Merge tag 'perf-core-2020-10-12' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull performance events updates from Ingo Molnar:
 "x86 Intel updates:

   - Add Jasper Lake support

   - Add support for TopDown metrics on Ice Lake

   - Fix Ice Lake & Tiger Lake uncore support, add Snow Ridge support

   - Add a PCI sub driver to support uncore PMUs where the PCI resources
     have been claimed already - extending the range of supported
     systems.

  x86 AMD updates:

   - Restore 'perf stat -a' behaviour to program the uncore PMU to count
     all CPU threads.

   - Fix setting the proper count when sampling Large Increment per
     Cycle events / 'paired' events.

   - Fix IBS Fetch sampling on F17h and some other IBS fine tuning,
     greatly reducing the number of interrupts when large sample periods
     are specified.

   - Extends Family 17h RAPL support to also work on compatible F19h
     machines.

  Core code updates:

   - Fix race in perf_mmap_close()

   - Add PERF_EV_CAP_SIBLING, to denote that sibling events should be
     closed if the leader is removed.

   - Smaller fixes and updates"

* tag 'perf-core-2020-10-12' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (45 commits)
  perf/core: Fix race in the perf_mmap_close() function
  perf/x86: Fix n_metric for cancelled txn
  perf/x86: Fix n_pair for cancelled txn
  x86/events/amd/iommu: Fix sizeof mismatch
  perf/x86/intel: Check perf metrics feature for each CPU
  perf/x86/intel: Fix Ice Lake event constraint table
  perf/x86/intel/uncore: Fix the scale of the IMC free-running events
  perf/x86/intel/uncore: Fix for iio mapping on Skylake Server
  perf/x86/msr: Add Jasper Lake support
  perf/x86/intel: Add Jasper Lake support
  perf/x86/intel/uncore: Reduce the number of CBOX counters
  perf/x86/intel/uncore: Update Ice Lake uncore units
  perf/x86/intel/uncore: Split the Ice Lake and Tiger Lake MSR uncore support
  perf/x86/intel/uncore: Support PCIe3 unit on Snow Ridge
  perf/x86/intel/uncore: Generic support for the PCI sub driver
  perf/x86/intel/uncore: Factor out uncore_pci_pmu_unregister()
  perf/x86/intel/uncore: Factor out uncore_pci_pmu_register()
  perf/x86/intel/uncore: Factor out uncore_pci_find_dev_pmu()
  perf/x86/intel/uncore: Factor out uncore_pci_get_dev_die_info()
  perf/amd/uncore: Inform the user how many counters each uncore PMU has
  ...

arch/x86/events/amd/ibs.c

@@ -89,6 +89,7 @@ struct perf_ibs {
 	u64				max_period;
 	unsigned long			offset_mask[1];
 	int				offset_max;
+	unsigned int			fetch_count_reset_broken : 1;
 	struct cpu_perf_ibs __percpu	*pcpu;
 
 	struct attribute		**format_attrs;
@@ -334,11 +335,18 @@ static u64 get_ibs_op_count(u64 config)
 {
 	u64 count = 0;
 
-	if (config & IBS_OP_VAL)
-		count += (config & IBS_OP_MAX_CNT) << 4; /* cnt rolled over */
-
-	if (ibs_caps & IBS_CAPS_RDWROPCNT)
-		count += (config & IBS_OP_CUR_CNT) >> 32;
+	/*
+	 * If the internal 27-bit counter rolled over, the count is MaxCnt
+	 * and the lower 7 bits of CurCnt are randomized.
+	 * Otherwise CurCnt has the full 27-bit current counter value.
+	 */
+	if (config & IBS_OP_VAL) {
+		count = (config & IBS_OP_MAX_CNT) << 4;
+		if (ibs_caps & IBS_CAPS_OPCNTEXT)
+			count += config & IBS_OP_MAX_CNT_EXT_MASK;
+	} else if (ibs_caps & IBS_CAPS_RDWROPCNT) {
+		count = (config & IBS_OP_CUR_CNT) >> 32;
+	}
 
 	return count;
 }
@@ -363,7 +371,12 @@ perf_ibs_event_update(struct perf_ibs *perf_ibs, struct perf_event *event,
 static inline void perf_ibs_enable_event(struct perf_ibs *perf_ibs,
 					 struct hw_perf_event *hwc, u64 config)
 {
-	wrmsrl(hwc->config_base, hwc->config | config | perf_ibs->enable_mask);
+	u64 tmp = hwc->config | config;
+
+	if (perf_ibs->fetch_count_reset_broken)
+		wrmsrl(hwc->config_base, tmp & ~perf_ibs->enable_mask);
+
+	wrmsrl(hwc->config_base, tmp | perf_ibs->enable_mask);
 }
 
 /*
@@ -394,7 +407,7 @@ static void perf_ibs_start(struct perf_event *event, int flags)
 	struct hw_perf_event *hwc = &event->hw;
 	struct perf_ibs *perf_ibs = container_of(event->pmu, struct perf_ibs, pmu);
 	struct cpu_perf_ibs *pcpu = this_cpu_ptr(perf_ibs->pcpu);
-	u64 period;
+	u64 period, config = 0;
 
 	if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
 		return;
@@ -403,13 +416,19 @@ static void perf_ibs_start(struct perf_event *event, int flags)
 	hwc->state = 0;
 
 	perf_ibs_set_period(perf_ibs, hwc, &period);
+	if (perf_ibs == &perf_ibs_op && (ibs_caps & IBS_CAPS_OPCNTEXT)) {
+		config |= period & IBS_OP_MAX_CNT_EXT_MASK;
+		period &= ~IBS_OP_MAX_CNT_EXT_MASK;
+	}
+	config |= period >> 4;
+
 	/*
 	 * Set STARTED before enabling the hardware, such that a subsequent NMI
 	 * must observe it.
 	 */
 	set_bit(IBS_STARTED,    pcpu->state);
 	clear_bit(IBS_STOPPING, pcpu->state);
-	perf_ibs_enable_event(perf_ibs, hwc, period >> 4);
+	perf_ibs_enable_event(perf_ibs, hwc, config);
 
 	perf_event_update_userpage(event);
 }
@@ -577,7 +596,7 @@ static int perf_ibs_handle_irq(struct perf_ibs *perf_ibs, struct pt_regs *iregs)
 	struct perf_ibs_data ibs_data;
 	int offset, size, check_rip, offset_max, throttle = 0;
 	unsigned int msr;
-	u64 *buf, *config, period;
+	u64 *buf, *config, period, new_config = 0;
 
 	if (!test_bit(IBS_STARTED, pcpu->state)) {
 fail:
@@ -626,18 +645,24 @@ static int perf_ibs_handle_irq(struct perf_ibs *perf_ibs, struct pt_regs *iregs)
 				       perf_ibs->offset_max,
 				       offset + 1);
 	} while (offset < offset_max);
+	/*
+	 * Read IbsBrTarget, IbsOpData4, and IbsExtdCtl separately
+	 * depending on their availability.
+	 * Can't add to offset_max as they are staggered
+	 */
 	if (event->attr.sample_type & PERF_SAMPLE_RAW) {
-		/*
-		 * Read IbsBrTarget and IbsOpData4 separately
-		 * depending on their availability.
-		 * Can't add to offset_max as they are staggered
-		 */
-		if (ibs_caps & IBS_CAPS_BRNTRGT) {
-			rdmsrl(MSR_AMD64_IBSBRTARGET, *buf++);
-			size++;
+		if (perf_ibs == &perf_ibs_op) {
+			if (ibs_caps & IBS_CAPS_BRNTRGT) {
+				rdmsrl(MSR_AMD64_IBSBRTARGET, *buf++);
+				size++;
+			}
+			if (ibs_caps & IBS_CAPS_OPDATA4) {
+				rdmsrl(MSR_AMD64_IBSOPDATA4, *buf++);
+				size++;
+			}
 		}
-		if (ibs_caps & IBS_CAPS_OPDATA4) {
-			rdmsrl(MSR_AMD64_IBSOPDATA4, *buf++);
+		if (perf_ibs == &perf_ibs_fetch && (ibs_caps & IBS_CAPS_FETCHCTLEXTD)) {
+			rdmsrl(MSR_AMD64_ICIBSEXTDCTL, *buf++);
 			size++;
 		}
 	}
@@ -666,13 +691,17 @@ static int perf_ibs_handle_irq(struct perf_ibs *perf_ibs, struct pt_regs *iregs)
 	if (throttle) {
 		perf_ibs_stop(event, 0);
 	} else {
-		period >>= 4;
-
-		if ((ibs_caps & IBS_CAPS_RDWROPCNT) &&
-		    (*config & IBS_OP_CNT_CTL))
-			period |= *config & IBS_OP_CUR_CNT_RAND;
+		if (perf_ibs == &perf_ibs_op) {
+			if (ibs_caps & IBS_CAPS_OPCNTEXT) {
+				new_config = period & IBS_OP_MAX_CNT_EXT_MASK;
+				period &= ~IBS_OP_MAX_CNT_EXT_MASK;
+			}
+			if ((ibs_caps & IBS_CAPS_RDWROPCNT) && (*config & IBS_OP_CNT_CTL))
+				new_config |= *config & IBS_OP_CUR_CNT_RAND;
+		}
+		new_config |= period >> 4;
 
-		perf_ibs_enable_event(perf_ibs, hwc, period);
+		perf_ibs_enable_event(perf_ibs, hwc, new_config);
 	}
 
 	perf_event_update_userpage(event);
@@ -733,12 +762,26 @@ static __init void perf_event_ibs_init(void)
 {
 	struct attribute **attr = ibs_op_format_attrs;
 
+	/*
+	 * Some chips fail to reset the fetch count when it is written; instead
+	 * they need a 0-1 transition of IbsFetchEn.
+	 */
+	if (boot_cpu_data.x86 >= 0x16 && boot_cpu_data.x86 <= 0x18)
+		perf_ibs_fetch.fetch_count_reset_broken = 1;
+
 	perf_ibs_pmu_init(&perf_ibs_fetch, "ibs_fetch");
 
 	if (ibs_caps & IBS_CAPS_OPCNT) {
 		perf_ibs_op.config_mask |= IBS_OP_CNT_CTL;
 		*attr++ = &format_attr_cnt_ctl.attr;
 	}
+
+	if (ibs_caps & IBS_CAPS_OPCNTEXT) {
+		perf_ibs_op.max_period  |= IBS_OP_MAX_CNT_EXT_MASK;
+		perf_ibs_op.config_mask	|= IBS_OP_MAX_CNT_EXT_MASK;
+		perf_ibs_op.cnt_mask    |= IBS_OP_MAX_CNT_EXT_MASK;
+	}
+
 	perf_ibs_pmu_init(&perf_ibs_op, "ibs_op");
 
 	register_nmi_handler(NMI_LOCAL, perf_ibs_nmi_handler, 0, "perf_ibs");

arch/x86/events/amd/iommu.c

@@ -379,7 +379,7 @@ static __init int _init_events_attrs(void)
 	while (amd_iommu_v2_event_descs[i].attr.attr.name)
 		i++;
 
-	attrs = kcalloc(i + 1, sizeof(struct attribute **), GFP_KERNEL);
+	attrs = kcalloc(i + 1, sizeof(*attrs), GFP_KERNEL);
 	if (!attrs)
 		return -ENOMEM;
 

arch/x86/events/amd/uncore.c

@@ -181,28 +181,28 @@ static void amd_uncore_del(struct perf_event *event, int flags)
 }
 
 /*
- * Convert logical CPU number to L3 PMC Config ThreadMask format
+ * Return a full thread and slice mask unless user
+ * has provided them
  */
-static u64 l3_thread_slice_mask(int cpu)
+static u64 l3_thread_slice_mask(u64 config)
 {
-	u64 thread_mask, core = topology_core_id(cpu);
-	unsigned int shift, thread = 0;
+	if (boot_cpu_data.x86 <= 0x18)
+		return ((config & AMD64_L3_SLICE_MASK) ? : AMD64_L3_SLICE_MASK) |
+		       ((config & AMD64_L3_THREAD_MASK) ? : AMD64_L3_THREAD_MASK);
 
-	if (topology_smt_supported() && !topology_is_primary_thread(cpu))
-		thread = 1;
-
-	if (boot_cpu_data.x86 <= 0x18) {
-		shift = AMD64_L3_THREAD_SHIFT + 2 * (core % 4) + thread;
-		thread_mask = BIT_ULL(shift);
-
-		return AMD64_L3_SLICE_MASK | thread_mask;
-	}
-
-	core = (core << AMD64_L3_COREID_SHIFT) & AMD64_L3_COREID_MASK;
-	shift = AMD64_L3_THREAD_SHIFT + thread;
-	thread_mask = BIT_ULL(shift);
+	/*
+	 * If the user doesn't specify a threadmask, they're not trying to
+	 * count core 0, so we enable all cores & threads.
+	 * We'll also assume that they want to count slice 0 if they specify
+	 * a threadmask and leave sliceid and enallslices unpopulated.
+	 */
+	if (!(config & AMD64_L3_F19H_THREAD_MASK))
+		return AMD64_L3_F19H_THREAD_MASK | AMD64_L3_EN_ALL_SLICES |
+		       AMD64_L3_EN_ALL_CORES;
 
-	return AMD64_L3_EN_ALL_SLICES | core | thread_mask;
+	return config & (AMD64_L3_F19H_THREAD_MASK | AMD64_L3_SLICEID_MASK |
+			 AMD64_L3_EN_ALL_CORES | AMD64_L3_EN_ALL_SLICES |
+			 AMD64_L3_COREID_MASK);
 }
 
 static int amd_uncore_event_init(struct perf_event *event)
@@ -232,7 +232,7 @@ static int amd_uncore_event_init(struct perf_event *event)
 	 * For other events, the two fields do not affect the count.
 	 */
 	if (l3_mask && is_llc_event(event))
-		hwc->config |= l3_thread_slice_mask(event->cpu);
+		hwc->config |= l3_thread_slice_mask(event->attr.config);
 
 	uncore = event_to_amd_uncore(event);
 	if (!uncore)
@@ -274,47 +274,72 @@ static struct attribute_group amd_uncore_attr_group = {
 	.attrs = amd_uncore_attrs,
 };
 
-/*
- * Similar to PMU_FORMAT_ATTR but allowing for format_attr to be assigned based
- * on family
- */
-#define AMD_FORMAT_ATTR(_dev, _name, _format)				     \
-static ssize_t								     \
-_dev##_show##_name(struct device *dev,					     \
-		struct device_attribute *attr,				     \
-		char *page)						     \
-{									     \
-	BUILD_BUG_ON(sizeof(_format) >= PAGE_SIZE);			     \
-	return sprintf(page, _format "\n");				     \
-}									     \
-static struct device_attribute format_attr_##_dev##_name = __ATTR_RO(_dev);
-
-/* Used for each uncore counter type */
-#define AMD_ATTRIBUTE(_name)						     \
-static struct attribute *amd_uncore_format_attr_##_name[] = {		     \
-	&format_attr_event_##_name.attr,				     \
-	&format_attr_umask.attr,					     \
-	NULL,								     \
-};									     \
-static struct attribute_group amd_uncore_format_group_##_name = {	     \
-	.name = "format",						     \
-	.attrs = amd_uncore_format_attr_##_name,			     \
-};									     \
-static const struct attribute_group *amd_uncore_attr_groups_##_name[] = {    \
-	&amd_uncore_attr_group,						     \
-	&amd_uncore_format_group_##_name,				     \
-	NULL,								     \
+#define DEFINE_UNCORE_FORMAT_ATTR(_var, _name, _format)			\
+static ssize_t __uncore_##_var##_show(struct kobject *kobj,		\
+				struct kobj_attribute *attr,		\
+				char *page)				\
+{									\
+	BUILD_BUG_ON(sizeof(_format) >= PAGE_SIZE);			\
+	return sprintf(page, _format "\n");				\
+}									\
+static struct kobj_attribute format_attr_##_var =			\
+	__ATTR(_name, 0444, __uncore_##_var##_show, NULL)
+
+DEFINE_UNCORE_FORMAT_ATTR(event12,	event,		"config:0-7,32-35");
+DEFINE_UNCORE_FORMAT_ATTR(event14,	event,		"config:0-7,32-35,59-60"); /* F17h+ DF */
+DEFINE_UNCORE_FORMAT_ATTR(event8,	event,		"config:0-7");		   /* F17h+ L3 */
+DEFINE_UNCORE_FORMAT_ATTR(umask,	umask,		"config:8-15");
+DEFINE_UNCORE_FORMAT_ATTR(coreid,	coreid,		"config:42-44");	   /* F19h L3 */
+DEFINE_UNCORE_FORMAT_ATTR(slicemask,	slicemask,	"config:48-51");	   /* F17h L3 */
+DEFINE_UNCORE_FORMAT_ATTR(threadmask8,	threadmask,	"config:56-63");	   /* F17h L3 */
+DEFINE_UNCORE_FORMAT_ATTR(threadmask2,	threadmask,	"config:56-57");	   /* F19h L3 */
+DEFINE_UNCORE_FORMAT_ATTR(enallslices,	enallslices,	"config:46");		   /* F19h L3 */
+DEFINE_UNCORE_FORMAT_ATTR(enallcores,	enallcores,	"config:47");		   /* F19h L3 */
+DEFINE_UNCORE_FORMAT_ATTR(sliceid,	sliceid,	"config:48-50");	   /* F19h L3 */
+
+static struct attribute *amd_uncore_df_format_attr[] = {
+	&format_attr_event12.attr, /* event14 if F17h+ */
+	&format_attr_umask.attr,
+	NULL,
+};
+
+static struct attribute *amd_uncore_l3_format_attr[] = {
+	&format_attr_event12.attr, /* event8 if F17h+ */
+	&format_attr_umask.attr,
+	NULL, /* slicemask if F17h,	coreid if F19h */
+	NULL, /* threadmask8 if F17h,	enallslices if F19h */
+	NULL, /*			enallcores if F19h */
+	NULL, /*			sliceid if F19h */
+	NULL, /*			threadmask2 if F19h */
+	NULL,
+};
+
+static struct attribute_group amd_uncore_df_format_group = {
+	.name = "format",
+	.attrs = amd_uncore_df_format_attr,
 };
 
-AMD_FORMAT_ATTR(event, , "config:0-7,32-35");
-AMD_FORMAT_ATTR(umask, , "config:8-15");
-AMD_FORMAT_ATTR(event, _df, "config:0-7,32-35,59-60");
-AMD_FORMAT_ATTR(event, _l3, "config:0-7");
-AMD_ATTRIBUTE(df);
-AMD_ATTRIBUTE(l3);
+static struct attribute_group amd_uncore_l3_format_group = {
+	.name = "format",
+	.attrs = amd_uncore_l3_format_attr,
+};
+
+static const struct attribute_group *amd_uncore_df_attr_groups[] = {
+	&amd_uncore_attr_group,
+	&amd_uncore_df_format_group,
+	NULL,
+};
+
+static const struct attribute_group *amd_uncore_l3_attr_groups[] = {
+	&amd_uncore_attr_group,
+	&amd_uncore_l3_format_group,
+	NULL,
+};
 
 static struct pmu amd_nb_pmu = {
 	.task_ctx_nr	= perf_invalid_context,
+	.attr_groups	= amd_uncore_df_attr_groups,
+	.name		= "amd_nb",
 	.event_init	= amd_uncore_event_init,
 	.add		= amd_uncore_add,
 	.del		= amd_uncore_del,
@@ -326,6 +351,8 @@ static struct pmu amd_nb_pmu = {
 
 static struct pmu amd_llc_pmu = {
 	.task_ctx_nr	= perf_invalid_context,
+	.attr_groups	= amd_uncore_l3_attr_groups,
+	.name		= "amd_l2",
 	.event_init	= amd_uncore_event_init,
 	.add		= amd_uncore_add,
 	.del		= amd_uncore_del,
@@ -529,6 +556,8 @@ static int amd_uncore_cpu_dead(unsigned int cpu)
 
 static int __init amd_uncore_init(void)
 {
+	struct attribute **df_attr = amd_uncore_df_format_attr;
+	struct attribute **l3_attr = amd_uncore_l3_format_attr;
 	int ret = -ENODEV;
 
 	if (boot_cpu_data.x86_vendor != X86_VENDOR_AMD &&
@@ -538,6 +567,8 @@ static int __init amd_uncore_init(void)
 	if (!boot_cpu_has(X86_FEATURE_TOPOEXT))
 		return -ENODEV;
 
+	num_counters_nb	= NUM_COUNTERS_NB;
+	num_counters_llc = NUM_COUNTERS_L2;
 	if (boot_cpu_data.x86 >= 0x17) {
 		/*
 		 * For F17h and above, the Northbridge counters are
@@ -545,27 +576,16 @@ static int __init amd_uncore_init(void)
 		 * counters are supported too. The PMUs are exported
 		 * based on family as either L2 or L3 and NB or DF.
 		 */
-		num_counters_nb		  = NUM_COUNTERS_NB;
 		num_counters_llc	  = NUM_COUNTERS_L3;
 		amd_nb_pmu.name		  = "amd_df";
 		amd_llc_pmu.name	  = "amd_l3";
-		format_attr_event_df.show = &event_show_df;
-		format_attr_event_l3.show = &event_show_l3;
 		l3_mask			  = true;
-	} else {
-		num_counters_nb		  = NUM_COUNTERS_NB;
-		num_counters_llc	  = NUM_COUNTERS_L2;
-		amd_nb_pmu.name		  = "amd_nb";
-		amd_llc_pmu.name	  = "amd_l2";
-		format_attr_event_df	  = format_attr_event;
-		format_attr_event_l3	  = format_attr_event;
-		l3_mask			  = false;
 	}
 
-	amd_nb_pmu.attr_groups	= amd_uncore_attr_groups_df;
-	amd_llc_pmu.attr_groups = amd_uncore_attr_groups_l3;
-
 	if (boot_cpu_has(X86_FEATURE_PERFCTR_NB)) {
+		if (boot_cpu_data.x86 >= 0x17)
+			*df_attr = &format_attr_event14.attr;
+
 		amd_uncore_nb = alloc_percpu(struct amd_uncore *);
 		if (!amd_uncore_nb) {
 			ret = -ENOMEM;
@@ -575,13 +595,29 @@ static int __init amd_uncore_init(void)
 		if (ret)
 			goto fail_nb;
 
-		pr_info("%s NB counters detected\n",
-			boot_cpu_data.x86_vendor == X86_VENDOR_HYGON ?
-				"HYGON" : "AMD");
+		pr_info("%d %s %s counters detected\n", num_counters_nb,
+			boot_cpu_data.x86_vendor == X86_VENDOR_HYGON ?  "HYGON" : "",
+			amd_nb_pmu.name);
+
 		ret = 0;
 	}
 
 	if (boot_cpu_has(X86_FEATURE_PERFCTR_LLC)) {
+		if (boot_cpu_data.x86 >= 0x19) {
+			*l3_attr++ = &format_attr_event8.attr;
+			*l3_attr++ = &format_attr_umask.attr;
+			*l3_attr++ = &format_attr_coreid.attr;
+			*l3_attr++ = &format_attr_enallslices.attr;
+			*l3_attr++ = &format_attr_enallcores.attr;
+			*l3_attr++ = &format_attr_sliceid.attr;
+			*l3_attr++ = &format_attr_threadmask2.attr;
+		} else if (boot_cpu_data.x86 >= 0x17) {
+			*l3_attr++ = &format_attr_event8.attr;
+			*l3_attr++ = &format_attr_umask.attr;
+			*l3_attr++ = &format_attr_slicemask.attr;
+			*l3_attr++ = &format_attr_threadmask8.attr;
+		}
+
 		amd_uncore_llc = alloc_percpu(struct amd_uncore *);
 		if (!amd_uncore_llc) {
 			ret = -ENOMEM;
@@ -591,9 +627,9 @@ static int __init amd_uncore_init(void)
 		if (ret)
 			goto fail_llc;
 
-		pr_info("%s LLC counters detected\n",
-			boot_cpu_data.x86_vendor == X86_VENDOR_HYGON ?
-				"HYGON" : "AMD");
+		pr_info("%d %s %s counters detected\n", num_counters_llc,
+			boot_cpu_data.x86_vendor == X86_VENDOR_HYGON ?  "HYGON" : "",
+			amd_llc_pmu.name);
 		ret = 0;
 	}
 

arch/x86/events/core.c

@@ -105,6 +105,9 @@ u64 x86_perf_event_update(struct perf_event *event)
 	if (unlikely(!hwc->event_base))
 		return 0;
 
+	if (unlikely(is_topdown_count(event)) && x86_pmu.update_topdown_event)
+		return x86_pmu.update_topdown_event(event);
+
 	/*
 	 * Careful: an NMI might modify the previous event value.
 	 *
@@ -1056,6 +1059,45 @@ int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
 	return unsched ? -EINVAL : 0;
 }
 
+static int add_nr_metric_event(struct cpu_hw_events *cpuc,
+			       struct perf_event *event)
+{
+	if (is_metric_event(event)) {
+		if (cpuc->n_metric == INTEL_TD_METRIC_NUM)
+			return -EINVAL;
+		cpuc->n_metric++;
+		cpuc->n_txn_metric++;
+	}
+
+	return 0;
+}
+
+static void del_nr_metric_event(struct cpu_hw_events *cpuc,
+				struct perf_event *event)
+{
+	if (is_metric_event(event))
+		cpuc->n_metric--;
+}
+
+static int collect_event(struct cpu_hw_events *cpuc, struct perf_event *event,
+			 int max_count, int n)
+{
+
+	if (x86_pmu.intel_cap.perf_metrics && add_nr_metric_event(cpuc, event))
+		return -EINVAL;
+
+	if (n >= max_count + cpuc->n_metric)
+		return -EINVAL;
+
+	cpuc->event_list[n] = event;
+	if (is_counter_pair(&event->hw)) {
+		cpuc->n_pair++;
+		cpuc->n_txn_pair++;
+	}
+
+	return 0;
+}
+
 /*
  * dogrp: true if must collect siblings events (group)
  * returns total number of events and error code
@@ -1092,28 +1134,22 @@ static int collect_events(struct cpu_hw_events *cpuc, struct perf_event *leader,
 	}
 
 	if (is_x86_event(leader)) {
-		if (n >= max_count)
+		if (collect_event(cpuc, leader, max_count, n))
 			return -EINVAL;
-		cpuc->event_list[n] = leader;
 		n++;
-		if (is_counter_pair(&leader->hw))
-			cpuc->n_pair++;
 	}
+
 	if (!dogrp)
 		return n;
 
 	for_each_sibling_event(event, leader) {
-		if (!is_x86_event(event) ||
-		    event->state <= PERF_EVENT_STATE_OFF)
+		if (!is_x86_event(event) || event->state <= PERF_EVENT_STATE_OFF)
 			continue;
 
-		if (n >= max_count)
+		if (collect_event(cpuc, event, max_count, n))
 			return -EINVAL;
 
-		cpuc->event_list[n] = event;
 		n++;
-		if (is_counter_pair(&event->hw))
-			cpuc->n_pair++;
 	}
 	return n;
 }
@@ -1135,11 +1171,16 @@ static inline void x86_assign_hw_event(struct perf_event *event,
 		hwc->event_base	= 0;
 		break;
 
+	case INTEL_PMC_IDX_METRIC_BASE ... INTEL_PMC_IDX_METRIC_END:
+		/* All the metric events are mapped onto the fixed counter 3. */
+		idx = INTEL_PMC_IDX_FIXED_SLOTS;
+		/* fall through */
 	case INTEL_PMC_IDX_FIXED ... INTEL_PMC_IDX_FIXED_BTS-1:
 		hwc->config_base = MSR_ARCH_PERFMON_FIXED_CTR_CTRL;
 		hwc->event_base = MSR_ARCH_PERFMON_FIXED_CTR0 +
 				(idx - INTEL_PMC_IDX_FIXED);
-		hwc->event_base_rdpmc = (idx - INTEL_PMC_IDX_FIXED) | 1<<30;
+		hwc->event_base_rdpmc = (idx - INTEL_PMC_IDX_FIXED) |
+					INTEL_PMC_FIXED_RDPMC_BASE;
 		break;
 
 	default:
@@ -1270,6 +1311,10 @@ int x86_perf_event_set_period(struct perf_event *event)
 	if (unlikely(!hwc->event_base))
 		return 0;
 
+	if (unlikely(is_topdown_count(event)) &&
+	    x86_pmu.set_topdown_event_period)
+		return x86_pmu.set_topdown_event_period(event);
+
 	/*
 	 * If we are way outside a reasonable range then just skip forward:
 	 */
@@ -1309,11 +1354,11 @@ int x86_perf_event_set_period(struct perf_event *event)
 	wrmsrl(hwc->event_base, (u64)(-left) & x86_pmu.cntval_mask);
 
 	/*
-	 * Clear the Merge event counter's upper 16 bits since
+	 * Sign extend the Merge event counter's upper 16 bits since
 	 * we currently declare a 48-bit counter width
 	 */
 	if (is_counter_pair(hwc))
-		wrmsrl(x86_pmu_event_addr(idx + 1), 0);
+		wrmsrl(x86_pmu_event_addr(idx + 1), 0xffff);
 
 	/*
 	 * Due to erratum on certan cpu we need
@@ -1551,6 +1596,8 @@ static void x86_pmu_del(struct perf_event *event, int flags)
 	}
 	cpuc->event_constraint[i-1] = NULL;
 	--cpuc->n_events;
+	if (x86_pmu.intel_cap.perf_metrics)
+		del_nr_metric_event(cpuc, event);
 
 	perf_event_update_userpage(event);
 
@@ -2018,6 +2065,8 @@ static void x86_pmu_start_txn(struct pmu *pmu, unsigned int txn_flags)
 
 	perf_pmu_disable(pmu);
 	__this_cpu_write(cpu_hw_events.n_txn, 0);
+	__this_cpu_write(cpu_hw_events.n_txn_pair, 0);
+	__this_cpu_write(cpu_hw_events.n_txn_metric, 0);
 }
 
 /*
@@ -2043,6 +2092,8 @@ static void x86_pmu_cancel_txn(struct pmu *pmu)
 	 */
 	__this_cpu_sub(cpu_hw_events.n_added, __this_cpu_read(cpu_hw_events.n_txn));
 	__this_cpu_sub(cpu_hw_events.n_events, __this_cpu_read(cpu_hw_events.n_txn));
+	__this_cpu_sub(cpu_hw_events.n_pair, __this_cpu_read(cpu_hw_events.n_txn_pair));
+	__this_cpu_sub(cpu_hw_events.n_metric, __this_cpu_read(cpu_hw_events.n_txn_metric));
 	perf_pmu_enable(pmu);
 }
 
@@ -2264,17 +2315,15 @@ static void x86_pmu_event_unmapped(struct perf_event *event, struct mm_struct *m
 
 static int x86_pmu_event_idx(struct perf_event *event)
 {
-	int idx = event->hw.idx;
+	struct hw_perf_event *hwc = &event->hw;
 
-	if (!(event->hw.flags & PERF_X86_EVENT_RDPMC_ALLOWED))
+	if (!(hwc->flags & PERF_X86_EVENT_RDPMC_ALLOWED))
 		return 0;
 
-	if (x86_pmu.num_counters_fixed && idx >= INTEL_PMC_IDX_FIXED) {
-		idx -= INTEL_PMC_IDX_FIXED;
-		idx |= 1 << 30;
-	}
-
-	return idx + 1;
+	if (is_metric_idx(hwc->idx))
+		return INTEL_PMC_FIXED_RDPMC_METRICS + 1;
+	else
+		return hwc->event_base_rdpmc + 1;
 }
 
 static ssize_t get_attr_rdpmc(struct device *cdev,

arch/x86/events/intel/core.c

@@ -243,10 +243,14 @@ static struct extra_reg intel_skl_extra_regs[] __read_mostly = {
 
 static struct event_constraint intel_icl_event_constraints[] = {
 	FIXED_EVENT_CONSTRAINT(0x00c0, 0),	/* INST_RETIRED.ANY */
-	INTEL_UEVENT_CONSTRAINT(0x1c0, 0),	/* INST_RETIRED.PREC_DIST */
+	FIXED_EVENT_CONSTRAINT(0x01c0, 0),	/* INST_RETIRED.PREC_DIST */
 	FIXED_EVENT_CONSTRAINT(0x003c, 1),	/* CPU_CLK_UNHALTED.CORE */
 	FIXED_EVENT_CONSTRAINT(0x0300, 2),	/* CPU_CLK_UNHALTED.REF */
 	FIXED_EVENT_CONSTRAINT(0x0400, 3),	/* SLOTS */
+	METRIC_EVENT_CONSTRAINT(INTEL_TD_METRIC_RETIRING, 0),
+	METRIC_EVENT_CONSTRAINT(INTEL_TD_METRIC_BAD_SPEC, 1),
+	METRIC_EVENT_CONSTRAINT(INTEL_TD_METRIC_FE_BOUND, 2),
+	METRIC_EVENT_CONSTRAINT(INTEL_TD_METRIC_BE_BOUND, 3),
 	INTEL_EVENT_CONSTRAINT_RANGE(0x03, 0x0a, 0xf),
 	INTEL_EVENT_CONSTRAINT_RANGE(0x1f, 0x28, 0xf),
 	INTEL_EVENT_CONSTRAINT(0x32, 0xf),	/* SW_PREFETCH_ACCESS.* */
@@ -309,6 +313,12 @@ EVENT_ATTR_STR_HT(topdown-recovery-bubbles, td_recovery_bubbles,
 EVENT_ATTR_STR_HT(topdown-recovery-bubbles.scale, td_recovery_bubbles_scale,
 	"4", "2");
 
+EVENT_ATTR_STR(slots,			slots,		"event=0x00,umask=0x4");
+EVENT_ATTR_STR(topdown-retiring,	td_retiring,	"event=0x00,umask=0x80");
+EVENT_ATTR_STR(topdown-bad-spec,	td_bad_spec,	"event=0x00,umask=0x81");
+EVENT_ATTR_STR(topdown-fe-bound,	td_fe_bound,	"event=0x00,umask=0x82");
+EVENT_ATTR_STR(topdown-be-bound,	td_be_bound,	"event=0x00,umask=0x83");
+
 static struct attribute *snb_events_attrs[] = {
 	EVENT_PTR(td_slots_issued),
 	EVENT_PTR(td_slots_retired),
@@ -2165,11 +2175,24 @@ static inline void intel_clear_masks(struct perf_event *event, int idx)
 static void intel_pmu_disable_fixed(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
-	int idx = hwc->idx - INTEL_PMC_IDX_FIXED;
 	u64 ctrl_val, mask;
+	int idx = hwc->idx;
 
-	mask = 0xfULL << (idx * 4);
+	if (is_topdown_idx(idx)) {
+		struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
+
+		/*
+		 * When there are other active TopDown events,
+		 * don't disable the fixed counter 3.
+		 */
+		if (*(u64 *)cpuc->active_mask & INTEL_PMC_OTHER_TOPDOWN_BITS(idx))
+			return;
+		idx = INTEL_PMC_IDX_FIXED_SLOTS;
+	}
 
+	intel_clear_masks(event, idx);
+
+	mask = 0xfULL << ((idx - INTEL_PMC_IDX_FIXED) * 4);
 	rdmsrl(hwc->config_base, ctrl_val);
 	ctrl_val &= ~mask;
 	wrmsrl(hwc->config_base, ctrl_val);
@@ -2180,17 +2203,28 @@ static void intel_pmu_disable_event(struct perf_event *event)
 	struct hw_perf_event *hwc = &event->hw;
 	int idx = hwc->idx;
 
-	if (idx < INTEL_PMC_IDX_FIXED) {
+	switch (idx) {
+	case 0 ... INTEL_PMC_IDX_FIXED - 1:
 		intel_clear_masks(event, idx);
 		x86_pmu_disable_event(event);
-	} else if (idx < INTEL_PMC_IDX_FIXED_BTS) {
-		intel_clear_masks(event, idx);
+		break;
+	case INTEL_PMC_IDX_FIXED ... INTEL_PMC_IDX_FIXED_BTS - 1:
+	case INTEL_PMC_IDX_METRIC_BASE ... INTEL_PMC_IDX_METRIC_END:
 		intel_pmu_disable_fixed(event);
-	} else if (idx == INTEL_PMC_IDX_FIXED_BTS) {
+		break;
+	case INTEL_PMC_IDX_FIXED_BTS:
 		intel_pmu_disable_bts();
 		intel_pmu_drain_bts_buffer();
-	} else if (idx == INTEL_PMC_IDX_FIXED_VLBR)
+		return;
+	case INTEL_PMC_IDX_FIXED_VLBR:
 		intel_clear_masks(event, idx);
+		break;
+	default:
+		intel_clear_masks(event, idx);
+		pr_warn("Failed to disable the event with invalid index %d\n",
+			idx);
+		return;
+	}
 
 	/*
 	 * Needs to be called after x86_pmu_disable_event,
@@ -2208,10 +2242,189 @@ static void intel_pmu_del_event(struct perf_event *event)
 		intel_pmu_pebs_del(event);
 }
 
+static int icl_set_topdown_event_period(struct perf_event *event)
+{
+	struct hw_perf_event *hwc = &event->hw;
+	s64 left = local64_read(&hwc->period_left);
+
+	/*
+	 * The values in PERF_METRICS MSR are derived from fixed counter 3.
+	 * Software should start both registers, PERF_METRICS and fixed
+	 * counter 3, from zero.
+	 * Clear PERF_METRICS and Fixed counter 3 in initialization.
+	 * After that, both MSRs will be cleared for each read.
+	 * Don't need to clear them again.
+	 */
+	if (left == x86_pmu.max_period) {
+		wrmsrl(MSR_CORE_PERF_FIXED_CTR3, 0);
+		wrmsrl(MSR_PERF_METRICS, 0);
+		hwc->saved_slots = 0;
+		hwc->saved_metric = 0;
+	}
+
+	if ((hwc->saved_slots) && is_slots_event(event)) {
+		wrmsrl(MSR_CORE_PERF_FIXED_CTR3, hwc->saved_slots);
+		wrmsrl(MSR_PERF_METRICS, hwc->saved_metric);
+	}
+
+	perf_event_update_userpage(event);
+
+	return 0;
+}
+
+static inline u64 icl_get_metrics_event_value(u64 metric, u64 slots, int idx)
+{
+	u32 val;
+
+	/*
+	 * The metric is reported as an 8bit integer fraction
+	 * suming up to 0xff.
+	 * slots-in-metric = (Metric / 0xff) * slots
+	 */
+	val = (metric >> ((idx - INTEL_PMC_IDX_METRIC_BASE) * 8)) & 0xff;
+	return  mul_u64_u32_div(slots, val, 0xff);
+}
+
+static u64 icl_get_topdown_value(struct perf_event *event,
+				       u64 slots, u64 metrics)
+{
+	int idx = event->hw.idx;
+	u64 delta;
+
+	if (is_metric_idx(idx))
+		delta = icl_get_metrics_event_value(metrics, slots, idx);
+	else
+		delta = slots;
+
+	return delta;
+}
+
+static void __icl_update_topdown_event(struct perf_event *event,
+				       u64 slots, u64 metrics,
+				       u64 last_slots, u64 last_metrics)
+{
+	u64 delta, last = 0;
+
+	delta = icl_get_topdown_value(event, slots, metrics);
+	if (last_slots)
+		last = icl_get_topdown_value(event, last_slots, last_metrics);
+
+	/*
+	 * The 8bit integer fraction of metric may be not accurate,
+	 * especially when the changes is very small.
+	 * For example, if only a few bad_spec happens, the fraction
+	 * may be reduced from 1 to 0. If so, the bad_spec event value
+	 * will be 0 which is definitely less than the last value.
+	 * Avoid update event->count for this case.
+	 */
+	if (delta > last) {
+		delta -= last;
+		local64_add(delta, &event->count);
+	}
+}
+
+static void update_saved_topdown_regs(struct perf_event *event,
+				      u64 slots, u64 metrics)
+{
+	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
+	struct perf_event *other;
+	int idx;
+
+	event->hw.saved_slots = slots;
+	event->hw.saved_metric = metrics;
+
+	for_each_set_bit(idx, cpuc->active_mask, INTEL_PMC_IDX_TD_BE_BOUND + 1) {
+		if (!is_topdown_idx(idx))
+			continue;
+		other = cpuc->events[idx];
+		other->hw.saved_slots = slots;
+		other->hw.saved_metric = metrics;
+	}
+}
+
+/*
+ * Update all active Topdown events.
+ *
+ * The PERF_METRICS and Fixed counter 3 are read separately. The values may be
+ * modify by a NMI. PMU has to be disabled before calling this function.
+ */
+static u64 icl_update_topdown_event(struct perf_event *event)
+{
+	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
+	struct perf_event *other;
+	u64 slots, metrics;
+	bool reset = true;
+	int idx;
+
+	/* read Fixed counter 3 */
+	rdpmcl((3 | INTEL_PMC_FIXED_RDPMC_BASE), slots);
+	if (!slots)
+		return 0;
+
+	/* read PERF_METRICS */
+	rdpmcl(INTEL_PMC_FIXED_RDPMC_METRICS, metrics);
+
+	for_each_set_bit(idx, cpuc->active_mask, INTEL_PMC_IDX_TD_BE_BOUND + 1) {
+		if (!is_topdown_idx(idx))
+			continue;
+		other = cpuc->events[idx];
+		__icl_update_topdown_event(other, slots, metrics,
+					   event ? event->hw.saved_slots : 0,
+					   event ? event->hw.saved_metric : 0);
+	}
+
+	/*
+	 * Check and update this event, which may have been cleared
+	 * in active_mask e.g. x86_pmu_stop()
+	 */
+	if (event && !test_bit(event->hw.idx, cpuc->active_mask)) {
+		__icl_update_topdown_event(event, slots, metrics,
+					   event->hw.saved_slots,
+					   event->hw.saved_metric);
+
+		/*
+		 * In x86_pmu_stop(), the event is cleared in active_mask first,
+		 * then drain the delta, which indicates context switch for
+		 * counting.
+		 * Save metric and slots for context switch.
+		 * Don't need to reset the PERF_METRICS and Fixed counter 3.
+		 * Because the values will be restored in next schedule in.
+		 */
+		update_saved_topdown_regs(event, slots, metrics);
+		reset = false;
+	}
+
+	if (reset) {
+		/* The fixed counter 3 has to be written before the PERF_METRICS. */
+		wrmsrl(MSR_CORE_PERF_FIXED_CTR3, 0);
+		wrmsrl(MSR_PERF_METRICS, 0);
+		if (event)
+			update_saved_topdown_regs(event, 0, 0);
+	}
+
+	return slots;
+}
+
+static void intel_pmu_read_topdown_event(struct perf_event *event)
+{
+	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
+
+	/* Only need to call update_topdown_event() once for group read. */
+	if ((cpuc->txn_flags & PERF_PMU_TXN_READ) &&
+	    !is_slots_event(event))
+		return;
+
+	perf_pmu_disable(event->pmu);
+	x86_pmu.update_topdown_event(event);
+	perf_pmu_enable(event->pmu);
+}
+
 static void intel_pmu_read_event(struct perf_event *event)
 {
 	if (event->hw.flags & PERF_X86_EVENT_AUTO_RELOAD)
 		intel_pmu_auto_reload_read(event);
+	else if (is_topdown_count(event) && x86_pmu.update_topdown_event)
+		intel_pmu_read_topdown_event(event);
 	else
 		x86_perf_event_update(event);
 }
@@ -2219,8 +2432,22 @@ static void intel_pmu_read_event(struct perf_event *event)
 static void intel_pmu_enable_fixed(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
-	int idx = hwc->idx - INTEL_PMC_IDX_FIXED;
 	u64 ctrl_val, mask, bits = 0;
+	int idx = hwc->idx;
+
+	if (is_topdown_idx(idx)) {
+		struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
+		/*
+		 * When there are other active TopDown events,
+		 * don't enable the fixed counter 3 again.
+		 */
+		if (*(u64 *)cpuc->active_mask & INTEL_PMC_OTHER_TOPDOWN_BITS(idx))
+			return;
+
+		idx = INTEL_PMC_IDX_FIXED_SLOTS;
+	}
+
+	intel_set_masks(event, idx);
 
 	/*
 	 * Enable IRQ generation (0x8), if not PEBS,
@@ -2240,6 +2467,7 @@ static void intel_pmu_enable_fixed(struct perf_event *event)
 	if (x86_pmu.version > 2 && hwc->config & ARCH_PERFMON_EVENTSEL_ANY)
 		bits |= 0x4;
 
+	idx -= INTEL_PMC_IDX_FIXED;
 	bits <<= (idx * 4);
 	mask = 0xfULL << (idx * 4);
 
@@ -2262,18 +2490,27 @@ static void intel_pmu_enable_event(struct perf_event *event)
 	if (unlikely(event->attr.precise_ip))
 		intel_pmu_pebs_enable(event);
 
-	if (idx < INTEL_PMC_IDX_FIXED) {
+	switch (idx) {
+	case 0 ... INTEL_PMC_IDX_FIXED - 1:
 		intel_set_masks(event, idx);
 		__x86_pmu_enable_event(hwc, ARCH_PERFMON_EVENTSEL_ENABLE);
-	} else if (idx < INTEL_PMC_IDX_FIXED_BTS) {
-		intel_set_masks(event, idx);
+		break;
+	case INTEL_PMC_IDX_FIXED ... INTEL_PMC_IDX_FIXED_BTS - 1:
+	case INTEL_PMC_IDX_METRIC_BASE ... INTEL_PMC_IDX_METRIC_END:
 		intel_pmu_enable_fixed(event);
-	} else if (idx == INTEL_PMC_IDX_FIXED_BTS) {
+		break;
+	case INTEL_PMC_IDX_FIXED_BTS:
 		if (!__this_cpu_read(cpu_hw_events.enabled))
 			return;
 		intel_pmu_enable_bts(hwc->config);
-	} else if (idx == INTEL_PMC_IDX_FIXED_VLBR)
+		break;
+	case INTEL_PMC_IDX_FIXED_VLBR:
 		intel_set_masks(event, idx);
+		break;
+	default:
+		pr_warn("Failed to enable the event with invalid index %d\n",
+			idx);
+	}
 }
 
 static void intel_pmu_add_event(struct perf_event *event)
@@ -2389,7 +2626,7 @@ static int handle_pmi_common(struct pt_regs *regs, u64 status)
 	/*
 	 * PEBS overflow sets bit 62 in the global status register
 	 */
-	if (__test_and_clear_bit(62, (unsigned long *)&status)) {
+	if (__test_and_clear_bit(GLOBAL_STATUS_BUFFER_OVF_BIT, (unsigned long *)&status)) {
 		u64 pebs_enabled = cpuc->pebs_enabled;
 
 		handled++;
@@ -2410,7 +2647,7 @@ static int handle_pmi_common(struct pt_regs *regs, u64 status)
 	/*
 	 * Intel PT
 	 */
-	if (__test_and_clear_bit(55, (unsigned long *)&status)) {
+	if (__test_and_clear_bit(GLOBAL_STATUS_TRACE_TOPAPMI_BIT, (unsigned long *)&status)) {
 		handled++;
 		if (unlikely(perf_guest_cbs && perf_guest_cbs->is_in_guest() &&
 			perf_guest_cbs->handle_intel_pt_intr))
@@ -2419,6 +2656,15 @@ static int handle_pmi_common(struct pt_regs *regs, u64 status)
 			intel_pt_interrupt();
 	}
 
+	/*
+	 * Intel Perf mertrics
+	 */
+	if (__test_and_clear_bit(GLOBAL_STATUS_PERF_METRICS_OVF_BIT, (unsigned long *)&status)) {
+		handled++;
+		if (x86_pmu.update_topdown_event)
+			x86_pmu.update_topdown_event(NULL);
+	}
+
 	/*
 	 * Checkpointed counters can lead to 'spurious' PMIs because the
 	 * rollback caused by the PMI will have cleared the overflow status
@@ -3355,6 +3601,56 @@ static int intel_pmu_hw_config(struct perf_event *event)
 	if (event->attr.type != PERF_TYPE_RAW)
 		return 0;
 
+	/*
+	 * Config Topdown slots and metric events
+	 *
+	 * The slots event on Fixed Counter 3 can support sampling,
+	 * which will be handled normally in x86_perf_event_update().
+	 *
+	 * Metric events don't support sampling and require being paired
+	 * with a slots event as group leader. When the slots event
+	 * is used in a metrics group, it too cannot support sampling.
+	 */
+	if (x86_pmu.intel_cap.perf_metrics && is_topdown_event(event)) {
+		if (event->attr.config1 || event->attr.config2)
+			return -EINVAL;
+
+		/*
+		 * The TopDown metrics events and slots event don't
+		 * support any filters.
+		 */
+		if (event->attr.config & X86_ALL_EVENT_FLAGS)
+			return -EINVAL;
+
+		if (is_metric_event(event)) {
+			struct perf_event *leader = event->group_leader;
+
+			/* The metric events don't support sampling. */
+			if (is_sampling_event(event))
+				return -EINVAL;
+
+			/* The metric events require a slots group leader. */
+			if (!is_slots_event(leader))
+				return -EINVAL;
+
+			/*
+			 * The leader/SLOTS must not be a sampling event for
+			 * metric use; hardware requires it starts at 0 when used
+			 * in conjunction with MSR_PERF_METRICS.
+			 */
+			if (is_sampling_event(leader))
+				return -EINVAL;
+
+			event->event_caps |= PERF_EV_CAP_SIBLING;
+			/*
+			 * Only once we have a METRICs sibling do we
+			 * need TopDown magic.
+			 */
+			leader->hw.flags |= PERF_X86_EVENT_TOPDOWN;
+			event->hw.flags  |= PERF_X86_EVENT_TOPDOWN;
+		}
+	}
+
 	if (!(event->attr.config & ARCH_PERFMON_EVENTSEL_ANY))
 		return 0;
 
@@ -3787,6 +4083,17 @@ static void intel_pmu_cpu_starting(int cpu)
 	if (x86_pmu.counter_freezing)
 		enable_counter_freeze();
 
+	/* Disable perf metrics if any added CPU doesn't support it. */
+	if (x86_pmu.intel_cap.perf_metrics) {
+		union perf_capabilities perf_cap;
+
+		rdmsrl(MSR_IA32_PERF_CAPABILITIES, perf_cap.capabilities);
+		if (!perf_cap.perf_metrics) {
+			x86_pmu.intel_cap.perf_metrics = 0;
+			x86_pmu.intel_ctrl &= ~(1ULL << GLOBAL_CTRL_EN_PERF_METRICS);
+		}
+	}
+
 	if (!cpuc->shared_regs)
 		return;
 
@@ -4355,6 +4662,15 @@ static struct attribute *icl_events_attrs[] = {
 	NULL,
 };
 
+static struct attribute *icl_td_events_attrs[] = {
+	EVENT_PTR(slots),
+	EVENT_PTR(td_retiring),
+	EVENT_PTR(td_bad_spec),
+	EVENT_PTR(td_fe_bound),
+	EVENT_PTR(td_be_bound),
+	NULL,
+};
+
 static struct attribute *icl_tsx_events_attrs[] = {
 	EVENT_PTR(tx_start),
 	EVENT_PTR(tx_abort),
@@ -4830,6 +5146,7 @@ __init int intel_pmu_init(void)
 
 	case INTEL_FAM6_ATOM_TREMONT_D:
 	case INTEL_FAM6_ATOM_TREMONT:
+	case INTEL_FAM6_ATOM_TREMONT_L:
 		x86_pmu.late_ack = true;
 		memcpy(hw_cache_event_ids, glp_hw_cache_event_ids,
 		       sizeof(hw_cache_event_ids));
@@ -5139,10 +5456,13 @@ __init int intel_pmu_init(void)
 			hsw_format_attr : nhm_format_attr;
 		extra_skl_attr = skl_format_attr;
 		mem_attr = icl_events_attrs;
+		td_attr = icl_td_events_attrs;
 		tsx_attr = icl_tsx_events_attrs;
 		x86_pmu.rtm_abort_event = X86_CONFIG(.event=0xca, .umask=0x02);
 		x86_pmu.lbr_pt_coexist = true;
 		intel_pmu_pebs_data_source_skl(pmem);
+		x86_pmu.update_topdown_event = icl_update_topdown_event;
+		x86_pmu.set_topdown_event_period = icl_set_topdown_event_period;
 		pr_cont("Icelake events, ");
 		name = "icelake";
 		break;
@@ -5198,6 +5518,15 @@ __init int intel_pmu_init(void)
 		 * counter, so do not extend mask to generic counters
 		 */
 		for_each_event_constraint(c, x86_pmu.event_constraints) {
+			/*
+			 * Don't extend the topdown slots and metrics
+			 * events to the generic counters.
+			 */
+			if (c->idxmsk64 & INTEL_PMC_MSK_TOPDOWN) {
+				c->weight = hweight64(c->idxmsk64);
+				continue;
+			}
+
 			if (c->cmask == FIXED_EVENT_FLAGS
 			    && c->idxmsk64 != INTEL_PMC_MSK_FIXED_REF_CYCLES) {
 				c->idxmsk64 |= (1ULL << x86_pmu.num_counters) - 1;
@@ -5253,6 +5582,9 @@ __init int intel_pmu_init(void)
 	if (x86_pmu.counter_freezing)
 		x86_pmu.handle_irq = intel_pmu_handle_irq_v4;
 
+	if (x86_pmu.intel_cap.perf_metrics)
+		x86_pmu.intel_ctrl |= 1ULL << GLOBAL_CTRL_EN_PERF_METRICS;
+
 	return 0;
 }
 

arch/x86/events/intel/ds.c

@@ -670,9 +670,7 @@ int intel_pmu_drain_bts_buffer(void)
 
 static inline void intel_pmu_drain_pebs_buffer(void)
 {
-	struct pt_regs regs;
-
-	x86_pmu.drain_pebs(&regs);
+	x86_pmu.drain_pebs(NULL);
 }
 
 /*
@@ -1737,6 +1735,7 @@ static void __intel_pmu_pebs_event(struct perf_event *event,
 	struct x86_perf_regs perf_regs;
 	struct pt_regs *regs = &perf_regs.regs;
 	void *at = get_next_pebs_record_by_bit(base, top, bit);
+	struct pt_regs dummy_iregs;
 
 	if (hwc->flags & PERF_X86_EVENT_AUTO_RELOAD) {
 		/*
@@ -1749,6 +1748,9 @@ static void __intel_pmu_pebs_event(struct perf_event *event,
 	} else if (!intel_pmu_save_and_restart(event))
 		return;
 
+	if (!iregs)
+		iregs = &dummy_iregs;
+
 	while (count > 1) {
 		setup_sample(event, iregs, at, &data, regs);
 		perf_event_output(event, &data, regs);
@@ -1758,16 +1760,22 @@ static void __intel_pmu_pebs_event(struct perf_event *event,
 	}
 
 	setup_sample(event, iregs, at, &data, regs);
-
-	/*
-	 * All but the last records are processed.
-	 * The last one is left to be able to call the overflow handler.
-	 */
-	if (perf_event_overflow(event, &data, regs)) {
-		x86_pmu_stop(event, 0);
-		return;
+	if (iregs == &dummy_iregs) {
+		/*
+		 * The PEBS records may be drained in the non-overflow context,
+		 * e.g., large PEBS + context switch. Perf should treat the
+		 * last record the same as other PEBS records, and doesn't
+		 * invoke the generic overflow handler.
+		 */
+		perf_event_output(event, &data, regs);
+	} else {
+		/*
+		 * All but the last records are processed.
+		 * The last one is left to be able to call the overflow handler.
+		 */
+		if (perf_event_overflow(event, &data, regs))
+			x86_pmu_stop(event, 0);
 	}
-
 }
 
 static void intel_pmu_drain_pebs_core(struct pt_regs *iregs)

arch/x86/events/intel/uncore.c

@@ -12,6 +12,8 @@ struct intel_uncore_type **uncore_mmio_uncores = empty_uncore;
 
 static bool pcidrv_registered;
 struct pci_driver *uncore_pci_driver;
+/* The PCI driver for the device which the uncore doesn't own. */
+struct pci_driver *uncore_pci_sub_driver;
 /* pci bus to socket mapping */
 DEFINE_RAW_SPINLOCK(pci2phy_map_lock);
 struct list_head pci2phy_map_head = LIST_HEAD_INIT(pci2phy_map_head);
@@ -989,65 +991,71 @@ uncore_types_init(struct intel_uncore_type **types, bool setid)
 }
 
 /*
- * add a pci uncore device
+ * Get the die information of a PCI device.
+ * @pdev: The PCI device.
+ * @phys_id: The physical socket id which the device maps to.
+ * @die: The die id which the device maps to.
  */
-static int uncore_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id)
+static int uncore_pci_get_dev_die_info(struct pci_dev *pdev,
+				       int *phys_id, int *die)
 {
-	struct intel_uncore_type *type;
-	struct intel_uncore_pmu *pmu = NULL;
-	struct intel_uncore_box *box;
-	int phys_id, die, ret;
-
-	phys_id = uncore_pcibus_to_physid(pdev->bus);
-	if (phys_id < 0)
+	*phys_id = uncore_pcibus_to_physid(pdev->bus);
+	if (*phys_id < 0)
 		return -ENODEV;
 
-	die = (topology_max_die_per_package() > 1) ? phys_id :
-					topology_phys_to_logical_pkg(phys_id);
-	if (die < 0)
+	*die = (topology_max_die_per_package() > 1) ? *phys_id :
+				topology_phys_to_logical_pkg(*phys_id);
+	if (*die < 0)
 		return -EINVAL;
 
-	if (UNCORE_PCI_DEV_TYPE(id->driver_data) == UNCORE_EXTRA_PCI_DEV) {
-		int idx = UNCORE_PCI_DEV_IDX(id->driver_data);
-
-		uncore_extra_pci_dev[die].dev[idx] = pdev;
-		pci_set_drvdata(pdev, NULL);
-		return 0;
-	}
-
-	type = uncore_pci_uncores[UNCORE_PCI_DEV_TYPE(id->driver_data)];
+	return 0;
+}
 
-	/*
-	 * Some platforms, e.g.  Knights Landing, use a common PCI device ID
-	 * for multiple instances of an uncore PMU device type. We should check
-	 * PCI slot and func to indicate the uncore box.
-	 */
-	if (id->driver_data & ~0xffff) {
-		struct pci_driver *pci_drv = pdev->driver;
-		const struct pci_device_id *ids = pci_drv->id_table;
-		unsigned int devfn;
-
-		while (ids && ids->vendor) {
-			if ((ids->vendor == pdev->vendor) &&
-			    (ids->device == pdev->device)) {
-				devfn = PCI_DEVFN(UNCORE_PCI_DEV_DEV(ids->driver_data),
-						  UNCORE_PCI_DEV_FUNC(ids->driver_data));
-				if (devfn == pdev->devfn) {
-					pmu = &type->pmus[UNCORE_PCI_DEV_IDX(ids->driver_data)];
-					break;
-				}
+/*
+ * Find the PMU of a PCI device.
+ * @pdev: The PCI device.
+ * @ids: The ID table of the available PCI devices with a PMU.
+ */
+static struct intel_uncore_pmu *
+uncore_pci_find_dev_pmu(struct pci_dev *pdev, const struct pci_device_id *ids)
+{
+	struct intel_uncore_pmu *pmu = NULL;
+	struct intel_uncore_type *type;
+	kernel_ulong_t data;
+	unsigned int devfn;
+
+	while (ids && ids->vendor) {
+		if ((ids->vendor == pdev->vendor) &&
+		    (ids->device == pdev->device)) {
+			data = ids->driver_data;
+			devfn = PCI_DEVFN(UNCORE_PCI_DEV_DEV(data),
+					  UNCORE_PCI_DEV_FUNC(data));
+			if (devfn == pdev->devfn) {
+				type = uncore_pci_uncores[UNCORE_PCI_DEV_TYPE(data)];
+				pmu = &type->pmus[UNCORE_PCI_DEV_IDX(data)];
+				break;
 			}
-			ids++;
 		}
-		if (pmu == NULL)
-			return -ENODEV;
-	} else {
-		/*
-		 * for performance monitoring unit with multiple boxes,
-		 * each box has a different function id.
-		 */
-		pmu = &type->pmus[UNCORE_PCI_DEV_IDX(id->driver_data)];
+		ids++;
 	}
+	return pmu;
+}
+
+/*
+ * Register the PMU for a PCI device
+ * @pdev: The PCI device.
+ * @type: The corresponding PMU type of the device.
+ * @pmu: The corresponding PMU of the device.
+ * @phys_id: The physical socket id which the device maps to.
+ * @die: The die id which the device maps to.
+ */
+static int uncore_pci_pmu_register(struct pci_dev *pdev,
+				   struct intel_uncore_type *type,
+				   struct intel_uncore_pmu *pmu,
+				   int phys_id, int die)
+{
+	struct intel_uncore_box *box;
+	int ret;
 
 	if (WARN_ON_ONCE(pmu->boxes[die] != NULL))
 		return -EINVAL;
@@ -1067,7 +1075,6 @@ static int uncore_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id
 	box->pci_dev = pdev;
 	box->pmu = pmu;
 	uncore_box_init(box);
-	pci_set_drvdata(pdev, box);
 
 	pmu->boxes[die] = box;
 	if (atomic_inc_return(&pmu->activeboxes) > 1)
@@ -1076,7 +1083,6 @@ static int uncore_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id
 	/* First active box registers the pmu */
 	ret = uncore_pmu_register(pmu);
 	if (ret) {
-		pci_set_drvdata(pdev, NULL);
 		pmu->boxes[die] = NULL;
 		uncore_box_exit(box);
 		kfree(box);
@@ -1084,18 +1090,87 @@ static int uncore_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id
 	return ret;
 }
 
+/*
+ * add a pci uncore device
+ */
+static int uncore_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id)
+{
+	struct intel_uncore_type *type;
+	struct intel_uncore_pmu *pmu = NULL;
+	int phys_id, die, ret;
+
+	ret = uncore_pci_get_dev_die_info(pdev, &phys_id, &die);
+	if (ret)
+		return ret;
+
+	if (UNCORE_PCI_DEV_TYPE(id->driver_data) == UNCORE_EXTRA_PCI_DEV) {
+		int idx = UNCORE_PCI_DEV_IDX(id->driver_data);
+
+		uncore_extra_pci_dev[die].dev[idx] = pdev;
+		pci_set_drvdata(pdev, NULL);
+		return 0;
+	}
+
+	type = uncore_pci_uncores[UNCORE_PCI_DEV_TYPE(id->driver_data)];
+
+	/*
+	 * Some platforms, e.g.  Knights Landing, use a common PCI device ID
+	 * for multiple instances of an uncore PMU device type. We should check
+	 * PCI slot and func to indicate the uncore box.
+	 */
+	if (id->driver_data & ~0xffff) {
+		struct pci_driver *pci_drv = pdev->driver;
+
+		pmu = uncore_pci_find_dev_pmu(pdev, pci_drv->id_table);
+		if (pmu == NULL)
+			return -ENODEV;
+	} else {
+		/*
+		 * for performance monitoring unit with multiple boxes,
+		 * each box has a different function id.
+		 */
+		pmu = &type->pmus[UNCORE_PCI_DEV_IDX(id->driver_data)];
+	}
+
+	ret = uncore_pci_pmu_register(pdev, type, pmu, phys_id, die);
+
+	pci_set_drvdata(pdev, pmu->boxes[die]);
+
+	return ret;
+}
+
+/*
+ * Unregister the PMU of a PCI device
+ * @pmu: The corresponding PMU is unregistered.
+ * @phys_id: The physical socket id which the device maps to.
+ * @die: The die id which the device maps to.
+ */
+static void uncore_pci_pmu_unregister(struct intel_uncore_pmu *pmu,
+				      int phys_id, int die)
+{
+	struct intel_uncore_box *box = pmu->boxes[die];
+
+	if (WARN_ON_ONCE(phys_id != box->pci_phys_id))
+		return;
+
+	pmu->boxes[die] = NULL;
+	if (atomic_dec_return(&pmu->activeboxes) == 0)
+		uncore_pmu_unregister(pmu);
+	uncore_box_exit(box);
+	kfree(box);
+}
+
 static void uncore_pci_remove(struct pci_dev *pdev)
 {
 	struct intel_uncore_box *box;
 	struct intel_uncore_pmu *pmu;
 	int i, phys_id, die;
 
-	phys_id = uncore_pcibus_to_physid(pdev->bus);
+	if (uncore_pci_get_dev_die_info(pdev, &phys_id, &die))
+		return;
 
 	box = pci_get_drvdata(pdev);
 	if (!box) {
-		die = (topology_max_die_per_package() > 1) ? phys_id :
-					topology_phys_to_logical_pkg(phys_id);
 		for (i = 0; i < UNCORE_EXTRA_PCI_DEV_MAX; i++) {
 			if (uncore_extra_pci_dev[die].dev[i] == pdev) {
 				uncore_extra_pci_dev[die].dev[i] = NULL;
@@ -1107,15 +1182,84 @@ static void uncore_pci_remove(struct pci_dev *pdev)
 	}
 
 	pmu = box->pmu;
-	if (WARN_ON_ONCE(phys_id != box->pci_phys_id))
-		return;
 
 	pci_set_drvdata(pdev, NULL);
-	pmu->boxes[box->dieid] = NULL;
-	if (atomic_dec_return(&pmu->activeboxes) == 0)
-		uncore_pmu_unregister(pmu);
-	uncore_box_exit(box);
-	kfree(box);
+
+	uncore_pci_pmu_unregister(pmu, phys_id, die);
+}
+
+static int uncore_bus_notify(struct notifier_block *nb,
+			     unsigned long action, void *data)
+{
+	struct device *dev = data;
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct intel_uncore_pmu *pmu;
+	int phys_id, die;
+
+	/* Unregister the PMU when the device is going to be deleted. */
+	if (action != BUS_NOTIFY_DEL_DEVICE)
+		return NOTIFY_DONE;
+
+	pmu = uncore_pci_find_dev_pmu(pdev, uncore_pci_sub_driver->id_table);
+	if (!pmu)
+		return NOTIFY_DONE;
+
+	if (uncore_pci_get_dev_die_info(pdev, &phys_id, &die))
+		return NOTIFY_DONE;
+
+	uncore_pci_pmu_unregister(pmu, phys_id, die);
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block uncore_notifier = {
+	.notifier_call = uncore_bus_notify,
+};
+
+static void uncore_pci_sub_driver_init(void)
+{
+	const struct pci_device_id *ids = uncore_pci_sub_driver->id_table;
+	struct intel_uncore_type *type;
+	struct intel_uncore_pmu *pmu;
+	struct pci_dev *pci_sub_dev;
+	bool notify = false;
+	unsigned int devfn;
+	int phys_id, die;
+
+	while (ids && ids->vendor) {
+		pci_sub_dev = NULL;
+		type = uncore_pci_uncores[UNCORE_PCI_DEV_TYPE(ids->driver_data)];
+		/*
+		 * Search the available device, and register the
+		 * corresponding PMU.
+		 */
+		while ((pci_sub_dev = pci_get_device(PCI_VENDOR_ID_INTEL,
+						     ids->device, pci_sub_dev))) {
+			devfn = PCI_DEVFN(UNCORE_PCI_DEV_DEV(ids->driver_data),
+					  UNCORE_PCI_DEV_FUNC(ids->driver_data));
+			if (devfn != pci_sub_dev->devfn)
+				continue;
+
+			pmu = &type->pmus[UNCORE_PCI_DEV_IDX(ids->driver_data)];
+			if (!pmu)
+				continue;
+
+			if (uncore_pci_get_dev_die_info(pci_sub_dev,
+							&phys_id, &die))
+				continue;
+
+			if (!uncore_pci_pmu_register(pci_sub_dev, type, pmu,
+						     phys_id, die))
+				notify = true;
+		}
+		ids++;
+	}
+
+	if (notify && bus_register_notifier(&pci_bus_type, &uncore_notifier))
+		notify = false;
+
+	if (!notify)
+		uncore_pci_sub_driver = NULL;
 }
 
 static int __init uncore_pci_init(void)
@@ -1141,6 +1285,9 @@ static int __init uncore_pci_init(void)
 	if (ret)
 		goto errtype;
 
+	if (uncore_pci_sub_driver)
+		uncore_pci_sub_driver_init();
+
 	pcidrv_registered = true;
 	return 0;
 
@@ -1158,6 +1305,8 @@ static void uncore_pci_exit(void)
 {
 	if (pcidrv_registered) {
 		pcidrv_registered = false;
+		if (uncore_pci_sub_driver)
+			bus_unregister_notifier(&pci_bus_type, &uncore_notifier);
 		pci_unregister_driver(uncore_pci_driver);
 		uncore_types_exit(uncore_pci_uncores);
 		kfree(uncore_extra_pci_dev);
@@ -1478,12 +1627,12 @@ static const struct intel_uncore_init_fun icl_uncore_init __initconst = {
 };
 
 static const struct intel_uncore_init_fun tgl_uncore_init __initconst = {
-	.cpu_init = icl_uncore_cpu_init,
+	.cpu_init = tgl_uncore_cpu_init,
 	.mmio_init = tgl_uncore_mmio_init,
 };
 
 static const struct intel_uncore_init_fun tgl_l_uncore_init __initconst = {
-	.cpu_init = icl_uncore_cpu_init,
+	.cpu_init = tgl_uncore_cpu_init,
 	.mmio_init = tgl_l_uncore_mmio_init,
 };
 

arch/x86/events/intel/uncore.h

@@ -552,6 +552,7 @@ extern struct intel_uncore_type **uncore_msr_uncores;
 extern struct intel_uncore_type **uncore_pci_uncores;
 extern struct intel_uncore_type **uncore_mmio_uncores;
 extern struct pci_driver *uncore_pci_driver;
+extern struct pci_driver *uncore_pci_sub_driver;
 extern raw_spinlock_t pci2phy_map_lock;
 extern struct list_head pci2phy_map_head;
 extern struct pci_extra_dev *uncore_extra_pci_dev;
@@ -567,6 +568,7 @@ void snb_uncore_cpu_init(void);
 void nhm_uncore_cpu_init(void);
 void skl_uncore_cpu_init(void);
 void icl_uncore_cpu_init(void);
+void tgl_uncore_cpu_init(void);
 void tgl_uncore_mmio_init(void);
 void tgl_l_uncore_mmio_init(void);
 int snb_pci2phy_map_init(int devid);

arch/x86/events/intel/uncore_snb.c

@@ -126,6 +126,10 @@
 #define ICL_UNC_CBO_0_PER_CTR0			0x702
 #define ICL_UNC_CBO_MSR_OFFSET			0x8
 
+/* ICL ARB register */
+#define ICL_UNC_ARB_PER_CTR			0x3b1
+#define ICL_UNC_ARB_PERFEVTSEL			0x3b3
+
 DEFINE_UNCORE_FORMAT_ATTR(event, event, "config:0-7");
 DEFINE_UNCORE_FORMAT_ATTR(umask, umask, "config:8-15");
 DEFINE_UNCORE_FORMAT_ATTR(edge, edge, "config:18");
@@ -313,15 +317,21 @@ void skl_uncore_cpu_init(void)
 	snb_uncore_arb.ops = &skl_uncore_msr_ops;
 }
 
+static struct intel_uncore_ops icl_uncore_msr_ops = {
+	.disable_event	= snb_uncore_msr_disable_event,
+	.enable_event	= snb_uncore_msr_enable_event,
+	.read_counter	= uncore_msr_read_counter,
+};
+
 static struct intel_uncore_type icl_uncore_cbox = {
 	.name		= "cbox",
-	.num_counters   = 4,
+	.num_counters   = 2,
 	.perf_ctr_bits	= 44,
 	.perf_ctr	= ICL_UNC_CBO_0_PER_CTR0,
 	.event_ctl	= SNB_UNC_CBO_0_PERFEVTSEL0,
 	.event_mask	= SNB_UNC_RAW_EVENT_MASK,
 	.msr_offset	= ICL_UNC_CBO_MSR_OFFSET,
-	.ops		= &skl_uncore_msr_ops,
+	.ops		= &icl_uncore_msr_ops,
 	.format_group	= &snb_uncore_format_group,
 };
 
@@ -350,13 +360,25 @@ static struct intel_uncore_type icl_uncore_clockbox = {
 	.single_fixed	= 1,
 	.event_mask	= SNB_UNC_CTL_EV_SEL_MASK,
 	.format_group	= &icl_uncore_clock_format_group,
-	.ops		= &skl_uncore_msr_ops,
+	.ops		= &icl_uncore_msr_ops,
 	.event_descs	= icl_uncore_events,
 };
 
+static struct intel_uncore_type icl_uncore_arb = {
+	.name		= "arb",
+	.num_counters   = 1,
+	.num_boxes	= 1,
+	.perf_ctr_bits	= 44,
+	.perf_ctr	= ICL_UNC_ARB_PER_CTR,
+	.event_ctl	= ICL_UNC_ARB_PERFEVTSEL,
+	.event_mask	= SNB_UNC_RAW_EVENT_MASK,
+	.ops		= &icl_uncore_msr_ops,
+	.format_group	= &snb_uncore_format_group,
+};
+
 static struct intel_uncore_type *icl_msr_uncores[] = {
 	&icl_uncore_cbox,
-	&snb_uncore_arb,
+	&icl_uncore_arb,
 	&icl_uncore_clockbox,
 	NULL,
 };
@@ -374,6 +396,21 @@ void icl_uncore_cpu_init(void)
 {
 	uncore_msr_uncores = icl_msr_uncores;
 	icl_uncore_cbox.num_boxes = icl_get_cbox_num();
+}
+
+static struct intel_uncore_type *tgl_msr_uncores[] = {
+	&icl_uncore_cbox,
+	&snb_uncore_arb,
+	&icl_uncore_clockbox,
+	NULL,
+};
+
+void tgl_uncore_cpu_init(void)
+{
+	uncore_msr_uncores = tgl_msr_uncores;
+	icl_uncore_cbox.num_boxes = icl_get_cbox_num();
+	icl_uncore_cbox.ops = &skl_uncore_msr_ops;
+	icl_uncore_clockbox.ops = &skl_uncore_msr_ops;
 	snb_uncore_arb.ops = &skl_uncore_msr_ops;
 }
 

arch/x86/events/intel/uncore_snbep.c

@@ -393,6 +393,11 @@
 #define SNR_M2M_PCI_PMON_BOX_CTL		0x438
 #define SNR_M2M_PCI_PMON_UMASK_EXT		0xff
 
+/* SNR PCIE3 */
+#define SNR_PCIE3_PCI_PMON_CTL0			0x508
+#define SNR_PCIE3_PCI_PMON_CTR0			0x4e8
+#define SNR_PCIE3_PCI_PMON_BOX_CTL		0x4e0
+
 /* SNR IMC */
 #define SNR_IMC_MMIO_PMON_FIXED_CTL		0x54
 #define SNR_IMC_MMIO_PMON_FIXED_CTR		0x38
@@ -3749,7 +3754,9 @@ static int skx_iio_set_mapping(struct intel_uncore_type *type)
 
 	ret = skx_iio_get_topology(type);
 	if (ret)
-		return ret;
+		goto clear_attr_update;
+
+	ret = -ENOMEM;
 
 	/* One more for NULL. */
 	attrs = kcalloc((uncore_max_dies() + 1), sizeof(*attrs), GFP_KERNEL);
@@ -3781,8 +3788,9 @@ static int skx_iio_set_mapping(struct intel_uncore_type *type)
 	kfree(eas);
 	kfree(attrs);
 	kfree(type->topology);
+clear_attr_update:
 	type->attr_update = NULL;
-	return -ENOMEM;
+	return ret;
 }
 
 static void skx_iio_cleanup_mapping(struct intel_uncore_type *type)
@@ -4551,12 +4559,46 @@ static struct intel_uncore_type snr_uncore_m2m = {
 	.format_group	= &snr_m2m_uncore_format_group,
 };
 
+static void snr_uncore_pci_enable_event(struct intel_uncore_box *box, struct perf_event *event)
+{
+	struct pci_dev *pdev = box->pci_dev;
+	struct hw_perf_event *hwc = &event->hw;
+
+	pci_write_config_dword(pdev, hwc->config_base, (u32)(hwc->config | SNBEP_PMON_CTL_EN));
+	pci_write_config_dword(pdev, hwc->config_base + 4, (u32)(hwc->config >> 32));
+}
+
+static struct intel_uncore_ops snr_pcie3_uncore_pci_ops = {
+	.init_box	= snr_m2m_uncore_pci_init_box,
+	.disable_box	= snbep_uncore_pci_disable_box,
+	.enable_box	= snbep_uncore_pci_enable_box,
+	.disable_event	= snbep_uncore_pci_disable_event,
+	.enable_event	= snr_uncore_pci_enable_event,
+	.read_counter	= snbep_uncore_pci_read_counter,
+};
+
+static struct intel_uncore_type snr_uncore_pcie3 = {
+	.name		= "pcie3",
+	.num_counters	= 4,
+	.num_boxes	= 1,
+	.perf_ctr_bits	= 48,
+	.perf_ctr	= SNR_PCIE3_PCI_PMON_CTR0,
+	.event_ctl	= SNR_PCIE3_PCI_PMON_CTL0,
+	.event_mask	= SKX_IIO_PMON_RAW_EVENT_MASK,
+	.event_mask_ext	= SKX_IIO_PMON_RAW_EVENT_MASK_EXT,
+	.box_ctl	= SNR_PCIE3_PCI_PMON_BOX_CTL,
+	.ops		= &snr_pcie3_uncore_pci_ops,
+	.format_group	= &skx_uncore_iio_format_group,
+};
+
 enum {
 	SNR_PCI_UNCORE_M2M,
+	SNR_PCI_UNCORE_PCIE3,
 };
 
 static struct intel_uncore_type *snr_pci_uncores[] = {
 	[SNR_PCI_UNCORE_M2M]		= &snr_uncore_m2m,
+	[SNR_PCI_UNCORE_PCIE3]		= &snr_uncore_pcie3,
 	NULL,
 };
 
@@ -4573,6 +4615,19 @@ static struct pci_driver snr_uncore_pci_driver = {
 	.id_table	= snr_uncore_pci_ids,
 };
 
+static const struct pci_device_id snr_uncore_pci_sub_ids[] = {
+	{ /* PCIe3 RP */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x334a),
+		.driver_data = UNCORE_PCI_DEV_FULL_DATA(4, 0, SNR_PCI_UNCORE_PCIE3, 0),
+	},
+	{ /* end: all zeroes */ }
+};
+
+static struct pci_driver snr_uncore_pci_sub_driver = {
+	.name		= "snr_uncore_sub",
+	.id_table	= snr_uncore_pci_sub_ids,
+};
+
 int snr_uncore_pci_init(void)
 {
 	/* SNR UBOX DID */
@@ -4584,6 +4639,7 @@ int snr_uncore_pci_init(void)
 
 	uncore_pci_uncores = snr_pci_uncores;
 	uncore_pci_driver = &snr_uncore_pci_driver;
+	uncore_pci_sub_driver = &snr_uncore_pci_sub_driver;
 	return 0;
 }
 
@@ -4751,10 +4807,10 @@ static struct uncore_event_desc snr_uncore_imc_freerunning_events[] = {
 	INTEL_UNCORE_EVENT_DESC(dclk,		"event=0xff,umask=0x10"),
 
 	INTEL_UNCORE_EVENT_DESC(read,		"event=0xff,umask=0x20"),
-	INTEL_UNCORE_EVENT_DESC(read.scale,	"3.814697266e-6"),
+	INTEL_UNCORE_EVENT_DESC(read.scale,	"6.103515625e-5"),
 	INTEL_UNCORE_EVENT_DESC(read.unit,	"MiB"),
 	INTEL_UNCORE_EVENT_DESC(write,		"event=0xff,umask=0x21"),
-	INTEL_UNCORE_EVENT_DESC(write.scale,	"3.814697266e-6"),
+	INTEL_UNCORE_EVENT_DESC(write.scale,	"6.103515625e-5"),
 	INTEL_UNCORE_EVENT_DESC(write.unit,	"MiB"),
 	{ /* end: all zeroes */ },
 };
@@ -5212,17 +5268,17 @@ static struct uncore_event_desc icx_uncore_imc_freerunning_events[] = {
 	INTEL_UNCORE_EVENT_DESC(dclk,			"event=0xff,umask=0x10"),
 
 	INTEL_UNCORE_EVENT_DESC(read,			"event=0xff,umask=0x20"),
-	INTEL_UNCORE_EVENT_DESC(read.scale,		"3.814697266e-6"),
+	INTEL_UNCORE_EVENT_DESC(read.scale,		"6.103515625e-5"),
 	INTEL_UNCORE_EVENT_DESC(read.unit,		"MiB"),
 	INTEL_UNCORE_EVENT_DESC(write,			"event=0xff,umask=0x21"),
-	INTEL_UNCORE_EVENT_DESC(write.scale,		"3.814697266e-6"),
+	INTEL_UNCORE_EVENT_DESC(write.scale,		"6.103515625e-5"),
 	INTEL_UNCORE_EVENT_DESC(write.unit,		"MiB"),
 
 	INTEL_UNCORE_EVENT_DESC(ddrt_read,		"event=0xff,umask=0x30"),
-	INTEL_UNCORE_EVENT_DESC(ddrt_read.scale,	"3.814697266e-6"),
+	INTEL_UNCORE_EVENT_DESC(ddrt_read.scale,	"6.103515625e-5"),
 	INTEL_UNCORE_EVENT_DESC(ddrt_read.unit,		"MiB"),
 	INTEL_UNCORE_EVENT_DESC(ddrt_write,		"event=0xff,umask=0x31"),
-	INTEL_UNCORE_EVENT_DESC(ddrt_write.scale,	"3.814697266e-6"),
+	INTEL_UNCORE_EVENT_DESC(ddrt_write.scale,	"6.103515625e-5"),
 	INTEL_UNCORE_EVENT_DESC(ddrt_write.unit,	"MiB"),
 	{ /* end: all zeroes */ },
 };

arch/x86/events/msr.c

@@ -78,6 +78,7 @@ static bool test_intel(int idx, void *data)
 	case INTEL_FAM6_ATOM_GOLDMONT_PLUS:
 	case INTEL_FAM6_ATOM_TREMONT_D:
 	case INTEL_FAM6_ATOM_TREMONT:
+	case INTEL_FAM6_ATOM_TREMONT_L:
 
 	case INTEL_FAM6_XEON_PHI_KNL:
 	case INTEL_FAM6_XEON_PHI_KNM:

arch/x86/events/perf_event.h

@@ -79,6 +79,31 @@ static inline bool constraint_match(struct event_constraint *c, u64 ecode)
 #define PERF_X86_EVENT_PEBS_VIA_PT	0x0800 /* use PT buffer for PEBS */
 #define PERF_X86_EVENT_PAIR		0x1000 /* Large Increment per Cycle */
 #define PERF_X86_EVENT_LBR_SELECT	0x2000 /* Save/Restore MSR_LBR_SELECT */
+#define PERF_X86_EVENT_TOPDOWN		0x4000 /* Count Topdown slots/metrics events */
+
+static inline bool is_topdown_count(struct perf_event *event)
+{
+	return event->hw.flags & PERF_X86_EVENT_TOPDOWN;
+}
+
+static inline bool is_metric_event(struct perf_event *event)
+{
+	u64 config = event->attr.config;
+
+	return ((config & ARCH_PERFMON_EVENTSEL_EVENT) == 0) &&
+		((config & INTEL_ARCH_EVENT_MASK) >= INTEL_TD_METRIC_RETIRING)  &&
+		((config & INTEL_ARCH_EVENT_MASK) <= INTEL_TD_METRIC_MAX);
+}
+
+static inline bool is_slots_event(struct perf_event *event)
+{
+	return (event->attr.config & INTEL_ARCH_EVENT_MASK) == INTEL_TD_SLOTS;
+}
+
+static inline bool is_topdown_event(struct perf_event *event)
+{
+	return is_metric_event(event) || is_slots_event(event);
+}
 
 struct amd_nb {
 	int nb_id;  /* NorthBridge id */
@@ -210,6 +235,8 @@ struct cpu_hw_events {
 					     they've never been enabled yet */
 	int			n_txn;    /* the # last events in the below arrays;
 					     added in the current transaction */
+	int			n_txn_pair;
+	int			n_txn_metric;
 	int			assign[X86_PMC_IDX_MAX]; /* event to counter assignment */
 	u64			tags[X86_PMC_IDX_MAX];
 
@@ -284,6 +311,12 @@ struct cpu_hw_events {
 	 */
 	u64				tfa_shadow;
 
+	/*
+	 * Perf Metrics
+	 */
+	/* number of accepted metrics events */
+	int				n_metric;
+
 	/*
 	 * AMD specific bits
 	 */
@@ -375,6 +408,19 @@ struct cpu_hw_events {
 #define FIXED_EVENT_CONSTRAINT(c, n)	\
 	EVENT_CONSTRAINT(c, (1ULL << (32+n)), FIXED_EVENT_FLAGS)
 
+/*
+ * The special metric counters do not actually exist. They are calculated from
+ * the combination of the FxCtr3 + MSR_PERF_METRICS.
+ *
+ * The special metric counters are mapped to a dummy offset for the scheduler.
+ * The sharing between multiple users of the same metric without multiplexing
+ * is not allowed, even though the hardware supports that in principle.
+ */
+
+#define METRIC_EVENT_CONSTRAINT(c, n)					\
+	EVENT_CONSTRAINT(c, (1ULL << (INTEL_PMC_IDX_METRIC_BASE + n)),	\
+			 INTEL_ARCH_EVENT_MASK)
+
 /*
  * Constraint on the Event code + UMask
  */
@@ -537,7 +583,7 @@ union perf_capabilities {
 		 */
 		u64	full_width_write:1;
 		u64     pebs_baseline:1;
-		u64	pebs_metrics_available:1;
+		u64	perf_metrics:1;
 		u64	pebs_output_pt_available:1;
 	};
 	u64	capabilities;
@@ -726,6 +772,12 @@ struct x86_pmu {
 	 */
 	atomic_t	lbr_exclusive[x86_lbr_exclusive_max];
 
+	/*
+	 * Intel perf metrics
+	 */
+	u64		(*update_topdown_event)(struct perf_event *event);
+	int		(*set_topdown_event_period)(struct perf_event *event);
+
 	/*
 	 * perf task context (i.e. struct perf_event_context::task_ctx_data)
 	 * switch helper to bridge calls from perf/core to perf/x86.

arch/x86/events/rapl.c

@@ -815,6 +815,7 @@ static const struct x86_cpu_id rapl_model_match[] __initconst = {
 	X86_MATCH_INTEL_FAM6_MODEL(SAPPHIRERAPIDS_X,	&model_spr),
 	X86_MATCH_VENDOR_FAM(AMD,	0x17,		&model_amd_fam17h),
 	X86_MATCH_VENDOR_FAM(HYGON,	0x18,		&model_amd_fam17h),
+	X86_MATCH_VENDOR_FAM(AMD,	0x19,		&model_amd_fam17h),
 	{},
 };
 MODULE_DEVICE_TABLE(x86cpu, rapl_model_match);

arch/x86/include/asm/msr-index.h

@@ -467,6 +467,7 @@
 #define MSR_AMD64_IBSOP_REG_MASK	((1UL<<MSR_AMD64_IBSOP_REG_COUNT)-1)
 #define MSR_AMD64_IBSCTL		0xc001103a
 #define MSR_AMD64_IBSBRTARGET		0xc001103b
+#define MSR_AMD64_ICIBSEXTDCTL		0xc001103c
 #define MSR_AMD64_IBSOPDATA4		0xc001103d
 #define MSR_AMD64_IBS_REG_COUNT_MAX	8 /* includes MSR_AMD64_IBSBRTARGET */
 #define MSR_AMD64_SEV			0xc0010131
@@ -860,11 +861,14 @@
 #define MSR_CORE_PERF_FIXED_CTR0	0x00000309
 #define MSR_CORE_PERF_FIXED_CTR1	0x0000030a
 #define MSR_CORE_PERF_FIXED_CTR2	0x0000030b
+#define MSR_CORE_PERF_FIXED_CTR3	0x0000030c
 #define MSR_CORE_PERF_FIXED_CTR_CTRL	0x0000038d
 #define MSR_CORE_PERF_GLOBAL_STATUS	0x0000038e
 #define MSR_CORE_PERF_GLOBAL_CTRL	0x0000038f
 #define MSR_CORE_PERF_GLOBAL_OVF_CTRL	0x00000390
 
+#define MSR_PERF_METRICS		0x00000329
+
 /* PERF_GLOBAL_OVF_CTL bits */
 #define MSR_CORE_PERF_GLOBAL_OVF_CTRL_TRACE_TOPA_PMI_BIT	55
 #define MSR_CORE_PERF_GLOBAL_OVF_CTRL_TRACE_TOPA_PMI		(1ULL << MSR_CORE_PERF_GLOBAL_OVF_CTRL_TRACE_TOPA_PMI_BIT)

arch/x86/include/asm/perf_event.h

@@ -196,13 +196,29 @@ struct x86_pmu_capability {
  * Fixed-purpose performance events:
  */
 
+/* RDPMC offset for Fixed PMCs */
+#define INTEL_PMC_FIXED_RDPMC_BASE		(1 << 30)
+#define INTEL_PMC_FIXED_RDPMC_METRICS		(1 << 29)
+
 /*
- * All 3 fixed-mode PMCs are configured via this single MSR:
+ * All the fixed-mode PMCs are configured via this single MSR:
  */
 #define MSR_ARCH_PERFMON_FIXED_CTR_CTRL	0x38d
 
 /*
- * The counts are available in three separate MSRs:
+ * There is no event-code assigned to the fixed-mode PMCs.
+ *
+ * For a fixed-mode PMC, which has an equivalent event on a general-purpose
+ * PMC, the event-code of the equivalent event is used for the fixed-mode PMC,
+ * e.g., Instr_Retired.Any and CPU_CLK_Unhalted.Core.
+ *
+ * For a fixed-mode PMC, which doesn't have an equivalent event, a
+ * pseudo-encoding is used, e.g., CPU_CLK_Unhalted.Ref and TOPDOWN.SLOTS.
+ * The pseudo event-code for a fixed-mode PMC must be 0x00.
+ * The pseudo umask-code is 0xX. The X equals the index of the fixed
+ * counter + 1, e.g., the fixed counter 2 has the pseudo-encoding 0x0300.
+ *
+ * The counts are available in separate MSRs:
  */
 
 /* Instr_Retired.Any: */
@@ -213,29 +229,84 @@ struct x86_pmu_capability {
 #define MSR_ARCH_PERFMON_FIXED_CTR1	0x30a
 #define INTEL_PMC_IDX_FIXED_CPU_CYCLES	(INTEL_PMC_IDX_FIXED + 1)
 
-/* CPU_CLK_Unhalted.Ref: */
+/* CPU_CLK_Unhalted.Ref: event=0x00,umask=0x3 (pseudo-encoding) */
 #define MSR_ARCH_PERFMON_FIXED_CTR2	0x30b
 #define INTEL_PMC_IDX_FIXED_REF_CYCLES	(INTEL_PMC_IDX_FIXED + 2)
 #define INTEL_PMC_MSK_FIXED_REF_CYCLES	(1ULL << INTEL_PMC_IDX_FIXED_REF_CYCLES)
 
+/* TOPDOWN.SLOTS: event=0x00,umask=0x4 (pseudo-encoding) */
+#define MSR_ARCH_PERFMON_FIXED_CTR3	0x30c
+#define INTEL_PMC_IDX_FIXED_SLOTS	(INTEL_PMC_IDX_FIXED + 3)
+#define INTEL_PMC_MSK_FIXED_SLOTS	(1ULL << INTEL_PMC_IDX_FIXED_SLOTS)
+
 /*
  * We model BTS tracing as another fixed-mode PMC.
  *
- * We choose a value in the middle of the fixed event range, since lower
+ * We choose the value 47 for the fixed index of BTS, since lower
  * values are used by actual fixed events and higher values are used
  * to indicate other overflow conditions in the PERF_GLOBAL_STATUS msr.
  */
-#define INTEL_PMC_IDX_FIXED_BTS				(INTEL_PMC_IDX_FIXED + 16)
+#define INTEL_PMC_IDX_FIXED_BTS			(INTEL_PMC_IDX_FIXED + 15)
+
+/*
+ * The PERF_METRICS MSR is modeled as several magic fixed-mode PMCs, one for
+ * each TopDown metric event.
+ *
+ * Internally the TopDown metric events are mapped to the FxCtr 3 (SLOTS).
+ */
+#define INTEL_PMC_IDX_METRIC_BASE		(INTEL_PMC_IDX_FIXED + 16)
+#define INTEL_PMC_IDX_TD_RETIRING		(INTEL_PMC_IDX_METRIC_BASE + 0)
+#define INTEL_PMC_IDX_TD_BAD_SPEC		(INTEL_PMC_IDX_METRIC_BASE + 1)
+#define INTEL_PMC_IDX_TD_FE_BOUND		(INTEL_PMC_IDX_METRIC_BASE + 2)
+#define INTEL_PMC_IDX_TD_BE_BOUND		(INTEL_PMC_IDX_METRIC_BASE + 3)
+#define INTEL_PMC_IDX_METRIC_END		INTEL_PMC_IDX_TD_BE_BOUND
+#define INTEL_PMC_MSK_TOPDOWN			((0xfull << INTEL_PMC_IDX_METRIC_BASE) | \
+						INTEL_PMC_MSK_FIXED_SLOTS)
 
-#define GLOBAL_STATUS_COND_CHG				BIT_ULL(63)
-#define GLOBAL_STATUS_BUFFER_OVF			BIT_ULL(62)
-#define GLOBAL_STATUS_UNC_OVF				BIT_ULL(61)
-#define GLOBAL_STATUS_ASIF				BIT_ULL(60)
-#define GLOBAL_STATUS_COUNTERS_FROZEN			BIT_ULL(59)
-#define GLOBAL_STATUS_LBRS_FROZEN_BIT			58
-#define GLOBAL_STATUS_LBRS_FROZEN			BIT_ULL(GLOBAL_STATUS_LBRS_FROZEN_BIT)
-#define GLOBAL_STATUS_TRACE_TOPAPMI			BIT_ULL(55)
+/*
+ * There is no event-code assigned to the TopDown events.
+ *
+ * For the slots event, use the pseudo code of the fixed counter 3.
+ *
+ * For the metric events, the pseudo event-code is 0x00.
+ * The pseudo umask-code starts from the middle of the pseudo event
+ * space, 0x80.
+ */
+#define INTEL_TD_SLOTS				0x0400	/* TOPDOWN.SLOTS */
+/* Level 1 metrics */
+#define INTEL_TD_METRIC_RETIRING		0x8000	/* Retiring metric */
+#define INTEL_TD_METRIC_BAD_SPEC		0x8100	/* Bad speculation metric */
+#define INTEL_TD_METRIC_FE_BOUND		0x8200	/* FE bound metric */
+#define INTEL_TD_METRIC_BE_BOUND		0x8300	/* BE bound metric */
+#define INTEL_TD_METRIC_MAX			INTEL_TD_METRIC_BE_BOUND
+#define INTEL_TD_METRIC_NUM			4
+
+static inline bool is_metric_idx(int idx)
+{
+	return (unsigned)(idx - INTEL_PMC_IDX_METRIC_BASE) < INTEL_TD_METRIC_NUM;
+}
+
+static inline bool is_topdown_idx(int idx)
+{
+	return is_metric_idx(idx) || idx == INTEL_PMC_IDX_FIXED_SLOTS;
+}
 
+#define INTEL_PMC_OTHER_TOPDOWN_BITS(bit)	\
+			(~(0x1ull << bit) & INTEL_PMC_MSK_TOPDOWN)
+
+#define GLOBAL_STATUS_COND_CHG			BIT_ULL(63)
+#define GLOBAL_STATUS_BUFFER_OVF_BIT		62
+#define GLOBAL_STATUS_BUFFER_OVF		BIT_ULL(GLOBAL_STATUS_BUFFER_OVF_BIT)
+#define GLOBAL_STATUS_UNC_OVF			BIT_ULL(61)
+#define GLOBAL_STATUS_ASIF			BIT_ULL(60)
+#define GLOBAL_STATUS_COUNTERS_FROZEN		BIT_ULL(59)
+#define GLOBAL_STATUS_LBRS_FROZEN_BIT		58
+#define GLOBAL_STATUS_LBRS_FROZEN		BIT_ULL(GLOBAL_STATUS_LBRS_FROZEN_BIT)
+#define GLOBAL_STATUS_TRACE_TOPAPMI_BIT		55
+#define GLOBAL_STATUS_TRACE_TOPAPMI		BIT_ULL(GLOBAL_STATUS_TRACE_TOPAPMI_BIT)
+#define GLOBAL_STATUS_PERF_METRICS_OVF_BIT	48
+
+#define GLOBAL_CTRL_EN_PERF_METRICS		48
 /*
  * We model guest LBR event tracing as another fixed-mode PMC like BTS.
  *
@@ -334,6 +405,7 @@ struct pebs_xmm {
 #define IBS_OP_ENABLE		(1ULL<<17)
 #define IBS_OP_MAX_CNT		0x0000FFFFULL
 #define IBS_OP_MAX_CNT_EXT	0x007FFFFFULL	/* not a register bit mask */
+#define IBS_OP_MAX_CNT_EXT_MASK	(0x7FULL<<20)	/* separate upper 7 bits */
 #define IBS_RIP_INVALID		(1ULL<<38)
 
 #ifdef CONFIG_X86_LOCAL_APIC

include/linux/perf_event.h

@@ -212,17 +212,26 @@ struct hw_perf_event {
 	 */
 	u64				sample_period;
 
-	/*
-	 * The period we started this sample with.
-	 */
-	u64				last_period;
+	union {
+		struct { /* Sampling */
+			/*
+			 * The period we started this sample with.
+			 */
+			u64				last_period;
 
-	/*
-	 * However much is left of the current period; note that this is
-	 * a full 64bit value and allows for generation of periods longer
-	 * than hardware might allow.
-	 */
-	local64_t			period_left;
+			/*
+			 * However much is left of the current period;
+			 * note that this is a full 64bit value and
+			 * allows for generation of periods longer
+			 * than hardware might allow.
+			 */
+			local64_t			period_left;
+		};
+		struct { /* Topdown events counting for context switch */
+			u64				saved_metric;
+			u64				saved_slots;
+		};
+	};
 
 	/*
 	 * State for throttling the event, see __perf_event_overflow() and
@@ -576,9 +585,13 @@ typedef void (*perf_overflow_handler_t)(struct perf_event *,
  * PERF_EV_CAP_SOFTWARE: Is a software event.
  * PERF_EV_CAP_READ_ACTIVE_PKG: A CPU event (or cgroup event) that can be read
  * from any CPU in the package where it is active.
+ * PERF_EV_CAP_SIBLING: An event with this flag must be a group sibling and
+ * cannot be a group leader. If an event with this flag is detached from the
+ * group it is scheduled out and moved into an unrecoverable ERROR state.
  */
 #define PERF_EV_CAP_SOFTWARE		BIT(0)
 #define PERF_EV_CAP_READ_ACTIVE_PKG	BIT(1)
+#define PERF_EV_CAP_SIBLING		BIT(2)
 
