sched_ext: Maintain an online cid mask in the scheduler arenagrafted

Schedulers on the default cid mapping treat [0, nr_online_cids) as the
online set and restart on hotplug. Schedulers that install their own mapping
with scx_bpf_cid_override() have no way to learn which cids are online: the
count no longer identifies members and the CPU-form cpumask is unusable from
cid programs. This is an obvious hole in the cid API.

Add scx_bpf_online_cmask(), a kernel-maintained cmask in the scheduler's
arena, allocated alongside the per-CPU scratch masks and populated after the
cid mapping is finalized and before ops.init(), for child schedulers too.
The pointer stays valid through ops.exit() with no reference to take. It is
the arena offset as a void pointer, the same form struct_ops arena arguments
arrive in. The verifier types the void return as a scalar for the program's
arena cast.

The mask follows the SCX hotplug notifications: seeded from cpu_active_mask
and updated before ops.cid_online/offline() runs, so it lags cpu_online_mask
only inside a hotplug transition. Updates walk the scheduler list under the
lock that also serializes unlinking. Reads are live, not atomic snapshots.
Root initialization excludes hotplug.

v2: Reworded the getter kerneldoc (Andrea Righi).

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>

1 files changed, 657 insertions, 0 deletions

diff --git a/sound/soc/sof/sof-client.c b/sound/soc/sof/sof-client.c
new file mode 100644
index 000000000..c7bbf09e5
--- /dev/null
+++ b/sound/soc/sof/sof-client.c
@@ -0,0 +1,657 @@
+// SPDX-License-Identifier: GPL-2.0-only
+//
+// Copyright(c) 2022 Intel Corporation
+//
+// Authors: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
+//	    Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
+//
+
+#include <linux/debugfs.h>
+#include <linux/errno.h>
+#include <linux/list.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/slab.h>
+#include <sound/sof/ipc4/header.h>
+#include "ops.h"
+#include "sof-client.h"
+#include "sof-priv.h"
+#include "ipc3-priv.h"
+#include "ipc4-priv.h"
+
+/**
+ * struct sof_ipc_event_entry - IPC client event description
+ * @ipc_msg_type:	IPC msg type of the event the client is interested
+ * @cdev:		sof_client_dev of the requesting client
+ * @callback:		Callback function of the client
+ * @list:		item in SOF core client event list
+ */
+struct sof_ipc_event_entry {
+	u32 ipc_msg_type;
+	struct sof_client_dev *cdev;
+	sof_client_event_callback callback;
+	struct list_head list;
+};
+
+/**
+ * struct sof_state_event_entry - DSP panic event subscription entry
+ * @cdev:		sof_client_dev of the requesting client
+ * @callback:		Callback function of the client
+ * @list:		item in SOF core client event list
+ */
+struct sof_state_event_entry {
+	struct sof_client_dev *cdev;
+	sof_client_fw_state_callback callback;
+	struct list_head list;
+};
+
+/**
+ * struct sof_client_dev_entry - client device entry for internal management use
+ * @sdev:	pointer to SOF core device struct
+ * @list:	item in SOF core client dev list
+ * @client_dev: SOF client device
+ */
+struct sof_client_dev_entry {
+	struct snd_sof_dev *sdev;
+	struct list_head list;
+
+	struct sof_client_dev client_dev;
+};
+
+#define cdev_to_centry(cdev) \
+	container_of(cdev, struct sof_client_dev_entry, client_dev)
+
+static void sof_client_auxdev_release(struct device *dev)
