diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c
index fd9f908debe5..c91434056c29 100644
--- a/arch/x86/events/intel/core.c
+++ b/arch/x86/events/intel/core.c
@@ -6236,6 +6236,19 @@ __init int intel_pmu_init(void)
 			pmu->num_counters = x86_pmu.num_counters;
 			pmu->num_counters_fixed = x86_pmu.num_counters_fixed;
 		}
+
+		/*
+		 * Quirk: For some Alder Lake machine, when all E-cores are disabled in
+		 * a BIOS, the leaf 0xA will enumerate all counters of P-cores. However,
+		 * the X86_FEATURE_HYBRID_CPU is still set. The above codes will
+		 * mistakenly add extra counters for P-cores. Correct the number of
+		 * counters here.
+		 */
+		if ((pmu->num_counters > 8) || (pmu->num_counters_fixed > 4)) {
+			pmu->num_counters = x86_pmu.num_counters;
+			pmu->num_counters_fixed = x86_pmu.num_counters_fixed;
+		}
+
 		pmu->max_pebs_events = min_t(unsigned, MAX_PEBS_EVENTS, pmu->num_counters);
 		pmu->unconstrained = (struct event_constraint)
 					__EVENT_CONSTRAINT(0, (1ULL << pmu->num_counters) - 1,
@@ -6340,6 +6353,8 @@ __init int intel_pmu_init(void)
 	}
 
 	if (x86_pmu.lbr_nr) {
+		intel_pmu_lbr_init();
+
 		pr_cont("%d-deep LBR, ", x86_pmu.lbr_nr);
 
 		/* only support branch_stack snapshot for perfmon >= v2 */
diff --git a/arch/x86/events/intel/lbr.c b/arch/x86/events/intel/lbr.c
index 8043213b75a5..669c2be14784 100644
--- a/arch/x86/events/intel/lbr.c
+++ b/arch/x86/events/intel/lbr.c
@@ -8,14 +8,6 @@
 
 #include "../perf_event.h"
 
-static const enum {
-	LBR_EIP_FLAGS		= 1,
-	LBR_TSX			= 2,
-} lbr_desc[LBR_FORMAT_MAX_KNOWN + 1] = {
-	[LBR_FORMAT_EIP_FLAGS]  = LBR_EIP_FLAGS,
-	[LBR_FORMAT_EIP_FLAGS2] = LBR_EIP_FLAGS | LBR_TSX,
-};
-
 /*
  * Intel LBR_SELECT bits
  * Intel Vol3a, April 2011, Section 16.7 Table 16-10
@@ -243,7 +235,7 @@ void intel_pmu_lbr_reset_64(void)
 	for (i = 0; i < x86_pmu.lbr_nr; i++) {
 		wrmsrl(x86_pmu.lbr_from + i, 0);
 		wrmsrl(x86_pmu.lbr_to   + i, 0);
-		if (x86_pmu.intel_cap.lbr_format == LBR_FORMAT_INFO)
+		if (x86_pmu.lbr_has_info)
 			wrmsrl(x86_pmu.lbr_info + i, 0);
 	}
 }
@@ -305,11 +297,10 @@ enum {
  */
 static inline bool lbr_from_signext_quirk_needed(void)
 {
-	int lbr_format = x86_pmu.intel_cap.lbr_format;
 	bool tsx_support = boot_cpu_has(X86_FEATURE_HLE) ||
 			   boot_cpu_has(X86_FEATURE_RTM);
 
-	return !tsx_support && (lbr_desc[lbr_format] & LBR_TSX);
+	return !tsx_support && x86_pmu.lbr_has_tsx;
 }
 
 static DEFINE_STATIC_KEY_FALSE(lbr_from_quirk_key);
@@ -427,12 +418,12 @@ rdlbr_all(struct lbr_entry *lbr, unsigned int idx, bool need_info)
 
 void intel_pmu_lbr_restore(void *ctx)
 {
-	bool need_info = x86_pmu.intel_cap.lbr_format == LBR_FORMAT_INFO;
 	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 	struct x86_perf_task_context *task_ctx = ctx;
-	int i;
-	unsigned lbr_idx, mask;
+	bool need_info = x86_pmu.lbr_has_info;
 	u64 tos = task_ctx->tos;
+	unsigned lbr_idx, mask;
+	int i;
 
 	mask = x86_pmu.lbr_nr - 1;
 	for (i = 0; i < task_ctx->valid_lbrs; i++) {
@@ -444,7 +435,7 @@ void intel_pmu_lbr_restore(void *ctx)
 		lbr_idx = (tos - i) & mask;
 		wrlbr_from(lbr_idx, 0);
 		wrlbr_to(lbr_idx, 0);
-		if (x86_pmu.intel_cap.lbr_format == LBR_FORMAT_INFO)
+		if (need_info)
 			wrlbr_info(lbr_idx, 0);
 	}
 
@@ -519,9 +510,9 @@ static void __intel_pmu_lbr_restore(void *ctx)
 
 void intel_pmu_lbr_save(void *ctx)
 {
-	bool need_info = x86_pmu.intel_cap.lbr_format == LBR_FORMAT_INFO;
 	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 	struct x86_perf_task_context *task_ctx = ctx;
+	bool need_info = x86_pmu.lbr_has_info;
 	unsigned lbr_idx, mask;
 	u64 tos;
 	int i;
@@ -816,7 +807,6 @@ void intel_pmu_lbr_read_64(struct cpu_hw_events *cpuc)
 {
 	bool need_info = false, call_stack = false;
 	unsigned long mask = x86_pmu.lbr_nr - 1;
-	int lbr_format = x86_pmu.intel_cap.lbr_format;
 	u64 tos = intel_pmu_lbr_tos();
 	int i;
 	int out = 0;
@@ -831,9 +821,7 @@ void intel_pmu_lbr_read_64(struct cpu_hw_events *cpuc)
 	for (i = 0; i < num; i++) {
 		unsigned long lbr_idx = (tos - i) & mask;
 		u64 from, to, mis = 0, pred = 0, in_tx = 0, abort = 0;
-		int skip = 0;
 		u16 cycles = 0;
-		int lbr_flags = lbr_desc[lbr_format];
 
 		from = rdlbr_from(lbr_idx, NULL);
 		to   = rdlbr_to(lbr_idx, NULL);
@@ -845,38 +833,40 @@ void intel_pmu_lbr_read_64(struct cpu_hw_events *cpuc)
 		if (call_stack && !from)
 			break;
 
