Skip to contentSkip to conversation

mm/memcontrol: Avoid stuck FLUSHING_CACHED_CHARGE on isolated CPU

PATCH RESEND Superseded Hui Zhu 3 participants 3 messages Jun 30

Subsystems (4)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Hui Zhu

author
#0 lore raw

PATCH RESEND v2 1/1 mm/memcontrol: Avoid stuck FLUSHING_CACHED_CHARGE on isolated CPU

From: Hui Zhu <[email protected]>

drain_all_stock() sets FLUSHING_CACHED_CHARGE before calling schedule_drain_work() to queue per-CPU drain work. When the target CPU is isolated (cpu_is_isolated() == true), the work is silently not queued, but FLUSHING_CACHED_CHARGE stays set. Every subsequent drain_all_stock() then sees the bit and skips this stock entirely, so the entry is effectively pinned until something else on that CPU runs drain_local_*_stock() and clears the bit -- which on a long- isolated CPU may never happen.

The original idea was to actually perform the drain from the calling CPU on behalf of the isolated one, by adding a lock around the per-CPU stock so that a remote drainer could safely touch it. In practice this turned out to be intrusive: the stock data structures and their fast paths (consume_stock(), refill_stock(), the obj_stock helpers) are deliberately designed around current-CPU-only access, and retrofitting cross-CPU serialisation onto them adds non-trivial locking and PREEMPT_RT concerns for very little gain.

Looking at the actual amount of charge that can accumulate in a single per-CPU stock, it is bounded and small, so leaving an isolated CPU's stock undrained for a while is not a real problem. The only real bug is that the stuck FLUSHING_CACHED_CHARGE bit prevents future drain_all_stock() callers from re-attempting once the CPU is no longer isolated.

Fix this minimally by clearing FLUSHING_CACHED_CHARGE when the work could not be queued because the target CPU is isolated. The cached charge itself is left in place; it will be released the next time the CPU runs drain_local_*_stock() (e.g. after leaving isolation, or if the isolated CPU itself calls drain_all_stock() -- in that case cpu == curcpu causes drain_local_memcg_stock() to be invoked directly), and the next drain_all_stock() call is free to retry instead of skipping the stock forever.

Fixes 6a792697a53a ("memcg: do not drain charge pcp caches on remote isolated cpus")

---
Changelog:
v2:
According to the comments of Waiman Long, updated fixes.

 mm/memcontrol.c | 28 ++++++++++++++++++++++------
 1 file changed, 22 insertions(+), 6 deletions(-)
mm/memcontrol.c +22 -6
index 6dc4888a90f3..2e66b4a2c25d 100644
@@ -2256,7 +2256,8 @@ static bool is_memcg_drain_needed(struct memcg_stock_pcp *stock,
 	return flush;
 }
 
-static void schedule_drain_work(int cpu, struct work_struct *work)
+static void
+schedule_drain_work(int cpu, struct work_struct *work, unsigned long *flags)
 {
 	/*
 	 * Protect housekeeping cpumask read and work enqueue together
@@ -2264,9 +2265,22 @@ static void schedule_drain_work(int cpu, struct work_struct *work)
 	 * partition update only need to wait for an RCU GP and flush the
 	 * pending work on newly isolated CPUs.
 	 */
-	guard(rcu)();
-	if (!cpu_is_isolated(cpu))
-		queue_work_on(cpu, memcg_wq, work);
+	scoped_guard(rcu) {
+		if (!cpu_is_isolated(cpu)) {
+			queue_work_on(cpu, memcg_wq, work);
+			return;
+		}
+	}
+
+	/*
+	 * The target CPU is isolated: the drain work was not queued.
+	 * Clear FLUSHING_CACHED_CHARGE so that future drain_all_stock()
+	 * callers can re-attempt instead of skipping this stock forever.
+	 * The cached charge is left in place; it will be released the
+	 * next time the CPU itself runs drain_local_*_stock() (e.g.
+	 * after leaving isolation), or by a follow-up mechanism.
+	 */
+	clear_bit(FLUSHING_CACHED_CHARGE, flags);
 }
 
