perf: x86, implement group scheduling transactional APIs

index 7de7061..d46e797 100644
@@ -110,6 +110,8 @@ struct cpu_hw_events {
 	u64			tags[X86_PMC_IDX_MAX];
 	struct perf_event	*event_list[X86_PMC_IDX_MAX]; /* in enabled order */
 
+	unsigned int		group_flag;
+
 	/*
 	 * Intel DebugStore bits
 	 */

@@ -944,6 +946,14 @@ static int x86_pmu_enable(struct perf_event *event)
 	if (n < 0)
 		return n;
 
+	/*
+	 * If group events scheduling transaction was started,
+	 * skip the schedulability test here, it will be peformed
+	 * at commit time(->commit_group_trans) as a whole
+	 */
+	if (cpuc->group_flag & PERF_EVENT_TRAN_STARTED)
+		goto out;
+
 	ret = x86_pmu.schedule_events(cpuc, n, assign);
 	if (ret)
 		return ret;

@@ -953,6 +963,7 @@ static int x86_pmu_enable(struct perf_event *event)
 	 */
 	memcpy(cpuc->assign, assign, n*sizeof(int));
 
+out:
 	cpuc->n_events = n;
 	cpuc->n_added += n - n0;
 

@@ -1210,119 +1221,6 @@ x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
 	return &unconstrained;
 }
 
-static int x86_event_sched_in(struct perf_event *event,
-			  struct perf_cpu_context *cpuctx)
-{
-	int ret = 0;
-
-	event->state = PERF_EVENT_STATE_ACTIVE;
-	event->oncpu = smp_processor_id();
-	event->tstamp_running += event->ctx->time - event->tstamp_stopped;
-
-	if (!is_x86_event(event))
-		ret = event->pmu->enable(event);
-
-	if (!ret && !is_software_event(event))
-		cpuctx->active_oncpu++;
-
-	if (!ret && event->attr.exclusive)
-		cpuctx->exclusive = 1;
-
-	return ret;
-}
-
-static void x86_event_sched_out(struct perf_event *event,
-			    struct perf_cpu_context *cpuctx)
-{
-	event->state = PERF_EVENT_STATE_INACTIVE;
-	event->oncpu = -1;
-
-	if (!is_x86_event(event))
-		event->pmu->disable(event);
-
-	event->tstamp_running -= event->ctx->time - event->tstamp_stopped;
-
-	if (!is_software_event(event))
-		cpuctx->active_oncpu--;
-
-	if (event->attr.exclusive || !cpuctx->active_oncpu)
-		cpuctx->exclusive = 0;
-}
-
-/*
- * Called to enable a whole group of events.
- * Returns 1 if the group was enabled, or -EAGAIN if it could not be.
- * Assumes the caller has disabled interrupts and has
- * frozen the PMU with hw_perf_save_disable.
- *
- * called with PMU disabled. If successful and return value 1,
- * then guaranteed to call perf_enable() and hw_perf_enable()
- */
-int hw_perf_group_sched_in(struct perf_event *leader,
-	       struct perf_cpu_context *cpuctx,
-	       struct perf_event_context *ctx)
-{
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
-	struct perf_event *sub;
-	int assign[X86_PMC_IDX_MAX];
-	int n0, n1, ret;
-
-	if (!x86_pmu_initialized())
-		return 0;
-
-	/* n0 = total number of events */
-	n0 = collect_events(cpuc, leader, true);
-	if (n0 < 0)
-		return n0;
-
-	ret = x86_pmu.schedule_events(cpuc, n0, assign);
-	if (ret)
-		return ret;
-
-	ret = x86_event_sched_in(leader, cpuctx);
-	if (ret)
-		return ret;
-
-	n1 = 1;
-	list_for_each_entry(sub, &leader->sibling_list, group_entry) {
-		if (sub->state > PERF_EVENT_STATE_OFF) {
-			ret = x86_event_sched_in(sub, cpuctx);
-			if (ret)
-				goto undo;
-			++n1;
-		}
-	}
-	/*
-	 * copy new assignment, now we know it is possible
-	 * will be used by hw_perf_enable()
-	 */
-	memcpy(cpuc->assign, assign, n0*sizeof(int));
-
-	cpuc->n_events  = n0;
-	cpuc->n_added  += n1;
-	ctx->nr_active += n1;
-
-	/*
-	 * 1 means successful and events are active
-	 * This is not quite true because we defer
-	 * actual activation until hw_perf_enable() but
-	 * this way we* ensure caller won't try to enable
-	 * individual events
-	 */
-	return 1;
-undo:
-	x86_event_sched_out(leader, cpuctx);
-	n0  = 1;
-	list_for_each_entry(sub, &leader->sibling_list, group_entry) {
-		if (sub->state == PERF_EVENT_STATE_ACTIVE) {
-			x86_event_sched_out(sub, cpuctx);
-			if (++n0 == n1)
-				break;
-		}
-	}
-	return ret;
-}
-
 #include "perf_event_amd.c"
 #include "perf_event_p6.c"
 #include "perf_event_p4.c"