-		if (lbr_format == LBR_FORMAT_INFO && need_info) {
-			u64 info;
+		if (x86_pmu.lbr_has_info) {
+			if (need_info) {
+				u64 info;
 
-			info = rdlbr_info(lbr_idx, NULL);
-			mis = !!(info & LBR_INFO_MISPRED);
-			pred = !mis;
-			in_tx = !!(info & LBR_INFO_IN_TX);
-			abort = !!(info & LBR_INFO_ABORT);
-			cycles = (info & LBR_INFO_CYCLES);
-		}
+				info = rdlbr_info(lbr_idx, NULL);
+				mis = !!(info & LBR_INFO_MISPRED);
+				pred = !mis;
+				cycles = (info & LBR_INFO_CYCLES);
+				if (x86_pmu.lbr_has_tsx) {
+					in_tx = !!(info & LBR_INFO_IN_TX);
+					abort = !!(info & LBR_INFO_ABORT);
+				}
+			}
+		} else {
+			int skip = 0;
 
-		if (lbr_format == LBR_FORMAT_TIME) {
-			mis = !!(from & LBR_FROM_FLAG_MISPRED);
-			pred = !mis;
-			skip = 1;
-			cycles = ((to >> 48) & LBR_INFO_CYCLES);
+			if (x86_pmu.lbr_from_flags) {
+				mis = !!(from & LBR_FROM_FLAG_MISPRED);
+				pred = !mis;
+				skip = 1;
+			}
+			if (x86_pmu.lbr_has_tsx) {
+				in_tx = !!(from & LBR_FROM_FLAG_IN_TX);
+				abort = !!(from & LBR_FROM_FLAG_ABORT);
+				skip = 3;
+			}
+			from = (u64)((((s64)from) << skip) >> skip);
 
-			to = (u64)((((s64)to) << 16) >> 16);
+			if (x86_pmu.lbr_to_cycles) {
+				cycles = ((to >> 48) & LBR_INFO_CYCLES);
+				to = (u64)((((s64)to) << 16) >> 16);
+			}
 		}
 
-		if (lbr_flags & LBR_EIP_FLAGS) {
-			mis = !!(from & LBR_FROM_FLAG_MISPRED);
-			pred = !mis;
-			skip = 1;
-		}
-		if (lbr_flags & LBR_TSX) {
-			in_tx = !!(from & LBR_FROM_FLAG_IN_TX);
-			abort = !!(from & LBR_FROM_FLAG_ABORT);
-			skip = 3;
-		}
-		from = (u64)((((s64)from) << skip) >> skip);
-
 		/*
 		 * Some CPUs report duplicated abort records,
 		 * with the second entry not having an abort bit set.
@@ -903,37 +893,40 @@ void intel_pmu_lbr_read_64(struct cpu_hw_events *cpuc)
 	cpuc->lbr_stack.hw_idx = tos;
 }
 
+static DEFINE_STATIC_KEY_FALSE(x86_lbr_mispred);
+static DEFINE_STATIC_KEY_FALSE(x86_lbr_cycles);
+static DEFINE_STATIC_KEY_FALSE(x86_lbr_type);
+
 static __always_inline int get_lbr_br_type(u64 info)
 {
-	if (!static_cpu_has(X86_FEATURE_ARCH_LBR) || !x86_pmu.lbr_br_type)
-		return 0;
+	int type = 0;
 
-	return (info & LBR_INFO_BR_TYPE) >> LBR_INFO_BR_TYPE_OFFSET;
+	if (static_branch_likely(&x86_lbr_type))
+		type = (info & LBR_INFO_BR_TYPE) >> LBR_INFO_BR_TYPE_OFFSET;
+
+	return type;
 }
 
 static __always_inline bool get_lbr_mispred(u64 info)
 {
-	if (static_cpu_has(X86_FEATURE_ARCH_LBR) && !x86_pmu.lbr_mispred)
-		return 0;
+	bool mispred = 0;
 
-	return !!(info & LBR_INFO_MISPRED);
-}
+	if (static_branch_likely(&x86_lbr_mispred))
+		mispred = !!(info & LBR_INFO_MISPRED);
 
-static __always_inline bool get_lbr_predicted(u64 info)
-{
-	if (static_cpu_has(X86_FEATURE_ARCH_LBR) && !x86_pmu.lbr_mispred)
-		return 0;
-
-	return !(info & LBR_INFO_MISPRED);
+	return mispred;
 }
 
 static __always_inline u16 get_lbr_cycles(u64 info)
 {
-	if (static_cpu_has(X86_FEATURE_ARCH_LBR) &&
-	    !(x86_pmu.lbr_timed_lbr && info & LBR_INFO_CYC_CNT_VALID))
-		return 0;
+	u16 cycles = info & LBR_INFO_CYCLES;
 
-	return info & LBR_INFO_CYCLES;
+	if (static_cpu_has(X86_FEATURE_ARCH_LBR) &&
+	    (!static_branch_likely(&x86_lbr_cycles) ||
+	     !(info & LBR_INFO_CYC_CNT_VALID)))
+		cycles = 0;
+
+	return cycles;
 }
 
 static void intel_pmu_store_lbr(struct cpu_hw_events *cpuc,
@@ -961,7 +954,7 @@ static void intel_pmu_store_lbr(struct cpu_hw_events *cpuc,
 		e->from		= from;
 		e->to		= to;
 		e->mispred	= get_lbr_mispred(info);
-		e->predicted	= get_lbr_predicted(info);
+		e->predicted	= !e->mispred;
 		e->in_tx	= !!(info & LBR_INFO_IN_TX);
 		e->abort	= !!(info & LBR_INFO_ABORT);
 		e->cycles	= get_lbr_cycles(info);
@@ -1120,7 +1113,7 @@ static int intel_pmu_setup_hw_lbr_filter(struct perf_event *event)
 
 	if ((br_type & PERF_SAMPLE_BRANCH_NO_CYCLES) &&
 	    (br_type & PERF_SAMPLE_BRANCH_NO_FLAGS) &&
-	    (x86_pmu.intel_cap.lbr_format == LBR_FORMAT_INFO))
+	    x86_pmu.lbr_has_info)
 		reg->config |= LBR_NO_INFO;
 
 	return 0;
@@ -1706,6 +1699,38 @@ void intel_pmu_lbr_init_knl(void)
 		x86_pmu.intel_cap.lbr_format = LBR_FORMAT_EIP_FLAGS;
 }
 
