[libvirt] [PATCH v3 01/12] conf: Add definitions for 'uid' and 'fid' PCI address attributes

Yi Min Zhao zyimin at linux.ibm.com
Mon Aug 20 08:21:40 UTC 2018



在 2018/8/16 下午10:38, Andrea Bolognani 写道:
> On Tue, 2018-08-07 at 17:10 +0800, Yi Min Zhao wrote:
>> +typedef struct _virZPCIDeviceAddress virZPCIDeviceAddress;
>> +typedef virZPCIDeviceAddress *virZPCIDeviceAddressPtr;
>> +struct _virZPCIDeviceAddress {
>> +    unsigned int zpci_fid;
>> +    unsigned int zpci_uid;
>> +    bool fid_assigned;
>> +    bool uid_assigned;
>> +};
> A couple of questions about the approach here, one of which I have
> mentioned already and one of which I probably haven't (my bad):
>
>    * do you really need to have separate booleans tracking whether
>      or not either id has been assigned? Wouldn't the same approach
>      as virPCIDeviceAddressIsEmpty() work, eg. consider the address
>      absent if both are zero and present otherwise?
>
>    * especially if you don't need the additional booleans, would it
>      be preferable to just add the two ids to the existing struct
>      instead of declaring a new one that you'll have to allocate
>      and make sure it's not NULL before accessing it? Again, I seem
>      to remember Laine feeling somewhat strongly about the topic.
>
> Cosmetic issues:
Sorry, forgot this comment.
>
>    * uid should be listed before fid;
Sure.
>
>    * the zpci_ prefix is unnecessary if you have a separate struct
>      that contains "ZPCI" in the name. But see above :)
>
Let's see your response to another mail.




More information about the libvir-list mailing list