diff mbox series

[v10,08/26] gunyah: rsc_mgr: Add resource manager RPC core

Message ID 20230214212327.3310128-1-quic_eberman@quicinc.com
State New
Headers show
Series Drivers for Gunyah hypervisor | expand

Commit Message

Elliot Berman Feb. 14, 2023, 9:23 p.m. UTC
The resource manager is a special virtual machine which is always
running on a Gunyah system. It provides APIs for creating and destroying
VMs, secure memory management, sharing/lending of memory between VMs,
and setup of inter-VM communication. Calls to the resource manager are
made via message queues.

This patch implements the basic probing and RPC mechanism to make those
API calls. Request/response calls can be made with gh_rm_call.
Drivers can also register to notifications pushed by RM via
gh_rm_register_notifier

Specific API calls that resource manager supports will be implemented in
subsequent patches.

Signed-off-by: Elliot Berman <quic_eberman@quicinc.com>
---
 drivers/virt/gunyah/Makefile   |   3 +
 drivers/virt/gunyah/rsc_mgr.c  | 604 +++++++++++++++++++++++++++++++++
 drivers/virt/gunyah/rsc_mgr.h  |  77 +++++
 include/linux/gunyah_rsc_mgr.h |  24 ++
 4 files changed, 708 insertions(+)
 create mode 100644 drivers/virt/gunyah/rsc_mgr.c
 create mode 100644 drivers/virt/gunyah/rsc_mgr.h
 create mode 100644 include/linux/gunyah_rsc_mgr.h

Comments

Greg Kroah-Hartman Feb. 16, 2023, 6:43 a.m. UTC | #1
On Tue, Feb 14, 2023 at 01:23:25PM -0800, Elliot Berman wrote:
> +struct gh_rm {
> +	struct device *dev;

What device does this point to?

> +	struct gunyah_resource tx_ghrsc, rx_ghrsc;
> +	struct gh_msgq msgq;
> +	struct mbox_client msgq_client;
> +	struct gh_rm_connection *active_rx_connection;
> +	int last_tx_ret;
> +
> +	struct idr call_idr;
> +	struct mutex call_idr_lock;
> +
> +	struct kmem_cache *cache;
> +	struct mutex send_lock;
> +	struct blocking_notifier_head nh;
> +};

This obviously is the "device" that your system works on, so what are
the lifetime rules of it?  Why isn't is just a real 'struct device' in
the system instead of a random memory blob with a pointer to a device?

What controls the lifetime of this structure and where is the reference
counting logic for it?

And why no documentation for this core structure?

thanks,

greg k-h
Elliot Berman Feb. 16, 2023, 5:40 p.m. UTC | #2
On 2/15/2023 10:43 PM, Greg Kroah-Hartman wrote:
> On Tue, Feb 14, 2023 at 01:23:25PM -0800, Elliot Berman wrote:
>> +struct gh_rm {
>> +	struct device *dev;
> 
> What device does this point to?
> 

The platform device.

>> +	struct gunyah_resource tx_ghrsc, rx_ghrsc;
>> +	struct gh_msgq msgq;
>> +	struct mbox_client msgq_client;
>> +	struct gh_rm_connection *active_rx_connection;
>> +	int last_tx_ret;
>> +
>> +	struct idr call_idr;
>> +	struct mutex call_idr_lock;
>> +
>> +	struct kmem_cache *cache;
>> +	struct mutex send_lock;
>> +	struct blocking_notifier_head nh;
>> +};
> 
> This obviously is the "device" that your system works on, so what are
> the lifetime rules of it?  Why isn't is just a real 'struct device' in
> the system instead of a random memory blob with a pointer to a device?
> 
> What controls the lifetime of this structure and where is the reference
> counting logic for it?
> 

The lifetime of the structure is bound by the platform device that above 
struct device *dev points to. get_gh_rm and put_gh_rm increments the 
device ref counter and ensures lifetime of the struct is also extended.

> And why no documentation for this core structure?
> 

Sure, I will add.

> thanks,
> 
> greg k-h
Srinivas Kandagatla Feb. 20, 2023, 6:10 p.m. UTC | #3
On 14/02/2023 21:23, Elliot Berman wrote:
> 
> The resource manager is a special virtual machine which is always
> running on a Gunyah system. It provides APIs for creating and destroying
> VMs, secure memory management, sharing/lending of memory between VMs,
> and setup of inter-VM communication. Calls to the resource manager are
> made via message queues.
> 
> This patch implements the basic probing and RPC mechanism to make those
> API calls. Request/response calls can be made with gh_rm_call.
> Drivers can also register to notifications pushed by RM via
> gh_rm_register_notifier
> 
> Specific API calls that resource manager supports will be implemented in
> subsequent patches.
> 
> Signed-off-by: Elliot Berman <quic_eberman@quicinc.com>
> ---
>   drivers/virt/gunyah/Makefile   |   3 +
>   drivers/virt/gunyah/rsc_mgr.c  | 604 +++++++++++++++++++++++++++++++++
>   drivers/virt/gunyah/rsc_mgr.h  |  77 +++++
>   include/linux/gunyah_rsc_mgr.h |  24 ++
>   4 files changed, 708 insertions(+)
>   create mode 100644 drivers/virt/gunyah/rsc_mgr.c
>   create mode 100644 drivers/virt/gunyah/rsc_mgr.h
>   create mode 100644 include/linux/gunyah_rsc_mgr.h
> 
> diff --git a/drivers/virt/gunyah/Makefile b/drivers/virt/gunyah/Makefile
> index 34f32110faf9..cc864ff5abbb 100644
> --- a/drivers/virt/gunyah/Makefile
> +++ b/drivers/virt/gunyah/Makefile
> @@ -1,3 +1,6 @@
>   # SPDX-License-Identifier: GPL-2.0
>   
>   obj-$(CONFIG_GUNYAH) += gunyah.o
> +
> +gunyah_rsc_mgr-y += rsc_mgr.o
> +obj-$(CONFIG_GUNYAH) += gunyah_rsc_mgr.o
> diff --git a/drivers/virt/gunyah/rsc_mgr.c b/drivers/virt/gunyah/rsc_mgr.c
> new file mode 100644
> index 000000000000..2a47139873a8
> --- /dev/null
> +++ b/drivers/virt/gunyah/rsc_mgr.c
> @@ -0,0 +1,604 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +/*
> + * Copyright (c) 2022-2023 Qualcomm Innovation Center, Inc. All rights reserved.
> + */
> +
> +#include <linux/of.h>
> +#include <linux/slab.h>
> +#include <linux/mutex.h>
> +#include <linux/sched.h>
> +#include <linux/gunyah.h>
> +#include <linux/module.h>
> +#include <linux/of_irq.h>
> +#include <linux/kthread.h>
why do we need this?

> +#include <linux/notifier.h>
> +#include <linux/workqueue.h>
> +#include <linux/completion.h>
> +#include <linux/gunyah_rsc_mgr.h>
> +#include <linux/platform_device.h>
> +
> +#include "rsc_mgr.h"
> +

...

> +struct gh_rm {
> +	struct device *dev;
> +	struct gunyah_resource tx_ghrsc, rx_ghrsc;
> +	struct gh_msgq msgq;
> +	struct mbox_client msgq_client;
> +	struct gh_rm_connection *active_rx_connection;
> +	int last_tx_ret;
> +

> +	struct idr call_idr;
> +	struct mutex call_idr_lock;

IDR interface is deprecated you should use Xarrays instead here,

Other good thing about Xarrays is that you need not worry about locking 
it uses RCU and internal spinlock, that should simiply code a bit here.

more info at
Documentation/core-api/xarray.rst

> +
> +	struct kmem_cache *cache;
> +	struct mutex send_lock;
> +	struct blocking_notifier_head nh;
> +};
> +
> +static struct gh_rm_connection *gh_rm_alloc_connection(__le32 msg_id, u8 type)
> +{
> +	struct gh_rm_connection *connection;
> +
> +	connection = kzalloc(sizeof(*connection), GFP_KERNEL);
> +	if (!connection)
> +		return ERR_PTR(-ENOMEM);
> +
> +	connection->type = type;
> +	connection->msg_id = msg_id;
> +
> +	return connection;
> +}
> +
> +static int gh_rm_init_connection_payload(struct gh_rm_connection *connection, void *msg,
> +					size_t hdr_size, size_t msg_size)
> +{
> +	size_t max_buf_size, payload_size;
> +	struct gh_rm_rpc_hdr *hdr = msg;
> +
> +	if (hdr_size > msg_size)
> +		return -EINVAL;
> +
> +	payload_size = msg_size - hdr_size;
> +
> +	connection->num_fragments = FIELD_GET(RM_RPC_FRAGMENTS_MASK, hdr->type);
> +	connection->fragments_received = 0;
> +
> +	/* There's not going to be any payload, no need to allocate buffer. */
> +	if (!payload_size && !connection->num_fragments)
> +		return 0;
> +
> +	if (connection->num_fragments > GH_RM_MAX_NUM_FRAGMENTS)
> +		return -EINVAL;
> +
> +	max_buf_size = payload_size + (connection->num_fragments * GH_RM_MAX_MSG_SIZE);
> +
> +	connection->payload = kzalloc(max_buf_size, GFP_KERNEL);
> +	if (!connection->payload)
> +		return -ENOMEM;
> +
> +	memcpy(connection->payload, msg + hdr_size, payload_size);
> +	connection->size = payload_size;
> +	return 0;
> +}
> +
> +static void gh_rm_notif_work(struct work_struct *work)
> +{
> +	struct gh_rm_connection *connection = container_of(work, struct gh_rm_connection,
> +								notification.work);
> +	struct gh_rm *rm = connection->notification.rm;
> +
> +	blocking_notifier_call_chain(&rm->nh, connection->msg_id, connection->payload);
> +
> +	put_gh_rm(rm);
> +	kfree(connection->payload);
if (connection->size)
	kfree(connection->payload);

should we check for payload size before freeing this, Normally kfree 
NULL should be safe, unless connection object is allocated uninitialized.

> +	kfree(connection);
> +}
> +
> +static struct gh_rm_connection *gh_rm_process_notif(struct gh_rm *rm, void *msg, size_t msg_size)
> +{
> +	struct gh_rm_connection *connection;
> +	struct gh_rm_rpc_hdr *hdr = msg;
> +	int ret;
> +
> +	connection = gh_rm_alloc_connection(hdr->msg_id, RM_RPC_TYPE_NOTIF);
> +	if (IS_ERR(connection)) {
> +		dev_err(rm->dev, "Failed to alloc connection for notification: %ld, dropping.\n",
> +			PTR_ERR(connection));
> +		return NULL;
> +	}
> +
> +	get_gh_rm(rm);
> +	connection->notification.rm = rm;
> +	INIT_WORK(&connection->notification.work, gh_rm_notif_work);
> +
> +	ret = gh_rm_init_connection_payload(connection, msg, sizeof(*hdr), msg_size);
> +	if (ret) {
> +		dev_err(rm->dev, "Failed to initialize connection buffer for notification: %d\n",
> +			ret);
put_gh_rm(rm);

is missing.
or move the get and other lines after this check

> +		kfree(connection);
> +		return NULL;
> +	}
> +
> +	return connection;
> +}
> +

> +static int gh_rm_send_request(struct gh_rm *rm, u32 message_id,
> +			      const void *req_buff, size_t req_buff_size,
> +			      struct gh_rm_connection *connection)
> +{
> +	u8 msg_type = FIELD_PREP(RM_RPC_TYPE_MASK, RM_RPC_TYPE_REQUEST);
> +	size_t buff_size_remaining = req_buff_size;
> +	const void *req_buff_curr = req_buff;
> +	struct gh_msgq_tx_data *msg;
> +	struct gh_rm_rpc_hdr *hdr;
> +	u32 cont_fragments = 0;
> +	size_t payload_size;
> +	void *payload;
> +	int ret;
> +
> +	if (req_buff_size)
> +		cont_fragments = (req_buff_size - 1) / GH_RM_MAX_MSG_SIZE;
> +
> +	if (req_buff_size > GH_RM_MAX_NUM_FRAGMENTS * GH_RM_MAX_MSG_SIZE) {
> +		pr_warn("Limit exceeded for the number of fragments: %u\n", cont_fragments);
> +		dump_stack();
> +		return -E2BIG;
> +	}
> +
> +	ret = mutex_lock_interruptible(&rm->send_lock);
> +	if (ret)
> +		return ret;
> +
> +	/* Consider also the 'request' packet for the loop count */
> +	do {
> +		msg = kmem_cache_zalloc(rm->cache, GFP_KERNEL);
> +		if (!msg) {
> +			ret = -ENOMEM;
> +			goto out;
> +		}
> +
> +		/* Fill header */
> +		hdr = (struct gh_rm_rpc_hdr *)msg->data;
> +		hdr->api = RM_RPC_API;
> +		hdr->type = msg_type | FIELD_PREP(RM_RPC_FRAGMENTS_MASK, cont_fragments);
> +		hdr->seq = cpu_to_le16(connection->reply.seq);
> +		hdr->msg_id = cpu_to_le32(message_id);
> +
> +		/* Copy payload */
> +		payload = hdr + 1;
> +		payload_size = min(buff_size_remaining, GH_RM_MAX_MSG_SIZE);
> +		memcpy(payload, req_buff_curr, payload_size);
> +		req_buff_curr += payload_size;
> +		buff_size_remaining -= payload_size;
> +
> +		/* Force the last fragment to immediately alert the receiver */
> +		msg->push = !buff_size_remaining;
> +		msg->length = sizeof(*hdr) + payload_size;
> +
> +		ret = mbox_send_message(gh_msgq_chan(&rm->msgq), msg);
> +		if (ret < 0) {
> +			kmem_cache_free(rm->cache, msg);
> +			break;
> +		}
> +
> +		if (rm->last_tx_ret) {
> +			ret = rm->last_tx_ret;
> +			break;
> +		}
> +
> +		msg_type = FIELD_PREP(RM_RPC_TYPE_MASK, RM_RPC_TYPE_CONTINUATION);
> +	} while (buff_size_remaining);
> +
> +out:
> +	mutex_unlock(&rm->send_lock);
> +	return ret < 0 ? ret : 0;
> +}
> +
> +/**
> + * gh_rm_call: Achieve request-response type communication with RPC
> + * @rm: Pointer to Gunyah resource manager internal data
> + * @message_id: The RM RPC message-id
> + * @req_buff: Request buffer that contains the payload
> + * @req_buff_size: Total size of the payload
> + * @resp_buf: Pointer to a response buffer
> + * @resp_buff_size: Size of the response buffer
> + *
> + * Make a request to the RM-VM and wait for reply back. For a successful
> + * response, the function returns the payload. The size of the payload is set in
> + * resp_buff_size. The resp_buf should be freed by the caller.
> + *
> + * req_buff should be not NULL for req_buff_size >0. If req_buff_size == 0,
> + * req_buff *can* be NULL and no additional payload is sent.
> + *
> + * Context: Process context. Will sleep waiting for reply.
> + * Return: 0 on success. <0 if error.
> + */
> +int gh_rm_call(struct gh_rm *rm, u32 message_id, void *req_buff, size_t req_buff_size,
> +		void **resp_buf, size_t *resp_buff_size)
> +{
> +	struct gh_rm_connection *connection;
> +	int ret;
> +
> +	/* message_id 0 is reserved. req_buff_size implies req_buf is not NULL */
> +	if (!message_id || (!req_buff && req_buff_size) || !rm)
> +		return -EINVAL;
> +
> +	connection = gh_rm_alloc_connection(cpu_to_le32(message_id), RM_RPC_TYPE_REPLY);
> +	if (IS_ERR(connection))
> +		return PTR_ERR(connection);
> +
> +	init_completion(&connection->reply.seq_done);
> +
> +	/* Allocate a new seq number for this connection */
> +	mutex_lock(&rm->call_idr_lock);
> +	ret = idr_alloc_cyclic(&rm->call_idr, connection, 0, U16_MAX,
> +						GFP_KERNEL);
> +	mutex_unlock(&rm->call_idr_lock);
> +	if (ret < 0)
> +		goto out;

new line.

