Re: [PATCH v1] virtio-net: support reset queue

[Xuan Zhuo <xuanzhuo@linux.alibaba.com>]

On Mon, 21 Feb 2022 11:02:54 +0800, Jason Wang <jasowang@redhat.com> wrote:
> On Fri, Feb 18, 2022 at 5:27 PM Michael S. Tsirkin <mst@redhat.com> wrote:
> >
> > On Thu, Feb 10, 2022 at 05:26:24PM +0800, Jason Wang wrote:
> > > On Thu, Feb 10, 2022 at 5:06 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
> > > >
> > > > This patch defines some requirements for virtio-net to support reset
> > > > queues.
> > > >
> > > > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
> > > > ---
> > > >  conformance.tex |  1 +
> > > >  content.tex     | 12 ++++++++++++
> > > >  2 files changed, 13 insertions(+)
> > > >
> > > > diff --git a/conformance.tex b/conformance.tex
> > > > index 42f8537..3656da5 100644
> > > > --- a/conformance.tex
> > > > +++ b/conformance.tex
> > > > @@ -396,6 +396,7 @@ \section{Conformance Targets}\label{sec:Conformance / Conformance Targets}
> > > >  \item \ref{devicenormative:Device Types / Network Device / Device Operation / Packet Transmission}
> > > >  \item \ref{devicenormative:Device Types / Network Device / Device Operation / Setting Up Receive Buffers}
> > > >  \item \ref{devicenormative:Device Types / Network Device / Device Operation / Processing of Incoming Packets}
> > > > +\item \ref{devicenormative:Device Types / Network Device / Device Operation / Reset Virtqueue}
> > > >  \item \ref{devicenormative:Device Types / Network Device / Device Operation / Control Virtqueue / Packet Receive Filtering}
> > > >  \item \ref{devicenormative:Device Types / Network Device / Device Operation / Control Virtqueue / Setting MAC Address Filtering}
> > > >  \item \ref{devicenormative:Device Types / Network Device / Device Operation / Control Virtqueue / Gratuitous Packet Sending}
> > > > diff --git a/content.tex b/content.tex
> > > > index c6f116c..af86689 100644
> > > > --- a/content.tex
> > > > +++ b/content.tex
> > > > @@ -4000,6 +4000,18 @@ \subsubsection{Processing of Incoming Packets}\label{sec:Device Types / Network
> > > >  #define VIRTIO_NET_HASH_REPORT_UDPv6_EX        9
> > > >  \end{lstlisting}
> > > >
> > > > +\subsubsection{Reset Virtqueue}\label{sec:Device Types / Network Device / Device Operation / Reset Virtqueue}
> > > > +
> > > > +The receive and transmission virtqueues and the control virtqueue can implement
> > > > +reset based on Virtqueue Reset (See \ref{sec:Basic Facilities of a Virtio Device / Virtqueues / Virtqueue Reset}).
> > > > +
> > > > +\devicenormative{\paragraph}{Gratuitous Packet Sending}{Device Types / Network Device / Device Operation / Reset Virtqueue}
> > > > +
> > > > +After automatic receive steering or RSS receive steering has completed the
> > > > +selection of the queue, if the destination receive queue is in reset state,
> > > > +the device SHOULD re-select a different random queue.
> > >
> > > This seems break the semantics of the indirection table?
> >
> > I don't really understand this random thing either. If we don't want to
> > drop then maybe use unclassified_queue?
> > And generally - is packet reordering better or worse than packet drops?
>
> Not sure, but this is something we need to clarify in the spec.
>
> Personally, drop looks more safe than unclassified_queue. It gives
> hints to administrator that something is wrong.

I also think drop can be selected in this case. My v2 has already been sent, and
you can see if there is any problem with v2.

Thanks.

>
> Thanks
>
> >
> > > For RSS, I
> > > think we probably need to drop.
> > >
> > > Thanks
> > >
> > > > If all receive queues are
> > > > +in reset state, the device can drop the packet.
> > > > +
> > > >  \subsubsection{Control Virtqueue}\label{sec:Device Types / Network Device / Device Operation / Control Virtqueue}
> > > >
> > > >  The driver uses the control virtqueue (if VIRTIO_NET_F_CTRL_VQ is
> > > > --
> > > > 2.31.0
> > > >
> >
>