OSDN Git Service

(split) DP: release pages (catch up to 3.50).
[linuxjm/LDP_man-pages.git] / release / man5 / proc.5
index a38257c..6d4828f 100644 (file)
@@ -6,6 +6,7 @@
 .\" and System V IPC (as well as various other) additions from
 .\" Michael Kerrisk <mtk.manpages@gmail.com>
 .\"
+.\" %%%LICENSE_START(GPLv2+_DOC_FULL)
 .\" This is free documentation; you can redistribute it and/or
 .\" modify it under the terms of the GNU General Public License as
 .\" published by the Free Software Foundation; either version 2 of
@@ -22,9 +23,9 @@
 .\" GNU General Public License for more details.
 .\"
 .\" You should have received a copy of the GNU General Public
-.\" License along with this manual; if not, write to the Free
-.\" Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111,
-.\" USA.
+.\" License along with this manual; if not, see
+.\" <http://www.gnu.org/licenses/>.
+.\" %%%LICENSE_END
 .\"
 .\" Modified 1995-05-17 by faith@cs.unc.edu
 .\" Minor changes by aeb and Marty Leisner (leisner@sdsp.mc.xerox.com).
@@ -63,7 +64,7 @@
 .\" This file was generated with po4a. Translate the source file.
 .\"
 .\"*******************************************************************
-.TH PROC 5 2012\-05\-03 Linux "Linux Programmer's Manual"
+.TH PROC 5 2013\-02\-11 Linux "Linux Programmer's Manual"
 .SH 名前
 proc \- プロセスの情報を含む擬似ファイルシステム
 .SH 説明
@@ -244,8 +245,12 @@ flags:  01002002
 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。
 .TP 
 \fI/proc/[pid]/limits\fP (kernel 2.6.24 以降)
-このファイルは、そのプロセスの各リソース制限について、 ソフト・リミット、ハード・リミット、計測単位を表示する (\fBgetrlimit\fP(2)
-参照)。 このファイルは、プロセスの実 UID を持つものだけが読み出すことができる。
+.\" commit 3036e7b490bf7878c6dae952eec5fb87b1106589
+This file displays the soft limit, hard limit, and units of measurement for
+each of the process's resource limits (see \fBgetrlimit\fP(2)).  Up to and
+including Linux 2.6.35, this file is protected to only allow reading by the
+real UID of the process.  Since Linux 2.6.36, this file is readable by all
+users on the system.
 .TP 
 \fI/proc/[pid]/maps\fP
 現在マップされているメモリ領域とそのアクセスパーミッションを含む。
@@ -353,7 +358,7 @@ unbindable
 root に dominant peer group がない場合は、 "master:X" フィールドだけが存在し、
 "propagate_from:X" フィールドは存在しない。
 
-mount propagation の詳細については、カーネルソースツリー内の
+mount propagation の詳細については、 Linux カーネルソースツリー内の
 \fIDocumentation/filesystems/sharedsubtree.txt\fP を参照。
 .TP 
 \fI/proc/[pid]/mounts\fP (Linux 2.4.19 以降)
@@ -425,15 +430,18 @@ 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
 \fBnuma\fP(7)  を参照。
 .TP 
 \fI/proc/[pid]/oom_adj\fP (Linux 2.6.11 以降)
-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
-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
-\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
-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
-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
-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。
-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。 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。
+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
+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
+\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
+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
+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
+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。
+.IP
+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。
+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。
+.IP
+Since Linux 2.6.36, use of this file is deprecated in favor of
+\fI/proc/[pid]/oom_score_adj\fP.
 .TP 
 \fI/proc/[pid]/oom_score\fP (Linux 2.6.11 以降)
 .\" See mm/oom_kill.c::badness() in the 2.6.25 sources
@@ -459,10 +467,59 @@ OOM\-killer により選択される 可能性が高いことを意味する。
 そのプロセスが direct hardware access を行っているか (\-)。
 .RE
 .IP
+\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。
+.TP 
+\fI/proc/[pid]/oom_score_adj\fP (Linux 2.6.36 以降)
+.\" Text taken from 3.7 Documentation/filesystems/proc.txt
+This file can be used to adjust the badness heuristic used to select which
+process gets killed in out\-of\-memory conditions.
+
+The badness heuristic assigns a value to each candidate task ranging from 0
+(never kill) to 1000 (always kill) to determine which process is targeted.
+The units are roughly a proportion along that range of allowed memory the
+process may allocate from, based on an estimation of its current memory and
+swap use.  For example, if a task is using all allowed memory, its badness
+score will be 1000.  If it is using half of its allowed memory, its score
+will be 500.
+
+There is an additional factor included in the badness score: root processes
+are given 3% extra memory over other tasks.
+
+The amount of "allowed" memory depends on the context in which the
+OOM\-killer was called.  If it is due to the memory assigned to the
+allocating task's cpuset being exhausted, the allowed memory represents the
+set of mems assigned to that cpuset (see \fBcpuset\fP(7)).  If it is due to a
+mempolicy's node(s) being exhausted, the allowed memory represents the set
+of mempolicy nodes.  If it is due to a memory limit (or swap limit) being
+reached, the allowed memory is that configured limit.  Finally, if it is due
+to the entire system being out of memory, the allowed memory represents all
+allocatable resources.
+
+The value of \fIoom_score_adj\fP is added to the badness score before it is
+used to determine which task to kill.  Acceptable values range from \-1000
+(OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX).  This allows user space to
+control the preference for OOM\-killing, ranging from always preferring a
+certain task or completely disabling it from OOM\-killing.  The lowest
+possible value, \-1000, is equivalent to disabling OOM\-killing entirely for
+that task, since it will always report a badness score of 0.