+void intel_pmu_lbr_init(void)
+{
+	switch (x86_pmu.intel_cap.lbr_format) {
+	case LBR_FORMAT_EIP_FLAGS2:
+		x86_pmu.lbr_has_tsx = 1;
+		fallthrough;
+	case LBR_FORMAT_EIP_FLAGS:
+		x86_pmu.lbr_from_flags = 1;
+		break;
+
+	case LBR_FORMAT_INFO:
+		x86_pmu.lbr_has_tsx = 1;
+		fallthrough;
+	case LBR_FORMAT_INFO2:
+		x86_pmu.lbr_has_info = 1;
+		break;
+
+	case LBR_FORMAT_TIME:
+		x86_pmu.lbr_from_flags = 1;
+		x86_pmu.lbr_to_cycles = 1;
+		break;
+	}
+
+	if (x86_pmu.lbr_has_info) {
+		/*
+		 * Only used in combination with baseline pebs.
+		 */
+		static_branch_enable(&x86_lbr_mispred);
+		static_branch_enable(&x86_lbr_cycles);
+	}
+}
+
 /*
  * LBR state size is variable based on the max number of registers.
  * This calculates the expected state size, which should match
@@ -1726,6 +1751,9 @@ static bool is_arch_lbr_xsave_available(void)
 	 * Check the LBR state with the corresponding software structure.
 	 * Disable LBR XSAVES support if the size doesn't match.
 	 */
+	if (xfeature_size(XFEATURE_LBR) == 0)
+		return false;
+
 	if (WARN_ON(xfeature_size(XFEATURE_LBR) != get_lbr_state_size()))
 		return false;
 
@@ -1765,6 +1793,12 @@ void __init intel_pmu_arch_lbr_init(void)
 	x86_pmu.lbr_br_type = ecx.split.lbr_br_type;
 	x86_pmu.lbr_nr = lbr_nr;
 
+	if (x86_pmu.lbr_mispred)
+		static_branch_enable(&x86_lbr_mispred);
+	if (x86_pmu.lbr_timed_lbr)
+		static_branch_enable(&x86_lbr_cycles);
+	if (x86_pmu.lbr_br_type)
+		static_branch_enable(&x86_lbr_type);
 
 	arch_lbr_xsave = is_arch_lbr_xsave_available();
 	if (arch_lbr_xsave) {
diff --git a/arch/x86/events/intel/uncore.c b/arch/x86/events/intel/uncore.c
index f1ba6ab2e97e..e497da9bf427 100644
--- a/arch/x86/events/intel/uncore.c
+++ b/arch/x86/events/intel/uncore.c
@@ -1762,7 +1762,7 @@ static const struct intel_uncore_init_fun rkl_uncore_init __initconst = {
 
 static const struct intel_uncore_init_fun adl_uncore_init __initconst = {
 	.cpu_init = adl_uncore_cpu_init,
-	.mmio_init = tgl_uncore_mmio_init,
+	.mmio_init = adl_uncore_mmio_init,
 };
 
 static const struct intel_uncore_init_fun icx_uncore_init __initconst = {
diff --git a/arch/x86/events/intel/uncore.h b/arch/x86/events/intel/uncore.h
index b9687980aab6..2adeaf4de4df 100644
--- a/arch/x86/events/intel/uncore.h
+++ b/arch/x86/events/intel/uncore.h
@@ -584,10 +584,11 @@ 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 adl_uncore_cpu_init(void);
 void tgl_uncore_cpu_init(void);
+void adl_uncore_cpu_init(void);
 void tgl_uncore_mmio_init(void);
 void tgl_l_uncore_mmio_init(void);
+void adl_uncore_mmio_init(void);
 int snb_pci2phy_map_init(int devid);
 
 /* uncore_snbep.c */
diff --git a/arch/x86/events/intel/uncore_discovery.c b/arch/x86/events/intel/uncore_discovery.c
index 3049c646fa20..6ddadb482f68 100644
--- a/arch/x86/events/intel/uncore_discovery.c
+++ b/arch/x86/events/intel/uncore_discovery.c
@@ -494,8 +494,8 @@ void intel_generic_uncore_mmio_enable_box(struct intel_uncore_box *box)
 	writel(0, box->io_addr);
 }
 
-static void intel_generic_uncore_mmio_enable_event(struct intel_uncore_box *box,
-					     struct perf_event *event)
+void intel_generic_uncore_mmio_enable_event(struct intel_uncore_box *box,
+					    struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
 
diff --git a/arch/x86/events/intel/uncore_discovery.h b/arch/x86/events/intel/uncore_discovery.h
index 6d735611c281..cfaf558bdb6b 100644
--- a/arch/x86/events/intel/uncore_discovery.h
+++ b/arch/x86/events/intel/uncore_discovery.h
@@ -139,6 +139,8 @@ void intel_generic_uncore_mmio_disable_box(struct intel_uncore_box *box);
 void intel_generic_uncore_mmio_enable_box(struct intel_uncore_box *box);
 void intel_generic_uncore_mmio_disable_event(struct intel_uncore_box *box,
 					     struct perf_event *event);
+void intel_generic_uncore_mmio_enable_event(struct intel_uncore_box *box,
+					    struct perf_event *event);
 
 void intel_generic_uncore_pci_init_box(struct intel_uncore_box *box);
 void intel_generic_uncore_pci_disable_box(struct intel_uncore_box *box);
diff --git a/arch/x86/events/intel/uncore_snb.c b/arch/x86/events/intel/uncore_snb.c
index 0f63706cdadf..f698a55bde81 100644
--- a/arch/x86/events/intel/uncore_snb.c
+++ b/arch/x86/events/intel/uncore_snb.c
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 /* Nehalem/SandBridge/Haswell/Broadwell/Skylake uncore support */
 #include "uncore.h"
+#include "uncore_discovery.h"
 
