From: Michel Lespinasse <walken@google.com>
To: Linus Torvalds <torvalds@linux-foundation.org>,
David Howells <dhowells@redhat.com>, Ingo Molnar <mingo@elte.hu>,
Thomas Gleixner <tglx@linutronix.de>
Cc: LKML <linux-kernel@vger.kernel.org>,
Andrew Morton <akpm@linux-foundation.org>,
Mike Waychison <mikew@google.com>,
Suleiman Souhlal <suleiman@google.com>,
Ying Han <yinghan@google.com>,
Michel Lespinasse <walken@google.com>
Subject: [PATCH 02/12] rwsem: use single atomic update for sem count when waking up readers.
Date: Tue, 11 May 2010 20:20:52 -0700 [thread overview]
Message-ID: <1273634462-2672-3-git-send-email-walken@google.com> (raw)
In-Reply-To: <1273634462-2672-1-git-send-email-walken@google.com>
When waking up queued readers, __rwsem_do_wake was using a pair of atomic
operations (one before looking at the first queued thread and one after
having looped through all readers). Restructure code to use a single
atomic operation instead.
Signed-off-by: Michel Lespinasse <walken@google.com>
---
lib/rwsem.c | 52 ++++++++++++++++++++++++++--------------------------
1 files changed, 26 insertions(+), 26 deletions(-)
diff --git a/lib/rwsem.c b/lib/rwsem.c
index 8d6a13e..caee9b7 100644
--- a/lib/rwsem.c
+++ b/lib/rwsem.c
@@ -41,7 +41,7 @@ struct rwsem_waiter {
* - if we come here from up_xxxx(), then:
* - the 'active part' of count (&0x0000ffff) reached 0 (but may have changed)
* - the 'waiting part' of count (&0xffff0000) is -ve (and will still be so)
- * - there must be someone on the queue
+ * - there must be someone on the queue
* - the spinlock must be held by the caller
* - woken process blocks are discarded from the list after having task zeroed
* - writers are only woken if downgrading is false
@@ -54,26 +54,26 @@ __rwsem_do_wake(struct rw_semaphore *sem, int downgrading)
struct list_head *next;
signed long oldcount, woken, loop;
+ waiter = list_entry(sem->wait_list.next, struct rwsem_waiter, list);
+ if (!(waiter->flags & RWSEM_WAITING_FOR_WRITE))
+ goto readers_only;
+
if (downgrading)
- goto dont_wake_writers;
+ /* Caller's lock is still active, so we can't possibly
+ * succeed waking writers.
+ */
+ goto out;
- /* if we came through an up_xxxx() call, we only only wake someone up
+ /* There's a writer at the front of the queue - try to grant it the
+ * write lock. However, we only only wake someone up
* if we can transition the active part of the count from 0 -> 1
*/
- try_again:
+ retry_writer:
oldcount = rwsem_atomic_update(RWSEM_ACTIVE_BIAS, sem)
- RWSEM_ACTIVE_BIAS;
if (oldcount & RWSEM_ACTIVE_MASK)
- goto undo;
-
- waiter = list_entry(sem->wait_list.next, struct rwsem_waiter, list);
-
- /* try to grant a single write lock if there's a writer at the front
- * of the queue - note we leave the 'active part' of the count
- * incremented by 1 and the waiting part incremented by 0x00010000
- */
- if (!(waiter->flags & RWSEM_WAITING_FOR_WRITE))
- goto readers_only;
+ /* Someone grabbed the sem already */
+ goto undo_writer;
/* We must be careful not to touch 'waiter' after we set ->task = NULL.
* It is an allocated on the waiter's stack and may become invalid at
@@ -87,12 +87,6 @@ __rwsem_do_wake(struct rw_semaphore *sem, int downgrading)
put_task_struct(tsk);
goto out;
- /* don't want to wake any writers */
- dont_wake_writers:
- waiter = list_entry(sem->wait_list.next, struct rwsem_waiter, list);
- if (waiter->flags & RWSEM_WAITING_FOR_WRITE)
- goto out;
-
/* grant an infinite number of read locks to the readers at the front
* of the queue
* - note we increment the 'active part' of the count by the number of
@@ -113,11 +107,12 @@ __rwsem_do_wake(struct rw_semaphore *sem, int downgrading)
loop = woken;
woken *= RWSEM_ACTIVE_BIAS - RWSEM_WAITING_BIAS;
- if (!downgrading)
- /* we'd already done one increment earlier */
- woken -= RWSEM_ACTIVE_BIAS;
- rwsem_atomic_add(woken, sem);
+ retry_readers:
+ oldcount = rwsem_atomic_update(woken, sem) - woken;
+ if (!downgrading && (oldcount & RWSEM_ACTIVE_MASK))
+ /* Someone grabbed the sem already */
+ goto undo_readers;
next = sem->wait_list.next;
for (; loop > 0; loop--) {
@@ -137,10 +132,15 @@ __rwsem_do_wake(struct rw_semaphore *sem, int downgrading)
return sem;
/* undo the change to count, but check for a transition 1->0 */
- undo:
+ undo_writer:
if (rwsem_atomic_update(-RWSEM_ACTIVE_BIAS, sem) & RWSEM_ACTIVE_MASK)
goto out;
- goto try_again;
+ goto retry_writer;
+
+ undo_readers:
+ if (rwsem_atomic_update(-woken, sem) & RWSEM_ACTIVE_MASK)
+ goto out;
+ goto retry_readers;
}
/*
--
1.7.0.1
next prev parent reply other threads:[~2010-05-12 3:22 UTC|newest]
Thread overview: 33+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-05-12 3:20 [PATCH 00/12] rwsem changes + down_read_unfair() proposal Michel Lespinasse
2010-05-12 3:20 ` [PATCH 01/12] rwsem: test for no active locks in __rwsem_do_wake undo code Michel Lespinasse
2010-05-12 3:20 ` Michel Lespinasse [this message]
2010-05-12 3:20 ` [PATCH 03/12] rwsem: let RWSEM_WAITING_BIAS represent any number of waiting threads Michel Lespinasse
2010-05-12 3:20 ` [PATCH 04/12] rwsem: consistently use adjustment variable Michel Lespinasse
2010-05-12 3:20 ` [PATCH 05/12] x86 rwsem: take advantage of new RWSEM_WAITING_BIAS semantics Michel Lespinasse
2010-05-12 3:20 ` [PATCH 06/12] rwsem: wake queued readers when other readers are active Michel Lespinasse
2010-05-12 3:20 ` [PATCH 07/12] rwsem: wake queued readers when writer blocks on active read lock Michel Lespinasse
2010-05-12 3:20 ` [PATCH 08/12] rwsem: smaller wrappers around rwsem_down_failed_common Michel Lespinasse
2010-05-12 3:20 ` [PATCH 09/12] generic rwsem: implement down_read_unfair Michel Lespinasse
2010-05-12 3:21 ` [PATCH 10/12] rwsem: down_read_unfair infrastructure support Michel Lespinasse
2010-05-12 3:21 ` [PATCH 11/12] x86 rwsem: down_read_unfair implementation Michel Lespinasse
2010-05-12 3:21 ` [PATCH 12/12] Use down_read_unfair() for /sys/<pid>/exe and /sys/<pid>/maps files Michel Lespinasse
2010-05-12 22:53 ` KOSAKI Motohiro
2010-05-12 23:35 ` Michel Lespinasse
2010-05-13 0:32 ` KOSAKI Motohiro
2010-05-12 10:39 ` [PATCH 01/12] rwsem: test for no active locks in __rwsem_do_wake undo code David Howells
2010-05-12 11:01 ` [PATCH 02/12] rwsem: use single atomic update for sem count when waking up readers David Howells
2010-05-13 0:54 ` Michel Lespinasse
2010-05-12 11:36 ` David Howells
2010-05-12 11:45 ` [PATCH 04/12] rwsem: consistently use adjustment variable David Howells
2010-05-13 1:12 ` Michel Lespinasse
2010-05-12 12:10 ` [PATCH 05/12] x86 rwsem: take advantage of new RWSEM_WAITING_BIAS semantics David Howells
2010-05-12 12:22 ` [PATCH 06/12] rwsem: wake queued readers when other readers are active David Howells
2010-05-13 2:39 ` Michel Lespinasse
2010-05-13 5:41 ` Michel Lespinasse
2010-05-12 12:33 ` [PATCH 07/12] rwsem: wake queued readers when writer blocks on active read lock David Howells
2010-05-12 12:36 ` [PATCH 08/12] rwsem: smaller wrappers around rwsem_down_failed_common David Howells
2010-05-12 12:42 ` David Howells
2010-05-13 2:54 ` Michel Lespinasse
2010-05-12 12:46 ` [PATCH 09/12] generic rwsem: implement down_read_unfair David Howells
2010-05-12 13:08 ` [PATCH 11/12] x86 rwsem: down_read_unfair implementation David Howells
2010-05-12 13:10 ` [PATCH 12/12] Use down_read_unfair() for /sys/<pid>/exe and /sys/<pid>/maps files David Howells
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=1273634462-2672-3-git-send-email-walken@google.com \
--to=walken@google.com \
--cc=akpm@linux-foundation.org \
--cc=dhowells@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mikew@google.com \
--cc=mingo@elte.hu \
--cc=suleiman@google.com \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.org \
--cc=yinghan@google.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
Powered by JetHome