> +	connection->reply.seq = ret;
> +
> +	/* Send the request to the Resource Manager */
> +	ret = gh_rm_send_request(rm, message_id, req_buff, req_buff_size, connection);
> +	if (ret < 0)
> +		goto out;
> +
> +	/* Wait for response */
> +	ret = wait_for_completion_interruptible(&connection->reply.seq_done);
> +	if (ret)
> +		goto out;
> +
> +	/* Check for internal (kernel) error waiting for the response */
> +	if (connection->reply.ret) {
> +		ret = connection->reply.ret;
> +		if (ret != -ENOMEM)
> +			kfree(connection->payload);
> +		goto out;
> +	}
> +
> +	/* Got a response, did resource manager give us an error? */
> +	if (connection->reply.rm_error != GH_RM_ERROR_OK) {
> +		pr_warn("RM rejected message %08x. Error: %d\n", message_id,
> +			connection->reply.rm_error);
> +		dump_stack();
> +		ret = gh_rm_remap_error(connection->reply.rm_error);
> +		kfree(connection->payload);
> +		goto out;
> +	}
> +
> +	/* Everything looks good, return the payload */
> +	*resp_buff_size = connection->size;
> +	if (connection->size)
> +		*resp_buf = connection->payload;
> +	else {
> +		/* kfree in case RM sent us multiple fragments but never any data in
> +		 * those fragments. We would've allocated memory for it, but connection->size == 0
> +		 */
> +		kfree(connection->payload);
> +	}
> +
> +out:
> +	mutex_lock(&rm->call_idr_lock);
> +	idr_remove(&rm->call_idr, connection->reply.seq);
> +	mutex_unlock(&rm->call_idr_lock);
> +	kfree(connection);
> +	return ret;
> +}
> +
> +
> +int gh_rm_notifier_register(struct gh_rm *rm, struct notifier_block *nb)
> +{
> +	return blocking_notifier_chain_register(&rm->nh, nb);
> +}
> +EXPORT_SYMBOL_GPL(gh_rm_notifier_register);
> +
> +int gh_rm_notifier_unregister(struct gh_rm *rm, struct notifier_block *nb)
> +{
> +	return blocking_notifier_chain_unregister(&rm->nh, nb);
> +}
> +EXPORT_SYMBOL_GPL(gh_rm_notifier_unregister);
> +
> +void get_gh_rm(struct gh_rm *rm)
> +{
> +	get_device(rm->dev);
> +}
> +EXPORT_SYMBOL_GPL(get_gh_rm);

Can we have some consistency in the exported symbol naming,
we have two combinations now.

EXPORT_SYMBOL_GPL(gh_rm_notifier_register);
EXPORT_SYMBOL_GPL(get_gh_rm);

lets stick to one.
> +
> +void put_gh_rm(struct gh_rm *rm)
> +{
> +	put_device(rm->dev);
> +}
> +EXPORT_SYMBOL_GPL(put_gh_rm);
> 
...

> +
> +static int gh_rm_drv_probe(struct platform_device *pdev)
> +{
> +	struct gh_msgq_tx_data *msg;
> +	struct gh_rm *rm;
> +	int ret;
> +
How are we ensuring that gunyah driver is probed before this driver?


> +	rm = devm_kzalloc(&pdev->dev, sizeof(*rm), GFP_KERNEL);
> +	if (!rm)
> +		return -ENOMEM;
> +
> +	platform_set_drvdata(pdev, rm);
> +	rm->dev = &pdev->dev;
> +
> +	mutex_init(&rm->call_idr_lock);
> +	idr_init(&rm->call_idr);
> +	rm->cache = kmem_cache_create("gh_rm", struct_size(msg, data, GH_MSGQ_MAX_MSG_SIZE), 0,
> +		SLAB_HWCACHE_ALIGN, NULL);
> +	if (!rm->cache)
> +		return -ENOMEM;
new line here would be nice.

> +	mutex_init(&rm->send_lock);
> +	BLOCKING_INIT_NOTIFIER_HEAD(&rm->nh);
> +
> +	ret = gh_msgq_platform_probe_direction(pdev, true, 0, &rm->tx_ghrsc);
> +	if (ret)
> +		goto err_cache;
> +
> +	ret = gh_msgq_platform_probe_direction(pdev, false, 1, &rm->rx_ghrsc);
> +	if (ret)
> +		goto err_cache;
> +
> +	rm->msgq_client.dev = &pdev->dev;
> +	rm->msgq_client.tx_block = true;
> +	rm->msgq_client.rx_callback = gh_rm_msgq_rx_data;
> +	rm->msgq_client.tx_done = gh_rm_msgq_tx_done;
> +
> +	return gh_msgq_init(&pdev->dev, &rm->msgq, &rm->msgq_client, &rm->tx_ghrsc, &rm->rx_ghrsc);
> +err_cache:
> +	kmem_cache_destroy(rm->cache);
> +	return ret;
> +}
> +
> +static int gh_rm_drv_remove(struct platform_device *pdev)
> +{
> +	struct gh_rm *rm = platform_get_drvdata(pdev);
> +
> +	mbox_free_channel(gh_msgq_chan(&rm->msgq));
> +	gh_msgq_remove(&rm->msgq);
> +	kmem_cache_destroy(rm->cache);
> +
> +	return 0;
> +}
> +
> +static const struct of_device_id gh_rm_of_match[] = {
> +	{ .compatible = "gunyah-resource-manager" },
> +	{}
> +};
> +MODULE_DEVICE_TABLE(of, gh_rm_of_match);
> +
> +static struct platform_driver gh_rm_driver = {
> +	.probe = gh_rm_drv_probe,
> +	.remove = gh_rm_drv_remove,
> +	.driver = {
> +		.name = "gh_rsc_mgr",
> +		.of_match_table = gh_rm_of_match,
> +	},
> +};
> +module_platform_driver(gh_rm_driver);
> +
> +MODULE_LICENSE("GPL");
> +MODULE_DESCRIPTION("Gunyah Resource Manager Driver");
> diff --git a/drivers/virt/gunyah/rsc_mgr.h b/drivers/virt/gunyah/rsc_mgr.h
> new file mode 100644
> index 000000000000..d4e799a7526f
> --- /dev/null
> +++ b/drivers/virt/gunyah/rsc_mgr.h
> @@ -0,0 +1,77 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +/*
> + * Copyright (c) 2022-2023 Qualcomm Innovation Center, Inc. All rights reserved.
> + */
> +#ifndef __GH_RSC_MGR_PRIV_H
> +#define __GH_RSC_MGR_PRIV_H
> +
> +#include <linux/gunyah.h>
> +#include <linux/gunyah_rsc_mgr.h>
> +#include <linux/types.h>
> +
<------------------
> +/* RM Error codes */
> +enum gh_rm_error {
> +	GH_RM_ERROR_OK			= 0x0,
> +	GH_RM_ERROR_UNIMPLEMENTED	= 0xFFFFFFFF,
> +	GH_RM_ERROR_NOMEM		= 0x1,
> +	GH_RM_ERROR_NORESOURCE		= 0x2,
> +	GH_RM_ERROR_DENIED		= 0x3,
> +	GH_RM_ERROR_INVALID		= 0x4,
> +	GH_RM_ERROR_BUSY		= 0x5,
> +	GH_RM_ERROR_ARGUMENT_INVALID	= 0x6,
> +	GH_RM_ERROR_HANDLE_INVALID	= 0x7,
> +	GH_RM_ERROR_VALIDATE_FAILED	= 0x8,
> +	GH_RM_ERROR_MAP_FAILED		= 0x9,
> +	GH_RM_ERROR_MEM_INVALID		= 0xA,
> +	GH_RM_ERROR_MEM_INUSE		= 0xB,
> +	GH_RM_ERROR_MEM_RELEASED	= 0xC,
> +	GH_RM_ERROR_VMID_INVALID	= 0xD,
> +	GH_RM_ERROR_LOOKUP_FAILED	= 0xE,
> +	GH_RM_ERROR_IRQ_INVALID		= 0xF,
> +	GH_RM_ERROR_IRQ_INUSE		= 0x10,
> +	GH_RM_ERROR_IRQ_RELEASED	= 0x11,
> +};
> +
> +/**
> + * gh_rm_remap_error() - Remap Gunyah resource manager errors into a Linux error code
> + * @gh_error: "Standard" return value from Gunyah resource manager
> + */
> +static inline int gh_rm_remap_error(enum gh_rm_error rm_error)
> +{
> +	switch (rm_error) {
> +	case GH_RM_ERROR_OK:
> +		return 0;
> +	case GH_RM_ERROR_UNIMPLEMENTED:
> +		return -EOPNOTSUPP;
> +	case GH_RM_ERROR_NOMEM:
> +		return -ENOMEM;
> +	case GH_RM_ERROR_NORESOURCE:
> +		return -ENODEV;
> +	case GH_RM_ERROR_DENIED:
> +		return -EPERM;
> +	case GH_RM_ERROR_BUSY:
> +		return -EBUSY;
> +	case GH_RM_ERROR_INVALID:
> +	case GH_RM_ERROR_ARGUMENT_INVALID:
> +	case GH_RM_ERROR_HANDLE_INVALID:
> +	case GH_RM_ERROR_VALIDATE_FAILED:
> +	case GH_RM_ERROR_MAP_FAILED:
> +	case GH_RM_ERROR_MEM_INVALID:
> +	case GH_RM_ERROR_MEM_INUSE:
> +	case GH_RM_ERROR_MEM_RELEASED:
> +	case GH_RM_ERROR_VMID_INVALID:
> +	case GH_RM_ERROR_LOOKUP_FAILED:
> +	case GH_RM_ERROR_IRQ_INVALID:
> +	case GH_RM_ERROR_IRQ_INUSE:
> +	case GH_RM_ERROR_IRQ_RELEASED:
> +		return -EINVAL;
> +	default:
> +		return -EBADMSG;
> +	}
> +}
> +
---------------->

Only user for the error code coversion is within the rm driver, you 
should just move this to the .c file, I see no value of this in .h 
unless there are some other users for this.



> +struct gh_rm;
> +int gh_rm_call(struct gh_rm *rsc_mgr, u32 message_id, void *req_buff, size_t req_buff_size,
> +		void **resp_buf, size_t *resp_buff_size);
> +
> +#endif
> diff --git a/include/linux/gunyah_rsc_mgr.h b/include/linux/gunyah_rsc_mgr.h
> new file mode 100644
> index 000000000000..c992b3188c8d
> --- /dev/null
> +++ b/include/linux/gunyah_rsc_mgr.h
> @@ -0,0 +1,24 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +/*
> + * Copyright (c) 2022-2023 Qualcomm Innovation Center, Inc. All rights reserved.
> + */
> +
> +#ifndef _GUNYAH_RSC_MGR_H
> +#define _GUNYAH_RSC_MGR_H
> +
> +#include <linux/list.h>
> +#include <linux/notifier.h>
> +#include <linux/gunyah.h>
> +
> +#define GH_VMID_INVAL	U16_MAX
> +
> +/* Gunyah recognizes VMID0 as an alias to the current VM's ID */
> +#define GH_VMID_SELF			0
> +
> +struct gh_rm;
> +int gh_rm_notifier_register(struct gh_rm *rm, struct notifier_block *nb);
> +int gh_rm_notifier_unregister(struct gh_rm *rm, struct notifier_block *nb);
> +void get_gh_rm(struct gh_rm *rm);
> +void put_gh_rm(struct gh_rm *rm);
> +
> +#endif
Srinivas Kandagatla Feb. 23, 2023, 10:29 a.m. UTC | #4
On 22/02/2023 23:18, Elliot Berman wrote:
>>>
>>> +EXPORT_SYMBOL_GPL(gh_rm_notifier_unregister);
>>> +
>>> +void get_gh_rm(struct gh_rm *rm)
>>> +{
>>> +    get_device(rm->dev);
>>> +}
>>> +EXPORT_SYMBOL_GPL(get_gh_rm);
>>
>> Can we have some consistency in the exported symbol naming,
>> we have two combinations now.
>>
>> EXPORT_SYMBOL_GPL(gh_rm_notifier_register);
>> EXPORT_SYMBOL_GPL(get_gh_rm);
>>
>> lets stick to one.
> 
> done.
> 
>>> +
>>> +void put_gh_rm(struct gh_rm *rm)
>>> +{
>>> +    put_device(rm->dev);
>>> +}
>>> +EXPORT_SYMBOL_GPL(put_gh_rm);
>>>
>> ...
>>
>>> +
>>> +static int gh_rm_drv_probe(struct platform_device *pdev)
>>> +{
>>> +    struct gh_msgq_tx_data *msg;
>>> +    struct gh_rm *rm;
>>> +    int ret;
>>> +
>> How are we ensuring that gunyah driver is probed before this driver?
>>
>>
> 
> Which driver?

Am referring to gunyah.ko

TBH, gunyah.c should be merged as part of resource manager, and check if 
uuids and features in probe before proceeding further.


