QEMU-Devel Archive on lore.kernel.org
 help / color / Atom feed
* [PATCH 0/5] virtio: enable VIRTIO_F_RING_PACKED for all devices
@ 2020-05-22 17:17 Stefan Hajnoczi
  2020-05-22 17:17 ` [PATCH 1/5] tests/libqos: mask out VIRTIO_F_RING_PACKED for now Stefan Hajnoczi
                   ` (4 more replies)
  0 siblings, 5 replies; 9+ messages in thread
From: Stefan Hajnoczi @ 2020-05-22 17:17 UTC (permalink / raw)
  To: qemu-devel
  Cc: Kevin Wolf, Laurent Vivier, Thomas Huth, Eduardo Habkost,
	qemu-block, Michael S. Tsirkin, jasowang, cohuck,
	Stefan Hajnoczi, Dr. David Alan Gilbert, Raphael Norwitz,
	Gonglei (Arei),
	Gerd Hoffmann, Paolo Bonzini, Marc-André Lureau, Fam Zheng,
	Max Reitz

The VIRTIO 1.1 packed virtqueue layout improves performance and guest driver
support has been available since Linux v5.0. virtio-blk benchmarks show it is
beneficial for non-net devices too so I wrote patches to enable it for all
devices.

It turned out to be trickier than I expected because vhost feature negotiation
is currently not ready for new virtqueue feature bits like
VIRTIO_F_RING_PACKED. Patches 2-4 solve this.

Patch 5 then enables packed virtqueues.

Stefan Hajnoczi (5):
  tests/libqos: mask out VIRTIO_F_RING_PACKED for now
  vhost: involve device backends in feature negotiation
  vhost-user-blk: add VIRTIO_F_RING_PACKED feature bit
  vhost-scsi: add VIRTIO_F_VERSION_1 and VIRTIO_F_RING_PACKED
  virtio: enable VIRTIO_F_RING_PACKED for all devices

 include/hw/virtio/vhost.h        |  1 +
 include/hw/virtio/virtio-gpu.h   |  2 ++
 include/hw/virtio/virtio.h       |  2 +-
 include/sysemu/cryptodev-vhost.h | 11 +++++++++++
 backends/cryptodev-vhost.c       | 19 +++++++++++++++++++
 hw/block/vhost-user-blk.c        |  1 +
 hw/core/machine.c                | 18 +++++++++++++++++-
 hw/display/vhost-user-gpu.c      | 17 +++++++++++++++++
 hw/display/virtio-gpu-base.c     |  2 +-
 hw/input/vhost-user-input.c      |  9 +++++++++
 hw/scsi/vhost-scsi.c             |  2 ++
 hw/scsi/vhost-user-scsi.c        |  2 ++
 hw/virtio/vhost-user-fs.c        |  5 +++--
 hw/virtio/vhost-vsock.c          |  5 +++--
 hw/virtio/vhost.c                | 22 ++++++++++++++++++++++
 hw/virtio/virtio-crypto.c        |  3 ++-
 tests/qtest/libqos/virtio.c      |  3 ++-
 17 files changed, 115 insertions(+), 9 deletions(-)

-- 
2.25.3


^ permalink raw reply	[flat|nested] 9+ messages in thread

* [PATCH 1/5] tests/libqos: mask out VIRTIO_F_RING_PACKED for now
  2020-05-22 17:17 [PATCH 0/5] virtio: enable VIRTIO_F_RING_PACKED for all devices Stefan Hajnoczi
@ 2020-05-22 17:17 ` Stefan Hajnoczi
  2020-05-22 19:21   ` Thomas Huth
  2020-05-22 17:17 ` [PATCH 2/5] vhost: involve device backends in feature negotiation Stefan Hajnoczi
                   ` (3 subsequent siblings)
  4 siblings, 1 reply; 9+ messages in thread
From: Stefan Hajnoczi @ 2020-05-22 17:17 UTC (permalink / raw)
  To: qemu-devel
  Cc: Kevin Wolf, Laurent Vivier, Thomas Huth, Eduardo Habkost,
	qemu-block, Michael S. Tsirkin, jasowang, cohuck,
	Stefan Hajnoczi, Dr. David Alan Gilbert, Raphael Norwitz,
	Gonglei (Arei),
	Gerd Hoffmann, Paolo Bonzini, Marc-André Lureau, Fam Zheng,
	Max Reitz

The libqos VIRTIO code does not implement the packed virtqueue layout
yet. Mask out the feature bit for now because tests have a habit of
enabling all device feature bits and we don't want packed virtqueues to
be enabled.

Later patches will enable VIRTIO_F_RING_PACKED so prepare libqos now.

Cc: Thomas Huth <thuth@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 tests/qtest/libqos/virtio.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tests/qtest/libqos/virtio.c b/tests/qtest/libqos/virtio.c
index 9aa360620c..1c3f4a0c8b 100644
--- a/tests/qtest/libqos/virtio.c
+++ b/tests/qtest/libqos/virtio.c
@@ -96,7 +96,8 @@ uint64_t qvirtio_config_readq(QVirtioDevice *d, uint64_t addr)
 
 uint64_t qvirtio_get_features(QVirtioDevice *d)
 {
-    return d->bus->get_features(d);
+    /* qvirtio does not support packed virtqueues yet */
+    return d->bus->get_features(d) & ~(1ull << VIRTIO_F_RING_PACKED);
 }
 
 void qvirtio_set_features(QVirtioDevice *d, uint64_t features)
-- 
2.25.3


^ permalink raw reply	[flat|nested] 9+ messages in thread

* [PATCH 2/5] vhost: involve device backends in feature negotiation
  2020-05-22 17:17 [PATCH 0/5] virtio: enable VIRTIO_F_RING_PACKED for all devices Stefan Hajnoczi
  2020-05-22 17:17 ` [PATCH 1/5] tests/libqos: mask out VIRTIO_F_RING_PACKED for now Stefan Hajnoczi
@ 2020-05-22 17:17 ` Stefan Hajnoczi
  2020-05-22 17:17 ` [PATCH 3/5] vhost-user-blk: add VIRTIO_F_RING_PACKED feature bit Stefan Hajnoczi
                   ` (2 subsequent siblings)
  4 siblings, 0 replies; 9+ messages in thread
From: Stefan Hajnoczi @ 2020-05-22 17:17 UTC (permalink / raw)
  To: qemu-devel
  Cc: Kevin Wolf, Laurent Vivier, Thomas Huth, Eduardo Habkost,
	qemu-block, Michael S. Tsirkin, jasowang, cohuck,
	Stefan Hajnoczi, Dr. David Alan Gilbert, Raphael Norwitz,
	Gonglei (Arei),
	Gerd Hoffmann, Paolo Bonzini, Marc-André Lureau, Fam Zheng,
	Max Reitz

Many vhost devices in QEMU currently do not involve the device backend
in feature negotiation. This seems fine at first glance for device types
without their own feature bits (virtio-net has many but other device
types have none).

This overlooks the fact that QEMU's virtqueue implementation and the
device backend's implementation may support different features.  QEMU
must not report features to the guest that the the device backend
doesn't support.

For example, QEMU supports VIRTIO 1.1 packed virtqueues while many
existing vhost device backends do not. When the user sets packed=on the
device backend breaks. This should have been handled gracefully by
feature negotiation instead.

Introduce vhost_get_default_features() and update all vhost devices in
QEMU to involve the device backend in feature negotiation.

This patch fixes the following error:

  $ x86_64-softmmu/qemu-system-x86_64 \
      -drive if=virtio,file=test.img,format=raw \
      -chardev socket,path=/tmp/vhost-user-blk.sock,id=char0 \
      -device vhost-user-blk-pci,chardev=char0,packed=on \
      -object memory-backend-memfd,size=1G,share=on,id=ram0 \
      -M accel=kvm,memory-backend=ram0
  qemu-system-x86_64: Failed to set msg fds.
  qemu-system-x86_64: vhost VQ 0 ring restore failed: -1: Success (0)

The vhost-user-blk backend failed as follows:

  $ ./vhost-user-blk --socket-path=/tmp/vhost-user-blk.sock -b test2.img
  vu_panic: virtio: zero sized buffers are not allowed
  virtio-blk request missing headers

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/hw/virtio/vhost.h        |  1 +
 include/hw/virtio/virtio-gpu.h   |  2 ++
 include/sysemu/cryptodev-vhost.h | 11 +++++++++++
 backends/cryptodev-vhost.c       | 19 +++++++++++++++++++
 hw/display/vhost-user-gpu.c      | 17 +++++++++++++++++
 hw/display/virtio-gpu-base.c     |  2 +-
 hw/input/vhost-user-input.c      |  9 +++++++++
 hw/virtio/vhost-user-fs.c        |  5 +++--
 hw/virtio/vhost-vsock.c          |  5 +++--
 hw/virtio/vhost.c                | 22 ++++++++++++++++++++++
 hw/virtio/virtio-crypto.c        |  3 ++-
 11 files changed, 90 insertions(+), 6 deletions(-)

diff --git a/include/hw/virtio/vhost.h b/include/hw/virtio/vhost.h
index 085450c6f8..d2e54dd4a8 100644
--- a/include/hw/virtio/vhost.h
+++ b/include/hw/virtio/vhost.h
@@ -112,6 +112,7 @@ void vhost_virtqueue_mask(struct vhost_dev *hdev, VirtIODevice *vdev, int n,
                           bool mask);
 uint64_t vhost_get_features(struct vhost_dev *hdev, const int *feature_bits,
                             uint64_t features);
+uint64_t vhost_get_default_features(struct vhost_dev *hdev, uint64_t features);
 void vhost_ack_features(struct vhost_dev *hdev, const int *feature_bits,
                         uint64_t features);
 bool vhost_has_free_slot(void);
diff --git a/include/hw/virtio/virtio-gpu.h b/include/hw/virtio/virtio-gpu.h
index 6dd57f2025..41d270d80e 100644
--- a/include/hw/virtio/virtio-gpu.h
+++ b/include/hw/virtio/virtio-gpu.h
@@ -192,6 +192,8 @@ bool virtio_gpu_base_device_realize(DeviceState *qdev,
 void virtio_gpu_base_reset(VirtIOGPUBase *g);
 void virtio_gpu_base_fill_display_info(VirtIOGPUBase *g,
                         struct virtio_gpu_resp_display_info *dpy_info);
+uint64_t virtio_gpu_base_get_features(VirtIODevice *vdev, uint64_t features,
+                                      Error **errp);
 
