On 4/22/2025 10:20 PM, Zhao Liu wrote:
+#ifdef CONFIG_TDX
+bool is_tdx_vm(void);
+#else
+#define is_tdx_vm() 0
+#endif /* CONFIG_TDX */
+
a little nit: could we rename it as "tdx_enabled"?
Then the cases like these would be neater?
When sev support was added, it was seen as a feature for the VMs that are
created on AMD platform. I think that's why it got called sev_enabled().
But for TDX, it is introduced as a different type of VM in contrast to the
legacy/normal VMX VMs. We need to pass specific TDX vm type to
KVM_CREATE_VM. Based on this, is_tdx_vm() was chosen.
But isn't AMD's SEV also defined as the VM type?
The initial SEV support that introduced sev_eanbled(), didn't introduce
the VM type.
The specific type for SEV* was added later.
static const char *vm_type_name[] = {
[KVM_X86_DEFAULT_VM] = "default",
[KVM_X86_SEV_VM] = "SEV",
[KVM_X86_SEV_ES_VM] = "SEV-ES",
[KVM_X86_SNP_VM] = "SEV-SNP",
[KVM_X86_TDX_VM] = "TDX",
};
Functionally, they are part of the coco functionality provided by
different vendors, so it's better thatt both could be in the same place
as much as possible, including file location, naming style. Of course,
it's not a big deal, and it can be cleaned up after merge if needed.
yes, cleanup can be a separate work if Paolo doesn't dislike.
I don't think the different name is a big issue, as nobody mentions it from
the initial RFC to current v8 until you.
The Chinese saying: There are a thousand Hamlets in a thousand people's
eyes :-).
