Performance events changes for v6.13:

- Uprobes:
     - Add BPF session support (Jiri Olsa)
     - Switch to RCU Tasks Trace flavor for better performance (Andrii Nakryiko)
     - Massively increase uretprobe SMP scalability by SRCU-protecting
       the uretprobe lifetime (Andrii Nakryiko)
     - Kill xol_area->slot_count (Oleg Nesterov)
 
  - Core facilities:
     - Implement targeted high-frequency profiling by adding the ability
       for an event to "pause" or "resume" AUX area tracing (Adrian Hunter)
 
  - VM profiling/sampling:
     - Correct perf sampling with guest VMs (Colton Lewis)
 
  - New hardware support:
     - x86/intel: Add PMU support for Intel ArrowLake-H CPUs (Dapeng Mi)
 
  - Misc fixes and enhancements:
     - x86/intel/pt: Fix buffer full but size is 0 case (Adrian Hunter)
     - x86/amd: Warn only on new bits set (Breno Leitao)
     - x86/amd/uncore: Avoid a false positive warning about snprintf
                       truncation in amd_uncore_umc_ctx_init (Jean Delvare)
     - uprobes: Re-order struct uprobe_task to save some space (Christophe JAILLET)
     - x86/rapl: Move the pmu allocation out of CPU hotplug (Kan Liang)
     - x86/rapl: Clean up cpumask and hotplug (Kan Liang)
     - uprobes: Deuglify xol_get_insn_slot/xol_free_insn_slot paths (Oleg Nesterov)
 
 Signed-off-by: Ingo Molnar <mingo@kernel.org>
 -----BEGIN PGP SIGNATURE-----
 
 iQJFBAABCgAvFiEEBpT5eoXrXCwVQwEKEnMQ0APhK1gFAmc7eKERHG1pbmdvQGtl
 cm5lbC5vcmcACgkQEnMQ0APhK1i57A/+KQ6TrIoICVTE+BPlDfUw8NU+N3DagVb0
 dzoyDxlDRsnsYzeXZipPn+3IitX1w+DrGxBNIojSoiFVCLnHIKgo4uHbj7cVrR7J
 fBTVSnoJ94SGAk5ySebvLwMLce/YhXBeHK2lx6W/pI6acNcxzDfIabjjETeqltUo
 g7hmT9lo10pzZEZyuUfYX9khlWBxda1dKHc9pMIq7baeLe4iz/fCGlJ0K4d4M4z3
 NPZw239Np6iHUwu3Lcs4gNKe4rcDe7Bt47hpedemHe0Y+7c4s2HaPxbXWxvDtE76
 mlsg93i28f8SYxeV83pREn0EOCptXcljhiek+US+GR7NSbltMnV+uUiDfPKIE9+Y
 vYP/DYF9hx73FsOucEFrHxYYcePorn3pne5/khBYWdQU6TnlrBYWpoLQsjgCKTTR
 4JhCFlBZ5cDpc6ihtpwCwVTQ4Q/H7vM1XOlDwx0hPhcIPPHDreaQD/wxo61jBdXf
 PY0EPAxh3BcQxfPYuDS+XiYjQ8qO8MtXMKz5bZyHBZlbHwccV6T4ExjsLKxFk5As
 6BG8pkBWLg7drXAgVdleIY0ux+34w/Zzv7gemdlQxvWLlZrVvpjiG93oU3PTpZeq
 A2UD9eAOuXVD6+HsF/dmn88sFmcLWbrMskFWujkvhEUmCvSGAnz3YSS/mLEawBiT
 2xI8xykNWSY=
 =ItOT
 -----END PGP SIGNATURE-----

Merge tag 'perf-core-2024-11-18' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull performance events updates from Ingo Molnar:
 "Uprobes:
    - Add BPF session support (Jiri Olsa)
    - Switch to RCU Tasks Trace flavor for better performance (Andrii
      Nakryiko)
    - Massively increase uretprobe SMP scalability by SRCU-protecting
      the uretprobe lifetime (Andrii Nakryiko)
    - Kill xol_area->slot_count (Oleg Nesterov)

  Core facilities:
    - Implement targeted high-frequency profiling by adding the ability
      for an event to "pause" or "resume" AUX area tracing (Adrian
      Hunter)

  VM profiling/sampling:
    - Correct perf sampling with guest VMs (Colton Lewis)

  New hardware support:
    - x86/intel: Add PMU support for Intel ArrowLake-H CPUs (Dapeng Mi)

  Misc fixes and enhancements:
    - x86/intel/pt: Fix buffer full but size is 0 case (Adrian Hunter)
    - x86/amd: Warn only on new bits set (Breno Leitao)
    - x86/amd/uncore: Avoid a false positive warning about snprintf
      truncation in amd_uncore_umc_ctx_init (Jean Delvare)
    - uprobes: Re-order struct uprobe_task to save some space
      (Christophe JAILLET)
    - x86/rapl: Move the pmu allocation out of CPU hotplug (Kan Liang)
    - x86/rapl: Clean up cpumask and hotplug (Kan Liang)
    - uprobes: Deuglify xol_get_insn_slot/xol_free_insn_slot paths (Oleg
      Nesterov)"

* tag 'perf-core-2024-11-18' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (32 commits)
  perf/core: Correct perf sampling with guest VMs
  perf/x86: Refactor misc flag assignments
  perf/powerpc: Use perf_arch_instruction_pointer()
  perf/core: Hoist perf_instruction_pointer() and perf_misc_flags()
  perf/arm: Drop unused functions
  uprobes: Re-order struct uprobe_task to save some space
  perf/x86/amd/uncore: Avoid a false positive warning about snprintf truncation in amd_uncore_umc_ctx_init
  perf/x86/intel: Do not enable large PEBS for events with aux actions or aux sampling
  perf/x86/intel/pt: Add support for pause / resume
  perf/core: Add aux_pause, aux_resume, aux_start_paused
  perf/x86/intel/pt: Fix buffer full but size is 0 case
  uprobes: SRCU-protect uretprobe lifetime (with timeout)
  uprobes: allow put_uprobe() from non-sleepable softirq context
  perf/x86/rapl: Clean up cpumask and hotplug
  perf/x86/rapl: Move the pmu allocation out of CPU hotplug
  uprobe: Add support for session consumer
  uprobe: Add data pointer to consumer handlers
  perf/x86/amd: Warn only on new bits set
  uprobes: fold xol_take_insn_slot() into xol_get_insn_slot()
  uprobes: kill xol_area->slot_count
  ...

arch/Kconfig

@@ -135,6 +135,7 @@ config KPROBES_ON_FTRACE
 config UPROBES
 	def_bool n
 	depends on ARCH_SUPPORTS_UPROBES
+	select TASKS_TRACE_RCU
 	help
 	  Uprobes is the user-space counterpart to kprobes: they
 	  enable instrumentation applications (such as 'perf probe')

arch/arm/include/asm/perf_event.h

@@ -8,13 +8,6 @@
 #ifndef __ARM_PERF_EVENT_H__
 #define __ARM_PERF_EVENT_H__
 
-#ifdef CONFIG_PERF_EVENTS
-struct pt_regs;
-extern unsigned long perf_instruction_pointer(struct pt_regs *regs);
-extern unsigned long perf_misc_flags(struct pt_regs *regs);
-#define perf_misc_flags(regs)	perf_misc_flags(regs)
-#endif
-
 #define perf_arch_fetch_caller_regs(regs, __ip) { \
 	(regs)->ARM_pc = (__ip); \
 	frame_pointer((regs)) = (unsigned long) __builtin_frame_address(0); \

arch/arm/kernel/perf_callchain.c

@@ -96,20 +96,3 @@ perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *re
 	arm_get_current_stackframe(regs, &fr);
 	walk_stackframe(&fr, callchain_trace, entry);
 }
-
-unsigned long perf_instruction_pointer(struct pt_regs *regs)
-{
-	return instruction_pointer(regs);
-}
-
-unsigned long perf_misc_flags(struct pt_regs *regs)
-{
-	int misc = 0;
-
-	if (user_mode(regs))
-		misc |= PERF_RECORD_MISC_USER;
-	else
-		misc |= PERF_RECORD_MISC_KERNEL;
-
-	return misc;
-}

arch/arm64/include/asm/perf_event.h

@@ -10,10 +10,6 @@
 #include <asm/ptrace.h>
 
 #ifdef CONFIG_PERF_EVENTS
-struct pt_regs;
-extern unsigned long perf_instruction_pointer(struct pt_regs *regs);
-extern unsigned long perf_misc_flags(struct pt_regs *regs);
-#define perf_misc_flags(regs)	perf_misc_flags(regs)
 #define perf_arch_bpf_user_pt_regs(regs) &regs->user_regs
 #endif
 

arch/arm64/kernel/perf_callchain.c

@@ -38,31 +38,3 @@ void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry,
 
 	arch_stack_walk(callchain_trace, entry, current, regs);
 }
-
-unsigned long perf_instruction_pointer(struct pt_regs *regs)
-{
-	if (perf_guest_state())
-		return perf_guest_get_ip();
-
-	return instruction_pointer(regs);
-}
-
-unsigned long perf_misc_flags(struct pt_regs *regs)
-{
-	unsigned int guest_state = perf_guest_state();
-	int misc = 0;
-
-	if (guest_state) {
-		if (guest_state & PERF_GUEST_USER)
-			misc |= PERF_RECORD_MISC_GUEST_USER;
-		else
-			misc |= PERF_RECORD_MISC_GUEST_KERNEL;
-	} else {
-		if (user_mode(regs))
-			misc |= PERF_RECORD_MISC_USER;
-		else
-			misc |= PERF_RECORD_MISC_KERNEL;
-	}
-
-	return misc;
-}

arch/powerpc/include/asm/perf_event_server.h

@@ -102,8 +102,8 @@ struct power_pmu {
 int __init register_power_pmu(struct power_pmu *pmu);
 
 struct pt_regs;
-extern unsigned long perf_misc_flags(struct pt_regs *regs);
+extern unsigned long perf_arch_misc_flags(struct pt_regs *regs);
-extern unsigned long perf_instruction_pointer(struct pt_regs *regs);
+extern unsigned long perf_arch_instruction_pointer(struct pt_regs *regs);
 extern unsigned long int read_bhrb(int n);
 
 /*
@@ -111,7 +111,7 @@ extern unsigned long int read_bhrb(int n);
  * if we have hardware PMU support.
  */
 #ifdef CONFIG_PPC_PERF_CTRS
-#define perf_misc_flags(regs)	perf_misc_flags(regs)
+#define perf_arch_misc_flags(regs)	perf_arch_misc_flags(regs)
 #endif
 
 /*

arch/powerpc/perf/callchain.c

@@ -51,7 +51,7 @@ perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *re
 
 	lr = regs->link;
 	sp = regs->gpr[1];
-	perf_callchain_store(entry, perf_instruction_pointer(regs));
+	perf_callchain_store(entry, perf_arch_instruction_pointer(regs));
 
 	if (!validate_sp(sp, current))
 		return;

arch/powerpc/perf/callchain_32.c

@@ -139,7 +139,7 @@ void perf_callchain_user_32(struct perf_callchain_entry_ctx *entry,
 	long level = 0;
 	unsigned int __user *fp, *uregs;
 
-	next_ip = perf_instruction_pointer(regs);
+	next_ip = perf_arch_instruction_pointer(regs);
 	lr = regs->link;
 	sp = regs->gpr[1];
 	perf_callchain_store(entry, next_ip);

arch/powerpc/perf/callchain_64.c

@@ -74,7 +74,7 @@ void perf_callchain_user_64(struct perf_callchain_entry_ctx *entry,
 	struct signal_frame_64 __user *sigframe;
 	unsigned long __user *fp, *uregs;
 
-	next_ip = perf_instruction_pointer(regs);
+	next_ip = perf_arch_instruction_pointer(regs);
 	lr = regs->link;
 	sp = regs->gpr[1];
 	perf_callchain_store(entry, next_ip);

arch/powerpc/perf/core-book3s.c

@@ -2332,7 +2332,7 @@ static void record_and_restart(struct perf_event *event, unsigned long val,
  * Called from generic code to get the misc flags (i.e. processor mode)
  * for an event_id.
  */
-unsigned long perf_misc_flags(struct pt_regs *regs)
+unsigned long perf_arch_misc_flags(struct pt_regs *regs)
 {
 	u32 flags = perf_get_misc_flags(regs);
 
@@ -2346,7 +2346,7 @@ unsigned long perf_misc_flags(struct pt_regs *regs)
  * Called from generic code to get the instruction pointer
  * for an event_id.
  */
-unsigned long perf_instruction_pointer(struct pt_regs *regs)
+unsigned long perf_arch_instruction_pointer(struct pt_regs *regs)
 {
 	unsigned long siar = mfspr(SPRN_SIAR);
 

arch/s390/include/asm/perf_event.h

@@ -37,9 +37,9 @@ extern ssize_t cpumf_events_sysfs_show(struct device *dev,
 
 /* Perf callbacks */
 struct pt_regs;
-extern unsigned long perf_instruction_pointer(struct pt_regs *regs);
+extern unsigned long perf_arch_instruction_pointer(struct pt_regs *regs);
-extern unsigned long perf_misc_flags(struct pt_regs *regs);
+extern unsigned long perf_arch_misc_flags(struct pt_regs *regs);
-#define perf_misc_flags(regs) perf_misc_flags(regs)
+#define perf_arch_misc_flags(regs) perf_arch_misc_flags(regs)
 #define perf_arch_bpf_user_pt_regs(regs) &regs->user_regs
 
 /* Perf pt_regs extension for sample-data-entry indicators */

arch/s390/kernel/perf_event.c

@@ -57,7 +57,7 @@ static unsigned long instruction_pointer_guest(struct pt_regs *regs)
 	return sie_block(regs)->gpsw.addr;
 }
 
-unsigned long perf_instruction_pointer(struct pt_regs *regs)
+unsigned long perf_arch_instruction_pointer(struct pt_regs *regs)
 {
 	return is_in_guest(regs) ? instruction_pointer_guest(regs)
 				 : instruction_pointer(regs);
@@ -84,7 +84,7 @@ static unsigned long perf_misc_flags_sf(struct pt_regs *regs)
 	return flags;
 }
 