 /*
@@ -2299,7 +2313,8 @@ void drain_all_stock(struct mem_cgroup *root_memcg)
 			if (cpu == curcpu)
 				drain_local_memcg_stock(&memcg_st->work);
 			else
-				schedule_drain_work(cpu, &memcg_st->work);
+				schedule_drain_work(cpu, &memcg_st->work,
+						    &memcg_st->flags);
 		}
 
 		if (!test_bit(FLUSHING_CACHED_CHARGE, &obj_st->flags) &&
@@ -2309,7 +2324,8 @@ void drain_all_stock(struct mem_cgroup *root_memcg)
 			if (cpu == curcpu)
 				drain_local_obj_stock(&obj_st->work);
 			else
-				schedule_drain_work(cpu, &obj_st->work);
+				schedule_drain_work(cpu, &obj_st->work,
+						    &obj_st->flags);
 		}
 	}
 	migrate_enable();
Signature · 1 line
2.43.0
maintainer
#1 lore raw
Quoted 73 lines · Hui Zhu

From: Hui Zhu <[email protected]>

drain_all_stock() sets FLUSHING_CACHED_CHARGE before calling schedule_drain_work() to queue per-CPU drain work. When the target CPU is isolated (cpu_is_isolated() == true), the work is silently not queued, but FLUSHING_CACHED_CHARGE stays set. Every subsequent drain_all_stock() then sees the bit and skips this stock entirely, so the entry is effectively pinned until something else on that CPU runs drain_local_*_stock() and clears the bit -- which on a long- isolated CPU may never happen.

The original idea was to actually perform the drain from the calling CPU on behalf of the isolated one, by adding a lock around the per-CPU stock so that a remote drainer could safely touch it. In practice this turned out to be intrusive: the stock data structures and their fast paths (consume_stock(), refill_stock(), the obj_stock helpers) are deliberately designed around current-CPU-only access, and retrofitting cross-CPU serialisation onto them adds non-trivial locking and PREEMPT_RT concerns for very little gain.

Looking at the actual amount of charge that can accumulate in a single per-CPU stock, it is bounded and small, so leaving an isolated CPU's stock undrained for a while is not a real problem. The only real bug is that the stuck FLUSHING_CACHED_CHARGE bit prevents future drain_all_stock() callers from re-attempting once the CPU is no longer isolated.

Fix this minimally by clearing FLUSHING_CACHED_CHARGE when the work could not be queued because the target CPU is isolated. The cached charge itself is left in place; it will be released the next time the CPU runs drain_local_*_stock() (e.g. after leaving isolation, or if the isolated CPU itself calls drain_all_stock() -- in that case cpu == curcpu causes drain_local_memcg_stock() to be invoked directly), and the next drain_all_stock() call is free to retry instead of skipping the stock forever.

Fixes 6a792697a53a ("memcg: do not drain charge pcp caches on remote isolated cpus")

---
Changelog:
v2:
According to the comments of Waiman Long, updated fixes.

 mm/memcontrol.c | 28 ++++++++++++++++++++++------
 1 file changed, 22 insertions(+), 6 deletions(-)
mm/memcontrol.c +8 -4
index 6dc4888a90f3..2e66b4a2c25d 100644
@@ -2256,7 +2256,8 @@ static bool is_memcg_drain_needed(struct memcg_stock_pcp *stock,
 	return flush;
 }
 