 /* Uncore IMC PCI IDs */
 #define PCI_DEVICE_ID_INTEL_SNB_IMC		0x0100
@@ -64,6 +65,20 @@
 #define PCI_DEVICE_ID_INTEL_RKL_2_IMC		0x4c53
 #define PCI_DEVICE_ID_INTEL_ADL_1_IMC		0x4660
 #define PCI_DEVICE_ID_INTEL_ADL_2_IMC		0x4641
+#define PCI_DEVICE_ID_INTEL_ADL_3_IMC		0x4601
+#define PCI_DEVICE_ID_INTEL_ADL_4_IMC		0x4602
+#define PCI_DEVICE_ID_INTEL_ADL_5_IMC		0x4609
+#define PCI_DEVICE_ID_INTEL_ADL_6_IMC		0x460a
+#define PCI_DEVICE_ID_INTEL_ADL_7_IMC		0x4621
+#define PCI_DEVICE_ID_INTEL_ADL_8_IMC		0x4623
+#define PCI_DEVICE_ID_INTEL_ADL_9_IMC		0x4629
+#define PCI_DEVICE_ID_INTEL_ADL_10_IMC		0x4637
+#define PCI_DEVICE_ID_INTEL_ADL_11_IMC		0x463b
+#define PCI_DEVICE_ID_INTEL_ADL_12_IMC		0x4648
+#define PCI_DEVICE_ID_INTEL_ADL_13_IMC		0x4649
+#define PCI_DEVICE_ID_INTEL_ADL_14_IMC		0x4650
+#define PCI_DEVICE_ID_INTEL_ADL_15_IMC		0x4668
+#define PCI_DEVICE_ID_INTEL_ADL_16_IMC		0x4670
 
 /* SNB event control */
 #define SNB_UNC_CTL_EV_SEL_MASK			0x000000ff
@@ -155,6 +170,7 @@
 
 DEFINE_UNCORE_FORMAT_ATTR(event, event, "config:0-7");
 DEFINE_UNCORE_FORMAT_ATTR(umask, umask, "config:8-15");
+DEFINE_UNCORE_FORMAT_ATTR(chmask, chmask, "config:8-11");
 DEFINE_UNCORE_FORMAT_ATTR(edge, edge, "config:18");
 DEFINE_UNCORE_FORMAT_ATTR(inv, inv, "config:23");
 DEFINE_UNCORE_FORMAT_ATTR(cmask5, cmask, "config:24-28");
@@ -1334,6 +1350,62 @@ static const struct pci_device_id tgl_uncore_pci_ids[] = {
 		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_2_IMC),
 		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
 	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_3_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_4_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_5_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_6_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_7_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_8_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_9_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_10_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_11_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_12_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_13_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_14_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_15_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
+	{ /* IMC */
+		PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ADL_16_IMC),
+		.driver_data = UNCORE_PCI_DEV_DATA(SNB_PCI_UNCORE_IMC, 0),
+	},
 	{ /* end: all zeroes */ }
 };
 
@@ -1390,7 +1462,8 @@ static struct pci_dev *tgl_uncore_get_mc_dev(void)
 #define TGL_UNCORE_MMIO_IMC_MEM_OFFSET		0x10000
 #define TGL_UNCORE_PCI_IMC_MAP_SIZE		0xe000
 
-static void tgl_uncore_imc_freerunning_init_box(struct intel_uncore_box *box)
+static void __uncore_imc_init_box(struct intel_uncore_box *box,
+				  unsigned int base_offset)
 {
 	struct pci_dev *pdev = tgl_uncore_get_mc_dev();
 	struct intel_uncore_pmu *pmu = box->pmu;
@@ -1417,11 +1490,17 @@ static void tgl_uncore_imc_freerunning_init_box(struct intel_uncore_box *box)
 	addr |= ((resource_size_t)mch_bar << 32);
 #endif
 
+	addr += base_offset;
 	box->io_addr = ioremap(addr, type->mmio_map_size);
 	if (!box->io_addr)
 		pr_warn("perf uncore: Failed to ioremap for %s.\n", type->name);
 }
 
+static void tgl_uncore_imc_freerunning_init_box(struct intel_uncore_box *box)
+{
+	__uncore_imc_init_box(box, 0);
+}
+
 static struct intel_uncore_ops tgl_uncore_imc_freerunning_ops = {
 	.init_box	= tgl_uncore_imc_freerunning_init_box,
 	.exit_box	= uncore_mmio_exit_box,
@@ -1469,3 +1548,136 @@ void tgl_uncore_mmio_init(void)
 }
 
