[libvirt] [PATCH 08/15] conf: utility function to convert PCI controller model into connect type

Laine Stump laine at laine.org
Tue Apr 12 14:50:20 UTC 2016


On 04/05/2016 08:33 AM, Ján Tomko wrote:
> On Thu, Mar 24, 2016 at 03:25:43PM -0400, Laine Stump wrote:
>> There are two places in qemu_domain_address.c where we have a switch
>> statement to convert PCI controller models
>> (VIR_DOMAIN_CONTROLLER_MODEL_PCI*) into the connection type flag that
>> is matched when looking for an upstream connection for that model of
>> controller (VIR_PCI_CONNECT_TYPE_*). This patch makes a utility
>> function in conf/domain_addr.c to do that, so that when a new PCI
>> controller is added, we only need to add the new model-->connect-type
>> in a single place.
>> ---
>>   src/conf/domain_addr.c         | 47 +++++++++++++++++++++++++
>>   src/conf/domain_addr.h         |  4 +++
>>   src/libvirt_private.syms       |  1 +
>>   src/qemu/qemu_domain_address.c | 80 +++++-------------------------------------
>>   4 files changed, 61 insertions(+), 71 deletions(-)
>>
>> diff --git a/src/conf/domain_addr.c b/src/conf/domain_addr.c
>> index 4408c4a..1bf2c9a 100644
>> --- a/src/conf/domain_addr.c
>> +++ b/src/conf/domain_addr.c
>> @@ -32,6 +32,53 @@
>>   
>>   VIR_LOG_INIT("conf.domain_addr");
>>   
>> +int
>> +virDomainPCIControllerModelToConnectType(virDomainControllerModelPCI model,
>> +                                         virDomainPCIConnectFlags *connectType)
>> +{
>> +    /* given a VIR_DOMAIN_CONTROLLER_MODEL_PCI*, set connectType to
>> +     * the equivalent VIR_PCI_CONNECT_TYPE_*. return 0 on success, -1
>> +     * if the model wasn't recognized.
>> +     */
>> +    switch (model) {
>> +    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
>> +    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
>> +        /* pci-root and pcie-root are implicit in the machine,
>> +         * and have no upstream connection
>> +         */
>> +        *connectType = 0;
>> +        break;
>> +    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE:
>> +        /* pci-bridge is treated like a standard PCI endpoint device, */
>> +        *connectType = VIR_PCI_CONNECT_TYPE_PCI_DEVICE;
>> +        break;
>> +    case VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE:
>> +        /* dmi-to-pci-bridge is treated like a PCIe device
>> +         * (e.g. it can be plugged directly into pcie-root)
>> +         */
>> +        *connectType = VIR_PCI_CONNECT_TYPE_PCIE_DEVICE;
>> +        break;
>> +    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT_PORT:
>> +        *connectType = VIR_PCI_CONNECT_TYPE_PCIE_ROOT_PORT;
>> +        break;
>> +    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_UPSTREAM_PORT:
>> +        *connectType = VIR_PCI_CONNECT_TYPE_PCIE_SWITCH_UPSTREAM_PORT;
>> +        break;
>> +    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_DOWNSTREAM_PORT:
>> +        *connectType = VIR_PCI_CONNECT_TYPE_PCIE_SWITCH_DOWNSTREAM_PORT;
>> +        break;
>> +    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
>> +        /* if this happens, there is an error in the code. A
>> +         * PCI controller should always have a proper model
>> +         * set
>> +         */
>> +        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
>> +                       _("PCI controller model incorrectly set to 'last'"));
>> +        return -1;
> This is dead code.
> The function can just return virDomainPCIConnectFlags directly.

I was originally going to do that, but since the switch is over an enum, 
all possible values must have cases. I suppose I could just hand-wave 
over that case saying "This will never happen" and have it return 0. 
Actually, I think I *will* do that; it would be reasonable to think that 
someone would screw up by failing to set the model (in which case it 
would be 0, i.e. "...PCI_ROOT"), but it would take some serious ill 
intent to set it to "...PCI_LAST"




More information about the libvir-list mailing list