+
+Consequently, it is very simple for user space to define the amount of
+memory to consider for each task.  Setting a \fIoom_score_adj\fP value of +500,
+for example, is roughly equivalent to allowing the remainder of tasks
+sharing the same system, cpuset, mempolicy, or memory controller resources
+to use at least 50% more memory.  A value of \-500, on the other hand, would
+be roughly equivalent to discounting 50% of the task's allowed memory from
+being considered as scoring against the task.
+
+For backward compatibility with previous kernels, \fI/proc/[pid]/oom_adj\fP can
+still be used to tune the badness score.  Its value is scaled linearly with
+\fIoom_score_adj\fP.
+
 .\" FIXME Describe /proc/[pid]/pagemap
 .\"       Added in 2.6.25
 .\"       CONFIG_PROC_PAGE_MONITOR
-\fIoom_score\fP は、そのプロセスの \fIoom_adj\fP 設定で規定されるビットシフト調整にも影響を与える。
+Writing to \fI/proc/[pid]/oom_score_adj\fP or \fI/proc/[pid]/oom_adj\fP will
+change the other with its scaled value.
 .TP 
 \fI/proc/[pid]/root\fP
 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
@@ -518,72 +575,72 @@ Private_Dirty:        0 kB
 .RS
 .TP  12
 \fIpid\fP %d
-プロセス ID。
+(1) プロセス ID。
 .TP 
 \fIcomm\fP %s
-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。
+(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。
 .TP 
 \fIstate\fP %c
-"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
+(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
 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
 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
 (stopped)、 W はページング中 (paging) を表している。
 .TP 
 \fIppid\fP %d
-親プロセスの PID。
+(4) 親プロセスの PID。
 .TP 
 \fIpgrp\fP %d
-プロセスのプロセスグループ ID。
+(5) プロセスのプロセスグループ ID。
 .TP 
 \fIsession\fP %d
-プロセスのセッション ID。
+(6) プロセスのセッション ID。
 .TP 
 \fItty_nr\fP %d
-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
+(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
 に格納される)。
 .TP 
 \fItpgid\fP %d
 .\" This field and following, up to and including wchan added 0.99.1
-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。
+(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。
 .TP 
 \fIflags\fP %u (Linux 2.6.22 より前は %lu)
-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_* 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。
+(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_*
+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。
 .TP 
 \fIminflt\fP %lu
-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。
+(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。
 .TP 
 .\" field 11
 \fIcminflt\fP %lu
-(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。
+(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。
 .TP 
 \fImajflt\fP %lu
-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。
+(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。
 .TP 
 \fIcmajflt\fP %lu
-(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。
+(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。
 .TP 
 \fIutime\fP %lu
-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
+(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
 (\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
-の実行に消費された時間)  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。
+の実行に消費された時間)
\82\82å\90«ã\81¾ã\82\8cã\82\8bã\80\82ã\81\93ã\82\8cã\81¯ã\80\81ã\82²ã\82¹ã\83\88æ\99\82é\96\93ã\81®ã\83\95ã\82£ã\83¼ã\83«ã\83\89ã\82\92èª\8dè­\98ã\81\97ã\81ªã\81\84ã\82¢ã\83\97ã\83ªã\82±ã\83¼ã\82·ã\83§ã\83³ã\81«ã\81\8aã\81\84ã\81¦ã\80\81ã\82²ã\82¹ã\83\88æ\99\82é\96\93å\88\86ã\82\92è¨\88ç®\97ã\81«å\85¥ã\82\8cæ\90\8dã\81­ã\81ªã\81\84ã\82\88ã\81\86ã\81«ã\81\99ã\82\8bã\81\9fã\82\81ã\81§ã\81\82ã\82\8bã\80\82
 .TP 
 \fIstime\fP %lu
-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 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)。
+(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
+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)。
 .TP 
 \fIcutime\fP %ld
-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 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。)
+(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
+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。)
 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。
 .TP 
 \fIcstime\fP %ld
-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 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)。
+(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
+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)。
 .TP 
 \fIpriority\fP %ld
-(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
+(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
 \fIpolicy ;\fP \fBsched_setscheduler\fP(2) 参照) で動作しているプロセスでは、 この
 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
 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
@@ -598,9 +655,6 @@ 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
 .TP 
 \fInice\fP %ld
 .\" Back in kernel 1.2 days things were different.
-nice 値 (\fBsetpriority\fP(2) 参照)。 19 (最低優先) から \-20 (最高優先)
-の範囲の値である。
-.TP 
 .\" .TP
 .\" \fIcounter\fP %ld
 .\" The current maximum size in jiffies of the process's next timeslice,
@@ -610,99 +664,107 @@ nice 値 (\fBsetpriority\fP(2) 参照)。 19 (最低優先) から \-20 (最高
 .\" \fItimeout\fP %u
 .\" The time in jiffies of the process's next timeout.
 .\" timeout was removed sometime around 2.1/2.2
+(19) nice 値 (\fBsetpriority\fP(2) 参照)。 19 (最低優先) から \-20 (最高優先)
+の範囲の値である。
+.TP 
 \fInum_threads\fP %ld
-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 0
-にハードコードされていた。
+(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
+にハードコードされていた。
 .TP 
 .\" field 21
 \fIitrealvalue\fP %ld
-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
+(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
 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。
 .TP 
 \fIstarttime\fP %llu (Linux 2.6 より前は %lu)
-プロセスの起動時刻。システムが起動した時刻が起点 (単位 jiffies)。
+(22) The time the process started after system boot.  In kernels before
+Linux 2.6, this value was expressed in jiffies.  Since Linux 2.6, the value
+is expressed in clock ticks (divide by \fIsysconf(_SC_CLK_TCK)\fP).
