diff --git a/fs/erofs/Kconfig b/fs/erofs/Kconfig
index 8549037..704fb59 100644
--- a/fs/erofs/Kconfig
+++ b/fs/erofs/Kconfig
@@ -108,3 +108,21 @@
 	  read support.
 
 	  If unsure, say N.
+
+config EROFS_FS_PCPU_KTHREAD
+	bool "EROFS per-cpu decompression kthread workers"
+	depends on EROFS_FS_ZIP
+	help
+	  Saying Y here enables per-CPU kthread workers pool to carry out
+	  async decompression for low latencies on some architectures.
+
+	  If unsure, say N.
+
+config EROFS_FS_PCPU_KTHREAD_HIPRI
+	bool "EROFS high priority per-CPU kthread workers"
+	depends on EROFS_FS_ZIP && EROFS_FS_PCPU_KTHREAD
+	help
+	  This permits EROFS to configure per-CPU kthread workers to run
+	  at higher priority.
+
+	  If unsure, say N.
diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
index 384f642..8ea3f5f 100644
--- a/fs/erofs/zdata.c
+++ b/fs/erofs/zdata.c
@@ -7,7 +7,7 @@
 #include "compress.h"
 #include <linux/prefetch.h>
 #include <linux/psi.h>
-
+#include <linux/cpuhotplug.h>
 #include <trace/events/erofs.h>
 
 #define Z_EROFS_PCLUSTER_MAX_PAGES	(Z_EROFS_PCLUSTER_MAX_SIZE / PAGE_SIZE)
@@ -109,6 +109,7 @@ struct z_erofs_decompressqueue {
 	union {
 		struct completion done;
 		struct work_struct work;
+		struct kthread_work kthread_work;
 	} u;
 	bool eio, sync;
 };
@@ -341,24 +342,128 @@ static void z_erofs_free_pcluster(struct z_erofs_pcluster *pcl)
 
 static struct workqueue_struct *z_erofs_workqueue __read_mostly;
 
-void z_erofs_exit_zip_subsystem(void)
+#ifdef CONFIG_EROFS_FS_PCPU_KTHREAD
+static struct kthread_worker __rcu **z_erofs_pcpu_workers;
+
+static void erofs_destroy_percpu_workers(void)
 {
-	destroy_workqueue(z_erofs_workqueue);
-	z_erofs_destroy_pcluster_pool();
+	struct kthread_worker *worker;
+	unsigned int cpu;
+
+	for_each_possible_cpu(cpu) {
+		worker = rcu_dereference_protected(
+					z_erofs_pcpu_workers[cpu], 1);
+		rcu_assign_pointer(z_erofs_pcpu_workers[cpu], NULL);
+		if (worker)
+			kthread_destroy_worker(worker);
+	}
+	kfree(z_erofs_pcpu_workers);
 }
 
-static inline int z_erofs_init_workqueue(void)
+static struct kthread_worker *erofs_init_percpu_worker(int cpu)
 {
-	const unsigned int onlinecpus = num_possible_cpus();
+	struct kthread_worker *worker =
+		kthread_create_worker_on_cpu(cpu, 0, "erofs_worker/%u", cpu);
 
