On Wed, Mar 27, 2024 at 05:42:54AM +0000, Manali Shukla wrote:
...
> diff --git a/tools/testing/selftests/kvm/include/kvm_util_base.h
> b/tools/testing/selftests/kvm/include/kvm_util_base.h
> index 4a40b332115d..00e37c376cf3 100644
> --- a/tools/testing/selftests/kvm/include/kvm_util_base.h
> +++ b/tools/testing/selftests/kvm/include/kvm_util_base.h
> @@ -879,7 +879,7 @@ static inline vm_paddr_t vm_phy_pages_alloc(struct kvm_vm
> *vm, size_t num,
> */
> struct kvm_vm *____vm_create(struct vm_shape shape);
> struct kvm_vm *__vm_create(struct vm_shape shape, uint32_t nr_runnable_vcpus,
> - uint64_t nr_extra_pages);
> + uint64_t nr_extra_pages, bool is_in_kernel_apic);
Adding boolean flag parameters to functions, which will 99% of the time be
called with the same value set for them, is not nice.
...
> diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c
> b/tools/testing/selftests/kvm/lib/kvm_util.c
> index adc51b0712ca..9c2a9e216d80 100644
> --- a/tools/testing/selftests/kvm/lib/kvm_util.c
> +++ b/tools/testing/selftests/kvm/lib/kvm_util.c
> @@ -354,7 +354,7 @@ static uint64_t vm_nr_pages_required(enum vm_guest_mode
> mode,
> }
>
> struct kvm_vm *__vm_create(struct vm_shape shape, uint32_t nr_runnable_vcpus,
> - uint64_t nr_extra_pages)
> + uint64_t nr_extra_pages, bool is_in_kernel_apic)
> {
> uint64_t nr_pages = vm_nr_pages_required(shape.mode, nr_runnable_vcpus,
> nr_extra_pages);
> @@ -382,7 +382,12 @@ struct kvm_vm *__vm_create(struct vm_shape shape,
> uint32_t nr_runnable_vcpus,
> slot0 = memslot2region(vm, 0);
> ucall_init(vm, slot0->region.guest_phys_addr +
> slot0->region.memory_size);
>
> - kvm_arch_vm_post_create(vm);
> + if (is_in_kernel_apic) {
> + kvm_arch_vm_post_create(vm);
> + } else {
> + sync_global_to_guest(vm, host_cpu_is_intel);
> + sync_global_to_guest(vm, host_cpu_is_amd);
> + }
__vm_create() is shared with other architectures, and duplicating part of
kvm_arch_vm_post_create() here is not a good approach, even if the
framework was only for x86.
I suggest:
1. Extend vm_shape to include a flags field and create a flag called
NO_IRQCHIP
2. Add a flags member to kvm_vm and set it to the value of vm_shape.flags
in ____vm_create()
3. Check !(vm.flags & NO_IRQCHIP) in x86's kvm_arch_vm_post_create()
before calling vm_create_irqchip()
Then, in your tests, you'll create your vm shape with the NO_IRQCHIP flag
set.
Thanks,
drew