Re: [Qemu-devel] [virtio-dev] Re: [PATCH virtio 1/1] Add "Group Identifier" support to virtio PCI capabilities.

["Michael S. Tsirkin" <mst@redhat.com>]

On Tue, Jun 19, 2018 at 12:54:06PM -0500, Venu Busireddy wrote:
> On 2018-06-19 20:30:06 +0300, Michael S. Tsirkin wrote:
> > On Tue, Jun 19, 2018 at 11:32:28AM -0500, Venu Busireddy wrote:
> > > Add VIRTIO_PCI_CAP_GROUP_ID_CFG (Group Identifier) capability to the
> > > virtio PCI capabilities to allow for the grouping of devices.
> > > 
> > > Signed-off-by: Venu Busireddy <venu.busireddy@oracle.com>
> > > ---
> > >  content.tex | 43 +++++++++++++++++++++++++++++++++++++++++++
> > >  1 file changed, 43 insertions(+)
> > > 
> > > diff --git a/content.tex b/content.tex
> > > index 7a92cb1..7ea6267 100644
> > > --- a/content.tex
> > > +++ b/content.tex
> > > @@ -599,6 +599,8 @@ The fields are interpreted as follows:
> > >  #define VIRTIO_PCI_CAP_DEVICE_CFG        4
> > >  /* PCI configuration access */
> > >  #define VIRTIO_PCI_CAP_PCI_CFG           5
> > > +/* Group Identifier */
> > > +#define VIRTIO_PCI_CAP_GROUP_ID_CFG      6
> > >  \end{lstlisting}
> > >  
> > >          Any other value is reserved for future use.
> > > @@ -997,6 +999,47 @@ address \field{cap.length} bytes within a BAR range
> > >  specified by some other Virtio Structure PCI Capability
> > >  of type other than \field{VIRTIO_PCI_CAP_PCI_CFG}.
> > >  
> > > +\subsubsection{Group Identifier capability}\label{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Group Identifier capability}
> > > +
> > > +The VIRTIO_PCI_CAP_GROUP_ID_CFG capability provides means for grouping devices together.
> > > +
> > > +The capability is immediately followed by an identifier of arbitrary size as below:
> > > +
> > > +\begin{lstlisting}
> > > +struct virtio_pci_group_id_cap {
> > > +        struct virtio_pci_cap cap;
> > > +        u8 group_id[]; /* Group Identifier */
> > > +};
> > > +\end{lstlisting}
> > > +
> > > +The fields \field{cap.bar}, \field{cap.length}, \field{cap.offset}
> > > +and \field{group_id} are read-only for the driver.
> > > +
> > > +The specification does not impose any restrictions on the size or
> > > +structure of group_id[].
> > 
> > I think it must be a multiple of 4 in size, as is
> > standard for all capabilities.
> 
> Sure. Would rephrasing it as below suffice?
> 
> The specification does not impose any restrictions on the size or
> structure of group_id[], except that the size must be a multiple of 4.
> 
> > 
> > 
> > > Vendors

Devices

> are free to declare this array as
> > > +large as needed, as long as the combined size of all capabilities can
> > > +be accommodated within the PCI configuration space.
> > > +
> > > +If there is enough room in the PCI configuration space to accommodate
> > > +the group identifier, the fields \field{cap.bar}, \field{cap.offset}
> > > +and \field{cap.length} should be set to 0.
> > > +
> > > +If there isn't enough room, some or all of the group identifier can be
> > > +presented in the BAR region, in which case the fields \field{cap.bar},
> > > +\field{cap.offset} and \field{cap.length} should be set appropriately.
> > 
> > And then how do you glue the two pieces?
> 
> How the user glues them up is up to the user. The specification should
> not impose rules on that, right?

We need to define how these are matched.
Let's assume device A has it all in config space, device B
has part in memory. How would we compare them?




> > 
> > > +
> > > +In either case, the field \field{cap.cap_len} indicates the length of
> > > +the group identifier information present in the configuration space
> > > +itself.
> > 
> > It seems like an overkill to me. Isn't it enough to have it in config
> > space? This would make comparisons easier.
> 
> I was trying to make the proposal permissive for expansion, in case
> the user needs the size to be larger than what can be accommodated in
> the config space. Would you like me to restrict that the capability be
> entirely present in the config space? I am fine with it. Please confirm,
> and I will change it so.

I think so, yes.

> > 
> > > +
> > > +\devicenormative{\paragraph}{Group Identifier capability}{Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Group Identifier capability}
> > > +
> > > +The device MAY present the VIRTIO_PCI_CAP_GROUP_ID_CFG capability.
> > > +
> > > +\drivernormative{\paragraph}{Group Identifier capability}{Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Group Identifier capability}
> > > +
> > > +The driver MUST NOT write to group_id[] area or the BAR region.
> > > +
> > >  \subsubsection{Legacy Interfaces: A Note on PCI Device Layout}\label{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Legacy Interfaces: A Note on PCI Device Layout}
> > >  
> > >  Transitional devices MUST present part of configuration
> > 
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: virtio-dev-unsubscribe@lists.oasis-open.org
> > For additional commands, e-mail: virtio-dev-help@lists.oasis-open.org
> >