[linux-2.6-block.git] / drivers / hv / connection.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 *
 * Copyright (c) 2009, Microsoft Corporation.
 *
 * Authors:
 *   Haiyang Zhang <haiyangz@microsoft.com>
 *   Hank Janssen  <hjanssen@microsoft.com>
 */
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/wait.h>
#include <linux/delay.h>
#include <linux/mm.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>
#include <linux/hyperv.h>
#include <linux/export.h>
#include <asm/mshyperv.h>

#include "hyperv_vmbus.h"


struct vmbus_connection vmbus_connection = {
	.conn_state		= DISCONNECTED,
	.next_gpadl_handle	= ATOMIC_INIT(0xE1E10),

	.ready_for_suspend_event= COMPLETION_INITIALIZER(
				  vmbus_connection.ready_for_suspend_event),
	.ready_for_resume_event	= COMPLETION_INITIALIZER(
				  vmbus_connection.ready_for_resume_event),
};
EXPORT_SYMBOL_GPL(vmbus_connection);

/*
 * Negotiated protocol version with the host.
 */
__u32 vmbus_proto_version;
EXPORT_SYMBOL_GPL(vmbus_proto_version);

static __u32 vmbus_get_next_version(__u32 current_version)
{
	switch (current_version) {
	case (VERSION_WIN7):
		return VERSION_WS2008;

	case (VERSION_WIN8):
		return VERSION_WIN7;

	case (VERSION_WIN8_1):
		return VERSION_WIN8;

	case (VERSION_WIN10):
		return VERSION_WIN8_1;

	case (VERSION_WIN10_V5):
		return VERSION_WIN10;

	case (VERSION_WS2008):
	default:
		return VERSION_INVAL;
	}
}

int vmbus_negotiate_version(struct vmbus_channel_msginfo *msginfo, u32 version)
{
	int ret = 0;
	unsigned int cur_cpu;
	struct vmbus_channel_initiate_contact *msg;
	unsigned long flags;

	init_completion(&msginfo->waitevent);

	msg = (struct vmbus_channel_initiate_contact *)msginfo->msg;

	memset(msg, 0, sizeof(*msg));
	msg->header.msgtype = CHANNELMSG_INITIATE_CONTACT;
	msg->vmbus_version_requested = version;

	/*
	 * VMBus protocol 5.0 (VERSION_WIN10_V5) requires that we must use
	 * VMBUS_MESSAGE_CONNECTION_ID_4 for the Initiate Contact Message,
	 * and for subsequent messages, we must use the Message Connection ID
	 * field in the host-returned Version Response Message. And, with
	 * VERSION_WIN10_V5, we don't use msg->interrupt_page, but we tell
	 * the host explicitly that we still use VMBUS_MESSAGE_SINT(2) for
	 * compatibility.
	 *
	 * On old hosts, we should always use VMBUS_MESSAGE_CONNECTION_ID (1).
	 */
	if (version >= VERSION_WIN10_V5) {
		msg->msg_sint = VMBUS_MESSAGE_SINT;
		vmbus_connection.msg_conn_id = VMBUS_MESSAGE_CONNECTION_ID_4;
	} else {
		msg->interrupt_page = virt_to_phys(vmbus_connection.int_page);
		vmbus_connection.msg_conn_id = VMBUS_MESSAGE_CONNECTION_ID;
	}

	msg->monitor_page1 = virt_to_phys(vmbus_connection.monitor_pages[0]);
	msg->monitor_page2 = virt_to_phys(vmbus_connection.monitor_pages[1]);
	/*
	 * We want all channel messages to be delivered on CPU 0.
	 * This has been the behavior pre-win8. This is not
	 * perf issue and having all channel messages delivered on CPU 0
	 * would be ok.
	 * For post win8 hosts, we support receiving channel messagges on
	 * all the CPUs. This is needed for kexec to work correctly where
	 * the CPU attempting to connect may not be CPU 0.
	 */
	if (version >= VERSION_WIN8_1) {
		cur_cpu = get_cpu();
		msg->target_vcpu = hv_cpu_number_to_vp_number(cur_cpu);
		vmbus_connection.connect_cpu = cur_cpu;
		put_cpu();
	} else {
		msg->target_vcpu = 0;
		vmbus_connection.connect_cpu = 0;
	}

	/*
	 * Add to list before we send the request since we may
	 * receive the response before returning from this routine
	 */
	spin_lock_irqsave(&vmbus_connection.channelmsg_lock, flags);
	list_add_tail(&msginfo->msglistentry,
		      &vmbus_connection.chn_msg_list);

	spin_unlock_irqrestore(&vmbus_connection.channelmsg_lock, flags);

	ret = vmbus_post_msg(msg,
			     sizeof(struct vmbus_channel_initiate_contact),
			     true);

	trace_vmbus_negotiate_version(msg, ret);

	if (ret != 0) {
		spin_lock_irqsave(&vmbus_connection.channelmsg_lock, flags);
		list_del(&msginfo->msglistentry);
		spin_unlock_irqrestore(&vmbus_connection.channelmsg_lock,
					flags);
		return ret;
	}

