perf/x86: Add constraint to create guest LBR event without hw counter

The hypervisor may request the perf subsystem to schedule a time window
to directly access the LBR records msrs for its own use. Normally, it would
create a guest LBR event with callstack mode enabled, which is scheduled
along with other ordinary LBR events on the host but in an exclusive way.

To avoid wasting a counter for the guest LBR event, the perf tracks its
hw->idx via INTEL_PMC_IDX_FIXED_VLBR and assigns it with a fake VLBR
counter with the help of new vlbr_constraint. As with the BTS event,
there is actually no hardware counter assigned for the guest LBR event.

Signed-off-by: Like Xu <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]

Author: Like Xu

arch/x86/events/core.c

@@ -1104,6 +1104,7 @@ static inline void x86_assign_hw_event(struct perf_event *event,
 
 	switch (hwc->idx) {
 	case INTEL_PMC_IDX_FIXED_BTS:
+	case INTEL_PMC_IDX_FIXED_VLBR:
 		hwc->config_base = 0;
 		hwc->event_base	= 0;
 		break;

arch/x86/events/intel/core.c

@@ -2621,6 +2621,20 @@ intel_bts_constraints(struct perf_event *event)
 	return NULL;
 }
 
+/*
+ * Note: matches a fake event, like Fixed2.
+ */
+static struct event_constraint *
+intel_vlbr_constraints(struct perf_event *event)
+{
+	struct event_constraint *c = &vlbr_constraint;
+
+	if (unlikely(constraint_match(c, event->hw.config)))
+		return c;
+
+	return NULL;
+}
+
 static int intel_alt_er(int idx, u64 config)
 {
 	int alt_idx = idx;
@@ -2811,6 +2825,10 @@ __intel_get_event_constraints(struct cpu_hw_events *cpuc, int idx,
 {
 	struct event_constraint *c;
 
+	c = intel_vlbr_constraints(event);
+	if (c)
+		return c;
+
 	c = intel_bts_constraints(event);
 	if (c)
 		return c;

arch/x86/events/intel/lbr.c

@@ -1363,3 +1363,7 @@ int x86_perf_get_lbr(struct x86_pmu_lbr *lbr)
 	return 0;
 }
 EXPORT_SYMBOL_GPL(x86_perf_get_lbr);
+
+struct event_constraint vlbr_constraint =
+	FIXED_EVENT_CONSTRAINT(INTEL_FIXED_VLBR_EVENT,
+			       (INTEL_PMC_IDX_FIXED_VLBR - INTEL_PMC_IDX_FIXED));

arch/x86/events/perf_event.h

@@ -990,6 +990,7 @@ void release_ds_buffers(void);
 void reserve_ds_buffers(void);
 
 extern struct event_constraint bts_constraint;
+extern struct event_constraint vlbr_constraint;
 
 void intel_pmu_enable_bts(u64 config);
 

arch/x86/include/asm/perf_event.h

@@ -192,9 +192,29 @@ struct x86_pmu_capability {
 #define GLOBAL_STATUS_UNC_OVF				BIT_ULL(61)
 #define GLOBAL_STATUS_ASIF				BIT_ULL(60)
 #define GLOBAL_STATUS_COUNTERS_FROZEN			BIT_ULL(59)
-#define GLOBAL_STATUS_LBRS_FROZEN			BIT_ULL(58)
+#define GLOBAL_STATUS_LBRS_FROZEN_BIT			58
+#define GLOBAL_STATUS_LBRS_FROZEN			BIT_ULL(GLOBAL_STATUS_LBRS_FROZEN_BIT)
 #define GLOBAL_STATUS_TRACE_TOPAPMI			BIT_ULL(55)
 
+/*
+ * We model guest LBR event tracing as another fixed-mode PMC like BTS.
+ *
+ * We choose bit 58 because it's used to indicate LBR stack frozen state
+ * for architectural perfmon v4, also we unconditionally mask that bit in
+ * the handle_pmi_common(), so it'll never be set in the overflow handling.
+ *
+ * With this fake counter assigned, the guest LBR event user (such as KVM),
+ * can program the LBR registers on its own, and we don't actually do anything
+ * with then in the host context.
+ */
+#define INTEL_PMC_IDX_FIXED_VLBR	(GLOBAL_STATUS_LBRS_FROZEN_BIT)
+
+/*
+ * Pseudo-encoding the guest LBR event as event=0x00,umask=0x1b,
+ * since it would claim bit 58 which is effectively Fixed26.
+ */
+#define INTEL_FIXED_VLBR_EVENT	0x1b00
+
 /*
  * Adaptive PEBS v4
  */