-unsigned long perf_misc_flags(struct pt_regs *regs)
+unsigned long perf_arch_misc_flags(struct pt_regs *regs)
 {
 	/* Check if the cpum_sf PMU has created the pt_regs structure.
 	 * In this case, perf misc flags can be easily extracted.  Otherwise,

arch/x86/events/amd/core.c

@@ -943,11 +943,12 @@ static int amd_pmu_v2_snapshot_branch_stack(struct perf_branch_entry *entries, u
 static int amd_pmu_v2_handle_irq(struct pt_regs *regs)
 {
 	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
+	static atomic64_t status_warned = ATOMIC64_INIT(0);
+	u64 reserved, status, mask, new_bits, prev_bits;
 	struct perf_sample_data data;
 	struct hw_perf_event *hwc;
 	struct perf_event *event;
 	int handled = 0, idx;
-	u64 reserved, status, mask;
 	bool pmu_enabled;
 
 	/*
@@ -1012,7 +1013,12 @@ static int amd_pmu_v2_handle_irq(struct pt_regs *regs)
 	 * the corresponding PMCs are expected to be inactive according to the
 	 * active_mask
 	 */
-	WARN_ON(status > 0);
+	if (status > 0) {
+		prev_bits = atomic64_fetch_or(status, &status_warned);
+		// A new bit was set for the very first time.
+		new_bits = status & ~prev_bits;
+		WARN(new_bits, "New overflows for inactive PMCs: %llx\n", new_bits);
+	}
 
 	/* Clear overflow and freeze bits */
 	amd_pmu_ack_global_status(~status);

arch/x86/events/amd/uncore.c

@@ -916,7 +916,8 @@ int amd_uncore_umc_ctx_init(struct amd_uncore *uncore, unsigned int cpu)
 	u8 group_num_pmcs[UNCORE_GROUP_MAX] = { 0 };
 	union amd_uncore_info info;
 	struct amd_uncore_pmu *pmu;
-	int index = 0, gid, i;
+	int gid, i;
+	u16 index = 0;
 
 	if (pmu_version < 2)
 		return 0;
@@ -948,7 +949,7 @@ int amd_uncore_umc_ctx_init(struct amd_uncore *uncore, unsigned int cpu)
 	for_each_set_bit(gid, gmask, UNCORE_GROUP_MAX) {
 		for (i = 0; i < group_num_pmus[gid]; i++) {
 			pmu = &uncore->pmus[index];
-			snprintf(pmu->name, sizeof(pmu->name), "amd_umc_%d", index);
+			snprintf(pmu->name, sizeof(pmu->name), "amd_umc_%hu", index);
 			pmu->num_counters = group_num_pmcs[gid] / group_num_pmus[gid];
 			pmu->msr_base = MSR_F19H_UMC_PERF_CTL + i * pmu->num_counters * 2;
 			pmu->rdpmc_base = -1;

arch/x86/events/core.c

@@ -3003,35 +3003,57 @@ static unsigned long code_segment_base(struct pt_regs *regs)
 	return 0;
 }
 
-unsigned long perf_instruction_pointer(struct pt_regs *regs)
+unsigned long perf_arch_instruction_pointer(struct pt_regs *regs)
 {
-	if (perf_guest_state())
-		return perf_guest_get_ip();
-
 	return regs->ip + code_segment_base(regs);
 }
 
-unsigned long perf_misc_flags(struct pt_regs *regs)
+static unsigned long common_misc_flags(struct pt_regs *regs)
 {
-	unsigned int guest_state = perf_guest_state();
-	int misc = 0;
-
-	if (guest_state) {
-		if (guest_state & PERF_GUEST_USER)
-			misc |= PERF_RECORD_MISC_GUEST_USER;
-		else
-			misc |= PERF_RECORD_MISC_GUEST_KERNEL;
-	} else {
-		if (user_mode(regs))
-			misc |= PERF_RECORD_MISC_USER;
-		else
-			misc |= PERF_RECORD_MISC_KERNEL;
-	}
-
 	if (regs->flags & PERF_EFLAGS_EXACT)
-		misc |= PERF_RECORD_MISC_EXACT_IP;
+		return PERF_RECORD_MISC_EXACT_IP;
 
-	return misc;
+	return 0;
+}
+
+static unsigned long guest_misc_flags(struct pt_regs *regs)
+{
+	unsigned long guest_state = perf_guest_state();
+
+	if (!(guest_state & PERF_GUEST_ACTIVE))
+		return 0;
+
+	if (guest_state & PERF_GUEST_USER)
+		return PERF_RECORD_MISC_GUEST_USER;
+	else
+		return PERF_RECORD_MISC_GUEST_KERNEL;
+
+}
+
+static unsigned long host_misc_flags(struct pt_regs *regs)
+{
+	if (user_mode(regs))
+		return PERF_RECORD_MISC_USER;
+	else
+		return PERF_RECORD_MISC_KERNEL;
+}
+
+unsigned long perf_arch_guest_misc_flags(struct pt_regs *regs)
+{
+	unsigned long flags = common_misc_flags(regs);
+
+	flags |= guest_misc_flags(regs);
+
+	return flags;
+}
+
+unsigned long perf_arch_misc_flags(struct pt_regs *regs)
+{
+	unsigned long flags = common_misc_flags(regs);
+
+	flags |= host_misc_flags(regs);
+
+	return flags;
 }
 
 void perf_get_x86_pmu_capability(struct x86_pmu_capability *cap)

arch/x86/events/intel/core.c

@@ -3962,8 +3962,8 @@ static int intel_pmu_hw_config(struct perf_event *event)
 
 		if (!(event->attr.freq || (event->attr.wakeup_events && !event->attr.watermark))) {
 			event->hw.flags |= PERF_X86_EVENT_AUTO_RELOAD;
-			if (!(event->attr.sample_type &
+			if (!(event->attr.sample_type & ~intel_pmu_large_pebs_flags(event)) &&
-			      ~intel_pmu_large_pebs_flags(event))) {
+			    !has_aux_action(event)) {
 				event->hw.flags |= PERF_X86_EVENT_LARGE_PEBS;
 				event->attach_state |= PERF_ATTACH_SCHED_CB;
 			}
@@ -4599,6 +4599,28 @@ static inline bool erratum_hsw11(struct perf_event *event)
 		X86_CONFIG(.event=0xc0, .umask=0x01);
 }
 
+static struct event_constraint *
+arl_h_get_event_constraints(struct cpu_hw_events *cpuc, int idx,
+			  struct perf_event *event)
+{
+	struct x86_hybrid_pmu *pmu = hybrid_pmu(event->pmu);
+
+	if (pmu->pmu_type == hybrid_tiny)
+		return cmt_get_event_constraints(cpuc, idx, event);
+
+	return mtl_get_event_constraints(cpuc, idx, event);
+}
+
+static int arl_h_hw_config(struct perf_event *event)
+{
+	struct x86_hybrid_pmu *pmu = hybrid_pmu(event->pmu);
+
+	if (pmu->pmu_type == hybrid_tiny)
+		return intel_pmu_hw_config(event);
+
+	return adl_hw_config(event);
+}
+
 /*
  * The HSW11 requires a period larger than 100 which is the same as the BDM11.
  * A minimum period of 128 is enforced as well for the INST_RETIRED.ALL.
@@ -4924,17 +4946,26 @@ static struct x86_hybrid_pmu *find_hybrid_pmu_for_cpu(void)
 
 	/*
 	 * This essentially just maps between the 'hybrid_cpu_type'
-	 * and 'hybrid_pmu_type' enums:
+	 * and 'hybrid_pmu_type' enums except for ARL-H processor
+	 * which needs to compare atom uarch native id since ARL-H
+	 * contains two different atom uarchs.
 	 */
 	for (i = 0; i < x86_pmu.num_hybrid_pmus; i++) {
 		enum hybrid_pmu_type pmu_type = x86_pmu.hybrid_pmu[i].pmu_type;
+		u32 native_id;
 
-		if (cpu_type == HYBRID_INTEL_CORE &&
+		if (cpu_type == HYBRID_INTEL_CORE && pmu_type == hybrid_big)
-		    pmu_type == hybrid_big)
 			return &x86_pmu.hybrid_pmu[i];
-		if (cpu_type == HYBRID_INTEL_ATOM &&
+		if (cpu_type == HYBRID_INTEL_ATOM) {
-		    pmu_type == hybrid_small)
+			if (x86_pmu.num_hybrid_pmus == 2 && pmu_type == hybrid_small)
 				return &x86_pmu.hybrid_pmu[i];
+
+			native_id = get_this_hybrid_cpu_native_id();
+			if (native_id == skt_native_id && pmu_type == hybrid_small)
+				return &x86_pmu.hybrid_pmu[i];
+			if (native_id == cmt_native_id && pmu_type == hybrid_tiny)
+				return &x86_pmu.hybrid_pmu[i];
+		}
 	}
 
 	return NULL;
@@ -5965,6 +5996,37 @@ static struct attribute *lnl_hybrid_events_attrs[] = {
 	NULL
 };
 
+/* The event string must be in PMU IDX order. */
+EVENT_ATTR_STR_HYBRID(topdown-retiring,
+		      td_retiring_arl_h,
+		      "event=0xc2,umask=0x02;event=0x00,umask=0x80;event=0xc2,umask=0x0",
+		      hybrid_big_small_tiny);
+EVENT_ATTR_STR_HYBRID(topdown-bad-spec,
+		      td_bad_spec_arl_h,
+		      "event=0x73,umask=0x0;event=0x00,umask=0x81;event=0x73,umask=0x0",
+		      hybrid_big_small_tiny);
+EVENT_ATTR_STR_HYBRID(topdown-fe-bound,
+		      td_fe_bound_arl_h,
+		      "event=0x9c,umask=0x01;event=0x00,umask=0x82;event=0x71,umask=0x0",
+		      hybrid_big_small_tiny);
+EVENT_ATTR_STR_HYBRID(topdown-be-bound,
+		      td_be_bound_arl_h,
+		      "event=0xa4,umask=0x02;event=0x00,umask=0x83;event=0x74,umask=0x0",
+		      hybrid_big_small_tiny);
+
+static struct attribute *arl_h_hybrid_events_attrs[] = {
+	EVENT_PTR(slots_adl),
+	EVENT_PTR(td_retiring_arl_h),
+	EVENT_PTR(td_bad_spec_arl_h),
+	EVENT_PTR(td_fe_bound_arl_h),
+	EVENT_PTR(td_be_bound_arl_h),
+	EVENT_PTR(td_heavy_ops_adl),
+	EVENT_PTR(td_br_mis_adl),
+	EVENT_PTR(td_fetch_lat_adl),
+	EVENT_PTR(td_mem_bound_adl),
+	NULL,
+};
+
 /* Must be in IDX order */
 EVENT_ATTR_STR_HYBRID(mem-loads,     mem_ld_adl,     "event=0xd0,umask=0x5,ldlat=3;event=0xcd,umask=0x1,ldlat=3", hybrid_big_small);
 EVENT_ATTR_STR_HYBRID(mem-stores,    mem_st_adl,     "event=0xd0,umask=0x6;event=0xcd,umask=0x2",                 hybrid_big_small);
@@ -5983,6 +6045,21 @@ static struct attribute *mtl_hybrid_mem_attrs[] = {
 	NULL
 };
 
+EVENT_ATTR_STR_HYBRID(mem-loads,
+		      mem_ld_arl_h,
+		      "event=0xd0,umask=0x5,ldlat=3;event=0xcd,umask=0x1,ldlat=3;event=0xd0,umask=0x5,ldlat=3",
+		      hybrid_big_small_tiny);
+EVENT_ATTR_STR_HYBRID(mem-stores,
+		      mem_st_arl_h,
+		      "event=0xd0,umask=0x6;event=0xcd,umask=0x2;event=0xd0,umask=0x6",
+		      hybrid_big_small_tiny);
+
+static struct attribute *arl_h_hybrid_mem_attrs[] = {
+	EVENT_PTR(mem_ld_arl_h),
+	EVENT_PTR(mem_st_arl_h),
+	NULL,
+};
+
 EVENT_ATTR_STR_HYBRID(tx-start,          tx_start_adl,          "event=0xc9,umask=0x1",          hybrid_big);
 EVENT_ATTR_STR_HYBRID(tx-commit,         tx_commit_adl,         "event=0xc9,umask=0x2",          hybrid_big);
 EVENT_ATTR_STR_HYBRID(tx-abort,          tx_abort_adl,          "event=0xc9,umask=0x4",          hybrid_big);
@@ -6006,8 +6083,8 @@ static struct attribute *adl_hybrid_tsx_attrs[] = {
 
 FORMAT_ATTR_HYBRID(in_tx,       hybrid_big);
 FORMAT_ATTR_HYBRID(in_tx_cp,    hybrid_big);
-FORMAT_ATTR_HYBRID(offcore_rsp, hybrid_big_small);
+FORMAT_ATTR_HYBRID(offcore_rsp, hybrid_big_small_tiny);
-FORMAT_ATTR_HYBRID(ldlat,       hybrid_big_small);
+FORMAT_ATTR_HYBRID(ldlat,       hybrid_big_small_tiny);
 FORMAT_ATTR_HYBRID(frontend,    hybrid_big);
 
 #define ADL_HYBRID_RTM_FORMAT_ATTR	\
@@ -6030,7 +6107,7 @@ static struct attribute *adl_hybrid_extra_attr[] = {
 	NULL
 };
 
-FORMAT_ATTR_HYBRID(snoop_rsp,	hybrid_small);
+FORMAT_ATTR_HYBRID(snoop_rsp,	hybrid_small_tiny);
 
 static struct attribute *mtl_hybrid_extra_attr_rtm[] = {
 	ADL_HYBRID_RTM_FORMAT_ATTR,
@@ -6240,6 +6317,7 @@ static inline int intel_pmu_v6_addr_offset(int index, bool eventsel)
 static const struct { enum hybrid_pmu_type id; char *name; } intel_hybrid_pmu_type_map[] __initconst = {
 	{ hybrid_small,	"cpu_atom" },
 	{ hybrid_big,	"cpu_core" },
+	{ hybrid_tiny,	"cpu_lowpower" },
 };
 
 static __always_inline int intel_pmu_init_hybrid(enum hybrid_pmu_type pmus)
@@ -6272,7 +6350,7 @@ static __always_inline int intel_pmu_init_hybrid(enum hybrid_pmu_type pmus)
 							0, x86_pmu_num_counters(&pmu->pmu), 0, 0);
 
 		pmu->intel_cap.capabilities = x86_pmu.intel_cap.capabilities;
-		if (pmu->pmu_type & hybrid_small) {
+		if (pmu->pmu_type & hybrid_small_tiny) {
 			pmu->intel_cap.perf_metrics = 0;
 			pmu->intel_cap.pebs_output_pt_available = 1;
 			pmu->mid_ack = true;
@@ -7111,6 +7189,37 @@ __init int intel_pmu_init(void)
 		name = "lunarlake_hybrid";
 		break;
 
