾンビプロセス
124 の場合は、このファイルは空となる。つまり、このファイルを読み出しても一文字
125 も返らない。このファイルではコマンドライン引き数が、
126 NULL バイト (\(aq\e0\(aq) で区切られた文字列として書かれており、
127 最後の文字列の後に NULL バイトが一つ置かれる。
128 .TP 
129 \fI/proc/[pid]/coredump_filter\fP (kernel 2.6.23 以降)
130 \fBcore\fP(5)  参照。
131 .TP 
132 \fI/proc/[pid]/cpuset\fP (kernel 2.6.12 以降)
133 .\" and/proc/[pid]/task/[tid]/cpuset
134 \fBcpuset\fP(7)  参照。
135 .TP 
136 \fI/proc/[pid]/cwd\fP
137 プロセスのカレントワーキングディレクトリへのシンボリックリンク。 例えば、プロセス 20 のカレントワーキングディレクトリを見つけるためには、
138 次のようにすればよい。
139
140 .in +4n
141 .nf
142 $\fB cd /proc/20/cwd; /bin/pwd\fP
143 .fi
144 .in
145
146 \fIpwd\fP コマンドはシェルの内部コマンドのことがよくあり、
147 うまく動作しないかもしれない。
148 \fBbash\fP(1) では \fIpwd\ \-P\fP を使ってもよい。
149
150 .\" The following was still true as at kernel 2.6.13
151 マルチスレッドプロセスでは、メインスレッドがすでに終了している場合、 このシンボリックリンクの内容は参照できない (スレッドの終了は通常
152 \fBpthread_exit\fP(3)  を呼び出しにより行われる)。
153 .TP 
154 \fI/proc/[pid]/environ\fP
155 このファイルはプロセスの環境変数を含んでいる。 各エントリは NULL バイト (\(aq\e0\(aq) で区切られていて、 末尾に NULL
156 バイトがあるかもしれない。 したがって、プロセス 1 の環境変数を表示するためには 次のようにすればよい。
157 .in +4n
158 .nf
159
160 $\fB strings /proc/1/environ\fP
161 .fi
162 .in
163 .TP 
164 \fI/proc/[pid]/exe\fP
165 .\" The following was still true as at kernel 2.6.13
166 Linux 2.2 以降では、このファイルはシンボリックリンクで、 実行可能コマンドの実際のパス名を格納している。
167 このシンボリックリンクは通常のように辿ることができる; これをオープンすると実行可能ファイルがオープンされる。 (コマンドラインで)
168 \fI/proc/[pid]/exe\fP と入力すると、プロセス番号 [pid] で実行されている 実行可能ファイルをもう一つ実行することができる。
169 マルチスレッドプロセスでは、メインスレッドがすでに終了している場合、 このシンボリックリンクの内容は参照できない (スレッドの終了は通常
170 \fBpthread_exit\fP(3)  を呼び出しにより行われる)。
171
172 Linux 2.0 以前では、 \fI/proc/[pid]/exe\fP は実行されたバイナリへのポインタで、シンボリックリンクのように見える。 Linux
173 2.0 以前では、このファイルに対して \fBreadlink\fP(2)  を実行すると、次のフォーマットの文字列が返る。
174
175     [デバイス番号]:iノード番号
176
177 たとえば、[0301]:1502 はメジャーデバイス番号 03 (IDE, MFM などのドライブ)  マイナーデバイス番号 01
178 (最初のドライブの最初のパーティション) の デバイス上の iノード番号 1502 である。
179
180 \fI\-inum\fP オプションをつけて \fBfind\fP(1) を使うと、
181 このファイルの所在を探すことができる。
182 .TP 
183 \fI/proc/[pid]/fd/\fP
184 プロセスがオープンしたファイル各々に対するエントリを含むサブディレクトリ。 ファイルディスクリプタがファイル名で、
185 実際のファイルへのシンボリックリンクになっている。 したがって 0 は標準入力、1 は標準出力、2 は標準エラー出力、などとなる。
186
187 For file descriptors for pipes and sockets, the entries will be symbolic
188 links whose content is the file type with the inode.  A \fBreadlink\fP(2)  call
189 on this file returns a string in the format:
190
191     type:[inode]
192
193 For example, \fIsocket:[2248868]\fP will be a socket and its inode is 2248868.
194 For sockets, that inode can be used to find more information in one of the
195 files under \fI/proc/net/\fP.
196
197 For file descriptors that have no corresponding inode (e.g., file
198 descriptors produced by \fBepoll_create\fP(2), \fBeventfd\fP(2),
199 \fBinotify_init\fP(2), \fBsignalfd\fP(2), and \fBtimerfd\fP(2)), the entry will be a
200 symbolic link with contents of the form
201
202     anon_inode:<file\-type>
203
204 In some cases, the \fIfile\-type\fP is surrounded by square brackets.
205
206 For example, an epoll file descriptor will have a symbolic link whose
207 content is the string \fIanon_inode:[eventpoll]\fP.
208
209 .\"The following was still true as at kernel 2.6.13
210 マルチスレッドプロセスでは、メインスレッドがすでに終了している場合、 このディレクトリの内容は参照できない (スレッドの終了は通常
211 \fBpthread_exit\fP(3)  を呼び出しにより行われる)。
212
213 コマンドライン引き数としてファイル名を受け取るが、引き数が 渡されなかった場合
214 に標準入力から入力を受け取らないようなプログラムや、 コマンドライン引き数とし
215 て書き込みファイルを受け取るが、引き数が 渡されなかった場合に標準出力に出力を
216 行わないようなプログラムであっても、 \fI/proc/[pid]/fd\fP を使うことで標準入力や
217 標準出力を使用できるようになる。 例えば、 \fI\-i\fP を入力ファイルを指定するフラ
218 グ、 \fI\-o\fP を出力ファイルを指定するフラグと仮定すると、
219 .in +4n
220 .nf
221
222 $\fB foobar \-i /proc/self/fd/0 \-o /proc/self/fd/1 ...\fP
223 .fi
224 .in
225
226 .\" The following is not true in my tests (MTK):
227 .\" Note that this will not work for
228 .\" programs that seek on their files, as the files in the fd directory
229 .\" are not seekable.
230 を実行することにより、フィルタとして動作させることができる。
231
232 \fI/proc/self/fd/N\fP は、ある種の UNIX や UNIX 風のシステムにある \fI/dev/fd/N\fP
233 とだいたい同じである。 事実 Linux のたいていの MAKEDEV スクリプトは、
234 \fI/dev/fd\fP を \fI/proc/self/fd\fP へのシンボリックリンクにしている。
235
236 ほとんどのシステムでは、シンボリックリンク \fI/dev/stdin\fP, \fI/dev/stdout\fP, \fI/dev/stderr\fP
237 が提供されており、それぞれ \fI/proc/self/fd\fP 内のファイル \fI0\fP, \fI1\fP, \fI2\fP にリンクされている。
238 したがって、上述のサンプルコマンドは次のようにも書くことができる。
239 .in +4n
240 .nf
241
242 $\fB foobar \-i /dev/stdin \-o /dev/stdout ...\fP
243 .fi
244 .in
245 .\" FIXME Describe /proc/[pid]/loginuid
246 .\"       Added in 2.6.11; updating requires CAP_AUDIT_CONTROL
247 .\"       CONFIG_AUDITSYSCALL
248 .TP 
249 \fI/proc/[pid]/fdinfo/\fP (kernel 2.6.22 以降)
250 このサブディレクトリには、そのプロセスがオープンしているファイル毎の エントリが入っており、ファイルディスクリプタがファイル名となっている。
251 各ファイルの内容を読み出すことで、対応するファイルディスクリプタに関する 情報を得ることができる。以下に例を示す。
252 .in +4n
253 .nf
254
255 $\fB cat /proc/12015/fdinfo/4\fP
256 pos:    1000
257 flags:  01002002
258 .fi
259 .in
260
261 \fIpos\fP フィールドは 10 進数で、現在のファイルオフセットを示す。 \fIflags\fP フィールドは 8 進数で、
262 ファイルのアクセスモードとファイル状態フラグを示す (\fBopen\fP(2)  参照)。
263
264 .\" FIXME document /proc/[pid]/io
265 .\" .TP
266 .\" .IR /proc/[pid]/io " (since kernel 2.6.20)"
267 このディレクトリのファイルは、プロセスの所有者だけが読み出すことができる。
268 .TP 
269 \fI/proc/[pid]/limits\fP (kernel 2.6.24 以降)
270 .\" commit 3036e7b490bf7878c6dae952eec5fb87b1106589
271 This file displays the soft limit, hard limit, and units of measurement for
272 each of the process's resource limits (see \fBgetrlimit\fP(2)).  Up to and
273 including Linux 2.6.35, this file is protected to allow reading only by the
274 real UID of the process.  Since Linux 2.6.36, this file is readable by all
275 users on the system.
276 .TP 
277 \fI/proc/[pid]/map_files/\fP (kernel 3.3 以降)
278 .\" commit 640708a2cff7f81e246243b0073c66e6ece7e53e
279 This subdirectory contains entries corresponding to memory\-mapped files (see
280 \fBmmap\fP(2)).  Entries are named by memory region start and end address pair
281 (expressed as hexadecimal numbers), and are symbolic links to the mapped
282 files themselves.  Here is an example, with the output wrapped and
283 reformatted to fit on an 80\-column display:
284 .in +4n
285 .nf
286
287 $\fB ls \-l /proc/self/map_files/\fP
288 lr\-\-\-\-\-\-\-\-. 1 root root 64 Apr 16 21:31
289             3252e00000\-3252e20000 \-> /usr/lib64/ld\-2.15.so
290 \&...
291 .fi
292 .in
293
294 Although these entries are present for memory regions that were mapped with
295 the \fBMAP_FILE\fP flag, the way anonymous shared memory (regions created with
296 the \fBMAP_ANON | MAP_SHARED\fP flags)  is implemented in Linux means that such
297 regions also appear on this directory.  Here is an example where the target
298 file is the deleted \fI/dev/zero\fP one:
299 .in +4n
300 .nf
301
302
303 lrw\-\-\-\-\-\-\-. 1 root root 64 Apr 16 21:33
304             7fc075d2f000\-7fc075e6f000 \-> /dev/zero (deleted)
305 .fi
306 .in
307
308 This directory appears only if the \fBCONFIG_CHECKPOINT_RESTORE\fP kernel
309 configuration option is enabled.
310 .TP 
311 \fI/proc/[pid]/maps\fP
312 A file containing the currently mapped memory regions and their access
313 permissions.  See \fBmmap\fP(2)  for some further information about memory
314 mappings.
315
316 ファイルのフォーマットは以下のとおり:
317
318 .in -7n
319 .nf
320 \fIaddress           perms offset  dev   inode       pathname\fP
321 00400000\-00452000 r\-xp 00000000 08:02 173521      /usr/bin/dbus\-daemon
322 00651000\-00652000 r\-\-p 00051000 08:02 173521      /usr/bin/dbus\-daemon
323 00652000\-00655000 rw\-p 00052000 08:02 173521      /usr/bin/dbus\-daemon
324 00e03000\-00e24000 rw\-p 00000000 00:00 0           [heap]
325 00e24000\-011f7000 rw\-p 00000000 00:00 0           [heap]
326 \&...
327 35b1800000\-35b1820000 r\-xp 00000000 08:02 135522  /usr/lib64/ld\-2.15.so
328 35b1a1f000\-35b1a20000 r\-\-p 0001f000 08:02 135522  /usr/lib64/ld\-2.15.so
329 35b1a20000\-35b1a21000 rw\-p 00020000 08:02 135522  /usr/lib64/ld\-2.15.so
330 35b1a21000\-35b1a22000 rw\-p 00000000 00:00 0
331 35b1c00000\-35b1dac000 r\-xp 00000000 08:02 135870  /usr/lib64/libc\-2.15.so
332 35b1dac000\-35b1fac000 \-\-\-p 001ac000 08:02 135870  /usr/lib64/libc\-2.15.so
333 35b1fac000\-35b1fb0000 r\-\-p 001ac000 08:02 135870  /usr/lib64/libc\-2.15.so
334 35b1fb0000\-35b1fb2000 rw\-p 001b0000 08:02 135870  /usr/lib64/libc\-2.15.so
335 \&...
336 f2c6ff8c000\-7f2c7078c000 rw\-p 00000000 00:00 0    [stack:986]
337 \&...
338 7fffb2c0d000\-7fffb2c2e000 rw\-p 00000000 00:00 0   [stack]
339 7fffb2d48000\-7fffb2d49000 r\-xp 00000000 00:00 0   [vdso]
340 .fi
341 .in
342
343 The \fIaddress\fP field is the address space in the process that the mapping
344 occupies.  The \fIperms\fP field is a set of permissions:
345
346 .nf
347 .in +5
348 r = read
349 w = write
350 x = execute
351 s = shared
352 p = private (copy on write)
353 .fi
354 .in
355
356 The \fIoffset\fP field is the offset into the file/whatever; \fIdev\fP is the
357 device (major:minor); \fIinode\fP is the inode on that device.  0 indicates
358 that no inode is associated with the memory region, as would be the case
359 with BSS (uninitialized data).
360
361 The \fIpathname\fP field will usually be the file that is backing the mapping.
362 For ELF files, you can easily coordinate with the \fIoffset\fP field by looking
363 at the Offset field in the ELF program headers (\fIreadelf\ \-l\fP).
364
365 There are additional helpful pseudo\-paths:
366 .RS 12
367 .TP 
368 \fI[stack]\fP
369 The initial process's (also known as the main thread's) stack.
370 .TP 
371 \fI[stack:<tid>]\fP (Linux 3.4 以降)
372 .\" commit b76437579d1344b612cf1851ae610c636cec7db0
373 A thread's stack (where the \fI<tid>\fP is a thread ID).  It
374 corresponds to the \fI/proc/[pid]/task/[tid]/\fP path.
375 .TP 
376 \fI[vdso]\fP
377 The virtual dynamically linked shared object.
378 .TP 
379 \fI[heap]\fP
380 The process's heap.
381 .in
382 .fi
383 .RE
384 .IP
385 If the \fIpathname\fP field is blank, this is an anonymous mapping as obtained
386 via the \fBmmap\fP(2)  function.  There is no easy way to coordinate this back
387 to a process's source, short of running it through \fBgdb\fP(1), \fBstrace\fP(1),
388 or similar.