 /* end of Tiger Lake MMIO uncore support */
+
+/* Alder Lake MMIO uncore support */
+#define ADL_UNCORE_IMC_BASE			0xd900
+#define ADL_UNCORE_IMC_MAP_SIZE			0x200
+#define ADL_UNCORE_IMC_CTR			0xe8
+#define ADL_UNCORE_IMC_CTRL			0xd0
+#define ADL_UNCORE_IMC_GLOBAL_CTL		0xc0
+#define ADL_UNCORE_IMC_BOX_CTL			0xc4
+#define ADL_UNCORE_IMC_FREERUNNING_BASE		0xd800
+#define ADL_UNCORE_IMC_FREERUNNING_MAP_SIZE	0x100
+
+#define ADL_UNCORE_IMC_CTL_FRZ			(1 << 0)
+#define ADL_UNCORE_IMC_CTL_RST_CTRL		(1 << 1)
+#define ADL_UNCORE_IMC_CTL_RST_CTRS		(1 << 2)
+#define ADL_UNCORE_IMC_CTL_INT			(ADL_UNCORE_IMC_CTL_RST_CTRL | \
+						ADL_UNCORE_IMC_CTL_RST_CTRS)
+
+static void adl_uncore_imc_init_box(struct intel_uncore_box *box)
+{
+	__uncore_imc_init_box(box, ADL_UNCORE_IMC_BASE);
+
+	/* The global control in MC1 can control both MCs. */
+	if (box->io_addr && (box->pmu->pmu_idx == 1))
+		writel(ADL_UNCORE_IMC_CTL_INT, box->io_addr + ADL_UNCORE_IMC_GLOBAL_CTL);
+}
+
+static void adl_uncore_mmio_disable_box(struct intel_uncore_box *box)
+{
+	if (!box->io_addr)
+		return;
+
+	writel(ADL_UNCORE_IMC_CTL_FRZ, box->io_addr + uncore_mmio_box_ctl(box));
+}
+
+static void adl_uncore_mmio_enable_box(struct intel_uncore_box *box)
+{
+	if (!box->io_addr)
+		return;
+
+	writel(0, box->io_addr + uncore_mmio_box_ctl(box));
+}
+
+static struct intel_uncore_ops adl_uncore_mmio_ops = {
+	.init_box	= adl_uncore_imc_init_box,
+	.exit_box	= uncore_mmio_exit_box,
+	.disable_box	= adl_uncore_mmio_disable_box,
+	.enable_box	= adl_uncore_mmio_enable_box,
+	.disable_event	= intel_generic_uncore_mmio_disable_event,
+	.enable_event	= intel_generic_uncore_mmio_enable_event,
+	.read_counter	= uncore_mmio_read_counter,
+};
+
+#define ADL_UNC_CTL_CHMASK_MASK			0x00000f00
+#define ADL_UNC_IMC_EVENT_MASK			(SNB_UNC_CTL_EV_SEL_MASK | \
+						 ADL_UNC_CTL_CHMASK_MASK | \
+						 SNB_UNC_CTL_EDGE_DET)
+
+static struct attribute *adl_uncore_imc_formats_attr[] = {
+	&format_attr_event.attr,
+	&format_attr_chmask.attr,
+	&format_attr_edge.attr,
+	NULL,
+};
+
+static const struct attribute_group adl_uncore_imc_format_group = {
+	.name		= "format",
+	.attrs		= adl_uncore_imc_formats_attr,
+};
+
+static struct intel_uncore_type adl_uncore_imc = {
+	.name		= "imc",
+	.num_counters   = 5,
+	.num_boxes	= 2,
+	.perf_ctr_bits	= 64,
+	.perf_ctr	= ADL_UNCORE_IMC_CTR,
+	.event_ctl	= ADL_UNCORE_IMC_CTRL,
+	.event_mask	= ADL_UNC_IMC_EVENT_MASK,
+	.box_ctl	= ADL_UNCORE_IMC_BOX_CTL,
+	.mmio_offset	= 0,
+	.mmio_map_size	= ADL_UNCORE_IMC_MAP_SIZE,
+	.ops		= &adl_uncore_mmio_ops,
+	.format_group	= &adl_uncore_imc_format_group,
+};
+
+enum perf_adl_uncore_imc_freerunning_types {
+	ADL_MMIO_UNCORE_IMC_DATA_TOTAL,
+	ADL_MMIO_UNCORE_IMC_DATA_READ,
+	ADL_MMIO_UNCORE_IMC_DATA_WRITE,
+	ADL_MMIO_UNCORE_IMC_FREERUNNING_TYPE_MAX
+};
+
+static struct freerunning_counters adl_uncore_imc_freerunning[] = {
+	[ADL_MMIO_UNCORE_IMC_DATA_TOTAL]	= { 0x40, 0x0, 0x0, 1, 64 },
+	[ADL_MMIO_UNCORE_IMC_DATA_READ]		= { 0x58, 0x0, 0x0, 1, 64 },
+	[ADL_MMIO_UNCORE_IMC_DATA_WRITE]	= { 0xA0, 0x0, 0x0, 1, 64 },
+};
+
+static void adl_uncore_imc_freerunning_init_box(struct intel_uncore_box *box)
+{
+	__uncore_imc_init_box(box, ADL_UNCORE_IMC_FREERUNNING_BASE);
+}
+
+static struct intel_uncore_ops adl_uncore_imc_freerunning_ops = {
+	.init_box	= adl_uncore_imc_freerunning_init_box,
+	.exit_box	= uncore_mmio_exit_box,
+	.read_counter	= uncore_mmio_read_counter,
+	.hw_config	= uncore_freerunning_hw_config,
+};
+
+static struct intel_uncore_type adl_uncore_imc_free_running = {
+	.name			= "imc_free_running",
+	.num_counters		= 3,
+	.num_boxes		= 2,
+	.num_freerunning_types	= ADL_MMIO_UNCORE_IMC_FREERUNNING_TYPE_MAX,
+	.mmio_map_size		= ADL_UNCORE_IMC_FREERUNNING_MAP_SIZE,
+	.freerunning		= adl_uncore_imc_freerunning,
+	.ops			= &adl_uncore_imc_freerunning_ops,
+	.event_descs		= tgl_uncore_imc_events,
+	.format_group		= &tgl_uncore_imc_format_group,
+};
+
+static struct intel_uncore_type *adl_mmio_uncores[] = {
+	&adl_uncore_imc,
+	&adl_uncore_imc_free_running,
+	NULL
+};
+
+void adl_uncore_mmio_init(void)
+{
+	uncore_mmio_uncores = adl_mmio_uncores;
+}
+
+/* end of Alder Lake MMIO uncore support */
diff --git a/arch/x86/events/intel/uncore_snbep.c b/arch/x86/events/intel/uncore_snbep.c
index 3660f698fb2a..ed869443efb2 100644
--- a/arch/x86/events/intel/uncore_snbep.c
+++ b/arch/x86/events/intel/uncore_snbep.c
@@ -5482,7 +5482,7 @@ static struct intel_uncore_type icx_uncore_imc = {
 	.fixed_ctr_bits	= 48,
 	.fixed_ctr	= SNR_IMC_MMIO_PMON_FIXED_CTR,
 	.fixed_ctl	= SNR_IMC_MMIO_PMON_FIXED_CTL,
-	.event_descs	= hswep_uncore_imc_events,
+	.event_descs	= snr_uncore_imc_events,
 	.perf_ctr	= SNR_IMC_MMIO_PMON_CTR0,
 	.event_ctl	= SNR_IMC_MMIO_PMON_CTL0,
 	.event_mask	= SNBEP_PMON_RAW_EVENT_MASK,
diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h
index 9d376e528dfc..150261d929b9 100644
--- a/arch/x86/events/perf_event.h
+++ b/arch/x86/events/perf_event.h
@@ -215,7 +215,8 @@ enum {
 	LBR_FORMAT_EIP_FLAGS2	= 0x04,
 	LBR_FORMAT_INFO		= 0x05,
 	LBR_FORMAT_TIME		= 0x06,
-	LBR_FORMAT_MAX_KNOWN    = LBR_FORMAT_TIME,
+	LBR_FORMAT_INFO2	= 0x07,
+	LBR_FORMAT_MAX_KNOWN    = LBR_FORMAT_INFO2,
 };
 
 enum {
@@ -840,6 +841,11 @@ struct x86_pmu {
 	bool		lbr_double_abort;	   /* duplicated lbr aborts */
 	bool		lbr_pt_coexist;		   /* (LBR|BTS) may coexist with PT */
 
