On Thu, Apr 11, 2019 at 06:48:26PM +0200, Paolo Bonzini wrote:
> Starting state migration after an IO exit without first completing IO
> may result in test failures. We already have two tests that need this
> (this patch in fact fixes evmcs_test, similar to what was fixed for
> state_test in commit 0f73bbc851ed, "KVM: selftests: complete IO before
> migrating guest state", 2019-03-13) and a third is coming. So, move the
> code to vcpu_save_state, and while at it do not access register state
> until after I/O is complete.
> ---
> tools/testing/selftests/kvm/lib/kvm_util.c | 5 +++++
> tools/testing/selftests/kvm/lib/x86_64/processor.c | 8 ++++++++
> tools/testing/selftests/kvm/x86_64/evmcs_test.c | 5 +++--
> tools/testing/selftests/kvm/x86_64/state_test.c | 15 +--------------
> 4 files changed, 17 insertions(+), 16 deletions(-)
>
> diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c
> index efa0aad8b3c6..4ca96b228e46 100644
> --- a/tools/testing/selftests/kvm/lib/kvm_util.c
> +++ b/tools/testing/selftests/kvm/lib/kvm_util.c
> @@ -91,6 +91,11 @@ static void vm_open(struct kvm_vm *vm, int perm, unsigned long type)
> if (vm->kvm_fd < 0)
> exit(KSFT_SKIP);
>
> + if (!kvm_check_cap(KVM_CAP_IMMEDIATE_EXIT)) {
> + fprintf(stderr, "immediate_exit not available, skipping test\n");
> + exit(KSFT_SKIP);
> + }
> +
> vm->fd = ioctl(vm->kvm_fd, KVM_CREATE_VM, type);
> TEST_ASSERT(vm->fd >= 0, "KVM_CREATE_VM ioctl failed, "
> "rc: %i errno: %i", vm->fd, errno);
> diff --git a/tools/testing/selftests/kvm/lib/x86_64/processor.c b/tools/testing/selftests/kvm/lib/x86_64/processor.c
> index f28127f4a3af..b363c9611bd6 100644
> --- a/tools/testing/selftests/kvm/lib/x86_64/processor.c
> +++ b/tools/testing/selftests/kvm/lib/x86_64/processor.c
> @@ -1030,6 +1030,14 @@ struct kvm_x86_state *vcpu_save_state(struct kvm_vm *vm, uint32_t vcpuid)
> nested_size, sizeof(state->nested_));
> }
>
> + /*
> + * When KVM exits to userspace with KVM_EXIT_IO, KVM guarantees
> + * guest state is consistent only after userspace re-enters the
> + * kernel with KVM_RUN. Complete IO prior to migrating state
> + * to a new VM.
> + */
> + vcpu_run_complete_io(vm, vcpuid);
> +
Since the need for IO completion also affects MMIO exits and there's no
reason to believe that vcpu_save_state() will always be called after
an IO exit, then shouldn't we instead put this in get_ucall()?