	/* Wait for the connection response */
	wait_for_completion(&msginfo->waitevent);

	spin_lock_irqsave(&vmbus_connection.channelmsg_lock, flags);
	list_del(&msginfo->msglistentry);
	spin_unlock_irqrestore(&vmbus_connection.channelmsg_lock, flags);

	/* Check if successful */
	if (msginfo->response.version_response.version_supported) {
		vmbus_connection.conn_state = CONNECTED;

		if (version >= VERSION_WIN10_V5)
			vmbus_connection.msg_conn_id =
				msginfo->response.version_response.msg_conn_id;
	} else {
		return -ECONNREFUSED;
	}

	return ret;
}

/*
 * vmbus_connect - Sends a connect request on the partition service connection
 */
int vmbus_connect(void)
{
	int ret = 0;
	struct vmbus_channel_msginfo *msginfo = NULL;
	__u32 version;

	/* Initialize the vmbus connection */
	vmbus_connection.conn_state = CONNECTING;
	vmbus_connection.work_queue = create_workqueue("hv_vmbus_con");
	if (!vmbus_connection.work_queue) {
		ret = -ENOMEM;
		goto cleanup;
	}

	vmbus_connection.handle_primary_chan_wq =
		create_workqueue("hv_pri_chan");
	if (!vmbus_connection.handle_primary_chan_wq) {
		ret = -ENOMEM;
		goto cleanup;
	}

	vmbus_connection.handle_sub_chan_wq =
		create_workqueue("hv_sub_chan");
	if (!vmbus_connection.handle_sub_chan_wq) {
		ret = -ENOMEM;
		goto cleanup;
	}

	INIT_LIST_HEAD(&vmbus_connection.chn_msg_list);
	spin_lock_init(&vmbus_connection.channelmsg_lock);

	INIT_LIST_HEAD(&vmbus_connection.chn_list);
	mutex_init(&vmbus_connection.channel_mutex);

	/*
	 * Setup the vmbus event connection for channel interrupt
	 * abstraction stuff
	 */
	vmbus_connection.int_page =
	(void *)__get_free_pages(GFP_KERNEL|__GFP_ZERO, 0);
	if (vmbus_connection.int_page == NULL) {
		ret = -ENOMEM;
		goto cleanup;
	}

	vmbus_connection.recv_int_page = vmbus_connection.int_page;
	vmbus_connection.send_int_page =
		(void *)((unsigned long)vmbus_connection.int_page +
			(PAGE_SIZE >> 1));

	/*
	 * Setup the monitor notification facility. The 1st page for
	 * parent->child and the 2nd page for child->parent
	 */
	vmbus_connection.monitor_pages[0] = (void *)__get_free_pages((GFP_KERNEL|__GFP_ZERO), 0);
	vmbus_connection.monitor_pages[1] = (void *)__get_free_pages((GFP_KERNEL|__GFP_ZERO), 0);
	if ((vmbus_connection.monitor_pages[0] == NULL) ||
	    (vmbus_connection.monitor_pages[1] == NULL)) {
		ret = -ENOMEM;
		goto cleanup;
	}

	msginfo = kzalloc(sizeof(*msginfo) +
			  sizeof(struct vmbus_channel_initiate_contact),
			  GFP_KERNEL);
	if (msginfo == NULL) {
		ret = -ENOMEM;
		goto cleanup;
	}

	/*
	 * Negotiate a compatible VMBUS version number with the
	 * host. We start with the highest number we can support
	 * and work our way down until we negotiate a compatible
	 * version.
	 */

	version = VERSION_CURRENT;

	do {
		ret = vmbus_negotiate_version(msginfo, version);
		if (ret == -ETIMEDOUT)
			goto cleanup;

		if (vmbus_connection.conn_state == CONNECTED)
			break;

		version = vmbus_get_next_version(version);
	} while (version != VERSION_INVAL);

	if (version == VERSION_INVAL)
		goto cleanup;

	vmbus_proto_version = version;
	pr_info("Vmbus version:%d.%d\n",
		version >> 16, version & 0xFFFF);

	kfree(msginfo);
	return 0;

cleanup:
	pr_err("Unable to connect to host\n");

	vmbus_connection.conn_state = DISCONNECTED;
	vmbus_disconnect();

	kfree(msginfo);

	return ret;
}

void vmbus_disconnect(void)
{
	/*
	 * First send the unload request to the host.
	 */
	vmbus_initiate_unload(false);

