Merge remote-tracking branch 'kvm/master' into HEAD

Pick commit fdba608f15e2 ("KVM: VMX: Wake vCPU when delivering posted
IRQ even if vCPU == this vCPU").  In addition to fixing a bug, it
also aligns the non-nested and nested usage of triggering posted
interrupts, allowing for additional cleanups.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

9 files changed, 299 insertions, 19 deletions

diff --git a/tools/testing/selftests/kvm/.gitignore b/tools/testing/selftests/kvm/.gitignore
index 3763105029fb..3cb5ac5da087 100644
--- a/tools/testing/selftests/kvm/.gitignore
+++ b/tools/testing/selftests/kvm/.gitignore
@@ -30,10 +30,12 @@
 /x86_64/svm_int_ctl_test
 /x86_64/sync_regs_test
 /x86_64/tsc_msrs_test
+/x86_64/userspace_io_test
 /x86_64/userspace_msr_exit_test
 /x86_64/vmx_apic_access_test
 /x86_64/vmx_close_while_nested_test
 /x86_64/vmx_dirty_log_test
+/x86_64/vmx_invalid_nested_guest_state
 /x86_64/vmx_preemption_timer_test
 /x86_64/vmx_set_nested_state_test
 /x86_64/vmx_tsc_adjust_test
diff --git a/tools/testing/selftests/kvm/Makefile b/tools/testing/selftests/kvm/Makefile
index c4e34717826a..17342b575e85 100644
--- a/tools/testing/selftests/kvm/Makefile
+++ b/tools/testing/selftests/kvm/Makefile
@@ -59,10 +59,12 @@ TEST_GEN_PROGS_x86_64 += x86_64/vmx_preemption_timer_test
 TEST_GEN_PROGS_x86_64 += x86_64/svm_vmcall_test
 TEST_GEN_PROGS_x86_64 += x86_64/svm_int_ctl_test
 TEST_GEN_PROGS_x86_64 += x86_64/sync_regs_test
+TEST_GEN_PROGS_x86_64 += x86_64/userspace_io_test
 TEST_GEN_PROGS_x86_64 += x86_64/userspace_msr_exit_test
 TEST_GEN_PROGS_x86_64 += x86_64/vmx_apic_access_test
 TEST_GEN_PROGS_x86_64 += x86_64/vmx_close_while_nested_test
 TEST_GEN_PROGS_x86_64 += x86_64/vmx_dirty_log_test
+TEST_GEN_PROGS_x86_64 += x86_64/vmx_invalid_nested_guest_state
 TEST_GEN_PROGS_x86_64 += x86_64/vmx_set_nested_state_test
 TEST_GEN_PROGS_x86_64 += x86_64/vmx_tsc_adjust_test
 TEST_GEN_PROGS_x86_64 += x86_64/vmx_nested_tsc_scaling_test
diff --git a/tools/testing/selftests/kvm/include/kvm_util.h b/tools/testing/selftests/kvm/include/kvm_util.h
index 6a1a37f30494..2d62edc49d67 100644
--- a/tools/testing/selftests/kvm/include/kvm_util.h
+++ b/tools/testing/selftests/kvm/include/kvm_util.h
@@ -321,6 +321,7 @@ bool vm_is_unrestricted_guest(struct kvm_vm *vm);
 
 unsigned int vm_get_page_size(struct kvm_vm *vm);
 unsigned int vm_get_page_shift(struct kvm_vm *vm);
+unsigned long vm_compute_max_gfn(struct kvm_vm *vm);
 uint64_t vm_get_max_gfn(struct kvm_vm *vm);
 int vm_get_fd(struct kvm_vm *vm);
 
diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c
index 8f2e0bb1ef96..53d2b5d04b82 100644
--- a/tools/testing/selftests/kvm/lib/kvm_util.c
+++ b/tools/testing/selftests/kvm/lib/kvm_util.c
@@ -302,7 +302,7 @@ struct kvm_vm *vm_create(enum vm_guest_mode mode, uint64_t phy_pages, int perm)
 		(1ULL << (vm->va_bits - 1)) >> vm->page_shift);
 
 	/* Limit physical addresses to PA-bits. */
-	vm->max_gfn = ((1ULL << vm->pa_bits) >> vm->page_shift) - 1;
+	vm->max_gfn = vm_compute_max_gfn(vm);
 