389
390 Linux 2.0 ではパス名を書いたフィールドがない。
391 .TP 
392 \fI/proc/[pid]/mem\fP
393 このファイルは、 \fBopen\fP(2), \fBread\fP(2), \fBlseek\fP(2)
394 を通して、プロセスのメモリのページにアクセスするために使われる。
395 .TP 
396 \fI/proc/[pid]/mountinfo\fP (Linux 2.6.26 以降)
397 .\" This info adapted from Documentation/filesystems/proc.txt
398 このファイルには、マウントポイントについての情報が入っている。 以下のような形式の行から構成される。
399 .nf
400
401 \f(CW36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 \- ext3 /dev/root rw,errors=continue
402 (1)(2)(3)   (4)   (5)      (6)      (7)   (8) (9)   (10)         (11)\fP
403 .fi
404 .IP
405 括弧付きの数字は、以下の説明のためのものである。
406 .RS 7
407 .TP  5
408 (1)
409 マウント ID: マウントの一意な識別子 (\fBumount\fP(2)  の後は再利用されるかもしれない)。
410 .TP 
411 (2)
412 parent ID: 親マウントの ID (マウントツリーの最上位の場合は自分自身の ID となる)。
413 .TP 
414 (3)
415 major:minor: ファイルシステム上のファイルの \fIst_dev\fP の値 (\fBstat\fP(2)  参照)。
416 .TP 
417 (4)
418 ルート: そのファイルシステム内のマウントのルート。
419 .TP 
420 (5)
421 マウントポイント: マウントポイントのそのプロセスのルートからの相対パス。
422 .TP 
423 (6)
424 マウントオプション: 各マウントのオプション。
425 .TP 
426 (7)
427 オプションフィールド: "tag[:value]" 形式のフィールドが 0 個以上並ぶ。
428 .TP 
429 (8)
430 セパレータ: オプションフィールドの終わりを示す。
431 .TP 
432 (9)
433 ファイルシステム種別: ファイルシステムの名前。 "type[.subtype]" という形式となる。
434 .TP 
435 (10)
436 マウント元: ファイルシステム固有の情報。ない場合は "none" となる。
437 .TP 
438 (11)
439 super options: スーパーブロック単位のオプション。
440 .RE
441 .IP
442 解釈する側は認識できないオプションフィールドは全て無視すべきである。 現在のところ、オプションフィールドとしては以下のようなものがある。
443 .RS 12
444 .TP  18
445 shared:X
446 マウントはピアグループ (peer group) X で共有されている。
447 .TP 
448 master:X
449 マウントはピアグループ (peer group) X のスレーブである。
450 .TP 
451 propagate_from:X
452 マウントはスレーブであり、ピアグループ X (*) から mount propagation を受信する。
453 .TP 
454 unbindable
455 マウントは unbind できない。
456 .RE
457 .IP
458 (*) X は、プロセスの root で直近の dominant peer group である。 X がマウントの直接のマスターである場合や、 同じ
459 root に dominant peer group がない場合は、 "master:X" フィールドだけが存在し、
460 "propagate_from:X" フィールドは存在しない。
461
462 mount propagation の詳細については、 Linux カーネルソースツリー内の
463 \fIDocumentation/filesystems/sharedsubtree.txt\fP を参照。
464 .TP 
465 \fI/proc/[pid]/mounts\fP (Linux 2.4.19 以降)
466 そのプロセスのマウント名前空間に現在マウントされている 全ファイルシステムのリスト。 このファイルのフォーマットは \fBfstab\fP(5)
467 に載っている。 カーネル 2.6.15 以降では、このファイルを監視することができる (pollable)。
468 このファイルを読み出し用にオープンした後で、このファイルに変更があると (つまりファイルシステムのマウントやアンマウントがあると)、
469 \fBselect\fP(2)  ではそのファイルディスクリプタは読み出し可能となり、 \fBpoll\fP(2)  と \fBepoll_wait\fP(2)
470 ではそのファイルはエラー状態として通知される。
471 .TP 
472 \fI/proc/[pid]/mountstats\fP (Linux 2.6.17 以降)
473 このファイルを通じて、そのプロセスの名前空間内のマウントポイントに関する 各種情報 (統計、設定情報) を参照できる。
474 ファイルの各行は以下のフォーマットである。
475 .nf
476
477 device /dev/sda7 mounted on /home with fstype ext3 [statistics]
478 (       1      )            ( 2 )             (3 ) (4)
479 .fi
480 .IP
481 各行のフィールドは以下の通りである。
482 .RS 7
483 .TP  5
484 (1)
485 マウントされているデバイス名 (対応するデバイスがない場合は "nodevice" となる)。
486 .TP 
487 (2)
488 マウントポイントのファイルシステムツリーにおけるパス名。
489 .TP 
490 (3)
491 ファイルシステム種別。
492 .TP 
493 (4)
494 追加の統計や設定情報。 現在のところ (Linux 2.6.26 時点では)、 このフィールドで情報が提供されているのは NFS
495 ファイルシステムだけである。
496 .RE
497 .IP
498 このファイルはプロセスの所有者だけが読み出すことができる。
499 .TP 
500 \fI/proc/[pid]/ns/\fP (Linux 3.0 以降)
501 .\" See commit 6b4e306aa3dc94a0545eb9279475b1ab6209a31f
502 このサブディレクトリには、名前空間毎に 1 エントリが置かれる。
503 各エントリは \fBsetns\fP(2) による操作をサポートしている。
504 名前空間に関する情報は \fBclone\fP(2) を参照。
505 .TP 
506 \fI/proc/[pid]/ns/ipc\fP (Linux 3.0 以降)
507 このファイルをファイルシステムのどこか他の場所に bind mount することで (\fBmount\fP(2)
508 参照)、現在この名前空間にいる全てのプロセスが終了したとしても、 \fIpid\fP で指定されたプロセスの IPC 名前空間は有効な状態で保たれる。
509
510 このファイルをオープンすると、 \fIpid\fP で指定されたプロセスの IPC
511 名前空間のファイルハンドルが返される。このファイルディスクリプタがオープンされている限り、この名前空間にいる全てのプロセスが終了したとしても、この
512 IPC 名前空間は有効なままとなる。このファイルディスクリプタは \fBsetns\fP(2) に渡すことができる。
513 .TP 
514 \fI/proc/[pid]/ns/net\fP (Linux 3.0 以降)
515 このファイルをファイルシステムのどこか他の場所に bind mount することで (\fBmount\fP(2)
516 参照)、現在この名前空間にいる全てのプロセスが終了したとしても、 \fIpid\fP で指定されたプロセスのネットワーク名前空間は有効な状態で保たれる。
517
518 このファイルをオープンすると、 \fIpid\fP
519 で指定されたプロセスのネットワーク名前空間のファイルハンドルが返される。このファイルディスクリプタがオープンされている限り、この名前空間にいる全てのプロセスが終了したとしても、このネットワーク名前空間は有効なままとなる。このファイルディスクリプタは
520 \fBsetns\fP(2) に渡すことができる。
521 .TP 
522 \fI/proc/[pid]/ns/uts\fP (Linux 3.0 以降)
523 このファイルをファイルシステムのどこか他の場所に bind mount することで (\fBmount\fP(2)
524 参照)、現在この名前空間にいる全てのプロセスが終了したとしても、 \fIpid\fP で指定されたプロセスの UTS 名前空間は有効な状態で保たれる。
525
526 このファイルをオープンすると、 \fIpid\fP で指定されたプロセスの UTS
527 名前空間のファイルハンドルが返される。このファイルディスクリプタがオープンされている限り、この名前空間にいる全てのプロセスが終了したとしても、この
528 UTS 名前空間は有効なままとなる。このファイルディスクリプタは \fBsetns\fP(2) に渡すことができる。
529 .TP 
530 \fI/proc/[pid]/numa_maps\fP (Linux 2.6.14 以降)
531 \fBnuma\fP(7)  を参照。
532 .TP 
533 \fI/proc/[pid]/oom_adj\fP (Linux 2.6.11 以降)
534 このファイルは、メモリ不足 (OOM) の状況下でどのプロセスを殺すべきかを選択す
535 るのに使用されるスコアを調整するのに使用される。カーネルは、プロセスの
536 \fIoom_score\fP 値のビットシフト操作に、この値を使用する。この値として有効な値
537 は \-16 から +15 までと、特別な意味を持つ \-17 である。 \-17 はそのプロセス
538 に対する OOM\-killing を完全に無効にすることを意味する。正の値ほど、そのプロ
539 セスが OOM\-killer により殺される可能性が高くなり、負の値ほど可能性が低くなる。
540 .IP
541 このファイルのデフォルト値は 0 である。 新しいプロセスは親プロセスの \fIoom_adj\fP の設定を継承する。
542 このファイルを変更するためには、プロセスは特権 (\fBCAP_SYS_RESOURCE\fP) を持っていなければならない。
543 .IP
544 Since Linux 2.6.36, use of this file is deprecated in favor of
545 \fI/proc/[pid]/oom_score_adj\fP.
546 .TP 
547 \fI/proc/[pid]/oom_score\fP (Linux 2.6.11 以降)
548 .\" See mm/oom_kill.c::badness() in the 2.6.25 sources
549 .\" See mm/oom_kill.c::badness() in the 2.6.25 sources
550 このファイルは、OOM\-killer のプロセス選択用として、カーネルが このプロセス
551 に対して与えた現在のスコアを表示する。 高いスコアは、そのプロセスが
552 OOM\-killer により選択される 可能性が高いことを意味する。 このスコアの基本は
553 そのプロセスが使用しているメモリ量であり、 以下の要因により加算 (+) 減算 (\-)
554 が行われる。
555 .RS
556 .IP * 2
557 そのプロセスが多くの子プロセスを \fBfork\fP(2)  を使って作成しているか (+)。
558 .IP *
559 そのプロセスが長時間実行されて来たか、
560 もしくは 多くの CPU 時間を使用しているか (\-)。
561 .IP *
562 そのプロセスが低い nice 値 (> 0) を持っているか (+)。
563 .IP *
564 .\" More precisely, if it has CAP_SYS_ADMIN or CAP_SYS_RESOURCE
565 そのプロセスが特権を持っているか (\-)。
566 .IP *
567 .\" More precisely, if it has CAP_SYS_RAWIO
568 そのプロセスが direct hardware access を行っているか (\-)。
569 .RE
570 .IP
571 \fIoom_score\fP は、そのプロセスの \fIoom_score_adj\fP や \fIoom_adj\fP 設定で規定される調整にも影響を与える。
572 .TP 
573 \fI/proc/[pid]/oom_score_adj\fP (Linux 2.6.36 以降)
574 .\" Text taken from 3.7 Documentation/filesystems/proc.txt
575 This file can be used to adjust the badness heuristic used to select which
576 process gets killed in out\-of\-memory conditions.
577
578 The badness heuristic assigns a value to each candidate task ranging from 0
579 (never kill) to 1000 (always kill) to determine which process is targeted.
580 The units are roughly a proportion along that range of allowed memory the
581 process may allocate from, based on an estimation of its current memory and
582 swap use.  For example, if a task is using all allowed memory, its badness
583 score will be 1000.  If it is using half of its allowed memory, its score
584 will be 500.
585
586 There is an additional factor included in the badness score: root processes
587 are given 3% extra memory over other tasks.
588
589 The amount of "allowed" memory depends on the context in which the
590 OOM\-killer was called.  If it is due to the memory assigned to the
591 allocating task's cpuset being exhausted, the allowed memory represents the
592 set of mems assigned to that cpuset (see \fBcpuset\fP(7)).  If it is due to a
593 mempolicy's node(s) being exhausted, the allowed memory represents the set
594 of mempolicy nodes.  If it is due to a memory limit (or swap limit) being
595 reached, the allowed memory is that configured limit.  Finally, if it is due
596 to the entire system being out of memory, the allowed memory represents all
597 allocatable resources.
598
599 The value of \fIoom_score_adj\fP is added to the badness score before it is
600 used to determine which task to kill.  Acceptable values range from \-1000
601 (OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX).  This allows user space to
602 control the preference for OOM\-killing, ranging from always preferring a
603 certain task or completely disabling it from OOM\-killing.  The lowest
604 possible value, \-1000, is equivalent to disabling OOM\-killing entirely for
605 that task, since it will always report a badness score of 0.
606
607 Consequently, it is very simple for user space to define the amount of
608 memory to consider for each task.  Setting a \fIoom_score_adj\fP value of +500,
609 for example, is roughly equivalent to allowing the remainder of tasks
610 sharing the same system, cpuset, mempolicy, or memory controller resources
611 to use at least 50% more memory.  A value of \-500, on the other hand, would
612 be roughly equivalent to discounting 50% of the task's allowed memory from
613 being considered as scoring against the task.
614
615 For backward compatibility with previous kernels, \fI/proc/[pid]/oom_adj\fP can
616 still be used to tune the badness score.  Its value is scaled linearly with
617 \fIoom_score_adj\fP.
618
619 .\" FIXME Describe /proc/[pid]/pagemap
620 .\"       Added in 2.6.25
621 .\"       CONFIG_PROC_PAGE_MONITOR
622 Writing to \fI/proc/[pid]/oom_score_adj\fP or \fI/proc/[pid]/oom_adj\fP will
623 change the other with its scaled value.
624 .TP 
625 \fI/proc/[pid]/root\fP
626 UNIX と Linux では、 ファイルシステムのルート (/) をプロセスごとに別々に
627 できる。これはシステムコール \fBchroot\fP(2) によって設定する。 このファイルは
628 プロセスのルートディレクトリを指すシンボリックリンクで、 exe や fd/* など
629 と同じような動作をする。
630
631 .\" The following was still true as at kernel 2.6.13
632 .\" FIXME Describe /proc/[pid]/seccomp
633 .\"       Added in 2.6.12
634 .\" FIXME Describe /proc/[pid]/sessionid
635 .\"       Added in 2.6.25; read-only; only readable by real UID
636 .\"       CONFIG_AUDITSYSCALL
637 .\" FIXME Describe /proc/[pid]/sched
638 .\"       Added in 2.6.23
639 .\"       CONFIG_SCHED_DEBUG, and additional fields if CONFIG_SCHEDSTATS
640 .\"       Displays various scheduling parameters
641 .\"       This file can be written, to reset stats
642 .\" FIXME Describe /proc/[pid]/schedstats and
643 .\"       /proc/[pid]/task/[tid]/schedstats
644 .\"       Added in 2.6.9
645 .\"       CONFIG_SCHEDSTATS
646 マルチスレッドプロセスでは、メインスレッドがすでに終了している場合、 このシンボリックリンクの内容は参照できない (スレッドの終了は通常
647 \fBpthread_exit\fP(3)  を呼び出しにより行われる)。
648 .TP 
649 \fI/proc/[pid]/smaps\fP (Linux 2.6.14 以降)
650 .\"       CONFIG_PROC_PAGE_MONITOR
651 このファイルは、そのプロセスの各マッピングのメモリ消費量を表示する。 マッピングのそれぞれについて、以下のような内容が表示される。
652 .in +4n
653 .nf
654
655 08048000\-080bc000 r\-xp 00000000 03:02 13130      /bin/bash
656 Size:               464 kB
657 Rss:                424 kB
658 Shared_Clean:       424 kB
659 Shared_Dirty:         0 kB
660 Private_Clean:        0 kB
661 Private_Dirty:        0 kB
662
663 .fi
664 .in
665 最初の行には、 \fI/proc/[pid]/maps\fP で表示されるマッピングと同じ情報が表示される。 残りの行には、マッピングのサイズ、現在 RAM
666 上に存在するマッピングの量、 マッピング内の共有ページのうちクリーンなページ数、ダーティなページ数、
667 マッピング内のプライベートページのうちクリーンなページ数、 ダーティなページ数、を示す。
668
669 このファイルが存在するのは、カーネルのコンフィギュレーション・オプション \fBCONFIG_MMU\fP を有効にした場合だけである。
670 .TP 
671 \fI/proc/[pid]/stat\fP
672 プロセスの状態についての情報。 これは \fBps\fP(1)  で使われ、 \fI/usr/src/linux/fs/proc/array.c\fP
673 で定義されている。
674
675 各フィールドを順番に、 \fBscanf\fP(3)  のフォーマット指定子付きで以下に示す。
676 .RS
677 .TP  12
678 \fIpid\fP %d
679 (1) プロセス ID。
680 .TP 
681 \fIcomm\fP %s
682 (2) 括弧でくくられた実行形式のファイル名。実行形式がスワップアウトされているかどうかによらず、見ることができる。
683 .TP 
684 \fIstate\fP %c
685 (3) "RSDZTW" のどれか 1 文字。 R は実行中 (running)、 S は割り込み可能な休眠状態 (sleeping in an
686 interruptible wait)、 D は割り込み不可能なディスクスリープの待機状態 (waiting in uninterruptible
687 disk sleep)、 Z はゾンビ状態 (zombie)、 T はトレースされている (traced) か (シグナルにより) 停止している状態
688 (stopped)、 W はページング中 (paging) を表している。
689 .TP 
690 \fIppid\fP %d
691 (4) 親プロセスの PID。
692 .TP 
693 \fIpgrp\fP %d
694 (5) プロセスのプロセスグループ ID。
695 .TP 
696 \fIsession\fP %d
697 (6) プロセスのセッション ID。
698 .TP 
699 \fItty_nr\fP %d
700 (7) プロセスの制御端末 (マイナー・デバイス番号はビット 31〜20 と 7〜0 にまたがって格納され、 メジャー・デバイス番号はビット 15〜8
701 に格納される)。
702 .TP 
703 \fItpgid\fP %d
704 .\" This field and following, up to and including wchan added 0.99.1
705 (8) プロセスの制御端末のフォアグランド・プロセス・グループの ID。
706 .TP 
707 \fIflags\fP %u (Linux 2.6.22 より前は %lu)
708 (9) プロセスのカーネルフラグワード。 ビットの意味は、 \fI<linux/sched.h>\fP で定義されている PF_*
709 を参照すること。 詳細はカーネルのバージョンに依存する。
710 .TP 
711 \fIminflt\fP %lu
712 (10) プロセスが引き起こしたマイナーフォールト (minor fault、ディスクから メモリページへのロードを必要としないフォールト) の回数。
713 .TP 
714 \fIcminflt\fP %lu
715 (11) (そのプロセスが終了を待っている) 子プロセスが引き起こしたマイナーフォールトの回数。
716 .TP 
717 \fImajflt\fP %lu
718 (12) プロセスが引き起こしたメジャーフォールト (major fault、ディスクからメモリページへのロードを必要とするフォールト) の回数。
719 .TP 
720 \fIcmajflt\fP %lu
721 (13) (そのプロセスが終了を待っている) 子プロセスが引き起こしたメジャーフォールトの回数。
722 .TP 
723 \fIutime\fP %lu
724 (14) このプロセスがユーザーモードでスケジューリングされた時間の合計。 clock tick 単位で計測される
725 (\fIsysconf(_SC_CLK_TCK)\fP で割った値が表示される)。 この値にはゲスト時間 \fIguest_time\fP (仮想 CPU
726 の実行に消費された時間)
727 も含まれる。これは、ゲスト時間のフィールドを認識しないアプリケーションにおいて、ゲスト時間分を計算に入れ損ねないようにするためである。
728 .TP 
729 \fIstime\fP %lu
730 (15) プロセスのカーネルモードでの実行時間 (単位 jiffies)。 このプロセスがカーネルモードでスケジューリングされた時間の合計。 clock
731 tick 単位で計測される (\fIsysconf(_SC_CLK_TCK)\fP で割った値が表示される)。
732 .TP 
733 \fIcutime\fP %ld
734 (16) このプロセスの子プロセスで、終了待ち (waited\-for) のプロセスが、 ユーザモードでスケジューリングされた時間の合計。 clock
735 tick 単位で計測される (\fIsysconf(_SC_CLK_TCK)\fP で割った値が表示される)。 (\fBtimes\fP(2)  も参照すること。)
736 この値にはゲスト時間 \fIcguest_time\fP (仮想 CPU を実行するのに消費した時間、下記参照) も含まれる。
737 .TP 
738 \fIcstime\fP %ld
739 (17) このプロセスの子プロセスで、終了待ち (waited\-for) のプロセスが、カーネルモードでスケジューリングされた時間の合計。 clock
740 tick 単位で計測される (\fIsysconf(_SC_CLK_TCK)\fP で割った値が表示される)。
741 .TP 
742 \fIpriority\fP %ld
743 (18) (Linux 2.6 の場合の説明) リアルタイム・スケジューリングポリシー (下記の
744 \fIpolicy ;\fP \fBsched_setscheduler\fP(2) 参照) で動作しているプロセスでは、 この
745 値はスケジューリング優先度を反転した値 (スケジューリング優先度を マイナスにし
746 た値) となる。値は \-2 から \-100 までの範囲の数値で、 それぞれリアルタイム優先
747 度の 1 から 99 に対応する。 リアルタイム以外のスケジューリングポリシーで動作し
748 ているプロセスでは、 この値はカーネル内で管理されている nice 値そのもの
749 (\fBsetpriority\fP(2)) となる。 カーネルは nice 値を 0 (高) から 39 (低) の範囲
750 の値として保持しており、 それぞれユーザに見える nice 値の \-20 から 19 に対応
751 する。
752
753 .\" And back in kernel 1.2 days things were different again.