+	case INTEL_ARROWLAKE_H:
+		intel_pmu_init_hybrid(hybrid_big_small_tiny);
+
+		x86_pmu.pebs_latency_data = arl_h_latency_data;
+		x86_pmu.get_event_constraints = arl_h_get_event_constraints;
+		x86_pmu.hw_config = arl_h_hw_config;
+
+		td_attr = arl_h_hybrid_events_attrs;
+		mem_attr = arl_h_hybrid_mem_attrs;
+		tsx_attr = adl_hybrid_tsx_attrs;
+		extra_attr = boot_cpu_has(X86_FEATURE_RTM) ?
+			mtl_hybrid_extra_attr_rtm : mtl_hybrid_extra_attr;
+
+		/* Initialize big core specific PerfMon capabilities. */
+		pmu = &x86_pmu.hybrid_pmu[X86_HYBRID_PMU_CORE_IDX];
+		intel_pmu_init_lnc(&pmu->pmu);
+
+		/* Initialize Atom core specific PerfMon capabilities. */
+		pmu = &x86_pmu.hybrid_pmu[X86_HYBRID_PMU_ATOM_IDX];
+		intel_pmu_init_skt(&pmu->pmu);
+
+		/* Initialize Lower Power Atom specific PerfMon capabilities. */
+		pmu = &x86_pmu.hybrid_pmu[X86_HYBRID_PMU_TINY_IDX];
+		intel_pmu_init_grt(&pmu->pmu);
+		pmu->extra_regs = intel_cmt_extra_regs;
+
+		intel_pmu_pebs_data_source_arl_h();
+		pr_cont("ArrowLake-H Hybrid events, ");
+		name = "arrowlake_h_hybrid";
+		break;
+
 	default:
 		switch (x86_pmu.version) {
 		case 1:

arch/x86/events/intel/ds.c

@@ -177,6 +177,17 @@ void __init intel_pmu_pebs_data_source_mtl(void)
 	__intel_pmu_pebs_data_source_cmt(data_source);
 }
 
+void __init intel_pmu_pebs_data_source_arl_h(void)
+{
+	u64 *data_source;
+
+	intel_pmu_pebs_data_source_lnl();
+
+	data_source = x86_pmu.hybrid_pmu[X86_HYBRID_PMU_TINY_IDX].pebs_data_source;
+	memcpy(data_source, pebs_data_source, sizeof(pebs_data_source));
+	__intel_pmu_pebs_data_source_cmt(data_source);
+}
+
 void __init intel_pmu_pebs_data_source_cmt(void)
 {
 	__intel_pmu_pebs_data_source_cmt(pebs_data_source);
@@ -388,6 +399,16 @@ u64 lnl_latency_data(struct perf_event *event, u64 status)
 	return lnc_latency_data(event, status);
 }
 
+u64 arl_h_latency_data(struct perf_event *event, u64 status)
+{
+	struct x86_hybrid_pmu *pmu = hybrid_pmu(event->pmu);
+
+	if (pmu->pmu_type == hybrid_tiny)
+		return cmt_latency_data(event, status);
+
+	return lnl_latency_data(event, status);
+}
+
 static u64 load_latency_data(struct perf_event *event, u64 status)
 {
 	union intel_x86_pebs_dse dse;

arch/x86/events/intel/pt.c

@@ -418,6 +418,9 @@ static void pt_config_start(struct perf_event *event)
 	struct pt *pt = this_cpu_ptr(&pt_ctx);
 	u64 ctl = event->hw.aux_config;
 
+	if (READ_ONCE(event->hw.aux_paused))
+		return;
+
 	ctl |= RTIT_CTL_TRACEEN;
 	if (READ_ONCE(pt->vmx_on))
 		perf_aux_output_flag(&pt->handle, PERF_AUX_FLAG_PARTIAL);
@@ -534,7 +537,24 @@ static void pt_config(struct perf_event *event)
 	reg |= (event->attr.config & PT_CONFIG_MASK);
 
 	event->hw.aux_config = reg;
+
+	/*
+	 * Allow resume before starting so as not to overwrite a value set by a
+	 * PMI.
+	 */
+	barrier();
+	WRITE_ONCE(pt->resume_allowed, 1);
+	/* Configuration is complete, it is now OK to handle an NMI */
+	barrier();
+	WRITE_ONCE(pt->handle_nmi, 1);
+	barrier();
 	pt_config_start(event);
+	barrier();
+	/*
+	 * Allow pause after starting so its pt_config_stop() doesn't race with
+	 * pt_config_start().
+	 */
+	WRITE_ONCE(pt->pause_allowed, 1);
 }
 
 static void pt_config_stop(struct perf_event *event)
@@ -828,11 +848,13 @@ static void pt_buffer_advance(struct pt_buffer *buf)
 	buf->cur_idx++;
 
 	if (buf->cur_idx == buf->cur->last) {
-		if (buf->cur == buf->last)
+		if (buf->cur == buf->last) {
 			buf->cur = buf->first;
-		else
+			buf->wrapped = true;
+		} else {
 			buf->cur = list_entry(buf->cur->list.next, struct topa,
 					      list);
+		}
 		buf->cur_idx = 0;
 	}
 }
@@ -846,8 +868,11 @@ static void pt_buffer_advance(struct pt_buffer *buf)
 static void pt_update_head(struct pt *pt)
 {
 	struct pt_buffer *buf = perf_get_aux(&pt->handle);
+	bool wrapped = buf->wrapped;
 	u64 topa_idx, base, old;
 
+	buf->wrapped = false;
+
 	if (buf->single) {
 		local_set(&buf->data_size, buf->output_off);
 		return;
@@ -865,7 +890,7 @@ static void pt_update_head(struct pt *pt)
 	} else {
 		old = (local64_xchg(&buf->head, base) &
 		       ((buf->nr_pages << PAGE_SHIFT) - 1));
-		if (base < old)
+		if (base < old || (base == old && wrapped))
 			base += buf->nr_pages << PAGE_SHIFT;
 
 		local_add(base - old, &buf->data_size);
@@ -1511,6 +1536,7 @@ void intel_pt_interrupt(void)
 		buf = perf_aux_output_begin(&pt->handle, event);
 		if (!buf) {
 			event->hw.state = PERF_HES_STOPPED;
+			WRITE_ONCE(pt->resume_allowed, 0);
 			return;
 		}
 
@@ -1519,6 +1545,7 @@ void intel_pt_interrupt(void)
 		ret = pt_buffer_reset_markers(buf, &pt->handle);
 		if (ret) {
 			perf_aux_output_end(&pt->handle, 0);
+			WRITE_ONCE(pt->resume_allowed, 0);
 			return;
 		}
 
@@ -1573,6 +1600,26 @@ static void pt_event_start(struct perf_event *event, int mode)
 	struct pt *pt = this_cpu_ptr(&pt_ctx);
 	struct pt_buffer *buf;
 
+	if (mode & PERF_EF_RESUME) {
+		if (READ_ONCE(pt->resume_allowed)) {
+			u64 status;
+
+			/*
+			 * Only if the trace is not active and the error and
+			 * stopped bits are clear, is it safe to start, but a
+			 * PMI might have just cleared these, so resume_allowed
+			 * must be checked again also.
+			 */
+			rdmsrl(MSR_IA32_RTIT_STATUS, status);
+			if (!(status & (RTIT_STATUS_TRIGGEREN |
+					RTIT_STATUS_ERROR |
+					RTIT_STATUS_STOPPED)) &&
+			   READ_ONCE(pt->resume_allowed))
+				pt_config_start(event);
+		}
+		return;
+	}
+
 	buf = perf_aux_output_begin(&pt->handle, event);
 	if (!buf)
 		goto fail_stop;
@@ -1583,7 +1630,6 @@ static void pt_event_start(struct perf_event *event, int mode)
 			goto fail_end_stop;
 	}
 
-	WRITE_ONCE(pt->handle_nmi, 1);
 	hwc->state = 0;
 
 	pt_config_buffer(buf);
@@ -1601,6 +1647,12 @@ static void pt_event_stop(struct perf_event *event, int mode)
 {
 	struct pt *pt = this_cpu_ptr(&pt_ctx);
 
+	if (mode & PERF_EF_PAUSE) {
+		if (READ_ONCE(pt->pause_allowed))
+			pt_config_stop(event);
+		return;
+	}
+
 	/*
 	 * Protect against the PMI racing with disabling wrmsr,
 	 * see comment in intel_pt_interrupt().
@@ -1608,6 +1660,15 @@ static void pt_event_stop(struct perf_event *event, int mode)
 	WRITE_ONCE(pt->handle_nmi, 0);
 	barrier();
 
+	/*
+	 * Prevent a resume from attempting to restart tracing, or a pause
+	 * during a subsequent start. Do this after clearing handle_nmi so that
+	 * pt_event_snapshot_aux() will not re-allow them.
+	 */
+	WRITE_ONCE(pt->pause_allowed, 0);
+	WRITE_ONCE(pt->resume_allowed, 0);
+	barrier();
+
 	pt_config_stop(event);
 
 	if (event->hw.state == PERF_HES_STOPPED)
@@ -1657,6 +1718,10 @@ static long pt_event_snapshot_aux(struct perf_event *event,
 	if (WARN_ON_ONCE(!buf->snapshot))
 		return 0;
 
+	/* Prevent pause/resume from attempting to start/stop tracing */
+	WRITE_ONCE(pt->pause_allowed, 0);
+	WRITE_ONCE(pt->resume_allowed, 0);
+	barrier();
 	/*
 	 * There is no PT interrupt in this mode, so stop the trace and it will
 	 * remain stopped while the buffer is copied.
@@ -1676,8 +1741,13 @@ static long pt_event_snapshot_aux(struct perf_event *event,
 	 * Here, handle_nmi tells us if the tracing was on.
 	 * If the tracing was on, restart it.
 	 */
-	if (READ_ONCE(pt->handle_nmi))
+	if (READ_ONCE(pt->handle_nmi)) {
+		WRITE_ONCE(pt->resume_allowed, 1);
+		barrier();
 		pt_config_start(event);
+		barrier();
+		WRITE_ONCE(pt->pause_allowed, 1);
+	}
 
 	return ret;
 }
@@ -1793,7 +1863,9 @@ static __init int pt_init(void)
 	if (!intel_pt_validate_hw_cap(PT_CAP_topa_multiple_entries))
 		pt_pmu.pmu.capabilities = PERF_PMU_CAP_AUX_NO_SG;
 
-	pt_pmu.pmu.capabilities	|= PERF_PMU_CAP_EXCLUSIVE | PERF_PMU_CAP_ITRACE;
+	pt_pmu.pmu.capabilities		|= PERF_PMU_CAP_EXCLUSIVE |
+					   PERF_PMU_CAP_ITRACE |
+					   PERF_PMU_CAP_AUX_PAUSE;
 	pt_pmu.pmu.attr_groups		 = pt_attr_groups;
 	pt_pmu.pmu.task_ctx_nr		 = perf_sw_context;
 	pt_pmu.pmu.event_init		 = pt_event_init;

arch/x86/events/intel/pt.h

@@ -65,6 +65,7 @@ struct pt_pmu {
  * @head:	logical write offset inside the buffer
  * @snapshot:	if this is for a snapshot/overwrite counter
  * @single:	use Single Range Output instead of ToPA
+ * @wrapped:	buffer advance wrapped back to the first topa table
  * @stop_pos:	STOP topa entry index
  * @intr_pos:	INT topa entry index
  * @stop_te:	STOP topa entry pointer
@@ -82,6 +83,7 @@ struct pt_buffer {
 	local64_t		head;
 	bool			snapshot;
 	bool			single;
+	bool			wrapped;
 	long			stop_pos, intr_pos;
 	struct topa_entry	*stop_te, *intr_te;
 	void			**data_pages;
@@ -117,6 +119,8 @@ struct pt_filters {
  * @filters:		last configured filters
  * @handle_nmi:		do handle PT PMI on this cpu, there's an active event
  * @vmx_on:		1 if VMX is ON on this cpu
+ * @pause_allowed:	PERF_EF_PAUSE is allowed to stop tracing
+ * @resume_allowed:	PERF_EF_RESUME is allowed to start tracing
  * @output_base:	cached RTIT_OUTPUT_BASE MSR value
  * @output_mask:	cached RTIT_OUTPUT_MASK MSR value
  */
@@ -125,6 +129,8 @@ struct pt {
 	struct pt_filters	filters;
 	int			handle_nmi;
 	int			vmx_on;
+	int			pause_allowed;
+	int			resume_allowed;
 	u64			output_base;
 	u64			output_mask;
 };

arch/x86/events/perf_event.h

@@ -668,24 +668,38 @@ enum {
 #define PERF_PEBS_DATA_SOURCE_GRT_MAX	0x10
 #define PERF_PEBS_DATA_SOURCE_GRT_MASK	(PERF_PEBS_DATA_SOURCE_GRT_MAX - 1)
 
+/*
+ * CPUID.1AH.EAX[31:0] uniquely identifies the microarchitecture
+ * of the core. Bits 31-24 indicates its core type (Core or Atom)
+ * and Bits [23:0] indicates the native model ID of the core.
+ * Core type and native model ID are defined in below enumerations.
+ */
 enum hybrid_cpu_type {
 	HYBRID_INTEL_NONE,
 	HYBRID_INTEL_ATOM	= 0x20,
 	HYBRID_INTEL_CORE	= 0x40,
 };
 
-enum hybrid_pmu_type {
-	not_hybrid,
-	hybrid_small		= BIT(0),
-	hybrid_big		= BIT(1),
-
-	hybrid_big_small	= hybrid_big | hybrid_small, /* only used for matching */
-};
-
 #define X86_HYBRID_PMU_ATOM_IDX		0
 #define X86_HYBRID_PMU_CORE_IDX		1
+#define X86_HYBRID_PMU_TINY_IDX		2
 