 	/* Allocate and setup memory for guest. */
 	vm->vpages_mapped = sparsebit_alloc();
@@ -2328,6 +2328,11 @@ unsigned int vm_get_page_shift(struct kvm_vm *vm)
 	return vm->page_shift;
 }
 
+unsigned long __attribute__((weak)) vm_compute_max_gfn(struct kvm_vm *vm)
+{
+	return ((1ULL << vm->pa_bits) >> vm->page_shift) - 1;
+}
+
 uint64_t vm_get_max_gfn(struct kvm_vm *vm)
 {
 	return vm->max_gfn;
diff --git a/tools/testing/selftests/kvm/lib/x86_64/processor.c b/tools/testing/selftests/kvm/lib/x86_64/processor.c
index 82c39db91369..eef7b34756d5 100644
--- a/tools/testing/selftests/kvm/lib/x86_64/processor.c
+++ b/tools/testing/selftests/kvm/lib/x86_64/processor.c
@@ -1431,3 +1431,71 @@ struct kvm_cpuid2 *vcpu_get_supported_hv_cpuid(struct kvm_vm *vm, uint32_t vcpui
 
 	return cpuid;
 }
+
+#define X86EMUL_CPUID_VENDOR_AuthenticAMD_ebx 0x68747541
+#define X86EMUL_CPUID_VENDOR_AuthenticAMD_ecx 0x444d4163
+#define X86EMUL_CPUID_VENDOR_AuthenticAMD_edx 0x69746e65
+
+static inline unsigned x86_family(unsigned int eax)
+{
+        unsigned int x86;
+
+        x86 = (eax >> 8) & 0xf;
+
+        if (x86 == 0xf)
+                x86 += (eax >> 20) & 0xff;
+
+        return x86;
+}
+
+unsigned long vm_compute_max_gfn(struct kvm_vm *vm)
+{
+	const unsigned long num_ht_pages = 12 << (30 - vm->page_shift); /* 12 GiB */
+	unsigned long ht_gfn, max_gfn, max_pfn;
+	uint32_t eax, ebx, ecx, edx, max_ext_leaf;
+
+	max_gfn = (1ULL << (vm->pa_bits - vm->page_shift)) - 1;
+
+	/* Avoid reserved HyperTransport region on AMD processors.  */
+	eax = ecx = 0;
+	cpuid(&eax, &ebx, &ecx, &edx);
+	if (ebx != X86EMUL_CPUID_VENDOR_AuthenticAMD_ebx ||
+	    ecx != X86EMUL_CPUID_VENDOR_AuthenticAMD_ecx ||
+	    edx != X86EMUL_CPUID_VENDOR_AuthenticAMD_edx)
+		return max_gfn;
+
+	/* On parts with <40 physical address bits, the area is fully hidden */
+	if (vm->pa_bits < 40)
+		return max_gfn;
+
+	/* Before family 17h, the HyperTransport area is just below 1T.  */
+	ht_gfn = (1 << 28) - num_ht_pages;
+	eax = 1;
+	cpuid(&eax, &ebx, &ecx, &edx);
+	if (x86_family(eax) < 0x17)
+		goto done;
+
+	/*
+	 * Otherwise it's at the top of the physical address space, possibly
+	 * reduced due to SME by bits 11:6 of CPUID[0x8000001f].EBX.  Use
+	 * the old conservative value if MAXPHYADDR is not enumerated.
+	 */
+	eax = 0x80000000;
+	cpuid(&eax, &ebx, &ecx, &edx);
+	max_ext_leaf = eax;
+	if (max_ext_leaf < 0x80000008)
+		goto done;
+
+	eax = 0x80000008;
+	cpuid(&eax, &ebx, &ecx, &edx);
+	max_pfn = (1ULL << ((eax & 0xff) - vm->page_shift)) - 1;
+	if (max_ext_leaf >= 0x8000001f) {
+		eax = 0x8000001f;
+		cpuid(&eax, &ebx, &ecx, &edx);
+		max_pfn >>= (ebx >> 6) & 0x3f;
+	}
+
+	ht_gfn = max_pfn - num_ht_pages;
+done:
+	return min(max_gfn, ht_gfn - 1);
+}
diff --git a/tools/testing/selftests/kvm/x86_64/svm_int_ctl_test.c b/tools/testing/selftests/kvm/x86_64/svm_int_ctl_test.c
index df04f56ce859..30a81038df46 100644
--- a/tools/testing/selftests/kvm/x86_64/svm_int_ctl_test.c
+++ b/tools/testing/selftests/kvm/x86_64/svm_int_ctl_test.c
@@ -75,7 +75,7 @@ static void l1_guest_code(struct svm_test_data *svm)
 	vmcb->control.int_ctl &= ~V_INTR_MASKING_MASK;
 
