Dan Smith wrote:
HE> struct virt_device {
HE> enum {
HE> - VIRT_DEV_NET = CIM_RASD_TYPE_NET,
HE> - VIRT_DEV_DISK = CIM_RASD_TYPE_DISK,
HE> - VIRT_DEV_MEM = CIM_RASD_TYPE_MEM,
HE> - VIRT_DEV_VCPU = CIM_RASD_TYPE_PROC,
HE> - VIRT_DEV_UNKNOWN = 1000,
HE> + VIRT_DEV_NET = CIM_RES_TYPE_NET,
HE> + VIRT_DEV_DISK = CIM_RES_TYPE_DISK,
HE> + VIRT_DEV_MEM = CIM_RES_TYPE_MEM,
HE> + VIRT_DEV_VCPU = CIM_RES_TYPE_PROC,
HE> + VIRT_DEV_UNKNOWN = CIM_RES_TYPE_UNKNOWN,
HE> VIRT_DEV_EMU,
HE> VIRT_DEV_GRAPHICS,
HE> } type;
Are the VIRT_DEV_foo constants used anywhere else? It looks to me
that they are mostly stripped out, with the exception of maybe some of
the device_parsing stuff.
I don't think it's a good idea to set dev->type=A in device_parsing,
but check for dev->type==B everywhere else, even though we know A==B.
Do you mind if we try to make this a clean change instead of an
incremental one?
I think breaking out the first few patches into a separate set to do
the switchover is appropriate.
It'll end up being one large set though, or several smaller sets all
dependent on one another. The "#2 - Add configurable enum_devices() to
Device provider and Adopt interface changes to SD & EAFP" also removes
some of the VIRT_DEV_* types.
--
Kaitlin Rupert
IBM Linux Technology Center
kaitlin(a)linux.vnet.ibm.com