 /* virtio-gpu.c */
 void virtio_gpu_ctrl_response(VirtIOGPU *g,
diff --git a/include/sysemu/cryptodev-vhost.h b/include/sysemu/cryptodev-vhost.h
index f42824fbde..e629446bfb 100644
--- a/include/sysemu/cryptodev-vhost.h
+++ b/include/sysemu/cryptodev-vhost.h
@@ -122,6 +122,17 @@ int cryptodev_vhost_start(VirtIODevice *dev, int total_queues);
  */
 void cryptodev_vhost_stop(VirtIODevice *dev, int total_queues);
 
+/**
+ * cryptodev_vhost_get_features:
+ * @dev: the virtio crypto object
+ * @requested_features: the features being offered
+ *
+ * Returns: the requested features bits that are supported by the vhost device,
+ * or the original request feature bits if vhost is disabled
+ *
+ */
+uint64_t cryptodev_vhost_get_features(VirtIODevice *dev, uint64_t features);
+
 /**
  * cryptodev_vhost_virtqueue_mask:
  * @dev: the virtio crypto object
diff --git a/backends/cryptodev-vhost.c b/backends/cryptodev-vhost.c
index 8337c9a495..5f5a4fda7b 100644
--- a/backends/cryptodev-vhost.c
+++ b/backends/cryptodev-vhost.c
@@ -266,6 +266,20 @@ void cryptodev_vhost_stop(VirtIODevice *dev, int total_queues)
     assert(r >= 0);
 }
 
+uint64_t cryptodev_vhost_get_features(VirtIODevice *dev, uint64_t features)
+{
+    VirtIOCrypto *vcrypto = VIRTIO_CRYPTO(dev);
+    CryptoDevBackend *b = vcrypto->cryptodev;
+    CryptoDevBackendClient *cc = b->conf.peers.ccs[0];
+    CryptoDevBackendVhost *vhost_crypto = cryptodev_get_vhost(cc, b, 0);
+
+    if (!vhost_crypto) {
+        return features; /* vhost disabled */
+    }
+
+    return vhost_get_default_features(&vhost_crypto->dev, features);
+}
+
 void cryptodev_vhost_virtqueue_mask(VirtIODevice *dev,
                                            int queue,
                                            int idx, bool mask)
@@ -333,6 +347,11 @@ void cryptodev_vhost_stop(VirtIODevice *dev, int total_queues)
 {
 }
 
+uint64_t cryptodev_vhost_get_features(VirtIODevice *dev, uint64_t features)
+{
+    return features;
+}
+
 void cryptodev_vhost_virtqueue_mask(VirtIODevice *dev,
                                     int queue,
                                     int idx, bool mask)
diff --git a/hw/display/vhost-user-gpu.c b/hw/display/vhost-user-gpu.c
index 4cdaee1bde..e483df2a9e 100644
--- a/hw/display/vhost-user-gpu.c
+++ b/hw/display/vhost-user-gpu.c
@@ -466,6 +466,22 @@ vhost_user_gpu_set_config(VirtIODevice *vdev,
     }
 }
 
+static uint64_t
+vhost_user_gpu_get_features(VirtIODevice *vdev, uint64_t features,
+                            Error **errp)
+{
+    VhostUserGPU *g = VHOST_USER_GPU(vdev);
+    Error *local_err = NULL;
+
+    features = virtio_gpu_base_get_features(vdev, features, &local_err);
+    if (local_err) {
+        error_propagate(errp, local_err);
+        return 0;
+    }
+
+    return vhost_get_default_features(&g->vhost->dev, features);
+}
+
 static void
 vhost_user_gpu_set_status(VirtIODevice *vdev, uint8_t val)
 {
@@ -582,6 +598,7 @@ vhost_user_gpu_class_init(ObjectClass *klass, void *data)
 
     vdc->realize = vhost_user_gpu_device_realize;
     vdc->reset = vhost_user_gpu_reset;
+    vdc->get_features = vhost_user_gpu_get_features;
     vdc->set_status   = vhost_user_gpu_set_status;
     vdc->guest_notifier_mask = vhost_user_gpu_guest_notifier_mask;
     vdc->guest_notifier_pending = vhost_user_gpu_guest_notifier_pending;
diff --git a/hw/display/virtio-gpu-base.c b/hw/display/virtio-gpu-base.c
index c159351be3..05d1ff2db2 100644
--- a/hw/display/virtio-gpu-base.c
+++ b/hw/display/virtio-gpu-base.c
@@ -176,7 +176,7 @@ virtio_gpu_base_device_realize(DeviceState *qdev,
     return true;
 }
 