754 Linux 2.6 より前では、このプロセスに割り当てられたスケジューリング 重みを変換した値が表示されていた。
755 .TP 
756 \fInice\fP %ld
757 .\" Back in kernel 1.2 days things were different.
758 .\" .TP
759 .\" \fIcounter\fP %ld
760 .\" The current maximum size in jiffies of the process's next timeslice,
761 .\" or what is currently left of its current timeslice, if it is the
762 .\" currently running process.
763 .\" .TP
764 .\" \fItimeout\fP %u
765 .\" The time in jiffies of the process's next timeout.
766 .\" timeout was removed sometime around 2.1/2.2
767 (19) nice 値 (\fBsetpriority\fP(2) 参照)。 19 (最低優先) から \-20 (最高優先)
768 の範囲の値である。
769 .TP 
770 \fInum_threads\fP %ld
771 (20) このプロセスのスレッド数 (Linux 2.6 以降)。 カーネル 2.6 より前では、このフィールドは削除されたフィールドの 場所埋めとして
772 0 にハードコードされていた。
773 .TP 
774 \fIitrealvalue\fP %ld
775 (21) インターバルタイマによって、次に \fBSIGALRM\fP がプロセスへ送られるまでの時間 (単位 jiffies)。 カーネル 2.6.17
776 以降では、このフィールドはメンテナンスされなくなり、 0 にハードコードされている。
777 .TP 
778 \fIstarttime\fP %llu (Linux 2.6 より前は %lu)
779 (22) The time the process started after system boot.  In kernels before
780 Linux 2.6, this value was expressed in jiffies.  Since Linux 2.6, the value
781 is expressed in clock ticks (divide by \fIsysconf(_SC_CLK_TCK)\fP).
782 .TP 
783 \fIvsize\fP %lu
784 (23) 仮想メモリのサイズ。単位はバイト。
785 .TP 
786 \fIrss\fP %ld
787 (24) Resident Set Size。プロセスが持っている実メモリ上のページ数。
788 これはちょうどテキスト、データ、スタック空間に使われているページ数である。 デマンドロードされていないページや
789 スワップアウトされたページの数は含んでいない。
790 .TP 
791 \fIrsslim\fP %lu
792 (25) このプロセスの rss の現在のソフト・リミット (バイト単位)。 \fBgetrlimit\fP(2) の \fBRLIMIT_RSS\fP
793 の説明を参照。
794 .TP 
795 \fIstartcode\fP %lu
796 (26) プログラムテキストが実行可能であるような領域の先頭アドレス。
797 .TP 
798 \fIendcode\fP %lu
799 (27) プログラムテキストが実行可能であるような領域の末尾アドレス。
800 .TP 
801 \fIstartstack\fP %lu
802 (28) スタックの開始アドレス (すなわち、スタックの底)。
803 .TP 
804 \fIkstkesp\fP %lu
805 (29) 現在の ESP (スタックポインタ) の値。 プロセスのカーネルスタックページにある。
806 .TP 
807 \fIkstkeip\fP %lu
808 (30) 現在の EIP (インストラクションポインタ) の値。
809 .TP 
810 \fIsignal\fP %lu
811 (31) 処理待ちのシグナルのビットマップ。 10
812 進数で表示される。このフィールドは廃止予定である。リアルタイム・シグナルに関する情報は表示されないからである。代わりに
813 \fI/proc/[pid]/status\fP を使うこと。
814 .TP 
815 \fIblocked\fP %lu
816 (32) ブロックされた (blocked) シグナルのビットマップ。 10 進数で表示される。 このフィールドは廃止予定である。
817 リアルタイム・シグナルに関する情報は表示されないからである。 代わりに \fI/proc/[pid]/status\fP を使うこと。
818 .TP 
819 \fIsigignore\fP %lu
820 (33) 無視された (ignored) シグナルのビットマップ。 10 進数で表示される。 このフィールドは廃止予定である。
821 リアルタイム・シグナルに関する情報は表示されないからである。 代わりに \fI/proc/[pid]/status\fP を使うこと。
822 .TP 
823 \fIsigcatch\fP %lu
824 (34) 捕捉された (caught) シグナルのビットマップ。 10 進数で表示される。 このフィールドは廃止予定である。
825 リアルタイム・シグナルに関する情報は表示されないからである。 代わりに \fI/proc/[pid]/status\fP を使うこと。
826 .TP 
827 \fIwchan\fP %lu
828 (35) プロセスが待っている「チャネル」。これはシステムコールのアドレスであり、
829 文字名が必要ならば (アドレスとシステムコール名との) 対応表から見つけられる
830 (もし \fI/etc/psdatabase\fP [訳注: このファイル名はパッケージによる] を更新
831 しているならば、 \fIps \-l\fP して WCHAN フィールドを見よ)。
832 .TP 
833 \fInswap\fP %lu
834 .\" nswap was added in 2.0
835 (36) スワップされたページ数 (メンテナンスされていない)。
836 .TP 
837 \fIcnswap\fP %lu
838 .\" cnswap was added in 2.0
839 (37) 子プロセスの \fInswap\fP の累計 (メンテナンスされていない)。
840 .TP 
841 \fIexit_signal\fP %d (Linux 2.1.22 以降)
842 (38) プロセスが死んだときに親プロセスに送られるシグナル。
843 .TP 
844 \fIprocessor\fP %d (Linux 2.2.8 以降)
845 (39) このプロセスを最後に実行した CPU の番号。
846 .TP 
847 \fIrt_priority\fP %u (Linux 2.5.19 以降; Linux 2.6.22 より前は %lu)
848 (40) リアルタイム・スケジューリングの優先度。 リアルタイム・ポリシーの元でスケジューリングされるプロセスでは 1 から 99 の範囲の値となり、
849 リアルタイム以外のスケジューリングポリシーのプロセスでは 0 となる (\fBsched_setscheduler\fP(2)  参照)。
850 .TP 
851 \fIpolicy\fP %u (Linux 2.5.19 以降; Linux 2.6.22 より前は %lu)
852 (41) スケジューリング・ポリシー (\fBsched_setscheduler\fP(2)  参照)。 値は、 \fIlinux/sched.h\fP の
853 SCHED_* 定数を使ってデコードすればよい。
854 .TP 
855 \fIdelayacct_blkio_ticks\fP %llu (Linux 2.6.18 以降)
856 (42) (clock tick (100分の1秒) 単位での) ブロック I/O の総遅延量。
857 .TP 
858 \fIguest_time\fP %lu (Linux 2.6.24 以降)
859 (43) プロセスのゲスト時間 (ゲスト OS の仮想 CPU を実行するのに消費された時間)。 clock tick 単位で計測される
860 (\fIsysconf(_SC_CLK_TCK)\fP で割った値が表示される)。
861 .TP 
862 \fIcguest_time\fP %ld (Linux 2.6.24 以降)
863 (44) プロセスの子プロセスのゲスト時間。 clock tick 単位で計測される (\fIsysconf(_SC_CLK_TCK)\fP
864 で割った値が表示される)。
865 .RE
866 .TP 
867 \fI/proc/[pid]/statm\fP
868 (ページ単位で計測した) メモリ使用量についての情報を提供する。 各列は以下の通りである。
869 .in +4n
870 .nf
871
872 .\" (not including libs; broken, includes data segment)
873 .\" (including libs; broken, includes library text)
874 size       プログラムサイズの総計
875            (\fI/proc/[pid]/status\fP の VmSize と同じ)
876 resident   実メモリ上に存在するページ
877            (\fI/proc/[pid]/status\fP の VmRSS と同じ)
878 share      共有ページ (ファイルと関連付けられているページ)
879 text       テキスト (コード)
880 lib        ライブラリ (Linux 2.6 では未使用)
881 data       データ + スタック
882 dt         ダーティページ (Linux 2.6 では未使用)
883 .fi
884 .in
885 .TP 
886 \fI/proc/[pid]/status\fP
887 \fI/proc/[pid]/stat\fP と \fI/proc/[pid]/statm\fP にある多くの情報を、人間が解析しやすい形式で提供する。
888 以下に例を示す。
889 .in +4n
890 .nf
891
892 $\fB cat /proc/$$/status\fP
893 Name:   bash
894 State:  S (sleeping)
895 Tgid:   3515
896 Pid:    3515
897 PPid:   3452
898 TracerPid:      0
899 Uid:    1000    1000    1000    1000
900 Gid:    100     100     100     100
901 FDSize: 256
902 Groups: 16 33 100
903 VmPeak:     9136 kB
904 VmSize:     7896 kB
905 VmLck:         0 kB
906 VmHWM:      7572 kB
907 VmRSS:      6316 kB
908 VmData:     5224 kB
909 VmStk:        88 kB
910 VmExe:       572 kB
911 VmLib:      1708 kB
912 VmPTE:        20 kB
913 Threads:        1
914 SigQ:   0/3067
915 SigPnd: 0000000000000000
916 ShdPnd: 0000000000000000
917 SigBlk: 0000000000010000
918 SigIgn: 0000000000384004
919 SigCgt: 000000004b813efb
920 CapInh: 0000000000000000
921 CapPrm: 0000000000000000
922 CapEff: 0000000000000000
923 CapBnd: ffffffffffffffff
924 Cpus_allowed:   00000001
925 Cpus_allowed_list:      0
926 Mems_allowed:   1
927 Mems_allowed_list:      0
928 voluntary_ctxt_switches:        150
929 nonvoluntary_ctxt_switches:     545
930 .fi
931 .in
932 .IP
933 フィールドの詳細は以下の通りである。
934 .RS
935 .IP * 2
936 \fIName\fP: このプロセスにより実行されたコマンド。
937 .IP *
938 \fIState\fP: プロセスの現在の状態。 "R (running; 実行中)", "S (sleeping; 休眠状態)", "D (disk
939 sleep; ディスク待ちの休眠状態)", "T (stopped; 停止状態)", "T (tracing stop; トレースによる停止)", "Z
940 (zombie; ゾンビ状態)", "X (dead; 死亡)" のいずれかである。
941 .IP *
942 \fITgid\fP: スレッドグループ ID (すなわち、プロセス ID)。
943 .IP *
944 \fIPid\fP: スレッド ID (\fBgettid\fP(2)  参照)。
945 .IP *
946 \fIPPid\fP: 親プロセスの PID。
947 .IP *
948 \fITracerPid\fP: このプロセスをトレースしているプロセスの PID (トレースされていない場合は 0)。
949 .IP *
950 \fIUid\fP, \fIGid\fP: 実 UID/GID、実効 UID/GID、保存 set\-UID/GID、ファイルシステム UID/GID。
951 .IP *
952 \fIFDSize\fP: 現在割り当てられているファイルディスクリプタのスロット数。
953 .IP *
954 \fIGroups\fP: 補助グループのリスト。
955 .IP *
956 \fIVmPeak\fP: 仮想メモリサイズのピーク値。
957 .IP *
958 \fIVmSize\fP: 仮想メモリサイズ。
959 .IP *
960 \fIVmLck\fP: ロックされているメモリサイズ (\fBmlock\fP(3) 参照)。
961 .IP *
962 \fIVmHWM\fP: 実メモリ上に存在するページサイズ (resident set size)  のピーク値 ("high water mark")。
963 .IP *
964 \fIVmRSS\fP: 実メモリ上に存在するページサイズ。
965 .IP *
966 \fIVmData\fP, \fIVmStk\fP, \fIVmExe\fP: データ、スタック、テキストセグメントのサイズ。
967 .IP *
968 \fIVmLib\fP: 共有ライブラリ・コードのサイズ。
969 .IP *
970 \fIVmPTE\fP: ページ・テーブル・エントリのサイズ (Linux 2.6.10 以降)。
971 .IP *
972 \fIThreads\fP: このスレッドが属するプロセスのスレッド数。
973 .IP *
974 \fISigQ\fP: このフィールドにはスラッシュで区切られた 2 つの数字が入っている。この数字はこのプロセスの実ユーザ ID
975 宛にキューイングされたシグナルに関するものである。一つ目の数字は、この実ユーザ ID
976 宛に現在キューイングされているシグナル数である。二つ目の数字は、このプロセス宛にキューイングされたシグナル数に関するリソース上限値である
977 (\fBgetrlimit\fP(2) の \fBRLIMIT_SIGPENDING\fP の説明を参照)。
978 .IP *
979 \fISigPnd\fP, \fIShdPnd\fP: スレッド宛およびプロセス全体宛の処理待ちシグナルの数 (\fBpthreads\fP(7),
980 \fBsignal\fP(7)  参照)。
981 .IP *
982 \fISigBlk\fP, \fISigIgn\fP, \fISigCgt\fP: ブロックされるシグナル、無視されるシグナル、捕捉待ちのシグナルを 示すマスク値
983 (\fBsignal\fP(7))。
984 .IP *
985 \fICapInh\fP, \fICapPrm\fP, \fICapEff\fP: 継承可能 (inheritable)、許可 (permitted)、実効
986 (effective)  の各ケーパビリティセットで有効になっているケーパビリティのマスク値 (\fBcapabilities\fP(7)  参照)。
987 .IP *
988 \fICapBnd\fP: ケーパビリティ・バウンディングセット (カーネル 2.6.26 以降、 \fBcapabilities\fP(7)  参照)。
989 .IP *
990 \fICpus_allowed\fP: このプロセスが実行を許可されている CPU のマスク値 (Linux 2.6.24 以降、 \fBcpuset\fP(7)
991 参照)。
992 .IP *
993 \fICpus_allowed_list\fP: 前項と同じだが、「リスト形式」での表示 (Linux 2.6.26 以降、 \fBcpuset\fP(7)
994 参照)。
995 .IP *
996 \fIMems_allowed\fP: このプロセスが使用できるメモリノードのマスク値 (Linux 2.6.24 以降、 \fBcpuset\fP(7)
997 参照)。
998 .IP *
999 \fIMems_allowed_list\fP: 前項と同じだが、「リスト形式」での表示 (Linux 2.6.26 以降、 \fBcpuset\fP(7)
1000 参照)。
1001 .IP *
1002 \fIvoluntary_context_switches\fP, \fInonvoluntary_context_switches\fP:
1003 自発的/非自発的なコンテキストスイッチの回数 (Linux 2.6.23 以降)。
1004 .RE
1005 .TP 
1006 \fI/proc/[pid]/task\fP (Linux 2.6.0\-test6 以降)
1007 このディレクトリには、そのプロセスのスレッド情報を含む サブディレクトリが 1 スレッドにつき 1 つ置かれる。
1008 各サブディレクトリの名前はスレッドのスレッド ID (\fI[tid]\fP)  を示す数字である \fB(\fPgettid\fB(2)\fP を参照)。
1009 これらの各サブディレクトリには、 \fI/proc/[pid]\fP ディレクトリ以下と同じ名前と内容のファイル群がある。
1010 すべてのスレッドで共有される属性の場合、 \fItask/[tid]\fP サブディレクトリ以下の各ファイルの内容は 親ディレクトリ
1011 \fI/proc/[pid]\fP の対応するファイルと同じになることだろう (例えば、マルチスレッド・プロセスではファイル
1012 \fItask/[tid]/cwd\fP はいずれも親ディレクトリ内の \fI/proc/[pid]/cwd\fP
1013 と同じ値を持つことになる。なぜなら、一つのプロセスに属すすべての スレッドは作業ディレクトリを共有するからである)。 スレッド毎に独立な属性の場合、
1014 \fItask/[tid]\fP サブディレクトリ以下の各ファイルは異なる値を持つことがある (例えば、ファイル \fItask/[tid]/status\fP
1015 はスレッド毎に異なる値を持つ可能性がある)。
1016
1017 .\" The following was still true as at kernel 2.6.13
1018 マルチスレッドプロセスでは、メインスレッドがすでに終了している場合、 \fI/proc/[pid]/task\fP ディレクトリの内容は参照できない
1019 (スレッドの終了は通常 \fBpthread_exit\fP(3)  を呼び出しにより行われる)。
1020 .TP 
1021 \fI/proc/apm\fP
1022 Advanced Power Management のバージョンとバッテリ情報。 カーネルのコンパイル時に \fBCONFIG_APM\fP
1023 を定義したときに存在する。
1024 .TP 
1025 \fI/proc/bus\fP
1026 インストールされている各バス用にサブディレクトリがある。
1027 .TP 
1028 \fI/proc/bus/pccard\fP
1029 PCMCIA デバイスの情報が書かれるサブディレクトリ。 カーネルのコンパイル時に \fBCONFIG_PCMCIA\fP を定義したときに存在する。
1030 .TP 
1031 \fI/proc/bus/pccard/drivers\fP
1032 .TP 
1033 \fI/proc/bus/pci\fP
1034 いくつかのサブディレクトリがあり、 PCI バス・インストールされているデバイス・ デバイスドライバの情報が書かれた仮想ファイルがある。
1035 これらのファイルのうちいくつかは ASCII フォーマットではない。
1036 .TP 
1037 \fI/proc/bus/pci/devices\fP
1038 PCI デバイスの情報。 \fBlspci\fP(8)  や \fBsetpci\fP(8)  でアクセスすることができる。
1039 .TP 
1040 \fI/proc/cmdline\fP
1041 ブート時に Linux カーネルに渡された引き数。 引き数の受け渡しは、たいてい \fBlilo\fP(8)  や \fBgrub\fP(8)
1042 といったブートマネージャを使って行われる。
1043 .TP 
1044 \fI/proc/config.gz\fP (Linux 2.6 以降)
1045 このファイルでは、現在実行中のカーネルの構築時に使用された 設定オプションを参照できる。 書式は、 (\fImake xconfig\fP, \fImake
1046 config\fP などを使って) カーネルの設定を変更した際に生成される \fI.config\fP ファイルのものと同じである。
1047 ファイルの内容は圧縮されており、 \fBzcat\fP(1), \fBzgrep\fP(1)  などを使うと、表示や検索ができる。 ファイルが変更されていない限り、
1048 \fI/proc/config.gz\fP の内容は次のコマンドで得られる内容と同じである。
1049 .in +4n
1050 .nf
1051
1052 cat /lib/modules/$(uname \-r)/build/.config
1053 .fi
1054 .in
1055 .IP
1056 \fI/proc/config.gz\fP が提供されるのは、カーネルの設定で \fBCONFIG_IKCONFIG_PROC\fP
1057 が有効になっている場合のみである。
1058 .TP 
1059 \fI/proc/cpuinfo\fP
1060 このファイルは、CPU およびシステムアーキテクチャに依存する項目を 集めたもので、リストの内容はサポートされているアーキテクチャ毎に異なる。 2
1061 つだけ共通の項目がある。 \fIprocessor\fP はプロセッサ番号で、 \fIbogomips\fP はカーネルの初期化時に計算されるシステム定数である。
1062 SMP マシンでは各 CPU についての情報が書かれている。 \fBlscpu\fP(1) コマンドはこのファイルから情報を収集する。
1063 .TP 
1064 \fI/proc/devices\fP
1065 メジャーデバイス番号とデバイスグループのテキスト形式のリスト。 MAKEDEV スクリプトはこのファイルを使って、
1066 カーネルとの整合性を保つことができる。
1067 .TP 
1068 \fI/proc/diskstats\fP (Linux 2.5.69 以降)
1069 このファイルには各ディスクデバイスのディスク I/O 統計情報が書かれている。 更に詳しい情報は、Linux カーネルソースファイル
1070 \fIDocumentation/iostats.txt\fP を参照すること。
1071 .TP 
1072 \fI/proc/dma\fP
1073 登録されている \fIISA\fP DMA (direct memory access) チャネルのリスト。
1074 .TP 
1075 \fI/proc/driver\fP
1076 空のサブディレクトリ。
1077 .TP 
1078 \fI/proc/execdomains\fP
1079 実行ドメインのリスト (ABI パーソナリティ)。
1080 .TP 
1081 \fI/proc/fb\fP
1082 カーネルのコンパイル時に \fBCONFIG_FB\fP が定義されている場合、フレームバッファの情報が書かれる。
1083 .TP 
1084 \fI/proc/filesystems\fP
1085 カーネルが対応しているファイルシステムのテキスト形式のリスト。 カーネルに組み込まれてコンパイルされたファイルシステムと、
1086 カーネルモジュールが現在ロードされているファイルシステムが列挙される (\fBfilesystems\fP(5)  参照)。 ファイルシステムに
1087 "nodev" という印が付いている場合、 そのファイルシステムがマウントするためのブロックデバイスを 必要としないことを意味する (例えば、
1088 仮想ファイルシステム、ネットワークファイルシステムなど)。
1089