-	/*
-	 * no need to spawn too many threads, limiting threads could minimum
-	 * scheduling overhead, perhaps per-CPU threads should be better?
-	 */
-	z_erofs_workqueue = alloc_workqueue("erofs_unzipd",
-					    WQ_UNBOUND | WQ_HIGHPRI,
-					    onlinecpus + onlinecpus / 4);
-	return z_erofs_workqueue ? 0 : -ENOMEM;
+	if (IS_ERR(worker))
+		return worker;
+	if (IS_ENABLED(CONFIG_EROFS_FS_PCPU_KTHREAD_HIPRI))
+		sched_set_fifo_low(worker->task);
+	else
+		sched_set_normal(worker->task, 0);
+	return worker;
+}
+
+static int erofs_init_percpu_workers(void)
+{
+	struct kthread_worker *worker;
+	unsigned int cpu;
+
+	z_erofs_pcpu_workers = kcalloc(num_possible_cpus(),
+			sizeof(struct kthread_worker *), GFP_ATOMIC);
+	if (!z_erofs_pcpu_workers)
+		return -ENOMEM;
+
+	for_each_online_cpu(cpu) {	/* could miss cpu{off,on}line? */
+		worker = erofs_init_percpu_worker(cpu);
+		if (!IS_ERR(worker))
+			rcu_assign_pointer(z_erofs_pcpu_workers[cpu], worker);
+	}
+	return 0;
+}
+#else
+static inline void erofs_destroy_percpu_workers(void) {}
+static inline int erofs_init_percpu_workers(void) { return 0; }
+#endif
+
+#if defined(CONFIG_HOTPLUG_CPU) && defined(CONFIG_EROFS_FS_PCPU_KTHREAD)
+static DEFINE_SPINLOCK(z_erofs_pcpu_worker_lock);
+static enum cpuhp_state erofs_cpuhp_state;
+
+static int erofs_cpu_online(unsigned int cpu)
+{
+	struct kthread_worker *worker, *old;
+
+	worker = erofs_init_percpu_worker(cpu);
+	if (IS_ERR(worker))
+		return PTR_ERR(worker);
+
+	spin_lock(&z_erofs_pcpu_worker_lock);
+	old = rcu_dereference_protected(z_erofs_pcpu_workers[cpu],
+			lockdep_is_held(&z_erofs_pcpu_worker_lock));
+	if (!old)
+		rcu_assign_pointer(z_erofs_pcpu_workers[cpu], worker);
+	spin_unlock(&z_erofs_pcpu_worker_lock);
+	if (old)
+		kthread_destroy_worker(worker);
+	return 0;
+}
+
+static int erofs_cpu_offline(unsigned int cpu)
+{
+	struct kthread_worker *worker;
+
+	spin_lock(&z_erofs_pcpu_worker_lock);
+	worker = rcu_dereference_protected(z_erofs_pcpu_workers[cpu],
+			lockdep_is_held(&z_erofs_pcpu_worker_lock));
+	rcu_assign_pointer(z_erofs_pcpu_workers[cpu], NULL);
+	spin_unlock(&z_erofs_pcpu_worker_lock);
+
+	synchronize_rcu();
+	if (worker)
+		kthread_destroy_worker(worker);
+	return 0;
+}
+
+static int erofs_cpu_hotplug_init(void)
+{
+	int state;
+
+	state = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN,
+			"fs/erofs:online", erofs_cpu_online, erofs_cpu_offline);
+	if (state < 0)
+		return state;
+
+	erofs_cpuhp_state = state;
+	return 0;
+}
+
+static void erofs_cpu_hotplug_destroy(void)
+{
+	if (erofs_cpuhp_state)
+		cpuhp_remove_state_nocalls(erofs_cpuhp_state);
+}
+#else /* !CONFIG_HOTPLUG_CPU || !CONFIG_EROFS_FS_PCPU_KTHREAD */
+static inline int erofs_cpu_hotplug_init(void) { return 0; }
+static inline void erofs_cpu_hotplug_destroy(void) {}
+#endif
+
+void z_erofs_exit_zip_subsystem(void)
+{
+	erofs_cpu_hotplug_destroy();
+	erofs_destroy_percpu_workers();
+	destroy_workqueue(z_erofs_workqueue);
+	z_erofs_destroy_pcluster_pool();
 }
 
 int __init z_erofs_init_zip_subsystem(void)
@@ -366,10 +471,29 @@ int __init z_erofs_init_zip_subsystem(void)
 	int err = z_erofs_create_pcluster_pool();
 