+	unsigned int	lbr_has_info:1;
+	unsigned int	lbr_has_tsx:1;
+	unsigned int	lbr_from_flags:1;
+	unsigned int	lbr_to_cycles:1;
+
 	/*
 	 * Intel Architectural LBR CPUID Enumeration
 	 */
@@ -1392,6 +1398,8 @@ void intel_pmu_lbr_init_skl(void);
 
 void intel_pmu_lbr_init_knl(void);
 
+void intel_pmu_lbr_init(void);
+
 void intel_pmu_arch_lbr_init(void);
 
 void intel_pmu_pebs_data_source_nhm(void);
diff --git a/arch/x86/events/rapl.c b/arch/x86/events/rapl.c
index 85feafacc445..77e3a47af5ad 100644
--- a/arch/x86/events/rapl.c
+++ b/arch/x86/events/rapl.c
@@ -536,11 +536,14 @@ static struct perf_msr intel_rapl_spr_msrs[] = {
  * - perf_msr_probe(PERF_RAPL_MAX)
  * - want to use same event codes across both architectures
  */
-static struct perf_msr amd_rapl_msrs[PERF_RAPL_MAX] = {
-	[PERF_RAPL_PKG]  = { MSR_AMD_PKG_ENERGY_STATUS,  &rapl_events_pkg_group,   test_msr },
+static struct perf_msr amd_rapl_msrs[] = {
+	[PERF_RAPL_PP0]  = { 0, &rapl_events_cores_group, 0, false, 0 },
+	[PERF_RAPL_PKG]  = { MSR_AMD_PKG_ENERGY_STATUS,  &rapl_events_pkg_group,   test_msr, false, RAPL_MSR_MASK },
+	[PERF_RAPL_RAM]  = { 0, &rapl_events_ram_group,   0, false, 0 },
+	[PERF_RAPL_PP1]  = { 0, &rapl_events_gpu_group,   0, false, 0 },
+	[PERF_RAPL_PSYS] = { 0, &rapl_events_psys_group,  0, false, 0 },
 };
 
-
 static int rapl_cpu_offline(unsigned int cpu)
 {
 	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index 117f230bcdfd..733649184b27 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -693,18 +693,6 @@ struct perf_event {
 	u64				total_time_running;
 	u64				tstamp;
 
-	/*
-	 * timestamp shadows the actual context timing but it can
-	 * be safely used in NMI interrupt context. It reflects the
-	 * context time as it was when the event was last scheduled in,
-	 * or when ctx_sched_in failed to schedule the event because we
-	 * run out of PMC.
-	 *
-	 * ctx_time already accounts for ctx->timestamp. Therefore to
-	 * compute ctx_time for a sample, simply add perf_clock().
-	 */
-	u64				shadow_ctx_time;
-
 	struct perf_event_attr		attr;
 	u16				header_size;
 	u16				id_header_size;
@@ -852,6 +840,7 @@ struct perf_event_context {
 	 */
 	u64				time;
 	u64				timestamp;
+	u64				timeoffset;
 
 	/*
 	 * These fields let us detect when two contexts have both
@@ -934,6 +923,8 @@ struct bpf_perf_event_data_kern {
 struct perf_cgroup_info {
 	u64				time;
 	u64				timestamp;
+	u64				timeoffset;
+	int				active;
 };
 
 struct perf_cgroup {
diff --git a/kernel/events/core.c b/kernel/events/core.c
index fc18664f49b0..479c9e672ec4 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -674,6 +674,23 @@ perf_event_set_state(struct perf_event *event, enum perf_event_state state)
 	WRITE_ONCE(event->state, state);
 }
 
+/*
+ * UP store-release, load-acquire
+ */
+
+#define __store_release(ptr, val)					\
+do {									\
+	barrier();							\
+	WRITE_ONCE(*(ptr), (val));					\
+} while (0)
+
+#define __load_acquire(ptr)						\
+({									\
+	__unqual_scalar_typeof(*(ptr)) ___p = READ_ONCE(*(ptr));	\
+	barrier();							\
+	___p;								\
+})
+
 #ifdef CONFIG_CGROUP_PERF
 
 static inline bool
@@ -719,34 +736,51 @@ static inline u64 perf_cgroup_event_time(struct perf_event *event)
 	return t->time;
 }
 
-static inline void __update_cgrp_time(struct perf_cgroup *cgrp)
+static inline u64 perf_cgroup_event_time_now(struct perf_event *event, u64 now)
 {
-	struct perf_cgroup_info *info;
-	u64 now;
+	struct perf_cgroup_info *t;
 
-	now = perf_clock();
-
-	info = this_cpu_ptr(cgrp->info);
-
-	info->time += now - info->timestamp;
-	info->timestamp = now;
+	t = per_cpu_ptr(event->cgrp->info, event->cpu);
+	if (!__load_acquire(&t->active))
+		return t->time;
+	now += READ_ONCE(t->timeoffset);
+	return now;
 }
 
-static inline void update_cgrp_time_from_cpuctx(struct perf_cpu_context *cpuctx)
+static inline void __update_cgrp_time(struct perf_cgroup_info *info, u64 now, bool adv)
+{
+	if (adv)
+		info->time += now - info->timestamp;
+	info->timestamp = now;
+	/*
+	 * see update_context_time()
+	 */
+	WRITE_ONCE(info->timeoffset, info->time - info->timestamp);
+}
+
+static inline void update_cgrp_time_from_cpuctx(struct perf_cpu_context *cpuctx, bool final)
 {
 	struct perf_cgroup *cgrp = cpuctx->cgrp;
 	struct cgroup_subsys_state *css;
+	struct perf_cgroup_info *info;
 
 	if (cgrp) {
+		u64 now = perf_clock();
+
 		for (css = &cgrp->css; css; css = css->parent) {
 			cgrp = container_of(css, struct perf_cgroup, css);
-			__update_cgrp_time(cgrp);
+			info = this_cpu_ptr(cgrp->info);
+
+			__update_cgrp_time(info, now, true);
+			if (final)
+				__store_release(&info->active, 0);
 		}
 	}
 }
 
 static inline void update_cgrp_time_from_event(struct perf_event *event)
 {
+	struct perf_cgroup_info *info;
 	struct perf_cgroup *cgrp;
 