1090 ちなみに、マウント時にファイルシステムが指定されず、 どうやってもファイルシステムの種類を判定できなかった際に、 このファイルを \fBmount\fP(8)
1091 が使用するかもしれない。 その場合、このファイルに含まれるファイルシステムが試される (ただし、"nodev" の印がついたものは除く)。
1092 .TP 
1093 \fI/proc/fs\fP
1094 空のサブディレクトリ。
1095 .TP 
1096 \fI/proc/ide\fP
1097 このディレクトリは IDE バスをもつシステムに存在する。 各 IDE チャネルとそれに取り付けられている各デバイスごとにディレクトリがあり、
1098 以下のファイルが含まれている。
1099
1100 .in +4n
1101 .nf
1102 cache              バッファサイズ (KB)
1103 capacity           セクタ数
1104 driver             ドライバのバージョン
1105 geometry           物理ジオメトリと論理ジオメトリ
1106 identify           16 進数表記
1107 media              メディアのタイプ
1108 model              製造者のモデル番号
1109 settings           ドライブの設定
1110 smart_thresholds   16 進数表記
1111 smart_values       16 進数表記
1112 .fi
1113 .in
1114
1115 \fBhdparm\fP(8)  ユーティリティは、分かりやすい形式で この情報にアクセスするための手段を提供する。
1116 .TP 
1117 \fI/proc/interrupts\fP
1118 IO デバイス毎の CPU 別の割り込み回数の記録に使われる。 Linux 2.6.24 以降、少なくとも i386 と x86_64
1119 アーキテクチャでは、 (デバイスと関連がない) システム内部の割り込みについても記録される。 システム内部の割り込みには、NMI
1120 (nonmaskable interrupt), LOC (local timer interrupt) や、SMP システムでは TLB (TLB
1121 flush interrupt), RES (rescheduling interrupt), CAL (remote function call
1122 interrupt)  などがある。 簡単に読むことのできるフォーマットで、ASCII で表記されている。
1123 .TP 
1124 \fI/proc/iomem\fP
1125 Linux 2.4 における I/O メモリマップ。
1126 .TP 
1127 \fI/proc/ioports\fP
1128 現在登録され使われている I/O ポート領域のリスト。
1129 .TP 
1130 \fI/proc/kallsyms\fP (Linux 2.5.71 以降)
1131 カーネルの外部シンボル定義を保持する。 \fBmodules\fP(X)  関係のツールがローダブルモジュールを動的にリンクしたり バインド (bind)
1132 するのに使われる。 Linux 2.5.47 以前では、微妙に異なる書式の似たようなファイルが \fIksyms\fP という名前であった。
1133 .TP 
1134 \fI/proc/kcore\fP
1135 このファイルはシステムの物理メモリを表現しており、 ELF コアファイル形式 (core file format) で保持されている。
1136 この擬似ファイルと strip されていないカーネルのバイナリ (\fI/usr/src/linux/vmlinux\fP)  [訳注:
1137 パッケージに依存する]) があれば、 GDB はカーネル内の任意のデータ構造の現在の状態を調べられる。
1138
1139 このファイルの大きさは物理メモリ (RAM) のサイズに 4KB を加えた値である。
1140 .TP 
1141 \fI/proc/kmsg\fP
1142 このファイルは \fBsyslog\fP(2)  システムコールでカーネルメッセージを読み出す代りに使える。
1143 プロセスがこのファイルを読むためにはスーパーユーザー権限が必要であり、 ファイルを読み出すのは 1 つのプロセスのみに限るべきである。
1144 カーネルメッセージを記録するために、 \fBsyslog\fP(2)  システムコールの機能を使う syslog プロセスが稼働している場合、
1145 このファイルを読み出すべきではない。
1146
1147 このファイルの中の情報は \fBdmesg\fP(1)  によって表示される。
1148 .TP 
1149 \fI/proc/ksyms\fP (Linux 1.1.23\-2.5.47)
1150 \fI/proc/kallsyms\fP を参照。
1151 .TP 
1152 \fI/proc/loadavg\fP
1153 このファイルの最初の 3 つのフィールドはロードアベレージの数値で、 1, 5, 15 分
1154 あたりの実行キュー内 (state R) または ディスク I/O 待ち (state D) のジョブ数
1155 を与える。 これは \fBuptime\fP(1) などのプログラムによって得られる値と同じである。
1156 4 番目のフィールドはスラッシュ (/) で区切られた 2 つの数値から構成される。
1157 この数値のうち最初のものは、現在実行可能なカーネルスケジュールエンティティ
1158 (プロセス、スレッド) の数である。スラッシュの後の数値は、現在システム上に
1159 存在するカーネルスケジュールエンティティの数である。 5 番目のフィールドは
1160 システム上に最も最近生成されたプロセスの PID である。
1161 .TP 
1162 \fI/proc/locks\fP
1163 このファイルは現在のファイルロック (\fBflock\fP(2) と \fBfcntl\fP(2))  とリース (\fBfcntl\fP(2))  を表示する。
1164 .TP 
1165 \fI/proc/malloc\fP (Linux 2.2 以前のみ)
1166 .\" It looks like this only ever did something back in 1.0 days
1167 コンパイルのときに \fBCONFIGDEBUGMALLOC\fP が定義されているときのみ、このファイルは存在する。
1168 .TP 
1169 \fI/proc/meminfo\fP
1170 This file reports statistics about memory usage on the system.  It is used
1171 by \fBfree\fP(1)  to report the amount of free and used memory (both physical
1172 and swap)  on the system as well as the shared memory and buffers used by
1173 the kernel.  Each line of the file consists of a parameter name, followed by
1174 a colon, the value of the parameter, and an option unit of measurement
1175 (e.g., "kB").  The list below describes the parameter names and the format
1176 specifier required to read the field value.  Except as noted below, all of
1177 the fields have been present since at least Linux 2.6.0.  Some fileds are
1178 displayed only if the kernel was configured with various options; those
1179 dependencies are noted in the list.
1180 .RS
1181 .TP 
1182 \fIMemTotal\fP %lu
1183 Total usable RAM (i.e. physical RAM minus a few reserved bits and the kernel
1184 binary code).
1185 .TP 
1186 \fIMemFree\fP %lu
1187 The sum of \fILowFree\fP+\fIHighFree\fP.
1188 .TP 
1189 \fIBuffers\fP %lu
1190 Relatively temporary storage for raw disk blocks that shouldn't get
1191 tremendously large (20MB or so).
1192 .TP 
1193 \fICached\fP %lu
1194 In\-memory cache for files read from the disk (the page cache).  Doesn't
1195 include \fISwapCached\fP.
1196 .TP 
1197 \fISwapCached\fP %lu
1198 Memory that once was swapped out, is swapped back in but still also is in
1199 the swap file.  (If memory pressure is high, these pages don't need to be
1200 swapped out again because they are already in the swap file.  This saves
1201 I/O.)
1202 .TP 
1203 \fIActive\fP %lu
1204 Memory that has been used more recently and usually not reclaimed unless
1205 absolutely necessary.
1206 .TP 
1207 \fIInactive\fP %lu
1208 Memory which has been less recently used.  It is more eligible to be
1209 reclaimed for other purposes.
1210 .TP 
1211 \fIActive(anon)\fP %lu (Linux 2.6.28 以降)
1212 [To be documented.]
1213 .TP 
1214 \fIInactive(anon)\fP %lu (Linux 2.6.28 以降)
1215 [To be documented.]
1216 .TP 
1217 \fIActive(file)\fP %lu (Linux 2.6.28 以降)
1218 [To be documented.]
1219 .TP 
1220 \fIInactive(file)\fP %lu (Linux 2.6.28 以降)
1221 [To be documented.]
1222 .TP 
1223 \fIUnevictable\fP %lu (Linux 2.6.28 以降)
1224 (From Linux 2.6.28 to 2.6.30, \fBCONFIG_UNEVICTABLE_LRU\fP was required.)  [To
1225 be documented.]
1226 .TP 
1227 \fIMlocked\fP %lu (Linux 2.6.28 以降)
1228 (From Linux 2.6.28 to 2.6.30, \fBCONFIG_UNEVICTABLE_LRU\fP was required.)  [To
1229 be documented.]
1230 .TP 
1231 \fIHighTotal\fP %lu
1232 (Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)  Total amount
1233 of highmem.  Highmem is all memory above ~860MB of physical memory.  Highmem
1234 areas are for use by user\-space programs, or for the page cache.  The kernel
1235 must use tricks to access this memory, making it slower to access than
1236 lowmem.
1237 .TP 
1238 \fIHighFree\fP %lu
1239 (Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)  Amount of free
1240 highmem.
1241 .TP 
1242 \fILowTotal\fP %lu
1243 (Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)  Total amount
1244 of lowmem.  Lowmem is memory which can be used for everything that highmem
1245 can be used for, but it is also available for the kernel's use for its own
1246 data structures.  Among many other things, it is where everything from
1247 \fISlab\fP is allocated.  Bad things happen when you're out of lowmem.
1248 .TP 
1249 \fILowFree\fP %lu
1250 (Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)  Amount of free
1251 lowmem.
1252 .TP 
1253 \fIMmapCopy\fP %lu (Linux 2.6.29 以降)
1254 (\fBCONFIG_MMU\fP is required.)  [To be documented.]
1255 .TP 
1256 \fISwapTotal\fP %lu
1257 Total amount of swap space available.
1258 .TP 
1259 \fISwapFree\fP %lu
1260 Amount of swap space that is currently unused.
1261 .TP 
1262 \fIDirty\fP %lu
1263 Memory which is waiting to get written back to the disk.
1264 .TP 
1265 \fIWriteback\fP %lu
1266 Memory which is actively being written back to the disk.
1267 .TP 
1268 \fIAnonPages\fP %lu (Linux 2.6.18 以降)
1269 Non\-file backed pages mapped into user\-space page tables.
1270 .TP 
1271 \fIMapped\fP %lu
1272 Files which have been mmaped, such as libraries.
1273 .TP 
1274 \fIShmem\fP %lu (Linux 2.6.32 以降)
1275 [To be documented.]
1276 .TP 
1277 \fISlab\fP %lu
1278 In\-kernel data structures cache.
1279 .TP 
1280 \fISReclaimable\fP %lu (Linux 2.6.19 以降)
1281 Part of \fISlab\fP, that might be reclaimed, such as caches.
1282 .TP 
1283 \fISUnreclaim\fP %lu (Linux 2.6.19 以降)
1284 Part of \fISlab\fP, that cannot be reclaimed on memory pressure.
1285 .TP 
1286 \fIKernelStack\fP %lu (Linux 2.6.32 以降)
1287 Amount of memory allocated to kernel stacks.
1288 .TP 
1289 \fIPageTables\fP %lu (Linux 2.6.18 以降)
1290 Amount of memory dedicated to the lowest level of page tables.
1291 .TP 
1292 \fIQuicklists\fP %lu (Linux 2.6.27 以降)
1293 (\fBCONFIG_QUICKLIST\fP is required.)  [To be documented.]
1294 .TP 
1295 \fINFS_Unstable\fP %lu (Linux 2.6.18 以降)
1296 NFS pages sent to the server, but not yet committed to stable storage.
1297 .TP 
1298 \fIBounce\fP %lu (Linux 2.6.18 以降)
1299 Memory used for block device "bounce buffers".
1300 .TP 
1301 \fIWritebackTmp\fP %lu (Linux 2.6.26 以降)
1302 Memory used by FUSE for temporary writeback buffers.
1303 .TP 
1304 \fICommitLimit\fP %lu (Linux 2.6.10 以降)
1305 Based on the overcommit ratio ('vm.overcommit_ratio'), this is the total
1306 amount of memory currently available to be allocated on the system.  This
1307 limit is adhered to only if strict overcommit accounting is enabled (mode 2
1308 in \fI/proc/sys/vm/overcommit_ratio\fP).  The \fICommitLimit\fP is calculated
1309 using the following formula:
1310
1311     CommitLimit = (overcommit_ratio * Physical RAM) + Swap
1312
1313 For example, on a system with 1GB of physical RAM and 7GB of swap with a
1314 \fIovercommit_ratio\fP of 30, this formula yields a \fICommitLimit\fP of 7.3GB.
1315 For more details, see the memory overcommit documentation in the kernel
1316 source file \fIDocumentation/vm/overcommit\-accounting\fP.
1317 .TP 
1318 \fICommitted_AS\fP %lu
1319 The amount of memory presently allocated on the system.  The committed
1320 memory is a sum of all of the memory which has been allocated by processes,
1321 even if it has not been "used" by them as of yet.  A process which allocates
1322 1GB of memory (using \fBmalloc\fP(3)  or similar), but touches only 300MB of
1323 that memory will show up as using only 300MB of memory even if it has the
1324 address space allocated for the entire 1GB.  This 1GB is memory which has
1325 been "committed" to by the VM and can be used at any time by the allocating
1326 application.  With strict overcommit enabled on the system (mode 2
1327 \fI/proc/sys/vm/overcommit_memory\fP), allocations which would exceed the
1328 \fICommitLimit\fP (detailed above) will not be permitted.  This is useful if
1329 one needs to guarantee that processes will not fail due to lack of memory
1330 once that memory has been successfully allocated.
1331 .TP 
1332 \fIVmallocTotal\fP %lu
1333 Total size of vmalloc memory area.
1334 .TP 
1335 \fIVmallocUsed\fP %lu
1336 Amount of vmalloc area which is used.
1337 .TP 
1338 \fIVmallocChunk\fP %lu
1339 Largest contiguous block of vmalloc area which is free.
1340 .TP 
1341 \fIHardwareCorrupted\fP %lu (Linux 2.6.32 以降)
1342 (\fBCONFIG_MEMORY_FAILURE\fP is required.)  [To be documented.]
1343 .TP 
1344 \fIAnonHugePages\fP %lu (Linux 2.6.38 以降)
1345 (\fBCONFIG_TRANSPARENT_HUGEPAGE\fP is required.)  Non\-file backed huge pages
1346 mapped into user\-space page tables.
1347 .TP 
1348 \fIHugePages_Total\fP %lu
1349 (\fBCONFIG_HUGETLB_PAGE\fP is required.)  The size of the pool of huge pages.
1350 .TP 
1351 \fIHugePages_Free\fP %lu
1352 (\fBCONFIG_HUGETLB_PAGE\fP is required.)  The number of huge pages in the pool
1353 that are not yet allocated.
1354 .TP 
1355 \fIHugePages_Rsvd\fP %lu (Linux 2.6.17 以降)
1356 (\fBCONFIG_HUGETLB_PAGE\fP is required.)  This is the number of huge pages for
1357 which a commitment to allocate from the pool has been made, but no
1358 allocation has yet been made.  These reserved huge pages guarantee that an
1359 application will be able to allocate a huge page from the pool of huge pages
1360 at fault time.
1361 .TP 
1362 \fIHugePages_Surp\fP %lu (Linux 2.6.24 以降)
1363 (\fBCONFIG_HUGETLB_PAGE\fP is required.)  This is the number of huge pages in
1364 the pool above the value in \fI/proc/sys/vm/nr_hugepages\fP.  The maximum
1365 number of surplus huge pages is controlled by
1366 \fI/proc/sys/vm/nr_overcommit_hugepages\fP.
1367 .TP 
1368 \fIHugepagesize\fP %lu
1369 (\fBCONFIG_HUGETLB_PAGE\fP is required.)  The size of huge pages.
