[libvirt] [PATCH 11/18] virpcimock: Store PCI address as ints not string

Michal Prívozník mprivozn at redhat.com
Sat Aug 17 11:01:39 UTC 2019


On 8/16/19 10:56 PM, Daniel Henrique Barboza wrote:
> 
> 
> On 8/14/19 8:57 AM, Michal Privoznik wrote:
>> In upcoming patches we will need only some portions of the PCI
>> address. To construct that easily, it's better if the PCI address
>> of a device is stored as four integers rather than one string.
>>
>> Signed-off-by: Michal Privoznik <mprivozn at redhat.com>
>> ---
>>   tests/virpcimock.c | 76 +++++++++++++++++++++++++++++++++++++---------
>>   1 file changed, 61 insertions(+), 15 deletions(-)
>>
>> diff --git a/tests/virpcimock.c b/tests/virpcimock.c
>> index de365cdb78..c10764dcdd 100644
>> --- a/tests/virpcimock.c
>> +++ b/tests/virpcimock.c
>> @@ -118,8 +118,16 @@ struct pciDriver {
>>       unsigned int fail;  /* Bitwise-OR of driverActions that should
>> fail */
>>   };
>>   +struct pciDeviceAddress {
>> +    unsigned int domain;
>> +    unsigned int bus;
>> +    unsigned int device;
>> +    unsigned int function;
>> +};
>> +# define ADDR_STR_FMT "%04x:%02x:%02x.%d"
>> +
> 
> I was going to complain "this stuff is similar to what we already have
> in utils/virpci.c, why can't we use it here", but in a second thought I
> realized virpci.h is too big of a import just for a macro and a couple
> of parse functions.

Yet again, great minds think alike :-) This was exactly my reasoning for
not including virpci.h. I mean, I wanted to but then realized it's
probably better if we keep the mock separate.

> 
> 
> Reviewed-by: Daniel Henrique Barboza <danielhb413 at gmail.com>

Thanks,

Michal




More information about the libvir-list mailing list