public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Thomas Gleixner <tglx@linutronix.de>
To: LKML <linux-kernel@vger.kernel.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
	Darren Hart <darren@dvhart.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@kernel.org>,
	Michael Kerrisk <mtk.manpages@googlemail.com>,
	Davidlohr Bueso <dave@stgolabs.net>, Chris Mason <clm@fb.com>,
	"Carlos O'Donell" <carlos@redhat.com>,
	Torvald Riegel <triegel@redhat.com>,
	Eric Dumazet <edumazet@google.com>
Subject: [RFC patch 6/7] futex.2: Document attached mode
Date: Sat, 02 Apr 2016 11:09:20 -0000	[thread overview]
Message-ID: <20160402110036.030897949@linutronix.de> (raw)
In-Reply-To: 20160402095108.894519835@linutronix.de

[-- Attachment #1: futex.2--Document-attached-mode.patch --]
[-- Type: text/plain, Size: 3620 bytes --]

At least an attempt to document the futex attached mode extension.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
 man2/futex.2 |   78 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 75 insertions(+), 3 deletions(-)

--- a/man2/futex.2
+++ b/man2/futex.2
@@ -138,8 +138,8 @@ Besides the basic wait and wake-up futex
 futex operations aimed at supporting more complex use cases.
 
 Note that
-no explicit initialization or destruction is necessary to use futexes;
-the kernel maintains a futex
+no explicit initialization or destruction is necessary to use futexes
+except when using futexes in attached mode; the kernel maintains a futex
 (i.e., the kernel-internal implementation artifact)
 only while operations such as
 .BR FUTEX_WAIT ,
@@ -253,6 +253,12 @@ as a relative time,
 measured against the
 .BR CLOCK_MONOTONIC
 clock.
+.TP
+.BR FUTEX_ATTACHED " (since Linux ?.?.?)"
+This option can be employed with all futex operations. It tells the kernel
+that the futex operates in attached mode. See the futex_op
+.BR FUTEX_ATTACH
+for further explanation.
 .PP
 The operation specified in
 .I futex_op
@@ -767,6 +773,63 @@ operations correspond to
 and
 .BR FUTEX_WAKE_BITSET
 operations where the bit masks are all ones.
+
+.\"
+.\""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
+.\"
+.TP
+.BR FUTEX_ATTACH " (since Linux ?.?.?)"
+This operation attaches the futex on the futex address provided by the
+.IR uaddr
+argument.
+
+This operation tells the kernel that the thread is going to operate on that
+futex and therefor it should create a kernel side state for it. This state is
+used in further futex operations instead of the global hash which is used
+without that optimization.
+
+There are two advantages for using attached futexes. Potential hash collisions
+on the global hash are avoided. The kernel state which replaces the global
+hash for attached futexes is allocated on the node on which the thread runs
+and therefor avoids cross node memory access.
+
+For this operation and all subsequent operations on the futex the flag
+.BR FUTEX_ATTACHED
+must be ORed to the
+.I futex_op
+
+The
+.IR val ,
+.IR timeout ,
+.IR uaddr2
+and
+.I val3
+arguments are ignored.
+
+.\"
+.\""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
+.\"
+.TP
+.BR FUTEX_DETACH " (since Linux ?.?.?)"
+This operation detaches the futex on the futex address provided by the
+.I uaddr
+argument from the kernel state.
+
+This operation reverses a previous
+.BR FUTEX_ATTACH
+operation.
+
+If the caller is the last user of the kernel side state, the kernel side state
+is destroyed.
+
+The
+.IR val ,
+.IR timeout ,
+.IR uaddr2
+and
+.I val3
+arguments are ignored.
+
 .\"
 .\""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 .\"
@@ -1402,6 +1465,10 @@ While requeueing a waiter to the PI fute
 .IR uaddr2 ,
 the kernel detected a deadlock.
 .TP
+.B EEXIST
+.RB ( FUTEX_ATTACH )
+Operation was attempted on a already attached futex.
+.TP
 .B EFAULT
 A required pointer argument (i.e.,
 .IR uaddr ,
@@ -1550,13 +1617,18 @@ Invalid argument.
 .BR ENOMEM
 .RB ( FUTEX_LOCK_PI ,
 .BR FUTEX_TRYLOCK_PI ,
-.BR FUTEX_CMP_REQUEUE_PI )
+.BR FUTEX_CMP_REQUEUE_PI,
+.BR FUTEX_ATTACH )
 The kernel could not allocate memory to hold state information.
 .TP
 .B ENFILE
 .RB ( FUTEX_FD )
 The system-wide limit on the total number of open files has been reached.
 .TP
+.B ENOSPC
+.RB ( FUTEX_ATTACH )
+Operation failed because the limit of attached futexes has been reached.
+.TP
 .B ENOSYS
 Invalid operation specified in
 .IR futex_op .

      parent reply	other threads:[~2016-04-02 11:12 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-04-02 11:09 [RFC patch 0/7] futex: Add support for attached futexes Thomas Gleixner
2016-04-02 11:09 ` [RFC patch 1/7] futex: Provide helpers for hash bucket add/remove Thomas Gleixner
2016-04-02 11:09 ` [RFC patch 2/7] futex: Add some more function commentry Thomas Gleixner
2016-04-02 11:09 ` [RFC patch 3/7] futex: Make key init a helper function Thomas Gleixner
2016-04-02 11:09 ` [RFC patch 4/7] futex: Add support for attached futexes Thomas Gleixner
2016-04-02 16:26   ` Peter Zijlstra
2016-04-02 18:01     ` Thomas Gleixner
2016-04-02 16:29   ` Peter Zijlstra
2016-04-03  9:59     ` Thomas Gleixner
2016-04-02 18:19   ` Andy Lutomirski
2016-04-03  9:57     ` Thomas Gleixner
2016-04-03 13:18       ` Andy Lutomirski
2016-04-03 15:56         ` Thomas Gleixner
2016-04-03 16:11           ` Andy Lutomirski
2016-04-02 23:48   ` Rasmus Villemoes
2016-04-03 10:05     ` Thomas Gleixner
2016-04-03 11:16   ` Ingo Molnar
2016-04-03 11:30     ` Linus Torvalds
2016-04-05  7:44       ` Torvald Riegel
2016-04-05 15:58       ` Carlos O'Donell
2016-04-02 11:09 ` [RFC patch 5/7] perf/bench/futex-hash: Support " Thomas Gleixner
2016-04-02 11:09 ` [RFC patch 7/7] [PATCH] glibc: nptl: Add support for attached pthread_mutexes Thomas Gleixner
2016-04-02 16:30   ` Peter Zijlstra
2016-04-02 16:32     ` Peter Zijlstra
2016-04-03 10:08     ` Thomas Gleixner
2016-04-02 11:09 ` Thomas Gleixner [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20160402110036.030897949@linutronix.de \
    --to=tglx@linutronix.de \
    --cc=bigeasy@linutronix.de \
    --cc=carlos@redhat.com \
    --cc=clm@fb.com \
    --cc=darren@dvhart.com \
    --cc=dave@stgolabs.net \
    --cc=edumazet@google.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@kernel.org \
    --cc=mtk.manpages@googlemail.com \
    --cc=peterz@infradead.org \
    --cc=triegel@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox