1 files changed, 194 insertions, 66 deletions

diff --git a/kernel/cgroup/rstat.c b/kernel/cgroup/rstat.c
index 6f87352f8219..793ecff29038 100644
--- a/kernel/cgroup/rstat.c
+++ b/kernel/cgroup/rstat.c
@@ -3,6 +3,10 @@
 
 #include <linux/sched/cputime.h>
 
+#include <linux/bpf.h>
+#include <linux/btf.h>
+#include <linux/btf_ids.h>
+
 static DEFINE_SPINLOCK(cgroup_rstat_lock);
 static DEFINE_PER_CPU(raw_spinlock_t, cgroup_rstat_cpu_lock);
 
@@ -25,34 +29,26 @@ static struct cgroup_rstat_cpu *cgroup_rstat_cpu(struct cgroup *cgrp, int cpu)
 void cgroup_rstat_updated(struct cgroup *cgrp, int cpu)
 {
 	raw_spinlock_t *cpu_lock = per_cpu_ptr(&cgroup_rstat_cpu_lock, cpu);
-	struct cgroup *parent;
 	unsigned long flags;
 
-	/* nothing to do for root */
-	if (!cgroup_parent(cgrp))
-		return;
-
-	/*
-	 * Paired with the one in cgroup_rstat_cpu_pop_updated().  Either we
-	 * see NULL updated_next or they see our updated stat.
-	 */
-	smp_mb();
-
 	/*
+	 * Speculative already-on-list test. This may race leading to
+	 * temporary inaccuracies, which is fine.
+	 *
 	 * Because @parent's updated_children is terminated with @parent
 	 * instead of NULL, we can tell whether @cgrp is on the list by
 	 * testing the next pointer for NULL.
 	 */
-	if (cgroup_rstat_cpu(cgrp, cpu)->updated_next)
+	if (data_race(cgroup_rstat_cpu(cgrp, cpu)->updated_next))
 		return;
 
 	raw_spin_lock_irqsave(cpu_lock, flags);
 
 	/* put @cgrp and all ancestors on the corresponding updated lists */
-	for (parent = cgroup_parent(cgrp); parent;
-	     cgrp = parent, parent = cgroup_parent(cgrp)) {
+	while (true) {
 		struct cgroup_rstat_cpu *rstatc = cgroup_rstat_cpu(cgrp, cpu);
-		struct cgroup_rstat_cpu *prstatc = cgroup_rstat_cpu(parent, cpu);
+		struct cgroup *parent = cgroup_parent(cgrp);
+		struct cgroup_rstat_cpu *prstatc;
 
 		/*
 		 * Both additions and removals are bottom-up.  If a cgroup
@@ -61,13 +57,21 @@ void cgroup_rstat_updated(struct cgroup *cgrp, int cpu)
 		if (rstatc->updated_next)
 			break;
 
+		/* Root has no parent to link it to, but mark it busy */
+		if (!parent) {
+			rstatc->updated_next = cgrp;
+			break;
+		}
+
+		prstatc = cgroup_rstat_cpu(parent, cpu);
 		rstatc->updated_next = prstatc->updated_children;
 		prstatc->updated_children = cgrp;
+
+		cgrp = parent;
 	}
 
 	raw_spin_unlock_irqrestore(cpu_lock, flags);
 }
-EXPORT_SYMBOL_GPL(cgroup_rstat_updated);
 
 /**
  * cgroup_rstat_cpu_pop_updated - iterate and dismantle rstat_cpu updated tree
@@ -75,7 +79,7 @@ EXPORT_SYMBOL_GPL(cgroup_rstat_updated);
  * @root: root of the tree to traversal
  * @cpu: target cpu
  *
- * Walks the udpated rstat_cpu tree on @cpu from @root.  %NULL @pos starts
+ * Walks the updated rstat_cpu tree on @cpu from @root.  %NULL @pos starts
  * the traversal and %NULL return indicates the end.  During traversal,
  * each returned cgroup is unlinked from the tree.  Must be called with the
  * matching cgroup_rstat_cpu_lock held.
@@ -88,6 +92,7 @@ static struct cgroup *cgroup_rstat_cpu_pop_updated(struct cgroup *pos,
 						   struct cgroup *root, int cpu)
 {
 	struct cgroup_rstat_cpu *rstatc;
+	struct cgroup *parent;
 
 	if (pos == root)
 		return NULL;
@@ -96,10 +101,14 @@ static struct cgroup *cgroup_rstat_cpu_pop_updated(struct cgroup *pos,
 	 * We're gonna walk down to the first leaf and visit/remove it.  We
 	 * can pick whatever unvisited node as the starting point.
 	 */
-	if (!pos)
+	if (!pos) {
 		pos = root;
-	else
+		/* return NULL if this subtree is not on-list */
+		if (!cgroup_rstat_cpu(pos, cpu)->updated_next)
+			return NULL;
+	} else {
 		pos = cgroup_parent(pos);
+	}
 
 	/* walk down to the first leaf */
 	while (true) {
@@ -115,39 +124,52 @@ static struct cgroup *cgroup_rstat_cpu_pop_updated(struct cgroup *pos,
 	 * However, due to the way we traverse, @pos will be the first
 	 * child in most cases. The only exception is @root.
 	 */
-	if (rstatc->updated_next) {
-		struct cgroup *parent = cgroup_parent(pos);
-		struct cgroup_rstat_cpu *prstatc = cgroup_rstat_cpu(parent, cpu);
-		struct cgroup_rstat_cpu *nrstatc;
+	parent = cgroup_parent(pos);
+	if (parent) {
+		struct cgroup_rstat_cpu *prstatc;
 		struct cgroup **nextp;
 
+		prstatc = cgroup_rstat_cpu(parent, cpu);
 		nextp = &prstatc->updated_children;
-		while (true) {
-			nrstatc = cgroup_rstat_cpu(*nextp, cpu);
-			if (*nextp == pos)
-				break;
+		while (*nextp != pos) {
+			struct cgroup_rstat_cpu *nrstatc;
 
+			nrstatc = cgroup_rstat_cpu(*nextp, cpu);
 			WARN_ON_ONCE(*nextp == parent);
 			nextp = &nrstatc->updated_next;
 		}
-
 		*nextp = rstatc->updated_next;
-		rstatc->updated_next = NULL;
+	}
 
-		/*
-		 * Paired with the one in cgroup_rstat_cpu_updated().
-		 * Either they see NULL updated_next or we see their
-		 * updated stat.
-		 */
-		smp_mb();
+	rstatc->updated_next = NULL;
+	return pos;
+}
 
-		return pos;
-	}
+/*
+ * A hook for bpf stat collectors to attach to and flush their stats.
+ * Together with providing bpf kfuncs for cgroup_rstat_updated() and
+ * cgroup_rstat_flush(), this enables a complete workflow where bpf progs that
+ * collect cgroup stats can integrate with rstat for efficient flushing.
+ *
+ * A static noinline declaration here could cause the compiler to optimize away
+ * the function. A global noinline declaration will keep the definition, but may
+ * optimize away the callsite. Therefore, __weak is needed to ensure that the
+ * call is still emitted, by telling the compiler that we don't know what the
+ * function might eventually be.
+ *
+ * __diag_* below are needed to dismiss the missing prototype warning.
+ */
+__diag_push();
+__diag_ignore_all("-Wmissing-prototypes",
+		  "kfuncs which will be used in BPF programs");
 
-	/* only happens for @root */
-	return NULL;
+__weak noinline void bpf_rstat_flush(struct cgroup *cgrp,
+				     struct cgroup *parent, int cpu)
+{
 }
 
+__diag_pop();
+
 /* see cgroup_rstat_flush() */
 static void cgroup_rstat_flush_locked(struct cgroup *cgrp, bool may_sleep)
 	__releases(&cgroup_rstat_lock) __acquires(&cgroup_rstat_lock)
@@ -160,12 +182,22 @@ static void cgroup_rstat_flush_locked(struct cgroup *cgrp, bool may_sleep)
 		raw_spinlock_t *cpu_lock = per_cpu_ptr(&cgroup_rstat_cpu_lock,
 						       cpu);
 		struct cgroup *pos = NULL;
+		unsigned long flags;
 
-		raw_spin_lock(cpu_lock);
+		/*
+		 * The _irqsave() is needed because cgroup_rstat_lock is
+		 * spinlock_t which is a sleeping lock on PREEMPT_RT. Acquiring
+		 * this lock with the _irq() suffix only disables interrupts on
+		 * a non-PREEMPT_RT kernel. The raw_spinlock_t below disables
+		 * interrupts on both configurations. The _irqsave() ensures
+		 * that interrupts are always disabled and later restored.
+		 */
+		raw_spin_lock_irqsave(cpu_lock, flags);
 		while ((pos = cgroup_rstat_cpu_pop_updated(pos, cgrp, cpu))) {
 			struct cgroup_subsys_state *css;
 
 			cgroup_base_stat_flush(pos, cpu);
+			bpf_rstat_flush(pos, cgroup_parent(pos), cpu);
 
 			rcu_read_lock();
 			list_for_each_entry_rcu(css, &pos->rstat_css_list,
@@ -173,7 +205,7 @@ static void cgroup_rstat_flush_locked(struct cgroup *cgrp, bool may_sleep)
 				css->ss->css_rstat_flush(css, cpu);
 			rcu_read_unlock();
 		}
-		raw_spin_unlock(cpu_lock);
+		raw_spin_unlock_irqrestore(cpu_lock, flags);
 
 		/* if @may_sleep, play nice and yield if necessary */
 		if (may_sleep && (need_resched() ||
@@ -224,7 +256,7 @@ void cgroup_rstat_flush_irqsafe(struct cgroup *cgrp)
 }
 
 /**
- * cgroup_rstat_flush_begin - flush stats in @cgrp's subtree and hold
+ * cgroup_rstat_flush_hold - flush stats in @cgrp's subtree and hold
  * @cgrp: target cgroup
  *
  * Flush stats in @cgrp's subtree and prevent further flushes.  Must be
@@ -296,8 +328,6 @@ void __init cgroup_rstat_boot(void)
 
 	for_each_possible_cpu(cpu)
 		raw_spin_lock_init(per_cpu_ptr(&cgroup_rstat_cpu_lock, cpu));
-
-	BUG_ON(cgroup_rstat_init(&cgrp_dfl_root.cgrp));
 }
 
 /*
@@ -310,6 +340,9 @@ static void cgroup_base_stat_add(struct cgroup_base_stat *dst_bstat,
 	dst_bstat->cputime.utime += src_bstat->cputime.utime;
 	dst_bstat->cputime.stime += src_bstat->cputime.stime;
 	dst_bstat->cputime.sum_exec_runtime += src_bstat->cputime.sum_exec_runtime;
+#ifdef CONFIG_SCHED_CORE
+	dst_bstat->forceidle_sum += src_bstat->forceidle_sum;
+#endif
 }
 
 static void cgroup_base_stat_sub(struct cgroup_base_stat *dst_bstat,
@@ -318,29 +351,35 @@ static void cgroup_base_stat_sub(struct cgroup_base_stat *dst_bstat,
 	dst_bstat->cputime.utime -= src_bstat->cputime.utime;
 	dst_bstat->cputime.stime -= src_bstat->cputime.stime;
 	dst_bstat->cputime.sum_exec_runtime -= src_bstat->cputime.sum_exec_runtime;
+#ifdef CONFIG_SCHED_CORE
+	dst_bstat->forceidle_sum -= src_bstat->forceidle_sum;
+#endif
 }
 
 static void cgroup_base_stat_flush(struct cgroup *cgrp, int cpu)
 {
-	struct cgroup *parent = cgroup_parent(cgrp);
 	struct cgroup_rstat_cpu *rstatc = cgroup_rstat_cpu(cgrp, cpu);
-	struct cgroup_base_stat cur, delta;
+	struct cgroup *parent = cgroup_parent(cgrp);
+	struct cgroup_base_stat delta;
 	unsigned seq;
 
+	/* Root-level stats are sourced from system-wide CPU stats */
+	if (!parent)
+		return;
+
 	/* fetch the current per-cpu values */
 	do {
 		seq = __u64_stats_fetch_begin(&rstatc->bsync);
-		cur.cputime = rstatc->bstat.cputime;
+		delta = rstatc->bstat;
 	} while (__u64_stats_fetch_retry(&rstatc->bsync, seq));
 
 	/* propagate percpu delta to global */
-	delta = cur;
 	cgroup_base_stat_sub(&delta, &rstatc->last_bstat);
 	cgroup_base_stat_add(&cgrp->bstat, &delta);
 	cgroup_base_stat_add(&rstatc->last_bstat, &delta);
 
-	/* propagate global delta to parent */
-	if (parent) {
+	/* propagate global delta to parent (unless that's root) */
+	if (cgroup_parent(parent)) {
 		delta = cgrp->bstat;
 		cgroup_base_stat_sub(&delta, &cgrp->last_bstat);
 		cgroup_base_stat_add(&parent->bstat, &delta);
@@ -349,19 +388,20 @@ static void cgroup_base_stat_flush(struct cgroup *cgrp, int cpu)
 }
 
 static struct cgroup_rstat_cpu *
-cgroup_base_stat_cputime_account_begin(struct cgroup *cgrp)
+cgroup_base_stat_cputime_account_begin(struct cgroup *cgrp, unsigned long *flags)
 {
 	struct cgroup_rstat_cpu *rstatc;
 
 	rstatc = get_cpu_ptr(cgrp->rstat_cpu);
-	u64_stats_update_begin(&rstatc->bsync);
+	*flags = u64_stats_update_begin_irqsave(&rstatc->bsync);
 	return rstatc;
 }
 
 static void cgroup_base_stat_cputime_account_end(struct cgroup *cgrp,
-						 struct cgroup_rstat_cpu *rstatc)
+						 struct cgroup_rstat_cpu *rstatc,
+						 unsigned long flags)
 {
-	u64_stats_update_end(&rstatc->bsync);
+	u64_stats_update_end_irqrestore(&rstatc->bsync, flags);
 	cgroup_rstat_updated(cgrp, smp_processor_id());
 	put_cpu_ptr(rstatc);
 }
@@ -369,18 +409,20 @@ static void cgroup_base_stat_cputime_account_end(struct cgroup *cgrp,
 void __cgroup_account_cputime(struct cgroup *cgrp, u64 delta_exec)
 {
 	struct cgroup_rstat_cpu *rstatc;
+	unsigned long flags;
 
-	rstatc = cgroup_base_stat_cputime_account_begin(cgrp);
+	rstatc = cgroup_base_stat_cputime_account_begin(cgrp, &flags);
 	rstatc->bstat.cputime.sum_exec_runtime += delta_exec;
-	cgroup_base_stat_cputime_account_end(cgrp, rstatc);
+	cgroup_base_stat_cputime_account_end(cgrp, rstatc, flags);
 }
 
 void __cgroup_account_cputime_field(struct cgroup *cgrp,
 				    enum cpu_usage_stat index, u64 delta_exec)
 {
 	struct cgroup_rstat_cpu *rstatc;
+	unsigned long flags;
 
-	rstatc = cgroup_base_stat_cputime_account_begin(cgrp);
+	rstatc = cgroup_base_stat_cputime_account_begin(cgrp, &flags);
 
 	switch (index) {
 	case CPUTIME_USER:
@@ -392,32 +434,118 @@ void __cgroup_account_cputime_field(struct cgroup *cgrp,
 	case CPUTIME_SOFTIRQ:
 		rstatc->bstat.cputime.stime += delta_exec;
 		break;
+#ifdef CONFIG_SCHED_CORE
+	case CPUTIME_FORCEIDLE:
+		rstatc->bstat.forceidle_sum += delta_exec;
+		break;
+#endif
 	default:
 		break;
 	}
 