@@ -1454,13 +1352,69 @@ static inline void x86_pmu_read(struct perf_event *event)
 	x86_perf_event_update(event);
 }
 
+/*
+ * Start group events scheduling transaction
+ * Set the flag to make pmu::enable() not perform the
+ * schedulability test, it will be performed at commit time
+ */
+static void x86_pmu_start_group_trans(const struct pmu *pmu)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	cpuc->group_flag |= PERF_EVENT_TRAN_STARTED;
+}
+
+/*
+ * Stop group events scheduling transaction
+ * Clear the flag and pmu::enable() will perform the
+ * schedulability test.
+ */
+static void x86_pmu_stop_group_trans(const struct pmu *pmu)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	cpuc->group_flag &= ~PERF_EVENT_TRAN_STARTED;
+}
+
+/*
+ * Commit group events scheduling transaction
+ * Perform the group schedulability test as a whole
+ * Return 0 if success
+ */
+static int x86_pmu_commit_group_trans(const struct pmu *pmu)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	int assign[X86_PMC_IDX_MAX];
+	int n, ret;
+
+	n = cpuc->n_events;
+
+	if (!x86_pmu_initialized())
+		return -EAGAIN;
+
+	ret = x86_pmu.schedule_events(cpuc, n, assign);
+	if (ret)
+		return ret;
+
+	/*
+	 * copy new assignment, now we know it is possible
+	 * will be used by hw_perf_enable()
+	 */
+	memcpy(cpuc->assign, assign, n*sizeof(int));
+
+	return 0;
+}
+
 static const struct pmu pmu = {
-	.enable		= x86_pmu_enable,
-	.disable	= x86_pmu_disable,
-	.start		= x86_pmu_start,
-	.stop		= x86_pmu_stop,
-	.read		= x86_pmu_read,
-	.unthrottle	= x86_pmu_unthrottle,
+	.enable			= x86_pmu_enable,
+	.disable		= x86_pmu_disable,
+	.start			= x86_pmu_start,
+	.stop			= x86_pmu_stop,
+	.read			= x86_pmu_read,
+	.unthrottle		= x86_pmu_unthrottle,
+	.start_group_trans	= x86_pmu_start_group_trans,
+	.stop_group_trans	= x86_pmu_stop_group_trans,
+	.commit_group_trans	= x86_pmu_commit_group_trans,
 };
 
 /*

index 27fa9ee..fd4db0d 100644
@@ -110,6 +110,8 @@ struct cpu_hw_events {
 	u64			tags[X86_PMC_IDX_MAX];
 	struct perf_event	*event_list[X86_PMC_IDX_MAX]; /* in enabled order */
 
+	unsigned int		group_flag;
+
 	/*
 	 * Intel DebugStore bits
 	 */

@@ -961,6 +963,14 @@ static int x86_pmu_enable(struct perf_event *event)
 	if (n < 0)
 		return n;
 
+	/*
+	 * If group events scheduling transaction was started,
+	 * skip the schedulability test here, it will be peformed
+	 * at commit time(->commit_txn) as a whole
+	 */
+	if (cpuc->group_flag & PERF_EVENT_TXN_STARTED)
+		goto out;
+
 	ret = x86_pmu.schedule_events(cpuc, n, assign);
 	if (ret)
 		return ret;

@@ -970,6 +980,7 @@ static int x86_pmu_enable(struct perf_event *event)
 	 */
 	memcpy(cpuc->assign, assign, n*sizeof(int));
 
+out:
 	cpuc->n_events = n;
 	cpuc->n_added += n - n0;
 

@@ -1227,119 +1238,6 @@ x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
 	return &unconstrained;
 }
 