 .TP 
 \fIvsize\fP %lu
-仮想メモリのサイズ。単位はバイト。
+(23) 仮想メモリのサイズ。単位はバイト。
 .TP 
 \fIrss\fP %ld
-Resident Set Size。プロセスが持っている実メモリ上のページ数。
+(24) Resident Set Size。プロセスが持っている実メモリ上のページ数。
 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
 スワップアウトされたページの数は含んでいない。
 .TP 
 \fIrsslim\fP %lu
-このプロセスの rss の現在のソフト・リミット (バイト単位)。 \fBgetpriority\fP(2)  の \fBRLIMIT_RSS\fP の説明を参照。
+(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
+の説明を参照。
 .TP 
 \fIstartcode\fP %lu
-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。
+(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。
 .TP 
 \fIendcode\fP %lu
-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。
+(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。
 .TP 
 \fIstartstack\fP %lu
-スタックの開始アドレス (すなわち、スタックの底)。
+(28) スタックの開始アドレス (すなわち、スタックの底)。
 .TP 
 \fIkstkesp\fP %lu
-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。
+(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。
 .TP 
 \fIkstkeip\fP %lu
-現在の EIP (インストラクションポインタ) の値。
+(30) 現在の EIP (インストラクションポインタ) の値。
 .TP 
 .\" field 31
 \fIsignal\fP %lu
-処理待ちの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。
-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。
+(31) 処理待ちの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。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。
 .TP 
 \fIblocked\fP %lu
-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。
+(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。
 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。
 .TP 
 \fIsigignore\fP %lu
-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。
+(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。
 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。
 .TP 
 \fIsigcatch\fP %lu
-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。
+(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。
 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。
 .TP 
 \fIwchan\fP %lu
-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、
+(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、
 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
 (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
 しているならば、 \fIps \-l\fP して WCHAN フィールドを見よ)。
 .TP 
 \fInswap\fP %lu
 .\" nswap was added in 2.0
-スワップされたページ数 (メンテナンスされていない)。
+(36) スワップされたページ数 (メンテナンスされていない)。
 .TP 
 \fIcnswap\fP %lu
 .\" cnswap was added in 2.0
-子プロセスの \fInswap\fP の累計 (メンテナンスされていない)。
+(37) 子プロセスの \fInswap\fP の累計 (メンテナンスされていない)。
 .TP 
 \fIexit_signal\fP %d (Linux 2.1.22 以降)
-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。
+(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。
 .TP 
 \fIprocessor\fP %d (Linux 2.2.8 以降)
-このプロセスを最後に実行した CPU の番号。
+(39) このプロセスを最後に実行した CPU の番号。
 .TP 
 \fIrt_priority\fP %u (Linux 2.5.19 以降; Linux 2.6.22 より前は %lu)
-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、
+(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、
 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)。
 .TP 
 .\" field 41
 \fIpolicy\fP %u (Linux 2.5.19 以降; Linux 2.6.22 より前は %lu)
-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 SCHED_*
-定数を使ってデコードすればよい。
+(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
+SCHED_* 定数を使ってデコードすればよい。
 .TP 
 \fIdelayacct_blkio_ticks\fP %llu (Linux 2.6.18 以降)
-(clock tick (100分の1秒) 単位での) ブロック I/O の総遅延量。
+(42) (clock tick (100分の1秒) 単位での) ブロック I/O の総遅延量。
 .TP 
 \fIguest_time\fP %lu (Linux 2.6.24 以降)
-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
+(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
 (\fIsysconf(_SC_CLK_TCK)\fP で割った値が表示される)。
 .TP 
 \fIcguest_time\fP %ld (Linux 2.6.24 以降)
-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 で割った値が表示される)。
+(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
+で割った値が表示される)。
 .RE
 .TP 
 \fI/proc/[pid]/statm\fP
@@ -716,7 +778,7 @@ size       プログラムサイズの総計
            (\fI/proc/[pid]/status\fP の VmSize と同じ)
 resident   実メモリ上に存在するページ
            (\fI/proc/[pid]/status\fP の VmRSS と同じ)
-share      共有ページ (共有マッピングされているページ)
+share      共有ページ (ファイルと関連付けられているページ)
 text       テキスト (コード)
 lib        ライブラリ (Linux 2.6 では未使用)
 data       データ + スタック
@@ -900,14 +962,14 @@ cat /lib/modules/$(uname \-r)/build/.config
 \fI/proc/cpuinfo\fP
 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
 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。
-SMP マシンでは各 CPU についての情報が書かれている。
+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。
 .TP 
 \fI/proc/devices\fP
 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、
 カーネルとの整合性を保つことができる。
 .TP 
 \fI/proc/diskstats\fP (Linux 2.5.69 以降)
-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、カーネルソースファイル
+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
 \fIDocumentation/iostats.txt\fP を参照すること。
 .TP 
 \fI/proc/dma\fP
@@ -1008,8 +1070,207 @@ Linux 2.4 における I/O メモリマップ。
 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。
 .TP 
 \fI/proc/meminfo\fP
-このファイルはシステム上のメモリ使用量に関する統計情報を表示する。 \fBfree\fP(1)  は、このファイルを参照し、
-システムの未使用および使用中のメモリ量 (物理メモリとスワップ) と、 カーネルに使われている共有メモリとバッファの情報を報告する。
+This file reports statistics about memory usage on the system.  It is used
+by \fBfree\fP(1)  to report the amount of free and used memory (both physical
+and swap)  on the system as well as the shared memory and buffers used by
+the kernel.  Each line of the file consists of a parameter name, followed by
+a colon, the value of the parameter, and an option unit of measurement
+(e.g., "kB").  The list below describes the parameter names and the format
+specifier required to read the field value.  Except as noted below, all of
+the fields have been present since at least Linux 2.6.0.  Some fileds are
+only displayed if the kernel was configured with various options; those
+dependencies are noted in the list.