 	/* No intercepts for real and virtual interrupts */
-	vmcb->control.intercept &= ~(1ULL << INTERCEPT_INTR | INTERCEPT_VINTR);
+	vmcb->control.intercept &= ~(BIT(INTERCEPT_INTR) | BIT(INTERCEPT_VINTR));
 
 	/* Make a virtual interrupt VINTR_IRQ_NUMBER pending */
 	vmcb->control.int_ctl |= V_IRQ_MASK | (0x1 << V_INTR_PRIO_SHIFT);
diff --git a/tools/testing/selftests/kvm/x86_64/userspace_io_test.c b/tools/testing/selftests/kvm/x86_64/userspace_io_test.c
new file mode 100644
index 000000000000..e4bef2e05686
--- /dev/null
+++ b/tools/testing/selftests/kvm/x86_64/userspace_io_test.c
@@ -0,0 +1,114 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <fcntl.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/ioctl.h>
+
+#include "test_util.h"
+
+#include "kvm_util.h"
+#include "processor.h"
+
+#define VCPU_ID			1
+
+static void guest_ins_port80(uint8_t *buffer, unsigned int count)
+{
+	unsigned long end;
+
+	if (count == 2)
+		end = (unsigned long)buffer + 1;
+	else
+		end = (unsigned long)buffer + 8192;
+
+	asm volatile("cld; rep; insb" : "+D"(buffer), "+c"(count) : "d"(0x80) : "memory");
+	GUEST_ASSERT_1(count == 0, count);
+	GUEST_ASSERT_2((unsigned long)buffer == end, buffer, end);
+}
+
+static void guest_code(void)
+{
+	uint8_t buffer[8192];
+	int i;
+
+	/*
+	 * Special case tests.  main() will adjust RCX 2 => 1 and 3 => 8192 to
+	 * test that KVM doesn't explode when userspace modifies the "count" on
+	 * a userspace I/O exit.  KVM isn't required to play nice with the I/O
+	 * itself as KVM doesn't support manipulating the count, it just needs
+	 * to not explode or overflow a buffer.
+	 */
+	guest_ins_port80(buffer, 2);
+	guest_ins_port80(buffer, 3);
+
+	/* Verify KVM fills the buffer correctly when not stuffing RCX. */
+	memset(buffer, 0, sizeof(buffer));
+	guest_ins_port80(buffer, 8192);
+	for (i = 0; i < 8192; i++)
+		GUEST_ASSERT_2(buffer[i] == 0xaa, i, buffer[i]);
+
+	GUEST_DONE();
+}
+
+int main(int argc, char *argv[])
+{
+	struct kvm_regs regs;
+	struct kvm_run *run;
+	struct kvm_vm *vm;
+	struct ucall uc;
+	int rc;
+
+	/* Tell stdout not to buffer its content */
+	setbuf(stdout, NULL);
+
+	/* Create VM */
+	vm = vm_create_default(VCPU_ID, 0, guest_code);
+	run = vcpu_state(vm, VCPU_ID);
+
+	memset(&regs, 0, sizeof(regs));
+
+	while (1) {
+		rc = _vcpu_run(vm, VCPU_ID);
+
+		TEST_ASSERT(rc == 0, "vcpu_run failed: %d\n", rc);
+		TEST_ASSERT(run->exit_reason == KVM_EXIT_IO,
+			    "Unexpected exit reason: %u (%s),\n",
+			    run->exit_reason,
+			    exit_reason_str(run->exit_reason));
+
+		if (get_ucall(vm, VCPU_ID, &uc))
+			break;
+
+		TEST_ASSERT(run->io.port == 0x80,
+			    "Expected I/O at port 0x80, got port 0x%x\n", run->io.port);
+
+		/*
+		 * Modify the rep string count in RCX: 2 => 1 and 3 => 8192.
+		 * Note, this abuses KVM's batching of rep string I/O to avoid
+		 * getting stuck in an infinite loop.  That behavior isn't in
+		 * scope from a testing perspective as it's not ABI in any way,
+		 * i.e. it really is abusing internal KVM knowledge.
+		 */
+		vcpu_regs_get(vm, VCPU_ID, &regs);
+		if (regs.rcx == 2)
+			regs.rcx = 1;
+		if (regs.rcx == 3)
+			regs.rcx = 8192;
+		memset((void *)run + run->io.data_offset, 0xaa, 4096);
+		vcpu_regs_set(vm, VCPU_ID, &regs);
+	}
+
+	switch (uc.cmd) {
+	case UCALL_DONE:
+		break;
+	case UCALL_ABORT:
+		TEST_FAIL("%s at %s:%ld : argN+1 = 0x%lx, argN+2 = 0x%lx",
+			  (const char *)uc.args[0], __FILE__, uc.args[1],
+			  uc.args[2], uc.args[3]);
+	default:
+		TEST_FAIL("Unknown ucall %lu", uc.cmd);
+	}
+
+	kvm_vm_free(vm);
+	return 0;
+}
diff --git a/tools/testing/selftests/kvm/x86_64/vmx_invalid_nested_guest_state.c b/tools/testing/selftests/kvm/x86_64/vmx_invalid_nested_guest_state.c