-srini

> 
>>> +    rm = devm_kzalloc(&pdev->dev, sizeof(*rm), GFP_KERNEL);
>>> +    if (!rm)
>>> +        return -ENOMEM;
>>> +
>>> +    platform_set_drvdata(pdev, rm);
>>> +    rm->dev = &pdev->dev;
>>> +
>>> +    mutex_init(&rm->call_idr_lock);
>>> +    idr_init(&rm->call_idr);
>>> +    rm->cache = kmem_cache_create("gh_rm", struct_size(msg, data, 
>>> GH_MSGQ_MAX_MSG_SIZE), 0,
>>> +        SLAB_HWCACHE_ALIGN, NULL);
>>> +    if (!rm->cache)
>>> +        return -ENOMEM;
>> new line here would be nice.
>>
> 
> done.
Alex Elder Feb. 23, 2023, 11:28 p.m. UTC | #5
On 2/14/23 3:23 PM, Elliot Berman wrote:
> 
> The resource manager is a special virtual machine which is always
> running on a Gunyah system. It provides APIs for creating and destroying
> VMs, secure memory management, sharing/lending of memory between VMs,
> and setup of inter-VM communication. Calls to the resource manager are
> made via message queues.
> 
> This patch implements the basic probing and RPC mechanism to make those
> API calls. Request/response calls can be made with gh_rm_call.
> Drivers can also register to notifications pushed by RM via
> gh_rm_register_notifier
> 
> Specific API calls that resource manager supports will be implemented in
> subsequent patches.
> 
> Signed-off-by: Elliot Berman <quic_eberman@quicinc.com>
> ---
>   drivers/virt/gunyah/Makefile   |   3 +
>   drivers/virt/gunyah/rsc_mgr.c  | 604 +++++++++++++++++++++++++++++++++
>   drivers/virt/gunyah/rsc_mgr.h  |  77 +++++
>   include/linux/gunyah_rsc_mgr.h |  24 ++
>   4 files changed, 708 insertions(+)
>   create mode 100644 drivers/virt/gunyah/rsc_mgr.c
>   create mode 100644 drivers/virt/gunyah/rsc_mgr.h
>   create mode 100644 include/linux/gunyah_rsc_mgr.h
> 
> diff --git a/drivers/virt/gunyah/Makefile b/drivers/virt/gunyah/Makefile
> index 34f32110faf9..cc864ff5abbb 100644
> --- a/drivers/virt/gunyah/Makefile
> +++ b/drivers/virt/gunyah/Makefile
> @@ -1,3 +1,6 @@
>   # SPDX-License-Identifier: GPL-2.0
>   
>   obj-$(CONFIG_GUNYAH) += gunyah.o
> +
> +gunyah_rsc_mgr-y += rsc_mgr.o
> +obj-$(CONFIG_GUNYAH) += gunyah_rsc_mgr.o
> diff --git a/drivers/virt/gunyah/rsc_mgr.c b/drivers/virt/gunyah/rsc_mgr.c
> new file mode 100644
> index 000000000000..2a47139873a8
> --- /dev/null
> +++ b/drivers/virt/gunyah/rsc_mgr.c
> @@ -0,0 +1,604 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +/*
> + * Copyright (c) 2022-2023 Qualcomm Innovation Center, Inc. All rights reserved.
> + */
> +
> +#include <linux/of.h>
> +#include <linux/slab.h>
> +#include <linux/mutex.h>
> +#include <linux/sched.h>
> +#include <linux/gunyah.h>
> +#include <linux/module.h>
> +#include <linux/of_irq.h>
> +#include <linux/kthread.h>
> +#include <linux/notifier.h>
> +#include <linux/workqueue.h>
> +#include <linux/completion.h>
> +#include <linux/gunyah_rsc_mgr.h>
> +#include <linux/platform_device.h>
> +
> +#include "rsc_mgr.h"
> +
> +#define RM_RPC_API_VERSION_MASK		GENMASK(3, 0)
> +#define RM_RPC_HEADER_WORDS_MASK	GENMASK(7, 4)
> +#define RM_RPC_API_VERSION		FIELD_PREP(RM_RPC_API_VERSION_MASK, 1)
> +#define RM_RPC_HEADER_WORDS		FIELD_PREP(RM_RPC_HEADER_WORDS_MASK, \
> +						(sizeof(struct gh_rm_rpc_hdr) / sizeof(u32)))
> +#define RM_RPC_API			(RM_RPC_API_VERSION | RM_RPC_HEADER_WORDS)
> +
> +#define RM_RPC_TYPE_CONTINUATION	0x0
> +#define RM_RPC_TYPE_REQUEST		0x1
> +#define RM_RPC_TYPE_REPLY		0x2
> +#define RM_RPC_TYPE_NOTIF		0x3
> +#define RM_RPC_TYPE_MASK		GENMASK(1, 0)
> +
> +#define GH_RM_MAX_NUM_FRAGMENTS		62
> +#define RM_RPC_FRAGMENTS_MASK		GENMASK(7, 2)
> +
> +struct gh_rm_rpc_hdr {
> +	u8 api;
> +	u8 type;
> +	__le16 seq;
> +	__le32 msg_id;
> +} __packed;
> +
> +struct gh_rm_rpc_reply_hdr {
> +	struct gh_rm_rpc_hdr hdr;
> +	__le32 err_code; /* GH_RM_ERROR_* */
> +} __packed;
> +
> +#define GH_RM_MAX_MSG_SIZE	(GH_MSGQ_MAX_MSG_SIZE - sizeof(struct gh_rm_rpc_hdr))
> +
> +/**
> + * struct gh_rm_connection - Represents a complete message from resource manager
> + * @payload: Combined payload of all the fragments (msg headers stripped off).
> + * @size: Size of the payload received so far.
> + * @msg_id: Message ID from the header.

You do not define the @type field here.

> + * @num_fragments: total number of fragments expected to be received.
> + * @fragments_received: fragments received so far.
> + * @reply: Fields used for request/reply sequences
> + * @ret: Linux return code, set in case there was an error processing connection
> + * @type: RM_RPC_TYPE_REPLY or RM_RPC_TYPE_NOTIF.

Oh, I guess you should move the above line up...

> + * @rm_error: For request/reply sequences with standard replies.
> + * @seq: Sequence ID for the main message.
> + * @seq_done: Signals caller that the RM reply has been received

Now that you've defined parts of these in a union, it
might be clearer to document them as separate struct
types defined earlier (above gh_rm_connection).

> + * @notification: Fields used for notifiations

You added a new rm pointer field, not documented here.

> + * @work: Triggered when all fragments of a notification received
> + */
> +struct gh_rm_connection {
> +	void *payload;
> +	size_t size;
> +	__le32 msg_id;
> +	u8 type;
> +
> +	u8 num_fragments;
> +	u8 fragments_received;
> +
> +	union {
> +		struct {
> +			int ret;
> +			u16 seq;
> +			enum gh_rm_error rm_error;
> +			struct completion seq_done;
> +		} reply;
> +
> +		struct {
> +			struct gh_rm *rm;
> +			struct work_struct work;
> +		} notification;
> +	};
> +};
> +
> +struct gh_rm {
> +	struct device *dev;
> +	struct gunyah_resource tx_ghrsc, rx_ghrsc;

Please split the above two definitions into two separate lines.

> +	struct gh_msgq msgq;
> +	struct mbox_client msgq_client;
> +	struct gh_rm_connection *active_rx_connection;
> +	int last_tx_ret;
> +

Maybe the next two fields can just be "idr" and "idr_lock".

> +	struct idr call_idr;
> +	struct mutex call_idr_lock;
> +
> +	struct kmem_cache *cache;
> +	struct mutex send_lock;
> +	struct blocking_notifier_head nh;
> +};

The next function is very simple.  You don't supply a
free_connection counterpart, so maybe it's not even
needed (just open-code it in the two places it's used)?

> +static struct gh_rm_connection *gh_rm_alloc_connection(__le32 msg_id, u8 type)
> +{
> +	struct gh_rm_connection *connection;
> +
> +	connection = kzalloc(sizeof(*connection), GFP_KERNEL);
> +	if (!connection)
> +		return ERR_PTR(-ENOMEM);
> +
> +	connection->type = type;
> +	connection->msg_id = msg_id;
> +
> +	return connection;
> +}
> +
> +static int gh_rm_init_connection_payload(struct gh_rm_connection *connection, void *msg,
> +					size_t hdr_size, size_t msg_size)

The value of hdr_size is *always* sizeof(*hdr), so you can
do without passing it as an argument.

> +{
> +	size_t max_buf_size, payload_size;
> +	struct gh_rm_rpc_hdr *hdr = msg;
> +

It probably sounds dumb, but I'd reverse the values
compared below (and the operator).

> +	if (hdr_size > msg_size)
> +		return -EINVAL;
> +
> +	payload_size = msg_size - hdr_size;
> +
> +	connection->num_fragments = FIELD_GET(RM_RPC_FRAGMENTS_MASK, hdr->type);
> +	connection->fragments_received = 0;
> +
> +	/* There's not going to be any payload, no need to allocate buffer. */
> +	if (!payload_size && !connection->num_fragments)

The payload size is the same across all messages in the
"connection" right?  As is the number of fragments?
It's not even possible/valid to have a zero payload size
and non-zero number of fragments.  I think the second
half of the above test can be dropped.

> +		return 0;
> +
> +	if (connection->num_fragments > GH_RM_MAX_NUM_FRAGMENTS)
> +		return -EINVAL;
> +
> +	max_buf_size = payload_size + (connection->num_fragments * GH_RM_MAX_MSG_SIZE);
> +
> +	connection->payload = kzalloc(max_buf_size, GFP_KERNEL);
> +	if (!connection->payload)
> +		return -ENOMEM;
> +
> +	memcpy(connection->payload, msg + hdr_size, payload_size);

I think I suggested (hdr + 1) rather than (msg + size) elsewhere
and you took that suggestion.  I'd say do it one way or the other,
consistently, everywhere.

> +	connection->size = payload_size;
> +	return 0;
> +}
> +
> +static void gh_rm_notif_work(struct work_struct *work)
> +{
> +	struct gh_rm_connection *connection = container_of(work, struct gh_rm_connection,
> +								notification.work);
> +	struct gh_rm *rm = connection->notification.rm;
> +
> +	blocking_notifier_call_chain(&rm->nh, connection->msg_id, connection->payload);
> +
> +	put_gh_rm(rm);
> +	kfree(connection->payload);
> +	kfree(connection);
> +}
> +
> +static struct gh_rm_connection *gh_rm_process_notif(struct gh_rm *rm, void *msg, size_t msg_size)

I think it might be better if you do some of what the caller
does here.  I.e., verify the current connection is null (and
abort if not and make it NULL), then assign it to the new
connection before you return success.  And return an errno.

> +{
> +	struct gh_rm_connection *connection;
> +	struct gh_rm_rpc_hdr *hdr = msg;
> +	int ret;
> +
> +	connection = gh_rm_alloc_connection(hdr->msg_id, RM_RPC_TYPE_NOTIF);
> +	if (IS_ERR(connection)) {
> +		dev_err(rm->dev, "Failed to alloc connection for notification: %ld, dropping.\n",
> +			PTR_ERR(connection));
> +		return NULL;
> +	}
> +
> +	get_gh_rm(rm);
> +	connection->notification.rm = rm;
> +	INIT_WORK(&connection->notification.work, gh_rm_notif_work);
> +
> +	ret = gh_rm_init_connection_payload(connection, msg, sizeof(*hdr), msg_size);
> +	if (ret) {
> +		dev_err(rm->dev, "Failed to initialize connection buffer for notification: %d\n",
> +			ret);
> +		kfree(connection);
> +		return NULL;
> +	}
> +
> +	return connection;
> +}
> +
> +static struct gh_rm_connection *gh_rm_process_rply(struct gh_rm *rm, void *msg, size_t msg_size)
> +{

Here too, make sure there is no active connection and then
set it within this function if the errno returned is 0.

> +	struct gh_rm_rpc_reply_hdr *reply_hdr = msg;
> +	struct gh_rm_connection *connection;
> +	u16 seq_id = le16_to_cpu(reply_hdr->hdr.seq);
> +
> +	mutex_lock(&rm->call_idr_lock);
> +	connection = idr_find(&rm->call_idr, seq_id);
> +	mutex_unlock(&rm->call_idr_lock);
> +
> +	if (!connection || connection->msg_id != reply_hdr->hdr.msg_id)
> +		return NULL;
> +
> +	if (gh_rm_init_connection_payload(connection, msg, sizeof(*reply_hdr), msg_size)) {
> +		dev_err(rm->dev, "Failed to alloc connection buffer for sequence %d\n", seq_id);
> +		/* Send connection complete and error the client. */
> +		connection->reply.ret = -ENOMEM;
> +		complete(&connection->reply.seq_done);
> +		return NULL;
> +	}
> +
> +	connection->reply.rm_error = le32_to_cpu(reply_hdr->err_code);
> +	return connection;
> +}
> +
> +static int gh_rm_process_cont(struct gh_rm *rm, struct gh_rm_connection *connection,
> +				void *msg, size_t msg_size)

Similar comment here.  Have this function verify there is
a non-null active connection.  Then process the message
and abort if there's an error (and null the active connection
pointer).

> +{
> +	struct gh_rm_rpc_hdr *hdr = msg;
> +	size_t payload_size = msg_size - sizeof(*hdr);
> +
> +	/*
> +	 * hdr->fragments and hdr->msg_id preserves the value from first reply
> +	 * or notif message. To detect mishandling, check it's still intact.
> +	 */
> +	if (connection->msg_id != hdr->msg_id ||
> +		connection->num_fragments != FIELD_GET(RM_RPC_FRAGMENTS_MASK, hdr->type))
> +		return -EINVAL;

Maybe -EBADMSG?

> +
> +	memcpy(connection->payload + connection->size, msg + sizeof(*hdr), payload_size);
> +	connection->size += payload_size;
> +	connection->fragments_received++;
> +	return 0;
> +}
> +
> +static void gh_rm_abort_connection(struct gh_rm_connection *connection)
> +{
> +	switch (connection->type) {
> +	case RM_RPC_TYPE_REPLY:
> +		connection->reply.ret = -EIO;
> +		complete(&connection->reply.seq_done);
> +		break;
> +	case RM_RPC_TYPE_NOTIF:
> +		fallthrough;
> +	default:
> +		kfree(connection->payload);
> +		kfree(connection);
> +	}
> +}
> +
> +static bool gh_rm_complete_connection(struct gh_rm *rm, struct gh_rm_connection *connection)

The only caller of this function passes rm->active_rx_connection
as the second argument.  It is available to you here, so you
can get rid of that argument.