+.RS
+.TP 
+\fIMemTotal\fP %lu
+Total usable RAM (i.e. physical RAM minus a few reserved bits and the kernel
+binary code).
+.TP 
+\fIMemFree\fP %lu
+The sum of \fILowFree\fP+\fIHighFree\fP.
+.TP 
+\fIBuffers\fP %lu
+Relatively temporary storage for raw disk blocks that shouldn't get
+tremendously large (20MB or so).
+.TP 
+\fICached\fP %lu
+In\-memory cache for files read from the disk (the page cache).  Doesn't
+include \fISwapCached\fP.
+.TP 
+\fISwapCached\fP %lu
+Memory that once was swapped out, is swapped back in but still also is in
+the swap file.  (If memory pressure is high, these pages don't need to be
+swapped out again because they are already in the swap file.  This saves
+I/O.)
+.TP 
+\fIActive\fP %lu
+Memory that has been used more recently and usually not reclaimed unless
+absolutely necessary.
+.TP 
+\fIInactive\fP %lu
+Memory which has been less recently used.  It is more eligible to be
+reclaimed for other purposes.
+.TP 
+\fIActive(anon)\fP %lu (Linux 2.6.28 以降)
+[To be documented.]
+.TP 
+\fIInactive(anon)\fP %lu (Linux 2.6.28 以降)
+[To be documented.]
+.TP 
+\fIActive(file)\fP %lu (Linux 2.6.28 以降)
+[To be documented.]
+.TP 
+\fIInactive(file)\fP %lu (Linux 2.6.28 以降)
+[To be documented.]
+.TP 
+\fIUnevictable\fP %lu (Linux 2.6.28 以降)
+(From Linux 2.6.28 to 2.6.30, \fBCONFIG_UNEVICTABLE_LRU\fP was required.)  [To
+be documented.]
+.TP 
+\fIMlocked\fP %lu (Linux 2.6.28 以降)
+(From Linux 2.6.28 to 2.6.30, \fBCONFIG_UNEVICTABLE_LRU\fP was required.)  [To
+be documented.]
+.TP 
+\fIHighTotal\fP %lu
+(Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)  Total amount
+of highmem.  Highmem is all memory above ~860MB of physical memory.  Highmem
+areas are for use by user\-space programs, or for the page cache.  The kernel
+must use tricks to access this memory, making it slower to access than
+lowmem.
+.TP 
+\fIHighFree\fP %lu
+(Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)  Amount of free
+highmem.
+.TP 
+\fILowTotal\fP %lu
+(Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)  Total amount
+of lowmem.  Lowmem is memory which can be used for everything that highmem
+can be used for, but it is also available for the kernel's use for its own
+data structures.  Among many other things, it is where everything from
+\fISlab\fP is allocated.  Bad things happen when you're out of lowmem.
+.TP 
+\fILowFree\fP %lu
+(Starting with Linux 2.6.19, \fBCONFIG_HIGHMEM\fP is required.)  Amount of free
+lowmem.
+.TP 
+\fIMmapCopy\fP %lu (Linux 2.6.29 以降)
+(\fBCONFIG_MMU\fP is required.)  [To be documented.]
+.TP 
+\fISwapTotal\fP %lu
+Total amount of swap space available.
+.TP 
+\fISwapFree\fP %lu
+Amount of swap space that is currently unused.
+.TP 
+\fIDirty\fP %lu
+Memory which is waiting to get written back to the disk.
+.TP 
+\fIWriteback\fP %lu
+Memory which is actively being written back to the disk.
+.TP 
+\fIAnonPages\fP %lu (Linux 2.6.18 以降)
+Non\-file backed pages mapped into user\-space page tables.
+.TP 
+\fIMapped\fP %lu
+Files which have been mmaped, such as libraries.
+.TP 
+\fIShmem\fP %lu (Linux 2.6.32 以降)
+[To be documented.]
+.TP 
+\fISlab\fP %lu
+In\-kernel data structures cache.
+.TP 
+\fISReclaimable\fP %lu (Linux 2.6.19 以降)
+Part of \fISlab\fP, that might be reclaimed, such as caches.
+.TP 
+\fISUnreclaim\fP %lu (Linux 2.6.19 以降)
+Part of \fISlab\fP, that cannot be reclaimed on memory pressure.
+.TP 
+\fIKernelStack\fP %lu (Linux 2.6.32 以降)
+Amount of memory allocated to kernel stacks.
+.TP 
+\fIPageTables\fP %lu (Linux 2.6.18 以降)
+Amount of memory dedicated to the lowest level of page tables.
+.TP 
+\fIQuicklists\fP %lu (Linux 2.6.27 以降)
+(\fBCONFIG_QUICKLIST\fP is required.)  [To be documented.]
+.TP 
+\fINFS_Unstable\fP %lu (Linux 2.6.18 以降)
+NFS pages sent to the server, but not yet committed to stable storage.
+.TP 
+\fIBounce\fP %lu (Linux 2.6.18 以降)
+Memory used for block device "bounce buffers".