 #define SWEVENT_HLIST_BITS		8
 #define SWEVENT_HLIST_SIZE		(1 << SWEVENT_HLIST_BITS)
@@ -859,7 +872,6 @@ struct perf_cpu_context {
 	struct list_head		cgrp_cpuctx_entry;
 #endif
 
-	struct list_head		sched_cb_entry;
 	int				sched_cb_usage;
 
 	int				online;

kernel/events/core.c

@@ -383,7 +383,6 @@ static DEFINE_MUTEX(perf_sched_mutex);
 static atomic_t perf_sched_count;
 
 static DEFINE_PER_CPU(atomic_t, perf_cgroup_events);
-static DEFINE_PER_CPU(int, perf_sched_cb_usages);
 static DEFINE_PER_CPU(struct pmu_event_list, pmu_sb_events);
 
 static atomic_t nr_mmap_events __read_mostly;
@@ -2134,8 +2133,24 @@ static inline struct list_head *get_event_list(struct perf_event *event)
 	return event->attr.pinned ? &ctx->pinned_active : &ctx->flexible_active;
 }
 
+/*
+ * Events that have PERF_EV_CAP_SIBLING require being part of a group and
+ * cannot exist on their own, schedule them out and move them into the ERROR
+ * state. Also see _perf_event_enable(), it will not be able to recover
+ * this ERROR state.
+ */
+static inline void perf_remove_sibling_event(struct perf_event *event)
+{
+	struct perf_event_context *ctx = event->ctx;
+	struct perf_cpu_context *cpuctx = __get_cpu_context(ctx);
+
+	event_sched_out(event, cpuctx, ctx);
+	perf_event_set_state(event, PERF_EVENT_STATE_ERROR);
+}
+
 static void perf_group_detach(struct perf_event *event)
 {
+	struct perf_event *leader = event->group_leader;
 	struct perf_event *sibling, *tmp;
 	struct perf_event_context *ctx = event->ctx;
 
@@ -2154,7 +2169,7 @@ static void perf_group_detach(struct perf_event *event)
 	/*
 	 * If this is a sibling, remove it from its group.
 	 */
-	if (event->group_leader != event) {
+	if (leader != event) {
 		list_del_init(&event->sibling_list);
 		event->group_leader->nr_siblings--;
 		goto out;
@@ -2167,6 +2182,9 @@ static void perf_group_detach(struct perf_event *event)
 	 */
 	list_for_each_entry_safe(sibling, tmp, &event->sibling_list, sibling_list) {
 
+		if (sibling->event_caps & PERF_EV_CAP_SIBLING)
+			perf_remove_sibling_event(sibling);
+
 		sibling->group_leader = sibling;
 		list_del_init(&sibling->sibling_list);
 
@@ -2184,10 +2202,10 @@ static void perf_group_detach(struct perf_event *event)
 	}
 
 out:
-	perf_event__header_size(event->group_leader);
-
-	for_each_sibling_event(tmp, event->group_leader)
+	for_each_sibling_event(tmp, leader)
 		perf_event__header_size(tmp);
+
+	perf_event__header_size(leader);
 }
 
 static bool is_orphaned_event(struct perf_event *event)
@@ -2980,6 +2998,7 @@ static void _perf_event_enable(struct perf_event *event)
 	raw_spin_lock_irq(&ctx->lock);
 	if (event->state >= PERF_EVENT_STATE_INACTIVE ||
 	    event->state <  PERF_EVENT_STATE_ERROR) {
+out:
 		raw_spin_unlock_irq(&ctx->lock);
 		return;
 	}
@@ -2991,8 +3010,16 @@ static void _perf_event_enable(struct perf_event *event)
 	 * has gone back into error state, as distinct from the task having
 	 * been scheduled away before the cross-call arrived.
 	 */
-	if (event->state == PERF_EVENT_STATE_ERROR)
+	if (event->state == PERF_EVENT_STATE_ERROR) {
+		/*
+		 * Detached SIBLING events cannot leave ERROR state.
+		 */
+		if (event->event_caps & PERF_EV_CAP_SIBLING &&
+		    event->group_leader == event)
+			goto out;
+
 		event->state = PERF_EVENT_STATE_OFF;
+	}
 	raw_spin_unlock_irq(&ctx->lock);
 
 	event_function_call(event, __perf_event_enable, NULL);
@@ -3357,10 +3384,12 @@ static void perf_event_context_sched_out(struct task_struct *task, int ctxn,
 	struct perf_event_context *parent, *next_parent;
 	struct perf_cpu_context *cpuctx;
 	int do_switch = 1;
+	struct pmu *pmu;
 
 	if (likely(!ctx))
 		return;
 
+	pmu = ctx->pmu;
 	cpuctx = __get_cpu_context(ctx);
 	if (!cpuctx->task_ctx)
 		return;
@@ -3390,11 +3419,15 @@ static void perf_event_context_sched_out(struct task_struct *task, int ctxn,
 		raw_spin_lock(&ctx->lock);
 		raw_spin_lock_nested(&next_ctx->lock, SINGLE_DEPTH_NESTING);
 		if (context_equiv(ctx, next_ctx)) {
-			struct pmu *pmu = ctx->pmu;
 
 			WRITE_ONCE(ctx->task, next);
 			WRITE_ONCE(next_ctx->task, task);
 
+			perf_pmu_disable(pmu);
+
+			if (cpuctx->sched_cb_usage && pmu->sched_task)
+				pmu->sched_task(ctx, false);
+
 			/*
 			 * PMU specific parts of task perf context can require
 			 * additional synchronization. As an example of such
@@ -3406,6 +3439,8 @@ static void perf_event_context_sched_out(struct task_struct *task, int ctxn,
 			else
 				swap(ctx->task_ctx_data, next_ctx->task_ctx_data);
 
+			perf_pmu_enable(pmu);
+
 			/*
 			 * RCU_INIT_POINTER here is safe because we've not
 			 * modified the ctx and the above modification of
@@ -3428,21 +3463,22 @@ static void perf_event_context_sched_out(struct task_struct *task, int ctxn,
 
 	if (do_switch) {
 		raw_spin_lock(&ctx->lock);
+		perf_pmu_disable(pmu);
+
+		if (cpuctx->sched_cb_usage && pmu->sched_task)
+			pmu->sched_task(ctx, false);
 		task_ctx_sched_out(cpuctx, ctx, EVENT_ALL);
+
+		perf_pmu_enable(pmu);
 		raw_spin_unlock(&ctx->lock);
 	}
 }
 
-static DEFINE_PER_CPU(struct list_head, sched_cb_list);
-
 void perf_sched_cb_dec(struct pmu *pmu)
 {
 	struct perf_cpu_context *cpuctx = this_cpu_ptr(pmu->pmu_cpu_context);
 
-	this_cpu_dec(perf_sched_cb_usages);
-
-	if (!--cpuctx->sched_cb_usage)
-		list_del(&cpuctx->sched_cb_entry);
+	--cpuctx->sched_cb_usage;
 }
 
 
@@ -3450,10 +3486,7 @@ void perf_sched_cb_inc(struct pmu *pmu)
 {
 	struct perf_cpu_context *cpuctx = this_cpu_ptr(pmu->pmu_cpu_context);
 
-	if (!cpuctx->sched_cb_usage++)
-		list_add(&cpuctx->sched_cb_entry, this_cpu_ptr(&sched_cb_list));
-
-	this_cpu_inc(perf_sched_cb_usages);
+	cpuctx->sched_cb_usage++;
 }
 
 /*
@@ -3464,30 +3497,22 @@ void perf_sched_cb_inc(struct pmu *pmu)
  * PEBS requires this to provide PID/TID information. This requires we flush
  * all queued PEBS records before we context switch to a new task.
  */
-static void perf_pmu_sched_task(struct task_struct *prev,
-				struct task_struct *next,
-				bool sched_in)
+static void __perf_pmu_sched_task(struct perf_cpu_context *cpuctx, bool sched_in)
 {
-	struct perf_cpu_context *cpuctx;
 	struct pmu *pmu;
 
-	if (prev == next)
-		return;
+	pmu = cpuctx->ctx.pmu; /* software PMUs will not have sched_task */
 
-	list_for_each_entry(cpuctx, this_cpu_ptr(&sched_cb_list), sched_cb_entry) {
-		pmu = cpuctx->ctx.pmu; /* software PMUs will not have sched_task */
-
-		if (WARN_ON_ONCE(!pmu->sched_task))
-			continue;
+	if (WARN_ON_ONCE(!pmu->sched_task))
+		return;
 
-		perf_ctx_lock(cpuctx, cpuctx->task_ctx);
-		perf_pmu_disable(pmu);
+	perf_ctx_lock(cpuctx, cpuctx->task_ctx);
+	perf_pmu_disable(pmu);
 
-		pmu->sched_task(cpuctx->task_ctx, sched_in);
+	pmu->sched_task(cpuctx->task_ctx, sched_in);
 
-		perf_pmu_enable(pmu);
-		perf_ctx_unlock(cpuctx, cpuctx->task_ctx);
-	}
+	perf_pmu_enable(pmu);
+	perf_ctx_unlock(cpuctx, cpuctx->task_ctx);
 }
 
 static void perf_event_switch(struct task_struct *task,
@@ -3512,9 +3537,6 @@ void __perf_event_task_sched_out(struct task_struct *task,
 {
 	int ctxn;
 
-	if (__this_cpu_read(perf_sched_cb_usages))
-		perf_pmu_sched_task(task, next, false);
-
 	if (atomic_read(&nr_switch_events))
 		perf_event_switch(task, next, false);
 
@@ -3746,10 +3768,14 @@ static void perf_event_context_sched_in(struct perf_event_context *ctx,
 					struct task_struct *task)
 {
 	struct perf_cpu_context *cpuctx;
+	struct pmu *pmu = ctx->pmu;
 
 	cpuctx = __get_cpu_context(ctx);
-	if (cpuctx->task_ctx == ctx)
+	if (cpuctx->task_ctx == ctx) {
+		if (cpuctx->sched_cb_usage)
+			__perf_pmu_sched_task(cpuctx, true);
 		return;
+	}
 
 	perf_ctx_lock(cpuctx, ctx);
 	/*
@@ -3759,7 +3785,7 @@ static void perf_event_context_sched_in(struct perf_event_context *ctx,
 	if (!ctx->nr_events)
 		goto unlock;
 
-	perf_pmu_disable(ctx->pmu);
+	perf_pmu_disable(pmu);
 	/*
 	 * We want to keep the following priority order:
 	 * cpu pinned (that don't need to move), task pinned,
@@ -3771,7 +3797,11 @@ static void perf_event_context_sched_in(struct perf_event_context *ctx,
 	if (!RB_EMPTY_ROOT(&ctx->pinned_groups.tree))
 		cpu_ctx_sched_out(cpuctx, EVENT_FLEXIBLE);
 	perf_event_sched_in(cpuctx, ctx, task);
-	perf_pmu_enable(ctx->pmu);
+
+	if (cpuctx->sched_cb_usage && pmu->sched_task)
+		pmu->sched_task(cpuctx->task_ctx, true);
+
+	perf_pmu_enable(pmu);
 
 unlock:
 	perf_ctx_unlock(cpuctx, ctx);
@@ -3814,9 +3844,6 @@ void __perf_event_task_sched_in(struct task_struct *prev,
 
 	if (atomic_read(&nr_switch_events))
 		perf_event_switch(task, prev, true);
-
-	if (__this_cpu_read(perf_sched_cb_usages))
-		perf_pmu_sched_task(prev, task, true);
 }
 
 static u64 perf_calculate_period(struct perf_event *event, u64 nsec, u64 count)
@@ -5869,11 +5896,11 @@ static void perf_pmu_output_stop(struct perf_event *event);
 static void perf_mmap_close(struct vm_area_struct *vma)
 {
 	struct perf_event *event = vma->vm_file->private_data;
-
 	struct perf_buffer *rb = ring_buffer_get(event);
 	struct user_struct *mmap_user = rb->mmap_user;
 	int mmap_locked = rb->mmap_locked;
 	unsigned long size = perf_data_size(rb);
+	bool detach_rest = false;
 
 	if (event->pmu->event_unmapped)
 		event->pmu->event_unmapped(event, vma->vm_mm);
@@ -5904,7 +5931,8 @@ static void perf_mmap_close(struct vm_area_struct *vma)
 		mutex_unlock(&event->mmap_mutex);
 	}
 
-	atomic_dec(&rb->mmap_count);
+	if (atomic_dec_and_test(&rb->mmap_count))
+		detach_rest = true;
 
 	if (!atomic_dec_and_mutex_lock(&event->mmap_count, &event->mmap_mutex))
 		goto out_put;
@@ -5913,7 +5941,7 @@ static void perf_mmap_close(struct vm_area_struct *vma)
 	mutex_unlock(&event->mmap_mutex);
 
 	/* If there's still other mmap()s of this buffer, we're done. */
-	if (atomic_read(&rb->mmap_count))
+	if (!detach_rest)
 		goto out_put;
 
 	/*
@@ -12829,7 +12857,6 @@ static void __init perf_event_init_all_cpus(void)
 #ifdef CONFIG_CGROUP_PERF
 		INIT_LIST_HEAD(&per_cpu(cgrp_cpuctx_list, cpu));
 #endif
-		INIT_LIST_HEAD(&per_cpu(sched_cb_list, cpu));
 	}
 }
 

kernel/trace/trace_kprobe.c

@@ -106,9 +106,10 @@ static nokprobe_inline bool trace_kprobe_has_gone(struct trace_kprobe *tk)
 static nokprobe_inline bool trace_kprobe_within_module(struct trace_kprobe *tk,
 						 struct module *mod)
 {
-	int len = strlen(mod->name);
+	int len = strlen(module_name(mod));
 	const char *name = trace_kprobe_symbol(tk);
-	return strncmp(mod->name, name, len) == 0 && name[len] == ':';
+
+	return strncmp(module_name(mod), name, len) == 0 && name[len] == ':';
 }
 
 static nokprobe_inline bool trace_kprobe_module_exist(struct trace_kprobe *tk)
@@ -688,7 +689,7 @@ static int trace_kprobe_module_callback(struct notifier_block *nb,
 			if (ret)
 				pr_warn("Failed to re-register probe %s on %s: %d\n",
 					trace_probe_name(&tk->tp),
-					mod->name, ret);
+					module_name(mod), ret);
 		}
 	}
 	mutex_unlock(&event_mutex);