1370 .RE
1371 .TP 
1372 \fI/proc/modules\fP
1373 現在システムにロードされているモジュールのテキスト形式のリスト。 \fBlsmod\fP(8)  も参照。
1374 .TP 
1375 \fI/proc/mounts\fP
1376 カーネル 2.4.19 より前では、このファイルは現在システムにマウントされている 全てのファイルシステムのリストであった。 Linux 2.4.19
1377 でプロセス単位のマウント名前空間が導入されたことに伴い、 このファイルは \fI/proc/self/mounts\fP へのリンクとなった。
1378 \fI/proc/self/mounts\fP はそのプロセス自身のマウント名前空間のマウントポイントのリストである。 このファイルのフォーマットは
1379 \fBfstab\fP(5)  に記述されている。
1380 .TP 
1381 \fI/proc/mtrr\fP
1382 Memory Type Range Registers。 詳細は、Linux カーネルソースファイル \fIDocumentation/mtrr.txt\fP
1383 を参照すること。
1384 .TP 
1385 \fI/proc/net\fP
1386 さまざまなネットワークについての擬似ファイルで、 それぞれがネットワーク層の各種の状態を与える。 これらのファイルの内容は ASCII 形式なので、
1387 \fBcat\fP(1)  で読み出せる。 とはいえ基本コマンドの \fBnetstat\fP(8)  はこれらのファイルの内容のよりすっきりとした表示を提供する。
1388 .TP 
1389 \fI/proc/net/arp\fP
1390 アドレス解決に使われるカーネルの ARP テーブルの ASCII 可読なダンプを保持している。 動的結合されたものと固定 (preprogrammed)
1391 の両方の APP エントリを見ることができる。フォーマットは以下のとおり:
1392
1393 .nf
1394 .in 8n
1395 \f(CWIP address     HW type   Flags     HW address          Mask   Device
1396 192.168.0.50   0x1       0x2       00:50:BF:25:68:F3   *      eth0
1397 192.168.0.250  0x1       0xc       00:00:00:00:00:00   *      eth0\fP
1398 .fi
1399 .in
1400
1401 ここで IP address はマシンの IPv4 アドレス、 HW type はそのアドレスの RFC\ 826 で定められているハードウェアの形式、
1402 Flags は ARP 構造体 (\fI/usr/include/linux/if_arp.h\fP 内で定義されている) の内部フラグ、 HW
1403 address はその IP アドレスにマップされているデータリンク層のアドレス (もしわかっていれば) である。
1404 .TP 
1405 \fI/proc/net/dev\fP
1406 擬似ファイル dev はネットワークデバイスの状態情報を含んでいる。 これは送受信したパケット数、エラーとコリジョン (collision) の回数、
1407 その他の基本的な統計を与える。 これらは \fBifconfig\fP(8)  がデバイスの状態を報告するのに使われる。 フォーマットは以下のとおり:
1408
1409 .nf
1410 .in 1n
1411 \f(CWInter\-|   Receive                                                |  Transmit
1412  face |bytes    packets errs drop fifo frame compressed multicast|bytes    packets errs drop fifo colls carrier compressed
1413     lo: 2776770   11307    0    0    0     0          0         0  2776770   11307    0    0    0     0       0          0
1414   eth0: 1215645    2751    0    0    0     0          0         0  1782404    4324    0    0    0   427       0          0
1415   ppp0: 1622270    5552    1    0    0     0          0         0   354130    5669    0    0    0     0       0          0
1416   tap0:    7714      81    0    0    0     0          0         0     7714      81    0    0    0     0       0          0\fP
1417 .in
1418 .fi
1419 .\" .TP
1420 .\" .I /proc/net/ipx
1421 .\" No information.
1422 .\" .TP
1423 .\" .I /proc/net/ipx_route
1424 .\" No information.
1425 .TP 
1426 \fI/proc/net/dev_mcast\fP
1427 \fI/usr/src/linux/net/core/dev_mcast.c\fP で定義されており、以下の形式である。
1428 .nf
1429 .in +5
1430 indx interface_name  dmi_u dmi_g dmi_address
1431 2    eth0            1     0     01005e000001
1432 3    eth1            1     0     01005e000001
1433 4    eth2            1     0     01005e000001
1434 .in
1435 .fi
1436 .TP 
1437 \fI/proc/net/igmp\fP
1438 Internet Group Management Protocol (インターネットグループ管理プロトコル)。
1439 \fI/usr/src/linux/net/core/igmp.c\fP で定義されている。
1440 .TP 
1441 \fI/proc/net/rarp\fP
1442 このファイルは \fIarp\fP と同じフォーマットで 逆アドレス解決サービス (reverse address lookup services)
1443 \fBrarp\fP(8)  に提供するために使われる現在の逆マップデータベースの内容を含んでいる。 RARP
1444 がカーネルコンフィグレーションに設定されていなければ、 このファイルは存在しない。
1445 .TP 
1446 \fI/proc/net/raw\fP
1447 .\" .TP
1448 .\" .I /proc/net/route
1449 .\" No information, but looks similar to
1450 .\" .BR route (8).
1451 RAW ソケットテーブルのダンプを保持している。 ほとんどの情報はデバッグ以外では
1452 使われない。 \&"sl" の値はソケットのカーネルハッシュスロット、
1453 \&"local_address" はローカルアドレスとプロトコル番号のペア
1454 [訳者追加: "rem_address" はリモートアドレスとプロトコル番号のペア]。
1455 \&"st" はソケットの内部状態。 \&"tx_queue" と "rx_queue" はカーネルメモリを
1456 消費している 送信/受信データキューのサイズ。 \&"tr" と "tm\->when" と
1457 "rexmits" フィールドは RAW では使われていない。
1458 \&"uid" フィールドはソケット生成者の実効 UID を保持している。
1459 .TP 
1460 \fI/proc/net/snmp\fP
1461 このファイルは SNMP エージェントが必要とする IP, ICMP, TCP, UDP 管理情報を ASCII データとして保持している。
1462 .TP 
1463 \fI/proc/net/tcp\fP
1464 TCP ソケットテーブルのダンプを保持している。 大部分の情報はデバッグ以外には
1465 使われない。 sl はソケットのカーネルハッシュスロットの値、 \&"local_address"
1466 はローカルアドレスとポート番号のペアである。 (ソケットが接続している場合は)
1467 \&"rem_address" はリモートアドレスとポート番号の対である。 \&"st" はソケット
1468 の内部状態である。 \&"tx_queue" と "rx_queue" はカーネルメモリを消費している
1469 送信/受信データキューのサイズ。 \&"tr" と "tm\->when" と "rexmits"
1470 フィールドはソケット状態のカーネル 内部情報を保持しているが、
1471 これらはデバッグのときにしか役に立たない。 \&"uid" フィールドはソケット
1472 生成者の実効 UID を保持している。
1473 .TP 
1474 \fI/proc/net/udp\fP
1475 UDP ソケットテーブルのダンプを保持している。 大部分の情報はデバッグ以外には
1476 使われない。 sl はソケットのカーネルハッシュスロットの値、 "local_address"
1477 はローカルアドレスとポート番号のペアである。 (ソケットが接続している場合は)
1478 "rem_address" はリモートアドレスとポート番号のペアである。 "st" はソケットの
1479 内部状態である。 "tx_queue" と "rx_queue" はカーネルメモリを消費している
1480 送信/受信データキューのサイズ。 "tr" と "tm\->when" と "rexmits"
1481 フィールドは UDP では使われていない。 "uid" フィールドはソケット生成者の
1482 実効 UID を保持している。 フォーマットは以下のとおり:
1483
1484 .nf
1485 .in 1n
1486 \f(CWsl  local_address rem_address   st tx_queue rx_queue tr rexmits  tm\->when uid
1487  1: 01642C89:0201 0C642C89:03FF 01 00000000:00000001 01:000071BA 00000000 0
1488  1: 00000000:0801 00000000:0000 0A 00000000:00000000 00:00000000 6F000100 0
1489  1: 00000000:0201 00000000:0000 0A 00000000:00000000 00:00000000 00000000 0\fP
1490 .in
1491 .fi
1492 .TP 
1493 \fI/proc/net/unix\fP
1494 UNIX ドメインソケットのリスト。 フォーマットは以下のとおり:
1495 .nf
1496 .sp .5
1497 \f(CWNum RefCount Protocol Flags    Type St Path
1498  0: 00000002 00000000 00000000 0001 03
1499  1: 00000001 00000000 00010000 0001 01 /dev/printer\fP
1500 .sp .5
1501 .fi
1502
1503 ここで、Num はカーネルのテーブルスロット数、 RefCount はソケットを使用して
1504 いるユーザー数、 Protocol はいまのところいつも 0 で、Flags はソケットの状態
1505 を保持している カーネル内部のフラグである。 Type はいまのところいつも 1
1506 (UNIX ドメインのデータグラムソケットは、現在のカーネルではサポートされていない
1507 [訳注: 2.0.34 ではサポートされているようだ])。
1508 St はソケットの内部状態で、Path は(もしあれば) ソケットのパス名である。
1509 .TP 
1510 \fI/proc/partitions\fP
1511 各パーティションのメジャー番号とマイナー番号が書かれている。 さらに、ブロック数とパーティション名も書かれている。
1512 .TP 
1513 \fI/proc/pci\fP
1514 カーネルの初期化時に見つかったすべての PCI デバイスのリストと その設定。
1515
1516 .\" FIXME /proc/sched_debug
1517 .\" .TP
1518 .\" .IR /proc/sched_debug " (since Linux 2.6.23)"
1519 .\" See also /proc/[pid]/sched
1520 このファイルは非推奨であり、新しい PCI 用の \fI/proc\fP インターフェイス (\fI/proc/bus/pci\fP)  を使うこと。
1521 このファイルは Linux 2.2 でオプションになった (カーネルのコンパイル時に \fBCONFIG_PCI_OLD_PROC\fP
1522 をセットすると利用可能であった)。 Linux 2.4 で再びオプションなしで有効に戻った。 さらに、Linux 2.6 で非推奨となり
1523 (\fBCONFIG_PCI_LEGACY_PROC\fP をセットするとまだ利用可能であった)、 最終的に Linux 2.6.17
1524 以降で完全に削除された。
1525 .TP 
1526 \fI/proc/profile\fP (Linux 2.4 以降)
1527 This file is present only if the kernel was booted with the \fIprofile=1\fP
1528 command\-line option.  It exposes kernel profiling information in a binary
1529 format for use by \fBreadprofile\fP(1).  Writing (e.g., an empty string) to
1530 this file resets the profiling counters; on some architectures, writing a
1531 binary integer "profiling multiplier" of size \fIsizeof(int)\fP sets the
1532 profiling interrupt frequency.
1533 .TP 
1534 \fI/proc/scsi\fP
1535 \fIscsi\fP 中間レベル擬似ファイルといくつかの SCSI 低レベルドライバの ディレクトリを含むディレクトリ。 これらのファイルは ASCII
1536 で表現されているので \fBcat\fP(1)  で読める。
1537
1538 いくつかのファイルは書き込み可能で、サブシステムの設定を変更したり、 特定の機能をオン/オフすることができる。
1539 .TP 
1540 \fI/proc/scsi/scsi\fP
1541 カーネルが知っているすべての SCSI デバイスのリスト。 このリストは起動時に (コンソールで) 見られるものとほぼ同じである。 scsi
1542 は現在のところ \fIadd\-single\-device\fP コマンドのみをサポートしている。 これによりルート (root)
1543 は既知のデバイスリストへ活線挿抜 (hotplugged)  デバイスを加えることができる。
1544
1545 次のコマンドを実行すると、
1546 .in +4n
1547 .nf
1548
1549 echo \(aqscsi add\-single\-device 1 0 5 0\(aq > /proc/scsi/scsi
1550
1551 .fi
1552 .in
1553 ホストアダプタ scsi1 は SCSI チャネル 0 で ID 5 LUN 0 のデバイスを探す。 もしこのアドレスに既知のデバイスがあるか、
1554 不正なアドレスであったならばエラーが返る。
1555 .TP 
1556 \fI/proc/scsi/[drivername]\fP
1557 いまのところ \fI[drivername]\fP は NCR53c7xx, aha152x, aha1542, aha1740, aic7xxx,
1558 buslogic, eata_dma, eata_pio, fdomain, in2000, pas16, qlogic, scsi_debug,
1559 seagate, t128, u15\-24f, ultrastore, wd7000 のどれかである。 少なくとも 1 つの SCSI
1560 ホストバスアダプタ (HBA) に ドライバが割り当てられていると、そのドライバに対応したディレクトリが現れる。 それぞれのディレクトリには、
1561 登録されたホストアダプタに対応してファイルが作られる。 このファイルの名前は、システムの初期化の際に ホストアダプタに割り当てられた番号になる。
1562
1563 これらのファイルを読めばドライバとホストアダプタの設定や 統計などを見ることができる。
1564
1565 これらのファイルへの書き込みはホストアダプタごとに異なる動作を引き起こす。 たとえば \fIlatency\fP と \fInolatency\fP
1566 コマンドを用いると、 ルート (root、スーパーユーザー) は eata_dma ドライバの隠し測定コードの オン/オフを切り替えることができる。
1567 また \fI lockup\fP と \fIunlock\fP コマンドを用いると、ルートは scsi_debug ドライバがシミュレートするバスロックアップ
1568 (bus lockup) を 制御することができる。
1569 .TP 
1570 \fI/proc/self\fP
1571 このディレクトリはプロセスに (プロセス自身の)  \fI/proc\fP ファイルシステムへのアクセスを参照させる。 これは \fI/proc\fP 内の
1572 (このプロセスの) プロセス ID が名前となっている ディレクトリと全く同一である。
1573 .TP 
1574 \fI/proc/slabinfo\fP
1575 Linux 2.6.16 以降では、 カーネル設定オプション \fBCONFIG_SLAB\fP が有効の場合にのみ、このファイルは存在する。
1576 カーネルキャッシュの情報。 \fI/proc/slabinfo\fP のフィールドは以下のとおり。
1577 .in +4n
1578 .nf
1579
1580 cache\-name
1581 num\-active\-objs
1582 total\-objs
1583 object\-size
1584 num\-active\-slabs
1585 total\-slabs
1586 num\-pages\-per\-slab
1587 .fi
1588 .in
1589
1590 詳細は \fBslabinfo\fP(5)  を参照すること。
1591 .TP 
1592 \fI/proc/stat\fP
1593 カーネル/システムの統計。 アーキテクチャによって異なる。 共通エントリには以下のものが含まれる。
1594 .RS
1595 .TP 
1596 \fIcpu  3357 0 4313 1362393\fP
1597 .\" 1024 on Alpha and ia64
1598 The amount of time, measured in units of USER_HZ (1/100ths of a second on
1599 most architectures, use \fIsysconf(_SC_CLK_TCK)\fP to obtain the right value),
1600 that the system spent in various states:
1601 .RS
1602 .TP 
1603 \fIuser\fP
1604 (1) Time spent in user mode.
1605 .TP 
1606 \fInice\fP
1607 (2) Time spent in user mode with low priority (nice).
1608 .TP 
1609 \fIsystem\fP
1610 (3) Time spent in system mode.
1611 .TP 
1612 \fIidle\fP
1613 .\" FIXME Actually, the following info about the /proc/stat 'cpu' field
1614 .\"       does not seem to be quite right (at least in 2.6.12 or 3.6):
1615 .\"       the idle time in /proc/uptime does not quite match this value
1616 (4) Time spent in the idle task.  This value should be USER_HZ times the
1617 second entry in the \fI/proc/uptime\fP pseudo\-file.
1618 .TP 
1619 \fIiowait\fP (Linux 2.5.41 以降)
1620 (5) Time waiting for I/O to complete.
1621 .TP 
1622 \fIirq\fP (Linux 2.6.0\-test4 以降)
1623 (6) Time servicing interrupts.
1624 .TP 
1625 \fIsoftirq\fP (Linux 2.6.0\-test4 以降)
1626 (7) Time servicing softirqs.
1627 .TP 
1628 \fIsteal\fP (Linux 2.6.11 以降)
1629 (8) 盗まれた時間 (stolen time)。仮想化環境での動作時に他のオペレーティングシステムにより消費された時間である。
1630 .TP 
1631 \fIguest\fP (Linux 2.6.24 以降)
1632 .\" See Changelog entry for 5e84cfde51cf303d368fcb48f22059f37b3872de
1633 (9) Linux カーネルの制御下のゲストオペレーティングシステムの仮想 CPU の 実行に消費された時間。
1634 .TP 
1635 \fIguest_nice\fP (Linux 2.6.33 以降)
1636 .\" commit ce0e7b28fb75cb003cfc8d0238613aaf1c55e797
1637 (10) nice が適用されたゲスト (Linux カーネルの制御下のゲストオペレーティングシステムの仮想 CPU) の 実行に消費された時間。
1638 .RE
1639 .TP 
1640 \fIpage 5741 1808\fP
1641 システムが (ディスクから) ページイン/ページアウトしたページ数。
1642 .TP 
1643 \fIswap 1 0\fP
1644 スワップイン/スワップアウトされたページ数。
1645 .TP 
1646 .\" FIXME The following is not the full picture for the 'intr' of
1647 .\"       /proc/stat on 2.6:
1648 \fIintr 1462898\fP
1649 この行はシステム起動時以降に処理された割り込みの回数を示す。 最初の欄は処理された割り込み全ての合計であり、それ以降の欄は
1650 個々の割り込みについての合計数である。
1651 .TP 
1652 \fIdisk_io: (2,0):(31,30,5764,1,2) (3,0):\fP...
