Merge tag 'perf-urgent-2020-11-15' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull perf fixes from Thomas Gleixner:
 "A set of fixes for perf:

    - A set of commits which reduce the stack usage of various perf
      event handling functions which allocated large data structs on
      stack causing stack overflows in the worst case

    - Use the proper mechanism for detecting soft interrupts in the
      recursion protection

    - Make the resursion protection simpler and more robust

    - Simplify the scheduling of event groups to make the code more
      robust and prepare for fixing the issues vs. scheduling of
      exclusive event groups

    - Prevent event multiplexing and rotation for exclusive event groups

    - Correct the perf event attribute exclusive semantics to take
      pinned events, e.g. the PMU watchdog, into account

    - Make the anythread filtering conditional for Intel's generic PMU
      counters as it is not longer guaranteed to be supported on newer
      CPUs. Check the corresponding CPUID leaf to make sure

    - Fixup a duplicate initialization in an array which was probably
      caused by the usual 'copy & paste - forgot to edit' mishap"

* tag 'perf-urgent-2020-11-15' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/x86/intel/uncore: Fix Add BW copypasta
  perf/x86/intel: Make anythread filter support conditional
  perf: Tweak perf_event_attr::exclusive semantics
  perf: Fix event multiplexing for exclusive groups
  perf: Simplify group_sched_in()
  perf: Simplify group_sched_out()
  perf/x86: Make dummy_iregs static
  perf/arch: Remove perf_sample_data::regs_user_copy
  perf: Optimize get_recursion_context()
  perf: Fix get_recursion_context()
  perf/x86: Reduce stack usage for x86_pmu::drain_pebs()
  perf: Reduce stack usage of perf_output_begin()

7 files changed, 59 insertions, 34 deletions

diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c
index f1926e9f2143..af457f8cb29d 100644
--- a/arch/x86/events/intel/core.c
+++ b/arch/x86/events/intel/core.c
@@ -2630,7 +2630,7 @@ static int handle_pmi_common(struct pt_regs *regs, u64 status)
 		u64 pebs_enabled = cpuc->pebs_enabled;
 
 		handled++;
-		x86_pmu.drain_pebs(regs);
+		x86_pmu.drain_pebs(regs, &data);
 		status &= x86_pmu.intel_ctrl | GLOBAL_STATUS_TRACE_TOPAPMI;
 
 		/*
@@ -4987,6 +4987,12 @@ __init int intel_pmu_init(void)
 
 	x86_add_quirk(intel_arch_events_quirk); /* Install first, so it runs last */
 
+	if (version >= 5) {
+		x86_pmu.intel_cap.anythread_deprecated = edx.split.anythread_deprecated;
+		if (x86_pmu.intel_cap.anythread_deprecated)
+			pr_cont(" AnyThread deprecated, ");
+	}
+
 	/*
 	 * Install the hw-cache-events table:
 	 */
@@ -5512,6 +5518,10 @@ __init int intel_pmu_init(void)
 	x86_pmu.intel_ctrl |=
 		((1LL << x86_pmu.num_counters_fixed)-1) << INTEL_PMC_IDX_FIXED;
 
+	/* AnyThread may be deprecated on arch perfmon v5 or later */
+	if (x86_pmu.intel_cap.anythread_deprecated)
+		x86_pmu.format_attrs = intel_arch_formats_attr;
+
 	if (x86_pmu.event_constraints) {
 		/*
 		 * event on fixed counter2 (REF_CYCLES) only works on this
diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c
index 404315df1e16..b47cc4226934 100644
--- a/arch/x86/events/intel/ds.c
+++ b/arch/x86/events/intel/ds.c
@@ -642,8 +642,8 @@ int intel_pmu_drain_bts_buffer(void)
 	rcu_read_lock();
 	perf_prepare_sample(&header, &data, event, &regs);
 
-	if (perf_output_begin(&handle, event, header.size *
-			      (top - base - skip)))
+	if (perf_output_begin(&handle, &data, event,
+			      header.size * (top - base - skip)))
 		goto unlock;
 
 	for (at = base; at < top; at++) {
@@ -670,7 +670,9 @@ unlock:
 
 static inline void intel_pmu_drain_pebs_buffer(void)
 {
-	x86_pmu.drain_pebs(NULL);
+	struct perf_sample_data data;
+
+	x86_pmu.drain_pebs(NULL, &data);
 }
 
 /*
@@ -1719,23 +1721,24 @@ intel_pmu_save_and_restart_reload(struct perf_event *event, int count)
 	return 0;
 }
 