+.TP 
+\fIWritebackTmp\fP %lu (Linux 2.6.26 以降)
+Memory used by FUSE for temporary writeback buffers.
+.TP 
+\fICommitLimit\fP %lu (Linux 2.6.10 以降)
+Based on the overcommit ratio ('vm.overcommit_ratio'), this is the total
+amount of memory currently available to be allocated on the system. This
+limit is only adhered to if strict overcommit accounting is enabled (mode 2
+in \fI/proc/sys/vm/overcommit_ratio\fP).  The \fICommitLimit\fP is calculated
+using the following formula:
+
+    CommitLimit = (overcommit_ratio * Physical RAM) + Swap
+
+For example, on a system with 1GB of physical RAM and 7GB of swap with a
+\fIovercommit_ratio\fP of 30, this formula yields a \fICommitLimit\fP of 7.3GB.
+For more details, see the memory overcommit documentation in the kernel
+source file \fIDocumentation/vm/overcommit\-accounting\fP.
+.TP 
+\fICommitted_AS\fP %lu
+The amount of memory presently allocated on the system.  The committed
+memory is a sum of all of the memory which has been allocated by processes,
+even if it has not been "used" by them as of yet.  A process which allocates
+1GB of memory (using \fBmalloc\fP(3)  or similar), but only touches 300MB of
+that memory will only show up as using 300MB of memory even if it has the
+address space allocated for the entire 1GB.  This 1GB is memory which has
+been "committed" to by the VM and can be used at any time by the allocating
+application.  With strict overcommit enabled on the system (mode 2
+\fI/proc/sys/vm/overcommit_memory\fP), allocations which would exceed the
+\fICommitLimit\fP (detailed above) will not be permitted.  This is useful if
+one needs to guarantee that processes will not fail due to lack of memory
+once that memory has been successfully allocated.
+.TP 
+\fIVmallocTotal\fP %lu
+Total size of vmalloc memory area.
+.TP 
+\fIVmallocUsed\fP %lu
+Amount of vmalloc area which is used.
+.TP 
+\fIVmallocChunk\fP %lu
+Largest contiguous block of vmalloc area which is free.
+.TP 
+\fIHardwareCorrupted\fP %lu (Linux 2.6.32 以降)
+(\fBCONFIG_MEMORY_FAILURE\fP is required.)  [To be documented.]
+.TP 
+\fIAnonHugePages\fP %lu (Linux 2.6.38 以降)
+(\fBCONFIG_TRANSPARENT_HUGEPAGE\fP is required.)  Non\-file backed huge pages
+mapped into user\-space page tables.
+.TP 
+\fIHugePages_Total\fP %lu
+(\fBCONFIG_HUGETLB_PAGE\fP is required.)  The size of the pool of huge pages.
+.TP 
+\fIHugePages_Free\fP %lu
+(\fBCONFIG_HUGETLB_PAGE\fP is required.)  The number of huge pages in the pool
+that are not yet allocated.
+.TP 
+\fIHugePages_Rsvd\fP %lu (Linux 2.6.17 以降)
+(\fBCONFIG_HUGETLB_PAGE\fP is required.)  This is the number of huge pages for
+which a commitment to allocate from the pool has been made, but no
+allocation has yet been made.  These reserved huge pages guarantee that an
+application will be able to allocate a huge page from the pool of huge pages
+at fault time.
+.TP 
+\fIHugePages_Surp\fP %lu (Linux 2.6.24 以降)
+(\fBCONFIG_HUGETLB_PAGE\fP is required.)  This is the number of huge pages in
+the pool above the value in \fI/proc/sys/vm/nr_hugepages\fP.  The maximum
+number of surplus huge pages is controlled by
+\fI/proc/sys/vm/nr_overcommit_hugepages\fP.
+.TP 
+\fIHugepagesize\fP %lu
+(\fBCONFIG_HUGETLB_PAGE\fP is required.)  The size of huge pages.
+.RE
 .TP 
 \fI/proc/modules\fP
 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。
@@ -1021,7 +1282,7 @@ Linux 2.4 における I/O メモリマップ。
 \fBfstab\fP(5)  に記述されている。
 .TP 
 \fI/proc/mtrr\fP
-Memory Type Range Registers。 詳細は、カーネルソースファイル \fIDocumentation/mtrr.txt\fP
+Memory Type Range Registers。 詳細は、Linux カーネルソースファイル \fIDocumentation/mtrr.txt\fP
 を参照すること。
 .TP 
 \fI/proc/net\fP
@@ -1165,6 +1426,14 @@ St はソケットの内部状態で、Path は(もしあれば) ソケットの
 (\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
 以降で完全に削除された。
 .TP 
+\fI/proc/profile\fP (since Linux 2.4)
+This file is present only if the kernel was booted with the \fIprofile=1\fP
+command\-line option.  It exposes kernel profiling information in a binary
+format for use by \fBreadprofile\fP(1).  Writing (e.g., an empty string) to
+this file resets the profiling counters; on some architectures, writing a
+binary integer "profiling multiplier" of size \fIsizeof(int)\fP sets the
+profiling interrupt frequency.
+.TP 
 \fI/proc/scsi\fP
 \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
 で表現されているので \fBcat\fP(1)  で読める。
@@ -1229,26 +1498,47 @@ num\-pages\-per\-slab
 .TP 
 \fIcpu  3357 0 4313 1362393\fP
 .\" 1024 on Alpha and ia64
+The amount of time, measured in units of USER_HZ (1/100ths of a second on
+most architectures, use \fIsysconf(_SC_CLK_TCK)\fP to obtain the right value),
+that the system spent in various states:
+.RS
+.TP 
+\fIuser\fP
+(1) Time spent in user mode.