-static void schedule_drain_work(int cpu, struct work_struct *work)
+static void
+schedule_drain_work(int cpu, struct work_struct *work, unsigned long *flags)
 {
 	/*
 	 * Protect housekeeping cpumask read and work enqueue together
@@ -2264,9 +2265,22 @@ static void schedule_drain_work(int cpu, struct work_struct *work)
 	 * partition update only need to wait for an RCU GP and flush the
 	 * pending work on newly isolated CPUs.
 	 */
-	guard(rcu)();
-	if (!cpu_is_isolated(cpu))
-		queue_work_on(cpu, memcg_wq, work);
+	scoped_guard(rcu) {
+		if (!cpu_is_isolated(cpu)) {
+			queue_work_on(cpu, memcg_wq, work);
+			return;
+		}
+	}

mm/memcontrol.c static void schedule_drain_work(int cpu, struct work_struct *work) · line 2283

Quoted 10 lines
mm/memcontrol.c +10 -0
+
+	/*
+	 * The target CPU is isolated: the drain work was not queued.
+	 * Clear FLUSHING_CACHED_CHARGE so that future drain_all_stock()
+	 * callers can re-attempt instead of skipping this stock forever.
+	 * The cached charge is left in place; it will be released the
+	 * next time the CPU itself runs drain_local_*_stock() (e.g.
+	 * after leaving isolation), or by a follow-up mechanism.
+	 */
+	clear_bit(FLUSHING_CACHED_CHARGE, flags);

Let's do something like the following.

	guard(rcu)();
	if (cpu_is_isolated(cpu)) {
		clear_bit(FLUSHING_CACHED_CHARGE, flags);
		return;
	}
	queue_work_on(cpu, memcg_wq, work);
Quoted 26 lines
mm/memcontrol.c +4 -2
 }
 
 /*
@@ -2299,7 +2313,8 @@ void drain_all_stock(struct mem_cgroup *root_memcg)
 			if (cpu == curcpu)
 				drain_local_memcg_stock(&memcg_st->work);
 			else
-				schedule_drain_work(cpu, &memcg_st->work);
+				schedule_drain_work(cpu, &memcg_st->work,
+						    &memcg_st->flags);
 		}
 
 		if (!test_bit(FLUSHING_CACHED_CHARGE, &obj_st->flags) &&
@@ -2309,7 +2324,8 @@ void drain_all_stock(struct mem_cgroup *root_memcg)
 			if (cpu == curcpu)
 				drain_local_obj_stock(&obj_st->work);
 			else
-				schedule_drain_work(cpu, &obj_st->work);
+				schedule_drain_work(cpu, &obj_st->work,
+						    &obj_st->flags);
 		}
 	}
 	migrate_enable();
Signature · 1 line
2.43.0
#2 lore raw
In reply to Hui Zhu
Quoted 10 lines · On Tue, 30 Jun 2026 15:00:04 +0800 Hui Zhu <[email protected]> wrote:

From: Hui Zhu <[email protected]>

drain_all_stock() sets FLUSHING_CACHED_CHARGE before calling schedule_drain_work() to queue per-CPU drain work. When the target CPU is isolated (cpu_is_isolated() == true), the work is silently not queued, but FLUSHING_CACHED_CHARGE stays set. Every subsequent drain_all_stock() then sees the bit and skips this stock entirely, so the entry is effectively pinned until something else on that CPU runs drain_local_*_stock() and clears the bit -- which on a long- isolated CPU may never happen.

Hello Hui,

I hope you're doing well! Thank you for the patch. I've also been thinking about memcg stock and specifically on how the flushing happens.

I haven't been thinking about the issue of isolated work taking a long time, but I have been reworking memcg stock by getting rid of it and pushing it down to the page_counter level [1].

I think that this patch is still important because of the obj_stock portion, which my series doens't touch. I just wanted to bring this to your attention! I don't think there should be any other problems.

Thank you again, have a great day!
Joshua

[1] https://lore.kernel.org/all/[email protected]/

Quoted 97 lines

The original idea was to actually perform the drain from the calling CPU on behalf of the isolated one, by adding a lock around the per-CPU stock so that a remote drainer could safely touch it. In practice this turned out to be intrusive: the stock data structures and their fast paths (consume_stock(), refill_stock(), the obj_stock helpers) are deliberately designed around current-CPU-only access, and retrofitting cross-CPU serialisation onto them adds non-trivial locking and PREEMPT_RT concerns for very little gain.