-static int x86_event_sched_in(struct perf_event *event,
-			  struct perf_cpu_context *cpuctx)
-{
-	int ret = 0;
-
-	event->state = PERF_EVENT_STATE_ACTIVE;
-	event->oncpu = smp_processor_id();
-	event->tstamp_running += event->ctx->time - event->tstamp_stopped;
-
-	if (!is_x86_event(event))
-		ret = event->pmu->enable(event);
-
-	if (!ret && !is_software_event(event))
-		cpuctx->active_oncpu++;
-
-	if (!ret && event->attr.exclusive)
-		cpuctx->exclusive = 1;
-
-	return ret;
-}
-
-static void x86_event_sched_out(struct perf_event *event,
-			    struct perf_cpu_context *cpuctx)
-{
-	event->state = PERF_EVENT_STATE_INACTIVE;
-	event->oncpu = -1;
-
-	if (!is_x86_event(event))
-		event->pmu->disable(event);
-
-	event->tstamp_running -= event->ctx->time - event->tstamp_stopped;
-
-	if (!is_software_event(event))
-		cpuctx->active_oncpu--;
-
-	if (event->attr.exclusive || !cpuctx->active_oncpu)
-		cpuctx->exclusive = 0;
-}
-
-/*
- * Called to enable a whole group of events.
- * Returns 1 if the group was enabled, or -EAGAIN if it could not be.
- * Assumes the caller has disabled interrupts and has
- * frozen the PMU with hw_perf_save_disable.
- *
- * called with PMU disabled. If successful and return value 1,
- * then guaranteed to call perf_enable() and hw_perf_enable()
- */
-int hw_perf_group_sched_in(struct perf_event *leader,
-	       struct perf_cpu_context *cpuctx,
-	       struct perf_event_context *ctx)
-{
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
-	struct perf_event *sub;
-	int assign[X86_PMC_IDX_MAX];
-	int n0, n1, ret;
-
-	if (!x86_pmu_initialized())
-		return 0;
-
-	/* n0 = total number of events */
-	n0 = collect_events(cpuc, leader, true);
-	if (n0 < 0)
-		return n0;
-
-	ret = x86_pmu.schedule_events(cpuc, n0, assign);
-	if (ret)
-		return ret;
-
-	ret = x86_event_sched_in(leader, cpuctx);
-	if (ret)
-		return ret;
-
-	n1 = 1;
-	list_for_each_entry(sub, &leader->sibling_list, group_entry) {
-		if (sub->state > PERF_EVENT_STATE_OFF) {
-			ret = x86_event_sched_in(sub, cpuctx);
-			if (ret)
-				goto undo;
-			++n1;
-		}
-	}
-	/*
-	 * copy new assignment, now we know it is possible
-	 * will be used by hw_perf_enable()
-	 */
-	memcpy(cpuc->assign, assign, n0*sizeof(int));
-
-	cpuc->n_events  = n0;
-	cpuc->n_added  += n1;
-	ctx->nr_active += n1;
-
-	/*
-	 * 1 means successful and events are active
-	 * This is not quite true because we defer
-	 * actual activation until hw_perf_enable() but
-	 * this way we* ensure caller won't try to enable
-	 * individual events
-	 */
-	return 1;
-undo:
-	x86_event_sched_out(leader, cpuctx);
-	n0  = 1;
-	list_for_each_entry(sub, &leader->sibling_list, group_entry) {
-		if (sub->state == PERF_EVENT_STATE_ACTIVE) {
-			x86_event_sched_out(sub, cpuctx);
-			if (++n0 == n1)
-				break;
-		}
-	}
-	return ret;
-}
-
 #include "perf_event_amd.c"
 #include "perf_event_p6.c"
 #include "perf_event_p4.c"

@@ -1471,6 +1369,59 @@ static inline void x86_pmu_read(struct perf_event *event)
 	x86_perf_event_update(event);
 }
 
+/*
+ * Start group events scheduling transaction
+ * Set the flag to make pmu::enable() not perform the
+ * schedulability test, it will be performed at commit time
+ */
+static void x86_pmu_start_txn(const struct pmu *pmu)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	cpuc->group_flag |= PERF_EVENT_TXN_STARTED;
+}
+
+/*
+ * Stop group events scheduling transaction
+ * Clear the flag and pmu::enable() will perform the
+ * schedulability test.
+ */
+static void x86_pmu_cancel_txn(const struct pmu *pmu)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+
+	cpuc->group_flag &= ~PERF_EVENT_TXN_STARTED;
+}
+
+/*
+ * Commit group events scheduling transaction
+ * Perform the group schedulability test as a whole
+ * Return 0 if success
+ */
+static int x86_pmu_commit_txn(const struct pmu *pmu)
+{
+	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	int assign[X86_PMC_IDX_MAX];
+	int n, ret;
+
+	n = cpuc->n_events;
+
+	if (!x86_pmu_initialized())
+		return -EAGAIN;
+
+	ret = x86_pmu.schedule_events(cpuc, n, assign);
+	if (ret)
+		return ret;
+
+	/*
+	 * copy new assignment, now we know it is possible
+	 * will be used by hw_perf_enable()
+	 */
+	memcpy(cpuc->assign, assign, n*sizeof(int));
+
+	return 0;
+}
+
 static const struct pmu pmu = {
 	.enable		= x86_pmu_enable,
 	.disable	= x86_pmu_disable,

@@ -1478,6 +1429,9 @@ static const struct pmu pmu = {
 	.stop		= x86_pmu_stop,
 	.read		= x86_pmu_read,
 	.unthrottle	= x86_pmu_unthrottle,
+	.start_txn	= x86_pmu_start_txn,
+	.cancel_txn	= x86_pmu_cancel_txn,
+	.commit_txn	= x86_pmu_commit_txn,
 };
 
 /*