 	if (err)
-		return err;
-	err = z_erofs_init_workqueue();
+		goto out_error_pcluster_pool;
+
+	z_erofs_workqueue = alloc_workqueue("erofs_worker",
+			WQ_UNBOUND | WQ_HIGHPRI, num_possible_cpus());
+	if (!z_erofs_workqueue)
+		goto out_error_workqueue_init;
+
+	err = erofs_init_percpu_workers();
 	if (err)
-		z_erofs_destroy_pcluster_pool();
+		goto out_error_pcpu_worker;
+
+	err = erofs_cpu_hotplug_init();
+	if (err < 0)
+		goto out_error_cpuhp_init;
+	return err;
+
+out_error_cpuhp_init:
+	erofs_destroy_percpu_workers();
+out_error_pcpu_worker:
+	destroy_workqueue(z_erofs_workqueue);
+out_error_workqueue_init:
+	z_erofs_destroy_pcluster_pool();
+out_error_pcluster_pool:
 	return err;
 }
 
@@ -1305,11 +1429,17 @@ static void z_erofs_decompressqueue_work(struct work_struct *work)
 
 	DBG_BUGON(bgq->head == Z_EROFS_PCLUSTER_TAIL_CLOSED);
 	z_erofs_decompress_queue(bgq, &pagepool);
-
 	erofs_release_pages(&pagepool);
 	kvfree(bgq);
 }
 
+#ifdef CONFIG_EROFS_FS_PCPU_KTHREAD
+static void z_erofs_decompressqueue_kthread_work(struct kthread_work *work)
+{
+	z_erofs_decompressqueue_work((struct work_struct *)work);
+}
+#endif
+
 static void z_erofs_decompress_kickoff(struct z_erofs_decompressqueue *io,
 				       int bios)
 {
@@ -1324,9 +1454,24 @@ static void z_erofs_decompress_kickoff(struct z_erofs_decompressqueue *io,
 
 	if (atomic_add_return(bios, &io->pending_bios))
 		return;
-	/* Use workqueue and sync decompression for atomic contexts only */
+	/* Use (kthread_)work and sync decompression for atomic contexts only */
 	if (in_atomic() || irqs_disabled()) {
+#ifdef CONFIG_EROFS_FS_PCPU_KTHREAD
+		struct kthread_worker *worker;
+
+		rcu_read_lock();
+		worker = rcu_dereference(
+				z_erofs_pcpu_workers[raw_smp_processor_id()]);
+		if (!worker) {
+			INIT_WORK(&io->u.work, z_erofs_decompressqueue_work);
+			queue_work(z_erofs_workqueue, &io->u.work);
+		} else {
+			kthread_queue_work(worker, &io->u.kthread_work);
+		}
+		rcu_read_unlock();
+#else
 		queue_work(z_erofs_workqueue, &io->u.work);
+#endif
 		/* enable sync decompression for readahead */
 		if (sbi->opt.sync_decompress == EROFS_SYNC_DECOMPRESS_AUTO)
 			sbi->opt.sync_decompress = EROFS_SYNC_DECOMPRESS_FORCE_ON;
@@ -1455,7 +1600,12 @@ static struct z_erofs_decompressqueue *jobqueue_init(struct super_block *sb,
 			*fg = true;
 			goto fg_out;
 		}
+#ifdef CONFIG_EROFS_FS_PCPU_KTHREAD
+		kthread_init_work(&q->u.kthread_work,
+				  z_erofs_decompressqueue_kthread_work);
+#else
 		INIT_WORK(&q->u.work, z_erofs_decompressqueue_work);
+#endif
 	} else {
 fg_out:
 		q = fgq;
@@ -1640,7 +1790,7 @@ static void z_erofs_submit_queue(struct z_erofs_decompress_frontend *f,
 
 	/*
 	 * although background is preferred, no one is pending for submission.
-	 * don't issue workqueue for decompression but drop it directly instead.
+	 * don't issue decompression but drop it directly instead.
 	 */
 	if (!*force_fg && !nr_bios) {
 		kvfree(q[JQ_SUBMIT]);