	if (vmbus_connection.handle_sub_chan_wq)
		destroy_workqueue(vmbus_connection.handle_sub_chan_wq);

	if (vmbus_connection.handle_primary_chan_wq)
		destroy_workqueue(vmbus_connection.handle_primary_chan_wq);

	if (vmbus_connection.work_queue)
		destroy_workqueue(vmbus_connection.work_queue);

	if (vmbus_connection.int_page) {
		free_pages((unsigned long)vmbus_connection.int_page, 0);
		vmbus_connection.int_page = NULL;
	}

	free_pages((unsigned long)vmbus_connection.monitor_pages[0], 0);
	free_pages((unsigned long)vmbus_connection.monitor_pages[1], 0);
	vmbus_connection.monitor_pages[0] = NULL;
	vmbus_connection.monitor_pages[1] = NULL;
}

/*
 * relid2channel - Get the channel object given its
 * child relative id (ie channel id)
 */
struct vmbus_channel *relid2channel(u32 relid)
{
	struct vmbus_channel *channel;
	struct vmbus_channel *found_channel  = NULL;
	struct list_head *cur, *tmp;
	struct vmbus_channel *cur_sc;

	BUG_ON(!mutex_is_locked(&vmbus_connection.channel_mutex));

	list_for_each_entry(channel, &vmbus_connection.chn_list, listentry) {
		if (channel->offermsg.child_relid == relid) {
			found_channel = channel;
			break;
		} else if (!list_empty(&channel->sc_list)) {
			/*
			 * Deal with sub-channels.
			 */
			list_for_each_safe(cur, tmp, &channel->sc_list) {
				cur_sc = list_entry(cur, struct vmbus_channel,
							sc_list);
				if (cur_sc->offermsg.child_relid == relid) {
					found_channel = cur_sc;
					break;
				}
			}
		}
	}

	return found_channel;
}

/*
 * vmbus_on_event - Process a channel event notification
 *
 * For batched channels (default) optimize host to guest signaling
 * by ensuring:
 * 1. While reading the channel, we disable interrupts from host.
 * 2. Ensure that we process all posted messages from the host
 *    before returning from this callback.
 * 3. Once we return, enable signaling from the host. Once this
 *    state is set we check to see if additional packets are
 *    available to read. In this case we repeat the process.
 *    If this tasklet has been running for a long time
 *    then reschedule ourselves.
 */
void vmbus_on_event(unsigned long data)
{
	struct vmbus_channel *channel = (void *) data;
	unsigned long time_limit = jiffies + 2;

	trace_vmbus_on_event(channel);

	do {
		void (*callback_fn)(void *);

		/* A channel once created is persistent even when
		 * there is no driver handling the device. An
		 * unloading driver sets the onchannel_callback to NULL.
		 */
		callback_fn = READ_ONCE(channel->onchannel_callback);
		if (unlikely(callback_fn == NULL))
			return;

		(*callback_fn)(channel->channel_callback_context);

		if (channel->callback_mode != HV_CALL_BATCHED)
			return;

		if (likely(hv_end_read(&channel->inbound) == 0))
			return;

		hv_begin_read(&channel->inbound);
	} while (likely(time_before(jiffies, time_limit)));

	/* The time limit (2 jiffies) has been reached */
	tasklet_schedule(&channel->callback_event);
}

/*
 * vmbus_post_msg - Send a msg on the vmbus's message connection
 */
int vmbus_post_msg(void *buffer, size_t buflen, bool can_sleep)
{
	struct vmbus_channel_message_header *hdr;
	union hv_connection_id conn_id;
	int ret = 0;
	int retries = 0;
	u32 usec = 1;

	conn_id.asu32 = 0;
	conn_id.u.id = vmbus_connection.msg_conn_id;

	/*
	 * hv_post_message() can have transient failures because of
	 * insufficient resources. Retry the operation a couple of
	 * times before giving up.
	 */
	while (retries < 100) {
		ret = hv_post_message(conn_id, 1, buffer, buflen);

		switch (ret) {
		case HV_STATUS_INVALID_CONNECTION_ID:
			/*
			 * See vmbus_negotiate_version(): VMBus protocol 5.0
			 * requires that we must use
			 * VMBUS_MESSAGE_CONNECTION_ID_4 for the Initiate
			 * Contact message, but on old hosts that only
			 * support VMBus protocol 4.0 or lower, here we get
			 * HV_STATUS_INVALID_CONNECTION_ID and we should
			 * return an error immediately without retrying.
			 */
			hdr = buffer;
			if (hdr->msgtype == CHANNELMSG_INITIATE_CONTACT)
				return -EINVAL;
			/*
			 * We could get this if we send messages too
			 * frequently.
			 */
			ret = -EAGAIN;
			break;
		case HV_STATUS_INSUFFICIENT_MEMORY:
		case HV_STATUS_INSUFFICIENT_BUFFERS:
			ret = -ENOBUFS;
			break;
		case HV_STATUS_SUCCESS:
			return ret;
		default:
			pr_err("hv_post_msg() failed; error code:%d\n", ret);
			return -EINVAL;
		}