> +{
> +	if (!connection || connection->fragments_received != connection->num_fragments)
> +		return false;
> +
> +	switch (connection->type) {
> +	case RM_RPC_TYPE_REPLY:
> +		complete(&connection->reply.seq_done);
> +		break;
> +	case RM_RPC_TYPE_NOTIF:
> +		schedule_work(&connection->notification.work);
> +		break;
> +	default:
> +		dev_err(rm->dev, "Invalid message type (%d) received\n", connection->type);
> +		gh_rm_abort_connection(connection);
> +		break;
> +	}
> +
> +	return true;
> +}
> +
> +static void gh_rm_msgq_rx_data(struct mbox_client *cl, void *mssg)
> +{
> +	struct gh_rm *rm = container_of(cl, struct gh_rm, msgq_client);
> +	struct gh_msgq_rx_data *rx_data = mssg;
> +	size_t msg_size = rx_data->length;
> +	void *msg = rx_data->data;
> +	struct gh_rm_rpc_hdr *hdr;
> +

Is it required that at least one byte (past the header) will
be received?  I.e., should the "<=" below just be "<"?

> +	if (msg_size <= sizeof(*hdr) || msg_size > GH_MSGQ_MAX_MSG_SIZE)
> +		return;

You previously reported a message here.  These seem like
errors, which if they occur, maybe should be reported.
They seem like "never happen" issues, but it's defensive
to make these checks (which is good).

> +
> +	hdr = msg;
> +	if (hdr->api != RM_RPC_API) {

If this ever happens, is the hardware failing?  It seems
like once Gunyah is initialized and you've checked the
API version once, there should be no need to check it
repeatedly.

> +		dev_err(rm->dev, "Unknown RM RPC API version: %x\n", hdr->api);
> +		return;
> +	}
> +
> +	switch (FIELD_GET(RM_RPC_TYPE_MASK, hdr->type)) {
> +	case RM_RPC_TYPE_NOTIF:
> +		rm->active_rx_connection = gh_rm_process_notif(rm, msg, msg_size);
> +		break;
> +	case RM_RPC_TYPE_REPLY:
> +		rm->active_rx_connection = gh_rm_process_rply(rm, msg, msg_size);
> +		break;
> +	case RM_RPC_TYPE_CONTINUATION:
> +		if (gh_rm_process_cont(rm, rm->active_rx_connection, msg, msg_size)) {
> +			gh_rm_abort_connection(rm->active_rx_connection);
> +			rm->active_rx_connection = NULL;
> +		}
> +		break;
> +	default:
> +		dev_err(rm->dev, "Invalid message type (%lu) received\n",
> +			FIELD_GET(RM_RPC_TYPE_MASK, hdr->type));
> +		return;
> +	}
> +
> +	if (gh_rm_complete_connection(rm, rm->active_rx_connection))
> +		rm->active_rx_connection = NULL;
> +}
> +
> +static void gh_rm_msgq_tx_done(struct mbox_client *cl, void *mssg, int r)
> +{
> +	struct gh_rm *rm = container_of(cl, struct gh_rm, msgq_client);
> +
> +	kmem_cache_free(rm->cache, mssg);
> +	rm->last_tx_ret = r;
> +}
> +
> +static int gh_rm_send_request(struct gh_rm *rm, u32 message_id,
> +			      const void *req_buff, size_t req_buff_size,
> +			      struct gh_rm_connection *connection)
> +{
> +	u8 msg_type = FIELD_PREP(RM_RPC_TYPE_MASK, RM_RPC_TYPE_REQUEST);
> +	size_t buff_size_remaining = req_buff_size;
> +	const void *req_buff_curr = req_buff;
> +	struct gh_msgq_tx_data *msg;
> +	struct gh_rm_rpc_hdr *hdr;
> +	u32 cont_fragments = 0;
> +	size_t payload_size;
> +	void *payload;
> +	int ret;
> +
> +	if (req_buff_size)
> +		cont_fragments = (req_buff_size - 1) / GH_RM_MAX_MSG_SIZE;

Compute this *after* verifying the size isn't too big.

> +
> +	if (req_buff_size > GH_RM_MAX_NUM_FRAGMENTS * GH_RM_MAX_MSG_SIZE) {
> +		pr_warn("Limit exceeded for the number of fragments: %u\n", cont_fragments);
> +		dump_stack();
> +		return -E2BIG;
> +	}
> +
> +	ret = mutex_lock_interruptible(&rm->send_lock);
> +	if (ret)
> +		return ret;
> +
> +	/* Consider also the 'request' packet for the loop count */
> +	do {
> +		msg = kmem_cache_zalloc(rm->cache, GFP_KERNEL);
> +		if (!msg) {
> +			ret = -ENOMEM;
> +			goto out;
> +		}
> +
> +		/* Fill header */
> +		hdr = (struct gh_rm_rpc_hdr *)msg->data;
> +		hdr->api = RM_RPC_API;
> +		hdr->type = msg_type | FIELD_PREP(RM_RPC_FRAGMENTS_MASK, cont_fragments);
> +		hdr->seq = cpu_to_le16(connection->reply.seq);
> +		hdr->msg_id = cpu_to_le32(message_id);

Most of the above are constant for every message.  I think the only
thing that changes is the type field.  It might not make a difference
but you could compute the "generic" header outside the loop and
assign it as a structure, then overwrite the type field.

> +
> +		/* Copy payload */
> +		payload = hdr + 1;
> +		payload_size = min(buff_size_remaining, GH_RM_MAX_MSG_SIZE);
> +		memcpy(payload, req_buff_curr, payload_size);
> +		req_buff_curr += payload_size;
> +		buff_size_remaining -= payload_size;
> +
> +		/* Force the last fragment to immediately alert the receiver */
> +		msg->push = !buff_size_remaining;
> +		msg->length = sizeof(*hdr) + payload_size;
> +
> +		ret = mbox_send_message(gh_msgq_chan(&rm->msgq), msg);
> +		if (ret < 0) {
> +			kmem_cache_free(rm->cache, msg);
> +			break;
> +		}
> +
> +		if (rm->last_tx_ret) {
> +			ret = rm->last_tx_ret;
> +			break;
> +		}
> +
> +		msg_type = FIELD_PREP(RM_RPC_TYPE_MASK, RM_RPC_TYPE_CONTINUATION);
> +	} while (buff_size_remaining);
> +
> +out:
> +	mutex_unlock(&rm->send_lock);
> +	return ret < 0 ? ret : 0;
> +}
> +
> +/**
> + * gh_rm_call: Achieve request-response type communication with RPC
> + * @rm: Pointer to Gunyah resource manager internal data
> + * @message_id: The RM RPC message-id
> + * @req_buff: Request buffer that contains the payload
> + * @req_buff_size: Total size of the payload
> + * @resp_buf: Pointer to a response buffer

Is it "buf" or is it "buff"?  I prefer the former, but you
should be consistent in your namings.

> + * @resp_buff_size: Size of the response buffer
> + *
> + * Make a request to the RM-VM and wait for reply back. For a successful
> + * response, the function returns the payload. The size of the payload is set in
> + * resp_buff_size. The resp_buf should be freed by the caller.

It might not matter, but resp_buf should be freed by the
caller *if 0 is returned* (no error).

> + *
> + * req_buff should be not NULL for req_buff_size >0. If req_buff_size == 0,
> + * req_buff *can* be NULL and no additional payload is sent.
> + *
> + * Context: Process context. Will sleep waiting for reply.
> + * Return: 0 on success. <0 if error.
> + */
> +int gh_rm_call(struct gh_rm *rm, u32 message_id, void *req_buff, size_t req_buff_size,
> +		void **resp_buf, size_t *resp_buff_size)
> +{
> +	struct gh_rm_connection *connection;
> +	int ret;
> +
> +	/* message_id 0 is reserved. req_buff_size implies req_buf is not NULL */
> +	if (!message_id || (!req_buff && req_buff_size) || !rm)
> +		return -EINVAL;
> +
> +	connection = gh_rm_alloc_connection(cpu_to_le32(message_id), RM_RPC_TYPE_REPLY);
> +	if (IS_ERR(connection))
> +		return PTR_ERR(connection);
> +
> +	init_completion(&connection->reply.seq_done);
> +
> +	/* Allocate a new seq number for this connection */
> +	mutex_lock(&rm->call_idr_lock);
> +	ret = idr_alloc_cyclic(&rm->call_idr, connection, 0, U16_MAX,
> +						GFP_KERNEL);
> +	mutex_unlock(&rm->call_idr_lock);
> +	if (ret < 0)
> +		goto out;

You need a different error path label here.  If there's an
error, the IDR allocation failed (so shoudln't be removed).
(Right?)

> +	connection->reply.seq = ret;
> +
> +	/* Send the request to the Resource Manager */
> +	ret = gh_rm_send_request(rm, message_id, req_buff, req_buff_size, connection);
> +	if (ret < 0)
> +		goto out;
> +
> +	/* Wait for response */
> +	ret = wait_for_completion_interruptible(&connection->reply.seq_done);
> +	if (ret)
> +		goto out;
> +
> +	/* Check for internal (kernel) error waiting for the response */
> +	if (connection->reply.ret) {
> +		ret = connection->reply.ret;
> +		if (ret != -ENOMEM)
> +			kfree(connection->payload);
> +		goto out;
> +	}
> +
> +	/* Got a response, did resource manager give us an error? */
> +	if (connection->reply.rm_error != GH_RM_ERROR_OK) {
> +		pr_warn("RM rejected message %08x. Error: %d\n", message_id,
> +			connection->reply.rm_error);
> +		dump_stack();
> +		ret = gh_rm_remap_error(connection->reply.rm_error);
> +		kfree(connection->payload);
> +		goto out;
> +	}
> +
> +	/* Everything looks good, return the payload */
> +	*resp_buff_size = connection->size;
> +	if (connection->size)
> +		*resp_buf = connection->payload;
> +	else {
> +		/* kfree in case RM sent us multiple fragments but never any data in
> +		 * those fragments. We would've allocated memory for it, but connection->size == 0
> +		 */
> +		kfree(connection->payload);
> +	}
> +
> +out:
> +	mutex_lock(&rm->call_idr_lock);
> +	idr_remove(&rm->call_idr, connection->reply.seq);
> +	mutex_unlock(&rm->call_idr_lock);
> +	kfree(connection);
> +	return ret;
> +}
> +
> +
> +int gh_rm_notifier_register(struct gh_rm *rm, struct notifier_block *nb)
> +{
> +	return blocking_notifier_chain_register(&rm->nh, nb);
> +}
> +EXPORT_SYMBOL_GPL(gh_rm_notifier_register);
> +
> +int gh_rm_notifier_unregister(struct gh_rm *rm, struct notifier_block *nb)
> +{
> +	return blocking_notifier_chain_unregister(&rm->nh, nb);
> +}
> +EXPORT_SYMBOL_GPL(gh_rm_notifier_unregister);
> +
> +void get_gh_rm(struct gh_rm *rm)

It is often pretty handy to return the argument in
functions like this.  It simultaneously takes the
reference and assigns the pointer the reference
represents.


> +{
> +	get_device(rm->dev);
> +}
> +EXPORT_SYMBOL_GPL(get_gh_rm);
> +
> +void put_gh_rm(struct gh_rm *rm)
> +{
> +	put_device(rm->dev);
> +}
> +EXPORT_SYMBOL_GPL(put_gh_rm);
> +
> +static int gh_msgq_platform_probe_direction(struct platform_device *pdev,
> +					bool tx, int idx, struct gunyah_resource *ghrsc)
> +{
> +	struct device_node *node = pdev->dev.of_node;
> +	int ret;

I think you should declare idx as a local variable.

	int idx = tx ? 1 : 0;

> +
> +	ghrsc->type = tx ? GUNYAH_RESOURCE_TYPE_MSGQ_TX : GUNYAH_RESOURCE_TYPE_MSGQ_RX;
> +
> +	ghrsc->irq = platform_get_irq(pdev, idx);

Do you suppose you could do platform_get_irq_byname(), and then
specify the names of the IRQs ("rm_tx_irq" and "rm_rx_irq" maybe)?

> +	if (ghrsc->irq < 0) {
> +		dev_err(&pdev->dev, "Failed to get irq%d: %d\n", idx, ghrsc->irq);

Maybe:	"Failed to get %cX IRQ: %d\n", tx ? 'T' : 'R', ghrsc->irq);

> +		return ghrsc->irq;
> +	}
> +
> +	ret = of_property_read_u64_index(node, "reg", idx, &ghrsc->capid);

Is a capability ID a simple (but large) number?

The *resource manager* (which is a very special VM) has to
have both TX and RX message queue capability IDs.  Is there
'any chance that these specific capability IDs have values
that are fixed by the design?  Like, 0 and 1?  I don't know
what they are, but it seems like it *could* be something
fixed by the design, and if that were the case, there would
be no need to specify the "reg" property to get the "capid"
values.

> +	if (ret) {
> +		dev_err(&pdev->dev, "Failed to get capid%d: %d\n", idx, ret);
> +		return ret;
> +	}
> +
> +	return 0;
> +}
> +
> +static int gh_rm_drv_probe(struct platform_device *pdev)
> +{
> +	struct gh_msgq_tx_data *msg;
> +	struct gh_rm *rm;
> +	int ret;
> +
> +	rm = devm_kzalloc(&pdev->dev, sizeof(*rm), GFP_KERNEL);
> +	if (!rm)
> +		return -ENOMEM;
> +
> +	platform_set_drvdata(pdev, rm);
> +	rm->dev = &pdev->dev;
> +
> +	mutex_init(&rm->call_idr_lock);
> +	idr_init(&rm->call_idr);
> +	rm->cache = kmem_cache_create("gh_rm", struct_size(msg, data, GH_MSGQ_MAX_MSG_SIZE), 0,
> +		SLAB_HWCACHE_ALIGN, NULL);
> +	if (!rm->cache)
> +		return -ENOMEM;

If you abstracted the allocation interface for these messages,
you could actually survive without the slab cache here.  But
if this fails, maybe you won't get far anyway.

> +	mutex_init(&rm->send_lock);
> +	BLOCKING_INIT_NOTIFIER_HEAD(&rm->nh);
> +
> +	ret = gh_msgq_platform_probe_direction(pdev, true, 0, &rm->tx_ghrsc);
> +	if (ret)
> +		goto err_cache;
> +
> +	ret = gh_msgq_platform_probe_direction(pdev, false, 1, &rm->rx_ghrsc);
> +	if (ret)
> +		goto err_cache;
> +
> +	rm->msgq_client.dev = &pdev->dev;
> +	rm->msgq_client.tx_block = true;
> +	rm->msgq_client.rx_callback = gh_rm_msgq_rx_data;
> +	rm->msgq_client.tx_done = gh_rm_msgq_tx_done;
> +
> +	return gh_msgq_init(&pdev->dev, &rm->msgq, &rm->msgq_client, &rm->tx_ghrsc, &rm->rx_ghrsc);
> +err_cache:
> +	kmem_cache_destroy(rm->cache);
> +	return ret;
> +}
> +
> +static int gh_rm_drv_remove(struct platform_device *pdev)
> +{
> +	struct gh_rm *rm = platform_get_drvdata(pdev);
> +
> +	mbox_free_channel(gh_msgq_chan(&rm->msgq));
> +	gh_msgq_remove(&rm->msgq);
> +	kmem_cache_destroy(rm->cache);
> +
> +	return 0;
> +}
> +
> +static const struct of_device_id gh_rm_of_match[] = {
> +	{ .compatible = "gunyah-resource-manager" },
> +	{}
> +};
> +MODULE_DEVICE_TABLE(of, gh_rm_of_match);
> +
> +static struct platform_driver gh_rm_driver = {
> +	.probe = gh_rm_drv_probe,
> +	.remove = gh_rm_drv_remove,
> +	.driver = {
> +		.name = "gh_rsc_mgr",
> +		.of_match_table = gh_rm_of_match,
> +	},
> +};
> +module_platform_driver(gh_rm_driver);
> +
> +MODULE_LICENSE("GPL");
> +MODULE_DESCRIPTION("Gunyah Resource Manager Driver");
> diff --git a/drivers/virt/gunyah/rsc_mgr.h b/drivers/virt/gunyah/rsc_mgr.h
> new file mode 100644
> index 000000000000..d4e799a7526f
> --- /dev/null
> +++ b/drivers/virt/gunyah/rsc_mgr.h
> @@ -0,0 +1,77 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +/*
> + * Copyright (c) 2022-2023 Qualcomm Innovation Center, Inc. All rights reserved.
> + */
> +#ifndef __GH_RSC_MGR_PRIV_H
> +#define __GH_RSC_MGR_PRIV_H
> +
> +#include <linux/gunyah.h>
> +#include <linux/gunyah_rsc_mgr.h>
> +#include <linux/types.h>
> +
> +/* RM Error codes */
> +enum gh_rm_error {
> +	GH_RM_ERROR_OK			= 0x0,
> +	GH_RM_ERROR_UNIMPLEMENTED	= 0xFFFFFFFF,
> +	GH_RM_ERROR_NOMEM		= 0x1,
> +	GH_RM_ERROR_NORESOURCE		= 0x2,
> +	GH_RM_ERROR_DENIED		= 0x3,
> +	GH_RM_ERROR_INVALID		= 0x4,
> +	GH_RM_ERROR_BUSY		= 0x5,
> +	GH_RM_ERROR_ARGUMENT_INVALID	= 0x6,
> +	GH_RM_ERROR_HANDLE_INVALID	= 0x7,
> +	GH_RM_ERROR_VALIDATE_FAILED	= 0x8,
> +	GH_RM_ERROR_MAP_FAILED		= 0x9,
> +	GH_RM_ERROR_MEM_INVALID		= 0xA,
> +	GH_RM_ERROR_MEM_INUSE		= 0xB,
> +	GH_RM_ERROR_MEM_RELEASED	= 0xC,
> +	GH_RM_ERROR_VMID_INVALID	= 0xD,
> +	GH_RM_ERROR_LOOKUP_FAILED	= 0xE,
> +	GH_RM_ERROR_IRQ_INVALID		= 0xF,
> +	GH_RM_ERROR_IRQ_INUSE		= 0x10,
> +	GH_RM_ERROR_IRQ_RELEASED	= 0x11,
> +};
> +
> +/**
> + * gh_rm_remap_error() - Remap Gunyah resource manager errors into a Linux error code
> + * @gh_error: "Standard" return value from Gunyah resource manager
> + */
> +static inline int gh_rm_remap_error(enum gh_rm_error rm_error)
> +{
> +	switch (rm_error) {
> +	case GH_RM_ERROR_OK:
> +		return 0;
> +	case GH_RM_ERROR_UNIMPLEMENTED:
> +		return -EOPNOTSUPP;
> +	case GH_RM_ERROR_NOMEM:
> +		return -ENOMEM;
> +	case GH_RM_ERROR_NORESOURCE:
> +		return -ENODEV;
> +	case GH_RM_ERROR_DENIED:
> +		return -EPERM;
> +	case GH_RM_ERROR_BUSY:
> +		return -EBUSY;
> +	case GH_RM_ERROR_INVALID:
> +	case GH_RM_ERROR_ARGUMENT_INVALID:
> +	case GH_RM_ERROR_HANDLE_INVALID:
> +	case GH_RM_ERROR_VALIDATE_FAILED:
> +	case GH_RM_ERROR_MAP_FAILED:
> +	case GH_RM_ERROR_MEM_INVALID:
> +	case GH_RM_ERROR_MEM_INUSE:
> +	case GH_RM_ERROR_MEM_RELEASED:
> +	case GH_RM_ERROR_VMID_INVALID:
> +	case GH_RM_ERROR_LOOKUP_FAILED:
> +	case GH_RM_ERROR_IRQ_INVALID:
> +	case GH_RM_ERROR_IRQ_INUSE:
> +	case GH_RM_ERROR_IRQ_RELEASED:
> +		return -EINVAL;
> +	default:
> +		return -EBADMSG;
> +	}
> +}
> +
> +struct gh_rm;

This might just be my preference, but I like to see declarations
like the one above grouped at the top of the file, under includes.

> +int gh_rm_call(struct gh_rm *rsc_mgr, u32 message_id, void *req_buff, size_t req_buff_size,
> +		void **resp_buf, size_t *resp_buff_size);
> +
> +#endif
> diff --git a/include/linux/gunyah_rsc_mgr.h b/include/linux/gunyah_rsc_mgr.h
> new file mode 100644
> index 000000000000..c992b3188c8d
> --- /dev/null
> +++ b/include/linux/gunyah_rsc_mgr.h
> @@ -0,0 +1,24 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +/*
> + * Copyright (c) 2022-2023 Qualcomm Innovation Center, Inc. All rights reserved.
> + */
> +
> +#ifndef _GUNYAH_RSC_MGR_H
> +#define _GUNYAH_RSC_MGR_H
> +
> +#include <linux/list.h>
> +#include <linux/notifier.h>
> +#include <linux/gunyah.h>
> +
> +#define GH_VMID_INVAL	U16_MAX
> +
> +/* Gunyah recognizes VMID0 as an alias to the current VM's ID */
> +#define GH_VMID_SELF			0

I haven't really checked very well, bur you should *use this*
definition where a VMID is being examined. I.e., if you're
going to define this, then never just compare a VMID against 0.

					-Alex

> +
> +struct gh_rm;
> +int gh_rm_notifier_register(struct gh_rm *rm, struct notifier_block *nb);
> +int gh_rm_notifier_unregister(struct gh_rm *rm, struct notifier_block *nb);
> +void get_gh_rm(struct gh_rm *rm);
> +void put_gh_rm(struct gh_rm *rm);
> +
> +#endif
Elliot Berman Feb. 24, 2023, 10:39 p.m. UTC | #6
On 2/23/2023 3:28 PM, Alex Elder wrote:
> On 2/14/23 3:23 PM, Elliot Berman wrote:
>>
<snip>
>> +}
>> +
>> +static int gh_rm_init_connection_payload(struct gh_rm_connection 
>> *connection, void *msg,
>> +                    size_t hdr_size, size_t msg_size)
> 
> The value of hdr_size is *always* sizeof(*hdr), so you can
> do without passing it as an argument.
> 

hdr_size is different when receiving reply (1 extra word) vs notification.

>> +{
>> +    size_t max_buf_size, payload_size;
>> +    struct gh_rm_rpc_hdr *hdr = msg;
>> +
> 
> It probably sounds dumb, but I'd reverse the values
> compared below (and the operator).
> 
>> +    if (hdr_size > msg_size)
>> +        return -EINVAL;
>> +
>> +    payload_size = msg_size - hdr_size;
>> +
>> +    connection->num_fragments = FIELD_GET(RM_RPC_FRAGMENTS_MASK, 
>> hdr->type);
>> +    connection->fragments_received = 0;
>> +
>> +    /* There's not going to be any payload, no need to allocate 
>> buffer. */
>> +    if (!payload_size && !connection->num_fragments)
> 
> The payload size is the same across all messages in the
> "connection" right?  As is the number of fragments?
> It's not even possible/valid to have a zero payload size
> and non-zero number of fragments.  I think the second
> half of the above test can be dropped.
> 

The RM RPC specification doesn't require that the first message have 
payload. (It makes sense to do it and it does, but that's implementation 
detail)

>> +        return 0;
>> +
>> +    if (connection->num_fragments > GH_RM_MAX_NUM_FRAGMENTS)
>> +        return -EINVAL;
>> +
>> +    max_buf_size = payload_size + (connection->num_fragments * 
>> GH_RM_MAX_MSG_SIZE);
>> +
>> +    connection->payload = kzalloc(max_buf_size, GFP_KERNEL);
>> +    if (!connection->payload)
>> +        return -ENOMEM;
>> +
>> +    memcpy(connection->payload, msg + hdr_size, payload_size);
> 
> I think I suggested (hdr + 1) rather than (msg + size) elsewhere
> and you took that suggestion.  I'd say do it one way or the other,
> consistently, everywhere.
> 

hdr_size != sizeof(*hdr) when we receive a reply message.

>> +    connection->size = payload_size;
>> +    return 0;
>> +}
>> +
>> +static void gh_rm_notif_work(struct work_struct *work)
>> +{
>> +    struct gh_rm_connection *connection = container_of(work, struct 
>> gh_rm_connection,
>> +                                notification.work);
>> +    struct gh_rm *rm = connection->notification.rm;
>> +
>> +    blocking_notifier_call_chain(&rm->nh, connection->msg_id, 
>> connection->payload);
>> +
>> +    put_gh_rm(rm);
>> +    kfree(connection->payload);
>> +    kfree(connection);
>> +}
>> +
>> +static struct gh_rm_connection *gh_rm_process_notif(struct gh_rm *rm, 
>> void *msg, size_t msg_size)
> 
> I think it might be better if you do some of what the caller
> does here.  I.e., verify the current connection is null (and
> abort if not and make it NULL), then assign it to the new
> connection before you return success.  And return an errno.
> 

Since you and Srini both suggest to do it, I'll cave. :-)

>> +{
>> +    struct gh_rm_connection *connection;
>> +    struct gh_rm_rpc_hdr *hdr = msg;
>> +    int ret;
>> +
>> +    connection = gh_rm_alloc_connection(hdr->msg_id, RM_RPC_TYPE_NOTIF);
>> +    if (IS_ERR(connection)) {
>> +        dev_err(rm->dev, "Failed to alloc connection for 
>> notification: %ld, dropping.\n",
>> +            PTR_ERR(connection));
>> +        return NULL;
>> +    }
>> +
>> +    get_gh_rm(rm);
>> +    connection->notification.rm = rm;
>> +    INIT_WORK(&connection->notification.work, gh_rm_notif_work);
>> +
>> +    ret = gh_rm_init_connection_payload(connection, msg, 
>> sizeof(*hdr), msg_size);
>> +    if (ret) {
>> +        dev_err(rm->dev, "Failed to initialize connection buffer for 
>> notification: %d\n",
>> +            ret);
>> +        kfree(connection);
>> +        return NULL;
>> +    }
>> +
>> +    return connection;
>> +}
>> +
>> +static struct gh_rm_connection *gh_rm_process_rply(struct gh_rm *rm, 
>> void *msg, size_t msg_size)
>> +{
> 
> Here too, make sure there is no active connection and then
> set it within this function if the errno returned is 0.
> 
>> +    struct gh_rm_rpc_reply_hdr *reply_hdr = msg;
>> +    struct gh_rm_connection *connection;
>> +    u16 seq_id = le16_to_cpu(reply_hdr->hdr.seq);
>> +
>> +    mutex_lock(&rm->call_idr_lock);
>> +    connection = idr_find(&rm->call_idr, seq_id);
>> +    mutex_unlock(&rm->call_idr_lock);
>> +
>> +    if (!connection || connection->msg_id != reply_hdr->hdr.msg_id)
>> +        return NULL;
>> +
>> +    if (gh_rm_init_connection_payload(connection, msg, 
>> sizeof(*reply_hdr), msg_size)) {
>> +        dev_err(rm->dev, "Failed to alloc connection buffer for 
>> sequence %d\n", seq_id);
>> +        /* Send connection complete and error the client. */
>> +        connection->reply.ret = -ENOMEM;
>> +        complete(&connection->reply.seq_done);
>> +        return NULL;
>> +    }
>> +
>> +    connection->reply.rm_error = le32_to_cpu(reply_hdr->err_code);
>> +    return connection;
>> +}
>> +
>> +static int gh_rm_process_cont(struct gh_rm *rm, struct 
>> gh_rm_connection *connection,
>> +                void *msg, size_t msg_size)
> 
> Similar comment here.  Have this function verify there is
> a non-null active connection.  Then process the message
> and abort if there's an error (and null the active connection
> pointer).
> 
>> +{
>> +    struct gh_rm_rpc_hdr *hdr = msg;
>> +    size_t payload_size = msg_size - sizeof(*hdr);
>> +
>> +    /*
>> +     * hdr->fragments and hdr->msg_id preserves the value from first 
>> reply
>> +     * or notif message. To detect mishandling, check it's still intact.
>> +     */
>> +    if (connection->msg_id != hdr->msg_id ||
>> +        connection->num_fragments != FIELD_GET(RM_RPC_FRAGMENTS_MASK, 
>> hdr->type))
>> +        return -EINVAL;
> 
> Maybe -EBADMSG?
> 
>> +
>> +    memcpy(connection->payload + connection->size, msg + 
>> sizeof(*hdr), payload_size);
>> +    connection->size += payload_size;
>> +    connection->fragments_received++;
>> +    return 0;
>> +}
>> +
>> +static void gh_rm_abort_connection(struct gh_rm_connection *connection)
>> +{
>> +    switch (connection->type) {
>> +    case RM_RPC_TYPE_REPLY:
>> +        connection->reply.ret = -EIO;
>> +        complete(&connection->reply.seq_done);
>> +        break;
>> +    case RM_RPC_TYPE_NOTIF:
>> +        fallthrough;
>> +    default:
>> +        kfree(connection->payload);
>> +        kfree(connection);
>> +    }
>> +}
>> +
>> +static bool gh_rm_complete_connection(struct gh_rm *rm, struct 
>> gh_rm_connection *connection)
> 
> The only caller of this function passes rm->active_rx_connection
> as the second argument.  It is available to you here, so you
> can get rid of that argument.
> 
>> +{
>> +    if (!connection || connection->fragments_received != 
>> connection->num_fragments)
>> +        return false;
>> +
>> +    switch (connection->type) {
>> +    case RM_RPC_TYPE_REPLY:
>> +        complete(&connection->reply.seq_done);
>> +        break;
>> +    case RM_RPC_TYPE_NOTIF:
>> +        schedule_work(&connection->notification.work);
>> +        break;
>> +    default:
>> +        dev_err(rm->dev, "Invalid message type (%d) received\n", 
>> connection->type);
>> +        gh_rm_abort_connection(connection);
>> +        break;
>> +    }
>> +
>> +    return true;
>> +}
>> +
>> +static void gh_rm_msgq_rx_data(struct mbox_client *cl, void *mssg)
>> +{
>> +    struct gh_rm *rm = container_of(cl, struct gh_rm, msgq_client);
>> +    struct gh_msgq_rx_data *rx_data = mssg;
>> +    size_t msg_size = rx_data->length;
>> +    void *msg = rx_data->data;
>> +    struct gh_rm_rpc_hdr *hdr;
>> +
> 
> Is it required that at least one byte (past the header) will
> be received?  I.e., should the "<=" below just be "<"?
> 
>> +    if (msg_size <= sizeof(*hdr) || msg_size > GH_MSGQ_MAX_MSG_SIZE)
>> +        return;
> 
> You previously reported a message here.  These seem like
> errors, which if they occur, maybe should be reported.
> They seem like "never happen" issues, but it's defensive
> to make these checks (which is good).
> 
>> +
>> +    hdr = msg;
>> +    if (hdr->api != RM_RPC_API) {
> 
> If this ever happens, is the hardware failing?  It seems
> like once Gunyah is initialized and you've checked the
> API version once, there should be no need to check it
> repeatedly.