-static uint64_t
+uint64_t
 virtio_gpu_base_get_features(VirtIODevice *vdev, uint64_t features,
                              Error **errp)
 {
diff --git a/hw/input/vhost-user-input.c b/hw/input/vhost-user-input.c
index 63984a8ba7..1371fb32cc 100644
--- a/hw/input/vhost-user-input.c
+++ b/hw/input/vhost-user-input.c
@@ -45,6 +45,14 @@ static void vhost_input_change_active(VirtIOInput *vinput)
     }
 }
 
+static uint64_t vhost_input_get_features(VirtIODevice *vdev, uint64_t features,
+                                         Error **errp)
+{
+    VHostUserInput *vhi = VHOST_USER_INPUT(vdev);
+
+    return vhost_get_default_features(&vhi->vhost->dev, features);
+}
+
 static void vhost_input_get_config(VirtIODevice *vdev, uint8_t *config_data)
 {
     VirtIOInput *vinput = VIRTIO_INPUT(vdev);
@@ -89,6 +97,7 @@ static void vhost_input_class_init(ObjectClass *klass, void *data)
     DeviceClass *dc = DEVICE_CLASS(klass);
 
     dc->vmsd = &vmstate_vhost_input;
+    vdc->get_features = vhost_input_get_features;
     vdc->get_config = vhost_input_get_config;
     vdc->set_config = vhost_input_set_config;
     vic->realize = vhost_input_realize;
diff --git a/hw/virtio/vhost-user-fs.c b/hw/virtio/vhost-user-fs.c
index 1bc5d03a00..56015ca3d4 100644
--- a/hw/virtio/vhost-user-fs.c
+++ b/hw/virtio/vhost-user-fs.c
@@ -130,8 +130,9 @@ static uint64_t vuf_get_features(VirtIODevice *vdev,
                                       uint64_t requested_features,
                                       Error **errp)
 {
-    /* No feature bits used yet */
-    return requested_features;
+    VHostUserFS *fs = VHOST_USER_FS(vdev);
+
+    return vhost_get_default_features(&fs->vhost_dev, requested_features);
 }
 
 static void vuf_handle_output(VirtIODevice *vdev, VirtQueue *vq)
diff --git a/hw/virtio/vhost-vsock.c b/hw/virtio/vhost-vsock.c
index 4a228f5168..7276587be6 100644
--- a/hw/virtio/vhost-vsock.c
+++ b/hw/virtio/vhost-vsock.c
@@ -180,8 +180,9 @@ static uint64_t vhost_vsock_get_features(VirtIODevice *vdev,
                                          uint64_t requested_features,
                                          Error **errp)
 {
-    /* No feature bits used yet */
-    return requested_features;
+    VHostVSock *vsock = VHOST_VSOCK(vdev);
+
+    return vhost_get_default_features(&vsock->vhost_dev, requested_features);
 }
 
 static void vhost_vsock_handle_output(VirtIODevice *vdev, VirtQueue *vq)
diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index aff98a0ede..f8a144dcd0 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -48,6 +48,23 @@ static unsigned int used_memslots;
 static QLIST_HEAD(, vhost_dev) vhost_devices =
     QLIST_HEAD_INITIALIZER(vhost_devices);
 
+/*
+ * Feature bits that device backends must explicitly report. Feature bits not
+ * listed here maybe set by QEMU without checking with the device backend.
+ * Ideally all feature bits would be listed here but existing vhost device
+ * implementations do not explicitly report bits like VIRTIO_F_VERSION_1, so we
+ * can only assume they are supported.
+ *
+ * New feature bits added to the VIRTIO spec should usually be included here
+ * so that existing vhost device backends that do not support them yet continue
+ * to work.
+ */
+static const int vhost_default_feature_bits[] = {
+    VIRTIO_F_IOMMU_PLATFORM,
+    VIRTIO_F_RING_PACKED,
+    VHOST_INVALID_FEATURE_BIT
+};
+
 bool vhost_has_free_slot(void)
 {
     unsigned int slots_limit = ~0U;
@@ -1468,6 +1485,11 @@ uint64_t vhost_get_features(struct vhost_dev *hdev, const int *feature_bits,
     return features;
 }
 
