From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: virtio-dev-return-6723-cohuck=redhat.com@lists.oasis-open.org Sender: List-Post: List-Help: List-Unsubscribe: List-Subscribe: Received: from lists.oasis-open.org (oasis-open.org [10.110.1.242]) by lists.oasis-open.org (Postfix) with ESMTP id 84705985D08 for ; Tue, 11 Feb 2020 11:33:18 +0000 (UTC) Date: Tue, 11 Feb 2020 06:33:05 -0500 From: "Michael S. Tsirkin" Message-ID: <20200211062205-mutt-send-email-mst@kernel.org> References: <8a4ea95d6d77a2814aaf6897b5517353289a098e.1581305609.git.zhabin@linux.alibaba.com> MIME-Version: 1.0 In-Reply-To: <8a4ea95d6d77a2814aaf6897b5517353289a098e.1581305609.git.zhabin@linux.alibaba.com> Subject: [virtio-dev] Re: [PATCH v2 1/5] virtio-mmio: add notify feature for per-queue Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable Content-Disposition: inline To: Zha Bin Cc: linux-kernel@vger.kernel.org, jasowang@redhat.com, slp@redhat.com, virtio-dev@lists.oasis-open.org, qemu-devel@nongnu.org, gerry@linux.alibaba.com, jing2.liu@linux.intel.com, chao.p.peng@linux.intel.com List-ID: On Mon, Feb 10, 2020 at 05:05:17PM +0800, Zha Bin wrote: > From: Liu Jiang >=20 > The standard virtio-mmio devices use notification register to signal > backend. This will cause vmexits and slow down the performance when we > passthrough the virtio-mmio devices to guest virtual machines. > We proposed to update virtio over MMIO spec to add the per-queue > notify feature VIRTIO_F_MMIO_NOTIFICATION[1]. It can allow the VMM to > configure notify location for each queue. >=20 > [1] https://lkml.org/lkml/2020/1/21/31 >=20 > Signed-off-by: Liu Jiang > Co-developed-by: Zha Bin > Signed-off-by: Zha Bin > Co-developed-by: Jing Liu > Signed-off-by: Jing Liu > Co-developed-by: Chao Peng > Signed-off-by: Chao Peng Hmm. Any way to make this static so we don't need base and multiplier? > --- > drivers/virtio/virtio_mmio.c | 37 ++++++++++++++++++++++++++++++++= +++-- > include/uapi/linux/virtio_config.h | 8 +++++++- > 2 files changed, 42 insertions(+), 3 deletions(-) >=20 > diff --git a/drivers/virtio/virtio_mmio.c b/drivers/virtio/virtio_mmio.c > index 97d5725..1733ab97 100644 > --- a/drivers/virtio/virtio_mmio.c > +++ b/drivers/virtio/virtio_mmio.c > @@ -90,6 +90,9 @@ struct virtio_mmio_device { > =09/* a list of queues so we can dispatch IRQs */ > =09spinlock_t lock; > =09struct list_head virtqueues; > + > +=09unsigned short notify_base; > +=09unsigned short notify_multiplier; > }; > =20 > struct virtio_mmio_vq_info { > @@ -98,6 +101,9 @@ struct virtio_mmio_vq_info { > =20 > =09/* the list node for the virtqueues list */ > =09struct list_head node; > + > +=09/* Notify Address*/ > +=09unsigned int notify_addr; > }; > =20 > =20 > @@ -119,13 +125,23 @@ static u64 vm_get_features(struct virtio_device *vd= ev) > =09return features; > } > =20 > +static void vm_transport_features(struct virtio_device *vdev, u64 featur= es) > +{ > +=09if (features & BIT_ULL(VIRTIO_F_MMIO_NOTIFICATION)) > +=09=09__virtio_set_bit(vdev, VIRTIO_F_MMIO_NOTIFICATION); > +} > + > static int vm_finalize_features(struct virtio_device *vdev) > { > =09struct virtio_mmio_device *vm_dev =3D to_virtio_mmio_device(vdev); > +=09u64 features =3D vdev->features; > =20 > =09/* Give virtio_ring a chance to accept features. */ > =09vring_transport_features(vdev); > =20 > +=09/* Give virtio_mmio a chance to accept features. */ > +=09vm_transport_features(vdev, features); > + > =09/* Make sure there is are no mixed devices */ > =09if (vm_dev->version =3D=3D 2 && > =09=09=09!