From: Peter Hurley <peter@hurleysoftware.com>
To: Michel Lespinasse <walken@google.com>
Cc: Alex Shi <alex.shi@intel.com>, Ingo Molnar <mingo@kernel.org>,
David Howells <dhowells@redhat.com>,
Peter Zijlstra <a.p.zijlstra@chello.nl>,
Thomas Gleixner <tglx@linutronix.de>,
Yuanhan Liu <yuanhan.liu@linux.intel.com>,
Rik van Riel <riel@redhat.com>,
Dave Chinner <david@fromorbit.com>,
Andrew Morton <akpm@linux-foundation.org>,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 10/13] rwsem: simplify __rwsem_do_wake
Date: Thu, 28 Mar 2013 12:51:32 -0400 [thread overview]
Message-ID: <1364489492.3559.18.camel@thor.lan> (raw)
In-Reply-To: <1363344869-15732-11-git-send-email-walken@google.com>
On Fri, 2013-03-15 at 03:54 -0700, Michel Lespinasse wrote:
> This is mostly for cleanup value:
>
> - We don't need several gotos to handle the case where the first
> waiter is a writer. Two simple tests will do (and generate very
> similar code).
>
> - In the remainder of the function, we know the first waiter is a reader,
> so we don't have to double check that. We can use do..while loops
> to iterate over the readers to wake (generates slightly better code).
>
> Signed-off-by: Michel Lespinasse <walken@google.com>
>
> ---
> lib/rwsem-spinlock.c | 25 ++++++++-----------------
> lib/rwsem.c | 26 ++++++++++++--------------
> 2 files changed, 20 insertions(+), 31 deletions(-)
>
> diff --git a/lib/rwsem-spinlock.c b/lib/rwsem-spinlock.c
> index 5f117f37ac0a..20207a6e8ac3 100644
> --- a/lib/rwsem-spinlock.c
> +++ b/lib/rwsem-spinlock.c
> @@ -70,26 +70,17 @@ __rwsem_do_wake(struct rw_semaphore *sem, int wakewrite)
>
> waiter = list_entry(sem->wait_list.next, struct rwsem_waiter, list);
>
> - if (!wakewrite) {
> - if (waiter->type == RWSEM_WAITING_FOR_WRITE)
> - goto out;
> - goto dont_wake_writers;
> - }
> -
> - /*
> - * as we support write lock stealing, we can't set sem->activity
> - * to -1 here to indicate we get the lock. Instead, we wake it up
> - * to let it go get it again.
> - */
> if (waiter->type == RWSEM_WAITING_FOR_WRITE) {
> - wake_up_process(waiter->task);
> + if (wakewrite)
> + /* Wake up a writer. Note that we do not grant it the
> + * lock - it will have to acquire it when it runs. */
> + wake_up_process(waiter->task);
> goto out;
> }
>
> - /* grant an infinite number of read locks to the front of the queue */
> - dont_wake_writers:
> + /* grant read locks to all queued readers. */
I think the original comment still applies here now?
> woken = 0;
> - while (waiter->type == RWSEM_WAITING_FOR_READ) {
> + do {
> struct list_head *next = waiter->list.next;
>
> list_del(&waiter->list);
> @@ -99,10 +90,10 @@ __rwsem_do_wake(struct rw_semaphore *sem, int wakewrite)
> wake_up_process(tsk);
> put_task_struct(tsk);
> woken++;
> - if (list_empty(&sem->wait_list))
> + if (next == &sem->wait_list)
> break;
> waiter = list_entry(next, struct rwsem_waiter, list);
> - }
> + } while (waiter->type != RWSEM_WAITING_FOR_WRITE);
Maybe convert this to list_for_each_entry_safe() since you're cleaning
it up?
>
> sem->activity += woken;
>
> diff --git a/lib/rwsem.c b/lib/rwsem.c
> index 0d50e46d5b0c..9a675fa9d78e 100644
> --- a/lib/rwsem.c
> +++ b/lib/rwsem.c
> @@ -68,20 +68,17 @@ __rwsem_do_wake(struct rw_semaphore *sem, int wake_type)
> signed long woken, loop, adjustment;
>
> waiter = list_entry(sem->wait_list.next, struct rwsem_waiter, list);
> - if (waiter->type != RWSEM_WAITING_FOR_WRITE)
> - goto readers_only;
> -
> - if (wake_type == RWSEM_WAKE_READ_OWNED)
> - /* Another active reader was observed, so wakeup is not
> - * likely to succeed. Save the atomic op.
> - */
> + if (waiter->type == RWSEM_WAITING_FOR_WRITE) {
> + if (wake_type != RWSEM_WAKE_READ_OWNED)
> + /* Wake writer at the front of the queue, but do not
> + * grant it the lock yet as we want other writers
> + * to be able to steal it. Readers, on the other hand,
> + * will block as they will notice the queued writer.