Looking at the actual amount of charge that can accumulate in a single per-CPU stock, it is bounded and small, so leaving an isolated CPU's stock undrained for a while is not a real problem. The only real bug is that the stuck FLUSHING_CACHED_CHARGE bit prevents future drain_all_stock() callers from re-attempting once the CPU is no longer isolated.

Fix this minimally by clearing FLUSHING_CACHED_CHARGE when the work could not be queued because the target CPU is isolated. The cached charge itself is left in place; it will be released the next time the CPU runs drain_local_*_stock() (e.g. after leaving isolation, or if the isolated CPU itself calls drain_all_stock() -- in that case cpu == curcpu causes drain_local_memcg_stock() to be invoked directly), and the next drain_all_stock() call is free to retry instead of skipping the stock forever.

Fixes 6a792697a53a ("memcg: do not drain charge pcp caches on remote isolated cpus")

---
Changelog:
v2:
According to the comments of Waiman Long, updated fixes.

 mm/memcontrol.c | 28 ++++++++++++++++++++++------
 1 file changed, 22 insertions(+), 6 deletions(-)
mm/memcontrol.c +22 -6
index 6dc4888a90f3..2e66b4a2c25d 100644
@@ -2256,7 +2256,8 @@ static bool is_memcg_drain_needed(struct memcg_stock_pcp *stock,
 	return flush;
 }
 
-static void schedule_drain_work(int cpu, struct work_struct *work)
+static void
+schedule_drain_work(int cpu, struct work_struct *work, unsigned long *flags)
 {
 	/*
 	 * Protect housekeeping cpumask read and work enqueue together
@@ -2264,9 +2265,22 @@ static void schedule_drain_work(int cpu, struct work_struct *work)
 	 * partition update only need to wait for an RCU GP and flush the
 	 * pending work on newly isolated CPUs.
 	 */
-	guard(rcu)();
-	if (!cpu_is_isolated(cpu))
-		queue_work_on(cpu, memcg_wq, work);
+	scoped_guard(rcu) {
+		if (!cpu_is_isolated(cpu)) {
+			queue_work_on(cpu, memcg_wq, work);
+			return;
+		}
+	}
+
+	/*
+	 * The target CPU is isolated: the drain work was not queued.
+	 * Clear FLUSHING_CACHED_CHARGE so that future drain_all_stock()
+	 * callers can re-attempt instead of skipping this stock forever.
+	 * The cached charge is left in place; it will be released the
+	 * next time the CPU itself runs drain_local_*_stock() (e.g.
+	 * after leaving isolation), or by a follow-up mechanism.
+	 */
+	clear_bit(FLUSHING_CACHED_CHARGE, flags);
 }
 
 /*
@@ -2299,7 +2313,8 @@ void drain_all_stock(struct mem_cgroup *root_memcg)
 			if (cpu == curcpu)
 				drain_local_memcg_stock(&memcg_st->work);
 			else
-				schedule_drain_work(cpu, &memcg_st->work);
+				schedule_drain_work(cpu, &memcg_st->work,
+						    &memcg_st->flags);
 		}
 
 		if (!test_bit(FLUSHING_CACHED_CHARGE, &obj_st->flags) &&
@@ -2309,7 +2324,8 @@ void drain_all_stock(struct mem_cgroup *root_memcg)
 			if (cpu == curcpu)
 				drain_local_obj_stock(&obj_st->work);
 			else
-				schedule_drain_work(cpu, &obj_st->work);
+				schedule_drain_work(cpu, &obj_st->work,
+						    &obj_st->flags);
 		}
 	}
 	migrate_enable();
Signature · 1 line
2.43.0