-#define X86_HYBRID_NUM_PMUS		2
+enum hybrid_pmu_type {
+	not_hybrid,
+	hybrid_small		= BIT(X86_HYBRID_PMU_ATOM_IDX),
+	hybrid_big		= BIT(X86_HYBRID_PMU_CORE_IDX),
+	hybrid_tiny		= BIT(X86_HYBRID_PMU_TINY_IDX),
+
+	/* The belows are only used for matching */
+	hybrid_big_small	= hybrid_big   | hybrid_small,
+	hybrid_small_tiny	= hybrid_small | hybrid_tiny,
+	hybrid_big_small_tiny	= hybrid_big   | hybrid_small_tiny,
+};
+
+enum atom_native_id {
+	cmt_native_id           = 0x2,  /* Crestmont */
+	skt_native_id           = 0x3,  /* Skymont */
+};
 
 struct x86_hybrid_pmu {
 	struct pmu			pmu;
@@ -1578,6 +1592,8 @@ u64 cmt_latency_data(struct perf_event *event, u64 status);
 
 u64 lnl_latency_data(struct perf_event *event, u64 status);
 
+u64 arl_h_latency_data(struct perf_event *event, u64 status);
+
 extern struct event_constraint intel_core2_pebs_event_constraints[];
 
 extern struct event_constraint intel_atom_pebs_event_constraints[];
@@ -1697,6 +1713,8 @@ void intel_pmu_pebs_data_source_grt(void);
 
 void intel_pmu_pebs_data_source_mtl(void);
 
+void intel_pmu_pebs_data_source_arl_h(void);
+
 void intel_pmu_pebs_data_source_cmt(void);
 
 void intel_pmu_pebs_data_source_lnl(void);

arch/x86/events/rapl.c

@@ -148,7 +148,6 @@ struct rapl_model {
  /* 1/2^hw_unit Joule */
 static int rapl_hw_unit[NR_RAPL_DOMAINS] __read_mostly;
 static struct rapl_pmus *rapl_pmus;
-static cpumask_t rapl_cpu_mask;
 static unsigned int rapl_cntr_mask;
 static u64 rapl_timer_ms;
 static struct perf_msr *rapl_msrs;
@@ -369,8 +368,6 @@ static int rapl_pmu_event_init(struct perf_event *event)
 	if (event->cpu < 0)
 		return -EINVAL;
 
-	event->event_caps |= PERF_EV_CAP_READ_ACTIVE_PKG;
-
 	if (!cfg || cfg >= NR_RAPL_DOMAINS + 1)
 		return -EINVAL;
 
@@ -389,7 +386,6 @@ static int rapl_pmu_event_init(struct perf_event *event)
 	pmu = cpu_to_rapl_pmu(event->cpu);
 	if (!pmu)
 		return -EINVAL;
-	event->cpu = pmu->cpu;
 	event->pmu_private = pmu;
 	event->hw.event_base = rapl_msrs[bit].msr;
 	event->hw.config = cfg;
@@ -403,23 +399,6 @@ static void rapl_pmu_event_read(struct perf_event *event)
 	rapl_event_update(event);
 }
 
-static ssize_t rapl_get_attr_cpumask(struct device *dev,
-				struct device_attribute *attr, char *buf)
-{
-	return cpumap_print_to_pagebuf(true, buf, &rapl_cpu_mask);
-}
-
-static DEVICE_ATTR(cpumask, S_IRUGO, rapl_get_attr_cpumask, NULL);
-
-static struct attribute *rapl_pmu_attrs[] = {
-	&dev_attr_cpumask.attr,
-	NULL,
-};
-
-static struct attribute_group rapl_pmu_attr_group = {
-	.attrs = rapl_pmu_attrs,
-};
-
 RAPL_EVENT_ATTR_STR(energy-cores, rapl_cores, "event=0x01");
 RAPL_EVENT_ATTR_STR(energy-pkg  ,   rapl_pkg, "event=0x02");
 RAPL_EVENT_ATTR_STR(energy-ram  ,   rapl_ram, "event=0x03");
@@ -467,7 +446,6 @@ static struct attribute_group rapl_pmu_format_group = {
 };
 
 static const struct attribute_group *rapl_attr_groups[] = {
-	&rapl_pmu_attr_group,
 	&rapl_pmu_format_group,
 	&rapl_pmu_events_group,
 	NULL,
@@ -570,65 +548,6 @@ static struct perf_msr amd_rapl_msrs[] = {
 	[PERF_RAPL_PSYS] = { 0, &rapl_events_psys_group,  NULL, false, 0 },
 };
 
-static int rapl_cpu_offline(unsigned int cpu)
-{
-	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
-	int target;
-
-	/* Check if exiting cpu is used for collecting rapl events */
-	if (!cpumask_test_and_clear_cpu(cpu, &rapl_cpu_mask))
-		return 0;
-
-	pmu->cpu = -1;
-	/* Find a new cpu to collect rapl events */
-	target = cpumask_any_but(get_rapl_pmu_cpumask(cpu), cpu);
-
-	/* Migrate rapl events to the new target */
-	if (target < nr_cpu_ids) {
-		cpumask_set_cpu(target, &rapl_cpu_mask);
-		pmu->cpu = target;
-		perf_pmu_migrate_context(pmu->pmu, cpu, target);
-	}
-	return 0;
-}
-
-static int rapl_cpu_online(unsigned int cpu)
-{
-	s32 rapl_pmu_idx = get_rapl_pmu_idx(cpu);
-	if (rapl_pmu_idx < 0) {
-		pr_err("topology_logical_(package/die)_id() returned a negative value");
-		return -EINVAL;
-	}
-	struct rapl_pmu *pmu = cpu_to_rapl_pmu(cpu);
-	int target;
-
-	if (!pmu) {
-		pmu = kzalloc_node(sizeof(*pmu), GFP_KERNEL, cpu_to_node(cpu));
-		if (!pmu)
-			return -ENOMEM;
-
-		raw_spin_lock_init(&pmu->lock);
-		INIT_LIST_HEAD(&pmu->active_list);
-		pmu->pmu = &rapl_pmus->pmu;
-		pmu->timer_interval = ms_to_ktime(rapl_timer_ms);
-		rapl_hrtimer_init(pmu);
-
-		rapl_pmus->pmus[rapl_pmu_idx] = pmu;
-	}
-
-	/*
-	 * Check if there is an online cpu in the package which collects rapl
-	 * events already.
-	 */
-	target = cpumask_any_and(&rapl_cpu_mask, get_rapl_pmu_cpumask(cpu));
-	if (target < nr_cpu_ids)
-		return 0;
-
-	cpumask_set_cpu(cpu, &rapl_cpu_mask);
-	pmu->cpu = cpu;
-	return 0;
-}
-
 static int rapl_check_hw_unit(struct rapl_model *rm)
 {
 	u64 msr_rapl_power_unit_bits;
@@ -707,12 +626,41 @@ static const struct attribute_group *rapl_attr_update[] = {
 	NULL,
 };
 
+static int __init init_rapl_pmu(void)
+{
+	struct rapl_pmu *pmu;
+	int idx;
+
+	for (idx = 0; idx < rapl_pmus->nr_rapl_pmu; idx++) {
+		pmu = kzalloc(sizeof(*pmu), GFP_KERNEL);
+		if (!pmu)
+			goto free;
+
+		raw_spin_lock_init(&pmu->lock);
+		INIT_LIST_HEAD(&pmu->active_list);
+		pmu->pmu = &rapl_pmus->pmu;
+		pmu->timer_interval = ms_to_ktime(rapl_timer_ms);
+		rapl_hrtimer_init(pmu);
+
+		rapl_pmus->pmus[idx] = pmu;
+	}
+
+	return 0;
+free:
+	for (; idx > 0; idx--)
+		kfree(rapl_pmus->pmus[idx - 1]);
+	return -ENOMEM;
+}
+
 static int __init init_rapl_pmus(void)
 {
 	int nr_rapl_pmu = topology_max_packages();
+	int rapl_pmu_scope = PERF_PMU_SCOPE_PKG;
 
-	if (!rapl_pmu_is_pkg_scope())
+	if (!rapl_pmu_is_pkg_scope()) {
 		nr_rapl_pmu *= topology_max_dies_per_package();
+		rapl_pmu_scope = PERF_PMU_SCOPE_DIE;
+	}
 
 	rapl_pmus = kzalloc(struct_size(rapl_pmus, pmus, nr_rapl_pmu), GFP_KERNEL);
 	if (!rapl_pmus)
@@ -728,9 +676,11 @@ static int __init init_rapl_pmus(void)
 	rapl_pmus->pmu.start		= rapl_pmu_event_start;
 	rapl_pmus->pmu.stop		= rapl_pmu_event_stop;
 	rapl_pmus->pmu.read		= rapl_pmu_event_read;
+	rapl_pmus->pmu.scope		= rapl_pmu_scope;
 	rapl_pmus->pmu.module		= THIS_MODULE;
 	rapl_pmus->pmu.capabilities	= PERF_PMU_CAP_NO_EXCLUDE;
-	return 0;
+
+	return init_rapl_pmu();
 }
 
 static struct rapl_model model_snb = {
@@ -876,24 +826,13 @@ static int __init rapl_pmu_init(void)
 	if (ret)
 		return ret;
 
-	/*
-	 * Install callbacks. Core will call them for each online cpu.
-	 */
-	ret = cpuhp_setup_state(CPUHP_AP_PERF_X86_RAPL_ONLINE,
-				"perf/x86/rapl:online",
-				rapl_cpu_online, rapl_cpu_offline);
-	if (ret)
-		goto out;
-
 	ret = perf_pmu_register(&rapl_pmus->pmu, "power", -1);
 	if (ret)
-		goto out1;
+		goto out;
 
 	rapl_advertise();
 	return 0;
 
-out1:
-	cpuhp_remove_state(CPUHP_AP_PERF_X86_RAPL_ONLINE);
 out:
 	pr_warn("Initialization failed (%d), disabled\n", ret);
 	cleanup_rapl_pmus();
@@ -903,7 +842,6 @@ module_init(rapl_pmu_init);
 
 static void __exit intel_rapl_exit(void)
 {
-	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_RAPL_ONLINE);
 	perf_pmu_unregister(&rapl_pmus->pmu);
 	cleanup_rapl_pmus();
 }

arch/x86/include/asm/cpu.h

@@ -32,6 +32,7 @@ extern bool handle_user_split_lock(struct pt_regs *regs, long error_code);
 extern bool handle_guest_split_lock(unsigned long ip);
 extern void handle_bus_lock(struct pt_regs *regs);
 u8 get_this_hybrid_cpu_type(void);
+u32 get_this_hybrid_cpu_native_id(void);
 #else
 static inline void __init sld_setup(struct cpuinfo_x86 *c) {}
 static inline bool handle_user_split_lock(struct pt_regs *regs, long error_code)
@@ -50,6 +51,11 @@ static inline u8 get_this_hybrid_cpu_type(void)
 {
 	return 0;
 }
+
+static inline u32 get_this_hybrid_cpu_native_id(void)
+{
+	return 0;
+}
 #endif
 #ifdef CONFIG_IA32_FEAT_CTL
 void init_ia32_feat_ctl(struct cpuinfo_x86 *c);

arch/x86/include/asm/perf_event.h

@@ -536,15 +536,17 @@ struct x86_perf_regs {
 	u64		*xmm_regs;
 };
 
-extern unsigned long perf_instruction_pointer(struct pt_regs *regs);
+extern unsigned long perf_arch_instruction_pointer(struct pt_regs *regs);
-extern unsigned long perf_misc_flags(struct pt_regs *regs);
+extern unsigned long perf_arch_misc_flags(struct pt_regs *regs);
-#define perf_misc_flags(regs)	perf_misc_flags(regs)
+extern unsigned long perf_arch_guest_misc_flags(struct pt_regs *regs);
+#define perf_arch_misc_flags(regs)	perf_arch_misc_flags(regs)
+#define perf_arch_guest_misc_flags(regs)	perf_arch_guest_misc_flags(regs)
 
 #include <asm/stacktrace.h>
 
 /*
- * We abuse bit 3 from flags to pass exact information, see perf_misc_flags
+ * We abuse bit 3 from flags to pass exact information, see
- * and the comment with PERF_EFLAGS_EXACT.
+ * perf_arch_misc_flags() and the comment with PERF_EFLAGS_EXACT.
  */
 #define perf_arch_fetch_caller_regs(regs, __ip)		{	\
 	(regs)->ip = (__ip);					\

arch/x86/kernel/cpu/intel.c

@@ -1299,3 +1299,18 @@ u8 get_this_hybrid_cpu_type(void)
 
 	return cpuid_eax(0x0000001a) >> X86_HYBRID_CPU_TYPE_ID_SHIFT;
 }
+
+/**
+ * get_this_hybrid_cpu_native_id() - Get the native id of this hybrid CPU
+ *
+ * Returns the uarch native ID [23:0] of a CPU in a hybrid processor.
+ * If the processor is not hybrid, returns 0.
+ */
+u32 get_this_hybrid_cpu_native_id(void)
+{
+	if (!cpu_feature_enabled(X86_FEATURE_HYBRID_CPU))
+		return 0;
+
+	return cpuid_eax(0x0000001a) &
+	       (BIT_ULL(X86_HYBRID_CPU_TYPE_ID_SHIFT) - 1);
+}

include/linux/cpuhotplug.h

@@ -208,7 +208,6 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_X86_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_AMD_UNCORE_ONLINE,
 	CPUHP_AP_PERF_X86_AMD_POWER_ONLINE,
-	CPUHP_AP_PERF_X86_RAPL_ONLINE,
 	CPUHP_AP_PERF_S390_CF_ONLINE,
 	CPUHP_AP_PERF_S390_SF_ONLINE,
 	CPUHP_AP_PERF_ARM_CCI_ONLINE,

include/linux/perf_event.h

@@ -170,6 +170,12 @@ struct hw_perf_event {
 		};
 		struct { /* aux / Intel-PT */
 			u64		aux_config;
+			/*
+			 * For AUX area events, aux_paused cannot be a state
+			 * flag because it can be updated asynchronously to
+			 * state.
+			 */
+			unsigned int	aux_paused;
 		};
 		struct { /* software */
 			struct hrtimer	hrtimer;
@@ -294,6 +300,7 @@ struct perf_event_pmu_context;
 #define PERF_PMU_CAP_NO_EXCLUDE			0x0040
 #define PERF_PMU_CAP_AUX_OUTPUT			0x0080
 #define PERF_PMU_CAP_EXTENDED_HW_TYPE		0x0100
+#define PERF_PMU_CAP_AUX_PAUSE			0x0200
 