new file mode 100644
index 000000000000..489fbed4ca6f
--- /dev/null
+++ b/tools/testing/selftests/kvm/x86_64/vmx_invalid_nested_guest_state.c
@@ -0,0 +1,105 @@
+// SPDX-License-Identifier: GPL-2.0-only
+#include "test_util.h"
+#include "kvm_util.h"
+#include "processor.h"
+#include "vmx.h"
+
+#include <string.h>
+#include <sys/ioctl.h>
+
+#include "kselftest.h"
+
+#define VCPU_ID	0
+#define ARBITRARY_IO_PORT 0x2000
+
+static struct kvm_vm *vm;
+
+static void l2_guest_code(void)
+{
+	/*
+	 * Generate an exit to L0 userspace, i.e. main(), via I/O to an
+	 * arbitrary port.
+	 */
+	asm volatile("inb %%dx, %%al"
+		     : : [port] "d" (ARBITRARY_IO_PORT) : "rax");
+}
+
+static void l1_guest_code(struct vmx_pages *vmx_pages)
+{
+#define L2_GUEST_STACK_SIZE 64
+	unsigned long l2_guest_stack[L2_GUEST_STACK_SIZE];
+
+	GUEST_ASSERT(prepare_for_vmx_operation(vmx_pages));
+	GUEST_ASSERT(load_vmcs(vmx_pages));
+
+	/* Prepare the VMCS for L2 execution. */
+	prepare_vmcs(vmx_pages, l2_guest_code,
+		     &l2_guest_stack[L2_GUEST_STACK_SIZE]);
+
+	/*
+	 * L2 must be run without unrestricted guest, verify that the selftests
+	 * library hasn't enabled it.  Because KVM selftests jump directly to
+	 * 64-bit mode, unrestricted guest support isn't required.
+	 */
+	GUEST_ASSERT(!(vmreadz(CPU_BASED_VM_EXEC_CONTROL) & CPU_BASED_ACTIVATE_SECONDARY_CONTROLS) ||
+		     !(vmreadz(SECONDARY_VM_EXEC_CONTROL) & SECONDARY_EXEC_UNRESTRICTED_GUEST));
+
+	GUEST_ASSERT(!vmlaunch());
+
+	/* L2 should triple fault after main() stuffs invalid guest state. */
+	GUEST_ASSERT(vmreadz(VM_EXIT_REASON) == EXIT_REASON_TRIPLE_FAULT);
+	GUEST_DONE();
+}
+
+int main(int argc, char *argv[])
+{
+	vm_vaddr_t vmx_pages_gva;
+	struct kvm_sregs sregs;
+	struct kvm_run *run;
+	struct ucall uc;
+
+	nested_vmx_check_supported();
+
+	vm = vm_create_default(VCPU_ID, 0, (void *) l1_guest_code);
+
+	/* Allocate VMX pages and shared descriptors (vmx_pages). */
+	vcpu_alloc_vmx(vm, &vmx_pages_gva);
+	vcpu_args_set(vm, VCPU_ID, 1, vmx_pages_gva);
+
+	vcpu_run(vm, VCPU_ID);
+
+	run = vcpu_state(vm, VCPU_ID);
+
+	/*
+	 * The first exit to L0 userspace should be an I/O access from L2.
+	 * Running L1 should launch L2 without triggering an exit to userspace.
+	 */
+	TEST_ASSERT(run->exit_reason == KVM_EXIT_IO,
+		    "Expected KVM_EXIT_IO, got: %u (%s)\n",
+		    run->exit_reason, exit_reason_str(run->exit_reason));
+
+	TEST_ASSERT(run->io.port == ARBITRARY_IO_PORT,
+		    "Expected IN from port %d from L2, got port %d",
+		    ARBITRARY_IO_PORT, run->io.port);
+
+	/*
+	 * Stuff invalid guest state for L2 by making TR unusuable.  The next
+	 * KVM_RUN should induce a TRIPLE_FAULT in L2 as KVM doesn't support
+	 * emulating invalid guest state for L2.
+	 */
+	memset(&sregs, 0, sizeof(sregs));
+	vcpu_sregs_get(vm, VCPU_ID, &sregs);
+	sregs.tr.unusable = 1;
+	vcpu_sregs_set(vm, VCPU_ID, &sregs);
+
+	vcpu_run(vm, VCPU_ID);
+
+	switch (get_ucall(vm, VCPU_ID, &uc)) {
+	case UCALL_DONE:
+		break;
+	case UCALL_ABORT:
+		TEST_FAIL("%s", (const char *)uc.args[0]);
+	default:
+		TEST_FAIL("Unexpected ucall: %lu", uc.cmd);
+	}
+}
diff --git a/tools/testing/selftests/kvm/x86_64/vmx_pmu_msrs_test.c b/tools/testing/selftests/kvm/x86_64/vmx_pmu_msrs_test.c
index 23051d84b907..2454a1f2ca0c 100644
--- a/tools/testing/selftests/kvm/x86_64/vmx_pmu_msrs_test.c
+++ b/tools/testing/selftests/kvm/x86_64/vmx_pmu_msrs_test.c
@@ -110,22 +110,5 @@ int main(int argc, char *argv[])
 	ret = _vcpu_set_msr(vm, 0, MSR_IA32_PERF_CAPABILITIES, PMU_CAP_LBR_FMT);
 	TEST_ASSERT(ret == 0, "Bad PERF_CAPABILITIES didn't fail.");
 