+uint64_t vhost_get_default_features(struct vhost_dev *hdev, uint64_t features)
+{
+    return vhost_get_features(hdev, vhost_default_feature_bits, features);
+}
+
 void vhost_ack_features(struct vhost_dev *hdev, const int *feature_bits,
                         uint64_t features)
 {
diff --git a/hw/virtio/virtio-crypto.c b/hw/virtio/virtio-crypto.c
index bd9165c565..ef711b56f4 100644
--- a/hw/virtio/virtio-crypto.c
+++ b/hw/virtio/virtio-crypto.c
@@ -739,7 +739,8 @@ static uint64_t virtio_crypto_get_features(VirtIODevice *vdev,
                                            uint64_t features,
                                            Error **errp)
 {
-    return features;
+    /* Just returns features when vhost is disabled */
+    return cryptodev_vhost_get_features(vdev, features);
 }
 
 static void virtio_crypto_reset(VirtIODevice *vdev)
-- 
2.25.3


^ permalink raw reply	[flat|nested] 9+ messages in thread

* [PATCH 3/5] vhost-user-blk: add VIRTIO_F_RING_PACKED feature bit
  2020-05-22 17:17 [PATCH 0/5] virtio: enable VIRTIO_F_RING_PACKED for all devices Stefan Hajnoczi
  2020-05-22 17:17 ` [PATCH 1/5] tests/libqos: mask out VIRTIO_F_RING_PACKED for now Stefan Hajnoczi
  2020-05-22 17:17 ` [PATCH 2/5] vhost: involve device backends in feature negotiation Stefan Hajnoczi
@ 2020-05-22 17:17 ` Stefan Hajnoczi
  2020-05-25  4:06   ` Raphael Norwitz
  2020-05-22 17:17 ` [PATCH 4/5] vhost-scsi: add VIRTIO_F_VERSION_1 and VIRTIO_F_RING_PACKED Stefan Hajnoczi
  2020-05-22 17:17 ` [PATCH 5/5] virtio: enable VIRTIO_F_RING_PACKED for all devices Stefan Hajnoczi
  4 siblings, 1 reply; 9+ messages in thread
From: Stefan Hajnoczi @ 2020-05-22 17:17 UTC (permalink / raw)
  To: qemu-devel
  Cc: Kevin Wolf, Laurent Vivier, Thomas Huth, Eduardo Habkost,
	qemu-block, Michael S. Tsirkin, jasowang, cohuck,
	Stefan Hajnoczi, Dr. David Alan Gilbert, Raphael Norwitz,
	Gonglei (Arei),
	Gerd Hoffmann, Paolo Bonzini, Marc-André Lureau, Fam Zheng,
	Max Reitz

Vhost devices have a list of feature bits that the device backend is
allowed to control. The VIRTIO_F_RING_PACKED feature is a feature that
must be negotiated through all the way to the device backend. Add it so
the device backend can declare whether or not it supports the packed
ring layout.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 hw/block/vhost-user-blk.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/hw/block/vhost-user-blk.c b/hw/block/vhost-user-blk.c
index 9d8c0b3909..10e114a19a 100644
--- a/hw/block/vhost-user-blk.c
+++ b/hw/block/vhost-user-blk.c
@@ -44,6 +44,7 @@ static const int user_feature_bits[] = {
     VIRTIO_BLK_F_DISCARD,
     VIRTIO_BLK_F_WRITE_ZEROES,
     VIRTIO_F_VERSION_1,
+    VIRTIO_F_RING_PACKED,
     VIRTIO_RING_F_INDIRECT_DESC,
     VIRTIO_RING_F_EVENT_IDX,
     VIRTIO_F_NOTIFY_ON_EMPTY,
-- 
2.25.3


^ permalink raw reply	[flat|nested] 9+ messages in thread

* [PATCH 4/5] vhost-scsi: add VIRTIO_F_VERSION_1 and VIRTIO_F_RING_PACKED
  2020-05-22 17:17 [PATCH 0/5] virtio: enable VIRTIO_F_RING_PACKED for all devices Stefan Hajnoczi
                   ` (2 preceding siblings ...)
  2020-05-22 17:17 ` [PATCH 3/5] vhost-user-blk: add VIRTIO_F_RING_PACKED feature bit Stefan Hajnoczi
@ 2020-05-22 17:17 ` Stefan Hajnoczi
  2020-05-25  4:02   ` Raphael Norwitz
  2020-05-22 17:17 ` [PATCH 5/5] virtio: enable VIRTIO_F_RING_PACKED for all devices Stefan Hajnoczi
  4 siblings, 1 reply; 9+ messages in thread
From: Stefan Hajnoczi @ 2020-05-22 17:17 UTC (permalink / raw)
  To: qemu-devel
  Cc: Kevin Wolf, Laurent Vivier, Thomas Huth, Eduardo Habkost,
	qemu-block, Michael S. Tsirkin, jasowang, cohuck,
	Stefan Hajnoczi, Dr. David Alan Gilbert, Raphael Norwitz,
	Gonglei (Arei),
	Gerd Hoffmann, Paolo Bonzini, Marc-André Lureau, Fam Zheng,
	Max Reitz

