Re: [PATCHv4] virtio-spec: virtio network device RFS support
From: "Michael S. Tsirkin" <mst@redhat.com>
Date: 2012-11-23 10:10:02
Also in:
kvm, virtualization
Subsystem:
the rest · Maintainer:
Linus Torvalds
On Fri, Nov 23, 2012 at 01:17:36PM +0800, Jason Wang wrote: ...
"specifying the number of the last transmit and receive queue that is going to be used; thus out of transmitq0..transmitqn and receiveq0..receiveqn where n=virtqueue_pairs will be used." In this description, looks like n+1 virtqueue pairs (include receiveq0 and transmitq0) could be used in RFS mode.
The intent was not to reserve any virt queue pairs. I hope I clarified this below. Thanks for the comments. Here's an incremental patch to address them.
diff --git a/virtio-spec.lyx b/virtio-spec.lyx
index e562335..53ddeec 100644
--- a/virtio-spec.lyx
+++ b/virtio-spec.lyx@@ -4384,7 +4384,7 @@ VIRTIO_NET_F_GUEST_ANNOUNCE(21) Guest can send gratuitous packets. \begin_layout Description \change_inserted 1986246365 1352742808 -VIRTIO_NET_F_RFS(2) Device supports Receive Flow Steering. +VIRTIO_NET_F_RFS(22) Device supports Receive Flow Steering. \change_unchanged \end_layout
@@ -4432,9 +4432,10 @@ N \emph default = \emph on -max_virtqueue_pairs +max_virtqueue_pairs - 1 \emph default ) that can be configured once VIRTIO_NET_F_RFS is negotiated. +Legal values for this field are 1 to 8000h. \change_unchanged
@@ -4496,7 +4497,7 @@ The initialization routine should identify the receive and transmission \change_inserted 1986246365 1352743942 If VIRTIO_NET_F_RFS feature bit is negotiated, \emph on -N=max_virtqueue_pairs +N=max_virtqueue_pairs-1 \emph default , otherwise identify \emph on
@@ -5464,7 +5465,7 @@ struct virtio_net_ctrl_rfs { \change_inserted 1986246365 1353594263 -#define VIRTIO_NET_CTRL_RFC 1 +#define VIRTIO_NET_CTRL_RFS 1 \end_layout \begin_layout Plain Layout
@@ -5474,6 +5475,19 @@ struct virtio_net_ctrl_rfs { #define VIRTIO_NET_CTRL_RFS_VQ_PAIRS_SET 0 \end_layout +\begin_layout Plain Layout + +\change_inserted 1986246365 1353594273 + + #define VIRTIO_NET_CTRL_RFS_VQ_PAIRS_MIN 1 +\end_layout + +\begin_layout Plain Layout + +\change_inserted 1986246365 1353594273 + + #define VIRTIO_NET_CTRL_RFS_VQ_PAIRS_MAX 0x8000 +\end_layout \end_inset
@@ -5484,14 +5498,19 @@ struct virtio_net_ctrl_rfs { \change_inserted 1986246365 1353594884 RFS acceleration is disabled by default. Driver enables RFS by executing the VIRTIO_NET_CTRL_RFS_VQ_PAIRS_SET command, - specifying the number of the last transmit and receive queue that is going + specifying the number of the transmit and receive queues that is going to be used; thus out of transmitq0..transmitqn and receiveq0..receiveqn where \emph on -n=virtqueue +n=virtqueue_pairs-1 \emph default -_pairs will be used. + will be used. All these virtqueues must have been pre-configured in advance. + The range of legal values for the +\emph on + virtqueue_pairs +\emph off + field is between 1 and 8000h. \end_layout \begin_layout Standard
@@ -5512,7 +5531,7 @@ Programming of the receive flow classificator is implicit. \change_inserted 1986246365 1353595040 RFS acceleration is disabled by setting \emph on -virtqueue_pairs = 0 +virtqueue_pairs = 1 \emph default (this is the default). Following this, driver should not transmit new packets on virtqueues other