page_counter: track failcnt only for legacy cgroups

Currently page_counter tracks failcnt for counters used by v1 and v2
controllers.  However failcnt is only exported for v1 deployment and thus
there is no need to maintain it in v2.  The oom report does expose failcnt
for memory and swap in v2 but v2 already maintains MEMCG_MAX and
MEMCG_SWAP_MAX event counters which can be used.

Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Shakeel Butt <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Muchun Song <[email protected]>
Cc: Roman Gushchin (Cruise) <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

include/linux/page_counter.h

@@ -28,12 +28,13 @@ struct page_counter {
 	unsigned long watermark;
 	/* Latest cg2 reset watermark */
 	unsigned long local_watermark;
-	unsigned long failcnt;
+	unsigned long failcnt; /* v1-only field */
 
 	/* Keep all the read most fields in a separete cacheline. */
 	CACHELINE_PADDING(_pad2_);
 
 	bool protection_support;
+	bool track_failcnt;
 	unsigned long min;
 	unsigned long low;
 	unsigned long high;
@@ -58,6 +59,7 @@ static inline void page_counter_init(struct page_counter *counter,
 	counter->max = PAGE_COUNTER_MAX;
 	counter->parent = parent;
 	counter->protection_support = protection_support;
+	counter->track_failcnt = false;
 }
 
 static inline unsigned long page_counter_read(struct page_counter *counter)

mm/hugetlb_cgroup.c

@@ -101,10 +101,9 @@ static void hugetlb_cgroup_init(struct hugetlb_cgroup *h_cgroup,
 	int idx;
 
 	for (idx = 0; idx < HUGE_MAX_HSTATE; idx++) {
-		struct page_counter *fault_parent = NULL;
-		struct page_counter *rsvd_parent = NULL;
+		struct page_counter *fault, *fault_parent = NULL;
+		struct page_counter *rsvd, *rsvd_parent = NULL;
 		unsigned long limit;
-		int ret;
 
 		if (parent_h_cgroup) {
 			fault_parent = hugetlb_cgroup_counter_from_cgroup(
@@ -112,24 +111,22 @@ static void hugetlb_cgroup_init(struct hugetlb_cgroup *h_cgroup,
 			rsvd_parent = hugetlb_cgroup_counter_from_cgroup_rsvd(
 				parent_h_cgroup, idx);
 		}
-		page_counter_init(hugetlb_cgroup_counter_from_cgroup(h_cgroup,
-								     idx),
-				  fault_parent, false);
-		page_counter_init(
-			hugetlb_cgroup_counter_from_cgroup_rsvd(h_cgroup, idx),
-			rsvd_parent, false);
+		fault = hugetlb_cgroup_counter_from_cgroup(h_cgroup, idx);
+		rsvd = hugetlb_cgroup_counter_from_cgroup_rsvd(h_cgroup, idx);
+
+		page_counter_init(fault, fault_parent, false);
+		page_counter_init(rsvd, rsvd_parent, false);
+
+		if (!cgroup_subsys_on_dfl(hugetlb_cgrp_subsys)) {
+			fault->track_failcnt = true;
+			rsvd->track_failcnt = true;
+		}
 
 		limit = round_down(PAGE_COUNTER_MAX,
 				   pages_per_huge_page(&hstates[idx]));
 
-		ret = page_counter_set_max(
-			hugetlb_cgroup_counter_from_cgroup(h_cgroup, idx),
-			limit);
-		VM_BUG_ON(ret);
-		ret = page_counter_set_max(
-			hugetlb_cgroup_counter_from_cgroup_rsvd(h_cgroup, idx),
-			limit);
-		VM_BUG_ON(ret);
+		VM_BUG_ON(page_counter_set_max(fault, limit));
+		VM_BUG_ON(page_counter_set_max(rsvd, limit));
 	}
 }
 

mm/memcontrol.c

@@ -1572,16 +1572,23 @@ void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg)
 	/* Use static buffer, for the caller is holding oom_lock. */
 	static char buf[SEQ_BUF_SIZE];
 	struct seq_buf s;
+	unsigned long memory_failcnt;
 
 	lockdep_assert_held(&oom_lock);
 
+	if (cgroup_subsys_on_dfl(memory_cgrp_subsys))
+		memory_failcnt = atomic_long_read(&memcg->memory_events[MEMCG_MAX]);
+	else
+		memory_failcnt = memcg->memory.failcnt;
+
 	pr_info("memory: usage %llukB, limit %llukB, failcnt %lu\n",
 		K((u64)page_counter_read(&memcg->memory)),
-		K((u64)READ_ONCE(memcg->memory.max)), memcg->memory.failcnt);
+		K((u64)READ_ONCE(memcg->memory.max)), memory_failcnt);
 	if (cgroup_subsys_on_dfl(memory_cgrp_subsys))
 		pr_info("swap: usage %llukB, limit %llukB, failcnt %lu\n",
 			K((u64)page_counter_read(&memcg->swap)),
-			K((u64)READ_ONCE(memcg->swap.max)), memcg->swap.failcnt);
+			K((u64)READ_ONCE(memcg->swap.max)),
+			atomic_long_read(&memcg->memory_events[MEMCG_SWAP_MAX]));
 #ifdef CONFIG_MEMCG_V1
 	else {
 		pr_info("memory+swap: usage %llukB, limit %llukB, failcnt %lu\n",
@@ -3631,6 +3638,7 @@ mem_cgroup_css_alloc(struct cgroup_subsys_state *parent_css)
 		page_counter_init(&memcg->memory, &parent->memory, memcg_on_dfl);
 		page_counter_init(&memcg->swap, &parent->swap, false);
 #ifdef CONFIG_MEMCG_V1
+		memcg->memory.track_failcnt = !memcg_on_dfl;
 		WRITE_ONCE(memcg->oom_kill_disable, READ_ONCE(parent->oom_kill_disable));
 		page_counter_init(&memcg->kmem, &parent->kmem, false);
 		page_counter_init(&memcg->tcpmem, &parent->tcpmem, false);

mm/page_counter.c

@@ -121,6 +121,7 @@ bool page_counter_try_charge(struct page_counter *counter,
 {
 	struct page_counter *c;
 	bool protection = track_protection(counter);
+	bool track_failcnt = counter->track_failcnt;
 
 	for (c = counter; c; c = c->parent) {
 		long new;
@@ -146,7 +147,8 @@ bool page_counter_try_charge(struct page_counter *counter,
 			 * inaccuracy in the failcnt which is only used
 			 * to report stats.
 			 */
-			data_race(c->failcnt++);
+			if (track_failcnt)
+				data_race(c->failcnt++);
 			*fail = c;
 			goto failed;
 		}