> + */
> + wake_up_process(waiter->task);
> goto out;
> + }
>
> - /* Wake up the writing waiter and let the task grab the sem: */
> - wake_up_process(waiter->task);
> - goto out;
> -
> - readers_only:
> /* If we come here from up_xxxx(), another thread might have reached
> * rwsem_down_failed_common() before we acquired the spinlock and
> * woken up a waiter, making it now active. We prefer to check for
> @@ -125,7 +122,8 @@ __rwsem_do_wake(struct rw_semaphore *sem, int wake_type)
> rwsem_atomic_add(adjustment, sem);
>
> next = sem->wait_list.next;
> - for (loop = woken; loop > 0; loop--) {
> + loop = woken;
> + do {
> waiter = list_entry(next, struct rwsem_waiter, list);
> next = waiter->list.next;
> tsk = waiter->task;
> @@ -133,7 +131,7 @@ __rwsem_do_wake(struct rw_semaphore *sem, int wake_type)
> waiter->task = NULL;
> wake_up_process(tsk);
> put_task_struct(tsk);
> - }
> + } while (--loop);
>
> sem->wait_list.next = next;
> next->prev = &sem->wait_list;
next prev parent reply other threads:[~2013-03-28 16:51 UTC|newest]
Thread overview: 33+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-03-15 10:54 [PATCH v2 00/13] rwsem fast-path write lock stealing Michel Lespinasse
2013-03-15 10:54 ` [PATCH v2 01/13] rwsem: make the waiter type an enumeration rather than a bitmask Michel Lespinasse
2013-03-28 18:59 ` Rik van Riel
2013-03-15 10:54 ` [PATCH v2 02/13] rwsem: shorter spinlocked section in rwsem_down_failed_common() Michel Lespinasse
2013-03-28 16:21 ` Peter Hurley
2013-03-28 19:03 ` Rik van Riel
2013-03-15 10:54 ` [PATCH v2 03/13] rwsem: move rwsem_down_failed_common code into rwsem_down_{read,write}_failed Michel Lespinasse
2013-03-28 17:05 ` Peter Hurley
2013-04-27 21:21 ` Davidlohr Bueso
2013-03-28 19:08 ` Rik van Riel
2013-03-15 10:54 ` [PATCH v2 04/13] rwsem: simplify rwsem_down_read_failed Michel Lespinasse
2013-03-28 19:22 ` Rik van Riel
2013-03-15 10:54 ` [PATCH v2 05/13] rwsem: simplify rwsem_down_write_failed Michel Lespinasse
2013-03-28 20:33 ` Rik van Riel
2013-03-15 10:54 ` [PATCH v2 06/13] rwsem: more agressive lock stealing in rwsem_down_write_failed Michel Lespinasse
2013-03-28 20:54 ` Rik van Riel
2013-03-15 10:54 ` [PATCH v2 07/13] rwsem: use cmpxchg for trying to steal write lock Michel Lespinasse
2013-03-28 16:44 ` Peter Hurley
2013-03-28 20:59 ` Rik van Riel
2013-03-15 10:54 ` [PATCH v2 08/13] rwsem: avoid taking wait_lock in rwsem_down_write_failed Michel Lespinasse
2013-03-28 21:01 ` Rik van Riel
2013-03-15 10:54 ` [PATCH v2 09/13] rwsem: skip initial trylock " Michel Lespinasse
2013-03-28 21:16 ` Rik van Riel
2013-03-15 10:54 ` [PATCH v2 10/13] rwsem: simplify __rwsem_do_wake Michel Lespinasse
2013-03-28 16:51 ` Peter Hurley [this message]
2013-03-15 10:54 ` [PATCH v2 11/13] rwsem: implement support for write lock stealing on the fastpath Michel Lespinasse
2013-03-28 16:17 ` Peter Hurley
2013-03-15 10:54 ` [PATCH v2 12/13] rwsem: do not block readers at head of queue if other readers are active Michel Lespinasse
2013-03-28 17:25 ` Peter Hurley
2013-03-15 10:54 ` [PATCH v2 13/13] x86 rwsem: avoid taking slow path when stealing write lock Michel Lespinasse
2013-03-28 19:19 ` [PATCH v2 00/13] rwsem fast-path write lock stealing Peter Hurley
[not found] ` <1367097352.1782.14.camel@buesod1.americas.hpqcorp.net>
2013-04-27 23:47 ` Michel Lespinasse
2013-05-07 9:18 ` Michel Lespinasse
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=1364489492.3559.18.camel@thor.lan \
--to=peter@hurleysoftware.com \
--cc=a.p.zijlstra@chello.nl \
--cc=akpm@linux-foundation.org \
--cc=alex.shi@intel.com \
--cc=david@fromorbit.com \
--cc=dhowells@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=riel@redhat.com \
--cc=tglx@linutronix.de \
--cc=walken@google.com \
--cc=yuanhan.liu@linux.intel.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