11 files changed, 181 insertions, 7 deletions
diff --git a/tools/perf/arch/x86/include/arch-tests.h b/tools/perf/arch/x86/include/arch-tests.h
index 6a1a1b3c0827..902e9ea9b99e 100644
--- a/tools/perf/arch/x86/include/arch-tests.h
+++ b/tools/perf/arch/x86/include/arch-tests.h
@@ -8,6 +8,7 @@ struct test_suite;
 int test__rdpmc(struct test_suite *test, int subtest);
 int test__insn_x86(struct test_suite *test, int subtest);
 int test__intel_pt_pkt_decoder(struct test_suite *test, int subtest);
+int test__intel_pt_hybrid_compat(struct test_suite *test, int subtest);
 int test__bp_modify(struct test_suite *test, int subtest);
 int test__x86_sample_parsing(struct test_suite *test, int subtest);
 
diff --git a/tools/perf/arch/x86/tests/Build b/tools/perf/arch/x86/tests/Build
index 70b5bcbc15df..6f4e8636c3bf 100644
--- a/tools/perf/arch/x86/tests/Build
+++ b/tools/perf/arch/x86/tests/Build
@@ -3,5 +3,5 @@ perf-$(CONFIG_DWARF_UNWIND) += dwarf-unwind.o
 
 perf-y += arch-tests.o
 perf-y += sample-parsing.o
-perf-$(CONFIG_AUXTRACE) += insn-x86.o intel-pt-pkt-decoder-test.o
+perf-$(CONFIG_AUXTRACE) += insn-x86.o intel-pt-test.o
 perf-$(CONFIG_X86_64) += bp-modify.o
diff --git a/tools/perf/arch/x86/tests/arch-tests.c b/tools/perf/arch/x86/tests/arch-tests.c
index 04018b8aa85b..aae6ea0fe52b 100644
--- a/tools/perf/arch/x86/tests/arch-tests.c
+++ b/tools/perf/arch/x86/tests/arch-tests.c
@@ -5,7 +5,18 @@
 
 #ifdef HAVE_AUXTRACE_SUPPORT
 DEFINE_SUITE("x86 instruction decoder - new instructions", insn_x86);
-DEFINE_SUITE("Intel PT packet decoder", intel_pt_pkt_decoder);
+
+static struct test_case intel_pt_tests[] = {
+	TEST_CASE("Intel PT packet decoder", intel_pt_pkt_decoder),
+	TEST_CASE("Intel PT hybrid CPU compatibility", intel_pt_hybrid_compat),
+	{ .name = NULL, }
+};
+
+struct test_suite suite__intel_pt = {
+	.desc = "Intel PT",
+	.test_cases = intel_pt_tests,
+};
+
 #endif
 #if defined(__x86_64__)
 DEFINE_SUITE("x86 bp modify", bp_modify);
@@ -18,7 +29,7 @@ struct test_suite *arch_tests[] = {
 #endif
 #ifdef HAVE_AUXTRACE_SUPPORT
 	&suite__insn_x86,
-	&suite__intel_pt_pkt_decoder,
+	&suite__intel_pt,
 #endif
 #if defined(__x86_64__)
 	&suite__bp_modify,
diff --git a/tools/perf/arch/x86/tests/intel-pt-pkt-decoder-test.c b/tools/perf/arch/x86/tests/intel-pt-test.c
index 42237656f453..70b7f79396b1 100644
--- a/tools/perf/arch/x86/tests/intel-pt-pkt-decoder-test.c
+++ b/tools/perf/arch/x86/tests/intel-pt-test.c
@@ -1,12 +1,17 @@
 // SPDX-License-Identifier: GPL-2.0
 
+#include <linux/compiler.h>
+#include <linux/bits.h>
 #include <string.h>
+#include <cpuid.h>
+#include <sched.h>
 
 #include "intel-pt-decoder/intel-pt-pkt-decoder.h"
 
 #include "debug.h"
 #include "tests/tests.h"
 #include "arch-tests.h"
+#include "cpumap.h"
 
 /**
  * struct test_data - Test data.
@@ -313,3 +318,152 @@ int test__intel_pt_pkt_decoder(struct test_suite *test __maybe_unused, int subte
 
 	return TEST_OK;
 }
+
+static int setaffinity(int cpu)
+{
+	cpu_set_t cpu_set;
+
+	CPU_ZERO(&cpu_set);
+	CPU_SET(cpu, &cpu_set);
+	if (sched_setaffinity(0, sizeof(cpu_set), &cpu_set)) {
+		pr_debug("sched_setaffinity() failed for CPU %d\n", cpu);
+		return -1;
+	}
+	return 0;
+}
+
+#define INTEL_PT_ADDR_FILT_CNT_MASK	GENMASK(2, 0)
+#define INTEL_PT_SUBLEAF_CNT		2
+#define CPUID_REG_CNT			4
+
+struct cpuid_result {
+	union {
+		struct {
+			unsigned int eax;
+			unsigned int ebx;
+			unsigned int ecx;
+			unsigned int edx;
+		};
+		unsigned int reg[CPUID_REG_CNT];
+	};
+};
+
+struct pt_caps {
+	struct cpuid_result subleaf[INTEL_PT_SUBLEAF_CNT];
+};
+
+static int get_pt_caps(int cpu, struct pt_caps *caps)
+{
+	struct cpuid_result r;
+	int i;
+
+	if (setaffinity(cpu))
+		return -1;
+
+	memset(caps, 0, sizeof(*caps));
+
+	for (i = 0; i < INTEL_PT_SUBLEAF_CNT; i++) {
+		__get_cpuid_count(20, i, &r.eax, &r.ebx, &r.ecx, &r.edx);
+		pr_debug("CPU %d CPUID leaf 20 subleaf %d\n", cpu, i);
+		pr_debug("eax = 0x%08x\n", r.eax);
+		pr_debug("ebx = 0x%08x\n", r.ebx);
+		pr_debug("ecx = 0x%08x\n", r.ecx);
+		pr_debug("edx = 0x%08x\n", r.edx);
+		caps->subleaf[i] = r;
+	}
+
+	return 0;
+}
+
+static bool is_hydrid(void)
+{
+	unsigned int eax, ebx, ecx, edx = 0;
+	bool result;
+
+	__get_cpuid_count(7, 0, &eax, &ebx, &ecx, &edx);
+	result = edx & BIT(15);
+	pr_debug("Is %shybrid : CPUID leaf 7 subleaf 0 edx %#x (bit-15 indicates hybrid)\n",
+		 result ? "" : "not ", edx);
+	return result;
+}
+
+static int compare_caps(int cpu, struct pt_caps *caps, struct pt_caps *caps0)
+{
+	struct pt_caps mask = { /* Mask of bits to check*/
+		.subleaf = {
+			[0] = {
+				.ebx = GENMASK(8, 0),
+				.ecx = GENMASK(3, 0),
+			},
+			[1] = {
+				.eax = GENMASK(31, 16),
+				.ebx = GENMASK(31, 0),
+			}
+		}
+	};
+	unsigned int m, reg, reg0;
+	int ret = 0;
+	int i, j;
+
+	for (i = 0; i < INTEL_PT_SUBLEAF_CNT; i++) {
+		for (j = 0; j < CPUID_REG_CNT; j++) {
+			m = mask.subleaf[i].reg[j];
+			reg = m & caps->subleaf[i].reg[j];
+			reg0 = m & caps0->subleaf[i].reg[j];
+			if ((reg & reg0) != reg0) {
+				pr_debug("CPU %d subleaf %d reg %d FAIL %#x vs %#x\n",
+					 cpu, i, j, reg, reg0);
+				ret = -1;
+			}
+		}
+	}
+
+	m = INTEL_PT_ADDR_FILT_CNT_MASK;
+	reg = m & caps->subleaf[1].eax;
+	reg0 = m & caps0->subleaf[1].eax;
+	if (reg < reg0) {
+		pr_debug("CPU %d subleaf 1 reg 0 FAIL address filter count %#x vs %#x\n",
+			 cpu, reg, reg0);
+		ret = -1;
+	}
+
+	if (!ret)
+		pr_debug("CPU %d OK\n", cpu);
+
+	return ret;
+}
+
+int test__intel_pt_hybrid_compat(struct test_suite *test, int subtest)
+{
+	int max_cpu = cpu__max_cpu().cpu;
+	struct pt_caps last_caps;
+	struct pt_caps caps0;
+	int ret = TEST_OK;
+	int cpu;
+
+	if (!is_hydrid()) {
+		test->test_cases[subtest].skip_reason = "not hybrid";
+		return TEST_SKIP;
+	}
+
+	if (get_pt_caps(0, &caps0))
+		return TEST_FAIL;
+
+	for (cpu = 1, last_caps = caps0; cpu < max_cpu; cpu++) {
+		struct pt_caps caps;
+
+		if (get_pt_caps(cpu, &caps)) {
+			pr_debug("CPU %d not found\n", cpu);
+			continue;
+		}
+		if (!memcmp(&caps, &last_caps, sizeof(caps))) {
+			pr_debug("CPU %d same caps as previous CPU\n", cpu);
+			continue;
+		}
+		if (compare_caps(cpu, &caps, &caps0))
+			ret = TEST_FAIL;
+		last_caps = caps;
+	}
+
+	return ret;
+}
diff --git a/tools/perf/arch/x86/tests/sample-parsing.c b/tools/perf/arch/x86/tests/sample-parsing.c
index bfbd3662b69e..690c7c07e90d 100644
--- a/tools/perf/arch/x86/tests/sample-parsing.c
+++ b/tools/perf/arch/x86/tests/sample-parsing.c
@@ -10,6 +10,7 @@
 #include "event.h"
 #include "evsel.h"
 #include "debug.h"
+#include "util/sample.h"
 #include "util/synthetic-events.h"
 
