Hi Arnd,

> On Tuesday 29 September 2009, Chris Wright wrote:
> > > +struct Vmxnet3_MiscConf {
> > > +       struct Vmxnet3_DriverInfo driverInfo;
> > > +       uint64_t             uptFeatures;
> > > +       uint64_t             ddPA;         /* driver data PA */
> > > +       uint64_t             queueDescPA;  /* queue descriptor
> table PA */
> > > +       uint32_t             ddLen;        /* driver data len */
> > > +       uint32_t             queueDescLen; /* queue desc. table len
> in bytes */
> > > +       uint32_t             mtu;
> > > +       uint16_t             maxNumRxSG;
> > > +       uint8_t              numTxQueues;
> > > +       uint8_t              numRxQueues;
> > > +       uint32_t             reserved[4];
> > > +};
> >
> > should this be packed (or others that are shared w/ device)?  i
> assume
> > you've already done 32 vs 64 here
> 
> I would not mark it packed, because it already is well-defined on all
> systems. You should add __packed only to the fields where you screwed
> up, but not to structures that already work fine.

You're exactly right; I reiterated as much in my response to Chris.

> One thing that should possibly be fixed is the naming of identifiers,
> e.g.
> 's/Vmxnet3_MiscConf/vmxnet3_misc_conf/g', unless these header files are
> shared with the host implementation.

These header files are indeed shared with the host implementation, as you've 
guessed. If it's not a big deal, we would like to keep the names the same, just 
for our own sanity's sake?

Thanks!

- Bhavesh

> 
>       Arnd <><
_______________________________________________
Virtualization mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/virtualization

Reply via email to