+.TP 
+\fInice\fP
+(2) Time spent in user mode with low priority (nice).
+.TP 
+\fIsystem\fP
+(3) Time spent in system mode.
+.TP 
+\fIidle\fP
 .\" FIXME Actually, the following info about the /proc/stat 'cpu' field
-.\"       does not seem to be quite right (at least in 2.6.12)
-ユーザーモード、 低い優先度 (nice) でのユーザーモード、 システムモード、 タスク待ち (idle task)、
-でそれぞれシステムが消費した時間を USER_HZ を単位として計測した積算値。 (ほとんどのアーキテクチャでは USER_HZ は 1/100
-秒である。 正しい値は \fIsysconf(_SC_CLK_TCK)\fP を使って取得できる。)  最後の値 (タスク待ち) は疑似ファイル uptime
-の 2番目のエントリの値を USER_HZ 倍したものである。
-
-Linux 2.6 では、この行に 3つの欄が追加されている:
-\fIiowait\fP (I/O の完了を待っていた時間; 2.5.41 以降);
-\fIirq\fP (割り込み処理を行った時間; 2.6.0\-test4 以降);
-\fIsoftirq\fP (ソフト割り込みの処理を行った時間; 2.6.0\-test4 以降)。
-
-Linux 2.6.11 以降では、
-8 個目の欄として \fIsteal\fP (盗まれた時間; stolen time) が存在する。
-これは、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。
-
+.\"       does not seem to be quite right (at least in 2.6.12 or 3.6):
+.\"       the idle time in /proc/uptime does not quite match this value
+(4) Time spent in the idle task.  This value should be USER_HZ times the
+second entry in the \fI/proc/uptime\fP pseudo\-file.
+.TP 
+\fIiowait\fP (Linux 2.5.41 以降)
+(5) Time waiting for I/O to complete.
+.TP 
+\fIirq\fP (Linux 2.6.0\-test4 以降)
+(6) Time servicing interrupts.
+.TP 
+\fIsoftirq\fP (Linux 2.6.0\-test4 以降)
+(7) Time servicing softirqs.
+.TP 
+\fIsteal\fP (Linux 2.6.11 以降)
+(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。
+.TP 
+\fIguest\fP (Linux 2.6.24 以降)
 .\" See Changelog entry for 5e84cfde51cf303d368fcb48f22059f37b3872de
-Linux 2.6.24 以降では、9 個目の欄として、 \fIguest\fP がある。これは、 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である。
+(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。
+.TP 
+\fIguest_nice\fP (Linux 2.6.33 以降)
+.\" commit ce0e7b28fb75cb003cfc8d0238613aaf1c55e797
+(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。
+.RE
 .TP 
 \fIpage 5741 1808\fP
 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。
@@ -1294,7 +1584,7 @@ I/O 完了待ちで停止 (blocked) しているプロセス数 (Linux 2.5.45 
 .TP 
 \fI/proc/sys/abi\fP (Linux 2.4.10 以降)
 .\" On some systems, it is not present.
-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、カーネルソースファイル
+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
 \fIDocumentation/sysctl/abi.txt\fP を参照すること。
 .TP 
 \fI/proc/sys/debug\fP
@@ -1309,7 +1599,8 @@ I/O 完了待ちで停止 (blocked) しているプロセス数 (Linux 2.5.45 
 ディレクトリとサブディレクトリが含まれる。
 .TP 
 \fI/proc/sys/fs/binfmt_misc\fP
-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、 カーネルソースの \fIDocumentation/binfmt_misc.txt\fP にある。
+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
+にある。
 .TP 
 \fI/proc/sys/fs/dentry\-state\fP (Linux 2.2 以降)
 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,
@@ -1347,9 +1638,12 @@ I/O 完了待ちで停止 (blocked) しているプロセス数 (Linux 2.5.45 
 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。
 .TP 
 \fI/proc/sys/fs/file\-max\fP
-このファイルはシステム全体でプロセスがオープンできる ファイル数の上限を定義する。 (各プロセスがオープンできるファイル数の上限を
-\fBRLIMIT_NOFILE\fP によって設定する \fBsetrlimit\fP(2)  も参照すること。)
-ファイルハンドルを使い果たして大量にエラーメッセージが出る場合は、 以下のようにしてこの値を増加させてみよ:
+This file defines a system\-wide limit on the number of open files for all
+processes.  (See also \fBsetrlimit\fP(2), which can be used by a process to set
+the per\-process limit, \fBRLIMIT_NOFILE\fP, on the number of files it may
+open.)  If you get lots of error messages in the kernel log about running
+out of file handles (look for "VFS: file\-max limit <number>
+reached"), try increasing this value:
 .br
 
 .br
@@ -1365,10 +1659,18 @@ I/O 完了待ちで停止 (blocked) しているプロセス数 (Linux 2.5.45 
 特権プロセス (\fBCAP_SYS_ADMIN\fP) は \fIfile\-max\fP 上限を上書きできる。
 .TP 
 \fI/proc/sys/fs/file\-nr\fP
-(このファイルは読み出し専用で) 読み出すと 現在オープンされているファイルの数が得られる。 このファイルには、割り当てられているファイルハンドル数・
-空いているファイルハンドル数・ファイルハンドル数の最大値、 という 3 つの数値が書かれている。 カーネルはファイルハンドルを動的に割り当てるが、
-それを再び解放しない。 割り当てられているファイル数が最大値に近づいた場合は、 最大値を大きくすることを考慮すべきである。
-空いているファイルハンドル数が多い場合は、 ファイルハンドルの使用のピークを経験したことがあり、 最大値を大きくする必要はないだろう。
+This (read\-only) file contains three numbers: the number of allocated file
+handles (i.e., the number of files presently opened); the number of free
+file handles; and the maximum number of file handles (i.e., the same value
+as \fI/proc/sys/fs/file\-max\fP).  If the number of allocated file handles is
+close to the maximum, you should consider increasing the maximum.  Before
+Linux 2.6, the kernel allocated file handles dynamically, but it didn't free
+them again.  Instead the free file handles were kept in a list for
+reallocation; the "free file handles" value indicates the size of that
+list.  A large number of free file handles indicates that there was a past
+peak in the usage of open file handles.  Since Linux 2.6, the kernel does
+deallocate freed file handles, and the "free file handles" value is always
+zero.
