From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754012AbZGFQdK (ORCPT ); Mon, 6 Jul 2009 12:33:10 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751875AbZGFQc4 (ORCPT ); Mon, 6 Jul 2009 12:32:56 -0400 Received: from victor.provo.novell.com ([137.65.250.26]:49566 "EHLO victor.provo.novell.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751171AbZGFQcz (ORCPT ); Mon, 6 Jul 2009 12:32:55 -0400 Message-ID: <4A52272A.2060101@novell.com> Date: Mon, 06 Jul 2009 12:32:42 -0400 From: Gregory Haskins User-Agent: Thunderbird 2.0.0.22 (Macintosh/20090605) MIME-Version: 1.0 To: "Michael S. Tsirkin" CC: kvm@vger.kernel.org, linux-kernel@vger.kernel.org, avi@redhat.com, davidel@xmailserver.org Subject: Re: [KVM PATCH v9 5/5] KVM: create irqfd-cleanup-wq on demand References: <20090702153454.20186.99191.stgit@dev.haskins.net> <20090702153821.20186.9344.stgit@dev.haskins.net> <20090706155850.GA12399@redhat.com> <4A522050.7030200@novell.com> <20090706161444.GC12399@redhat.com> In-Reply-To: <20090706161444.GC12399@redhat.com> X-Enigmail-Version: 0.95.7 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig4534F80F3EEEC52661E832DA" Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig4534F80F3EEEC52661E832DA Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Michael S. Tsirkin wrote: > On Mon, Jul 06, 2009 at 12:03:28PM -0400, Gregory Haskins wrote: > =20 >> Michael S. Tsirkin wrote: >> =20 >>> On Thu, Jul 02, 2009 at 11:38:22AM -0400, Gregory Haskins wrote: >>> =20 >>> =20 >>>> We currently create this wq on module_init, which may be wasteful if= the >>>> host never creates a guest that uses irqfd. This patch changes the >>>> algorithm so that the workqueue is only created when at least one gu= est >>>> is using irqfd. The queue is cleaned up when the last guest using i= rqfd >>>> is shutdown. >>>> >>>> To keep things simple, we only check whether the guest has tried to = create >>>> an irqfd, not whether there are actually irqfds active. >>>> >>>> Signed-off-by: Gregory Haskins >>>> --- >>>> >>>> include/linux/kvm_host.h | 1=20 >>>> virt/kvm/eventfd.c | 100 ++++++++++++++++++++++++++++++++++-= ----------- >>>> 2 files changed, 75 insertions(+), 26 deletions(-) >>>> >>>> diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h >>>> index 8e04a34..cd1a0f3 100644 >>>> --- a/include/linux/kvm_host.h >>>> +++ b/include/linux/kvm_host.h >>>> @@ -149,6 +149,7 @@ struct kvm { >>>> struct { >>>> spinlock_t lock; >>>> struct list_head items; >>>> + bool init; >>>> } irqfds; >>>> #endif >>>> struct kvm_vm_stat stat; >>>> diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c >>>> index 4092b8d..fcc3469 100644 >>>> --- a/virt/kvm/eventfd.c >>>> +++ b/virt/kvm/eventfd.c >>>> @@ -49,7 +49,16 @@ struct _irqfd { >>>> struct work_struct shutdown; >>>> }; >>>> =20 >>>> -static struct workqueue_struct *irqfd_cleanup_wq; >>>> +struct _irqfd_cleanup { >>>> + struct mutex lock; >>>> + int refs; >>>> + struct workqueue_struct *wq; >>>> +}; >>>> + >>>> +static struct _irqfd_cleanup irqfd_cleanup =3D { >>>> + .lock =3D __MUTEX_INITIALIZER(irqfd_cleanup.lock), >>>> + .refs =3D 0, >>>> +}; >>>> =20 >>>> static void >>>> irqfd_inject(struct work_struct *work) >>>> @@ -110,7 +119,7 @@ irqfd_deactivate(struct _irqfd *irqfd) >>>> =20 >>>> list_del_init(&irqfd->list); >>>> =20 >>>> - queue_work(irqfd_cleanup_wq, &irqfd->shutdown); >>>> + queue_work(irqfd_cleanup.wq, &irqfd->shutdown); >>>> } >>>> =20 >>>> /* >>>> @@ -161,6 +170,62 @@ irqfd_ptable_queue_proc(struct file *file, wait= _queue_head_t *wqh, >>>> add_wait_queue(wqh, &irqfd->wait); >>>> } >>>> =20 >>>> +/* >>>> + * create a host-wide workqueue for issuing deferred shutdown reque= sts >>>> + * aggregated from all vm* instances. We need our own isolated sing= le-thread >>>> + * queue to prevent deadlock against flushing the normal work-queue= =2E >>>> + */ >>>> +static int >>>> +irqfd_cleanup_init(struct kvm *kvm) >>>> +{ >>>> + int ret =3D 0; >>>> + >>>> + mutex_lock(&irqfd_cleanup.lock); >>>> + >>>> + /* >>>> + * Check the current init state from within the lock so that we >>>> + * sync all users to the thread creation. >>>> + */ >>>> + if (kvm->irqfds.init) >>>> + goto out; >>>> + >>>> + if (!irqfd_cleanup.refs) { >>>> + struct workqueue_struct *wq; >>>> + >>>> + wq =3D