1653 (メジャー番号, ディスクインデックス番号):(情報なし (noinfo), 読み込み回数, 読み込みブロック数, 書き出し回数, 書き出しブロック数)
1654 .br
1655 (Linux 2.4 のみ)
1656 .TP 
1657 \fIctxt 115315\fP
1658 コンテクスト・スイッチの延べ回数。
1659 .TP 
1660 \fIbtime 769041601\fP
1661 起動時刻、紀元 (Epoch; 1970\-01\-01 00:00:00 +0000 (UTC)) からの秒数。
1662 .TP 
1663 \fIprocesses 86031\fP
1664 システム起動時からの延べフォーク (fork) 数。
1665 .TP 
1666 \fIprocs_running 6\fP
1667 実行中状態のプロセス数 (Linux 2.5.45 以降)。
1668 .TP 
1669 \fIprocs_blocked 2\fP
1670 I/O 完了待ちで停止 (blocked) しているプロセス数 (Linux 2.5.45 以降)。
1671 .RE
1672 .TP 
1673 \fI/proc/swaps\fP
1674 使用中のスワップ領域。 \fBswapon\fP(8)  も参照すること。
1675 .TP 
1676 \fI/proc/sys\fP
1677 このディレクトリ (1.3.57 以降に存在) はカーネル変数に対応するいくつかの
1678 ファイルとサブディレクトリを含む。 これらの変数は読み出し可能である。
1679 また場合によっては \fI/proc\fP ファイルシステムや、 (非推奨の) システムコール
1680 \fBsysctl\fP(2) を用いて書き換えることもできる。
1681 .TP 
1682 \fI/proc/sys/abi\fP (Linux 2.4.10 以降)
1683 .\" On some systems, it is not present.
1684 このディレクトリにはアプリケーションのバイナリ情報が入ったファイルが置かれる。 更に詳しい情報は、 Linux カーネルソースファイル
1685 \fIDocumentation/sysctl/abi.txt\fP を参照すること。
1686 .TP 
1687 \fI/proc/sys/debug\fP
1688 このディレクトリは空の場合もある。
1689 .TP 
1690 \fI/proc/sys/dev\fP
1691 このディレクトリにはデバイス特有の情報 (たとえば \fIdev/cdrom/info\fP)  が含まれる。
1692 このディレクトリが空になっているシステムもある。
1693 .TP 
1694 \fI/proc/sys/fs\fP
1695 このディレクトリには、ファイルシステムに関連するカーネル変数用の
1696 ディレクトリとサブディレクトリが含まれる。
1697 .TP 
1698 \fI/proc/sys/fs/binfmt_misc\fP
1699 このディレクトリ以下のファイルについてのドキュメントは、 Linux カーネルソースの \fIDocumentation/binfmt_misc.txt\fP
1700 にある。
1701 .TP 
1702 \fI/proc/sys/fs/dentry\-state\fP (Linux 2.2 以降)
1703 このファイルには、ディレクトリキャッシュ (dcache) の状態に関する情報が 入っている。ファイルには、 \fInr_dentry\fP,
1704 \fInr_unused\fP, \fIage_limit\fP (秒単位の age), \fIwant_pages\fP (システムがリクエストしたページ数),
1705 ダミーの 2 つの値、 という 6 つの数字が書かれている。
1706 .RS
1707 .IP * 2
1708 \fInr_dentry\fP は割り当てられた dentry (dcache エントリ) の数である。 このフィールドは Linux 2.2
1709 では使用されない。
1710 .IP *
1711 \fInr_unused\fP は未使用の dentry 数である。
1712 .IP *
1713 .\" looks like this is unused in kernels 2.2 to 2.6
1714 \fIage_limit\fP は、メモリが不足している場合に次に dcache entry を再要求できるように なるまでの残り時間 (秒数) である。
1715 .IP *
1716 .\" looks like this is unused in kernels 2.2 to 2.6
1717 \fIwant_pages\fP は、カーネルが shrink_dcache_pages() を呼び出したが dcache がまだ縮小されていない場合に、0
1718 以外の値となる。
1719 .RE
1720 .TP 
1721 \fI/proc/sys/fs/dir\-notify\-enable\fP
1722 このファイルは \fBfcntl\fP(2)  に記述されている \fIdnotify\fP インターフェースをシステム全体で無効にしたり有効にしたりする。
1723 このファイルに値 0 が書かれている場合はインターフェースが無効になり、 値 1 の場合は有効になる。
1724 .TP 
1725 \fI/proc/sys/fs/dquot\-max\fP
1726 このファイルにはキャッシュされるディスク quota エントリの最大数が書かれている。 (2.4 系の)
1727 システムの中には、このファイルが存在しないものもある。 キャッシュされるディスク quota エントリの空きが非常に少なく、
1728 とても多くのシステムユーザーが同時に存在する場合、 この制限を上げるといいかもしれない。
1729 .TP 
1730 \fI/proc/sys/fs/dquot\-nr\fP
1731 このファイルには割り当てられているディスク quota のエントリ数と、 空いているディスク quota のエントリ数が書かれている。
1732 .TP 
1733 \fI/proc/sys/fs/epoll\fP (Linux 2.6.28 以降)
1734 このディレクトリには、ファイル \fImax_user_watches\fP がある。 これらは、 \fIepoll\fP
1735 インタフェースが消費するカーネルメモリ量を制限するのに使用できる。 詳細は \fBepoll\fP(7)  を参照。
1736 .TP 
1737 \fI/proc/sys/fs/file\-max\fP
1738 This file defines a system\-wide limit on the number of open files for all
1739 processes.  (See also \fBsetrlimit\fP(2), which can be used by a process to set
1740 the per\-process limit, \fBRLIMIT_NOFILE\fP, on the number of files it may
1741 open.)  If you get lots of error messages in the kernel log about running
1742 out of file handles (look for "VFS: file\-max limit <number>
1743 reached"), try increasing this value:
1744 .br
1745
1746 .br
1747 .nf
1748 \f(CW    echo 100000 > /proc/sys/fs/file\-max\fP
1749 .fi
1750
1751 \fIfile\-max\fP に書かれている値は、カーネル定数 \fBNR_OPEN\fP に制限される。
1752
1753 \fI/proc/sys/fs/file\-max\fP を増やした場合は、 \fI/proc/sys/fs/inode\-max\fP を新しい
1754 \fI/proc/sys/fs/file\-max\fP の値の 3\-4 倍に増やしておくこと。 こうしないと inode を使い果たしてしまうだろう。
1755
1756 特権プロセス (\fBCAP_SYS_ADMIN\fP) は \fIfile\-max\fP 上限を上書きできる。
1757 .TP 
1758 \fI/proc/sys/fs/file\-nr\fP
1759 This (read\-only) file contains three numbers: the number of allocated file
1760 handles (i.e., the number of files presently opened); the number of free
1761 file handles; and the maximum number of file handles (i.e., the same value
1762 as \fI/proc/sys/fs/file\-max\fP).  If the number of allocated file handles is
1763 close to the maximum, you should consider increasing the maximum.  Before
1764 Linux 2.6, the kernel allocated file handles dynamically, but it didn't free
1765 them again.  Instead the free file handles were kept in a list for
1766 reallocation; the "free file handles" value indicates the size of that
1767 list.  A large number of free file handles indicates that there was a past
1768 peak in the usage of open file handles.  Since Linux 2.6, the kernel does
1769 deallocate freed file handles, and the "free file handles" value is always
1770 zero.
1771 .TP 
1772 \fI/proc/sys/fs/inode\-max\fP
1773 このファイルには、メモリ内 inode の最大値が書かれている。 (2.4 系の) システムによっては、このファイルが存在しないかもしれない。 この値は
1774 \fIfile\-max\fP の値の 3\-4 倍にすべきである。 これは \fIstdin\fP, \fIstdout\fP, ネットワークソケットを扱うにも inode
1775 が必要なためである。 日常的に inode を使い果たしている場合は、この値を増やす必要がある。
1776 .TP 
1777 \fI/proc/sys/fs/inode\-nr\fP
1778 このファイルには、 \fIinode\-state\fP の最初の 2 つの値が書かれている。
1779 .TP 
1780 \fI/proc/sys/fs/inode\-state\fP
1781 このファイルには 7 個の値が書かれている: \fInr_inodes,\fP \fInr_free_inodes\fP, \fIpreshrink\fP と 4
1782 つのダミーの値である。 \fInr_inodes\fP はシステムが確保する inode の数である。 Linux は 1 度に 1 ページ分いっぱいに
1783 nr_inode を確保するので、この値が \fIinode\-max\fP より幾分大きくなることもある。 \fInr_free_inodes\fP は空いている
1784 inode の数を表す。 \fInr_inodes\fP > \fIinode\-max\fP の場合、 \fIpreshrink\fP は 0 以外の値になる。
1785 この場合システムは inode をさらに確保するのではなく、 inode リストを切り詰める必要がある。
1786 .TP 
1787 \fI/proc/sys/fs/inotify\fP (Linux 2.6.13 以降)
1788 このディレクトリには、ファイル \fImax_queued_events\fP, \fImax_user_instances\fP, and
1789 \fImax_user_watches\fP がある。これらは、 \fIinotify\fP インタフェースが消費するカーネルメモリ量を制限するのに利用できる。
1790 詳細は \fBinotify\fP(7)  を参照。
1791 .TP 
1792 \fI/proc/sys/fs/lease\-break\-time\fP
1793 このファイルは、ファイルを保持しているプロセスに対して カーネルがシグナルを送り、
1794 他のプロセスがそのファイルをオープンするのを待っていることを通知してから、 そのプロセスに対してカーネルがファイルのリース (lease, 貸し出し)
1795 (\fBfcntl\fP(2)  を参照) を許す猶予期間を指定する。 リースホルダ (lease holder: ファイルの貸し出しを受けているプロセス)
1796 が 猶予期間中にリースを削除するか階級を低くしない場合、 カーネルはファイルのリースを強制的に止める。
1797 .TP 
1798 \fI/proc/sys/fs/leases\-enable\fP
1799 このファイルはシステム全体でのファイルリース (\fBfcntl\fP(2)  を参照) を有効または無効にする。 ファイルに値 0
1800 が書き込まれている場合、リースは無効である。 0 以外の場合にはリースは有効である。
1801 .TP 
1802 \fI/proc/sys/fs/mqueue\fP (Linux 2.6.6 以降)
1803 このディレクトリにはファイル \fImsg_max\fP, \fImsgsize_max\fP, and \fIqueues_max\fP がある。これらは POSIX
1804 メッセージキューで使用されるリソースの 制御を行う。 詳細は \fBmq_overview\fP(7)  を参照。
1805 .TP 
1806 \fI/proc/sys/fs/overflowgid\fP と \fI/proc/sys/fs/overflowuid\fP
1807 これらのファイルにより、ユーザーは固定 UID と固定 GID の値を変更できる。 デフォルトは 65534 である。 Linux の UID と
1808 GID は 32 ビットであるが、 16 ビットの UID と GID しかサポートしないファイルシステムもある。
1809 このようなファイルシステムが書き込みを許可してマウントされた場合、 65535 を超える UID と GID は、
1810 ディスクに書き込まれる前にオーバーフロー値に変換される。
1811 .TP 
1812 \fI/proc/sys/fs/pipe\-max\-size\fP (Linux 2.6.35 以降)
1813 このファイルの値により、 \fBfcntl\fP(2) の \fBF_SETPIPE_SZ\fP
1814 操作で増やすことができるパイプ容量の上限値が定義される。この上限は非特権プロセスにのみ適用される。このファイルのデフォルト値は 1,048,576
1815 である。このファイルに設定した値は切り上げられて、実装側で利用するのに都合のよい値に変更される場合がある。切り上げられた値を確認するには、値を設定した後でこのファイルの内容を表示すればよい。このファイルに設定できる最小値はシステムのページサイズである。
1816 .TP 
1817 \fI/proc/sys/fs/protected_hardlinks\fP (Linux 3.6 以降)
1818 .\" commit 800179c9b8a1e796e441674776d11cd4c05d61d7
1819 When the value in this file is 0, no restrictions are placed on the creation
1820 of hard links (i.e., this is the historical behaviour before Linux 3.6).
1821 When the value in this file is 1, a hard link can be created to a target
1822 file only if one of the following conditions is true:
1823 .RS
1824 .IP * 3
1825 The caller has the \fBCAP_FOWNER\fP capability.
1826 .IP *
1827 The file system UID of the process creating the link matches the owner (UID)
1828 of the target file (as described in \fBcredentials\fP(7), a process's file
1829 system UID is normally the same as its effective UID).
1830 .IP *
1831 All of the following conditions are true:
1832 .RS 4
1833 .IP \(bu 3
1834 the target is a regular file;
1835 .IP \(bu
1836 the target file does not have its set\-user\-ID permission bit enabled;
1837 .IP \(bu
1838 the target file does not have both its set\-group\-ID and group\-executable
1839 permission bits enabled; and
1840 .IP \(bu
1841 the caller has permission to read and write the target file (either via the
1842 file's permissions mask or because it has suitable capabilities).
1843 .RE
1844 .RE
1845 .IP
1846 The default value in this file is 0.  Setting the value to 1 prevents a
1847 longstanding class of security issues caused by hard\-link\-based
1848 time\-of\-check, time\-of\-use races, most commonly seen in world\-writable
1849 directories such as \fI/tmp\fP.  The common method of exploiting this flaw is
1850 to cross privilege boundaries when following a given hard link (i.e., a root
1851 process follows a hard link created by another user).  Additionally, on
1852 systems without separated partitions, this stops unauthorized users from
1853 "pinning" vulnerable set\-user\-ID and set\-group\-ID files against being
1854 upgraded by the administrator, or linking to special files.
1855 .TP 
1856 \fI/proc/sys/fs/protected_symlinks\fP (Linux 3.6 以降)
1857 .\" commit 800179c9b8a1e796e441674776d11cd4c05d61d7
1858 When the value in this file is 0, no restrictions are placed on following
1859 symbolic links (i.e., this is the historical behaviour before Linux 3.6).
1860 When the value in this file is 1, symbolic links are followed only in the
1861 following circumstances:
1862 .RS
1863 .IP * 3
1864 the file system UID of the process following the link matches the owner
1865 (UID) of the symbolic link (as described in \fBcredentials\fP(7), a process's
1866 file system UID is normally the same as its effective UID);
1867 .IP *
1868 the link is not in a sticky world\-writable directory; or
1869 .IP *
1870 the symbolic link and and its parent directory have the same owner (UID)
1871 .RE
1872 .IP
1873 A system call that fails to follow a symbolic link because of the above
1874 restrictions returns the error \fBEACCES\fP in \fIerrno\fP.
1875 .IP
1876 The default value in this file is 0.  Setting the value to 1 avoids a
1877 longstanding class of security issues based on time\-of\-check, time\-of\-use
1878 races when accessing symbolic links.
1879 .TP 
1880 \fI/proc/sys/fs/suid_dumpable\fP (Linux 2.6.13 以降)
1881 .\" The following is based on text from Documentation/sysctl/kernel.txt
1882 このファイルの値により、set\-user\-ID されたバイナリや、 保護がかかった (protected) バイナリ / tainted な
1883 (汚染された; ライセンスがカーネルと適合しない) バイナリに対して、コアダンプファイルを 生成するかどうかが決定される。 以下の
1884 3つの値を指定することができる:
1885 .RS
1886 .TP 
1887 \fI0\ (default)\fP
1888 この値を指定すると、以前と同じ (Linux 2.6.13 より前の) 動作をする。 (\fBseteuid\fP(2), \fBsetgid\fP(2)
1889 などを呼び出すことや、set\-user\-ID や set\-group\-ID されたプログラムを 実行することで) 資格情報 (credentials)
1890 が変更されているプロセスや、 プロセスの実行バイナリの読み出し許可がないプロセスに対して、 コアダンプを生成しない。
1891 .TP 
1892 \fI1\ ("debug")\fP
1893 すべてのプロセスで、可能であればコアダンプを行う。 コアダンプファイルの所有者は、ダンプを行うプロセスのファイルシステム UID
1894 となり、セキュリティ上の考慮は行われない。 この値は、システムデバッグの場面だけを想定して設けられている。 ptrace のチェックも行われない。
1895 .TP 
1896 \fI2\ ("suidsafe")\fP
1897 通常はダンプされないようなバイナリ (上記の "0" 参照) を root だけが読み出し可能な形でダンプする。
1898 この場合、ユーザはそのコアダンプファイルを削除することはできるが、 読むことはできない。 セキュリティ上の理由から、このモードのコアダンプでは、
1899 既存のダンプファイルや他のファイルを上書きすることはない。 このモードは、管理者が通常の環境で問題を解析しようとする際に 適している。
1900 .IP
1901 .\" 9520628e8ceb69fa9a4aee6b57f22675d9e1b709
1902 .\" 54b501992dd2a839e94e76aa392c392b55080ce8
1903 Additionally, since Linux 3.6, \fI/proc/sys/kernel/core_pattern\fP must either
1904 be an absolute pathname or a pipe command, as detailed in \fBcore\fP(5).
1905 Warnings will be written to the kernel log if \fIcore_pattern\fP does not
1906 follow these rules, and no core dump will be produced.
