Merge tag 'kvm-x86-selftests-6.11' of https://github.com/kvm-x86/linux into HEAD

KVM selftests for 6.11

 - Remove dead code in the memslot modification stress test.

 - Treat "branch instructions retired" as supported on all AMD Family 17h+ CPUs.

 - Print the guest pseudo-RNG seed only when it changes, to avoid spamming the
   log for tests that create lots of VMs.

 - Make the PMU counters test less flaky when counting LLC cache misses by
   doing CLFLUSH{OPT} in every loop iteration.

Author: bonzini

tools/testing/selftests/kvm/lib/kvm_util.c

@@ -21,6 +21,7 @@
 
 uint32_t guest_random_seed;
 struct guest_random_state guest_rng;
+static uint32_t last_guest_seed;
 
 static int vcpu_mmap_sz(void);
 
@@ -434,7 +435,10 @@ 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);
 
-	pr_info("Random seed: 0x%x\n", guest_random_seed);
+	if (guest_random_seed != last_guest_seed) {
+		pr_info("Random seed: 0x%x\n", guest_random_seed);
+		last_guest_seed = guest_random_seed;
+	}
 	guest_rng = new_guest_random_state(guest_random_seed);
 	sync_global_to_guest(vm, guest_rng);
 
@@ -2319,7 +2323,8 @@ void __attribute((constructor)) kvm_selftest_init(void)
 	/* Tell stdout not to buffer its content. */
 	setbuf(stdout, NULL);
 
-	guest_random_seed = random();
+	guest_random_seed = last_guest_seed = random();
+	pr_info("Random seed: 0x%x\n", guest_random_seed);
 
 	kvm_selftest_arch_init();
 }

tools/testing/selftests/kvm/memslot_modification_stress_test.c

@@ -53,12 +53,6 @@ static void vcpu_worker(struct memstress_vcpu_args *vcpu_args)
 	}
 }
 
-struct memslot_antagonist_args {
-	struct kvm_vm *vm;
-	useconds_t delay;
-	uint64_t nr_modifications;
-};
-
 static void add_remove_memslot(struct kvm_vm *vm, useconds_t delay,
 			       uint64_t nr_modifications)
 {

tools/testing/selftests/kvm/x86_64/pmu_counters_test.c

@@ -7,15 +7,28 @@
 #include "pmu.h"
 #include "processor.h"
 
-/* Number of LOOP instructions for the guest measurement payload. */
-#define NUM_BRANCHES		10
+/* Number of iterations of the loop for the guest measurement payload. */
+#define NUM_LOOPS			10
+
+/* Each iteration of the loop retires one branch instruction. */
+#define NUM_BRANCH_INSNS_RETIRED	(NUM_LOOPS)
+
+/*
+ * Number of instructions in each loop. 1 CLFLUSH/CLFLUSHOPT/NOP, 1 MFENCE,
+ * 1 LOOP.
+ */
+#define NUM_INSNS_PER_LOOP		3
+
 /*
  * Number of "extra" instructions that will be counted, i.e. the number of
- * instructions that are needed to set up the loop and then disabled the
- * counter.  1 CLFLUSH/CLFLUSHOPT/NOP, 1 MFENCE, 2 MOV, 2 XOR, 1 WRMSR.
+ * instructions that are needed to set up the loop and then disable the
+ * counter.  2 MOV, 2 XOR, 1 WRMSR.
  */
-#define NUM_EXTRA_INSNS		7
-#define NUM_INSNS_RETIRED	(NUM_BRANCHES + NUM_EXTRA_INSNS)
+#define NUM_EXTRA_INSNS			5
+
+/* Total number of instructions retired within the measured section. */
+#define NUM_INSNS_RETIRED		(NUM_LOOPS * NUM_INSNS_PER_LOOP + NUM_EXTRA_INSNS)
+
 
 static uint8_t kvm_pmu_version;
 static bool kvm_has_perf_caps;
@@ -100,7 +113,7 @@ static void guest_assert_event_count(uint8_t idx,
 		GUEST_ASSERT_EQ(count, NUM_INSNS_RETIRED);
 		break;
 	case INTEL_ARCH_BRANCHES_RETIRED_INDEX:
-		GUEST_ASSERT_EQ(count, NUM_BRANCHES);
+		GUEST_ASSERT_EQ(count, NUM_BRANCH_INSNS_RETIRED);
 		break;
 	case INTEL_ARCH_LLC_REFERENCES_INDEX:
 	case INTEL_ARCH_LLC_MISSES_INDEX:
@@ -120,7 +133,7 @@ static void guest_assert_event_count(uint8_t idx,
 	}
 
 sanity_checks:
-	__asm__ __volatile__("loop ." : "+c"((int){NUM_BRANCHES}));
+	__asm__ __volatile__("loop ." : "+c"((int){NUM_LOOPS}));
 	GUEST_ASSERT_EQ(_rdpmc(pmc), count);
 