I'd need to check the API version for the first message. On subsequent 
messages, I'd need to check if I already checked. Might as well just 
check the version every time?

<done for all the comments snipped>

>> +
>> +void get_gh_rm(struct gh_rm *rm)
> 
> It is often pretty handy to return the argument in
> functions like this.  It simultaneously takes the
> reference and assigns the pointer the reference
> represents.
> 
> 

I've updated so that gh_rm_get() returns a struct device * (the 
miscdev's device). Is this too unusual?

>> +{
>> +    get_device(rm->dev);
>> +}
>> +EXPORT_SYMBOL_GPL(get_gh_rm);
>> +
>> +void put_gh_rm(struct gh_rm *rm)
>> +{
>> +    put_device(rm->dev);
>> +}
>> +EXPORT_SYMBOL_GPL(put_gh_rm);
>> +
>> +static int gh_msgq_platform_probe_direction(struct platform_device 
>> *pdev,
>> +                    bool tx, int idx, struct gunyah_resource *ghrsc)
>> +{
>> +    struct device_node *node = pdev->dev.of_node;
>> +    int ret;
> 
> I think you should declare idx as a local variable.
> 
>      int idx = tx ? 1 : 0;
> >> +
>> +    ghrsc->type = tx ? GUNYAH_RESOURCE_TYPE_MSGQ_TX : 
>> GUNYAH_RESOURCE_TYPE_MSGQ_RX;
>> +
>> +    ghrsc->irq = platform_get_irq(pdev, idx);
> 
> Do you suppose you could do platform_get_irq_byname(), and then
> specify the names of the IRQs ("rm_tx_irq" and "rm_rx_irq" maybe)?
> 
>> +    if (ghrsc->irq < 0) {
>> +        dev_err(&pdev->dev, "Failed to get irq%d: %d\n", idx, 
>> ghrsc->irq);
> 
> Maybe:    "Failed to get %cX IRQ: %d\n", tx ? 'T' : 'R', ghrsc->irq);
> 
>> +        return ghrsc->irq;
>> +    }
>> +
>> +    ret = of_property_read_u64_index(node, "reg", idx, &ghrsc->capid);
> 
> Is a capability ID a simple (but large) number?
> 
> The *resource manager* (which is a very special VM) has to
> have both TX and RX message queue capability IDs.  Is there
> 'any chance that these specific capability IDs have values
> that are fixed by the design?  Like, 0 and 1?  I don't know
> what they are, but it seems like it *could* be something
> fixed by the design, and if that were the case, there would
> be no need to specify the "reg" property to get the "capid"
> values.
> 

They aren't fixed by the design in a production version of Gunyah.

>> +    if (ret) {
>> +        dev_err(&pdev->dev, "Failed to get capid%d: %d\n", idx, ret);
>> +        return ret;
>> +    }
>> +
>> +    return 0;
>> +}
>> +
>> +static int gh_rm_drv_probe(struct platform_device *pdev)
>> +{
>> +    struct gh_msgq_tx_data *msg;
>> +    struct gh_rm *rm;
>> +    int ret;
>> +
>> +    rm = devm_kzalloc(&pdev->dev, sizeof(*rm), GFP_KERNEL);
>> +    if (!rm)
>> +        return -ENOMEM;
>> +
>> +    platform_set_drvdata(pdev, rm);
>> +    rm->dev = &pdev->dev;
>> +
>> +    mutex_init(&rm->call_idr_lock);
>> +    idr_init(&rm->call_idr);
>> +    rm->cache = kmem_cache_create("gh_rm", struct_size(msg, data, 
>> GH_MSGQ_MAX_MSG_SIZE), 0,
>> +        SLAB_HWCACHE_ALIGN, NULL);
>> +    if (!rm->cache)
>> +        return -ENOMEM;
> 
> If you abstracted the allocation interface for these messages,
> you could actually survive without the slab cache here.  But
> if this fails, maybe you won't get far anyway.
> 
>> +    mutex_init(&rm->send_lock);
>> +    BLOCKING_INIT_NOTIFIER_HEAD(&rm->nh);
>> +
>> +    ret = gh_msgq_platform_probe_direction(pdev, true, 0, 
>> &rm->tx_ghrsc);
>> +    if (ret)
>> +        goto err_cache;
>> +
>> +    ret = gh_msgq_platform_probe_direction(pdev, false, 1, 
>> &rm->rx_ghrsc);
>> +    if (ret)
>> +        goto err_cache;
>> +
>> +    rm->msgq_client.dev = &pdev->dev;
>> +    rm->msgq_client.tx_block = true;
>> +    rm->msgq_client.rx_callback = gh_rm_msgq_rx_data;
>> +    rm->msgq_client.tx_done = gh_rm_msgq_tx_done;
>> +
>> +    return gh_msgq_init(&pdev->dev, &rm->msgq, &rm->msgq_client, 
>> &rm->tx_ghrsc, &rm->rx_ghrsc);
>> +err_cache:
>> +    kmem_cache_destroy(rm->cache);
>> +    return ret;
>> +}
>> +
>> +static int gh_rm_drv_remove(struct platform_device *pdev)
>> +{
>> +    struct gh_rm *rm = platform_get_drvdata(pdev);
>> +
>> +    mbox_free_channel(gh_msgq_chan(&rm->msgq));
>> +    gh_msgq_remove(&rm->msgq);
>> +    kmem_cache_destroy(rm->cache);
>> +
>> +    return 0;
>> +}
>> +
>> +static const struct of_device_id gh_rm_of_match[] = {
>> +    { .compatible = "gunyah-resource-manager" },
>> +    {}
>> +};
>> +MODULE_DEVICE_TABLE(of, gh_rm_of_match);
>> +
>> +static struct platform_driver gh_rm_driver = {
>> +    .probe = gh_rm_drv_probe,
>> +    .remove = gh_rm_drv_remove,
>> +    .driver = {
>> +        .name = "gh_rsc_mgr",
>> +        .of_match_table = gh_rm_of_match,
>> +    },
>> +};
>> +module_platform_driver(gh_rm_driver);
>> +
>> +MODULE_LICENSE("GPL");
>> +MODULE_DESCRIPTION("Gunyah Resource Manager Driver");
>> diff --git a/drivers/virt/gunyah/rsc_mgr.h 
>> b/drivers/virt/gunyah/rsc_mgr.h
>> new file mode 100644
>> index 000000000000..d4e799a7526f
>> --- /dev/null
>> +++ b/drivers/virt/gunyah/rsc_mgr.h
>> @@ -0,0 +1,77 @@
>> +/* SPDX-License-Identifier: GPL-2.0-only */
>> +/*
>> + * Copyright (c) 2022-2023 Qualcomm Innovation Center, Inc. All 
>> rights reserved.
>> + */
>> +#ifndef __GH_RSC_MGR_PRIV_H
>> +#define __GH_RSC_MGR_PRIV_H
>> +
>> +#include <linux/gunyah.h>
>> +#include <linux/gunyah_rsc_mgr.h>
>> +#include <linux/types.h>
>> +
>> +/* RM Error codes */
>> +enum gh_rm_error {
>> +    GH_RM_ERROR_OK            = 0x0,
>> +    GH_RM_ERROR_UNIMPLEMENTED    = 0xFFFFFFFF,
>> +    GH_RM_ERROR_NOMEM        = 0x1,
>> +    GH_RM_ERROR_NORESOURCE        = 0x2,
>> +    GH_RM_ERROR_DENIED        = 0x3,
>> +    GH_RM_ERROR_INVALID        = 0x4,
>> +    GH_RM_ERROR_BUSY        = 0x5,
>> +    GH_RM_ERROR_ARGUMENT_INVALID    = 0x6,
>> +    GH_RM_ERROR_HANDLE_INVALID    = 0x7,
>> +    GH_RM_ERROR_VALIDATE_FAILED    = 0x8,
>> +    GH_RM_ERROR_MAP_FAILED        = 0x9,
>> +    GH_RM_ERROR_MEM_INVALID        = 0xA,
>> +    GH_RM_ERROR_MEM_INUSE        = 0xB,
>> +    GH_RM_ERROR_MEM_RELEASED    = 0xC,
>> +    GH_RM_ERROR_VMID_INVALID    = 0xD,
>> +    GH_RM_ERROR_LOOKUP_FAILED    = 0xE,
>> +    GH_RM_ERROR_IRQ_INVALID        = 0xF,
>> +    GH_RM_ERROR_IRQ_INUSE        = 0x10,
>> +    GH_RM_ERROR_IRQ_RELEASED    = 0x11,
>> +};
>> +
>> +/**
>> + * gh_rm_remap_error() - Remap Gunyah resource manager errors into a 
>> Linux error code
>> + * @gh_error: "Standard" return value from Gunyah resource manager
>> + */
>> +static inline int gh_rm_remap_error(enum gh_rm_error rm_error)
>> +{
>> +    switch (rm_error) {
>> +    case GH_RM_ERROR_OK:
>> +        return 0;
>> +    case GH_RM_ERROR_UNIMPLEMENTED:
>> +        return -EOPNOTSUPP;
>> +    case GH_RM_ERROR_NOMEM:
>> +        return -ENOMEM;
>> +    case GH_RM_ERROR_NORESOURCE:
>> +        return -ENODEV;
>> +    case GH_RM_ERROR_DENIED:
>> +        return -EPERM;
>> +    case GH_RM_ERROR_BUSY:
>> +        return -EBUSY;
>> +    case GH_RM_ERROR_INVALID:
>> +    case GH_RM_ERROR_ARGUMENT_INVALID:
>> +    case GH_RM_ERROR_HANDLE_INVALID:
>> +    case GH_RM_ERROR_VALIDATE_FAILED:
>> +    case GH_RM_ERROR_MAP_FAILED:
>> +    case GH_RM_ERROR_MEM_INVALID:
>> +    case GH_RM_ERROR_MEM_INUSE:
>> +    case GH_RM_ERROR_MEM_RELEASED:
>> +    case GH_RM_ERROR_VMID_INVALID:
>> +    case GH_RM_ERROR_LOOKUP_FAILED:
>> +    case GH_RM_ERROR_IRQ_INVALID:
>> +    case GH_RM_ERROR_IRQ_INUSE:
>> +    case GH_RM_ERROR_IRQ_RELEASED:
>> +        return -EINVAL;
>> +    default:
>> +        return -EBADMSG;
>> +    }
>> +}
>> +
>> +struct gh_rm;
> 
> This might just be my preference, but I like to see declarations
> like the one above grouped at the top of the file, under includes.
> 
>> +int gh_rm_call(struct gh_rm *rsc_mgr, u32 message_id, void *req_buff, 
>> size_t req_buff_size,
>> +        void **resp_buf, size_t *resp_buff_size);
>> +
>> +#endif
>> diff --git a/include/linux/gunyah_rsc_mgr.h 
>> b/include/linux/gunyah_rsc_mgr.h
>> new file mode 100644
>> index 000000000000..c992b3188c8d
>> --- /dev/null
>> +++ b/include/linux/gunyah_rsc_mgr.h
>> @@ -0,0 +1,24 @@
>> +/* SPDX-License-Identifier: GPL-2.0-only */
>> +/*
>> + * Copyright (c) 2022-2023 Qualcomm Innovation Center, Inc. All 
>> rights reserved.
>> + */
>> +
>> +#ifndef _GUNYAH_RSC_MGR_H
>> +#define _GUNYAH_RSC_MGR_H
>> +
>> +#include <linux/list.h>
>> +#include <linux/notifier.h>
>> +#include <linux/gunyah.h>
>> +
>> +#define GH_VMID_INVAL    U16_MAX
>> +
>> +/* Gunyah recognizes VMID0 as an alias to the current VM's ID */
>> +#define GH_VMID_SELF            0
> 
> I haven't really checked very well, bur you should *use this*
> definition where a VMID is being examined. I.e., if you're
> going to define this, then never just compare a VMID against 0.
> 

I realize now the only place I *could* use GH_VMID_SELF is the one 
exception to usage of VMID -- in gh_rm_vmid_alloc. There, vmid of 0 
means "use dynamic allocation". Since there aren't any users of the 
GH_VMID_SELF, I'll drop it.

Thanks,
Elliot

>                      -Alex
> 
>> +
>> +struct gh_rm;
>> +int gh_rm_notifier_register(struct gh_rm *rm, struct notifier_block 
>> *nb);
>> +int gh_rm_notifier_unregister(struct gh_rm *rm, struct notifier_block 
>> *nb);
>> +void get_gh_rm(struct gh_rm *rm);
>> +void put_gh_rm(struct gh_rm *rm);
>> +
>> +#endif
>
Elliot Berman Feb. 28, 2023, 10:49 p.m. UTC | #7
On 2/27/2023 4:52 PM, Alex Elder wrote:
> On 2/23/23 5:13 PM, Elliot Berman wrote:
>>> TBH, gunyah.c should be merged as part of resource manager, and check 
>>> if uuids and features in probe before proceeding further.
>>>
>>
>>
>> Ah -- gunyah_rsc_mgr.ko has symbol dependency on gunyah-msgq.ko. 
>> gunyah-msgq.ko has symbol dependency on gunyah.ko. gunyah.ko doesn't 
>> have any probe and does all its work on module_init.
>>
>> In order to merge gunyah.c with resource manager, I would need to 
>> incorporate message queue mailbox into resource manager. IMO, this 
>> rapidly moves towards a mega-module which was discouraged previously.
> 
> I missed this discussion; why was it discouraged?
> 
> I can think of some reasons why I guess.  But I don't see what
> problem comes from linking together a "mega module" that's made
> up of well-isolated source files that expose minimal APIs to
> one another.  All inter-dependent modules will required at once
> anyway; I don't understand the benefit of implementing them
> separately.  Can you explain, or provide some context?  Thanks.