 /**
  * pmu::scope
@@ -384,6 +391,8 @@ struct pmu {
 #define PERF_EF_START	0x01		/* start the counter when adding    */
 #define PERF_EF_RELOAD	0x02		/* reload the counter when starting */
 #define PERF_EF_UPDATE	0x04		/* update the counter when stopping */
+#define PERF_EF_PAUSE	0x08		/* AUX area event, pause tracing */
+#define PERF_EF_RESUME	0x10		/* AUX area event, resume tracing */
 
 	/*
 	 * Adds/Removes a counter to/from the PMU, can be done inside a
@@ -423,6 +432,18 @@ struct pmu {
 	 *
 	 * ->start() with PERF_EF_RELOAD will reprogram the counter
 	 *  value, must be preceded by a ->stop() with PERF_EF_UPDATE.
+	 *
+	 * ->stop() with PERF_EF_PAUSE will stop as simply as possible. Will not
+	 * overlap another ->stop() with PERF_EF_PAUSE nor ->start() with
+	 * PERF_EF_RESUME.
+	 *
+	 * ->start() with PERF_EF_RESUME will start as simply as possible but
+	 * only if the counter is not otherwise stopped. Will not overlap
+	 * another ->start() with PERF_EF_RESUME nor ->stop() with
+	 * PERF_EF_PAUSE.
+	 *
+	 * Notably, PERF_EF_PAUSE/PERF_EF_RESUME *can* be concurrent with other
+	 * ->stop()/->start() invocations, just not itself.
 	 */
 	void (*start)			(struct perf_event *event, int flags);
 	void (*stop)			(struct perf_event *event, int flags);
@@ -1655,15 +1676,35 @@ extern void perf_tp_event(u16 event_type, u64 count, void *record,
 			  struct task_struct *task);
 extern void perf_bp_event(struct perf_event *event, void *data);
 
-#ifndef perf_misc_flags
+extern unsigned long perf_misc_flags(struct perf_event *event, struct pt_regs *regs);
-# define perf_misc_flags(regs) \
+extern unsigned long perf_instruction_pointer(struct perf_event *event,
+					      struct pt_regs *regs);
+
+#ifndef perf_arch_misc_flags
+# define perf_arch_misc_flags(regs) \
 		(user_mode(regs) ? PERF_RECORD_MISC_USER : PERF_RECORD_MISC_KERNEL)
-# define perf_instruction_pointer(regs)	instruction_pointer(regs)
+# define perf_arch_instruction_pointer(regs)	instruction_pointer(regs)
 #endif
 #ifndef perf_arch_bpf_user_pt_regs
 # define perf_arch_bpf_user_pt_regs(regs) regs
 #endif
 
+#ifndef perf_arch_guest_misc_flags
+static inline unsigned long perf_arch_guest_misc_flags(struct pt_regs *regs)
+{
+	unsigned long guest_state = perf_guest_state();
+
+	if (!(guest_state & PERF_GUEST_ACTIVE))
+		return 0;
+
+	if (guest_state & PERF_GUEST_USER)
+		return PERF_RECORD_MISC_GUEST_USER;
+	else
+		return PERF_RECORD_MISC_GUEST_KERNEL;
+}
+# define perf_arch_guest_misc_flags(regs)	perf_arch_guest_misc_flags(regs)
+#endif
+
 static inline bool has_branch_stack(struct perf_event *event)
 {
 	return event->attr.sample_type & PERF_SAMPLE_BRANCH_STACK;
@@ -1679,6 +1720,13 @@ static inline bool has_aux(struct perf_event *event)
 	return event->pmu->setup_aux;
 }
 
+static inline bool has_aux_action(struct perf_event *event)
+{
+	return event->attr.aux_sample_size ||
+	       event->attr.aux_pause ||
+	       event->attr.aux_resume;
+}
+
 static inline bool is_write_backward(struct perf_event *event)
 {
 	return !!event->attr.write_backward;

include/linux/uprobes.h

@@ -15,6 +15,7 @@
 #include <linux/rbtree.h>
 #include <linux/types.h>
 #include <linux/wait.h>
+#include <linux/timer.h>
 
 struct uprobe;
 struct vm_area_struct;
@@ -23,8 +24,17 @@ struct inode;
 struct notifier_block;
 struct page;
 
+/*
+ * Allowed return values from uprobe consumer's handler callback
+ * with following meaning:
+ *
+ * UPROBE_HANDLER_REMOVE
+ * - Remove the uprobe breakpoint from current->mm.
+ * UPROBE_HANDLER_IGNORE
+ * - Ignore ret_handler callback for this consumer.
+ */
 #define UPROBE_HANDLER_REMOVE		1
-#define UPROBE_HANDLER_MASK		1
+#define UPROBE_HANDLER_IGNORE		2
 
 #define MAX_URETPROBE_DEPTH		64
 
@@ -37,13 +47,15 @@ struct uprobe_consumer {
 	 * for the current process. If filter() is omitted or returns true,
 	 * UPROBE_HANDLER_REMOVE is effectively ignored.
 	 */
-	int (*handler)(struct uprobe_consumer *self, struct pt_regs *regs);
+	int (*handler)(struct uprobe_consumer *self, struct pt_regs *regs, __u64 *data);
 	int (*ret_handler)(struct uprobe_consumer *self,
 				unsigned long func,
-				struct pt_regs *regs);
+				struct pt_regs *regs, __u64 *data);
 	bool (*filter)(struct uprobe_consumer *self, struct mm_struct *mm);
 
 	struct list_head cons_node;
+
+	__u64 id;	/* set when uprobe_consumer is registered */
 };
 
 #ifdef CONFIG_UPROBES
@@ -56,12 +68,62 @@ enum uprobe_task_state {
 	UTASK_SSTEP_TRAPPED,
 };
 
+/* The state of hybrid-lifetime uprobe inside struct return_instance */
+enum hprobe_state {
+	HPROBE_LEASED,		/* uretprobes_srcu-protected uprobe */
+	HPROBE_STABLE,		/* refcounted uprobe */
+	HPROBE_GONE,		/* NULL uprobe, SRCU expired, refcount failed */
+	HPROBE_CONSUMED,	/* uprobe "consumed" by uretprobe handler */
+};
+
+/*
+ * Hybrid lifetime uprobe. Represents a uprobe instance that could be either
+ * SRCU protected (with SRCU protection eventually potentially timing out),
+ * refcounted using uprobe->ref, or there could be no valid uprobe (NULL).
+ *
+ * hprobe's internal state is setup such that background timer thread can
+ * atomically "downgrade" temporarily RCU-protected uprobe into refcounted one
+ * (or no uprobe, if refcounting failed).
+ *
+ * *stable* pointer always point to the uprobe (or could be NULL if there is
+ * was no valid underlying uprobe to begin with).
+ *
+ * *leased* pointer is the key to achieving race-free atomic lifetime state
+ * transition and can have three possible states:
+ *   - either the same non-NULL value as *stable*, in which case uprobe is
+ *     SRCU-protected;
+ *   - NULL, in which case uprobe (if there is any) is refcounted;
+ *   - special __UPROBE_DEAD value, which represents an uprobe that was SRCU
+ *     protected initially, but SRCU period timed out and we attempted to
+ *     convert it to refcounted, but refcount_inc_not_zero() failed, because
+ *     uprobe effectively went away (the last consumer unsubscribed). In this
+ *     case it's important to know that *stable* pointer (which still has
+ *     non-NULL uprobe pointer) shouldn't be used, because lifetime of
+ *     underlying uprobe is not guaranteed anymore. __UPROBE_DEAD is just an
+ *     internal marker and is handled transparently by hprobe_fetch() helper.
+ *
+ * When uprobe is SRCU-protected, we also record srcu_idx value, necessary for
+ * SRCU unlocking.
+ *
+ * See hprobe_expire() and hprobe_fetch() for details of race-free uprobe
+ * state transitioning details. It all hinges on atomic xchg() over *leaded*
+ * pointer. *stable* pointer, once initially set, is not modified concurrently.
+ */
+struct hprobe {
+	enum hprobe_state state;
+	int srcu_idx;
+	struct uprobe *uprobe;
+};
+
 /*
  * uprobe_task: Metadata of a task while it singlesteps.
  */
 struct uprobe_task {
 	enum uprobe_task_state		state;
 
+	unsigned int			depth;
+	struct return_instance		*return_instances;
+
 	union {
 		struct {
 			struct arch_uprobe_task	autask;
@@ -75,23 +137,30 @@ struct uprobe_task {
 	};
 
 	struct uprobe			*active_uprobe;
+	struct timer_list		ri_timer;
 	unsigned long			xol_vaddr;
 
 	struct arch_uprobe              *auprobe;
+};
 
-	struct return_instance		*return_instances;
+struct return_consumer {
-	unsigned int			depth;
+	__u64	cookie;
+	__u64	id;
 };
 
 struct return_instance {
-	struct uprobe		*uprobe;
+	struct hprobe		hprobe;
 	unsigned long		func;
 	unsigned long		stack;		/* stack pointer */
 	unsigned long		orig_ret_vaddr; /* original return address */
 	bool			chained;	/* true, if instance is nested */
+	int			consumers_cnt;
 
 	struct return_instance	*next;		/* keep as stack */
-};
+	struct rcu_head		rcu;
+
+	struct return_consumer	consumers[] __counted_by(consumers_cnt);
+} ____cacheline_aligned;
 
 enum rp_check {
 	RP_CHECK_CALL,

include/uapi/linux/perf_event.h

@@ -511,7 +511,16 @@ struct perf_event_attr {
 	__u16	sample_max_stack;
 	__u16	__reserved_2;
 	__u32	aux_sample_size;
-	__u32	__reserved_3;
+
+	union {
+		__u32	aux_action;
+		struct {
+			__u32	aux_start_paused :  1, /* start AUX area tracing paused */
+				aux_pause        :  1, /* on overflow, pause AUX area tracing */
+				aux_resume       :  1, /* on overflow, resume AUX area tracing */
+				__reserved_3     : 29;
+		};
+	};
 
 	/*
 	 * User provided data if sigtrap=1, passed back to user via

kernel/events/core.c

@@ -2142,7 +2142,7 @@ static void perf_put_aux_event(struct perf_event *event)
 
 static bool perf_need_aux_event(struct perf_event *event)
 {
-	return !!event->attr.aux_output || !!event->attr.aux_sample_size;
+	return event->attr.aux_output || has_aux_action(event);
 }
 
 static int perf_get_aux_event(struct perf_event *event,
@@ -2167,6 +2167,10 @@ static int perf_get_aux_event(struct perf_event *event,
 	    !perf_aux_output_match(event, group_leader))
 		return 0;
 
+	if ((event->attr.aux_pause || event->attr.aux_resume) &&
+	    !(group_leader->pmu->capabilities & PERF_PMU_CAP_AUX_PAUSE))
+		return 0;
+
 	if (event->attr.aux_sample_size && !group_leader->pmu->snapshot_aux)
 		return 0;
 
@@ -7003,6 +7007,29 @@ void perf_unregister_guest_info_callbacks(struct perf_guest_info_callbacks *cbs)
 EXPORT_SYMBOL_GPL(perf_unregister_guest_info_callbacks);
 #endif
 
+static bool should_sample_guest(struct perf_event *event)
+{
+	return !event->attr.exclude_guest && perf_guest_state();
+}
+
+unsigned long perf_misc_flags(struct perf_event *event,
+			      struct pt_regs *regs)
+{
+	if (should_sample_guest(event))
+		return perf_arch_guest_misc_flags(regs);
+
+	return perf_arch_misc_flags(regs);
+}
+
+unsigned long perf_instruction_pointer(struct perf_event *event,
+				       struct pt_regs *regs)
+{
+	if (should_sample_guest(event))
+		return perf_guest_get_ip();
+
+	return perf_arch_instruction_pointer(regs);
+}
+
 static void
 perf_output_sample_regs(struct perf_output_handle *handle,
 			struct pt_regs *regs, u64 mask)
@@ -7820,7 +7847,7 @@ void perf_prepare_sample(struct perf_sample_data *data,
 	__perf_event_header__init_id(data, event, filtered_sample_type);
 
 	if (filtered_sample_type & PERF_SAMPLE_IP) {
-		data->ip = perf_instruction_pointer(regs);
+		data->ip = perf_instruction_pointer(event, regs);
 		data->sample_flags |= PERF_SAMPLE_IP;
 	}
 
@@ -7984,7 +8011,7 @@ void perf_prepare_header(struct perf_event_header *header,
 {
 	header->type = PERF_RECORD_SAMPLE;
 	header->size = perf_sample_data_size(data, event);
-	header->misc = perf_misc_flags(regs);
+	header->misc = perf_misc_flags(event, regs);
 
 	/*
 	 * If you're adding more sample types here, you likely need to do
@@ -7997,6 +8024,49 @@ void perf_prepare_header(struct perf_event_header *header,
 	WARN_ON_ONCE(header->size & 7);
 }
 
+static void __perf_event_aux_pause(struct perf_event *event, bool pause)
+{
+	if (pause) {
+		if (!event->hw.aux_paused) {
+			event->hw.aux_paused = 1;
+			event->pmu->stop(event, PERF_EF_PAUSE);
+		}
+	} else {
+		if (event->hw.aux_paused) {
+			event->hw.aux_paused = 0;
+			event->pmu->start(event, PERF_EF_RESUME);
+		}
+	}
+}
+
+static void perf_event_aux_pause(struct perf_event *event, bool pause)
+{
+	struct perf_buffer *rb;
+
+	if (WARN_ON_ONCE(!event))
+		return;
+
+	rb = ring_buffer_get(event);
+	if (!rb)
+		return;
+
+	scoped_guard (irqsave) {
+		/*
+		 * Guard against self-recursion here. Another event could trip
+		 * this same from NMI context.
+		 */
+		if (READ_ONCE(rb->aux_in_pause_resume))
+			break;
+
+		WRITE_ONCE(rb->aux_in_pause_resume, 1);
+		barrier();
+		__perf_event_aux_pause(event, pause);
+		barrier();
+		WRITE_ONCE(rb->aux_in_pause_resume, 0);
+	}
+	ring_buffer_put(rb);
+}
+
 static __always_inline int
 __perf_event_output(struct perf_event *event,
 		    struct perf_sample_data *data,
@@ -9799,9 +9869,12 @@ static int __perf_event_overflow(struct perf_event *event,
 
 	ret = __perf_event_account_interrupt(event, throttle);
 