create_singlethread_workqueue("kvm-irqfd-cleanup"); >>>> + if (!wq) { >>>> + ret =3D -ENOMEM; >>>> + goto out; >>>> + } >>>> + >>>> + irqfd_cleanup.wq =3D wq; >>>> + } >>>> + >>>> + irqfd_cleanup.refs++; >>>> + kvm->irqfds.init =3D true; >>>> + >>>> +out: >>>> + mutex_unlock(&irqfd_cleanup.lock); >>>> + >>>> + return ret; >>>> +} >>>> + >>>> +static void >>>> +irqfd_cleanup_release(struct kvm *kvm) >>>> +{ >>>> + if (!kvm->irqfds.init) >>>> + return; >>>> =20 >>>> =20 >>> init is checked outside the lock here. >>> Why? >>> =20 >>> =20 >> Guest is shutting down via vmfd->f_ops->release() and is already >> guaranteed to be single-threaded, so proper locking is not really >> important. Probably should document that, though ;) >> =20 > > Is this an impirtant optimization? Simple locking is better.. > =20 Its not "important" from a performance perspective or anything like that. Its just that locking here is inappropriate. The init value is per-vm and we only need to guard it going from false->true in the assign path to cover the unlikely case that more than one threads tries to assign an irqfd at the same time. The shutdown path has no such requirement since the VM has already implicitly ceased to exist and can therefore not possibly assign another irqfd. Therefore, locking here, while harmless in every dimension, is simply gratuitous. I would rather just document it, personally: i.e. I think overlocking is just sloppy.=20 Otherwise we should be documenting that the lock isn't really needed, etc= =2E For the record, I would have rather just used my originally proposed slow-work thread instead of all this ;) -Greg > =20 >>>> + >>>> + mutex_lock(&irqfd_cleanup.lock); >>>> + >>>> + if (!(--irqfd_cleanup.refs)) >>>> + destroy_workqueue(irqfd_cleanup.wq); >>>> + >>>> + mutex_unlock(&irqfd_cleanup.lock); >>>> + >>>> + kvm->irqfds.init =3D false; >>>> =20 >>>> =20 >>> ... and cleaned outside the lock as well. >>> >>> =20 >>> =20 >> Ditto >> >> =20 >>>> +} >>>> + >>>> static int >>>> kvm_irqfd_assign(struct kvm *kvm, int fd, int gsi) >>>> { >>>> @@ -170,6 +235,10 @@ kvm_irqfd_assign(struct kvm *kvm, int fd, int g= si) >>>> int ret; >>>> unsigned int events; >>>> =20 >>>> + ret =3D irqfd_cleanup_init(kvm); >>>> + if (ret < 0) >>>> + return ret; >>>> + >>>> irqfd =3D kzalloc(sizeof(*irqfd), GFP_KERNEL); >>>> if (!irqfd) >>>> return -ENOMEM; >>>> @@ -268,7 +337,7 @@ kvm_irqfd_deassign(struct kvm *kvm, int fd, int = gsi) >>>> * so that we guarantee there will not be any more interrupts on t= his >>>> * gsi once this deassign function returns. >>>> */ >>>> - flush_workqueue(irqfd_cleanup_wq); >>>> + flush_workqueue(irqfd_cleanup.wq); >>>> =20 >>>> return 0; >>>> } >>>> @@ -302,28 +371,7 @@ kvm_irqfd_release(struct kvm *kvm) >>>> * Block until we know all outstanding shutdown jobs have complete= d >>>> * since we do not take a kvm* reference. >>>> */ >>>> - flush_workqueue(irqfd_cleanup_wq); >>>> - >>>> -} >>>> - >>>> -/* >>>> - * create a host-wide workqueue for issuing deferred shutdown reque= sts >>>> - * aggregated from all vm* instances. We need our own isolated sing= le-thread >>>> - * queue to prevent deadlock against flushing the normal work-queue= =2E >>>> - */ >>>> -static int __init irqfd_module_init(void) >>>> -{ >>>> - irqfd_cleanup_wq =3D create_singlethread_workqueue("kvm-irqfd-clea= nup"); >>>> - if (!irqfd_cleanup_wq) >>>> - return -ENOMEM; >>>> - >>>> - return 0; >>>> -} >>>> + flush_workqueue(irqfd_cleanup.wq); >>>> + irqfd_cleanup_release(kvm); >>>> =20 >>>> -static void __exit irqfd_module_exit(void) >>>> -{ >>>> - destroy_workqueue(irqfd_cleanup_wq); >>>> } >>>> - >>>> -module_init(irqfd_module_init); >>>> -module_exit(irqfd_module_exit); >>>> =20 >>>> =20 >>> -- >>> To unsubscribe from this list: send the line "unsubscribe linux-kerne= l" in >>> the body of a message to majordomo@vger.kernel.org >>> More majordomo info at http://vger.kernel.org/majordomo-info.html >>> Please read the FAQ at http://www.tux.org/lkml/ >>> =20 >>> =20 >> =20 > > > =20 --------------enig4534F80F3EEEC52661E832DA Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.11 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkpSJy4ACgkQlOSOBdgZUxlOiACeIOWI2WQPL8ZLI9FDzp1hqN72 kNQAn1HOSYeykMqPx0EmzY6PERyfsu6G =2kyJ -----END PGP SIGNATURE----- --------------enig4534F80F3EEEC52661E832DA--