I came from some earlier comments from Dmitry:

https://lore.kernel.org/all/250945d2-3940-9830-63e5-beec5f44010b@linaro.org/

Earlier comments from Dmitry were about having bus and drivers in same 
module. I think same comment applies with mailbox built into the 
gunyah.ko (message queue is the provider and rsr_mgr is the consumer).

Thanks,
Elliot
diff mbox series

Patch

diff --git a/drivers/virt/gunyah/Makefile b/drivers/virt/gunyah/Makefile
index 34f32110faf9..cc864ff5abbb 100644
--- a/drivers/virt/gunyah/Makefile
+++ b/drivers/virt/gunyah/Makefile
@@ -1,3 +1,6 @@ 
 # SPDX-License-Identifier: GPL-2.0
 
 obj-$(CONFIG_GUNYAH) += gunyah.o
+
+gunyah_rsc_mgr-y += rsc_mgr.o
+obj-$(CONFIG_GUNYAH) += gunyah_rsc_mgr.o
diff --git a/drivers/virt/gunyah/rsc_mgr.c b/drivers/virt/gunyah/rsc_mgr.c
new file mode 100644
index 000000000000..2a47139873a8
--- /dev/null
+++ b/drivers/virt/gunyah/rsc_mgr.c
@@ -0,0 +1,604 @@ 
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2022-2023 Qualcomm Innovation Center, Inc. All rights reserved.
+ */
+
+#include <linux/of.h>
+#include <linux/slab.h>
+#include <linux/mutex.h>
+#include <linux/sched.h>
+#include <linux/gunyah.h>
+#include <linux/module.h>
+#include <linux/of_irq.h>
+#include <linux/kthread.h>
+#include <linux/notifier.h>
+#include <linux/workqueue.h>
+#include <linux/completion.h>
+#include <linux/gunyah_rsc_mgr.h>
+#include <linux/platform_device.h>
+
+#include "rsc_mgr.h"
+
+#define RM_RPC_API_VERSION_MASK		GENMASK(3, 0)
+#define RM_RPC_HEADER_WORDS_MASK	GENMASK(7, 4)
+#define RM_RPC_API_VERSION		FIELD_PREP(RM_RPC_API_VERSION_MASK, 1)
+#define RM_RPC_HEADER_WORDS		FIELD_PREP(RM_RPC_HEADER_WORDS_MASK, \
+						(sizeof(struct gh_rm_rpc_hdr) / sizeof(u32)))
+#define RM_RPC_API			(RM_RPC_API_VERSION | RM_RPC_HEADER_WORDS)
+
+#define RM_RPC_TYPE_CONTINUATION	0x0
+#define RM_RPC_TYPE_REQUEST		0x1
+#define RM_RPC_TYPE_REPLY		0x2
+#define RM_RPC_TYPE_NOTIF		0x3
+#define RM_RPC_TYPE_MASK		GENMASK(1, 0)
+
+#define GH_RM_MAX_NUM_FRAGMENTS		62
+#define RM_RPC_FRAGMENTS_MASK		GENMASK(7, 2)
+
+struct gh_rm_rpc_hdr {
+	u8 api;
+	u8 type;
+	__le16 seq;
+	__le32 msg_id;
+} __packed;
+
+struct gh_rm_rpc_reply_hdr {
+	struct gh_rm_rpc_hdr hdr;
+	__le32 err_code; /* GH_RM_ERROR_* */
+} __packed;
+
+#define GH_RM_MAX_MSG_SIZE	(GH_MSGQ_MAX_MSG_SIZE - sizeof(struct gh_rm_rpc_hdr))
+
+/**
+ * struct gh_rm_connection - Represents a complete message from resource manager
+ * @payload: Combined payload of all the fragments (msg headers stripped off).
+ * @size: Size of the payload received so far.
+ * @msg_id: Message ID from the header.
+ * @num_fragments: total number of fragments expected to be received.
+ * @fragments_received: fragments received so far.
+ * @reply: Fields used for request/reply sequences
+ * @ret: Linux return code, set in case there was an error processing connection
+ * @type: RM_RPC_TYPE_REPLY or RM_RPC_TYPE_NOTIF.
+ * @rm_error: For request/reply sequences with standard replies.
+ * @seq: Sequence ID for the main message.
+ * @seq_done: Signals caller that the RM reply has been received
+ * @notification: Fields used for notifiations
+ * @work: Triggered when all fragments of a notification received
+ */
+struct gh_rm_connection {
+	void *payload;
+	size_t size;
+	__le32 msg_id;
+	u8 type;
+
+	u8 num_fragments;
+	u8 fragments_received;
+
+	union {
+		struct {
+			int ret;
+			u16 seq;
+			enum gh_rm_error rm_error;
+			struct completion seq_done;
+		} reply;
+
+		struct {
+			struct gh_rm *rm;
+			struct work_struct work;
+		} notification;
+	};
+};
+
+struct gh_rm {
+	struct device *dev;
+	struct gunyah_resource tx_ghrsc, rx_ghrsc;
+	struct gh_msgq msgq;
+	struct mbox_client msgq_client;
+	struct gh_rm_connection *active_rx_connection;
+	int last_tx_ret;
+
+	struct idr call_idr;
+	struct mutex call_idr_lock;
+
+	struct kmem_cache *cache;
+	struct mutex send_lock;
+	struct blocking_notifier_head nh;
+};
+
+static struct gh_rm_connection *gh_rm_alloc_connection(__le32 msg_id, u8 type)
+{
+	struct gh_rm_connection *connection;
+
+	connection = kzalloc(sizeof(*connection), GFP_KERNEL);
+	if (!connection)
+		return ERR_PTR(-ENOMEM);
+
+	connection->type = type;
+	connection->msg_id = msg_id;
+
+	return connection;
+}
+
+static int gh_rm_init_connection_payload(struct gh_rm_connection *connection, void *msg,
+					size_t hdr_size, size_t msg_size)
+{
+	size_t max_buf_size, payload_size;
+	struct gh_rm_rpc_hdr *hdr = msg;
+
+	if (hdr_size > msg_size)
+		return -EINVAL;
+
+	payload_size = msg_size - hdr_size;
+
+	connection->num_fragments = FIELD_GET(RM_RPC_FRAGMENTS_MASK, hdr->type);
+	connection->fragments_received = 0;
+
+	/* There's not going to be any payload, no need to allocate buffer. */
+	if (!payload_size && !connection->num_fragments)
+		return 0;
+
+	if (connection->num_fragments > GH_RM_MAX_NUM_FRAGMENTS)
+		return -EINVAL;
+
+	max_buf_size = payload_size + (connection->num_fragments * GH_RM_MAX_MSG_SIZE);
+
+	connection->payload = kzalloc(max_buf_size, GFP_KERNEL);
+	if (!connection->payload)
+		return -ENOMEM;
+
+	memcpy(connection->payload, msg + hdr_size, payload_size);
+	connection->size = payload_size;
+	return 0;
+}
+
+static void gh_rm_notif_work(struct work_struct *work)
+{
+	struct gh_rm_connection *connection = container_of(work, struct gh_rm_connection,
+								notification.work);
+	struct gh_rm *rm = connection->notification.rm;
+
+	blocking_notifier_call_chain(&rm->nh, connection->msg_id, connection->payload);
+
+	put_gh_rm(rm);
+	kfree(connection->payload);
+	kfree(connection);
+}
+
+static struct gh_rm_connection *gh_rm_process_notif(struct gh_rm *rm, void *msg, size_t msg_size)
+{
+	struct gh_rm_connection *connection;
+	struct gh_rm_rpc_hdr *hdr = msg;
+	int ret;
+
+	connection = gh_rm_alloc_connection(hdr->msg_id, RM_RPC_TYPE_NOTIF);
+	if (IS_ERR(connection)) {
+		dev_err(rm->dev, "Failed to alloc connection for notification: %ld, dropping.\n",
+			PTR_ERR(connection));
+		return NULL;
+	}
+
+	get_gh_rm(rm);
+	connection->notification.rm = rm;
+	INIT_WORK(&connection->notification.work, gh_rm_notif_work);
+
+	ret = gh_rm_init_connection_payload(connection, msg, sizeof(*hdr), msg_size);
+	if (ret) {
+		dev_err(rm->dev, "Failed to initialize connection buffer for notification: %d\n",
+			ret);
+		kfree(connection);
+		return NULL;
+	}
+
+	return connection;
+}
+
+static struct gh_rm_connection *gh_rm_process_rply(struct gh_rm *rm, void *msg, size_t msg_size)
+{
+	struct gh_rm_rpc_reply_hdr *reply_hdr = msg;
+	struct gh_rm_connection *connection;
+	u16 seq_id = le16_to_cpu(reply_hdr->hdr.seq);
+
+	mutex_lock(&rm->call_idr_lock);
+	connection = idr_find(&rm->call_idr, seq_id);
+	mutex_unlock(&rm->call_idr_lock);
+
+	if (!connection || connection->msg_id != reply_hdr->hdr.msg_id)
+		return NULL;
+
+	if (gh_rm_init_connection_payload(connection, msg, sizeof(*reply_hdr), msg_size)) {
+		dev_err(rm->dev, "Failed to alloc connection buffer for sequence %d\n", seq_id);
+		/* Send connection complete and error the client. */
+		connection->reply.ret = -ENOMEM;
+		complete(&connection->reply.seq_done);
+		return NULL;
+	}
+
+	connection->reply.rm_error = le32_to_cpu(reply_hdr->err_code);
+	return connection;
+}
+
+static int gh_rm_process_cont(struct gh_rm *rm, struct gh_rm_connection *connection,
+				void *msg, size_t msg_size)
+{
+	struct gh_rm_rpc_hdr *hdr = msg;
+	size_t payload_size = msg_size - sizeof(*hdr);
+
+	/*
+	 * hdr->fragments and hdr->msg_id preserves the value from first reply
+	 * or notif message. To detect mishandling, check it's still intact.
+	 */
+	if (connection->msg_id != hdr->msg_id ||
+		connection->num_fragments != FIELD_GET(RM_RPC_FRAGMENTS_MASK, hdr->type))
+		return -EINVAL;
+
+	memcpy(connection->payload + connection->size, msg + sizeof(*hdr), payload_size);
+	connection->size += payload_size;
+	connection->fragments_received++;
+	return 0;
+}
+
+static void gh_rm_abort_connection(struct gh_rm_connection *connection)
+{
+	switch (connection->type) {
+	case RM_RPC_TYPE_REPLY:
+		connection->reply.ret = -EIO;
+		complete(&connection->reply.seq_done);
+		break;
+	case RM_RPC_TYPE_NOTIF:
+		fallthrough;
+	default:
+		kfree(connection->payload);
+		kfree(connection);
+	}
+}
+
+static bool gh_rm_complete_connection(struct gh_rm *rm, struct gh_rm_connection *connection)
+{
+	if (!connection || connection->fragments_received != connection->num_fragments)
+		return false;
+
+	switch (connection->type) {
+	case RM_RPC_TYPE_REPLY:
+		complete(&connection->reply.seq_done);
+		break;
+	case RM_RPC_TYPE_NOTIF:
+		schedule_work(&connection->notification.work);
+		break;
+	default:
+		dev_err(rm->dev, "Invalid message type (%d) received\n", connection->type);
+		gh_rm_abort_connection(connection);
+		break;
+	}
+
+	return true;
+}
+
+static void gh_rm_msgq_rx_data(struct mbox_client *cl, void *mssg)
+{
+	struct gh_rm *rm = container_of(cl, struct gh_rm, msgq_client);
+	struct gh_msgq_rx_data *rx_data = mssg;
+	size_t msg_size = rx_data->length;
+	void *msg = rx_data->data;
+	struct gh_rm_rpc_hdr *hdr;
+
+	if (msg_size <= sizeof(*hdr) || msg_size > GH_MSGQ_MAX_MSG_SIZE)
+		return;
+
+	hdr = msg;
+	if (hdr->api != RM_RPC_API) {
+		dev_err(rm->dev, "Unknown RM RPC API version: %x\n", hdr->api);
+		return;
+	}
+
+	switch (FIELD_GET(RM_RPC_TYPE_MASK, hdr->type)) {
+	case RM_RPC_TYPE_NOTIF:
+		rm->active_rx_connection = gh_rm_process_notif(rm, msg, msg_size);
+		break;
+	case RM_RPC_TYPE_REPLY:
+		rm->active_rx_connection = gh_rm_process_rply(rm, msg, msg_size);
+		break;
+	case RM_RPC_TYPE_CONTINUATION:
+		if (gh_rm_process_cont(rm, rm->active_rx_connection, msg, msg_size)) {
+			gh_rm_abort_connection(rm->active_rx_connection);
+			rm->active_rx_connection = NULL;
+		}
+		break;
+	default:
+		dev_err(rm->dev, "Invalid message type (%lu) received\n",
+			FIELD_GET(RM_RPC_TYPE_MASK, hdr->type));
+		return;
+	}
+
+	if (gh_rm_complete_connection(rm, rm->active_rx_connection))
+		rm->active_rx_connection = NULL;
+}
+
+static void gh_rm_msgq_tx_done(struct mbox_client *cl, void *mssg, int r)
+{
+	struct gh_rm *rm = container_of(cl, struct gh_rm, msgq_client);
+
+	kmem_cache_free(rm->cache, mssg);
+	rm->last_tx_ret = r;
+}
+
+static int gh_rm_send_request(struct gh_rm *rm, u32 message_id,
+			      const void *req_buff, size_t req_buff_size,
+			      struct gh_rm_connection *connection)
+{
+	u8 msg_type = FIELD_PREP(RM_RPC_TYPE_MASK, RM_RPC_TYPE_REQUEST);
+	size_t buff_size_remaining = req_buff_size;
+	const void *req_buff_curr = req_buff;
+	struct gh_msgq_tx_data *msg;
+	struct gh_rm_rpc_hdr *hdr;
+	u32 cont_fragments = 0;
+	size_t payload_size;
+	void *payload;
+	int ret;
+
+	if (req_buff_size)