diff --git a/tools/testing/selftests/kvm/lib/ucall.c b/tools/testing/selftests/kvm/lib/ucall.c
index a2ab38be2f47..e8c6f2741ce7 100644
--- a/tools/testing/selftests/kvm/lib/ucall.c
+++ b/tools/testing/selftests/kvm/lib/ucall.c
@@ -132,6 +132,7 @@ uint64_t get_ucall(struct kvm_vm *vm, uint32_t vcpu_id, struct ucall *uc)
if (ucall_type == UCALL_PIO && run->exit_reason == KVM_EXIT_IO &&
run->io.port == UCALL_PIO_PORT) {
struct kvm_regs regs;
+ vcpu_run_complete_io(vm, vcpu_id);
vcpu_regs_get(vm, vcpu_id, ®s);
memcpy(uc, addr_gva2hva(vm, (vm_vaddr_t)regs.rdi), sizeof(*uc));
return uc->cmd;
@@ -144,6 +145,7 @@ uint64_t get_ucall(struct kvm_vm *vm, uint32_t vcpu_id, struct ucall *uc)
"Unexpected ucall exit mmio address access");
gva = *(vm_vaddr_t *)run->mmio.data;
memcpy(uc, addr_gva2hva(vm, gva), sizeof(*uc));
+ vcpu_run_complete_io(vm, vcpu_id);
}
return uc->cmd;
> nmsrs = kvm_get_num_msrs(vm);
> list = malloc(sizeof(*list) + nmsrs * sizeof(list->indices[0]));
> list->nmsrs = nmsrs;
> diff --git a/tools/testing/selftests/kvm/x86_64/evmcs_test.c b/tools/testing/selftests/kvm/x86_64/evmcs_test.c
> index c49c2a28b0eb..36669684eca5 100644
> --- a/tools/testing/selftests/kvm/x86_64/evmcs_test.c
> +++ b/tools/testing/selftests/kvm/x86_64/evmcs_test.c
> @@ -123,8 +123,6 @@ int main(int argc, char *argv[])
> stage, run->exit_reason,
> exit_reason_str(run->exit_reason));
>
> - memset(®s1, 0, sizeof(regs1));
> - vcpu_regs_get(vm, VCPU_ID, ®s1);
> switch (get_ucall(vm, VCPU_ID, &uc)) {
> case UCALL_ABORT:
> TEST_ASSERT(false, "%s at %s:%d", (const char *)uc.args[0],
> @@ -144,6 +142,9 @@ int main(int argc, char *argv[])
> stage, (ulong)uc.args[1]);
>
> state = vcpu_save_state(vm, VCPU_ID);
> + memset(®s1, 0, sizeof(regs1));
> + vcpu_regs_get(vm, VCPU_ID, ®s1);
> +
We would still need this change to get the vcpu_regs_get() below the
get_ucall().
> kvm_vm_release(vm);
>
> /* Restore state in a new VM. */
> diff --git a/tools/testing/selftests/kvm/x86_64/state_test.c b/tools/testing/selftests/kvm/x86_64/state_test.c
> index 30f75856cf39..e0a3c0204b7c 100644
> --- a/tools/testing/selftests/kvm/x86_64/state_test.c
> +++ b/tools/testing/selftests/kvm/x86_64/state_test.c
> @@ -134,11 +134,6 @@ int main(int argc, char *argv[])
>
> struct kvm_cpuid_entry2 *entry = kvm_get_supported_cpuid_entry(1);
>
> - if (!kvm_check_cap(KVM_CAP_IMMEDIATE_EXIT)) {
> - fprintf(stderr, "immediate_exit not available, skipping test\n");
> - exit(KSFT_SKIP);
> - }
> -
> /* Create VM */
> vm = vm_create_default(VCPU_ID, 0, guest_code);
> vcpu_set_cpuid(vm, VCPU_ID, kvm_get_supported_cpuid());
> @@ -179,18 +174,10 @@ int main(int argc, char *argv[])
> uc.args[1] == stage, "Unexpected register values vmexit #%lx, got %lx",
> stage, (ulong)uc.args[1]);
>
> - /*
> - * When KVM exits to userspace with KVM_EXIT_IO, KVM guarantees
> - * guest state is consistent only after userspace re-enters the
> - * kernel with KVM_RUN. Complete IO prior to migrating state
> - * to a new VM.
> - */
> - vcpu_run_complete_io(vm, VCPU_ID);
> -
> + state = vcpu_save_state(vm, VCPU_ID);
> memset(®s1, 0, sizeof(regs1));
> vcpu_regs_get(vm, VCPU_ID, ®s1);
>
> - state = vcpu_save_state(vm, VCPU_ID);
> kvm_vm_release(vm);
>
> /* Restore state in a new VM. */
> --
> 1.8.3.1
>
>
Thanks,
drew