Let vhost-scsi and vhost-user-scsi device backends determine whether
VIRTIO 1.0 and packed virtqueues are supported. It doesn't make sense to
handle these feature bits in QEMU since the device backend needs to
support them if we want to use them.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 hw/scsi/vhost-scsi.c      | 2 ++
 hw/scsi/vhost-user-scsi.c | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/hw/scsi/vhost-scsi.c b/hw/scsi/vhost-scsi.c
index c1b012aea4..a7fb788af5 100644
--- a/hw/scsi/vhost-scsi.c
+++ b/hw/scsi/vhost-scsi.c
@@ -34,6 +34,8 @@
 
 /* Features supported by host kernel. */
 static const int kernel_feature_bits[] = {
+    VIRTIO_F_VERSION_1,
+    VIRTIO_F_RING_PACKED,
     VIRTIO_F_NOTIFY_ON_EMPTY,
     VIRTIO_RING_F_INDIRECT_DESC,
     VIRTIO_RING_F_EVENT_IDX,
diff --git a/hw/scsi/vhost-user-scsi.c b/hw/scsi/vhost-user-scsi.c
index cbb5d97599..6aa0d5ded2 100644
--- a/hw/scsi/vhost-user-scsi.c
+++ b/hw/scsi/vhost-user-scsi.c
@@ -32,6 +32,8 @@
 
 /* Features supported by the host application */
 static const int user_feature_bits[] = {
+    VIRTIO_F_VERSION_1,
+    VIRTIO_F_RING_PACKED,
     VIRTIO_F_NOTIFY_ON_EMPTY,
     VIRTIO_RING_F_INDIRECT_DESC,
     VIRTIO_RING_F_EVENT_IDX,
-- 
2.25.3


^ permalink raw reply	[flat|nested] 9+ messages in thread

* [PATCH 5/5] virtio: enable VIRTIO_F_RING_PACKED for all devices
  2020-05-22 17:17 [PATCH 0/5] virtio: enable VIRTIO_F_RING_PACKED for all devices Stefan Hajnoczi
                   ` (3 preceding siblings ...)
  2020-05-22 17:17 ` [PATCH 4/5] vhost-scsi: add VIRTIO_F_VERSION_1 and VIRTIO_F_RING_PACKED Stefan Hajnoczi
@ 2020-05-22 17:17 ` Stefan Hajnoczi
  4 siblings, 0 replies; 9+ messages in thread
From: Stefan Hajnoczi @ 2020-05-22 17:17 UTC (permalink / raw)
  To: qemu-devel
  Cc: Kevin Wolf, Laurent Vivier, Thomas Huth, Eduardo Habkost,
	qemu-block, Michael S. Tsirkin, jasowang, cohuck,
	Stefan Hajnoczi, Dr. David Alan Gilbert, Raphael Norwitz,
	Gonglei (Arei),
	Gerd Hoffmann, Paolo Bonzini, Marc-André Lureau, Fam Zheng,
	Max Reitz

The packed virtqueue layout was introduced in VIRTIO 1.1. It is a single
ring instead of a split avail/used ring design. There are CPU cache
advantages to this layout and it is also suited better to hardware
implementation.

The vhost-net backend has already supported packed virtqueues for some
time. Performance benchmarks show that virtio-blk performance on NVMe
drives is also improved.

Go ahead and enable this feature for all VIRTIO devices. Keep it
disabled for QEMU 5.0 and earlier machine types.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/hw/virtio/virtio.h |  2 +-
 hw/core/machine.c          | 18 +++++++++++++++++-
 2 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/include/hw/virtio/virtio.h b/include/hw/virtio/virtio.h
index b69d517496..fd5b4a2044 100644
--- a/include/hw/virtio/virtio.h
+++ b/include/hw/virtio/virtio.h
@@ -292,7 +292,7 @@ typedef struct VirtIORNGConf VirtIORNGConf;
     DEFINE_PROP_BIT64("iommu_platform", _state, _field, \
                       VIRTIO_F_IOMMU_PLATFORM, false), \
     DEFINE_PROP_BIT64("packed", _state, _field, \
-                      VIRTIO_F_RING_PACKED, false)
+                      VIRTIO_F_RING_PACKED, true)
 
 hwaddr virtio_queue_get_desc_addr(VirtIODevice *vdev, int n);
 bool virtio_queue_enabled(VirtIODevice *vdev, int n);
diff --git a/hw/core/machine.c b/hw/core/machine.c
index bb3a7b18b1..3598c3c825 100644
--- a/hw/core/machine.c
+++ b/hw/core/machine.c
@@ -28,7 +28,23 @@
 #include "hw/mem/nvdimm.h"
 #include "migration/vmstate.h"
 
-GlobalProperty hw_compat_5_0[] = {};
+GlobalProperty hw_compat_5_0[] = {
+    { "vhost-user-blk", "packed", "off" },
+    { "vhost-user-fs-device", "packed", "off" },
+    { "vhost-vsock-device", "packed", "off" },
+    { "virtio-9p-device", "packed", "off" },
+    { "virtio-balloon-device", "packed", "off" },
+    { "virtio-blk-device", "packed", "off" },
+    { "virtio-crypto-device", "packed", "off" },
+    { "virtio-gpu-device", "packed", "off" },
+    { "virtio-input-device", "packed", "off" },
+    { "virtio-iommu-device", "packed", "off" },
+    { "virtio-net-device", "packed", "off" },
+    { "virtio-pmem", "packed", "off" },
+    { "virtio-rng-device", "packed", "off" },
+    { "virtio-scsi-common", "packed", "off" },
+    { "virtio-serial-device", "packed", "off" },
+};
 const size_t hw_compat_5_0_len = G_N_ELEMENTS(hw_compat_5_0);
 