+		cont_fragments = (req_buff_size - 1) / GH_RM_MAX_MSG_SIZE;
+
+	if (req_buff_size > GH_RM_MAX_NUM_FRAGMENTS * GH_RM_MAX_MSG_SIZE) {
+		pr_warn("Limit exceeded for the number of fragments: %u\n", cont_fragments);
+		dump_stack();
+		return -E2BIG;
+	}
+
+	ret = mutex_lock_interruptible(&rm->send_lock);
+	if (ret)
+		return ret;
+
+	/* Consider also the 'request' packet for the loop count */
+	do {
+		msg = kmem_cache_zalloc(rm->cache, GFP_KERNEL);
+		if (!msg) {
+			ret = -ENOMEM;
+			goto out;
+		}
+
+		/* Fill header */
+		hdr = (struct gh_rm_rpc_hdr *)msg->data;
+		hdr->api = RM_RPC_API;
+		hdr->type = msg_type | FIELD_PREP(RM_RPC_FRAGMENTS_MASK, cont_fragments);
+		hdr->seq = cpu_to_le16(connection->reply.seq);
+		hdr->msg_id = cpu_to_le32(message_id);
+
+		/* Copy payload */
+		payload = hdr + 1;
+		payload_size = min(buff_size_remaining, GH_RM_MAX_MSG_SIZE);
+		memcpy(payload, req_buff_curr, payload_size);
+		req_buff_curr += payload_size;
+		buff_size_remaining -= payload_size;
+
+		/* Force the last fragment to immediately alert the receiver */
+		msg->push = !buff_size_remaining;
+		msg->length = sizeof(*hdr) + payload_size;
+
+		ret = mbox_send_message(gh_msgq_chan(&rm->msgq), msg);
+		if (ret < 0) {
+			kmem_cache_free(rm->cache, msg);
+			break;
+		}
+
+		if (rm->last_tx_ret) {
+			ret = rm->last_tx_ret;
+			break;
+		}
+
+		msg_type = FIELD_PREP(RM_RPC_TYPE_MASK, RM_RPC_TYPE_CONTINUATION);
+	} while (buff_size_remaining);
+
+out:
+	mutex_unlock(&rm->send_lock);
+	return ret < 0 ? ret : 0;
+}
+
+/**
+ * gh_rm_call: Achieve request-response type communication with RPC
+ * @rm: Pointer to Gunyah resource manager internal data
+ * @message_id: The RM RPC message-id
+ * @req_buff: Request buffer that contains the payload
+ * @req_buff_size: Total size of the payload
+ * @resp_buf: Pointer to a response buffer
+ * @resp_buff_size: Size of the response buffer
+ *
+ * Make a request to the RM-VM and wait for reply back. For a successful
+ * response, the function returns the payload. The size of the payload is set in
+ * resp_buff_size. The resp_buf should be freed by the caller.
+ *
+ * req_buff should be not NULL for req_buff_size >0. If req_buff_size == 0,
+ * req_buff *can* be NULL and no additional payload is sent.
+ *
+ * Context: Process context. Will sleep waiting for reply.
+ * Return: 0 on success. <0 if error.
+ */
+int gh_rm_call(struct gh_rm *rm, u32 message_id, void *req_buff, size_t req_buff_size,
+		void **resp_buf, size_t *resp_buff_size)
+{
+	struct gh_rm_connection *connection;
+	int ret;
+
+	/* message_id 0 is reserved. req_buff_size implies req_buf is not NULL */
+	if (!message_id || (!req_buff && req_buff_size) || !rm)
+		return -EINVAL;
+
+	connection = gh_rm_alloc_connection(cpu_to_le32(message_id), RM_RPC_TYPE_REPLY);
+	if (IS_ERR(connection))
+		return PTR_ERR(connection);
+
+	init_completion(&connection->reply.seq_done);
+
+	/* Allocate a new seq number for this connection */
+	mutex_lock(&rm->call_idr_lock);
+	ret = idr_alloc_cyclic(&rm->call_idr, connection, 0, U16_MAX,
+						GFP_KERNEL);
+	mutex_unlock(&rm->call_idr_lock);
+	if (ret < 0)
+		goto out;
+	connection->reply.seq = ret;
+
+	/* Send the request to the Resource Manager */
+	ret = gh_rm_send_request(rm, message_id, req_buff, req_buff_size, connection);
+	if (ret < 0)
+		goto out;
+
+	/* Wait for response */
+	ret = wait_for_completion_interruptible(&connection->reply.seq_done);
+	if (ret)
+		goto out;
+
+	/* Check for internal (kernel) error waiting for the response */
+	if (connection->reply.ret) {
+		ret = connection->reply.ret;
+		if (ret != -ENOMEM)
+			kfree(connection->payload);
+		goto out;
+	}
+
+	/* Got a response, did resource manager give us an error? */
+	if (connection->reply.rm_error != GH_RM_ERROR_OK) {
+		pr_warn("RM rejected message %08x. Error: %d\n", message_id,
+			connection->reply.rm_error);
+		dump_stack();
+		ret = gh_rm_remap_error(connection->reply.rm_error);
+		kfree(connection->payload);
+		goto out;
+	}
+
+	/* Everything looks good, return the payload */
+	*resp_buff_size = connection->size;
+	if (connection->size)
+		*resp_buf = connection->payload;
+	else {
+		/* kfree in case RM sent us multiple fragments but never any data in
+		 * those fragments. We would've allocated memory for it, but connection->size == 0
+		 */
+		kfree(connection->payload);
+	}
+
+out:
+	mutex_lock(&rm->call_idr_lock);
+	idr_remove(&rm->call_idr, connection->reply.seq);
+	mutex_unlock(&rm->call_idr_lock);
+	kfree(connection);
+	return ret;
+}
+
+
+int gh_rm_notifier_register(struct gh_rm *rm, struct notifier_block *nb)
+{
+	return blocking_notifier_chain_register(&rm->nh, nb);
+}
+EXPORT_SYMBOL_GPL(gh_rm_notifier_register);
+
+int gh_rm_notifier_unregister(struct gh_rm *rm, struct notifier_block *nb)
+{
+	return blocking_notifier_chain_unregister(&rm->nh, nb);
+}
+EXPORT_SYMBOL_GPL(gh_rm_notifier_unregister);
+
+void get_gh_rm(struct gh_rm *rm)
+{
+	get_device(rm->dev);
+}
+EXPORT_SYMBOL_GPL(get_gh_rm);
+
+void put_gh_rm(struct gh_rm *rm)
+{
+	put_device(rm->dev);
+}
+EXPORT_SYMBOL_GPL(put_gh_rm);
+
+static int gh_msgq_platform_probe_direction(struct platform_device *pdev,
+					bool tx, int idx, struct gunyah_resource *ghrsc)
+{
+	struct device_node *node = pdev->dev.of_node;
+	int ret;
+
+	ghrsc->type = tx ? GUNYAH_RESOURCE_TYPE_MSGQ_TX : GUNYAH_RESOURCE_TYPE_MSGQ_RX;
+
+	ghrsc->irq = platform_get_irq(pdev, idx);
+	if (ghrsc->irq < 0) {
+		dev_err(&pdev->dev, "Failed to get irq%d: %d\n", idx, ghrsc->irq);
+		return ghrsc->irq;
+	}
+
+	ret = of_property_read_u64_index(node, "reg", idx, &ghrsc->capid);
+	if (ret) {
+		dev_err(&pdev->dev, "Failed to get capid%d: %d\n", idx, ret);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int gh_rm_drv_probe(struct platform_device *pdev)
+{
+	struct gh_msgq_tx_data *msg;
+	struct gh_rm *rm;
+	int ret;
+
+	rm = devm_kzalloc(&pdev->dev, sizeof(*rm), GFP_KERNEL);
+	if (!rm)
+		return -ENOMEM;
+
+	platform_set_drvdata(pdev, rm);
+	rm->dev = &pdev->dev;
+
+	mutex_init(&rm->call_idr_lock);
+	idr_init(&rm->call_idr);
+	rm->cache = kmem_cache_create("gh_rm", struct_size(msg, data, GH_MSGQ_MAX_MSG_SIZE), 0,
+		SLAB_HWCACHE_ALIGN, NULL);
+	if (!rm->cache)
+		return -ENOMEM;
+	mutex_init(&rm->send_lock);
+	BLOCKING_INIT_NOTIFIER_HEAD(&rm->nh);
+
+	ret = gh_msgq_platform_probe_direction(pdev, true, 0, &rm->tx_ghrsc);
+	if (ret)
+		goto err_cache;
+
+	ret = gh_msgq_platform_probe_direction(pdev, false, 1, &rm->rx_ghrsc);
+	if (ret)
+		goto err_cache;
+
+	rm->msgq_client.dev = &pdev->dev;
+	rm->msgq_client.tx_block = true;
+	rm->msgq_client.rx_callback = gh_rm_msgq_rx_data;
+	rm->msgq_client.tx_done = gh_rm_msgq_tx_done;
+
+	return gh_msgq_init(&pdev->dev, &rm->msgq, &rm->msgq_client, &rm->tx_ghrsc, &rm->rx_ghrsc);
+err_cache:
+	kmem_cache_destroy(rm->cache);
+	return ret;
+}
+
+static int gh_rm_drv_remove(struct platform_device *pdev)
+{
+	struct gh_rm *rm = platform_get_drvdata(pdev);
+
+	mbox_free_channel(gh_msgq_chan(&rm->msgq));
+	gh_msgq_remove(&rm->msgq);
+	kmem_cache_destroy(rm->cache);
+
+	return 0;
+}
+
+static const struct of_device_id gh_rm_of_match[] = {
+	{ .compatible = "gunyah-resource-manager" },
+	{}
+};
+MODULE_DEVICE_TABLE(of, gh_rm_of_match);
+
+static struct platform_driver gh_rm_driver = {
+	.probe = gh_rm_drv_probe,
+	.remove = gh_rm_drv_remove,
+	.driver = {
+		.name = "gh_rsc_mgr",
+		.of_match_table = gh_rm_of_match,
+	},
+};
+module_platform_driver(gh_rm_driver);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("Gunyah Resource Manager Driver");
diff --git a/drivers/virt/gunyah/rsc_mgr.h b/drivers/virt/gunyah/rsc_mgr.h
new file mode 100644
index 000000000000..d4e799a7526f
--- /dev/null
+++ b/drivers/virt/gunyah/rsc_mgr.h
@@ -0,0 +1,77 @@ 
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (c) 2022-2023 Qualcomm Innovation Center, Inc. All rights reserved.
+ */
+#ifndef __GH_RSC_MGR_PRIV_H
+#define __GH_RSC_MGR_PRIV_H
+
+#include <linux/gunyah.h>
+#include <linux/gunyah_rsc_mgr.h>
+#include <linux/types.h>
+
+/* RM Error codes */
+enum gh_rm_error {
+	GH_RM_ERROR_OK			= 0x0,
+	GH_RM_ERROR_UNIMPLEMENTED	= 0xFFFFFFFF,
+	GH_RM_ERROR_NOMEM		= 0x1,
+	GH_RM_ERROR_NORESOURCE		= 0x2,
+	GH_RM_ERROR_DENIED		= 0x3,
+	GH_RM_ERROR_INVALID		= 0x4,
+	GH_RM_ERROR_BUSY		= 0x5,
+	GH_RM_ERROR_ARGUMENT_INVALID	= 0x6,
+	GH_RM_ERROR_HANDLE_INVALID	= 0x7,
+	GH_RM_ERROR_VALIDATE_FAILED	= 0x8,
+	GH_RM_ERROR_MAP_FAILED		= 0x9,
+	GH_RM_ERROR_MEM_INVALID		= 0xA,
+	GH_RM_ERROR_MEM_INUSE		= 0xB,
+	GH_RM_ERROR_MEM_RELEASED	= 0xC,
+	GH_RM_ERROR_VMID_INVALID	= 0xD,
+	GH_RM_ERROR_LOOKUP_FAILED	= 0xE,
+	GH_RM_ERROR_IRQ_INVALID		= 0xF,
+	GH_RM_ERROR_IRQ_INUSE		= 0x10,
+	GH_RM_ERROR_IRQ_RELEASED	= 0x11,
+};
+
+/**
+ * gh_rm_remap_error() - Remap Gunyah resource manager errors into a Linux error code
+ * @gh_error: "Standard" return value from Gunyah resource manager
+ */
+static inline int gh_rm_remap_error(enum gh_rm_error rm_error)
+{
+	switch (rm_error) {
+	case GH_RM_ERROR_OK:
+		return 0;
+	case GH_RM_ERROR_UNIMPLEMENTED:
+		return -EOPNOTSUPP;
+	case GH_RM_ERROR_NOMEM:
+		return -ENOMEM;
+	case GH_RM_ERROR_NORESOURCE:
+		return -ENODEV;
+	case GH_RM_ERROR_DENIED:
+		return -EPERM;
+	case GH_RM_ERROR_BUSY:
+		return -EBUSY;
+	case GH_RM_ERROR_INVALID:
+	case GH_RM_ERROR_ARGUMENT_INVALID:
+	case GH_RM_ERROR_HANDLE_INVALID:
+	case GH_RM_ERROR_VALIDATE_FAILED:
+	case GH_RM_ERROR_MAP_FAILED:
+	case GH_RM_ERROR_MEM_INVALID:
+	case GH_RM_ERROR_MEM_INUSE:
+	case GH_RM_ERROR_MEM_RELEASED:
+	case GH_RM_ERROR_VMID_INVALID:
+	case GH_RM_ERROR_LOOKUP_FAILED:
+	case GH_RM_ERROR_IRQ_INVALID:
+	case GH_RM_ERROR_IRQ_INUSE:
+	case GH_RM_ERROR_IRQ_RELEASED:
+		return -EINVAL;
+	default:
+		return -EBADMSG;
+	}
+}
+
+struct gh_rm;
+int gh_rm_call(struct gh_rm *rsc_mgr, u32 message_id, void *req_buff, size_t req_buff_size,
+		void **resp_buf, size_t *resp_buff_size);
+
+#endif
diff --git a/include/linux/gunyah_rsc_mgr.h b/include/linux/gunyah_rsc_mgr.h
new file mode 100644
index 000000000000..c992b3188c8d
--- /dev/null
+++ b/include/linux/gunyah_rsc_mgr.h
@@ -0,0 +1,24 @@ 
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (c) 2022-2023 Qualcomm Innovation Center, Inc. All rights reserved.
+ */
+
+#ifndef _GUNYAH_RSC_MGR_H
+#define _GUNYAH_RSC_MGR_H
+
+#include <linux/list.h>
+#include <linux/notifier.h>
+#include <linux/gunyah.h>
+
+#define GH_VMID_INVAL	U16_MAX
+
+/* Gunyah recognizes VMID0 as an alias to the current VM's ID */
+#define GH_VMID_SELF			0
+
+struct gh_rm;
+int gh_rm_notifier_register(struct gh_rm *rm, struct notifier_block *nb);
+int gh_rm_notifier_unregister(struct gh_rm *rm, struct notifier_block *nb);
+void get_gh_rm(struct gh_rm *rm);
+void put_gh_rm(struct gh_rm *rm);
+
+#endif