 	/*
@@ -760,8 +794,10 @@ static inline void update_cgrp_time_from_event(struct perf_event *event)
 	/*
 	 * Do not update time when cgroup is not active
 	 */
-	if (cgroup_is_descendant(cgrp->css.cgroup, event->cgrp->css.cgroup))
-		__update_cgrp_time(event->cgrp);
+	if (cgroup_is_descendant(cgrp->css.cgroup, event->cgrp->css.cgroup)) {
+		info = this_cpu_ptr(event->cgrp->info);
+		__update_cgrp_time(info, perf_clock(), true);
+	}
 }
 
 static inline void
@@ -785,7 +821,8 @@ perf_cgroup_set_timestamp(struct task_struct *task,
 	for (css = &cgrp->css; css; css = css->parent) {
 		cgrp = container_of(css, struct perf_cgroup, css);
 		info = this_cpu_ptr(cgrp->info);
-		info->timestamp = ctx->timestamp;
+		__update_cgrp_time(info, ctx->timestamp, false);
+		__store_release(&info->active, 1);
 	}
 }
 
@@ -981,14 +1018,6 @@ out:
 	return ret;
 }
 
-static inline void
-perf_cgroup_set_shadow_time(struct perf_event *event, u64 now)
-{
-	struct perf_cgroup_info *t;
-	t = per_cpu_ptr(event->cgrp->info, event->cpu);
-	event->shadow_ctx_time = now - t->timestamp;
-}
-
 static inline void
 perf_cgroup_event_enable(struct perf_event *event, struct perf_event_context *ctx)
 {
@@ -1066,7 +1095,8 @@ static inline void update_cgrp_time_from_event(struct perf_event *event)
 {
 }
 
-static inline void update_cgrp_time_from_cpuctx(struct perf_cpu_context *cpuctx)
+static inline void update_cgrp_time_from_cpuctx(struct perf_cpu_context *cpuctx,
+						bool final)
 {
 }
 
@@ -1098,12 +1128,12 @@ perf_cgroup_switch(struct task_struct *task, struct task_struct *next)
 {
 }
 
-static inline void
-perf_cgroup_set_shadow_time(struct perf_event *event, u64 now)
+static inline u64 perf_cgroup_event_time(struct perf_event *event)
 {
+	return 0;
 }
 
-static inline u64 perf_cgroup_event_time(struct perf_event *event)
+static inline u64 perf_cgroup_event_time_now(struct perf_event *event, u64 now)
 {
 	return 0;
 }
@@ -1525,22 +1555,59 @@ static void perf_unpin_context(struct perf_event_context *ctx)
 /*
  * Update the record of the current time in a context.
  */
-static void update_context_time(struct perf_event_context *ctx)
+static void __update_context_time(struct perf_event_context *ctx, bool adv)
 {
 	u64 now = perf_clock();
 
-	ctx->time += now - ctx->timestamp;
+	if (adv)
+		ctx->time += now - ctx->timestamp;
 	ctx->timestamp = now;
+
+	/*
+	 * The above: time' = time + (now - timestamp), can be re-arranged
+	 * into: time` = now + (time - timestamp), which gives a single value
+	 * offset to compute future time without locks on.
+	 *
+	 * See perf_event_time_now(), which can be used from NMI context where
+	 * it's (obviously) not possible to acquire ctx->lock in order to read
+	 * both the above values in a consistent manner.
+	 */
+	WRITE_ONCE(ctx->timeoffset, ctx->time - ctx->timestamp);
+}
+
+static void update_context_time(struct perf_event_context *ctx)
+{
+	__update_context_time(ctx, true);
 }
 
 static u64 perf_event_time(struct perf_event *event)
 {
 	struct perf_event_context *ctx = event->ctx;
 
+	if (unlikely(!ctx))
+		return 0;
+
 	if (is_cgroup_event(event))
 		return perf_cgroup_event_time(event);
 
-	return ctx ? ctx->time : 0;
+	return ctx->time;
+}
+
+static u64 perf_event_time_now(struct perf_event *event, u64 now)
+{
+	struct perf_event_context *ctx = event->ctx;
+
+	if (unlikely(!ctx))
+		return 0;
+
+	if (is_cgroup_event(event))
+		return perf_cgroup_event_time_now(event, now);
+
+	if (!(__load_acquire(&ctx->is_active) & EVENT_TIME))
+		return ctx->time;
+
+	now += READ_ONCE(ctx->timeoffset);
+	return now;
 }
 
 static enum event_type_t get_event_type(struct perf_event *event)
@@ -2350,7 +2417,7 @@ __perf_remove_from_context(struct perf_event *event,
 
 	if (ctx->is_active & EVENT_TIME) {
 		update_context_time(ctx);
-		update_cgrp_time_from_cpuctx(cpuctx);
+		update_cgrp_time_from_cpuctx(cpuctx, false);
 	}
 
 	event_sched_out(event, cpuctx, ctx);
@@ -2361,6 +2428,9 @@ __perf_remove_from_context(struct perf_event *event,
 	list_del_event(event, ctx);
 
 	if (!ctx->nr_events && ctx->is_active) {
+		if (ctx == &cpuctx->ctx)
+			update_cgrp_time_from_cpuctx(cpuctx, true);
+
 		ctx->is_active = 0;
 		ctx->rotate_necessary = 0;
 		if (ctx->task) {
@@ -2482,40 +2552,6 @@ void perf_event_disable_inatomic(struct perf_event *event)
 	irq_work_queue(&event->pending);
 }
 
-static void perf_set_shadow_time(struct perf_event *event,
-				 struct perf_event_context *ctx)
-{
-	/*
-	 * use the correct time source for the time snapshot
-	 *
-	 * We could get by without this by leveraging the
-	 * fact that to get to this function, the caller
-	 * has most likely already called update_context_time()
-	 * and update_cgrp_time_xx() and thus both timestamp
-	 * are identical (or very close). Given that tstamp is,
-	 * already adjusted for cgroup, we could say that:
-	 *    tstamp - ctx->timestamp
-	 * is equivalent to
-	 *    tstamp - cgrp->timestamp.
-	 *
-	 * Then, in perf_output_read(), the calculation would
-	 * work with no changes because:
-	 * - event is guaranteed scheduled in
-	 * - no scheduled out in between
-	 * - thus the timestamp would be the same
-	 *
-	 * But this is a bit hairy.
-	 *
-	 * So instead, we have an explicit cgroup call to remain
-	 * within the time source all along. We believe it
-	 * is cleaner and simpler to understand.
-	 */
-	if (is_cgroup_event(event))
-		perf_cgroup_set_shadow_time(event, event->tstamp);
-	else
-		event->shadow_ctx_time = event->tstamp - ctx->timestamp;
-}
-
 #define MAX_INTERRUPTS (~0ULL)
 
 static void perf_log_throttle(struct perf_event *event, int enable);
@@ -2556,8 +2592,6 @@ event_sched_in(struct perf_event *event,
 
 	perf_pmu_disable(event->pmu);
 
