On Thu, 2019-01-24 at 17:48 +0100, Andrea Bolognani wrote:
On Wed, 2019-01-23 at 16:32 -0500, Cole Robinson wrote: [...]
+/** + * qemuBuildVirtioDevStr + * @buf: virBufferPtr to append the built string + * @baseName: qemu virtio device basename string. Ex: virtio-rng for <rng> + * @devtype: virDomainDeviceType of the device. Ex: VIR_DOMAIN_DEVICE_TYPE_RNG + * @devdata: *DefPtr of the device definition + * + * Build the qemu virtio -device name from the passed parameters. Currently + * this is mostly about attaching the correct string prefix to @baseName for + * the passed @type. So for @baseName "virtio-rng" and devdata->info.type + * VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI, generate "virtio-rng-pci" + * + * Returns: -1 on failure, 0 on success + */ static int qemuBuildVirtioDevStr(virBufferPtr buf, const char *baseName, - virDomainDeviceAddressType type) + virDomainDeviceType devtype, + void *devdata) { const char *implName = NULL; + virDomainDeviceDef device = { .type = devtype }; + virDomainDeviceInfoPtr info;
- switch (type) { + virDomainDeviceSetData(&device, devdata); + info = virDomainDeviceGetInfo(&device); + + switch (info->type) {
You should cast info->type to virDomainDeviceAddressType here to force the compiler to check the switch statements covers all possible values.
With that fixed,
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Small additional note: I won't be able to review more patches until next week, but since everything up until here makes sense even in isolation and there were only minor / cosmetic issues, feel free to take care of those and then push away :) -- Andrea Bolognani / Red Hat / Virtualization