Skip to contentSkip to conversation

lib/group_cpus.c: avoid to acquire cpu hotplug lock in group_cpus_evenly

PATCH Superseded Ming Lei 2 participants 3 messages Aug 18, 2023

Subsystems (4)

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Ming Lei

author
#0 lore raw

PATCH v2 1/1 lib/group_cpus.c: avoid to acquire cpu hotplug lock in group_cpus_evenly

group_cpus_evenly() could be part of storage driver's error handler, such as nvme driver, when may happen during CPU hotplug, in which storage queue has to drain its pending IOs because all CPUs associated with the queue are offline and the queue is becoming inactive. And handling IO needs error handler to provide forward progress.

Then dead lock is caused:

1) inside CPU hotplug handler, CPU hotplug lock is held, and blk-mq's handler is waiting for inflight IO

2) error handler is waiting for CPU hotplug lock

3) inflight IO can't be completed in blk-mq's CPU hotplug handler because error handling can't provide forward progress.

Solve the deadlock by not holding CPU hotplug lock in group_cpus_evenly(), in which two stage spreads are taken: 1) the 1st stage is over all present CPUs; 2) the end stage is over all other CPUs.

Turns out the two stage spread just needs consistent 'cpu_present_mask', and remove the CPU hotplug lock by storing it into one local cache. This way doesn't change correctness, because all CPUs are still covered.

---
V2:
- fix "Cc: block list"
- add tested-by tag

 lib/group_cpus.c | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)
lib/group_cpus.c +16 -6
index aa3f6815bb12..15006e79196f 100644
@@ -348,6 +348,7 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 {
 	unsigned int curgrp = 0, nr_present = 0, nr_others = 0;
 	cpumask_var_t *node_to_cpumask;
+	cpumask_var_t local_cpu_present_mask;
 	cpumask_var_t nmsk, npresmsk;
 	int ret = -ENOMEM;
 	struct cpumask *masks = NULL;
@@ -355,6 +356,16 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 	if (!zalloc_cpumask_var(&nmsk, GFP_KERNEL))
 		return NULL;
 
+	if (!zalloc_cpumask_var(&local_cpu_present_mask, GFP_KERNEL))
+		goto fail_local_pres_mask;
+
+	/*
+	 * Make a local cache of 'cpu_present_mask', so the two stages
+	 * spread can observe consistent 'cpu_present_mask' without holding
+	 * cpu hotplug lock.
+	 */
+	cpumask_copy(local_cpu_present_mask, cpu_present_mask);
+
 	if (!zalloc_cpumask_var(&npresmsk, GFP_KERNEL))
 		goto fail_nmsk;
 
@@ -366,13 +377,11 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 	if (!masks)
 		goto fail_node_to_cpumask;
 
-	/* Stabilize the cpumasks */
-	cpus_read_lock();
 	build_node_to_cpumask(node_to_cpumask);
 
 	/* grouping present CPUs first */
 	ret = __group_cpus_evenly(curgrp, numgrps, node_to_cpumask,
-				  cpu_present_mask, nmsk, masks);
+				  local_cpu_present_mask, nmsk, masks);
 	if (ret < 0)
 		goto fail_build_affinity;
 	nr_present = ret;
@@ -387,15 +396,13 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 		curgrp = 0;
 	else
 		curgrp = nr_present;
-	cpumask_andnot(npresmsk, cpu_possible_mask, cpu_present_mask);
+	cpumask_andnot(npresmsk, cpu_possible_mask, local_cpu_present_mask);
 	ret = __group_cpus_evenly(curgrp, numgrps, node_to_cpumask,
 				  npresmsk, nmsk, masks);
 	if (ret >= 0)
 		nr_others = ret;
 
  fail_build_affinity:
-	cpus_read_unlock();
-
 	if (ret >= 0)
 		WARN_ON(nr_present + nr_others < numgrps);
 
@@ -406,6 +413,9 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 	free_cpumask_var(npresmsk);
 
  fail_nmsk:
+	free_cpumask_var(local_cpu_present_mask);
+
+ fail_local_pres_mask:
 	free_cpumask_var(nmsk);
 	if (ret < 0) {
 		kfree(masks);
Signature · 1 line
2.40.1
Reviewed-by on v2 1/1
#1 lore raw

Hi,

Quoted 55 lines · Ming Lei

group_cpus_evenly() could be part of storage driver's error handler, such as nvme driver, when may happen during CPU hotplug, in which storage queue has to drain its pending IOs because all CPUs associated with the queue are offline and the queue is becoming inactive. And handling IO needs error handler to provide forward progress.

Then dead lock is caused:

1) inside CPU hotplug handler, CPU hotplug lock is held, and blk-mq's handler is waiting for inflight IO

2) error handler is waiting for CPU hotplug lock

3) inflight IO can't be completed in blk-mq's CPU hotplug handler because error handling can't provide forward progress.

Solve the deadlock by not holding CPU hotplug lock in group_cpus_evenly(), in which two stage spreads are taken: 1) the 1st stage is over all present CPUs; 2) the end stage is over all other CPUs.

Turns out the two stage spread just needs consistent 'cpu_present_mask', and remove the CPU hotplug lock by storing it into one local cache. This way doesn't change correctness, because all CPUs are still covered.