 #include "tests/tests.h"
diff --git a/tools/perf/arch/x86/util/Build b/tools/perf/arch/x86/util/Build
index dbeb04cb336e..195ccfdef7aa 100644
--- a/tools/perf/arch/x86/util/Build
+++ b/tools/perf/arch/x86/util/Build
@@ -1,7 +1,7 @@
 perf-y += header.o
 perf-y += tsc.o
 perf-y += pmu.o
-perf-y += kvm-stat.o
+perf-$(CONFIG_LIBTRACEEVENT) += kvm-stat.o
 perf-y += perf_regs.o
 perf-y += topdown.o
 perf-y += machine.o
diff --git a/tools/perf/arch/x86/util/event.c b/tools/perf/arch/x86/util/event.c
index e670f3547581..a3acefe6d0c6 100644
--- a/tools/perf/arch/x86/util/event.c
+++ b/tools/perf/arch/x86/util/event.c
@@ -2,6 +2,7 @@
 #include <linux/types.h>
 #include <linux/string.h>
 #include <linux/zalloc.h>
+#include <stdlib.h>
 
 #include "../../../util/event.h"
 #include "../../../util/synthetic-events.h"
@@ -9,6 +10,7 @@
 #include "../../../util/tool.h"
 #include "../../../util/map.h"
 #include "../../../util/debug.h"
+#include "util/sample.h"
 
 #if defined(__x86_64__)
 
diff --git a/tools/perf/arch/x86/util/intel-pt.c b/tools/perf/arch/x86/util/intel-pt.c
index af102f471e9f..1e39a034cee9 100644
--- a/tools/perf/arch/x86/util/intel-pt.c
+++ b/tools/perf/arch/x86/util/intel-pt.c
@@ -418,6 +418,7 @@ static int intel_pt_info_fill(struct auxtrace_record *itr,
 	return 0;
 }
 
+#ifdef HAVE_LIBTRACEEVENT
 static int intel_pt_track_switches(struct evlist *evlist)
 {
 	const char *sched_switch = "sched:sched_switch";
@@ -439,6 +440,7 @@ static int intel_pt_track_switches(struct evlist *evlist)
 
 	return 0;
 }
+#endif
 
 static void intel_pt_valid_str(char *str, size_t len, u64 valid)
 {
@@ -829,6 +831,7 @@ static int intel_pt_recording_options(struct auxtrace_record *itr,
 					ptr->have_sched_switch = 2;
 			}
 		} else {
+#ifdef HAVE_LIBTRACEEVENT
 			err = intel_pt_track_switches(evlist);
 			if (err == -EPERM)
 				pr_debug2("Unable to select sched:sched_switch\n");
@@ -836,6 +839,7 @@ static int intel_pt_recording_options(struct auxtrace_record *itr,
 				return err;
 			else
 				ptr->have_sched_switch = 1;
+#endif
 		}
 	}
 
diff --git a/tools/perf/arch/x86/util/iostat.c b/tools/perf/arch/x86/util/iostat.c
index 404de795ec0b..7eb0a7b00b95 100644
--- a/tools/perf/arch/x86/util/iostat.c
+++ b/tools/perf/arch/x86/util/iostat.c
@@ -449,7 +449,7 @@ void iostat_print_metric(struct perf_stat_config *config, struct evsel *evsel,
 
 void iostat_print_counters(struct evlist *evlist,
 			   struct perf_stat_config *config, struct timespec *ts,
-			   char *prefix, iostat_print_counter_t print_cnt_cb)
+			   char *prefix, iostat_print_counter_t print_cnt_cb, void *arg)
 {
 	void *perf_device = NULL;
 	struct evsel *counter = evlist__first(evlist);
@@ -464,7 +464,7 @@ void iostat_print_counters(struct evlist *evlist,
 			iostat_prefix(evlist, config, prefix, ts);
 			fprintf(config->output, "\n%s", prefix);
 		}
-		print_cnt_cb(config, counter, prefix);
+		print_cnt_cb(config, counter, arg);
 	}
 	fputc('\n', config->output);
 }
diff --git a/tools/perf/arch/x86/util/tsc.c b/tools/perf/arch/x86/util/tsc.c
index eb2b5195bd02..9b99f48b923c 100644
--- a/tools/perf/arch/x86/util/tsc.c
+++ b/tools/perf/arch/x86/util/tsc.c
@@ -2,6 +2,7 @@
 #include <linux/types.h>
 #include <math.h>
 #include <string.h>
+#include <stdlib.h>
 
 #include "../../../util/debug.h"
 #include "../../../util/tsc.h"
diff --git a/tools/perf/arch/x86/util/unwind-libdw.c b/tools/perf/arch/x86/util/unwind-libdw.c
index eea2bf87232b..ef71e8bf80bf 100644
--- a/tools/perf/arch/x86/util/unwind-libdw.c
+++ b/tools/perf/arch/x86/util/unwind-libdw.c
@@ -2,7 +2,7 @@
 #include <elfutils/libdwfl.h>
 #include "../../../util/unwind-libdw.h"
 #include "../../../util/perf_regs.h"
-#include "../../../util/event.h"
+#include "util/sample.h"
 
 bool libdw__arch_set_initial_registers(Dwfl_Thread *thread, void *arg)
 {