-static void __intel_pmu_pebs_event(struct perf_event *event,
-				   struct pt_regs *iregs,
-				   void *base, void *top,
-				   int bit, int count,
-				   void (*setup_sample)(struct perf_event *,
-						struct pt_regs *,
-						void *,
-						struct perf_sample_data *,
-						struct pt_regs *))
+static __always_inline void
+__intel_pmu_pebs_event(struct perf_event *event,
+		       struct pt_regs *iregs,
+		       struct perf_sample_data *data,
+		       void *base, void *top,
+		       int bit, int count,
+		       void (*setup_sample)(struct perf_event *,
+					    struct pt_regs *,
+					    void *,
+					    struct perf_sample_data *,
+					    struct pt_regs *))
 {
 	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 	struct hw_perf_event *hwc = &event->hw;
-	struct perf_sample_data data;
 	struct x86_perf_regs perf_regs;
 	struct pt_regs *regs = &perf_regs.regs;
 	void *at = get_next_pebs_record_by_bit(base, top, bit);
-	struct pt_regs dummy_iregs;
+	static struct pt_regs dummy_iregs;
 
 	if (hwc->flags & PERF_X86_EVENT_AUTO_RELOAD) {
 		/*
@@ -1752,14 +1755,14 @@ static void __intel_pmu_pebs_event(struct perf_event *event,
 		iregs = &dummy_iregs;
 
 	while (count > 1) {
-		setup_sample(event, iregs, at, &data, regs);
-		perf_event_output(event, &data, regs);
+		setup_sample(event, iregs, at, data, regs);
+		perf_event_output(event, data, regs);
 		at += cpuc->pebs_record_size;
 		at = get_next_pebs_record_by_bit(at, top, bit);
 		count--;
 	}
 
-	setup_sample(event, iregs, at, &data, regs);
+	setup_sample(event, iregs, at, data, regs);
 	if (iregs == &dummy_iregs) {
 		/*
 		 * The PEBS records may be drained in the non-overflow context,
@@ -1767,18 +1770,18 @@ static void __intel_pmu_pebs_event(struct perf_event *event,
 		 * last record the same as other PEBS records, and doesn't
 		 * invoke the generic overflow handler.
 		 */
-		perf_event_output(event, &data, regs);
+		perf_event_output(event, data, regs);
 	} else {
 		/*
 		 * All but the last records are processed.
 		 * The last one is left to be able to call the overflow handler.
 		 */
-		if (perf_event_overflow(event, &data, regs))
+		if (perf_event_overflow(event, data, regs))
 			x86_pmu_stop(event, 0);
 	}
 }
 
-static void intel_pmu_drain_pebs_core(struct pt_regs *iregs)
+static void intel_pmu_drain_pebs_core(struct pt_regs *iregs, struct perf_sample_data *data)
 {
 	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 	struct debug_store *ds = cpuc->ds;
@@ -1812,7 +1815,7 @@ static void intel_pmu_drain_pebs_core(struct pt_regs *iregs)
 		return;
 	}
 
-	__intel_pmu_pebs_event(event, iregs, at, top, 0, n,
+	__intel_pmu_pebs_event(event, iregs, data, at, top, 0, n,
 			       setup_pebs_fixed_sample_data);
 }
 
@@ -1835,7 +1838,7 @@ static void intel_pmu_pebs_event_update_no_drain(struct cpu_hw_events *cpuc, int
 	}
 }
 
-static void intel_pmu_drain_pebs_nhm(struct pt_regs *iregs)
+static void intel_pmu_drain_pebs_nhm(struct pt_regs *iregs, struct perf_sample_data *data)
 {
 	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
 	struct debug_store *ds = cpuc->ds;
@@ -1942,14 +1945,14 @@ static void intel_pmu_drain_pebs_nhm(struct pt_regs *iregs)
 		}
 
 		if (counts[bit]) {
-			__intel_pmu_pebs_event(event, iregs, base,
+			__intel_pmu_pebs_event(event, iregs, data, base,
 					       top, bit, counts[bit],
 					       setup_pebs_fixed_sample_data);
 		}
 	}
 }
 
-static void intel_pmu_drain_pebs_icl(struct pt_regs *iregs)
+static void intel_pmu_drain_pebs_icl(struct pt_regs *iregs, struct perf_sample_data *data)
 {
 	short counts[INTEL_PMC_IDX_FIXED + MAX_FIXED_PEBS_EVENTS] = {};
 	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
@@ -1997,7 +2000,7 @@ static void intel_pmu_drain_pebs_icl(struct pt_regs *iregs)
 		if (WARN_ON_ONCE(!event->attr.precise_ip))
 			continue;
 