		retries++;
		if (can_sleep && usec > 1000)
			msleep(usec / 1000);
		else if (usec < MAX_UDELAY_MS * 1000)
			udelay(usec);
		else
			mdelay(usec / 1000);

		if (retries < 22)
			usec *= 2;
	}
	return ret;
}

/*
 * vmbus_set_event - Send an event notification to the parent
 */
void vmbus_set_event(struct vmbus_channel *channel)
{
	u32 child_relid = channel->offermsg.child_relid;

	if (!channel->is_dedicated_interrupt)
		vmbus_send_interrupt(child_relid);

	++channel->sig_events;

	hv_do_fast_hypercall8(HVCALL_SIGNAL_EVENT, channel->sig_event);
}
EXPORT_SYMBOL_GPL(vmbus_set_event);
Commit	Line	Data
3b20eb23	1	// SPDX-License-Identifier: GPL-2.0-only
3e7ee490 HJ	2	/*
	3	*
	4	* Copyright (c) 2009, Microsoft Corporation.
	5	*
3e7ee490 HJ	6	* Authors:
	7	* Haiyang Zhang <haiyangz@microsoft.com>
	8	* Hank Janssen <hjanssen@microsoft.com>
3e7ee490	9	*/
0a46618d HJ	10	#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
0a46618d HJ	11
a0086dc5	12	#include <linux/kernel.h>
0c3b7b2f S	13	#include <linux/sched.h>
0c3b7b2f S	14	#include <linux/wait.h>
5289d3d1	15	#include <linux/delay.h>
a0086dc5	16	#include <linux/mm.h>
5a0e3ad6	17	#include <linux/slab.h>
a0086dc5	18	#include <linux/vmalloc.h>
46a97191	19	#include <linux/hyperv.h>
37f7278b	20	#include <linux/export.h>
fc53662f VK	21	#include <asm/mshyperv.h>
fc53662f VK	22
0f2a6619	23	#include "hyperv_vmbus.h"
3e7ee490	24
3e7ee490	25
da9fcb72 HZ	26	struct vmbus_connection vmbus_connection = {
	27	.conn_state = DISCONNECTED,
	28	.next_gpadl_handle = ATOMIC_INIT(0xE1E10),
b307b389 DC	29
	30	.ready_for_suspend_event= COMPLETION_INITIALIZER(
	31	vmbus_connection.ready_for_suspend_event),
d8bd2d44 DC	32	.ready_for_resume_event = COMPLETION_INITIALIZER(
d8bd2d44 DC	33	vmbus_connection.ready_for_resume_event),
3e7ee490	34	};
95096f2f	35	EXPORT_SYMBOL_GPL(vmbus_connection);
3e7ee490	36
37f7278b S	37	/*
	38	* Negotiated protocol version with the host.
	39	*/
	40	__u32 vmbus_proto_version;
	41	EXPORT_SYMBOL_GPL(vmbus_proto_version);
	42
610071c3 S	43	static __u32 vmbus_get_next_version(__u32 current_version)
	44	{
	45	switch (current_version) {
	46	case (VERSION_WIN7):
	47	return VERSION_WS2008;
	48
	49	case (VERSION_WIN8):
	50	return VERSION_WIN7;
	51
03367ef5 S	52	case (VERSION_WIN8_1):
	53	return VERSION_WIN8;
	54
6c4e5f9c KM	55	case (VERSION_WIN10):
	56	return VERSION_WIN8_1;
	57
ae20b254 DC	58	case (VERSION_WIN10_V5):
	59	return VERSION_WIN10;
	60
610071c3 S	61	case (VERSION_WS2008):
	62	default:
	63	return VERSION_INVAL;
	64	}
	65	}
	66
f53335e3	67	int vmbus_negotiate_version(struct vmbus_channel_msginfo *msginfo, u32 version)
610071c3 S	68	{
610071c3 S	69	int ret = 0;
41e270f6	70	unsigned int cur_cpu;
610071c3 S	71	struct vmbus_channel_initiate_contact *msg;
610071c3 S	72	unsigned long flags;
610071c3 S	73
	74	init_completion(&msginfo->waitevent);
	75
	76	msg = (struct vmbus_channel_initiate_contact *)msginfo->msg;
	77
ae20b254	78	memset(msg, 0, sizeof(*msg));
610071c3 S	79	msg->header.msgtype = CHANNELMSG_INITIATE_CONTACT;
610071c3 S	80	msg->vmbus_version_requested = version;
ae20b254 DC	81
	82	/*
	83	* VMBus protocol 5.0 (VERSION_WIN10_V5) requires that we must use
	84	* VMBUS_MESSAGE_CONNECTION_ID_4 for the Initiate Contact Message,
	85	* and for subsequent messages, we must use the Message Connection ID
	86	* field in the host-returned Version Response Message. And, with