__virtio_test_bit(vdev, VIRTIO_F_VERSION_1)) { > @@ -272,10 +288,13 @@ static void vm_reset(struct virtio_device *vdev) > static bool vm_notify(struct virtqueue *vq) > { > =09struct virtio_mmio_device *vm_dev =3D to_virtio_mmio_device(vq->vdev)= ; > +=09struct virtio_mmio_vq_info *info =3D vq->priv; > =20 > -=09/* We write the queue's selector into the notification register to > +=09/* We write the queue's selector into the Notify Address to > =09 * signal the other end */ > -=09writel(vq->index, vm_dev->base + VIRTIO_MMIO_QUEUE_NOTIFY); > +=09if (info) > +=09=09writel(vq->index, vm_dev->base + info->notify_addr); > + > =09return true; > } > =20 > @@ -434,6 +453,12 @@ static struct virtqueue *vm_setup_vq(struct virtio_d= evice *vdev, unsigned index, > =09vq->priv =3D info; > =09info->vq =3D vq; > =20 > +=09if (__virtio_test_bit(vdev, VIRTIO_F_MMIO_NOTIFICATION)) > +=09=09info->notify_addr =3D vm_dev->notify_base + > +=09=09=09=09vm_dev->notify_multiplier * vq->index; > +=09else > +=09=09info->notify_addr =3D VIRTIO_MMIO_QUEUE_NOTIFY; > + > =09spin_lock_irqsave(&vm_dev->lock, flags); > =09list_add(&info->node, &vm_dev->virtqueues); > =09spin_unlock_irqrestore(&vm_dev->lock, flags); > @@ -471,6 +496,14 @@ static int vm_find_vqs(struct virtio_device *vdev, u= nsigned nvqs, > =09=09return irq; > =09} > =20 > +=09if (__virtio_test_bit(vdev, VIRTIO_F_MMIO_NOTIFICATION)) { > +=09=09unsigned int notify =3D readl(vm_dev->base + > +=09=09=09=09VIRTIO_MMIO_QUEUE_NOTIFY); that register is documented as: /* Queue notifier - Write Only */ #define VIRTIO_MMIO_QUEUE_NOTIFY 0x050 so at least you need to update the doc. > + > +=09=09vm_dev->notify_base =3D notify & 0xffff; > +=09=09vm_dev->notify_multiplier =3D (notify >> 16) & 0xffff; are 16 bit base/limit always enough? In fact won't we be short on 16 bit address space in a rather short order if queues use up a page of space at a time? > +=09} > + > =09err =3D request_irq(irq, vm_interrupt, IRQF_SHARED, > =09=09=09dev_name(&vdev->dev), vm_dev); > =09if (err) > diff --git a/include/uapi/linux/virtio_config.h b/include/uapi/linux/virt= io_config.h > index ff8e7dc..5d93c01 100644 > --- a/include/uapi/linux/virtio_config.h > +++ b/include/uapi/linux/virtio_config.h > @@ -52,7 +52,7 @@ > * rest are per-device feature bits. > */ > #define VIRTIO_TRANSPORT_F_START=0928 > -#define VIRTIO_TRANSPORT_F_END=09=0938 > +#define VIRTIO_TRANSPORT_F_END=09=0940 > =20 > #ifndef VIRTIO_CONFIG_NO_LEGACY > /* Do we get callbacks when the ring is completely used, even if we've > @@ -88,4 +88,10 @@ > * Does the device support Single Root I/O Virtualization? > */ > #define VIRTIO_F_SR_IOV=09=09=0937 > + > +/* > + * This feature indicates the enhanced notification support on MMIO tran= sport > + * layer. Let's replace this with an actual description of the enhancement please otherwise it will not make sense in a couple of months. e.g. "Per queue notification address"? > + */ > +#define VIRTIO_F_MMIO_NOTIFICATION=0939 > #endif /* _UAPI_LINUX_VIRTIO_CONFIG_H */ > --=20 > 1.8.3.1 --------------------------------------------------------------------- To unsubscribe, e-mail: virtio-dev-unsubscribe@lists.oasis-open.org For additional commands, e-mail: virtio-dev-help@lists.oasis-open.org