1907 .RE
1908 .TP 
1909 このファイルはスーパブロックの値を制御する。
1910 この値はカーネルがマウントできるファイルシステムの最大値になる。 現在、 \fIsuper\-max\fP で許可されているファイルシステム数以上に
1911 マウントする必要がある場合は、この値を増加させるだけでよい。
1912 .TP 
1913 \fI/proc/sys/fs/super\-nr\fP
1914 このファイルには現在マウントされているファイルシステム数が書かれている。
1915 .TP 
1916 \fI/proc/sys/kernel\fP
1917 このディレクトリには、以下で説明する様々なカーネルパラメータを 制御するためのファイルが配置されている。
1918 .TP 
1919 \fI/proc/sys/kernel/acct\fP
1920 このファイルには 3 つの値が含まれている: \fIhighwater\fP, \fIlowwater\fP, \fIfrequency\fP である。
1921 BSD\-style process accounting が有効になっている場合、 これら 3 つの値が動作を制御する。
1922 ログファイルのあるファイルシステムの空き領域が \fIlowwater\fP パーセント以下になった場合は、ログ記録を一時停止する。 空き領域が
1923 \fIhighwater\fP パーセント以上になった場合に、ログ記録を再開する。 \fIfrequency\fP はカーネルが空き領域のチェックをする頻度である
1924 (単位は秒)。 デフォルトの値は、4, 2, 30 である。 つまり、空き領域が 2% 以下になるとログ記録を一時停止し、 空き領域が 4%
1925 以上となったときに再開する。 空き領域についての情報は 30 秒間有効である点に注意すること。
1926 .TP 
1927 \fI/proc/sys/kernel/cap_last_cap\fP (Linux 3.2 以降)
1928 \fBcapabilities\fP(7) 参照。
1929 .TP 
1930 \fI/proc/sys/kernel/cap\-bound\fP (Linux 2.2 to 2.6.24 以降)
1931 \fI/proc/sys/kernel/cap\-bound\fP このファイルにはカーネルの \fIcapability bounding set\fP (符号付き
1932 10 進数表現) の値が書かれている。 \fBexecve\fP(2)  中は、このセットとプロセスに許可されている権限の AND がとられる。 Linux
1933 2.6.25 以降では、システム全体のケーパビリティバウンディングセットは なくなり、スレッド単位のバウンディングセットに置き換えられた。
1934 \fBcapabilities\fP(7)  を参照。
1935 .TP 
1936 \fI/proc/sys/kernel/core_pattern\fP
1937 \fBcore\fP(5)  参照。
1938 .TP 
1939 \fI/proc/sys/kernel/core_uses_pid\fP
1940 \fBcore\fP(5)  参照。
1941 .TP 
1942 \fI/proc/sys/kernel/ctrl\-alt\-del\fP
1943 このファイルはキーボードの Ctrl\-Alt\-Del の扱いを制御する。 このファイルにある値が 0 の場合、 Ctrl\-Alt\-Del が捕捉されると
1944 \fBinit\fP(8)  プログラムに送られて、正しく再起動される。 値が 0 より大きい場合、Vulcan Nerve Pinch (tm)
1945 に反応して、 Linux はダーティバッファを同期させることなく、すぐに再起動を行う。 注意: プログラム (dosemu など) に "raw"
1946 モードのキーボードがある場合、 ctrl\-alt\-del はカーネルの tty レイヤーに到達する前に プログラムに遮断され、
1947 プログラムに送られてどのように扱うかが決められる。
1948 .TP 
1949 \fI/proc/sys/kernel/dmesg_restrict\fP (Linux 2.6.37 以降)
1950 .\" commit 620f6e8e855d6d447688a5f67a4e176944a084e8
1951 The value in this file determines who can see kernel syslog contents.  A
1952 value of 0 in this file imposes no restrictions.  If the value is 1, only
1953 privileged users can read the kernel syslog.  (See \fBsyslog\fP(2)  for more
1954 details.)  Since Linux 3.4, only users with the \fBCAP_SYS_ADMIN\fP capability
1955 may change the value in this file.
1956 .TP 
1957 \fI/proc/sys/kernel/domainname\fP と \fI/proc/sys/kernel/hostname\fP
1958 これらのファイルは、コマンド \fBdomainname\fP(1), \fBhostname\fP(1)  と全く同じ方法で、 マシンの NIS/YP
1959 ドメイン名とホスト名の設定に使える。 すなわち
1960
1961 .in +4n
1962 .nf
1963 #\fB echo \(aqdarkstar\(aq > /proc/sys/kernel/hostname\fP
1964 #\fB echo \(aqmydomain\(aq > /proc/sys/kernel/domainname\fP
1965 .fi
1966 .in
1967
1968 は、以下と同じ効果がある。
1969
1970 .in +4n
1971 .nf
1972 #\fB hostname \(aqdarkstar\(aq\fP
1973 #\fB domainname \(aqmydomain\(aq\fP
1974 .fi
1975 .in
1976
1977 注意: 典型的な darkstar.frop.org という名前には、 ホスト名 "darkstar" と DNS (Internet Domain
1978 Name Server)  ドメイン名 "frop.org" が含まれているが、DNS ドメイン名と NIS (Network Information
1979 Service) または YP (Yellow Pages) のドメイン名を混同してはならない。 一般にこれら 2 つのドメイン名は異なる。
1980 詳細な議論は、 \fBhostname\fP(1)  の man ページを参照すること。
1981 .TP 
1982 \fI/proc/sys/kernel/hotplug\fP
1983 このファイルはホットプラグ・ポリシー・エージェントのパスが書かれている。 このファイルのデフォルト値は \fI/sbin/hotplug\fP である。
1984 .TP 
1985 \fI/proc/sys/kernel/htab\-reclaim\fP
1986 (PowerPC のみ) このファイルを 0 以外の値に設定すると、 PowerPC htab (カーネルソースファイル
1987 \fIDocumentation/powerpc/ppc_htab.txt\fP 参照) を、システムがアイドルループになるたびに切り詰める。
1988 .TP 
1989 \fI/proc/sys/kernel/kptr_restrict\fP (Linux 2.6.38 以降)
1990 .\" 455cd5ab305c90ffc422dd2e0fb634730942b257
1991 .\" commit 411f05f123cbd7f8aa1edcae86970755a6e2a9d9
1992 .\" commit 620f6e8e855d6d447688a5f67a4e176944a084e8
1993 The value in this file determines whether kernel addresses are exposed via
1994 \fI/proc\fP files and other interfaces.  A value of 0 in this file imposes no
1995 restrictions.  If the value is 1, kernel pointers printed using the \fI%pK\fP
1996 format specifier will be replaced with zeros unless the user has the
1997 \fBCAP_SYSLOG\fP capability.  If the value is 2, kernel pointers printed using
1998 the \fI%pK\fP format specifier will be replaced with zeros regardless of the
1999 user's capabilities.  The initial default value for this file was 1, but the
2000 default was changed to 0 in Linux 2.6.39.  Since Linux 3.4, only users with
2001 the \fBCAP_SYS_ADMIN\fP capability can change the value in this file.
2002 .TP 
2003 \fI/proc/sys/kernel/l2cr\fP
2004 (PowerPC のみ) このファイルには G3 プロセッサボードの L2 キャッシュを制御するフラグが含まれる。 0 の場合、キャッシュは無効になる。
2005 0 以外の場合は有効になる。
2006 .TP 
2007 \fI/proc/sys/kernel/modprobe\fP
2008 このファイルには、カーネルモジュールローダへのパスが含まれる。 デフォルトの値は \fI/sbin/modprobe\fP
2009 である。このファイルは、\fBCONFIG_MODULES\fP オプション (Linux 2.6.26 以前では \fBCONFIG_KMOD\fP)
2010 を有効にしてカーネルが作成されている場合にのみ存在する。 このファイルについては、Linux カーネルソースファイル
2011 \fIDocumentation/kmod.txt\fP (カーネル 2.4 以前のみに存在) に記述されている。
2012 .TP 
2013 \fI/proc/sys/kernel/modules_disabled\fP (Linux 2.6.31 以降)
2014 .\" 3d43321b7015387cfebbe26436d0e9d299162ea1
2015 .\" From Documentation/sysctl/kernel.txt
2016 A toggle value indicating if modules are allowed to be loaded in an
2017 otherwise modular kernel.  This toggle defaults to off (0), but can be set
2018 true (1).  Once true, modules can be neither loaded nor unloaded, and the
2019 toggle cannot be set back to false.  The file is present only if the kernel
2020 is built with the \fBCONFIG_MODULES\fP option enabled.
2021 .TP 
2022 \fI/proc/sys/kernel/msgmax\fP
2023 このファイルは、System V メッセージキューに書き込まれる 1 つのメッセージの 最大バイト数を、システム全体で制限する。
2024 .TP 
2025 \fI/proc/sys/kernel/msgmni\fP (Linux 2.4 以降)
2026 このファイルはメッセージキュー識別子の最大数をシステム全体で制限する。
2027 .TP 
2028 \fI/proc/sys/kernel/msgmnb\fP
2029 このファイルは、 \fImsg_qbytes\fP の設定を初期化するシステム全体のパラメータで
2030 ある。 \fImsg_qbytes\fP は以降で作成されるメッセージキューで使われる。
2031 \fImsg_qbytes\fP 設定では、メッセージキューに書き込まれる最大バイト数を指定する。
2032 .TP 
2033 \fI/proc/sys/kernel/ostype\fP と \fI/proc/sys/kernel/osrelease\fP
2034 これらのファイルは文字列 \fI/proc/version\fP の各部分を与える。
2035 .TP 
2036 \fI/proc/sys/kernel/overflowgid\fP と \fI/proc/sys/kernel/overflowuid\fP
2037 これらのファイルは \fI/proc/sys/fs/overflowgid\fP と \fI/proc/sys/fs/overflowuid\fP
2038 を複製したものである。
2039 .TP 
2040 \fI/proc/sys/kernel/panic\fP
2041 このファイルはカーネル変数 \fIpanic_timeout\fP への読み出しと書き込みのアクセスを与える。 この値が 0 ならば、パニック時にカーネルは
2042 (無限) ループに入る。 0 でなければ、その秒数だけ待ってから自動的に再起動する。 ソフトウェア watchdog ドライバを使っている場合、
2043 推奨される設定は 60 である。
2044 .TP 
2045 \fI/proc/sys/kernel/panic_on_oops\fP (Linux 2.5.68 以降)
2046 このファイルは、oops や BUG が起こった場合のカーネルの動作を制御する。 ファイルに 0 が書かれている場合、システムは操作を続行しようとする。
2047 1 が書かれている場合、システムは (klogd が oops 出力を記録する時間を与えるために) 数秒間遅延した後、 panic を起こす。
2048 \fI/proc/sys/kernel/panic\fP ファイルも 0 でない場合、マシンは再起動される。
2049 .TP 
2050 \fI/proc/sys/kernel/pid_max\fP (Linux 2.5.34 以降)
2051 .\" Prior to 2.6.10, pid_max could also be raised above 32768 on 32-bit
2052 .\" platforms, but this broke /proc/[pid]
2053 .\" See http://marc.theaimsgroup.com/?l=linux-kernel&m=109513010926152&w=2
2054 このファイルは、PID をいくつで終了にするかを指定する (すなわち、このファイルの値は最大 PID より 1 大きい)。 このファイルのデフォルト値は
2055 32768 であり、 その場合には以前のカーネルと同じ PID の範囲になる。 32ビットのプラットフォームでは、 \fIpid_max\fP の最大値は
2056 32768 である。 64ビットのプラットフォームでは、 2^22 (\fBPID_MAX_LIMIT\fP, 約 4,000,000)
2057 までの任意の値を設定できる。
2058 .TP 
2059 \fI/proc/sys/kernel/powersave\-nap\fP (PowerPC のみ)
2060 このファイルにはフラグが書かれている。 フラグが設定されると、Linux\-PPC は 省電力の "nap" モードを使う。
2061 設定されない場合は、"doze" モードが使われる。
2062 .TP 
2063 \fI/proc/sys/kernel/printk\fP
2064 このファイルにある 4 つの値は、 \fIconsole_loglevel\fP, \fIdefault_message_loglevel\fP,
2065 \fIminimum_console_loglevel\fP, \fIdefault_console_loglevel\fP である。
2066 これらの値はエラーメッセージを表示したり記録したりする \fIprintk()\fP の動作に影響する。 各 loglevel の情報については、
2067 \fBsyslog\fP(2)  を参照すること。 優先度が \fIconsole_loglevel\fP 以上のメッセージは、コンソールに表示される。
2068 優先度が明示されていないメッセージは、優先度が \fIdefault_message_level\fP のときに表示される。
2069 \fIminimum_console_loglevel\fP は \fIconsole_loglevel\fP に設定できる最小 (最高) の値である。
2070 \fIdefault_console_loglevel\fP は \fIconsole_loglevel\fP のデフォルトの値である。
2071 .TP 
2072 \fI/proc/sys/kernel/pty\fP (Linux 2.6.4 以降)
2073 このディレクトリは、UNIX 98 疑似端末 (\fBpts\fP(4) を参照) の数に関連する
2074 2 つのファイルを含む。
2075 .TP 
2076 \fI/proc/sys/kernel/pty/max\fP
2077 このファイルは疑似端末の最大数を定義する。
2078 .TP 
2079 \fI/proc/sys/kernel/pty/nr\fP
2080 この読み出し専用のファイルは、現在いくつの疑似端末が使われているかを表す。
2081 .TP 
2082 \fI/proc/sys/kernel/random\fP
2083 このディレクトリは、ファイル \fI/dev/random\fP の操作を制御する様々なパラメータが書かれている。 詳細は \fBrandom\fP(4)
2084 を参照。
2085 .TP 
2086 \fI/proc/sys/kernel/real\-root\-dev\fP
2087 このファイルは Linux カーネルソースファイル \fIDocumentation/initrd.txt\fP に記述されている。
2088 .TP 
2089 \fI/proc/sys/kernel/reboot\-cmd\fP (Sparc のみ) 
2090 このファイルは SPARC ROM/Flash ブートローダに引き数を渡す方法を 提供しているように思われる。
2091 再起動後に何をするかを指定しているのだろうか?
2092 .TP 
2093 \fI/proc/sys/kernel/rtsig\-max\fP
2094 (2.6.7 までのカーネルにのみ存在する。 \fBsetrlimit\fP(2)  を参照すること)  このファイルはシステムで発行される POSIX
2095 real\-time (queued) signal の 最大数を調整するのに使用される。
2096 .TP 
2097 \fI/proc/sys/kernel/rtsig\-nr\fP
2098 (2.6.7 までのカーネルにのみ存在する)  このファイルは現在キューに入っている POSIX real\-time signal の数を表す。
2099 .TP 
2100 \fI/proc/sys/kernel/sched_rr_timeslice_ms\fP (Linux 3.9 以降)
2101 \fBsched_rr_get_interval\fP(2) 参照。
2102 .TP 
2103 \fI/proc/sys/kernel/sem\fP (Linux 2.4 以降)
2104 このファイルには System V IPC セマフォを制限する 4 つの値が書かれている。 これらのフィールドは次の順番に並んでいる:
2105 .RS
2106 .IP SEMMSL 8
2107 セマフォ集合ごとのセマフォ数の最大値。
2108 .IP SEMMNS 8
2109 システム全体での、全てのセマフォ集合におけるセマフォ数の制限。
2110 .IP SEMOPM 8
2111 \fBsemop\fP(2)  コールに指定されるオペレーション数の最大値。
2112 .IP SEMMNI 8
2113 システム全体でのセマフォ識別子の最大値。
2114 .RE
2115 .TP 
2116 \fI/proc/sys/kernel/sg\-big\-buff\fP
2117 このファイルは、汎用 SCSI デバイス (sg) のバッファサイズの最大値を表す。 今はこれを変更することはできないが、 コンパイル時に
2118 \fIinclude/scsi/sg.h\fP を編集して \fBSG_BIG_BUFF\fP の値を変えれば変更できる。
2119 ただし、この値を変更する理由はないだろう。
2120 .TP 
2121 \fI/proc/sys/kernel/shm_rmid_forced\fP (Linux 3.1 以降)
2122 .\" commit b34a6b1da371ed8af1221459a18c67970f7e3d53
2123 .\" See also Documentation/sysctl/kernel.txt
2124 If this file is set to 1, all System V shared memory segments will be marked
2125 for destruction as soon as the number of attached processes falls to zero;
2126 in other words, it is no longer possible to create shared memory segments
2127 that exist independently of any attached process.
2128 .IP
2129 The effect is as though a \fBshmctl\fP(2)  \fBIPC_RMID\fP is performed on all
2130 existing segments as well as all segments created in the future (until this
2131 file is reset to 0).  Note that existing segments that are attached to no
2132 process will be immediately destroyed when this file is set to 1.  Setting
2133 this option will also destroy segments that were created, but never
2134 attached, upon termination of the process that created the segment with
2135 \fBshmget\fP(2).
2136 .IP
2137 Setting this file to 1 provides a way of ensuring that all System V shared
2138 memory segments are counted against the resource usage and resource limits
2139 (see the description of \fBRLIMIT_AS\fP in \fBgetrlimit\fP(2))  of at least one
2140 process.
2141 .IP
2142 Because setting this file to 1 produces behavior that is nonstandard and
2143 could also break existing applications, the default value in this file is
2144 0.  Only set this file to 1 if you have a good understanding of the
2145 semantics of the applications using System V shared memory on your system.
