Re: [virtio-comment] [PATCH 1/1] content: Reserve virtio-nsm device ID

["Eftime, Petre" <epetre@amazon.com>]

On 2020-05-26 12:05, Stefan Hajnoczi wrote:
> On Thu, May 21, 2020 at 01:49:42PM +0300, Petre Eftime wrote:
> > The NitroSecureModule is a device with a very stripped down
> > Trusted Platform Module functionality, which is used in the
> > context of a Nitro Enclave (see https://lkml.org/lkml/2020/4/21/1020)
> > to provide boot time measurement and attestation.
> >
> > Since this device provides some critical cryptographic operations,
> > there are a series of operations which are required to have guarantees
> > of atomicity, ordering and consistency: operations fully succeed or fully
> > fail, including when some external events might interfere in the
> > process: live migration, crashes, etc; any failure in the critical
> > section requires termination of the enclave it is attached to, so
> > the device needs to be as resilient as possible, simplicity is
> > strongly desired.
> >
> > To account for that, the device and driver are made to have very few
> > error cases in the critical path and the operations themselves can be
> > rolled back and retried if events happen outside the critical
> > area, while processing a request. The driver itself can be made very
> > simple and thus is easily portable.
> >
> > Since the requests can be handled directly in the virtio queue, serving
> > most requests requires no additional buffering or memory allocations
> > on the host side.
> >
> > Signed-off-by: Petre Eftime <epetre@amazon.com>
> > ---
> >   content.tex | 2 ++
> >   1 file changed, 2 insertions(+)
> >
> > diff --git a/content.tex b/content.tex
> > index 91735e3..6eee371 100644
> > --- a/content.tex
> > +++ b/content.tex
> > @@ -2801,6 +2801,8 @@ \chapter{Device Types}\label{sec:Device Types}
> >   \hline
> >   31         &   Video decoder device \\
> >   \hline
> > +0xec2000   &   NitroSecureModule \\
> The PCI and CCW transports represent VIRTIO device type IDs with 16-bit
> fields. Is there a reason why the NitroSecureModule device should only
> work with VIRTIO transports that support >16-bit device type IDs?
>
> If not, then please pick the next low-numbered ID instead.
>
> Stefan

I saw that there's already a vote open for device ID 32, so I went with 
33 instead. I was under the impression that virtio device IDs are always 
32bit, based on the data types used in Linux to express them.

Best,

Petre Eftime




Amazon Development Center (Romania) S.R.L. registered office: 27A Sf. Lazar Street, UBC5, floor 2, Iasi, Iasi County, 700045, Romania. Registered in Romania. Registration number J22/2621/2005.