From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([140.186.70.92]:53298)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <stefanha@gmail.com>) id 1RXv1n-0002os-Km
	for qemu-devel@nongnu.org; Tue, 06 Dec 2011 08:25:48 -0500
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <stefanha@gmail.com>) id 1RXv1h-0007qg-UU
	for qemu-devel@nongnu.org; Tue, 06 Dec 2011 08:25:47 -0500
Received: from mail-fx0-f45.google.com ([209.85.161.45]:46991)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <stefanha@gmail.com>) id 1RXv1h-0007qa-Me
	for qemu-devel@nongnu.org; Tue, 06 Dec 2011 08:25:41 -0500
Received: by faao26 with SMTP id o26so1286768faa.4
	for <qemu-devel@nongnu.org>; Tue, 06 Dec 2011 05:25:40 -0800 (PST)
MIME-Version: 1.0
In-Reply-To: <1323105612-32419-1-git-send-email-avi@redhat.com>
References: <1323105612-32419-1-git-send-email-avi@redhat.com>
Date: Tue, 6 Dec 2011 13:25:40 +0000
Message-ID: <CAJSP0QWm7xtftXNgkaQBAfW7OXMnp22K39yZQhMpq03S++d6aQ@mail.gmail.com>
From: Stefan Hajnoczi <stefanha@gmail.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Subject: Re: [Qemu-devel] [PATCH] coroutine: switch per-thread free pool to
 a global pool
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Avi Kivity <avi@redhat.com>
Cc: Kevin Wolf <kwolf@redhat.com>, qemu-devel@nongnu.org, Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

On Mon, Dec 5, 2011 at 5:20 PM, Avi Kivity <avi@redhat.com> wrote:
> ucontext-based coroutines use a free pool to reduce allocations and
> deallocations of coroutine objects. =A0The pool is per-thread, presumably
> to improve locality. =A0However, as coroutines are usually allocated in
> a vcpu thread and freed in the I/O thread, the pool accounting gets
> screwed up and we end allocating and freeing a coroutine for every I/O
> request. =A0This is expensive since large objects are allocated via the
> kernel, and are not cached by the C runtime.
>
> Fix by switching to a global pool. =A0This is safe since we're protected
> by the global mutex.

Looks good to me.  I did check how hw/9pfs/ uses coroutines because it
bounces them into worker threads that are not under the QEMU mutex but
they are not created/destroyed there so we should be okay.

Stefan