2146 .TP 
2147 \fI/proc/sys/kernel/shmall\fP
2148 このファイルには System V 共有メモリの総ページ数の システム全体での制限が書かれている。
2149 .TP 
2150 \fI/proc/sys/kernel/shmmax\fP
2151 このファイルを通じて、(System V IPC) 共有メモリセグメントを作成するときの 最大サイズの実行時上限 (run\-time limit)
2152 を取得または設定できる。 現在は 1GB までの共有メモリセグメントが カーネルでサポートされている。 この値のデフォルトは \fBSHMMAX\fP
2153 である。
2154 .TP 
2155 \fI/proc/sys/kernel/shmmni\fP (Linux 2.4 以降)
2156 このファイルは、システム全体で作成可能な System V 共有メモリセグメント数を指定する。
2157 .TP 
2158 \fI/proc/sys/kernel/sysrq\fP
2159 このファイルは、SysRq キーにより起動が許可されている関数群を制御する ものである。デフォルトでは、ファイルの内容は 1 であり、
2160 これは起こり得る全ての SysRq リクエストが許可されることを意味する (古いバージョンのカーネルでは、SysRq はデフォルトでは無効になっており、
2161 実行時に明示的に有効にする必要があったが、今はそうではない)。 このファイルで指定可能な値は以下の通り。
2162
2163    0 \- sysrq を完全に無効にする
2164    1 \- sysrq の全ての関数を有効にする
2165   >1 \- 許可する sysrq 関数のビットマスク。内訳は以下の通り。
2166           2 \- コンソールのログ・レベルの制御を有効にする
2167           4 \- キーボードの制御を有効にする (SAK, unraw)
2168           8 \- プロセスなどのデバッグ・ダンプを有効にする
2169          16 \- sync コマンドを有効にする
2170          32 \- 読み出し専用での再マウントを有効にする
2171          64 \- プロセスへのシグナル発行を有効にする (term, kill, oom\-kill)
2172         128 \- リブート/電源オフを許可する
2173         256 \- 全てのリアルタイム・タスクの nice 値の変更を許可する
2174
2175 カーネル設定オプション \fBCONFIG_MAGIC_SYSRQ\fP が有効な場合のみ、このファイルは存在する。 詳細は、Linux
2176 カーネルソースファイル \fIDocumentation/sysrq.txt\fP を参照のこと。
2177 .TP 
2178 \fI/proc/sys/kernel/version\fP
2179 このファイルには、以下のような文字列が書かれている:
2180
2181     #5 Wed Feb 25 21:49:24 MET 1998
2182
2183 \&"#5" はこのソースで構築された 5 番目のカーネルであることを意味する。 その後にある日付はカーネルが構築された時刻を表す。
2184 .TP 
2185 \fI/proc/sys/kernel/threads\-max\fP (Linux 2.3.11 以降)
2186 このファイルは、システム全体で作成可能なスレッド数 (タスク数) の上限を指定する。
2187 .TP 
2188 \fI/proc/sys/kernel/zero\-paged\fP (PowerPC のみ) 
2189 このファイルはフラグを含む。 (0 以外の値で) 有効された場合、Linux\-PPC はアイドルループで pre\-zero page
2190 を行うので、get_free_pages の速度が向上する可能性がある。
2191 .TP 
2192 \fI/proc/sys/net\fP
2193 このディレクトリにはネットワーク関係の情報が入っている。 このディレクトリにあるファイルのいくつかについては、 \fBtcp\fP(7)  や
2194 \fBip\fP(7)  に説明がある。
2195 .TP 
2196 \fI/proc/sys/net/core/somaxconn\fP
2197 このファイルは \fBlisten\fP(2)  の \fIbacklog\fP 引き数の上限値を規定する。 詳細は \fBlisten\fP(2)
2198 のマニュアルページを参照。
2199 .TP 
2200 \fI/proc/sys/proc\fP
2201 このディレクトリは空の場合もある。
2202 .TP 
2203 \fI/proc/sys/sunrpc\fP
2204 このディレクトリはネットワークファイルシステム (NFS) への Sun remote procedure call (遠隔手続き呼び出し)
2205 をサポートする。 これが存在しないシステムもある。
2206 .TP 
2207 \fI/proc/sys/vm\fP
2208 このディレクトリにはメモリ管理の調整、バッファやキャッシュ管理のための ファイルがある。
2209 .TP 
2210 \fI/proc/sys/vm/drop_caches\fP (Linux 2.6.16 以降)
2211 このファイルに書き込みを行うことで、クリーンなキャッシュ、dentry、 inode をメモリ上から外し、そのメモリを解放する。
2212
2213 ページキャッシュを解放するには、 \fIecho 1 > /proc/sys/vm/drop_caches\fP とする。 dentry、inode
2214 を解放するには、 \fIecho 2 > /proc/sys/vm/drop_caches\fP とする。ページキャッシュ、dentry、inode
2215 を解放するには、 \fIecho 3 > /proc/sys/vm/drop_caches\fP とする。
2216
2217 この操作は非破壊的な操作で、ダーティな (dirty) オブジェクトは 解放されないので、この操作を行う際は最初に \fBsync\fP(8)
2218 を実行しておくべきである。
2219 .TP 
2220 \fI/proc/sys/vm/legacy_va_layout\fP (Linux 2.6.9 以降)
2221 .\" The following is from Documentation/filesystems/proc.txt
2222 0 以外の場合、新しい 32ビットメモリマッピング配置が無効になり、 カーネルは全てのプロセスに対して従来の (カーネル 2.4 の) 配置方法を
2223 使用する。
2224 .TP 
2225 \fI/proc/sys/vm/memory_failure_early_kill\fP (Linux 2.6.32 以降)
2226 .\" The following is based on the text in Documentation/sysctl/vm.txt
2227 このファイルは、カーネルが処理できない訂正不能なメモリエラー (通常はメモリモジュールでの 2 ビットエラー)
2228 がバックグラウンドでハードウェアにより検出された際に、プロセスをどのように kill するかを制御する。
2229 (ページの有効なコピーがディスク上にも存在するなど)
2230 いくつかの場面では、カーネルはアプリケーションに影響を与えずに透過的に故障を処理する。しかし、データの最新のコピーが他にはない場合には、データ破壊が波及するのを防ぐため、カーネルはプロセスを
2231 kill する。
2232
2233 このファイルは以下のいずれかの値を持つ。
2234 .RS
2235 .IP 1: 4
2236 データ破壊が検出されるとすぐに、故障が検出され復元できないページ (corrupted\-and\-not\-reloadable page)
2237 をマップしている全てのプロセスを kill する。この機能は、(カーネル内部で割り当てられたデータやスワップのキャッシュなど)
2238 少数のいくつかのタイプのページではサポートされていないが、大半のユーザページではこの機能は働く。
2239 .IP 0: 4
2240 データ破壊が検出されたページの全てのプロセスからの unmap のみを行い、そのページにアクセスしようとしたプロセスのみを kill する。
2241 .RE
2242 .IP
2243 この kill は \fIsi_code\fP に \fBBUS_MCEERR_AO\fP を設定した \fBSIGBUS\fP
2244 シグナルを使って行われる。プロセス側では必要であればこのシグナルを処理することができる。詳細は \fBsigaction\fP(2) を参照。
2245
2246 この機能は、高度なマシンチェック機構を持ったアーキテクチャ/プラットフォームにおいてのみ有効であり、ハードウェア機能にも依存している。
2247
2248 アプリケーションは \fBprctl\fP(2) の \fBPR_MCE_KILL\fP 操作を使って個別に
2249 \fImemory_failure_early_kill\fP の設定を上書きすることができる。
2250 .IP
2251 カーネルの設定で \fBCONFIG_MEMORY_FAILURE\fP が有効になっている場合にのみ
2252 存在する。
2253 .TP 
2254 \fI/proc/sys/vm/memory_failure_recovery\fP (Linux 2.6.32 以降)
2255 .\" The following is based on the text in Documentation/sysctl/vm.txt
2256 メモリ故障回復 (memory failure recovery) を有効にする
2257  (プラットフォームがサポートしている場合)
2258 .RS
2259 .IP 1: 4
2260 回復を試みる。
2261 .IP 0: 4
2262 メモリ故障時には常に panic を起こす。
2263 .RE
2264 .IP
2265 カーネルの設定で \fBCONFIG_MEMORY_FAILURE\fP が有効になっている場合にのみ
2266 存在する。
2267 .TP 
2268 \fI/proc/sys/vm/oom_dump_tasks\fP (Linux 2.6.25 以降)
2269 .\" The following is from Documentation/sysctl/vm.txt
2270 カーネルが OOM\-killing を実行する際に、システム全体のタスク・ダンプ (カーネルスレッドを除く) を生成するかを制御する。
2271 ダンプには、タスク (スレッド、プロセス) 毎に以下の情報が出力される: スレッド ID、実ユーザ ID、スレッドグループ ID (プロセス ID)、
2272 仮想メモリサイズ、Resident Set Size (実メモリ上に存在するページサイズ)、 タスクがスケジューリングされた CPU、 oom_adj
2273 スコア (\fI/proc/[pid]/oom_adj\fP の説明を参照)、コマンド名。 このダンプ情報は、なぜ OOM\-killer
2274 が起動されたかを知り、 その原因となったならず者のタスクを特定するのに役に立つ。
2275
2276 このファイルの内容が値 0 の場合、ダンプ情報の出力は行われない。 タスクが何千もある非常に巨大なシステムでは、
2277 各々のタスクについてメモリ状態をダンプするのは適切でないかもしれない。 そのようなシステムでは、ダンプ情報が必要でもないときに メモリ不足 (OOM)
2278 の状況で性能面の不利益が起こらないようにすべきだろう。
2279
2280 このファイルの内容が 0 以外の場合、 OOM\-killer が実際にメモリを占有したタスクを kill する度に ダンプ情報が出力される。
2281
2282 デフォルト値は 0 である。
2283 .TP 
2284 \fI/proc/sys/vm/oom_kill_allocating_task\fP (Linux 2.6.24 以降)
2285 .\" The following is from Documentation/sysctl/vm.txt
2286 このファイルは、メモリ不足 (OOM) の状況が起こった際に、 メモリ不足のきっかけとなったタスクを kill するかどうかを制御する。
2287
2288 このファイルが 0 に設定された場合、 OOM\-killer はタスクリスト全体をスキャンし、経験則に基づき kill するタスクを選択する。
2289 通常は、kill した場合に多くのメモリが解放できる、 ならず者のメモリ占有タスクが選択される。
2290
2291 このファイルが 0 以外に設定された場合、 OOM\-killer はメモリ不足の状況が発生するきっかけとなったタスクを 単純に kill
2292 するだけである。 これにより、たいていは重い処理となるタスクリストのスキャンを回避できる。
2293
2294 \fI/proc/sys/vm/panic_on_oom\fP が 0 以外の場合、
2295 \fI/proc/sys/vm/oom_kill_allocating_task\fP にどのような値が入っていたとしても、
2296 \fI/proc/sys/vm/panic_on_oom\fP の方が優先される。
2297
2298 デフォルト値は 0 である。
2299 .TP 
2300 \fI/proc/sys/vm/overcommit_memory\fP
2301 このファイルにはカーネル仮想メモリのアカウントモードが書かれている。 値は以下の通り:
2302 .RS
2303 .IP
2304 0: 発見的なオーバーコミット (heuristic overcommit) (これがデフォルトである)
2305 .br
2306 1: 常にオーバーコミットし、チェックしない。
2307 .br
2308 2: 常にチェックし、オーバーコミットしない。
2309 .RE
2310 .IP
2311 モード 0 では、 \fBMAP_NORESERVE\fP を設定して呼び出された \fBmmap\fP(2)  はチェックされない。
2312 またデフォルトのチェックはとても脆弱で、 プロセスを "OOM\-kill" してしまうリスクを引き起こす。 Linux 2.4 では 0
2313 以外の値はモード 1 を意味する。 (Linux 2.6 以降で利用可能な) モード 2 では、 システム上の仮想アドレス空間の合計が (SS +
2314 RAM*(r/100)) に制限されている。 ここで、SS はスワップ空間のサイズ、 RAM は物理メモリのサイズ、r はファイル
2315 \fI/proc/sys/vm/overcommit_ratio\fP の内容である。
2316 .TP 
2317 \fI/proc/sys/vm/overcommit_ratio\fP
2318 \fI/proc/sys/vm/overcommit_memory\fP の説明を参照すること。
2319 .TP 
2320 \fI/proc/sys/vm/panic_on_oom\fP (Linux 2.6.18 以降)
2321 .\" The following is adapted from Documentation/sysctl/vm.txt
2322 このファイルは、メモリ不足時にカーネルパニックを 起こすか起こさないかを制御する。
2323
2324 このファイルに値 0 を設定すると、 カーネルの OOM\-killer がならず者のプロセスを kill する。 普通は、OOM\-killer
2325 がならず者のプロセスを kill することができ、 システムは何とか動き続けることができる。
2326
2327 このファイルに値 1 を設定すると、 メモリ不足の状況が発生すると、カーネルは普通はパニックする。 しかしながら、プロセスが メモリポリシー
2328 (\fBmbind\fP(2)  の \fBMPOL_BIND\fP)  や cpusets (\fBcpuset\fP(7))
2329 を使って特定のノードへのメモリ割り当てを制限していて、 それらのノードでメモリ枯渇状態に至った場合は、 一つのプロセスが OOM\-killer により
2330 kill されるだけかもしれない。 この場合には、カーネルパニックは発生しない。 なぜなら、他のノードのメモリには空きがあるかもしれず、
2331 したがって、システム全体としてはメモリ不足の状況にまだ達していないかも しれないからである。
2332
2333 このファイルに値 2 を設定すると、 メモリ不足の状況が発生するとカーネルは常にパニックを起こす。
2334
2335 デフォルト値は 0 である。 1 と 2 はクラスタリングのフェイルオーバー用である。 フェイルオーバーの方針に応じてどちらかの値を選択すること。
2336 .TP 
2337 \fI/proc/sys/vm/swappiness\fP
2338 .\" The following is from Documentation/sysctl/vm.txt
2339 このファイルの値により、カーネルがどの程度激しくメモリページの スワップを行う
2340 かが制御される。 大きな値ほどスワップが激しくなり、小さい値ほど激しくなくなる。
2341 デフォルト値は 60 である。
2342 .TP 
2343 \fI/proc/sysrq\-trigger\fP (Linux 2.4.21 以降)
2344 このファイルに文字 character を書き込むと、 キーボードから ALT\-SysRq\-<character> を入力した場合と
2345 同じ SysRq 関数が起動される (\fI/proc/sys/kernel/sysrq\fP の説明を参照)。 通常、このファイルへ書き込みができるのは
2346 \fIroot\fP だけである。詳細については、Linux カーネルソースファイルの \fIDocumentation/sysrq.txt\fP を参照のこと。
2347 .TP 
2348 \fI/proc/sysvipc\fP
2349 疑似ファイル \fImsg\fP, \fIsem\fP, \fIshm\fP を含むサブディレクトリ。 これらのファイルは、現在システム上に存在する System V
2350 プロセス間通信 (Interprocess Communication, IPC) オブジェクト (それぞれ: メッセージキュー、セマフォ、共有メモリ)
2351 のリストであり、 \fBipcs\fP(1)  で取得できる情報と同じものを提供する。 これらのファイルにはヘッダがあり、理解しやすいように (1 行につき
2352 1 個の IPC オブジェクトの形式で) フォーマットされている。 \fBsvipc\fP(7)
2353 にはこれらのファイルから分かる情報の詳細な背景が書かれている。
2354 .TP 
2355 \fI/proc/tty\fP
2356 疑似ファイルを含むサブディレクトリ。 tty ドライバとライン設定 (line discipline) の書かれた サブディレクトリも含まれる。
2357 .TP 
2358 \fI/proc/uptime\fP
2359 このファイルは システム起動時から経過した時間 (秒) と アイドル (idle) しているプロセスが消費した時間 (秒) の 2 つの数を含む。
2360 .TP 
2361 \fI/proc/version\fP
2362 現在稼働しているカーネルのバージョン識別子である文字列。 これには \fI/proc/sys/ostype\fP,
2363 \fI/proc/sys/osrelease\fP, \fI/proc/sys/version\fP の内容が含まれる。 たとえばこのように:
2364 .nf
2365 .in -2
2366 \f(CWLinux version 1.0.9 (quinlan@phaze) #1 Sat May 14 01:51:54 EDT 1994\fP
2367 .in +2
2368 .fi
2369 .\" FIXME Document /proc/timer_list
2370 .\" .TP
2371 .\" .IR /proc/timer_list " (since Linux 2.6.21)"
2372 .\" See the 2.6.21 Change log
2373 .\" FIXME Document /proc/timer_stats
2374 .\" .TP
2375 .\" .IR /proc/timer_stats " (since Linux 2.6.21)"
2376 .\" See the 2.6.21 Change log
2377 .TP 
2378 \fI/proc/vmstat\fP (Linux 2.6 以降)
2379 このファイルは仮想メモリの様々な統計情報を表示する。
2380 .TP 
2381 \fI/proc/zoneinfo\fP (Linux 2.6.13 以降)
2382 .\" FIXME more should be said about /proc/zoneinfo
2383 このファイルはメモリのゾーン (memory zone) に関する情報を表示する。 仮想メモリの振舞いを分析するのに役立つ。
2384 .SH 注意
2385 ほとんどの文字列 (たとえば環境変数やコマンド行) は内部表現のままなので、
2386 各フィールドは NULL バイト (\(aq\e0\(aq) で区切られている。だから、
2387 \fIod \-c\fP や \fItr "\e000" "\en"\fP を使えば、それらはより読みやすくなる。
2388 また \fIecho \`cat <file>\`\fP でもよい。
2389
2390 .\" .SH ACKNOWLEDGEMENTS
2391 .\" The material on /proc/sys/fs and /proc/sys/kernel is closely based on
2392 .\" kernel source documentation files written by Rik van Riel.
2393 このマニュアルは不完全であり、たぶん不正確で、しばしば更新される必要がある。
2394 .SH 関連項目
2395 \fBcat\fP(1), \fBdmesg\fP(1), \fBfind\fP(1), \fBfree\fP(1), \fBps\fP(1), \fBtr\fP(1),
2396 \fBuptime\fP(1), \fBchroot\fP(2), \fBmmap\fP(2), \fBreadlink\fP(2), \fBsyslog\fP(2),
2397 \fBslabinfo\fP(5), \fBhier\fP(7), \fBtime\fP(7), \fBarp\fP(8), \fBhdparm\fP(8),
2398 \fBifconfig\fP(8), \fBinit\fP(8), \fBlsmod\fP(8), \fBlspci\fP(8), \fBmount\fP(8),
2399 \fBnetstat\fP(8), \fBprocinfo\fP(8), \fBroute\fP(8), \fBsysctl\fP(8)
2400
2401 Linux カーネルのソースファイル: \fIDocumentation/filesystems/proc.txt\fP,
2402 \fIDocumentation/sysctl/vm.txt\fP
2403 .SH この文書について
2404 この man ページは Linux \fIman\-pages\fP プロジェクトのリリース 3.52 の一部
2405 である。プロジェクトの説明とバグ報告に関する情報は
2406 http://www.kernel.org/doc/man\-pages/ に書かれている。