	87	* VERSION_WIN10_V5, we don't use msg->interrupt_page, but we tell
	88	* the host explicitly that we still use VMBUS_MESSAGE_SINT(2) for
	89	* compatibility.
	90	*
	91	* On old hosts, we should always use VMBUS_MESSAGE_CONNECTION_ID (1).
	92	*/
	93	if (version >= VERSION_WIN10_V5) {
	94	msg->msg_sint = VMBUS_MESSAGE_SINT;
	95	vmbus_connection.msg_conn_id = VMBUS_MESSAGE_CONNECTION_ID_4;
	96	} else {
	97	msg->interrupt_page = virt_to_phys(vmbus_connection.int_page);
	98	vmbus_connection.msg_conn_id = VMBUS_MESSAGE_CONNECTION_ID;
	99	}
	100
8681db44 GKH	101	msg->monitor_page1 = virt_to_phys(vmbus_connection.monitor_pages[0]);
8681db44 GKH	102	msg->monitor_page2 = virt_to_phys(vmbus_connection.monitor_pages[1]);
b282e4c0 S	103	/*
	104	* We want all channel messages to be delivered on CPU 0.
	105	* This has been the behavior pre-win8. This is not
	106	* perf issue and having all channel messages delivered on CPU 0
	107	* would be ok.
72686447 AN	108	* For post win8 hosts, we support receiving channel messagges on
	109	* all the CPUs. This is needed for kexec to work correctly where
	110	* the CPU attempting to connect may not be CPU 0.
b282e4c0	111	*/
54a66265	112	if (version >= VERSION_WIN8_1) {
41e270f6 DC	113	cur_cpu = get_cpu();
	114	msg->target_vcpu = hv_cpu_number_to_vp_number(cur_cpu);
	115	vmbus_connection.connect_cpu = cur_cpu;
	116	put_cpu();
54a66265	117	} else {
72686447	118	msg->target_vcpu = 0;
54a66265 S	119	vmbus_connection.connect_cpu = 0;
54a66265 S	120	}
610071c3 S	121
	122	/*
	123	* Add to list before we send the request since we may
	124	* receive the response before returning from this routine
	125	*/
	126	spin_lock_irqsave(&vmbus_connection.channelmsg_lock, flags);
	127	list_add_tail(&msginfo->msglistentry,
	128	&vmbus_connection.chn_msg_list);
	129
	130	spin_unlock_irqrestore(&vmbus_connection.channelmsg_lock, flags);
	131
	132	ret = vmbus_post_msg(msg,
c0bb0392 VK	133	sizeof(struct vmbus_channel_initiate_contact),
c0bb0392 VK	134	true);
034ebf55 VK	135
	136	trace_vmbus_negotiate_version(msg, ret);
	137
610071c3 S	138	if (ret != 0) {
	139	spin_lock_irqsave(&vmbus_connection.channelmsg_lock, flags);
	140	list_del(&msginfo->msglistentry);
	141	spin_unlock_irqrestore(&vmbus_connection.channelmsg_lock,
	142	flags);
	143	return ret;
	144	}
	145
	146	/* Wait for the connection response */
269f9794	147	wait_for_completion(&msginfo->waitevent);
610071c3 S	148
	149	spin_lock_irqsave(&vmbus_connection.channelmsg_lock, flags);
	150	list_del(&msginfo->msglistentry);
	151	spin_unlock_irqrestore(&vmbus_connection.channelmsg_lock, flags);
	152
	153	/* Check if successful */
	154	if (msginfo->response.version_response.version_supported) {
	155	vmbus_connection.conn_state = CONNECTED;
ae20b254 DC	156
	157	if (version >= VERSION_WIN10_V5)
	158	vmbus_connection.msg_conn_id =
	159	msginfo->response.version_response.msg_conn_id;
610071c3	160	} else {
610071c3 S	161	return -ECONNREFUSED;
	162	}
	163
	164	return ret;
	165	}
	166
3e189519	167	/*
c6977677	168	* vmbus_connect - Sends a connect request on the partition service connection
fd8b85ea	169	*/
c6977677	170	int vmbus_connect(void)
3e7ee490	171	{
fd8b85ea	172	int ret = 0;
15b2f647	173	struct vmbus_channel_msginfo *msginfo = NULL;
610071c3	174	__u32 version;
3e7ee490	175
454f18a9	176	/* Initialize the vmbus connection */
da9fcb72 HZ	177	vmbus_connection.conn_state = CONNECTING;