+	if (event->attr.aux_pause)
+		perf_event_aux_pause(event->aux_event, true);
+
 	if (event->prog && event->prog->type == BPF_PROG_TYPE_PERF_EVENT &&
 	    !bpf_overflow_handler(event, data, regs))
-		return ret;
+		goto out;
 
 	/*
 	 * XXX event_limit might not quite work as expected on inherited
@@ -9863,6 +9936,9 @@ static int __perf_event_overflow(struct perf_event *event,
 		event->pending_wakeup = 1;
 		irq_work_queue(&event->pending_irq);
 	}
+out:
+	if (event->attr.aux_resume)
+		perf_event_aux_pause(event->aux_event, false);
 
 	return ret;
 }
@@ -12254,11 +12330,25 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu,
 	}
 
 	if (event->attr.aux_output &&
-	    !(pmu->capabilities & PERF_PMU_CAP_AUX_OUTPUT)) {
+	    (!(pmu->capabilities & PERF_PMU_CAP_AUX_OUTPUT) ||
+	     event->attr.aux_pause || event->attr.aux_resume)) {
 		err = -EOPNOTSUPP;
 		goto err_pmu;
 	}
 
+	if (event->attr.aux_pause && event->attr.aux_resume) {
+		err = -EINVAL;
+		goto err_pmu;
+	}
+
+	if (event->attr.aux_start_paused) {
+		if (!(pmu->capabilities & PERF_PMU_CAP_AUX_PAUSE)) {
+			err = -EOPNOTSUPP;
+			goto err_pmu;
+		}
+		event->hw.aux_paused = 1;
+	}
+
 	if (cgroup_fd != -1) {
 		err = perf_cgroup_connect(cgroup_fd, event, attr, group_leader);
 		if (err)
@@ -13052,7 +13142,7 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu,
 	 * Grouping is not supported for kernel events, neither is 'AUX',
 	 * make sure the caller's intentions are adjusted.
 	 */
-	if (attr->aux_output)
+	if (attr->aux_output || attr->aux_action)
 		return ERR_PTR(-EINVAL);
 
 	event = perf_event_alloc(attr, cpu, task, NULL, NULL,

kernel/events/internal.h

@@ -52,6 +52,7 @@ struct perf_buffer {
 	void				(*free_aux)(void *);
 	refcount_t			aux_refcount;
 	int				aux_in_sampling;
+	int				aux_in_pause_resume;
 	void				**aux_pages;
 	void				*aux_priv;
 

kernel/events/uprobes.c

@@ -26,6 +26,9 @@
 #include <linux/task_work.h>
 #include <linux/shmem_fs.h>
 #include <linux/khugepaged.h>
+#include <linux/rcupdate_trace.h>
+#include <linux/workqueue.h>
+#include <linux/srcu.h>
 
 #include <linux/uprobes.h>
 
@@ -42,8 +45,6 @@ static struct rb_root uprobes_tree = RB_ROOT;
 static DEFINE_RWLOCK(uprobes_treelock);	/* serialize rbtree access */
 static seqcount_rwlock_t uprobes_seqcount = SEQCNT_RWLOCK_ZERO(uprobes_seqcount, &uprobes_treelock);
 
-DEFINE_STATIC_SRCU(uprobes_srcu);
-
 #define UPROBES_HASH_SZ	13
 /* serialize uprobe->pending_list */
 static struct mutex uprobes_mmap_mutex[UPROBES_HASH_SZ];
@@ -51,6 +52,9 @@ static struct mutex uprobes_mmap_mutex[UPROBES_HASH_SZ];
 
 DEFINE_STATIC_PERCPU_RWSEM(dup_mmap_sem);
 
+/* Covers return_instance's uprobe lifetime. */
+DEFINE_STATIC_SRCU(uretprobes_srcu);
+
 /* Have a copy of original instruction */
 #define UPROBE_COPY_INSN	0
 
@@ -62,10 +66,13 @@ struct uprobe {
 	struct list_head	pending_list;
 	struct list_head	consumers;
 	struct inode		*inode;		/* Also hold a ref to inode */
+	union {
 		struct rcu_head		rcu;
+		struct work_struct	work;
+	};
 	loff_t			offset;
 	loff_t			ref_ctr_offset;
-	unsigned long		flags;
+	unsigned long		flags;		/* "unsigned long" so bitops work */
 
 	/*
 	 * The generic code assumes that it has two members of unknown type
@@ -100,7 +107,6 @@ static LIST_HEAD(delayed_uprobe_list);
  */
 struct xol_area {
 	wait_queue_head_t 		wq;		/* if all slots are busy */
-	atomic_t 			slot_count;	/* number of in-use slots */
 	unsigned long 			*bitmap;	/* 0 = free slot */
 
 	struct page			*page;
@@ -620,17 +626,23 @@ static inline bool uprobe_is_active(struct uprobe *uprobe)
 	return !RB_EMPTY_NODE(&uprobe->rb_node);
 }
 
-static void uprobe_free_rcu(struct rcu_head *rcu)
+static void uprobe_free_rcu_tasks_trace(struct rcu_head *rcu)
 {
 	struct uprobe *uprobe = container_of(rcu, struct uprobe, rcu);
 
 	kfree(uprobe);
 }
 
-static void put_uprobe(struct uprobe *uprobe)
+static void uprobe_free_srcu(struct rcu_head *rcu)
 {
-	if (!refcount_dec_and_test(&uprobe->ref))
+	struct uprobe *uprobe = container_of(rcu, struct uprobe, rcu);
-		return;
+
+	call_rcu_tasks_trace(&uprobe->rcu, uprobe_free_rcu_tasks_trace);
+}
+
+static void uprobe_free_deferred(struct work_struct *work)
+{
+	struct uprobe *uprobe = container_of(work, struct uprobe, work);
 
 	write_lock(&uprobes_treelock);
 
@@ -651,7 +663,162 @@ static void put_uprobe(struct uprobe *uprobe)
 	delayed_uprobe_remove(uprobe, NULL);
 	mutex_unlock(&delayed_uprobe_lock);
 
-	call_srcu(&uprobes_srcu, &uprobe->rcu, uprobe_free_rcu);
+	/* start srcu -> rcu_tasks_trace -> kfree chain */
+	call_srcu(&uretprobes_srcu, &uprobe->rcu, uprobe_free_srcu);
+}
+
+static void put_uprobe(struct uprobe *uprobe)
+{
+	if (!refcount_dec_and_test(&uprobe->ref))
+		return;
+
+	INIT_WORK(&uprobe->work, uprobe_free_deferred);
+	schedule_work(&uprobe->work);
+}
+
+/* Initialize hprobe as SRCU-protected "leased" uprobe */
+static void hprobe_init_leased(struct hprobe *hprobe, struct uprobe *uprobe, int srcu_idx)
+{
+	WARN_ON(!uprobe);
+	hprobe->state = HPROBE_LEASED;
+	hprobe->uprobe = uprobe;
+	hprobe->srcu_idx = srcu_idx;
+}
+
+/* Initialize hprobe as refcounted ("stable") uprobe (uprobe can be NULL). */
+static void hprobe_init_stable(struct hprobe *hprobe, struct uprobe *uprobe)
+{
+	hprobe->state = uprobe ? HPROBE_STABLE : HPROBE_GONE;
+	hprobe->uprobe = uprobe;
+	hprobe->srcu_idx = -1;
+}
+
+/*
+ * hprobe_consume() fetches hprobe's underlying uprobe and detects whether
+ * uprobe is SRCU protected or is refcounted. hprobe_consume() can be
+ * used only once for a given hprobe.
+ *
+ * Caller has to call hprobe_finalize() and pass previous hprobe_state, so
+ * that hprobe_finalize() can perform SRCU unlock or put uprobe, whichever
+ * is appropriate.
+ */
+static inline struct uprobe *hprobe_consume(struct hprobe *hprobe, enum hprobe_state *hstate)
+{
+	*hstate = xchg(&hprobe->state, HPROBE_CONSUMED);
+	switch (*hstate) {
+	case HPROBE_LEASED:
+	case HPROBE_STABLE:
+		return hprobe->uprobe;
+	case HPROBE_GONE:	/* uprobe is NULL, no SRCU */
+	case HPROBE_CONSUMED:	/* uprobe was finalized already, do nothing */
+		return NULL;
+	default:
+		WARN(1, "hprobe invalid state %d", *hstate);
+		return NULL;
+	}
+}
+
+/*
+ * Reset hprobe state and, if hprobe was LEASED, release SRCU lock.
+ * hprobe_finalize() can only be used from current context after
+ * hprobe_consume() call (which determines uprobe and hstate value).
+ */
+static void hprobe_finalize(struct hprobe *hprobe, enum hprobe_state hstate)
+{
+	switch (hstate) {
+	case HPROBE_LEASED:
+		__srcu_read_unlock(&uretprobes_srcu, hprobe->srcu_idx);
+		break;
+	case HPROBE_STABLE:
+		put_uprobe(hprobe->uprobe);
+		break;
+	case HPROBE_GONE:
+	case HPROBE_CONSUMED:
+		break;
+	default:
+		WARN(1, "hprobe invalid state %d", hstate);
+		break;
+	}
+}
+
+/*
+ * Attempt to switch (atomically) uprobe from being SRCU protected (LEASED)
+ * to refcounted (STABLE) state. Competes with hprobe_consume(); only one of
+ * them can win the race to perform SRCU unlocking. Whoever wins must perform
+ * SRCU unlock.
+ *
+ * Returns underlying valid uprobe or NULL, if there was no underlying uprobe
+ * to begin with or we failed to bump its refcount and it's going away.
+ *
+ * Returned non-NULL uprobe can be still safely used within an ongoing SRCU
+ * locked region. If `get` is true, it's guaranteed that non-NULL uprobe has
+ * an extra refcount for caller to assume and use. Otherwise, it's not
+ * guaranteed that returned uprobe has a positive refcount, so caller has to
+ * attempt try_get_uprobe(), if it needs to preserve uprobe beyond current
+ * SRCU lock region. See dup_utask().
+ */
+static struct uprobe *hprobe_expire(struct hprobe *hprobe, bool get)
+{
+	enum hprobe_state hstate;
+
+	/*
+	 * return_instance's hprobe is protected by RCU.
+	 * Underlying uprobe is itself protected from reuse by SRCU.
+	 */
+	lockdep_assert(rcu_read_lock_held() && srcu_read_lock_held(&uretprobes_srcu));
+
+	hstate = READ_ONCE(hprobe->state);
+	switch (hstate) {
+	case HPROBE_STABLE:
+		/* uprobe has positive refcount, bump refcount, if necessary */
+		return get ? get_uprobe(hprobe->uprobe) : hprobe->uprobe;
+	case HPROBE_GONE:
+		/*
+		 * SRCU was unlocked earlier and we didn't manage to take
+		 * uprobe refcnt, so it's effectively NULL
+		 */
+		return NULL;
+	case HPROBE_CONSUMED:
+		/*
+		 * uprobe was consumed, so it's effectively NULL as far as
+		 * uretprobe processing logic is concerned
+		 */
+		return NULL;
+	case HPROBE_LEASED: {
+		struct uprobe *uprobe = try_get_uprobe(hprobe->uprobe);
+		/*
+		 * Try to switch hprobe state, guarding against
+		 * hprobe_consume() or another hprobe_expire() racing with us.
+		 * Note, if we failed to get uprobe refcount, we use special
+		 * HPROBE_GONE state to signal that hprobe->uprobe shouldn't
+		 * be used as it will be freed after SRCU is unlocked.
+		 */
+		if (try_cmpxchg(&hprobe->state, &hstate, uprobe ? HPROBE_STABLE : HPROBE_GONE)) {
+			/* We won the race, we are the ones to unlock SRCU */
+			__srcu_read_unlock(&uretprobes_srcu, hprobe->srcu_idx);
+			return get ? get_uprobe(uprobe) : uprobe;
+		}
+
+		/*
+		 * We lost the race, undo refcount bump (if it ever happened),
+		 * unless caller would like an extra refcount anyways.
+		 */
+		if (uprobe && !get)
+			put_uprobe(uprobe);
+		/*
+		 * Even if hprobe_consume() or another hprobe_expire() wins
+		 * the state update race and unlocks SRCU from under us, we
+		 * still have a guarantee that underyling uprobe won't be
+		 * freed due to ongoing caller's SRCU lock region, so we can
+		 * return it regardless. Also, if `get` was true, we also have
+		 * an extra ref for the caller to own. This is used in dup_utask().
+		 */
+		return uprobe;
+	}
+	default:
+		WARN(1, "unknown hprobe state %d", hstate);
+		return NULL;
+	}
 }
 
 static __always_inline
@@ -706,7 +873,7 @@ static struct uprobe *find_uprobe_rcu(struct inode *inode, loff_t offset)
 	struct rb_node *node;
 	unsigned int seq;
 
-	lockdep_assert(srcu_read_lock_held(&uprobes_srcu));
+	lockdep_assert(rcu_read_lock_trace_held());
 
 	do {
 		seq = read_seqcount_begin(&uprobes_seqcount);
@@ -825,8 +992,11 @@ static struct uprobe *alloc_uprobe(struct inode *inode, loff_t offset,
 
 static void consumer_add(struct uprobe *uprobe, struct uprobe_consumer *uc)
 {
+	static atomic64_t id;
+
 	down_write(&uprobe->consumer_rwsem);
 	list_add_rcu(&uc->cons_node, &uprobe->consumers);
+	uc->id = (__u64) atomic64_inc_return(&id);
 	up_write(&uprobe->consumer_rwsem);
 }
 