+{
+	struct auxiliary_device *auxdev = to_auxiliary_dev(dev);
+	struct sof_client_dev *cdev = auxiliary_dev_to_sof_client_dev(auxdev);
+	struct sof_client_dev_entry *centry = cdev_to_centry(cdev);
+
+	kfree(cdev->auxdev.dev.platform_data);
+	kfree(centry);
+}
+
+static int sof_client_dev_add_data(struct sof_client_dev *cdev, const void *data,
+				   size_t size)
+{
+	void *d = NULL;
+
+	if (data) {
+		d = kmemdup(data, size, GFP_KERNEL);
+		if (!d)
+			return -ENOMEM;
+	}
+
+	cdev->auxdev.dev.platform_data = d;
+	return 0;
+}
+
+#if IS_ENABLED(CONFIG_SND_SOC_SOF_DEBUG_IPC_FLOOD_TEST)
+static int sof_register_ipc_flood_test(struct snd_sof_dev *sdev)
+{
+	int ret = 0;
+	int i;
+
+	if (sdev->pdata->ipc_type != SOF_IPC_TYPE_3)
+		return 0;
+
+	for (i = 0; i < CONFIG_SND_SOC_SOF_DEBUG_IPC_FLOOD_TEST_NUM; i++) {
+		ret = sof_client_dev_register(sdev, "ipc_flood", i, NULL, 0);
+		if (ret < 0)
+			break;
+	}
+
+	if (ret) {
+		for (; i >= 0; --i)
+			sof_client_dev_unregister(sdev, "ipc_flood", i);
+	}
+
+	return ret;
+}
+
+static void sof_unregister_ipc_flood_test(struct snd_sof_dev *sdev)
+{
+	int i;
+
+	for (i = 0; i < CONFIG_SND_SOC_SOF_DEBUG_IPC_FLOOD_TEST_NUM; i++)
+		sof_client_dev_unregister(sdev, "ipc_flood", i);
+}
+#else
+static inline int sof_register_ipc_flood_test(struct snd_sof_dev *sdev)
+{
+	return 0;
+}
+
+static inline void sof_unregister_ipc_flood_test(struct snd_sof_dev *sdev) {}
+#endif /* CONFIG_SND_SOC_SOF_DEBUG_IPC_FLOOD_TEST */
+
+#if IS_ENABLED(CONFIG_SND_SOC_SOF_DEBUG_IPC_MSG_INJECTOR)
+static int sof_register_ipc_msg_injector(struct snd_sof_dev *sdev)
+{
+	return sof_client_dev_register(sdev, "msg_injector", 0, NULL, 0);
+}
+
+static void sof_unregister_ipc_msg_injector(struct snd_sof_dev *sdev)
+{
+	sof_client_dev_unregister(sdev, "msg_injector", 0);
+}
+#else
+static inline int sof_register_ipc_msg_injector(struct snd_sof_dev *sdev)
+{
+	return 0;
+}
+
+static inline void sof_unregister_ipc_msg_injector(struct snd_sof_dev *sdev) {}
+#endif /* CONFIG_SND_SOC_SOF_DEBUG_IPC_MSG_INJECTOR */
+
+#if IS_ENABLED(CONFIG_SND_SOC_SOF_DEBUG_IPC_KERNEL_INJECTOR)
+static int sof_register_ipc_kernel_injector(struct snd_sof_dev *sdev)
+{
+	/* Only IPC3 supported right now */
+	if (sdev->pdata->ipc_type != SOF_IPC_TYPE_3)
+		return 0;
+
+	return sof_client_dev_register(sdev, "kernel_injector", 0, NULL, 0);
+}
+
+static void sof_unregister_ipc_kernel_injector(struct snd_sof_dev *sdev)
+{
+	sof_client_dev_unregister(sdev, "kernel_injector", 0);
+}
+#else
+static inline int sof_register_ipc_kernel_injector(struct snd_sof_dev *sdev)
+{
+	return 0;
+}
+
+static inline void sof_unregister_ipc_kernel_injector(struct snd_sof_dev *sdev) {}
+#endif /* CONFIG_SND_SOC_SOF_DEBUG_IPC_KERNEL_INJECTOR */
+
+int sof_register_clients(struct snd_sof_dev *sdev)
+{
+	int ret;
+
+	if (sdev->dspless_mode_selected)
+		return 0;
+
+	/* Register platform independent client devices */
+	ret = sof_register_ipc_flood_test(sdev);
+	if (ret) {
+		dev_err(sdev->dev, "IPC flood test client registration failed\n");
+		return ret;
+	}
+
+	ret = sof_register_ipc_msg_injector(sdev);
+	if (ret) {
+		dev_err(sdev->dev, "IPC message injector client registration failed\n");
+		goto err_msg_injector;
+	}
+
+	ret = sof_register_ipc_kernel_injector(sdev);
+	if (ret) {