	178	vmbus_connection.work_queue = create_workqueue("hv_vmbus_con");
	179	if (!vmbus_connection.work_queue) {
3a7546d9	180	ret = -ENOMEM;
b0043863	181	goto cleanup;
de65a384	182	}
3e7ee490	183
37c2578c DC	184	vmbus_connection.handle_primary_chan_wq =
	185	create_workqueue("hv_pri_chan");
	186	if (!vmbus_connection.handle_primary_chan_wq) {
	187	ret = -ENOMEM;
	188	goto cleanup;
	189	}
	190
	191	vmbus_connection.handle_sub_chan_wq =
	192	create_workqueue("hv_sub_chan");
	193	if (!vmbus_connection.handle_sub_chan_wq) {
	194	ret = -ENOMEM;
	195	goto cleanup;
	196	}
	197
da9fcb72	198	INIT_LIST_HEAD(&vmbus_connection.chn_msg_list);
15b2f647	199	spin_lock_init(&vmbus_connection.channelmsg_lock);
3e7ee490	200
da9fcb72	201	INIT_LIST_HEAD(&vmbus_connection.chn_list);
d6f591e3	202	mutex_init(&vmbus_connection.channel_mutex);
3e7ee490	203
454f18a9 BP	204	/*
	205	* Setup the vmbus event connection for channel interrupt
	206	* abstraction stuff
	207	*/
df3493e0 S	208	vmbus_connection.int_page =
df3493e0 S	209	(void *)__get_free_pages(GFP_KERNEL\|__GFP_ZERO, 0);
da9fcb72	210	if (vmbus_connection.int_page == NULL) {
3a7546d9	211	ret = -ENOMEM;
b0043863	212	goto cleanup;
3e7ee490 HJ	213	}
3e7ee490 HJ	214
da9fcb72 HZ	215	vmbus_connection.recv_int_page = vmbus_connection.int_page;
	216	vmbus_connection.send_int_page =
	217	(void *)((unsigned long)vmbus_connection.int_page +
fd8b85ea	218	(PAGE_SIZE >> 1));
3e7ee490	219
fd8b85ea GKH	220	/*
	221	* Setup the monitor notification facility. The 1st page for
	222	* parent->child and the 2nd page for child->parent
454f18a9	223	*/
8681db44 GKH	224	vmbus_connection.monitor_pages[0] = (void *)__get_free_pages((GFP_KERNEL\|__GFP_ZERO), 0);
	225	vmbus_connection.monitor_pages[1] = (void *)__get_free_pages((GFP_KERNEL\|__GFP_ZERO), 0);
	226	if ((vmbus_connection.monitor_pages[0] == NULL) \|\|
	227	(vmbus_connection.monitor_pages[1] == NULL)) {
3a7546d9	228	ret = -ENOMEM;
b0043863	229	goto cleanup;
3e7ee490 HJ	230	}
3e7ee490 HJ	231
15b2f647	232	msginfo = kzalloc(sizeof(*msginfo) +
fd8b85ea GKH	233	sizeof(struct vmbus_channel_initiate_contact),
fd8b85ea GKH	234	GFP_KERNEL);
15b2f647	235	if (msginfo == NULL) {
8cad0af9	236	ret = -ENOMEM;
b0043863	237	goto cleanup;
3e7ee490 HJ	238	}
3e7ee490 HJ	239
454f18a9	240	/*
610071c3 S	241	* Negotiate a compatible VMBUS version number with the
	242	* host. We start with the highest number we can support
	243	* and work our way down until we negotiate a compatible
	244	* version.
454f18a9	245	*/
3e7ee490	246
2a5c43a8	247	version = VERSION_CURRENT;
3e7ee490	248
610071c3 S	249	do {
610071c3 S	250	ret = vmbus_negotiate_version(msginfo, version);
8bbf9f44	251	if (ret == -ETIMEDOUT)
666b9adc S	252	goto cleanup;
	253
	254	if (vmbus_connection.conn_state == CONNECTED)
610071c3	255	break;
3e7ee490	256
610071c3 S	257	version = vmbus_get_next_version(version);
610071c3 S	258	} while (version != VERSION_INVAL);
3e7ee490	259
610071c3	260	if (version == VERSION_INVAL)
b0043863	261	goto cleanup;
3e7ee490	262
37f7278b	263	vmbus_proto_version = version;
8de8af7e S	264	pr_info("Vmbus version:%d.%d\n",
8de8af7e S	265	version >> 16, version & 0xFFFF);
3bacaf0c	266
15b2f647	267	kfree(msginfo);
3e7ee490 HJ	268	return 0;
3e7ee490 HJ	269
b0043863	270	cleanup:
3bacaf0c	271	pr_err("Unable to connect to host\n");
09a19628	272
da9fcb72	273	vmbus_connection.conn_state = DISCONNECTED;
09a19628 VK	274	vmbus_disconnect();