@@ -934,8 +1104,7 @@ static bool filter_chain(struct uprobe *uprobe, struct mm_struct *mm)
 	bool ret = false;
 
 	down_read(&uprobe->consumer_rwsem);
-	list_for_each_entry_srcu(uc, &uprobe->consumers, cons_node,
+	list_for_each_entry_rcu(uc, &uprobe->consumers, cons_node, rcu_read_lock_trace_held()) {
-				 srcu_read_lock_held(&uprobes_srcu)) {
 		ret = consumer_filter(uc, mm);
 		if (ret)
 			break;
@@ -1156,7 +1325,8 @@ void uprobe_unregister_sync(void)
 	 * unlucky enough caller can free consumer's memory and cause
 	 * handler_chain() or handle_uretprobe_chain() to do an use-after-free.
 	 */
-	synchronize_srcu(&uprobes_srcu);
+	synchronize_rcu_tasks_trace();
+	synchronize_srcu(&uretprobes_srcu);
 }
 EXPORT_SYMBOL_GPL(uprobe_unregister_sync);
 
@@ -1240,19 +1410,18 @@ EXPORT_SYMBOL_GPL(uprobe_register);
 int uprobe_apply(struct uprobe *uprobe, struct uprobe_consumer *uc, bool add)
 {
 	struct uprobe_consumer *con;
-	int ret = -ENOENT, srcu_idx;
+	int ret = -ENOENT;
 
 	down_write(&uprobe->register_rwsem);
 
-	srcu_idx = srcu_read_lock(&uprobes_srcu);
+	rcu_read_lock_trace();
-	list_for_each_entry_srcu(con, &uprobe->consumers, cons_node,
+	list_for_each_entry_rcu(con, &uprobe->consumers, cons_node, rcu_read_lock_trace_held()) {
-				 srcu_read_lock_held(&uprobes_srcu)) {
 		if (con == uc) {
 			ret = register_for_each_vma(uprobe, add ? uc : NULL);
 			break;
 		}
 	}
-	srcu_read_unlock(&uprobes_srcu, srcu_idx);
+	rcu_read_unlock_trace();
 
 	up_write(&uprobe->register_rwsem);
 
@@ -1475,9 +1644,15 @@ static vm_fault_t xol_fault(const struct vm_special_mapping *sm,
 	return 0;
 }
 
+static int xol_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma)
+{
+	return -EPERM;
+}
+
 static const struct vm_special_mapping xol_mapping = {
 	.name = "[uprobes]",
 	.fault = xol_fault,
+	.mremap = xol_mremap,
 };
 
 /* Slot allocation for XOL */
@@ -1553,7 +1728,6 @@ static struct xol_area *__create_xol_area(unsigned long vaddr)
 	init_waitqueue_head(&area->wq);
 	/* Reserve the 1st slot for get_trampoline_vaddr() */
 	set_bit(0, area->bitmap);
-	atomic_set(&area->slot_count, 1);
 	insns = arch_uprobe_trampoline(&insns_size);
 	arch_uprobe_copy_ixol(area->page, 0, insns, insns_size);
 
@@ -1626,92 +1800,57 @@ void uprobe_dup_mmap(struct mm_struct *oldmm, struct mm_struct *newmm)
 	}
 }
 
-/*
+static unsigned long xol_get_slot_nr(struct xol_area *area)
- *  - search for a free slot.
- */
-static unsigned long xol_take_insn_slot(struct xol_area *area)
 {
-	unsigned long slot_addr;
+	unsigned long slot_nr;
-	int slot_nr;
 
-	do {
 	slot_nr = find_first_zero_bit(area->bitmap, UINSNS_PER_PAGE);
 	if (slot_nr < UINSNS_PER_PAGE) {
 		if (!test_and_set_bit(slot_nr, area->bitmap))
-				break;
+			return slot_nr;
-
-			slot_nr = UINSNS_PER_PAGE;
-			continue;
 	}
-		wait_event(area->wq, (atomic_read(&area->slot_count) < UINSNS_PER_PAGE));
-	} while (slot_nr >= UINSNS_PER_PAGE);
 
-	slot_addr = area->vaddr + (slot_nr * UPROBE_XOL_SLOT_BYTES);
+	return UINSNS_PER_PAGE;
-	atomic_inc(&area->slot_count);
-
-	return slot_addr;
 }
 
 /*
  * xol_get_insn_slot - allocate a slot for xol.
- * Returns the allocated slot address or 0.
  */
-static unsigned long xol_get_insn_slot(struct uprobe *uprobe)
+static bool xol_get_insn_slot(struct uprobe *uprobe, struct uprobe_task *utask)
 {
-	struct xol_area *area;
+	struct xol_area *area = get_xol_area();
-	unsigned long xol_vaddr;
+	unsigned long slot_nr;
 
-	area = get_xol_area();
 	if (!area)
-		return 0;
+		return false;
 
-	xol_vaddr = xol_take_insn_slot(area);
+	wait_event(area->wq, (slot_nr = xol_get_slot_nr(area)) < UINSNS_PER_PAGE);
-	if (unlikely(!xol_vaddr))
-		return 0;
 
-	arch_uprobe_copy_ixol(area->page, xol_vaddr,
+	utask->xol_vaddr = area->vaddr + slot_nr * UPROBE_XOL_SLOT_BYTES;
+	arch_uprobe_copy_ixol(area->page, utask->xol_vaddr,
 			      &uprobe->arch.ixol, sizeof(uprobe->arch.ixol));
-
+	return true;
-	return xol_vaddr;
 }
 
 /*
- * xol_free_insn_slot - If slot was earlier allocated by
+ * xol_free_insn_slot - free the slot allocated by xol_get_insn_slot()
- * @xol_get_insn_slot(), make the slot available for
- * subsequent requests.
  */
-static void xol_free_insn_slot(struct task_struct *tsk)
+static void xol_free_insn_slot(struct uprobe_task *utask)
 {
-	struct xol_area *area;
+	struct xol_area *area = current->mm->uprobes_state.xol_area;
-	unsigned long vma_end;
+	unsigned long offset = utask->xol_vaddr - area->vaddr;
-	unsigned long slot_addr;
+	unsigned int slot_nr;
 
-	if (!tsk->mm || !tsk->mm->uprobes_state.xol_area || !tsk->utask)
+	utask->xol_vaddr = 0;
+	/* xol_vaddr must fit into [area->vaddr, area->vaddr + PAGE_SIZE) */
+	if (WARN_ON_ONCE(offset >= PAGE_SIZE))
 		return;
 
-	slot_addr = tsk->utask->xol_vaddr;
-	if (unlikely(!slot_addr))
-		return;
-
-	area = tsk->mm->uprobes_state.xol_area;
-	vma_end = area->vaddr + PAGE_SIZE;
-	if (area->vaddr <= slot_addr && slot_addr < vma_end) {
-		unsigned long offset;
-		int slot_nr;
-
-		offset = slot_addr - area->vaddr;
 	slot_nr = offset / UPROBE_XOL_SLOT_BYTES;
-		if (slot_nr >= UINSNS_PER_PAGE)
-			return;
-
 	clear_bit(slot_nr, area->bitmap);
-		atomic_dec(&area->slot_count);
 	smp_mb__after_atomic(); /* pairs with prepare_to_wait() */
 	if (waitqueue_active(&area->wq))
 		wake_up(&area->wq);
-
-		tsk->utask->xol_vaddr = 0;
-	}
 }
 
 void __weak arch_uprobe_copy_ixol(struct page *page, unsigned long vaddr,
@@ -1750,11 +1889,18 @@ unsigned long uprobe_get_trap_addr(struct pt_regs *regs)
 	return instruction_pointer(regs);
 }
 
-static struct return_instance *free_ret_instance(struct return_instance *ri)
+static struct return_instance *free_ret_instance(struct return_instance *ri, bool cleanup_hprobe)
 {
 	struct return_instance *next = ri->next;
-	put_uprobe(ri->uprobe);
+
-	kfree(ri);
+	if (cleanup_hprobe) {
+		enum hprobe_state hstate;
+
+		(void)hprobe_consume(&ri->hprobe, &hstate);
+		hprobe_finalize(&ri->hprobe, hstate);
+	}
+
+	kfree_rcu(ri, rcu);
 	return next;
 }
 
@@ -1770,18 +1916,50 @@ void uprobe_free_utask(struct task_struct *t)
 	if (!utask)
 		return;
 
-	if (utask->active_uprobe)
+	WARN_ON_ONCE(utask->active_uprobe || utask->xol_vaddr);
-		put_uprobe(utask->active_uprobe);
+
+	timer_delete_sync(&utask->ri_timer);
 
 	ri = utask->return_instances;
 	while (ri)
-		ri = free_ret_instance(ri);
+		ri = free_ret_instance(ri, true /* cleanup_hprobe */);
 
-	xol_free_insn_slot(t);
 	kfree(utask);
 	t->utask = NULL;
 }
 
+#define RI_TIMER_PERIOD (HZ / 10) /* 100 ms */
+
+#define for_each_ret_instance_rcu(pos, head) \
+	for (pos = rcu_dereference_raw(head); pos; pos = rcu_dereference_raw(pos->next))
+
+static void ri_timer(struct timer_list *timer)
+{
+	struct uprobe_task *utask = container_of(timer, struct uprobe_task, ri_timer);
+	struct return_instance *ri;
+
+	/* SRCU protects uprobe from reuse for the cmpxchg() inside hprobe_expire(). */
+	guard(srcu)(&uretprobes_srcu);
+	/* RCU protects return_instance from freeing. */
+	guard(rcu)();
+
+	for_each_ret_instance_rcu(ri, utask->return_instances)
+		hprobe_expire(&ri->hprobe, false);
+}
+
+static struct uprobe_task *alloc_utask(void)
+{
+	struct uprobe_task *utask;
+
+	utask = kzalloc(sizeof(*utask), GFP_KERNEL);
+	if (!utask)
+		return NULL;
+
+	timer_setup(&utask->ri_timer, ri_timer, 0);
+
+	return utask;
+}
+
 /*
  * Allocate a uprobe_task object for the task if necessary.
  * Called when the thread hits a breakpoint.
@@ -1793,38 +1971,73 @@ void uprobe_free_utask(struct task_struct *t)
 static struct uprobe_task *get_utask(void)
 {
 	if (!current->utask)
-		current->utask = kzalloc(sizeof(struct uprobe_task), GFP_KERNEL);
+		current->utask = alloc_utask();
 	return current->utask;
 }
 
+static size_t ri_size(int consumers_cnt)
+{
+	struct return_instance *ri;
+
+	return sizeof(*ri) + sizeof(ri->consumers[0]) * consumers_cnt;
+}
+
+#define DEF_CNT 4
+
+static struct return_instance *alloc_return_instance(void)
+{
+	struct return_instance *ri;
+
+	ri = kzalloc(ri_size(DEF_CNT), GFP_KERNEL);
+	if (!ri)
+		return ZERO_SIZE_PTR;
+
+	ri->consumers_cnt = DEF_CNT;
+	return ri;
+}
+
+static struct return_instance *dup_return_instance(struct return_instance *old)
+{
+	size_t size = ri_size(old->consumers_cnt);
+
+	return kmemdup(old, size, GFP_KERNEL);
+}
+
 static int dup_utask(struct task_struct *t, struct uprobe_task *o_utask)
 {
 	struct uprobe_task *n_utask;
 	struct return_instance **p, *o, *n;
+	struct uprobe *uprobe;
 
-	n_utask = kzalloc(sizeof(struct uprobe_task), GFP_KERNEL);
+	n_utask = alloc_utask();
 	if (!n_utask)
 		return -ENOMEM;
 	t->utask = n_utask;
 
+	/* protect uprobes from freeing, we'll need try_get_uprobe() them */
+	guard(srcu)(&uretprobes_srcu);
+
 	p = &n_utask->return_instances;
 	for (o = o_utask->return_instances; o; o = o->next) {
-		n = kmalloc(sizeof(struct return_instance), GFP_KERNEL);
+		n = dup_return_instance(o);
 		if (!n)
 			return -ENOMEM;
 
-		*n = *o;
+		/* if uprobe is non-NULL, we'll have an extra refcount for uprobe */
-		/*
+		uprobe = hprobe_expire(&o->hprobe, true);
-		 * uprobe's refcnt has to be positive at this point, kept by
-		 * utask->return_instances items; return_instances can't be
-		 * removed right now, as task is blocked due to duping; so
-		 * get_uprobe() is safe to use here.
-		 */
-		get_uprobe(n->uprobe);
-		n->next = NULL;
 
-		*p = n;
+		/*
+		 * New utask will have stable properly refcounted uprobe or
+		 * NULL. Even if we failed to get refcounted uprobe, we still
+		 * need to preserve full set of return_instances for proper
+		 * uretprobe handling and nesting in forked task.
+		 */
+		hprobe_init_stable(&n->hprobe, uprobe);
+
+		n->next = NULL;
+		rcu_assign_pointer(*p, n);
 		p = &n->next;
+
 		n_utask->depth++;
 	}
 
@@ -1900,45 +2113,34 @@ static void cleanup_return_instances(struct uprobe_task *utask, bool chained,
 	enum rp_check ctx = chained ? RP_CHECK_CHAIN_CALL : RP_CHECK_CALL;
 
 	while (ri && !arch_uretprobe_is_alive(ri, ctx, regs)) {
-		ri = free_ret_instance(ri);
+		ri = free_ret_instance(ri, true /* cleanup_hprobe */);
 		utask->depth--;
 	}
-	utask->return_instances = ri;
+	rcu_assign_pointer(utask->return_instances, ri);
 }
 
-static void prepare_uretprobe(struct uprobe *uprobe, struct pt_regs *regs)
+static void prepare_uretprobe(struct uprobe *uprobe, struct pt_regs *regs,
+			      struct return_instance *ri)
 {
-	struct return_instance *ri;
+	struct uprobe_task *utask = current->utask;
-	struct uprobe_task *utask;
 	unsigned long orig_ret_vaddr, trampoline_vaddr;
 	bool chained;
+	int srcu_idx;
 
 	if (!get_xol_area())
-		return;
+		goto free;
-
-	utask = get_utask();
-	if (!utask)
-		return;
 
 	if (utask->depth >= MAX_URETPROBE_DEPTH) {
 		printk_ratelimited(KERN_INFO "uprobe: omit uretprobe due to"
 				" nestedness limit pid/tgid=%d/%d\n",
 				current->pid, current->tgid);
-		return;
+		goto free;
 	}
 