 	wrmsr(pmc_msr, 0xdead);
@@ -134,8 +147,8 @@ static void guest_assert_event_count(uint8_t idx,
  * before the end of the sequence.
  *
  * If CLFUSH{,OPT} is supported, flush the cacheline containing (at least) the
- * start of the loop to force LLC references and misses, i.e. to allow testing
- * that those events actually count.
+ * CLFUSH{,OPT} instruction on each loop iteration to force LLC references and
+ * misses, i.e. to allow testing that those events actually count.
  *
  * If forced emulation is enabled (and specified), force emulation on a subset
  * of the measured code to verify that KVM correctly emulates instructions and
@@ -145,10 +158,11 @@ static void guest_assert_event_count(uint8_t idx,
 #define GUEST_MEASURE_EVENT(_msr, _value, clflush, FEP)				\
 do {										\
 	__asm__ __volatile__("wrmsr\n\t"					\
+			     " mov $" __stringify(NUM_LOOPS) ", %%ecx\n\t"	\
+			     "1:\n\t"						\
 			     clflush "\n\t"					\
 			     "mfence\n\t"					\
-			     "1: mov $" __stringify(NUM_BRANCHES) ", %%ecx\n\t"	\
-			     FEP "loop .\n\t"					\
+			     FEP "loop 1b\n\t"					\
 			     FEP "mov %%edi, %%ecx\n\t"				\
 			     FEP "xor %%eax, %%eax\n\t"				\
 			     FEP "xor %%edx, %%edx\n\t"				\
@@ -163,9 +177,9 @@ do {										\
 	wrmsr(pmc_msr, 0);							\
 										\
 	if (this_cpu_has(X86_FEATURE_CLFLUSHOPT))				\
-		GUEST_MEASURE_EVENT(_ctrl_msr, _value, "clflushopt 1f", FEP);	\
+		GUEST_MEASURE_EVENT(_ctrl_msr, _value, "clflushopt .", FEP);	\
 	else if (this_cpu_has(X86_FEATURE_CLFLUSH))				\
-		GUEST_MEASURE_EVENT(_ctrl_msr, _value, "clflush 1f", FEP);	\
+		GUEST_MEASURE_EVENT(_ctrl_msr, _value, "clflush .", FEP);	\
 	else									\
 		GUEST_MEASURE_EVENT(_ctrl_msr, _value, "nop", FEP);		\
 										\
@@ -500,7 +514,7 @@ static void guest_test_fixed_counters(void)
 		wrmsr(MSR_CORE_PERF_FIXED_CTR0 + i, 0);
 		wrmsr(MSR_CORE_PERF_FIXED_CTR_CTRL, FIXED_PMC_CTRL(i, FIXED_PMC_KERNEL));
 		wrmsr(MSR_CORE_PERF_GLOBAL_CTRL, FIXED_PMC_GLOBAL_CTRL_ENABLE(i));
-		__asm__ __volatile__("loop ." : "+c"((int){NUM_BRANCHES}));
+		__asm__ __volatile__("loop ." : "+c"((int){NUM_LOOPS}));
 		wrmsr(MSR_CORE_PERF_GLOBAL_CTRL, 0);
 		val = rdmsr(MSR_CORE_PERF_FIXED_CTR0 + i);
 

tools/testing/selftests/kvm/x86_64/pmu_event_filter_test.c

@@ -32,8 +32,8 @@ struct __kvm_pmu_event_filter {
 
 /*
  * This event list comprises Intel's known architectural events, plus AMD's
- * "retired branch instructions" for Zen1-Zen3 (and* possibly other AMD CPUs).
- * Note, AMD and Intel use the same encoding for instructions retired.
+ * Branch Instructions Retired for Zen CPUs.  Note, AMD and Intel use the
+ * same encoding for Instructions Retired.
  */
 kvm_static_assert(INTEL_ARCH_INSTRUCTIONS_RETIRED == AMD_ZEN_INSTRUCTIONS_RETIRED);
 
@@ -353,38 +353,13 @@ static bool use_intel_pmu(void)
 	       kvm_pmu_has(X86_PMU_FEATURE_BRANCH_INSNS_RETIRED);
 }
 
-static bool is_zen1(uint32_t family, uint32_t model)
-{
-	return family == 0x17 && model <= 0x0f;
-}
-
-static bool is_zen2(uint32_t family, uint32_t model)
-{
-	return family == 0x17 && model >= 0x30 && model <= 0x3f;
-}
-
-static bool is_zen3(uint32_t family, uint32_t model)
-{
-	return family == 0x19 && model <= 0x0f;
-}
-
 /*
- * Determining AMD support for a PMU event requires consulting the AMD
- * PPR for the CPU or reference material derived therefrom. The AMD
- * test code herein has been verified to work on Zen1, Zen2, and Zen3.
- *
- * Feel free to add more AMD CPUs that are documented to support event
- * select 0xc2 umask 0 as "retired branch instructions."
+ * On AMD, all Family 17h+ CPUs (Zen and its successors) use event encoding
+ * 0xc2,0 for Branch Instructions Retired.
  */
 static bool use_amd_pmu(void)
 {
-	uint32_t family = kvm_cpu_family();
-	uint32_t model = kvm_cpu_model();
-
-	return host_cpu_is_amd &&
-		(is_zen1(family, model) ||
-		 is_zen2(family, model) ||
-		 is_zen3(family, model));
+	return host_cpu_is_amd && kvm_cpu_family() >= 0x17;
 }
 
 /*