Re: [PATCH 3/8] kvm: Detect guest-memfd flags supported

[Xiaoyao Li <xiaoyao.li@intel.com>]

On 11/14/2025 5:19 AM, Peter Xu wrote:
> On Thu, Nov 13, 2025 at 04:15:50PM -0500, Peter Xu wrote:
>> On Fri, Oct 24, 2025 at 11:52:01AM +0800, Xiaoyao Li wrote:
>>>> @@ -2781,6 +2782,11 @@ static int kvm_init(AccelState *as, MachineState *ms)
>>>>        kvm_supported_memory_attributes = kvm_vm_check_extension(s, KVM_CAP_MEMORY_ATTRIBUTES);
>>>>        kvm_guest_memfd_supported = kvm_vm_check_extension(s, KVM_CAP_GUEST_MEMFD) &&
>>>>            kvm_vm_check_extension(s, KVM_CAP_USER_MEMORY2);
>>>> +    ret = kvm_vm_check_extension(s, KVM_CAP_GUEST_MEMFD_FLAGS);
>>>> +    if (ret > 0)
>>>> +        kvm_guest_memfd_flags_supported = (uint64_t)ret;
>>>> +    else
>>>> +        kvm_guest_memfd_flags_supported = 0;
>>>
>>> Nit:
>>> 1. QEMU's coding style always requires curly braces.
>>> 2. is the (uint64_t) necessary?
>>> 3. can we name it "kvm_supported_guest_memfd_flags" to make it consistent
>>> with "kvm_supported_memory_attributes"?
>>>
>>> so how about
>>>
>>> kvm_supported_guest_memfd_flags = kvm_vm_check_extension(s,
>>> KVM_CAP_GUEST_MEMFD_FLAGS);
>>>      if (kvm_supported_guest_memfd_flags < 0) {
>>>          kvm_supported_guest_memfd_flags = 0;
>>>      }
>>
>> Yep this looks good, I'll use it, thanks.
> 
> About naming: note that we already have different styles (both below
> variables introduced by your previous commits):
> 
> static uint64_t kvm_supported_memory_attributes;
> static bool kvm_guest_memfd_supported;

The kvm_supported_* is for multiple bits, while
kvm_*_supported is for a boolean.

> I personally preferred kvm_guest_memfd* as prefix, so I kept it.

I'm fine with it.