	275
	276	kfree(msginfo);
	277
	278	return ret;
	279	}
3e7ee490	280
09a19628 VK	281	void vmbus_disconnect(void)
09a19628 VK	282	{
2db84eff S	283	/*
	284	* First send the unload request to the host.
	285	*/
75ff3a8a	286	vmbus_initiate_unload(false);
2db84eff	287
37c2578c DC	288	if (vmbus_connection.handle_sub_chan_wq)
	289	destroy_workqueue(vmbus_connection.handle_sub_chan_wq);
	290
	291	if (vmbus_connection.handle_primary_chan_wq)
	292	destroy_workqueue(vmbus_connection.handle_primary_chan_wq);
	293
	294	if (vmbus_connection.work_queue)
da9fcb72	295	destroy_workqueue(vmbus_connection.work_queue);
3e7ee490	296
da9fcb72	297	if (vmbus_connection.int_page) {
df3493e0	298	free_pages((unsigned long)vmbus_connection.int_page, 0);
da9fcb72	299	vmbus_connection.int_page = NULL;
3e7ee490 HJ	300	}
3e7ee490 HJ	301
a100d88d RK	302	free_pages((unsigned long)vmbus_connection.monitor_pages[0], 0);
a100d88d RK	303	free_pages((unsigned long)vmbus_connection.monitor_pages[1], 0);
8681db44 GKH	304	vmbus_connection.monitor_pages[0] = NULL;
8681db44 GKH	305	vmbus_connection.monitor_pages[1] = NULL;
3e7ee490 HJ	306	}
3e7ee490 HJ	307
3e189519	308	/*
c6977677 HZ	309	* relid2channel - Get the channel object given its
c6977677 HZ	310	* child relative id (ie channel id)
fd8b85ea	311	*/
d43e2fe7	312	struct vmbus_channel *relid2channel(u32 relid)
3e7ee490	313	{
aded7165	314	struct vmbus_channel *channel;
15b2f647	315	struct vmbus_channel *found_channel = NULL;
e68d2971 S	316	struct list_head cur, tmp;
e68d2971 S	317	struct vmbus_channel *cur_sc;
3e7ee490	318
85d9aa70 DC	319	BUG_ON(!mutex_is_locked(&vmbus_connection.channel_mutex));
85d9aa70 DC	320
da9fcb72	321	list_for_each_entry(channel, &vmbus_connection.chn_list, listentry) {
15b2f647 HZ	322	if (channel->offermsg.child_relid == relid) {
15b2f647 HZ	323	found_channel = channel;
3e7ee490	324	break;
e68d2971 S	325	} else if (!list_empty(&channel->sc_list)) {
	326	/*
	327	* Deal with sub-channels.
	328	*/
	329	list_for_each_safe(cur, tmp, &channel->sc_list) {
	330	cur_sc = list_entry(cur, struct vmbus_channel,
	331	sc_list);
	332	if (cur_sc->offermsg.child_relid == relid) {
	333	found_channel = cur_sc;
	334	break;
	335	}
	336	}
3e7ee490 HJ	337	}
3e7ee490 HJ	338	}
3e7ee490	339
15b2f647	340	return found_channel;
3e7ee490 HJ	341	}
3e7ee490 HJ	342
3e189519	343	/*
631e63a9	344	* vmbus_on_event - Process a channel event notification
ada6eb11 SH	345	*
	346	* For batched channels (default) optimize host to guest signaling
	347	* by ensuring:
	348	* 1. While reading the channel, we disable interrupts from host.
	349	* 2. Ensure that we process all posted messages from the host
	350	* before returning from this callback.
	351	* 3. Once we return, enable signaling from the host. Once this
	352	* state is set we check to see if additional packets are
	353	* available to read. In this case we repeat the process.
	354	* If this tasklet has been running for a long time
	355	* then reschedule ourselves.
fd8b85ea	356	*/
631e63a9	357	void vmbus_on_event(unsigned long data)
3e7ee490	358	{
631e63a9	359	struct vmbus_channel channel = (void ) data;
ada6eb11	360	unsigned long time_limit = jiffies + 2;
3e7ee490	361
991f8f1c VK	362	trace_vmbus_on_event(channel);
991f8f1c VK	363
ada6eb11 SH	364	do {
	365	void (callback_fn)(void );
	366
	367	/* A channel once created is persistent even when
	368	* there is no driver handling the device. An