 GlobalProperty hw_compat_4_2[] = {
-- 
2.25.3


^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [PATCH 1/5] tests/libqos: mask out VIRTIO_F_RING_PACKED for now
  2020-05-22 17:17 ` [PATCH 1/5] tests/libqos: mask out VIRTIO_F_RING_PACKED for now Stefan Hajnoczi
@ 2020-05-22 19:21   ` Thomas Huth
  0 siblings, 0 replies; 9+ messages in thread
From: Thomas Huth @ 2020-05-22 19:21 UTC (permalink / raw)
  To: Stefan Hajnoczi
  Cc: Kevin Wolf, Laurent Vivier, Eduardo Habkost, qemu-block,
	Michael S. Tsirkin, jasowang, cohuck, qemu-devel,
	Raphael Norwitz, Gonglei (Arei),
	Gerd Hoffmann, Paolo Bonzini, Marc-André Lureau, Fam Zheng,
	Max Reitz, Dr. David Alan Gilbert

> From: "Stefan Hajnoczi" <stefanha@redhat.com>
> Sent: Friday, May 22, 2020 7:17:22 PM
> 
> The libqos VIRTIO code does not implement the packed virtqueue layout
> yet. Mask out the feature bit for now because tests have a habit of
> enabling all device feature

Sounds like we should rather fix these tests in the long run - they
should really only enable the bits that they support...

> bits and we don't want packed virtqueues to
> be enabled.
> 
> Later patches will enable VIRTIO_F_RING_PACKED so prepare libqos now.
> 
> Cc: Thomas Huth <thuth@redhat.com>
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> ---
>  tests/qtest/libqos/virtio.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/tests/qtest/libqos/virtio.c b/tests/qtest/libqos/virtio.c
> index 9aa360620c..1c3f4a0c8b 100644
> --- a/tests/qtest/libqos/virtio.c
> +++ b/tests/qtest/libqos/virtio.c
> @@ -96,7 +96,8 @@ uint64_t qvirtio_config_readq(QVirtioDevice *d, uint64_t
> addr)
>  
>  uint64_t qvirtio_get_features(QVirtioDevice *d)
>  {
> -    return d->bus->get_features(d);
> +    /* qvirtio does not support packed virtqueues yet */
> +    return d->bus->get_features(d) & ~(1ull << VIRTIO_F_RING_PACKED);
>  }

... but as a temporary work-around, that should be fine, too.
(in case you respin, maybe add a TODO comment here, too, to remind us to fix
the tests later).

Acked-by: Thomas Huth <thuth@redhat.com>



^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [PATCH 4/5] vhost-scsi: add VIRTIO_F_VERSION_1 and VIRTIO_F_RING_PACKED
  2020-05-22 17:17 ` [PATCH 4/5] vhost-scsi: add VIRTIO_F_VERSION_1 and VIRTIO_F_RING_PACKED Stefan Hajnoczi
@ 2020-05-25  4:02   ` Raphael Norwitz
  0 siblings, 0 replies; 9+ messages in thread
From: Raphael Norwitz @ 2020-05-25  4:02 UTC (permalink / raw)
  To: Stefan Hajnoczi
  Cc: Kevin Wolf, Laurent Vivier, Thomas Huth, Eduardo Habkost,
	qemu-block, Michael S. Tsirkin, jasowang, cohuck, QEMU,
	Dr. David Alan Gilbert, Gonglei (Arei),
	Gerd Hoffmann, Marc-André Lureau, Paolo Bonzini, Fam Zheng,
	Raphael Norwitz, Max Reitz

On Fri, May 22, 2020 at 1:19 PM Stefan Hajnoczi <stefanha@redhat.com> wrote:
>
> Let vhost-scsi and vhost-user-scsi device backends determine whether
> VIRTIO 1.0 and packed virtqueues are supported. It doesn't make sense to
> handle these feature bits in QEMU since the device backend needs to
> support them if we want to use them.
>
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Reviewed-by: Raphael Norwitz <raphael.norwitz@nutanix.com>