 .TP 
 \fI/proc/sys/fs/inode\-max\fP
 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
@@ -1415,23 +1717,97 @@ GID は 32 ビットであるが、 16 ビットの UID と GID しかサポー
 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
 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。
 .TP 
+\fI/proc/sys/fs/protected_hardlinks\fP (Linux 3.6 以降)
+.\" commit 800179c9b8a1e796e441674776d11cd4c05d61d7
+When the value in this file is 0, no restrictions are placed on the creation
+of hard links (i.e., this is the historical behaviour before Linux 3.6).
+When the value in this file is 1, a hard link can be created to a target
+file only if one of the following conditions is true:
+.RS
+.IP * 3
+The caller has the \fBCAP_FOWNER\fP capability.
+.IP *
+The file system UID of the process creating the link matches the owner (UID)
+of the target file (as described in \fBcredentials\fP(7), a process's file
+system UID is normally the same as its effective UID).
+.IP *
+All of the following conditions are true:
+.RS 4
+.IP \(bu 3
+the target is a regular file;
+.IP \(bu
+the target file does not have its set\-user\-ID permission bit enabled;
+.IP \(bu
+the target file does not have both its set\-group\-ID and group\-executable
+permission bits enabled; and
+.IP \(bu
+the caller has permission to read and write the target file (either via the
+file's permissions mask or because it has suitable capabilities).
+.RE
+.RE
+.IP
+The default value in this file is 0.  Setting the value to 1 prevents a
+longstanding class of security issues caused by hard\-link\-based
+time\-of\-check, time\-of\-use races, most commonly seen in world\-writable
+directories such as \fI/tmp\fP.  The common method of exploiting this flaw is
+to cross privilege boundaries when following a given hard link (i.e., a root
+process follows a hard link created by another user).  Additionally, on
+systems without separated partitions, this stops unauthorized users from
+"pinning" vulnerable set\-user\-ID and set\-group\-ID files against being
+upgraded by the administrator, or linking to special files.
+.TP 
+\fI/proc/sys/fs/protected_symlinks\fP (Linux 3.6 以降)
+.\" commit 800179c9b8a1e796e441674776d11cd4c05d61d7
+When the value in this file is 0, no restrictions are placed on following
+symbolic links (i.e., this is the historical behaviour before Linux 3.6).
+When the value in this file is 1, symbolic links are followed only in the
+following circumstances:
+.RS
+.IP * 3
+the file system UID of the process following the link matches the owner
+(UID) of the symbolic link (as described in \fBcredentials\fP(7), a process's
+file system UID is normally the same as its effective UID);
+.IP *
+the link is not in a sticky world\-writable directory; or
+.IP *
+the symbolic link and and its parent directory have the same owner (UID)
+.RE
+.IP
+A system call that fails to follow a symbolic link because of the above
+restrictions returns the error \fBEACCES\fP in \fIerrno\fP.
+.IP
+The default value in this file is 0.  Setting the value to 1 avoids a
+longstanding class of security issues based on time\-of\-check, time\-of\-use
+races when accessing symbolic links.
+.TP 
 \fI/proc/sys/fs/suid_dumpable\fP (Linux 2.6.13 以降)
 .\" The following is based on text from Documentation/sysctl/kernel.txt
 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
 (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
 3つの値を指定することができる:
-.sp
-\fI0\ (default)\fP 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)  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) 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。
-.sp
-\fI1\ ("debug")\fP 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 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。
-.sp
-\fI2\ ("suidsafe")\fP 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。
+.RS
+.TP 
+\fI0\ (default)\fP
+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)
+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)
+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。
+.TP 
+\fI1\ ("debug")\fP
+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
+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。
+.TP 
+\fI2\ ("suidsafe")\fP
+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。
 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、
 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。
+.IP
+.\" 9520628e8ceb69fa9a4aee6b57f22675d9e1b709
+.\" 54b501992dd2a839e94e76aa392c392b55080ce8
+Additionally, since Linux 3.6, \fI/proc/sys/kernel/core_pattern\fP must either
+be an absolute pathname or a pipe command, as detailed in \fBcore\fP(5).
+Warnings will be written to the kernel log if \fIcore_pattern\fP does not
+follow these rules, and no core dump will be produced.
+.RE
 .TP 
 このファイルはスーパブロックの値を制御する。
 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
@@ -1470,8 +1846,13 @@ BSD\-style process accounting が有効になっている場合、 これら 3 
 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、
 プログラムに送られてどのように扱うかが決められる。
 .TP 
-\fI/proc/sys/kernel/hotplug\fP
-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。
+\fI/proc/sys/kernel/dmesg_restrict\fP (Linux 2.6.37 以降)
+.\" commit 620f6e8e855d6d447688a5f67a4e176944a084e8
+The value in this file determines who can see kernel syslog contents.  A
+value of 0 in this file imposes no restrictions.  If the value is 1, only
+privileged users can read the kernel syslog.  (See \fBsyslog\fP(2)  for more
+details.)  Since Linux 3.4, only users with the \fBCAP_SYS_ADMIN\fP capability
+may change the value in this file.
