Re: [PATCH v2 01/10] KVM: selftests: Use gva_t instead of vm_vaddr_t

[Sean Christopherson <seanjc@google.com>]

On Fri, Feb 20, 2026, David Matlack wrote:
> Replace all occurrences of vm_vaddr_t with gva_t to align with KVM code
> and with the conversion helpers (e.g. addr_gva2hva()). Also replace
> vm_vaddr in function names with gva to align with the new type name.

...

> @@ -716,22 +716,22 @@ void vm_mem_region_move(struct kvm_vm *vm, uint32_t slot, uint64_t new_gpa);
>  void vm_mem_region_delete(struct kvm_vm *vm, uint32_t slot);
>  struct kvm_vcpu *__vm_vcpu_add(struct kvm_vm *vm, uint32_t vcpu_id);
>  void vm_populate_vaddr_bitmap(struct kvm_vm *vm);
> -vm_vaddr_t vm_vaddr_unused_gap(struct kvm_vm *vm, size_t sz, vm_vaddr_t vaddr_min);
> -vm_vaddr_t vm_vaddr_alloc(struct kvm_vm *vm, size_t sz, vm_vaddr_t vaddr_min);
> -vm_vaddr_t __vm_vaddr_alloc(struct kvm_vm *vm, size_t sz, vm_vaddr_t vaddr_min,
> -			    enum kvm_mem_region_type type);
> -vm_vaddr_t vm_vaddr_alloc_shared(struct kvm_vm *vm, size_t sz,
> -				 vm_vaddr_t vaddr_min,
> -				 enum kvm_mem_region_type type);
> -vm_vaddr_t vm_vaddr_alloc_pages(struct kvm_vm *vm, int nr_pages);
> -vm_vaddr_t __vm_vaddr_alloc_page(struct kvm_vm *vm,
> -				 enum kvm_mem_region_type type);
> -vm_vaddr_t vm_vaddr_alloc_page(struct kvm_vm *vm);
> +gva_t gva_unused_gap(struct kvm_vm *vm, size_t sz, gva_t vaddr_min);
> +gva_t gva_alloc(struct kvm_vm *vm, size_t sz, gva_t vaddr_min);
> +gva_t __gva_alloc(struct kvm_vm *vm, size_t sz, gva_t vaddr_min,
> +		  enum kvm_mem_region_type type);
> +gva_t gva_alloc_shared(struct kvm_vm *vm, size_t sz,
> +		       gva_t vaddr_min,
> +		       enum kvm_mem_region_type type);
> +gva_t gva_alloc_pages(struct kvm_vm *vm, int nr_pages);
> +gva_t __gva_alloc_page(struct kvm_vm *vm,
> +		       enum kvm_mem_region_type type);
> +gva_t gva_alloc_page(struct kvm_vm *vm);

The existing vm_vaddr_alloc() and friends are pretty bad names.  gva_alloc() is
far, far worse.  The APIs aren't just allocation a guest virtual address, they're
allocating guest physical memory, finding a usable virtual address, and creating
mappings.

I don't see any reason to have vaddr or gva in the name.  E.g. malloc() isn't
virt_malloc().  But I do think they need to be explicitly scoped to KVM, and to
a VM.  I'll drop API renames from this patch, and rename them to vm_alloc() and
friends in a separate patch.  Amusingly, that naming scheme will still work if
"vm" is misconstrued as "virtual memory" instead of "virtual machine".

P.S. This is a great example of why I insist on one logical change per patch, with
judicious exemptions for opportunistic cleanups/changes.  If this has been a
separate patch, it would have taken me all of two seconds to unwind.  As it was,
I spent a good 10-15 minutes dealing with this.  In large part because I kept
making goofs, but that's the whole point: there was no reason to put me in a
position to make goofs.

The other argument against these sorts of "Also do xyz" add-ons is that of a
slippery slope.  Why rename these APIs in this patch, but not the myriad vaddr
variables?  Then after a few "I'll just clean this up too" changes, there's an
entire series in what is purportedly just a typedef rename.

-- 
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv