On Mon, Feb 09, 2015 at 03:01:15PM +0800, Chen, Tiejun wrote: > On 2015/2/8 18:48, Michael S. Tsirkin wrote: > >On Fri, Feb 06, 2015 at 01:14:46PM +0100, Cornelia Huck wrote: > >>On Fri, 6 Feb 2015 13:41:26 +0800 > >>Tiejun Chen <tiejun.c...@intel.com> wrote: > >> > >>>Actually we define these device IDs in virtio standard, so > >>>we'd better put them into one common place to manage conveniently. > >>>Here I also add VIRTIO_ID_RESERVE according to virtio spec. > >>> > >>>Signed-off-by: Tiejun Chen <tiejun.c...@intel.com> > > > >We really should just write a script to import the headers > >from the linux kernel. > >They will need some tweaks to avoid dependencies on > >linux/types, but this seems easy to do - better than > >trying to keep things in sync manually. > > I prefer Cornelia's comment since actually we're trying to define a little > bit according to a spec, so the following may be enough? > > diff --git a/include/hw/virtio/virtio.h b/include/hw/virtio/virtio.h > index f24997d..4afb0b7 100644 > --- a/include/hw/virtio/virtio.h > +++ b/include/hw/virtio/virtio.h > @@ -23,6 +23,22 @@ > #include "hw/virtio/virtio-9p.h" > #endif > > +/* Refer to VirtIO Spec 1.0. */ > + > +#define VIRTIO_ID_RESERVED 0 /* reserved (invalid)*/ > +#define VIRTIO_ID_NET 1 /* network card */ > +#define VIRTIO_ID_BLOCK 2 /* block device */ > +#define VIRTIO_ID_CONSOLE 3 /* console */ > +#define VIRTIO_ID_RNG 4 /* entropy source */ > +#define VIRTIO_ID_BALLOON 5 /* memory ballooning */ > +#define VIRTIO_ID_IOMEMORY 6 /* ioMemory */ > +#define VIRTIO_ID_RPMSG 7 /* rpmsg */ > +#define VIRTIO_ID_SCSI 8 /* SCSI host */ > +#define VIRTIO_ID_9P 9 /* 9P transport */ > +#define VIRTIO_ID_MAC80211_WALN 10 /* mac80211 wlan */ > +#define VIRTIO_ID_RPROC_SERIAL 11 /* rproc seria */ > +#define VIRTIO_ID_CAIF 12 /* virtio CAIF */ > + > /* from Linux's linux/virtio_config.h */ > > /* Status byte for guest to report progress, and synchronize features. */ > > Thanks > Tiejun
This still means each change has to be done in two places. An automated script for copying headers would be much better IMHO. -- MST
