linux-kernel - Re: [PATCH] selftests/kvm: add set_boot_cpu

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <c4adbfcd-1c5a-3f7a-7d61-206964256963@redhat.com>
Date:   Wed, 17 Mar 2021 12:01:46 +0100
From:   Paolo Bonzini <pbonzini@...hat.com>
To:     Emanuele Giuseppe Esposito <eesposit@...hat.com>,
        linux-kselftest@...r.kernel.org
Cc:     Shuah Khan <shuah@...nel.org>,
        Vitaly Kuznetsov <vkuznets@...hat.com>,
        linux-kernel@...r.kernel.org, kvm@...r.kernel.org
Subject: Re: [PATCH] selftests/kvm: add set_boot_cpu_id test

On 17/03/21 08:44, Emanuele Giuseppe Esposito wrote:
> +	printf("vcpu executing...\n");
> +	vcpu_run(vm, vcpuid);
> +	printf("vcpu executed\n");
> +
> +	switch (get_ucall(vm, vcpuid, &uc)) {
> +	case UCALL_SYNC:
> +		printf("stage %d sync %ld\n", stage, uc.args[1]);
> +		TEST_ASSERT(!strcmp((const char *)uc.args[0], "hello") &&
> +			    uc.args[1] == stage + 1,
> +			    "Stage %d: Unexpected register values vmexit, got %lx",
> +			    stage + 1, (ulong)uc.args[1]);
> +		return;
> +	case UCALL_DONE:
> +		printf("got done\n");

You can remove the printfs here.

> +		return;
> +	case UCALL_ABORT:
> +		TEST_ASSERT(false, "%s at %s:%ld\n\tvalues: %#lx, %#lx", (const char *)uc.args[0],
> +			    __FILE__, uc.args[1], uc.args[2], uc.args[3]);
> +	default:
> +		TEST_ASSERT(false, "Unexpected exit: %s",
> +			    exit_reason_str(vcpu_state(vm, vcpuid)->exit_reason));
> +	}
> +}
> +
> +static void add_x86_vcpu(struct kvm_vm *vm, uint32_t vcpuid, void *code)
> +{
> +	vm_vcpu_add_default(vm, vcpuid, code);
> +	vcpu_set_cpuid(vm, vcpuid, kvm_get_supported_cpuid());
> +}
> +
> +static void run_vm_bsp(uint32_t bsp_vcpu)
> +{
> +	struct kvm_vm *vm;
> +	int stage;
> +	void *vcpu0_code, *vcpu1_code;
> +
> +	vm = create_vm();
> +
> +	vcpu0_code = guest_bsp_vcpu;
> +	vcpu1_code = guest_not_bsp_vcpu;
> +
> +	if (bsp_vcpu == VCPU_ID1) {
> +		vcpu0_code = guest_not_bsp_vcpu;
> +		vcpu1_code = guest_bsp_vcpu;
> +
> +		vm_ioctl(vm, KVM_SET_BOOT_CPU_ID, (void *) VCPU_ID1);
> +	}
> +
> +	add_x86_vcpu(vm, VCPU_ID0, vcpu0_code);
> +	add_x86_vcpu(vm, VCPU_ID1, vcpu1_code);
> +
> +	for (stage = 0; stage < 2; stage++) {
> +		run_vcpu(vm, VCPU_ID0, stage);
> +		run_vcpu(vm, VCPU_ID1, stage);
> +	}

These are just stylistic nits, but:

1) you could pass a bool to add_x86_vcpu, like

	add_x86_vcpu(vm, VCPU_ID0, bsp_vcpu == VCPU_ID0);
	add_x86_vcpu(vm, VCPU_ID1, bsp_vcpu == VCPU_ID1);

instead of having the vcpu0_code and vcpu1_code pointers.

2) you could move the for loop inside run_vcpu.  This way you can assert 
that you get UCALL_DONE when stage == 2.

> +	kvm_vm_free(vm);
> +}
> +
> +int main(int argc, char *argv[])
> +{
> +	if (!kvm_check_cap(KVM_CAP_SET_BOOT_CPU_ID)) {
> +		print_skip("set_boot_cpu_id not available");
> +		return 0;
> +	}
> +
> +	run_vm_bsp(VCPU_ID0);
> +	run_vm_bsp(VCPU_ID1);
> +	run_vm_bsp(VCPU_ID0);

Looks good!  Just another small change: since this third test is not 
adding much, you could instead invoke KVM_SET_BOOT_CPU_ID(VCPU_ID1) 
*after* the creation of the vCPUs instead of before.  This should fail 
with -EBUSY, and vcpu 0 will remain the BSP.

Paolo