	369	* unloading driver sets the onchannel_callback to NULL.
f878f3d5	370	*/
ada6eb11 SH	371	callback_fn = READ_ONCE(channel->onchannel_callback);
	372	if (unlikely(callback_fn == NULL))
	373	return;
f878f3d5	374
ada6eb11 SH	375	(*callback_fn)(channel->channel_callback_context);
	376
	377	if (channel->callback_mode != HV_CALL_BATCHED)
	378	return;
	379
	380	if (likely(hv_end_read(&channel->inbound) == 0))
	381	return;
	382
	383	hv_begin_read(&channel->inbound);
	384	} while (likely(time_before(jiffies, time_limit)));
	385
	386	/* The time limit (2 jiffies) has been reached */
	387	tasklet_schedule(&channel->callback_event);
3e7ee490 HJ	388	}
3e7ee490 HJ	389
3e189519	390	/*
c6977677	391	* vmbus_post_msg - Send a msg on the vmbus's message connection
fd8b85ea	392	*/
c0bb0392	393	int vmbus_post_msg(void *buffer, size_t buflen, bool can_sleep)
3e7ee490	394	{
ae20b254	395	struct vmbus_channel_message_header *hdr;
15b2f647	396	union hv_connection_id conn_id;
5289d3d1 S	397	int ret = 0;
5289d3d1 S	398	int retries = 0;
8de0d7e9	399	u32 usec = 1;
3e7ee490	400
15b2f647	401	conn_id.asu32 = 0;
ae20b254	402	conn_id.u.id = vmbus_connection.msg_conn_id;
5289d3d1 S	403
	404	/*
	405	* hv_post_message() can have transient failures because of
	406	* insufficient resources. Retry the operation a couple of
	407	* times before giving up.
	408	*/
c0bb0392	409	while (retries < 100) {
fdeebcc6 S	410	ret = hv_post_message(conn_id, 1, buffer, buflen);
	411
	412	switch (ret) {
89f9f679	413	case HV_STATUS_INVALID_CONNECTION_ID:
ae20b254 DC	414	/*
	415	* See vmbus_negotiate_version(): VMBus protocol 5.0
	416	* requires that we must use
	417	* VMBUS_MESSAGE_CONNECTION_ID_4 for the Initiate
	418	* Contact message, but on old hosts that only
	419	* support VMBus protocol 4.0 or lower, here we get
	420	* HV_STATUS_INVALID_CONNECTION_ID and we should
	421	* return an error immediately without retrying.
	422	*/
89760937	423	hdr = buffer;
ae20b254 DC	424	if (hdr->msgtype == CHANNELMSG_INITIATE_CONTACT)
ae20b254 DC	425	return -EINVAL;
89f9f679 DC	426	/*
	427	* We could get this if we send messages too
	428	* frequently.
	429	*/
	430	ret = -EAGAIN;
	431	break;
	432	case HV_STATUS_INSUFFICIENT_MEMORY:
fdeebcc6	433	case HV_STATUS_INSUFFICIENT_BUFFERS:
48f4ccdf	434	ret = -ENOBUFS;
fdeebcc6 S	435	break;
fdeebcc6 S	436	case HV_STATUS_SUCCESS:
5289d3d1	437	return ret;
fdeebcc6 S	438	default:
	439	pr_err("hv_post_msg() failed; error code:%d\n", ret);
	440	return -EINVAL;
	441	}
	442
5289d3d1	443	retries++;
c0bb0392 VK	444	if (can_sleep && usec > 1000)
	445	msleep(usec / 1000);
	446	else if (usec < MAX_UDELAY_MS * 1000)
	447	udelay(usec);
	448	else
	449	mdelay(usec / 1000);
	450
e917a5e2	451	if (retries < 22)
8de0d7e9	452	usec *= 2;
5289d3d1 S	453	}
5289d3d1 S	454	return ret;
3e7ee490 HJ	455	}
3e7ee490 HJ	456
3e189519	457	/*
c6977677	458	* vmbus_set_event - Send an event notification to the parent
fd8b85ea	459	*/
1b807e10	460	void vmbus_set_event(struct vmbus_channel *channel)
3e7ee490	461	{
21c3bef5	462	u32 child_relid = channel->offermsg.child_relid;
7c369f40	463
5c1bec61 SH	464	if (!channel->is_dedicated_interrupt)
5c1bec61 SH	465	vmbus_send_interrupt(child_relid);
3be77774	466
6981fbf3 SH	467	++channel->sig_events;
6981fbf3 SH	468
05784171	469	hv_do_fast_hypercall8(HVCALL_SIGNAL_EVENT, channel->sig_event);
3e7ee490	470	}
5cc47247	471	EXPORT_SYMBOL_GPL(vmbus_set_event);