-		__intel_pmu_pebs_event(event, iregs, base,
+		__intel_pmu_pebs_event(event, iregs, data, base,
 				       top, bit, counts[bit],
 				       setup_pebs_adaptive_sample_data);
 	}
diff --git a/arch/x86/events/intel/uncore_snb.c b/arch/x86/events/intel/uncore_snb.c
index 39e632ed6ca9..bbd1120ae161 100644
--- a/arch/x86/events/intel/uncore_snb.c
+++ b/arch/x86/events/intel/uncore_snb.c
@@ -475,7 +475,7 @@ enum perf_snb_uncore_imc_freerunning_types {
 static struct freerunning_counters snb_uncore_imc_freerunning[] = {
 	[SNB_PCI_UNCORE_IMC_DATA_READS]		= { SNB_UNCORE_PCI_IMC_DATA_READS_BASE,
 							0x0, 0x0, 1, 32 },
-	[SNB_PCI_UNCORE_IMC_DATA_READS]		= { SNB_UNCORE_PCI_IMC_DATA_WRITES_BASE,
+	[SNB_PCI_UNCORE_IMC_DATA_WRITES]	= { SNB_UNCORE_PCI_IMC_DATA_WRITES_BASE,
 							0x0, 0x0, 1, 32 },
 	[SNB_PCI_UNCORE_IMC_GT_REQUESTS]	= { SNB_UNCORE_PCI_IMC_GT_REQUESTS_BASE,
 							0x0, 0x0, 1, 32 },
diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h
index ee2b9b9fc2a5..6a8edfe59b09 100644
--- a/arch/x86/events/perf_event.h
+++ b/arch/x86/events/perf_event.h
@@ -585,6 +585,7 @@ union perf_capabilities {
 		u64     pebs_baseline:1;
 		u64	perf_metrics:1;
 		u64	pebs_output_pt_available:1;
+		u64	anythread_deprecated:1;
 	};
 	u64	capabilities;
 };
@@ -727,7 +728,7 @@ struct x86_pmu {
 	int		pebs_record_size;
 	int		pebs_buffer_size;
 	int		max_pebs_events;
-	void		(*drain_pebs)(struct pt_regs *regs);
+	void		(*drain_pebs)(struct pt_regs *regs, struct perf_sample_data *data);
 	struct event_constraint *pebs_constraints;
 	void		(*pebs_aliases)(struct perf_event *event);
 	unsigned long	large_pebs_flags;
diff --git a/arch/x86/include/asm/perf_event.h b/arch/x86/include/asm/perf_event.h
index 6960cd6d1f23..b9a7fd0a27e2 100644
--- a/arch/x86/include/asm/perf_event.h
+++ b/arch/x86/include/asm/perf_event.h
@@ -137,7 +137,9 @@ union cpuid10_edx {
 	struct {
 		unsigned int num_counters_fixed:5;
 		unsigned int bit_width_fixed:8;
-		unsigned int reserved:19;
+		unsigned int reserved1:2;
+		unsigned int anythread_deprecated:1;
+		unsigned int reserved2:16;
 	} split;
 	unsigned int full;
 };
diff --git a/arch/x86/kernel/perf_regs.c b/arch/x86/kernel/perf_regs.c
index bb7e1132290b..f9e5352b3bef 100644
--- a/arch/x86/kernel/perf_regs.c
+++ b/arch/x86/kernel/perf_regs.c
@@ -101,8 +101,7 @@ u64 perf_reg_abi(struct task_struct *task)
 }
 
 void perf_get_regs_user(struct perf_regs *regs_user,
-			struct pt_regs *regs,
-			struct pt_regs *regs_user_copy)
+			struct pt_regs *regs)
 {
 	regs_user->regs = task_pt_regs(current);
 	regs_user->abi = perf_reg_abi(current);
@@ -129,12 +128,20 @@ u64 perf_reg_abi(struct task_struct *task)
 		return PERF_SAMPLE_REGS_ABI_64;
 }
 
+static DEFINE_PER_CPU(struct pt_regs, nmi_user_regs);
+
 void perf_get_regs_user(struct perf_regs *regs_user,
-			struct pt_regs *regs,
-			struct pt_regs *regs_user_copy)
+			struct pt_regs *regs)
 {
+	struct pt_regs *regs_user_copy = this_cpu_ptr(&nmi_user_regs);
 	struct pt_regs *user_regs = task_pt_regs(current);
 
+	if (!in_nmi()) {
+		regs_user->regs = user_regs;
+		regs_user->abi = perf_reg_abi(current);
+		return;
+	}
+
 	/*
 	 * If we're in an NMI that interrupted task_pt_regs setup, then
 	 * we can't sample user regs at all.  This check isn't really
diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
index d50041f570e8..e83bfe2daf82 100644
--- a/arch/x86/kvm/cpuid.c
+++ b/arch/x86/kvm/cpuid.c
@@ -681,7 +681,9 @@ static inline int __do_cpuid_func(struct kvm_cpuid_array *array, u32 function)
 
 		edx.split.num_counters_fixed = min(cap.num_counters_fixed, MAX_FIXED_COUNTERS);
 		edx.split.bit_width_fixed = cap.bit_width_fixed;
-		edx.split.reserved = 0;
+		edx.split.anythread_deprecated = 1;
+		edx.split.reserved1 = 0;
+		edx.split.reserved2 = 0;
 
 		entry->eax = eax.full;
 		entry->ebx = cap.events_mask;