On Sun, 2026-08-16 at 11:03 +0200, Philippe Mathieu-Daudé wrote:
> On 15/6/26 22:38, Marc-André Lureau wrote:
> > Replace the hand-written SpdmTransportType C enum and QEnumLookup
> > table with a QAPI schema definition in qapi/sockets.json. Update
> > all users in hw/nvme/ctrl.c to use the QAPI-generated enum
> > constants
> > (SPDM_TRANSPORT_TYPE_* instead of SPDM_SOCKET_TRANSPORT_TYPE_*).
> > 
> > Signed-off-by: Marc-André Lureau <[email protected]>

Reviewed-by: Alistair Francis <[email protected]>

Alistair

> > ---
> >   qapi/sockets.json            | 20 ++++++++++++++++++++
> >   include/system/spdm-socket.h | 28 +++++++++++-----------------
> >   backends/spdm-socket.c       | 23 +++++++----------------
> >   hw/nvme/ctrl.c               | 26 +++++++++++++-------------
> >   4 files changed, 51 insertions(+), 46 deletions(-)
> > 
> > diff --git a/qapi/sockets.json b/qapi/sockets.json
> > index 473be2ac58a..dbb452cfb72 100644
> > --- a/qapi/sockets.json
> > +++ b/qapi/sockets.json
> > @@ -252,3 +252,23 @@
> >               'unix': 'UnixSocketAddress',
> >               'vsock': 'VsockSocketAddress',
> >               'fd': 'FdSocketAddress' } }
> > +
> > +##
> > +# @SpdmTransportType:
> > +#
> > +# SPDM transport type
> > +#
> > +# @unspecified: Unspecified transport
> > +#
> > +# @mctp: MCTP transport
> > +#
> > +# @doe: PCI Data Object Exchange (DOE) transport
> > +#
> > +# @scsi: SCSI transport
> > +#
> > +# @nvme: NVMe transport
> > +#
> > +# Since: 11.1
> > +##
> > +{ 'enum': 'SpdmTransportType',
> > +  'data': [ 'unspecified', 'mctp', 'doe', 'scsi', 'nvme' ] }
> 
> The change is correct, so:
> 
> Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
> 
> But I believe SPDM_SOCKET_TRANSPORT_TYPE_UNSPEC was a mistake
> and shouldn't be included. Not sure if worth a distinct patch
> or mention here.
> 
> (Cc'ing SPDM maintainers to get confirmation)

It's included to match the enums from the original spdm-emu:

https://github.com/DMTF/spdm-emu/blob/64cf999e8037bd3e0fa95f69332ff4c1d277f705/spdm_emu/spdm_emu_common/command.h#L29

So I think it should stay

Alistair

Reply via email to