 .TP 
 \fI/proc/sys/kernel/domainname\fP と \fI/proc/sys/kernel/hostname\fP
 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
@@ -1498,19 +1879,46 @@ Name Server)  ドメイン名 "frop.org" が含まれているが、DNS ドメ
 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。
 詳細な議論は、 \fBhostname\fP(1)  の man ページを参照すること。
 .TP 
+\fI/proc/sys/kernel/hotplug\fP
+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。
+.TP 
 \fI/proc/sys/kernel/htab\-reclaim\fP
 (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
 \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。
 .TP 
+\fI/proc/sys/kernel/kptr_restrict\fP (Linux 2.6.38 以降)
+.\" 455cd5ab305c90ffc422dd2e0fb634730942b257
+.\" commit 411f05f123cbd7f8aa1edcae86970755a6e2a9d9
+.\" commit 620f6e8e855d6d447688a5f67a4e176944a084e8
+The value in this file determines whether kernel addresses are exposed via
+\fI/proc\fP files and other interfaces.  A value of 0 in this file imposes no
+restrictions.  If the value is 1, kernel pointers printed using the \fI%pK\fP
+format specifier will be replaced with zeros unless the user has the
+\fBCAP_SYSLOG\fP capability.  If the value is 2, kernel pointers printed using
+the \fI%pK\fP format specifier will be replaced with zeros regardless of the
+user's capabilities.  The initial default value for this file was 1, but the
+default was changed to 0 in Linux 2.6.39.  Since Linux 3.4, only users with
+the \fBCAP_SYS_ADMIN\fP capability can change the value in this file.
+.TP 
 \fI/proc/sys/kernel/l2cr\fP
 (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。
 0 以外の場合は有効になる。
 .TP 
 \fI/proc/sys/kernel/modprobe\fP
-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 である。このファイルは、
-\fBCONFIG_KMOD\fP オプション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、カーネルソースファイル
+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
+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)
+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
 \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。
 .TP 
+\fI/proc/sys/kernel/modules_disabled\fP (Linux 2.6.31 以降)
+.\" 3d43321b7015387cfebbe26436d0e9d299162ea1
+.\" From Documentation/sysctl/kernel.txt
+A toggle value indicating if modules are allowed to be loaded in an
+otherwise modular kernel.  This toggle defaults to off (0), but can be set
+true (1).  Once true, modules can be neither loaded nor unloaded, and the
+toggle cannot be set back to false.  The file is only present if the kernel
+is built with the \fBCONFIG_MODULES\fP option enabled.
+.TP 
 \fI/proc/sys/kernel/msgmax\fP
 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。
 .TP 
@@ -1577,7 +1985,7 @@ Service) または YP (Yellow Pages) のドメイン名を混同してはなら
 を参照。
 .TP 
 \fI/proc/sys/kernel/real\-root\-dev\fP
-このファイルはカーネルソースファイル \fIDocumentation/initrd.txt\fP に記述されている。
+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。
 .TP 
 \fI/proc/sys/kernel/reboot\-cmd\fP (Sparc のみ) 
 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。
@@ -1636,8 +2044,8 @@ real\-time (queued) signal の 最大数を調整するのに使用される。
         128 \- リブート/電源オフを許可する
         256 \- 全てのリアルタイム・タスクの nice 値の変更を許可する
 
-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、カーネルソースファイル
-\fIDocumentation/sysrq.txt\fP を参照のこと。
+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
+カーネルソースファイル \fIDocumentation/sysrq.txt\fP を参照のこと。
 .TP 
 \fI/proc/sys/kernel/version\fP
 このファイルには、以下のような文字列が書かれている:
@@ -1807,7 +2215,7 @@ kill されるだけかもしれない。 この場合には、カーネルパ
 \fI/proc/sysrq\-trigger\fP (Linux 2.4.21 以降)
 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
 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
-\fIroot\fP だけである。詳細については、カーネルソースファイルの \fIDocumentation/sysrq.txt\fP を参照のこと。
+\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。
 .TP 
 \fI/proc/sysvipc\fP
 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
@@ -1861,10 +2269,10 @@ kill されるだけかもしれない。 この場合には、カーネルパ
 \fBslabinfo\fP(5), \fBhier\fP(7), \fBtime\fP(7), \fBarp\fP(8), \fBhdparm\fP(8),
 \fBifconfig\fP(8), \fBinit\fP(8), \fBlsmod\fP(8), \fBlspci\fP(8), \fBmount\fP(8),
 \fBnetstat\fP(8), \fBprocinfo\fP(8), \fBroute\fP(8)
-.br
-カーネルのソースファイル: \fIDocumentation/filesystems/proc.txt\fP,
+
+Linux カーネルのソースファイル: \fIDocumentation/filesystems/proc.txt\fP,
 \fIDocumentation/sysctl/vm.txt\fP
 .SH この文書について
-この man ページは Linux \fIman\-pages\fP プロジェクトのリリース 3.41 の一部
+この man ページは Linux \fIman\-pages\fP プロジェクトのリリース 3.50 の一部
 である。プロジェクトの説明とバグ報告に関する情報は
 http://www.kernel.org/doc/man\-pages/ に書かれている。