+		dev_err(sdev->dev, "IPC kernel injector client registration failed\n");
+		goto err_kernel_injector;
+	}
+
+	/* Platform dependent client device registration */
+
+	if (sof_ops(sdev) && sof_ops(sdev)->register_ipc_clients)
+		ret = sof_ops(sdev)->register_ipc_clients(sdev);
+
+	if (!ret)
+		return 0;
+
+	sof_unregister_ipc_kernel_injector(sdev);
+
+err_kernel_injector:
+	sof_unregister_ipc_msg_injector(sdev);
+
+err_msg_injector:
+	sof_unregister_ipc_flood_test(sdev);
+
+	return ret;
+}
+
+void sof_unregister_clients(struct snd_sof_dev *sdev)
+{
+	if (sof_ops(sdev) && sof_ops(sdev)->unregister_ipc_clients)
+		sof_ops(sdev)->unregister_ipc_clients(sdev);
+
+	sof_unregister_ipc_kernel_injector(sdev);
+	sof_unregister_ipc_msg_injector(sdev);
+	sof_unregister_ipc_flood_test(sdev);
+}
+
+int sof_client_dev_register(struct snd_sof_dev *sdev, const char *name, u32 id,
+			    const void *data, size_t size)
+{
+	struct sof_client_dev_entry *centry;
+	struct auxiliary_device *auxdev;
+	struct sof_client_dev *cdev;
+	int ret;
+
+	centry = kzalloc(sizeof(*centry), GFP_KERNEL);
+	if (!centry)
+		return -ENOMEM;
+
+	cdev = &centry->client_dev;
+
+	centry->sdev = sdev;
+	auxdev = &cdev->auxdev;
+	auxdev->name = name;
+	auxdev->dev.parent = sdev->dev;
+	auxdev->dev.release = sof_client_auxdev_release;
+	auxdev->id = id;
+
+	ret = sof_client_dev_add_data(cdev, data, size);
+	if (ret < 0)
+		goto err_dev_add_data;
+
+	ret = auxiliary_device_init(auxdev);
+	if (ret < 0) {
+		dev_err(sdev->dev, "failed to initialize client dev %s.%d\n", name, id);
+		goto err_dev_init;
+	}
+
+	ret = auxiliary_device_add(&cdev->auxdev);
+	if (ret < 0) {
+		dev_err(sdev->dev, "failed to add client dev %s.%d\n", name, id);
+		/*
+		 * sof_client_auxdev_release() will be invoked to free up memory
+		 * allocations through put_device()
+		 */
+		auxiliary_device_uninit(&cdev->auxdev);
+		return ret;
+	}
+
+	/* add to list of SOF client devices */
+	scoped_guard(mutex, &sdev->ipc_client_mutex)
+		list_add(&centry->list, &sdev->ipc_client_list);
+
+	return 0;
+
+err_dev_init:
+	kfree(cdev->auxdev.dev.platform_data);
+
+err_dev_add_data:
+	kfree(centry);
+
+	return ret;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_dev_register, "SND_SOC_SOF_CLIENT");
+
+void sof_client_dev_unregister(struct snd_sof_dev *sdev, const char *name, u32 id)
+{
+	struct sof_client_dev_entry *centry;
+
+	guard(mutex)(&sdev->ipc_client_mutex);
+
+	/*
+	 * sof_client_auxdev_release() will be invoked to free up memory
+	 * allocations through put_device()
+	 */
+	list_for_each_entry(centry, &sdev->ipc_client_list, list) {
+		struct sof_client_dev *cdev = &centry->client_dev;
+
+		if (!strcmp(cdev->auxdev.name, name) && cdev->auxdev.id == id) {
+			list_del(&centry->list);
+			auxiliary_device_delete(&cdev->auxdev);
+			auxiliary_device_uninit(&cdev->auxdev);
+			break;
+		}
+	}
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_dev_unregister, "SND_SOC_SOF_CLIENT");
+
+int sof_client_ipc_tx_message(struct sof_client_dev *cdev, void *ipc_msg,
+			      void *reply_data, size_t reply_bytes)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	if (sdev->pdata->ipc_type == SOF_IPC_TYPE_3) {
+		struct sof_ipc_cmd_hdr *hdr = ipc_msg;
+
+		return sof_ipc_tx_message(sdev->ipc, ipc_msg, hdr->size,
+					  reply_data, reply_bytes);
+	} else if (sdev->pdata->ipc_type == SOF_IPC_TYPE_4) {