-	cgroup_base_stat_cputime_account_end(cgrp, rstatc);
+	cgroup_base_stat_cputime_account_end(cgrp, rstatc, flags);
+}
+
+/*
+ * compute the cputime for the root cgroup by getting the per cpu data
+ * at a global level, then categorizing the fields in a manner consistent
+ * with how it is done by __cgroup_account_cputime_field for each bit of
+ * cpu time attributed to a cgroup.
+ */
+static void root_cgroup_cputime(struct cgroup_base_stat *bstat)
+{
+	struct task_cputime *cputime = &bstat->cputime;
+	int i;
+
+	cputime->stime = 0;
+	cputime->utime = 0;
+	cputime->sum_exec_runtime = 0;
+	for_each_possible_cpu(i) {
+		struct kernel_cpustat kcpustat;
+		u64 *cpustat = kcpustat.cpustat;
+		u64 user = 0;
+		u64 sys = 0;
+
+		kcpustat_cpu_fetch(&kcpustat, i);
+
+		user += cpustat[CPUTIME_USER];
+		user += cpustat[CPUTIME_NICE];
+		cputime->utime += user;
+
+		sys += cpustat[CPUTIME_SYSTEM];
+		sys += cpustat[CPUTIME_IRQ];
+		sys += cpustat[CPUTIME_SOFTIRQ];
+		cputime->stime += sys;
+
+		cputime->sum_exec_runtime += user;
+		cputime->sum_exec_runtime += sys;
+		cputime->sum_exec_runtime += cpustat[CPUTIME_STEAL];
+
+#ifdef CONFIG_SCHED_CORE
+		bstat->forceidle_sum += cpustat[CPUTIME_FORCEIDLE];
+#endif
+	}
 }
 
 void cgroup_base_stat_cputime_show(struct seq_file *seq)
 {
 	struct cgroup *cgrp = seq_css(seq)->cgroup;
 	u64 usage, utime, stime;
-
-	if (!cgroup_parent(cgrp))
-		return;
-
-	cgroup_rstat_flush_hold(cgrp);
-	usage = cgrp->bstat.cputime.sum_exec_runtime;
-	cputime_adjust(&cgrp->bstat.cputime, &cgrp->prev_cputime, &utime, &stime);
-	cgroup_rstat_flush_release();
+	struct cgroup_base_stat bstat;
+#ifdef CONFIG_SCHED_CORE
+	u64 forceidle_time;
+#endif
+
+	if (cgroup_parent(cgrp)) {
+		cgroup_rstat_flush_hold(cgrp);
+		usage = cgrp->bstat.cputime.sum_exec_runtime;
+		cputime_adjust(&cgrp->bstat.cputime, &cgrp->prev_cputime,
+			       &utime, &stime);
+#ifdef CONFIG_SCHED_CORE
+		forceidle_time = cgrp->bstat.forceidle_sum;
+#endif
+		cgroup_rstat_flush_release();
+	} else {
+		root_cgroup_cputime(&bstat);
+		usage = bstat.cputime.sum_exec_runtime;
+		utime = bstat.cputime.utime;
+		stime = bstat.cputime.stime;
+#ifdef CONFIG_SCHED_CORE
+		forceidle_time = bstat.forceidle_sum;
+#endif
+	}
 
 	do_div(usage, NSEC_PER_USEC);
 	do_div(utime, NSEC_PER_USEC);
 	do_div(stime, NSEC_PER_USEC);
+#ifdef CONFIG_SCHED_CORE
+	do_div(forceidle_time, NSEC_PER_USEC);
+#endif
 
 	seq_printf(seq, "usage_usec %llu\n"
 		   "user_usec %llu\n"
 		   "system_usec %llu\n",
 		   usage, utime, stime);
+
+#ifdef CONFIG_SCHED_CORE
+	seq_printf(seq, "core_sched.force_idle_usec %llu\n", forceidle_time);
+#endif
+}
+
+/* Add bpf kfuncs for cgroup_rstat_updated() and cgroup_rstat_flush() */
+BTF_SET8_START(bpf_rstat_kfunc_ids)
+BTF_ID_FLAGS(func, cgroup_rstat_updated)
+BTF_ID_FLAGS(func, cgroup_rstat_flush, KF_SLEEPABLE)
+BTF_SET8_END(bpf_rstat_kfunc_ids)
+
+static const struct btf_kfunc_id_set bpf_rstat_kfunc_set = {
+	.owner          = THIS_MODULE,
+	.set            = &bpf_rstat_kfunc_ids,
+};
+
+static int __init bpf_rstat_kfunc_init(void)
+{
+	return register_btf_kfunc_id_set(BPF_PROG_TYPE_TRACING,
+					 &bpf_rstat_kfunc_set);
 }
+late_initcall(bpf_rstat_kfunc_init);