---
V2:
- fix "Cc: block list"
- add tested-by tag

 lib/group_cpus.c | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)
lib/group_cpus.c +1 -0
index aa3f6815bb12..15006e79196f 100644
@@ -348,6 +348,7 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 {
 	unsigned int curgrp = 0, nr_present = 0, nr_others = 0;
 	cpumask_var_t *node_to_cpumask;
+	cpumask_var_t local_cpu_present_mask;
 	cpumask_var_t nmsk, npresmsk;
 	int ret = -ENOMEM;
 	struct cpumask *masks = NULL;
@@ -355,6 +356,16 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 	if (!zalloc_cpumask_var(&nmsk, GFP_KERNEL))
 		return NULL;
 

lib/group_cpus.c struct cpumask *group_cpus_evenly(unsigned int numgrps) · line 368

Quoted 10 lines
lib/group_cpus.c +10 -0
+	if (!zalloc_cpumask_var(&local_cpu_present_mask, GFP_KERNEL))
+		goto fail_local_pres_mask;
+
+	/*
+	 * Make a local cache of 'cpu_present_mask', so the two stages
+	 * spread can observe consistent 'cpu_present_mask' without holding
+	 * cpu hotplug lock.
+	 */
+	cpumask_copy(local_cpu_present_mask, cpu_present_mask);
+

Maybe we can reuse npresmsk instead of allocating another cpumask? In the first stage: npresmsk = cpu_present_mask
In the second stage: npresmsk = cpu_possible_mask & ~npresmsk

Quoted 24 lines
lib/group_cpus.c +2 -4
 	if (!zalloc_cpumask_var(&npresmsk, GFP_KERNEL))
 		goto fail_nmsk;
 
@@ -366,13 +377,11 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 	if (!masks)
 		goto fail_node_to_cpumask;
 
-	/* Stabilize the cpumasks */
-	cpus_read_lock();
 	build_node_to_cpumask(node_to_cpumask);
 
 	/* grouping present CPUs first */
 	ret = __group_cpus_evenly(curgrp, numgrps, node_to_cpumask,
-				  cpu_present_mask, nmsk, masks);
+				  local_cpu_present_mask, nmsk, masks);
 	if (ret < 0)
 		goto fail_build_affinity;
 	nr_present = ret;
@@ -387,15 +396,13 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 		curgrp = 0;
 	else
 		curgrp = nr_present;
-	cpumask_andnot(npresmsk, cpu_possible_mask, cpu_present_mask);
+	cpumask_andnot(npresmsk, cpu_possible_mask, local_cpu_present_mask);

lib/group_cpus.c struct cpumask *group_cpus_evenly(unsigned int numgrps) · line 407

Quoted 10 lines
lib/group_cpus.c +0 -2
 	ret = __group_cpus_evenly(curgrp, numgrps, node_to_cpumask,
 				  npresmsk, nmsk, masks);
 	if (ret >= 0)
 		nr_others = ret;
 
  fail_build_affinity:
-	cpus_read_unlock();
-
 	if (ret >= 0)
 		WARN_ON(nr_present + nr_others < numgrps);

This fail_build_affinity tag seems unneeded anymore.

The patch looks good to me:

Thanks.

Quoted 11 lines
lib/group_cpus.c +3 -0
@@ -406,6 +413,9 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 	free_cpumask_var(npresmsk);
 
  fail_nmsk:
+	free_cpumask_var(local_cpu_present_mask);
+
+ fail_local_pres_mask:
 	free_cpumask_var(nmsk);
 	if (ret < 0) {
 		kfree(masks);
author
#2 lore raw
Quoted 58 lines · Chengming Zhou

Hi,

Quoted 55 lines · Ming Lei

group_cpus_evenly() could be part of storage driver's error handler, such as nvme driver, when may happen during CPU hotplug, in which storage queue has to drain its pending IOs because all CPUs associated with the queue are offline and the queue is becoming inactive. And handling IO needs error handler to provide forward progress.

Then dead lock is caused:

1) inside CPU hotplug handler, CPU hotplug lock is held, and blk-mq's handler is waiting for inflight IO

2) error handler is waiting for CPU hotplug lock

3) inflight IO can't be completed in blk-mq's CPU hotplug handler because error handling can't provide forward progress.

Solve the deadlock by not holding CPU hotplug lock in group_cpus_evenly(), in which two stage spreads are taken: 1) the 1st stage is over all present CPUs; 2) the end stage is over all other CPUs.

Turns out the two stage spread just needs consistent 'cpu_present_mask', and remove the CPU hotplug lock by storing it into one local cache. This way doesn't change correctness, because all CPUs are still covered.

---
V2:
- fix "Cc: block list"
- add tested-by tag

 lib/group_cpus.c | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)
lib/group_cpus.c +1 -0
index aa3f6815bb12..15006e79196f 100644
@@ -348,6 +348,7 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 {
 	unsigned int curgrp = 0, nr_present = 0, nr_others = 0;
 	cpumask_var_t *node_to_cpumask;
+	cpumask_var_t local_cpu_present_mask;
 	cpumask_var_t nmsk, npresmsk;
 	int ret = -ENOMEM;
 	struct cpumask *masks = NULL;
@@ -355,6 +356,16 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
 	if (!zalloc_cpumask_var(&nmsk, GFP_KERNEL))
 		return NULL;
 
Quoted 14 lines

lib/group_cpus.c struct cpumask *group_cpus_evenly(unsigned int numgrps) · line 368

Quoted 10 lines
lib/group_cpus.c +10 -0
+	if (!zalloc_cpumask_var(&local_cpu_present_mask, GFP_KERNEL))
+		goto fail_local_pres_mask;
+
+	/*
+	 * Make a local cache of 'cpu_present_mask', so the two stages
+	 * spread can observe consistent 'cpu_present_mask' without holding
+	 * cpu hotplug lock.
+	 */
+	cpumask_copy(local_cpu_present_mask, cpu_present_mask);
+

Maybe we can reuse npresmsk instead of allocating another cpumask? In the first stage: npresmsk = cpu_present_mask
In the second stage: npresmsk = cpu_possible_mask & ~npresmsk

Good idea!

Thanks,
Ming