[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [RFC 2/9] virtio-pci: support host notifiers in TCG mod
From: |
Paolo Bonzini |
Subject: |
Re: [Qemu-devel] [RFC 2/9] virtio-pci: support host notifiers in TCG mode |
Date: |
Wed, 25 Jul 2012 08:50:51 +0200 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:13.0) Gecko/20120615 Thunderbird/13.0.1 |
Il 25/07/2012 00:33, Nicholas A. Bellinger ha scritto:
> From: Stefan Hajnoczi <address@hidden>
>
> Normally host notifiers are only used together with vhost-net in KVM
> mode. It is occassionally useful to use vhost with TCG mode, mainly for
> testing and development. This isn't hard to achieve, simply fall back
> to notifying the host notifier manually from qemu if KVM mode is
> disabled.
>
> Signed-off-by: Stefan Hajnoczi <address@hidden>
> Cc: Anthony Liguori <address@hidden>
> Cc: Paolo Bonzini <address@hidden>
> Signed-off-by: Nicholas Bellinger <address@hidden>
> ---
> hw/virtio-pci.c | 23 ++++++++++++++++++++---
> hw/virtio.c | 7 +++++++
> 2 files changed, 27 insertions(+), 3 deletions(-)
>
> diff --git a/hw/virtio-pci.c b/hw/virtio-pci.c
> index 4e03f0b..538eef4 100644
> --- a/hw/virtio-pci.c
> +++ b/hw/virtio-pci.c
> @@ -249,6 +249,25 @@ void virtio_pci_reset(DeviceState *d)
> proxy->flags &= ~VIRTIO_PCI_FLAG_BUS_MASTER_BUG;
> }
>
> +static void virtio_pci_queue_notify(VirtIOPCIProxy *proxy, uint32_t n)
> +{
> + VirtQueue *vq;
> + EventNotifier *notifier;
> +
> + if (n >= VIRTIO_PCI_QUEUE_MAX) {
> + return;
> + }
> +
> + vq = virtio_get_queue(proxy->vdev, n);
> + notifier = virtio_queue_get_host_notifier(vq);
> + if (event_notifier_valid(notifier)) {
> + printf("notifying vq %u host notifier from userspace\n", n);
Debug printf.
> + event_notifier_notify(notifier);
> + } else {
> + virtio_queue_notify_vq(vq);
> + }
This can be done directly in virtio_queue_notify, there is nothing
specific to virtio-pci.
> +}
> +
> static void virtio_ioport_write(void *opaque, uint32_t addr, uint32_t val)
> {
> VirtIOPCIProxy *proxy = opaque;
> @@ -278,9 +297,7 @@ static void virtio_ioport_write(void *opaque, uint32_t
> addr, uint32_t val)
> vdev->queue_sel = val;
> break;
> case VIRTIO_PCI_QUEUE_NOTIFY:
> - if (val < VIRTIO_PCI_QUEUE_MAX) {
> - virtio_queue_notify(vdev, val);
> - }
> + virtio_pci_queue_notify(proxy, val);
> break;
> case VIRTIO_PCI_STATUS:
> if (!(val & VIRTIO_CONFIG_S_DRIVER_OK)) {
> diff --git a/hw/virtio.c b/hw/virtio.c
> index d146f86..36a18b5 100644
> --- a/hw/virtio.c
> +++ b/hw/virtio.c
> @@ -536,6 +536,11 @@ void virtio_reset(void *opaque)
> vdev->vq[i].signalled_used = 0;
> vdev->vq[i].signalled_used_valid = false;
> vdev->vq[i].notification = true;
> +
> + assert(!event_notifier_valid(&vdev->vq[i].guest_notifier));
> + assert(!event_notifier_valid(&vdev->vq[i].host_notifier));
> + vdev->vq[i].guest_notifier = EVENT_NOTIFIER_INITIALIZER;
> + vdev->vq[i].host_notifier = EVENT_NOTIFIER_INITIALIZER;
Given the assertions, the assignments should be no-ops.
> }
> }
>
> @@ -905,6 +910,8 @@ VirtIODevice *virtio_common_init(const char *name,
> uint16_t device_id,
> for(i = 0; i < VIRTIO_PCI_QUEUE_MAX; i++) {
> vdev->vq[i].vector = VIRTIO_NO_VECTOR;
> vdev->vq[i].vdev = vdev;
> + vdev->vq[i].guest_notifier = EVENT_NOTIFIER_INITIALIZER;
> + vdev->vq[i].host_notifier = EVENT_NOTIFIER_INITIALIZER;
> }
>
> vdev->name = name;
>
- [Qemu-devel] [RFC 0/9] vhost-scsi: Add support for host virtualized target, Nicholas A. Bellinger, 2012/07/24
- [Qemu-devel] [RFC 1/9] notifier: add validity check and notify function, Nicholas A. Bellinger, 2012/07/24
- [Qemu-devel] [RFC 2/9] virtio-pci: support host notifiers in TCG mode, Nicholas A. Bellinger, 2012/07/24
- Re: [Qemu-devel] [RFC 2/9] virtio-pci: support host notifiers in TCG mode,
Paolo Bonzini <=
- [Qemu-devel] [RFC 3/9] virtio-pci: check that event notification worked, Nicholas A. Bellinger, 2012/07/24
- [Qemu-devel] [RFC 4/9] vhost: Pass device path to vhost_dev_init(), Nicholas A. Bellinger, 2012/07/24
- [Qemu-devel] [RFC 5/9] virtio-scsi: Add wwpn and tgpt properties, Nicholas A. Bellinger, 2012/07/24
- [Qemu-devel] [RFC 6/9] virtio-scsi: Open and initialize /dev/vhost-scsi, Nicholas A. Bellinger, 2012/07/24
- [Qemu-devel] [RFC 7/9] virtio-scsi: Start/stop vhost, Nicholas A. Bellinger, 2012/07/24
- [Qemu-devel] [RFC 8/9] virtio-scsi: Set max_target=0 during vhost-scsi operation, Nicholas A. Bellinger, 2012/07/24
- [Qemu-devel] [RFC 9/9] vhost-scsi: add -vhost-scsi host device, Nicholas A. Bellinger, 2012/07/24