-	perf_set_shadow_time(event, ctx);
-
 	perf_log_itrace_start(event);
 
 	if (event->pmu->add(event, PERF_EF_START)) {
@@ -3251,16 +3285,6 @@ static void ctx_sched_out(struct perf_event_context *ctx,
 		return;
 	}
 
-	ctx->is_active &= ~event_type;
-	if (!(ctx->is_active & EVENT_ALL))
-		ctx->is_active = 0;
-
-	if (ctx->task) {
-		WARN_ON_ONCE(cpuctx->task_ctx != ctx);
-		if (!ctx->is_active)
-			cpuctx->task_ctx = NULL;
-	}
-
 	/*
 	 * Always update time if it was set; not only when it changes.
 	 * Otherwise we can 'forget' to update time for any but the last
@@ -3274,7 +3298,22 @@ static void ctx_sched_out(struct perf_event_context *ctx,
 	if (is_active & EVENT_TIME) {
 		/* update (and stop) ctx time */
 		update_context_time(ctx);
-		update_cgrp_time_from_cpuctx(cpuctx);
+		update_cgrp_time_from_cpuctx(cpuctx, ctx == &cpuctx->ctx);
+		/*
+		 * CPU-release for the below ->is_active store,
+		 * see __load_acquire() in perf_event_time_now()
+		 */
+		barrier();
+	}
+
+	ctx->is_active &= ~event_type;
+	if (!(ctx->is_active & EVENT_ALL))
+		ctx->is_active = 0;
+
+	if (ctx->task) {
+		WARN_ON_ONCE(cpuctx->task_ctx != ctx);
+		if (!ctx->is_active)
+			cpuctx->task_ctx = NULL;
 	}
 
 	is_active ^= ctx->is_active; /* changed bits */
@@ -3711,13 +3750,19 @@ static noinline int visit_groups_merge(struct perf_cpu_context *cpuctx,
 	return 0;
 }
 
+/*
+ * Because the userpage is strictly per-event (there is no concept of context,
+ * so there cannot be a context indirection), every userpage must be updated
+ * when context time starts :-(
+ *
+ * IOW, we must not miss EVENT_TIME edges.
+ */
 static inline bool event_update_userpage(struct perf_event *event)
 {
 	if (likely(!atomic_read(&event->mmap_count)))
 		return false;
 
 	perf_event_update_time(event);
-	perf_set_shadow_time(event, event->ctx);
 	perf_event_update_userpage(event);
 
 	return true;
@@ -3801,13 +3846,23 @@ ctx_sched_in(struct perf_event_context *ctx,
 	     struct task_struct *task)
 {
 	int is_active = ctx->is_active;
-	u64 now;
 
 	lockdep_assert_held(&ctx->lock);
 
 	if (likely(!ctx->nr_events))
 		return;
 
+	if (is_active ^ EVENT_TIME) {
+		/* start ctx time */
+		__update_context_time(ctx, false);
+		perf_cgroup_set_timestamp(task, ctx);
+		/*
+		 * CPU-release for the below ->is_active store,
+		 * see __load_acquire() in perf_event_time_now()
+		 */
+		barrier();
+	}
+
 	ctx->is_active |= (event_type | EVENT_TIME);
 	if (ctx->task) {
 		if (!is_active)
@@ -3818,13 +3873,6 @@ ctx_sched_in(struct perf_event_context *ctx,
 
 	is_active ^= ctx->is_active; /* changed bits */
 
-	if (is_active & EVENT_TIME) {
-		/* start ctx time */
-		now = perf_clock();
-		ctx->timestamp = now;
-		perf_cgroup_set_timestamp(task, ctx);
-	}
-
 	/*
 	 * First go through the list and put on any pinned groups
 	 * in order to give them the best chance of going on.
@@ -4418,6 +4466,18 @@ static inline u64 perf_event_count(struct perf_event *event)
 	return local64_read(&event->count) + atomic64_read(&event->child_count);
 }
 
+static void calc_timer_values(struct perf_event *event,
+				u64 *now,
+				u64 *enabled,
+				u64 *running)
+{
+	u64 ctx_time;
+
+	*now = perf_clock();
+	ctx_time = perf_event_time_now(event, *now);
+	__perf_update_times(event, ctx_time, enabled, running);
+}
+
 /*
  * NMI-safe method to read a local event, that is an event that
  * is:
@@ -4477,10 +4537,9 @@ int perf_event_read_local(struct perf_event *event, u64 *value,
 
 	*value = local64_read(&event->count);
 	if (enabled || running) {
-		u64 now = event->shadow_ctx_time + perf_clock();
-		u64 __enabled, __running;
+		u64 __enabled, __running, __now;;
 
-		__perf_update_times(event, now, &__enabled, &__running);
+		calc_timer_values(event, &__now, &__enabled, &__running);
 		if (enabled)
 			*enabled = __enabled;
 		if (running)
@@ -5802,18 +5861,6 @@ static int perf_event_index(struct perf_event *event)
 	return event->pmu->event_idx(event);
 }
 
-static void calc_timer_values(struct perf_event *event,
-				u64 *now,
-				u64 *enabled,
-				u64 *running)
-{
-	u64 ctx_time;
-
-	*now = perf_clock();
-	ctx_time = event->shadow_ctx_time + *now;
-	__perf_update_times(event, ctx_time, enabled, running);
-}
-
 static void perf_event_init_userpage(struct perf_event *event)
 {
 	struct perf_event_mmap_page *userpg;
@@ -6353,7 +6400,6 @@ accounting:
 		ring_buffer_attach(event, rb);
 
 		perf_event_update_time(event);
-		perf_set_shadow_time(event, event->ctx);
 		perf_event_init_userpage(event);
 		perf_event_update_userpage(event);
 	} else {