-	/* testcase 4, set capabilities when we don't have PDCM bit */
-	entry_1_0->ecx &= ~X86_FEATURE_PDCM;
-	vcpu_set_cpuid(vm, VCPU_ID, cpuid);
-	ret = _vcpu_set_msr(vm, 0, MSR_IA32_PERF_CAPABILITIES, host_cap.capabilities);
-	TEST_ASSERT(ret == 0, "Bad PERF_CAPABILITIES didn't fail.");
-
-	/* testcase 5, set capabilities when we don't have PMU version bits */
-	entry_1_0->ecx |= X86_FEATURE_PDCM;
-	eax.split.version_id = 0;
-	entry_1_0->ecx = eax.full;
-	vcpu_set_cpuid(vm, VCPU_ID, cpuid);
-	ret = _vcpu_set_msr(vm, 0, MSR_IA32_PERF_CAPABILITIES, PMU_CAP_FW_WRITES);
-	TEST_ASSERT(ret == 0, "Bad PERF_CAPABILITIES didn't fail.");
-
-	vcpu_set_msr(vm, 0, MSR_IA32_PERF_CAPABILITIES, 0);
-	ASSERT_EQ(vcpu_get_msr(vm, VCPU_ID, MSR_IA32_PERF_CAPABILITIES), 0);
-
 	kvm_vm_free(vm);
 }