On 04/06/2019 13:55, Julien Grall wrote:
>> @@ -476,8 +474,6 @@ struct arch_domain
>>   #define has_pirq(d)        (!!((d)->arch.emulation_flags &
>> X86_EMU_USE_PIRQ))
>>   #define has_vpci(d)        (!!((d)->arch.emulation_flags &
>> X86_EMU_VPCI))
>>   -#define has_arch_pdevs(d)    (!list_empty(&(d)->arch.pdev_list))
>> -
>>   #define gdt_ldt_pt_idx(v) \
>>         ((v)->vcpu_id >> (PAGETABLE_ORDER - GDT_LDT_VCPU_SHIFT))
>>   #define pv_gdt_ptes(v) \
>> --- a/xen/include/xen/pci.h
>> +++ b/xen/include/xen/pci.h
>> @@ -121,7 +121,9 @@ struct pci_dev {
>>   };
>>     #define for_each_pdev(domain, pdev) \
>> -    list_for_each_entry(pdev, &(domain->arch.pdev_list), domain_list)
>> +    list_for_each_entry(pdev, &(domain)->pdev_list, domain_list)
>> +
>> +#define has_arch_pdevs(d) (!list_empty(&(d)->pdev_list))
>
> This feels a bit strange to keep "arch" in the macro name when the
> code is now generic. How about "domain_has_pdevs"?

I agree that keeping arch in the name is a little weird, and this is
definitely the patch to rename it in - there are only 9 instances in the
entire codebase.

For the patch overall, Acked-by: Andrew Cooper
<andrew.cooper3@citrix.com>, however this helper ends up being named.

~Andrew