> ---
>  hw/scsi/vhost-scsi.c      | 2 ++
>  hw/scsi/vhost-user-scsi.c | 2 ++
>  2 files changed, 4 insertions(+)
>
> diff --git a/hw/scsi/vhost-scsi.c b/hw/scsi/vhost-scsi.c
> index c1b012aea4..a7fb788af5 100644
> --- a/hw/scsi/vhost-scsi.c
> +++ b/hw/scsi/vhost-scsi.c
> @@ -34,6 +34,8 @@
>
>  /* Features supported by host kernel. */
>  static const int kernel_feature_bits[] = {
> +    VIRTIO_F_VERSION_1,
> +    VIRTIO_F_RING_PACKED,
>      VIRTIO_F_NOTIFY_ON_EMPTY,
>      VIRTIO_RING_F_INDIRECT_DESC,
>      VIRTIO_RING_F_EVENT_IDX,
> diff --git a/hw/scsi/vhost-user-scsi.c b/hw/scsi/vhost-user-scsi.c
> index cbb5d97599..6aa0d5ded2 100644
> --- a/hw/scsi/vhost-user-scsi.c
> +++ b/hw/scsi/vhost-user-scsi.c
> @@ -32,6 +32,8 @@
>
>  /* Features supported by the host application */
>  static const int user_feature_bits[] = {
> +    VIRTIO_F_VERSION_1,
> +    VIRTIO_F_RING_PACKED,
>      VIRTIO_F_NOTIFY_ON_EMPTY,
>      VIRTIO_RING_F_INDIRECT_DESC,
>      VIRTIO_RING_F_EVENT_IDX,
> --
> 2.25.3
>


^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [PATCH 3/5] vhost-user-blk: add VIRTIO_F_RING_PACKED feature bit
  2020-05-22 17:17 ` [PATCH 3/5] vhost-user-blk: add VIRTIO_F_RING_PACKED feature bit Stefan Hajnoczi
@ 2020-05-25  4:06   ` Raphael Norwitz
  0 siblings, 0 replies; 9+ messages in thread
From: Raphael Norwitz @ 2020-05-25  4:06 UTC (permalink / raw)
  To: Stefan Hajnoczi
  Cc: Kevin Wolf, Laurent Vivier, Thomas Huth, Eduardo Habkost,
	qemu-block, Michael S. Tsirkin, jasowang, cohuck, QEMU,
	Dr. David Alan Gilbert, Gonglei (Arei),
	Gerd Hoffmann, Marc-André Lureau, Paolo Bonzini, Fam Zheng,
	Raphael Norwitz, Max Reitz

On Fri, May 22, 2020 at 1:20 PM Stefan Hajnoczi <stefanha@redhat.com> wrote:
>
> Vhost devices have a list of feature bits that the device backend is
> allowed to control. The VIRTIO_F_RING_PACKED feature is a feature that
> must be negotiated through all the way to the device backend. Add it so
> the device backend can declare whether or not it supports the packed
> ring layout.
>
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Reviewed-by: Raphael Norwitz <raphael.norwitz@nutanix.com>

> ---
>  hw/block/vhost-user-blk.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/hw/block/vhost-user-blk.c b/hw/block/vhost-user-blk.c
> index 9d8c0b3909..10e114a19a 100644
> --- a/hw/block/vhost-user-blk.c
> +++ b/hw/block/vhost-user-blk.c
> @@ -44,6 +44,7 @@ static const int user_feature_bits[] = {
>      VIRTIO_BLK_F_DISCARD,
>      VIRTIO_BLK_F_WRITE_ZEROES,
>      VIRTIO_F_VERSION_1,
> +    VIRTIO_F_RING_PACKED,
>      VIRTIO_RING_F_INDIRECT_DESC,
>      VIRTIO_RING_F_EVENT_IDX,
>      VIRTIO_F_NOTIFY_ON_EMPTY,
> --
> 2.25.3
>


^ permalink raw reply	[flat|nested] 9+ messages in thread

end of thread, back to index

Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-05-22 17:17 [PATCH 0/5] virtio: enable VIRTIO_F_RING_PACKED for all devices Stefan Hajnoczi
2020-05-22 17:17 ` [PATCH 1/5] tests/libqos: mask out VIRTIO_F_RING_PACKED for now Stefan Hajnoczi
2020-05-22 19:21   ` Thomas Huth
2020-05-22 17:17 ` [PATCH 2/5] vhost: involve device backends in feature negotiation Stefan Hajnoczi
2020-05-22 17:17 ` [PATCH 3/5] vhost-user-blk: add VIRTIO_F_RING_PACKED feature bit Stefan Hajnoczi
2020-05-25  4:06   ` Raphael Norwitz
2020-05-22 17:17 ` [PATCH 4/5] vhost-scsi: add VIRTIO_F_VERSION_1 and VIRTIO_F_RING_PACKED Stefan Hajnoczi
2020-05-25  4:02   ` Raphael Norwitz
2020-05-22 17:17 ` [PATCH 5/5] virtio: enable VIRTIO_F_RING_PACKED for all devices Stefan Hajnoczi

QEMU-Devel Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/qemu-devel/0 qemu-devel/git/0.git
	git clone --mirror https://lore.kernel.org/qemu-devel/1 qemu-devel/git/1.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 qemu-devel qemu-devel/ https://lore.kernel.org/qemu-devel \
		qemu-devel@nongnu.org
	public-inbox-index qemu-devel

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.nongnu.qemu-devel


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git