-	/* we need to bump refcount to store uprobe in utask */
-	if (!try_get_uprobe(uprobe))
-		return;
-
-	ri = kmalloc(sizeof(struct return_instance), GFP_KERNEL);
-	if (!ri)
-		goto fail;
-
 	trampoline_vaddr = uprobe_get_trampoline_vaddr();
 	orig_ret_vaddr = arch_uretprobe_hijack_return_addr(trampoline_vaddr, regs);
 	if (orig_ret_vaddr == -1)
-		goto fail;
+		goto free;
 
 	/* drop the entries invalidated by longjmp() */
 	chained = (orig_ret_vaddr == trampoline_vaddr);
@@ -1956,53 +2158,51 @@ static void prepare_uretprobe(struct uprobe *uprobe, struct pt_regs *regs)
 			 * attack from user-space.
 			 */
 			uprobe_warn(current, "handle tail call");
-			goto fail;
+			goto free;
 		}
 		orig_ret_vaddr = utask->return_instances->orig_ret_vaddr;
 	}
-	ri->uprobe = uprobe;
+
+	/* __srcu_read_lock() because SRCU lock survives switch to user space */
+	srcu_idx = __srcu_read_lock(&uretprobes_srcu);
+
 	ri->func = instruction_pointer(regs);
 	ri->stack = user_stack_pointer(regs);
 	ri->orig_ret_vaddr = orig_ret_vaddr;
 	ri->chained = chained;
 
 	utask->depth++;
+
+	hprobe_init_leased(&ri->hprobe, uprobe, srcu_idx);
 	ri->next = utask->return_instances;
-	utask->return_instances = ri;
+	rcu_assign_pointer(utask->return_instances, ri);
+
+	mod_timer(&utask->ri_timer, jiffies + RI_TIMER_PERIOD);
 
 	return;
-fail:
+free:
 	kfree(ri);
-	put_uprobe(uprobe);
 }
 
 /* Prepare to single-step probed instruction out of line. */
 static int
 pre_ssout(struct uprobe *uprobe, struct pt_regs *regs, unsigned long bp_vaddr)
 {
-	struct uprobe_task *utask;
+	struct uprobe_task *utask = current->utask;
-	unsigned long xol_vaddr;
 	int err;
 
-	utask = get_utask();
-	if (!utask)
-		return -ENOMEM;
-
 	if (!try_get_uprobe(uprobe))
 		return -EINVAL;
 
-	xol_vaddr = xol_get_insn_slot(uprobe);
+	if (!xol_get_insn_slot(uprobe, utask)) {
-	if (!xol_vaddr) {
 		err = -ENOMEM;
 		goto err_out;
 	}
 
-	utask->xol_vaddr = xol_vaddr;
 	utask->vaddr = bp_vaddr;
-
 	err = arch_uprobe_pre_xol(&uprobe->arch, regs);
 	if (unlikely(err)) {
-		xol_free_insn_slot(current);
+		xol_free_insn_slot(utask);
 		goto err_out;
 	}
 
@@ -2125,35 +2325,90 @@ static struct uprobe *find_active_uprobe_rcu(unsigned long bp_vaddr, int *is_swb
 	return uprobe;
 }
 
+static struct return_instance*
+push_consumer(struct return_instance *ri, int idx, __u64 id, __u64 cookie)
+{
+	if (unlikely(ri == ZERO_SIZE_PTR))
+		return ri;
+
+	if (unlikely(idx >= ri->consumers_cnt)) {
+		struct return_instance *old_ri = ri;
+
+		ri->consumers_cnt += DEF_CNT;
+		ri = krealloc(old_ri, ri_size(old_ri->consumers_cnt), GFP_KERNEL);
+		if (!ri) {
+			kfree(old_ri);
+			return ZERO_SIZE_PTR;
+		}
+	}
+
+	ri->consumers[idx].id = id;
+	ri->consumers[idx].cookie = cookie;
+	return ri;
+}
+
+static struct return_consumer *
+return_consumer_find(struct return_instance *ri, int *iter, int id)
+{
+	struct return_consumer *ric;
+	int idx = *iter;
+
+	for (ric = &ri->consumers[idx]; idx < ri->consumers_cnt; idx++, ric++) {
+		if (ric->id == id) {
+			*iter = idx + 1;
+			return ric;
+		}
+	}
+	return NULL;
+}
+
+static bool ignore_ret_handler(int rc)
+{
+	return rc == UPROBE_HANDLER_REMOVE || rc == UPROBE_HANDLER_IGNORE;
+}
+
 static void handler_chain(struct uprobe *uprobe, struct pt_regs *regs)
 {
 	struct uprobe_consumer *uc;
-	int remove = UPROBE_HANDLER_REMOVE;
+	bool has_consumers = false, remove = true;
-	bool need_prep = false; /* prepare return uprobe, when needed */
+	struct return_instance *ri = NULL;
-	bool has_consumers = false;
+	int push_idx = 0;
 
 	current->utask->auprobe = &uprobe->arch;
 
-	list_for_each_entry_srcu(uc, &uprobe->consumers, cons_node,
+	list_for_each_entry_rcu(uc, &uprobe->consumers, cons_node, rcu_read_lock_trace_held()) {
-				 srcu_read_lock_held(&uprobes_srcu)) {
+		bool session = uc->handler && uc->ret_handler;
+		__u64 cookie = 0;
 		int rc = 0;
 
 		if (uc->handler) {
-			rc = uc->handler(uc, regs);
+			rc = uc->handler(uc, regs, &cookie);
-			WARN(rc & ~UPROBE_HANDLER_MASK,
+			WARN(rc < 0 || rc > 2,
 				"bad rc=0x%x from %ps()\n", rc, uc->handler);
 		}
 
-		if (uc->ret_handler)
+		remove &= rc == UPROBE_HANDLER_REMOVE;
-			need_prep = true;
-
-		remove &= rc;
 		has_consumers = true;
+
+		if (!uc->ret_handler || ignore_ret_handler(rc))
+			continue;
+
+		if (!ri)
+			ri = alloc_return_instance();
+
+		if (session)
+			ri = push_consumer(ri, push_idx++, uc->id, cookie);
 	}
 	current->utask->auprobe = NULL;
 
-	if (need_prep && !remove)
+	if (!ZERO_OR_NULL_PTR(ri)) {
-		prepare_uretprobe(uprobe, regs); /* put bp at return */
+		/*
+		 * The push_idx value has the final number of return consumers,
+		 * and ri->consumers_cnt has number of allocated consumers.
+		 */
+		ri->consumers_cnt = push_idx;
+		prepare_uretprobe(uprobe, regs, ri);
+	}
 
 	if (remove && has_consumers) {
 		down_read(&uprobe->register_rwsem);
@@ -2169,19 +2424,27 @@ static void handler_chain(struct uprobe *uprobe, struct pt_regs *regs)
 }
 
 static void
-handle_uretprobe_chain(struct return_instance *ri, struct pt_regs *regs)
+handle_uretprobe_chain(struct return_instance *ri, struct uprobe *uprobe, struct pt_regs *regs)
 {
-	struct uprobe *uprobe = ri->uprobe;
+	struct return_consumer *ric;
 	struct uprobe_consumer *uc;
-	int srcu_idx;
+	int ric_idx = 0;
 
-	srcu_idx = srcu_read_lock(&uprobes_srcu);
+	/* all consumers unsubscribed meanwhile */
-	list_for_each_entry_srcu(uc, &uprobe->consumers, cons_node,
+	if (unlikely(!uprobe))
-				 srcu_read_lock_held(&uprobes_srcu)) {
+		return;
-		if (uc->ret_handler)
+
-			uc->ret_handler(uc, ri->func, regs);
+	rcu_read_lock_trace();
+	list_for_each_entry_rcu(uc, &uprobe->consumers, cons_node, rcu_read_lock_trace_held()) {
+		bool session = uc->handler && uc->ret_handler;
+
+		if (uc->ret_handler) {
+			ric = return_consumer_find(ri, &ric_idx, uc->id);
+			if (!session || ric)
+				uc->ret_handler(uc, ri->func, regs, ric ? &ric->cookie : NULL);
 		}
-	srcu_read_unlock(&uprobes_srcu, srcu_idx);
+	}
+	rcu_read_unlock_trace();
 }
 
 static struct return_instance *find_next_ret_chain(struct return_instance *ri)
@@ -2200,6 +2463,8 @@ void uprobe_handle_trampoline(struct pt_regs *regs)
 {
 	struct uprobe_task *utask;
 	struct return_instance *ri, *next;
+	struct uprobe *uprobe;
+	enum hprobe_state hstate;
 	bool valid;
 
 	utask = current->utask;
@@ -2230,21 +2495,24 @@ void uprobe_handle_trampoline(struct pt_regs *regs)
 			 * trampoline addresses on the stack are replaced with correct
 			 * original return addresses
 			 */
-			utask->return_instances = ri->next;
+			rcu_assign_pointer(utask->return_instances, ri->next);
+
+			uprobe = hprobe_consume(&ri->hprobe, &hstate);
 			if (valid)
-				handle_uretprobe_chain(ri, regs);
+				handle_uretprobe_chain(ri, uprobe, regs);
-			ri = free_ret_instance(ri);
+			hprobe_finalize(&ri->hprobe, hstate);
+
+			/* We already took care of hprobe, no need to waste more time on that. */
+			ri = free_ret_instance(ri, false /* !cleanup_hprobe */);
 			utask->depth--;
 		} while (ri != next);
 	} while (!valid);
 
-	utask->return_instances = ri;
 	return;
 
- sigill:
+sigill:
 	uprobe_warn(current, "handle uretprobe, sending SIGILL.");
 	force_sig(SIGILL);
-
 }
 
 bool __weak arch_uprobe_ignore(struct arch_uprobe *aup, struct pt_regs *regs)
@@ -2266,13 +2534,13 @@ static void handle_swbp(struct pt_regs *regs)
 {
 	struct uprobe *uprobe;
 	unsigned long bp_vaddr;
-	int is_swbp, srcu_idx;
+	int is_swbp;
 
 	bp_vaddr = uprobe_get_swbp_addr(regs);
 	if (bp_vaddr == uprobe_get_trampoline_vaddr())
 		return uprobe_handle_trampoline(regs);
 
-	srcu_idx = srcu_read_lock(&uprobes_srcu);
+	rcu_read_lock_trace();
 
 	uprobe = find_active_uprobe_rcu(bp_vaddr, &is_swbp);
 	if (!uprobe) {
@@ -2330,7 +2598,7 @@ static void handle_swbp(struct pt_regs *regs)
 
 out:
 	/* arch_uprobe_skip_sstep() succeeded, or restart if can't singlestep */
-	srcu_read_unlock(&uprobes_srcu, srcu_idx);
+	rcu_read_unlock_trace();
 }
 
 /*
@@ -2353,7 +2621,7 @@ static void handle_singlestep(struct uprobe_task *utask, struct pt_regs *regs)
 	put_uprobe(uprobe);
 	utask->active_uprobe = NULL;
 	utask->state = UTASK_RUNNING;
-	xol_free_insn_slot(current);
+	xol_free_insn_slot(utask);
 
 	spin_lock_irq(&current->sighand->siglock);
 	recalc_sigpending(); /* see uprobe_deny_signal() */

kernel/trace/bpf_trace.c

@@ -3240,7 +3240,8 @@ uprobe_multi_link_filter(struct uprobe_consumer *con, struct mm_struct *mm)
 }
 
 static int
-uprobe_multi_link_handler(struct uprobe_consumer *con, struct pt_regs *regs)
+uprobe_multi_link_handler(struct uprobe_consumer *con, struct pt_regs *regs,
+			  __u64 *data)
 {
 	struct bpf_uprobe *uprobe;
 
@@ -3249,7 +3250,8 @@ uprobe_multi_link_handler(struct uprobe_consumer *con, struct pt_regs *regs)
 }
 
 static int
-uprobe_multi_link_ret_handler(struct uprobe_consumer *con, unsigned long func, struct pt_regs *regs)
+uprobe_multi_link_ret_handler(struct uprobe_consumer *con, unsigned long func, struct pt_regs *regs,
+			      __u64 *data)
 {
 	struct bpf_uprobe *uprobe;
 

kernel/trace/trace_uprobe.c

@@ -89,9 +89,11 @@ static struct trace_uprobe *to_trace_uprobe(struct dyn_event *ev)
 static int register_uprobe_event(struct trace_uprobe *tu);
 static int unregister_uprobe_event(struct trace_uprobe *tu);
 
-static int uprobe_dispatcher(struct uprobe_consumer *con, struct pt_regs *regs);
+static int uprobe_dispatcher(struct uprobe_consumer *con, struct pt_regs *regs,
+			     __u64 *data);
 static int uretprobe_dispatcher(struct uprobe_consumer *con,
-				unsigned long func, struct pt_regs *regs);
+				unsigned long func, struct pt_regs *regs,
+				__u64 *data);
 
 #ifdef CONFIG_STACK_GROWSUP
 static unsigned long adjust_stack_addr(unsigned long addr, unsigned int n)
@@ -1522,7 +1524,8 @@ trace_uprobe_register(struct trace_event_call *event, enum trace_reg type,
 	}
 }
 
-static int uprobe_dispatcher(struct uprobe_consumer *con, struct pt_regs *regs)
+static int uprobe_dispatcher(struct uprobe_consumer *con, struct pt_regs *regs,
+			     __u64 *data)
 {
 	struct trace_uprobe *tu;
 	struct uprobe_dispatch_data udd;
@@ -1553,7 +1556,8 @@ static int uprobe_dispatcher(struct uprobe_consumer *con, struct pt_regs *regs)
 }
 
 static int uretprobe_dispatcher(struct uprobe_consumer *con,
-				unsigned long func, struct pt_regs *regs)
+				unsigned long func, struct pt_regs *regs,
+				__u64 *data)
 {
 	struct trace_uprobe *tu;
 	struct uprobe_dispatch_data udd;

tools/testing/selftests/bpf/bpf_testmod/bpf_testmod.c

@@ -461,7 +461,7 @@ static struct bin_attribute bin_attr_bpf_testmod_file __ro_after_init = {
 
 static int
 uprobe_ret_handler(struct uprobe_consumer *self, unsigned long func,
-		   struct pt_regs *regs)
+		   struct pt_regs *regs, __u64 *data)
 
 {
 	regs->ax  = 0x12345678deadbeef;