+		struct sof_ipc4_msg *msg = ipc_msg;
+
+		return sof_ipc_tx_message(sdev->ipc, ipc_msg, msg->data_size,
+					  reply_data, reply_bytes);
+	}
+
+	return -EINVAL;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_ipc_tx_message, "SND_SOC_SOF_CLIENT");
+
+int sof_client_ipc_rx_message(struct sof_client_dev *cdev, void *ipc_msg, void *msg_buf)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	if (IS_ENABLED(CONFIG_SND_SOC_SOF_IPC3) &&
+	    sdev->pdata->ipc_type == SOF_IPC_TYPE_3) {
+		struct sof_ipc_cmd_hdr *hdr = ipc_msg;
+
+		if (hdr->size < sizeof(hdr)) {
+			dev_err(sdev->dev, "The received message size is invalid\n");
+			return -EINVAL;
+		}
+
+		sof_ipc3_do_rx_work(sdev, ipc_msg, msg_buf);
+		return 0;
+	}
+
+	return -EOPNOTSUPP;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_ipc_rx_message, "SND_SOC_SOF_CLIENT");
+
+int sof_client_ipc_set_get_data(struct sof_client_dev *cdev, void *ipc_msg,
+				bool set)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	if (sdev->pdata->ipc_type == SOF_IPC_TYPE_3) {
+		struct sof_ipc_cmd_hdr *hdr = ipc_msg;
+
+		return sof_ipc_set_get_data(sdev->ipc, ipc_msg, hdr->size, set);
+	} else if (sdev->pdata->ipc_type == SOF_IPC_TYPE_4) {
+		struct sof_ipc4_msg *msg = ipc_msg;
+
+		return sof_ipc_set_get_data(sdev->ipc, ipc_msg, msg->data_size,
+					    set);
+	}
+
+	return -EINVAL;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_ipc_set_get_data, "SND_SOC_SOF_CLIENT");
+
+#ifdef CONFIG_SND_SOC_SOF_IPC4
+struct sof_ipc4_fw_module *sof_client_ipc4_find_module(struct sof_client_dev *c, const guid_t *uuid)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(c);
+
+	if (sdev->pdata->ipc_type == SOF_IPC_TYPE_4)
+		return sof_ipc4_find_module_by_uuid(sdev, uuid);
+	dev_err(sdev->dev, "Only supported with IPC4\n");
+
+	return NULL;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_ipc4_find_module, "SND_SOC_SOF_CLIENT");
+
+struct snd_sof_widget *sof_client_ipc4_find_swidget_by_id(struct sof_client_dev *cdev,
+							  u32 module_id, int instance_id)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	if (sdev->pdata->ipc_type == SOF_IPC_TYPE_4)
+		return sof_ipc4_find_swidget_by_ids(sdev, module_id, instance_id);
+	dev_err(sdev->dev, "Only supported with IPC4\n");
+
+	return NULL;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_ipc4_find_swidget_by_id, "SND_SOC_SOF_CLIENT");
+#endif
+
+int sof_suspend_clients(struct snd_sof_dev *sdev, pm_message_t state)
+{
+	const struct auxiliary_driver *adrv;
+	struct sof_client_dev_entry *centry;
+
+	guard(mutex)(&sdev->ipc_client_mutex);
+
+	list_for_each_entry(centry, &sdev->ipc_client_list, list) {
+		struct sof_client_dev *cdev = &centry->client_dev;
+
+		/* Skip devices without loaded driver */
+		if (!cdev->auxdev.dev.driver)
+			continue;
+
+		adrv = to_auxiliary_drv(cdev->auxdev.dev.driver);
+		if (adrv->suspend)
+			adrv->suspend(&cdev->auxdev, state);
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_NS_GPL(sof_suspend_clients, "SND_SOC_SOF_CLIENT");
+
+int sof_resume_clients(struct snd_sof_dev *sdev)
+{
+	const struct auxiliary_driver *adrv;
+	struct sof_client_dev_entry *centry;
+
+	guard(mutex)(&sdev->ipc_client_mutex);
+
+	list_for_each_entry(centry, &sdev->ipc_client_list, list) {
+		struct sof_client_dev *cdev = &centry->client_dev;
+
+		/* Skip devices without loaded driver */
+		if (!cdev->auxdev.dev.driver)
+			continue;
+
+		adrv = to_auxiliary_drv(cdev->auxdev.dev.driver);
+		if (adrv->resume)
+			adrv->resume(&cdev->auxdev);
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_NS_GPL(sof_resume_clients, "SND_SOC_SOF_CLIENT");
+
+struct dentry *sof_client_get_debugfs_root(struct sof_client_dev *cdev)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	return sdev->debugfs_root;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_get_debugfs_root, "SND_SOC_SOF_CLIENT");
+
+/* DMA buffer allocation in client drivers must use the core SOF device */
+struct device *sof_client_get_dma_dev(struct sof_client_dev *cdev)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	return sdev->dev;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_get_dma_dev, "SND_SOC_SOF_CLIENT");
+
+const struct sof_ipc_fw_version *sof_client_get_fw_version(struct sof_client_dev *cdev)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	return &sdev->fw_ready.version;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_get_fw_version, "SND_SOC_SOF_CLIENT");
+
+size_t sof_client_get_ipc_max_payload_size(struct sof_client_dev *cdev)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	return sdev->ipc->max_payload_size;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_get_ipc_max_payload_size, "SND_SOC_SOF_CLIENT");
+
+enum sof_ipc_type sof_client_get_ipc_type(struct sof_client_dev *cdev)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	return sdev->pdata->ipc_type;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_get_ipc_type, "SND_SOC_SOF_CLIENT");
+
+int sof_client_boot_dsp(struct sof_client_dev *cdev)
+{
+	return snd_sof_boot_dsp_firmware(sof_client_dev_to_sof_dev(cdev));
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_boot_dsp, "SND_SOC_SOF_CLIENT");
+
+/* module refcount management of SOF core */
+int sof_client_core_module_get(struct sof_client_dev *cdev)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	if (!try_module_get(sdev->dev->driver->owner))
+		return -ENODEV;
+
+	return 0;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_core_module_get, "SND_SOC_SOF_CLIENT");
+
+void sof_client_core_module_put(struct sof_client_dev *cdev)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	module_put(sdev->dev->driver->owner);
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_core_module_put, "SND_SOC_SOF_CLIENT");
+
+/* IPC event handling */
+void sof_client_ipc_rx_dispatcher(struct snd_sof_dev *sdev, void *msg_buf)
+{
+	struct sof_ipc_event_entry *event;
+	u32 msg_type;
+
+	if (sdev->pdata->ipc_type == SOF_IPC_TYPE_3) {
+		struct sof_ipc_cmd_hdr *hdr = msg_buf;
+
+		msg_type = hdr->cmd & SOF_GLB_TYPE_MASK;
+	} else if (sdev->pdata->ipc_type == SOF_IPC_TYPE_4) {
+		struct sof_ipc4_msg *msg = msg_buf;
+
+		msg_type = SOF_IPC4_NOTIFICATION_TYPE_GET(msg->primary);
+	} else {
+		dev_dbg_once(sdev->dev, "Not supported IPC version: %d\n",
+			     sdev->pdata->ipc_type);
+		return;
+	}
+
+	guard(mutex)(&sdev->client_event_handler_mutex);
+	list_for_each_entry(event, &sdev->ipc_rx_handler_list, list) {
+		if (event->ipc_msg_type == msg_type)
+			event->callback(event->cdev, msg_buf);
+	}
+}
+
+int sof_client_register_ipc_rx_handler(struct sof_client_dev *cdev,
+				       u32 ipc_msg_type,
+				       sof_client_event_callback callback)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+	struct sof_ipc_event_entry *event;
+
+	if (!callback)
+		return -EINVAL;
+
+	if (sdev->pdata->ipc_type == SOF_IPC_TYPE_3) {
+		if (!(ipc_msg_type & SOF_GLB_TYPE_MASK))
+			return -EINVAL;
+	} else if (sdev->pdata->ipc_type == SOF_IPC_TYPE_4) {
+		if (!(ipc_msg_type & SOF_IPC4_NOTIFICATION_TYPE_MASK))
+			return -EINVAL;
+	} else {
+		dev_warn(sdev->dev, "%s: Not supported IPC version: %d\n",
+			 __func__, sdev->pdata->ipc_type);
+		return -EINVAL;
+	}
+
+	event = kmalloc_obj(*event);
+	if (!event)
+		return -ENOMEM;
+
+	event->ipc_msg_type = ipc_msg_type;
+	event->cdev = cdev;
+	event->callback = callback;
+
+	/* add to list of SOF client devices */
+	guard(mutex)(&sdev->client_event_handler_mutex);
+	list_add(&event->list, &sdev->ipc_rx_handler_list);
+
+	return 0;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_register_ipc_rx_handler, "SND_SOC_SOF_CLIENT");
+
+void sof_client_unregister_ipc_rx_handler(struct sof_client_dev *cdev,
+					  u32 ipc_msg_type)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+	struct sof_ipc_event_entry *event;
+
+	guard(mutex)(&sdev->ipc_client_mutex);
+
+	list_for_each_entry(event, &sdev->ipc_rx_handler_list, list) {
+		if (event->cdev == cdev && event->ipc_msg_type == ipc_msg_type) {
+			list_del(&event->list);
+			kfree(event);
+			break;
+		}
+	}
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_unregister_ipc_rx_handler, "SND_SOC_SOF_CLIENT");
+
+/*DSP state notification and query */
+void sof_client_fw_state_dispatcher(struct snd_sof_dev *sdev)
+{
+	struct sof_state_event_entry *event;
+
+	guard(mutex)(&sdev->ipc_client_mutex);
+
+	list_for_each_entry(event, &sdev->fw_state_handler_list, list)
+		event->callback(event->cdev, sdev->fw_state);
+}
+
+int sof_client_register_fw_state_handler(struct sof_client_dev *cdev,
+					 sof_client_fw_state_callback callback)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+	struct sof_state_event_entry *event;
+
+	if (!callback)
+		return -EINVAL;
+
+	event = kmalloc_obj(*event);
+	if (!event)
+		return -ENOMEM;
+
+	event->cdev = cdev;
+	event->callback = callback;
+
+	/* add to list of SOF client devices */
+	guard(mutex)(&sdev->client_event_handler_mutex);
+	list_add(&event->list, &sdev->fw_state_handler_list);
+
+	return 0;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_register_fw_state_handler, "SND_SOC_SOF_CLIENT");
+
+void sof_client_unregister_fw_state_handler(struct sof_client_dev *cdev)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+	struct sof_state_event_entry *event;
+
+	guard(mutex)(&sdev->ipc_client_mutex);
+
+	list_for_each_entry(event, &sdev->fw_state_handler_list, list) {
+		if (event->cdev == cdev) {
+			list_del(&event->list);
+			kfree(event);
+			break;
+		}
+	}
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_unregister_fw_state_handler, "SND_SOC_SOF_CLIENT");
+
+enum sof_fw_state sof_client_get_fw_state(struct sof_client_dev *cdev)
+{
+	struct snd_sof_dev *sdev = sof_client_dev_to_sof_dev(cdev);
+
+	return sdev->fw_state;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_get_fw_state, "SND_SOC_SOF_CLIENT");
+
+struct snd_sof_dev *sof_client_dev_to_sof_dev(struct sof_client_dev *cdev)
+{
+	struct sof_client_dev_entry *centry = cdev_to_centry(cdev);
+
+	return centry->sdev;
+}
+EXPORT_SYMBOL_NS_GPL(sof_client_dev_to_sof_dev, "SND_SOC_SOF_CLIENT");