Thread (4 messages) flat view 4 messages, 3 authors, 2012-12-10

Re: [PATCHv6] virtio-spec: virtio network device multiqueue support

From: Jason Wang <hidden>
Date: 2012-12-10 05:54:36
Also in: kvm, virtualization

On Friday, December 07, 2012 04:18:56 PM Michael S. Tsirkin wrote:
quoted hunk ↗ jump to hunk
Add multiqueue support to virtio network device.
Add a new feature flag VIRTIO_NET_F_MQ for this feature, a new
configuration field max_virtqueue_pairs to detect supported number of
virtqueues as well as a new command VIRTIO_NET_CTRL_MQ to program
packet steering for unidirectional protocols.

---

Changes in v6:
- rename RFS -> multiqueue to avoid confusion with RFS in linux
  mention automatic receive steering as Rusty suggested

Changes in v5:
- Address Rusty's comments.
  Changes are only in the text, not the ideas.
- Some minor formatting changes.

Changes in v4:
- address Jason's comments
- have configuration specify the number of VQ pairs and not pairs - 1

Changes in v3:
- rename multiqueue -> rfs this is what we support
- Be more explicit about what driver should do.
- Simplify layout making VQs functionality depend on feature.
- Remove unused commands, only leave in programming # of queues

Changes in v2:
Address Jason's comments on v2:
- Changed STEERING_HOST to STEERING_RX_FOLLOWS_TX:
   this is both clearer and easier to support.
   It does not look like we need a separate steering command
   since host can just watch tx packets as they go.
- Moved RX and TX steering sections near each other.
- Add motivation for other changes in v2

Changes in v1 (from Jason's rfc):
- reserved vq 3: this makes all rx vqs even and tx vqs odd, which
   looks nicer to me.
- documented packet steering, added a generalized steering programming
   command. Current modes are single queue and host driven multiqueue,
   but I envision support for guest driven multiqueue in the future.
- make default vqs unused when in mq mode - this wastes some memory
   but makes it more efficient to switch between modes as
   we can avoid this causing packet reordering.

Rusty, could you please take a look and comment soon?
If this looks OK to everyone, we can proceed with finalizing the
implementation. Would be nice to try and put it in 3.8.
diff --git a/virtio-spec.lyx b/virtio-spec.lyx
index 83f2771..c5b32c4 100644
--- a/virtio-spec.lyx
+++ b/virtio-spec.lyx
@@ -59,6 +59,7 @@
 \author -608949062 "Rusty Russell,,,"
 \author -385801441 "Cornelia Huck" cornelia.huck@de.ibm.com
 \author 1531152142 "Paolo Bonzini,,,"
+\author 1986246365 "Michael S. Tsirkin"
 \end_header
[...]
+
+\begin_layout Plain Layout
+
+\change_inserted 1986246365 1353594263
+
+#define VIRTIO_NET_CTRL_MQ    1
Should be 4 here.
+\end_layout
+
+\begin_layout Plain Layout
+
+\change_inserted 1986246365 1353594273
+
+ #define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_SET        0
+\end_layout
+
+\begin_layout Plain Layout
+
+\change_inserted 1986246365 1353594273
+
+ #define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MIN        1
+\end_layout
+
[...]
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help