Merge commit '8700c95adb03' into timers/nohz

The full dynticks tree needs the latest RCU and sched
upstream updates in order to fix some dependencies.

Merge a common upstream merge point that has these
updates.

Conflicts:
	include/linux/perf_event.h
	kernel/rcutree.h
	kernel/rcutree_plugin.h

Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>

7 files changed, 800 insertions, 60 deletions

diff --git a/drivers/hv/Makefile b/drivers/hv/Makefile
index e6abfa02d8b7..0a74b5661186 100644
--- a/drivers/hv/Makefile
+++ b/drivers/hv/Makefile
@@ -5,4 +5,4 @@ obj-$(CONFIG_HYPERV_BALLOON)	+= hv_balloon.o
 hv_vmbus-y := vmbus_drv.o \
 		 hv.o connection.o channel.o \
 		 channel_mgmt.o ring_buffer.o
-hv_utils-y := hv_util.o hv_kvp.o
+hv_utils-y := hv_util.o hv_kvp.o hv_snapshot.o
diff --git a/drivers/hv/channel_mgmt.c b/drivers/hv/channel_mgmt.c
index ff1be167eb04..bad8128b283a 100644
--- a/drivers/hv/channel_mgmt.c
+++ b/drivers/hv/channel_mgmt.c
@@ -165,8 +165,19 @@ static void vmbus_process_rescind_offer(struct work_struct *work)
 	struct vmbus_channel *channel = container_of(work,
 						     struct vmbus_channel,
 						     work);
+	unsigned long flags;
+	struct vmbus_channel_relid_released msg;
 
 	vmbus_device_unregister(channel->device_obj);
+	memset(&msg, 0, sizeof(struct vmbus_channel_relid_released));
+	msg.child_relid = channel->offermsg.child_relid;
+	msg.header.msgtype = CHANNELMSG_RELID_RELEASED;
+	vmbus_post_msg(&msg, sizeof(struct vmbus_channel_relid_released));
+
+	spin_lock_irqsave(&vmbus_connection.channel_lock, flags);
+	list_del(&channel->listentry);
+	spin_unlock_irqrestore(&vmbus_connection.channel_lock, flags);
+	free_channel(channel);
 }
 
 void vmbus_free_channels(void)
diff --git a/drivers/hv/hv.c b/drivers/hv/hv.c
index 731158910c1e..ae4923756d98 100644
--- a/drivers/hv/hv.c
+++ b/drivers/hv/hv.c
@@ -289,9 +289,8 @@ void hv_synic_init(void *arg)
 	/* Check the version */
 	rdmsrl(HV_X64_MSR_SVERSION, version);
 
-	hv_context.event_dpc[cpu] = (struct tasklet_struct *)
-					kmalloc(sizeof(struct tasklet_struct),
-						GFP_ATOMIC);
+	hv_context.event_dpc[cpu] = kmalloc(sizeof(struct tasklet_struct),
+					    GFP_ATOMIC);
 	if (hv_context.event_dpc[cpu] == NULL) {
 		pr_err("Unable to allocate event dpc\n");
 		goto cleanup;
diff --git a/drivers/hv/hv_balloon.c b/drivers/hv/hv_balloon.c
index 37873213e24f..4c605c70ebf9 100644
--- a/drivers/hv/hv_balloon.c
+++ b/drivers/hv/hv_balloon.c
@@ -117,7 +117,14 @@ union dm_caps {
 	struct {
 		__u64 balloon:1;
 		__u64 hot_add:1;
-		__u64 reservedz:62;
+		/*
+		 * To support guests that may have alignment
+		 * limitations on hot-add, the guest can specify
+		 * its alignment requirements; a value of n
+		 * represents an alignment of 2^n in mega bytes.
+		 */
+		__u64 hot_add_alignment:4;
+		__u64 reservedz:58;
 	} cap_bits;
 	__u64 caps;
 } __packed;
@@ -412,13 +419,45 @@ struct dm_info_msg {
  * End protocol definitions.
  */
 
-static bool hot_add;
+/*
+ * State to manage hot adding memory into the guest.
+ * The range start_pfn : end_pfn specifies the range
+ * that the host has asked us to hot add. The range
+ * start_pfn : ha_end_pfn specifies the range that we have
+ * currently hot added. We hot add in multiples of 128M
+ * chunks; it is possible that we may not be able to bring
+ * online all the pages in the region. The range
+ * covered_start_pfn : covered_end_pfn defines the pages that can
+ * be brough online.
+ */
+
+struct hv_hotadd_state {
+	struct list_head list;
+	unsigned long start_pfn;
+	unsigned long covered_start_pfn;
+	unsigned long covered_end_pfn;
+	unsigned long ha_end_pfn;
+	unsigned long end_pfn;
+};
+
+struct balloon_state {
+	__u32 num_pages;
+	struct work_struct wrk;
+};
+
+struct hot_add_wrk {
+	union dm_mem_page_range ha_page_range;
+	union dm_mem_page_range ha_region_range;
+	struct work_struct wrk;
+};
+
+static bool hot_add = true;
 static bool do_hot_add;
 /*
  * Delay reporting memory pressure by
  * the specified number of seconds.
  */
-static uint pressure_report_delay = 30;
+static uint pressure_report_delay = 45;
 
 module_param(hot_add, bool, (S_IRUGO | S_IWUSR));
 MODULE_PARM_DESC(hot_add, "If set attempt memory hot_add");
@@ -446,6 +485,7 @@ enum hv_dm_state {
 static __u8 recv_buffer[PAGE_SIZE];
 static __u8 *send_buffer;
 #define PAGES_IN_2M	512
+#define HA_CHUNK (32 * 1024)
 
 struct hv_dynmem_device {
 	struct hv_device *dev;
@@ -459,7 +499,28 @@ struct hv_dynmem_device {
 	unsigned int num_pages_ballooned;
 
 	/*
-	 * This thread handles both balloon/hot-add
+	 * State to manage the ballooning (up) operation.
+	 */
+	struct balloon_state balloon_wrk;
+
+	/*
+	 * State to execute the "hot-add" operation.
+	 */
+	struct hot_add_wrk ha_wrk;
+
+	/*
+	 * This state tracks if the host has specified a hot-add
+	 * region.
+	 */
+	bool host_specified_ha_region;
+
+	/*
+	 * State to synchronize hot-add.
+	 */
+	struct completion  ol_waitevent;
+	bool ha_waiting;
+	/*
+	 * This thread handles hot-add
 	 * requests from the host as well as notifying
 	 * the host with regards to memory pressure in
 	 * the guest.
@@ -467,6 +528,11 @@ struct hv_dynmem_device {
 	struct task_struct *thread;
 
 	/*
+	 * A list of hot-add regions.
+	 */
+	struct list_head ha_region_list;
+
+	/*
 	 * We start with the highest version we can support
 	 * and downgrade based on the host; we save here the
 	 * next version to try.
@@ -476,35 +542,358 @@ struct hv_dynmem_device {
 
 static struct hv_dynmem_device dm_device;
 
-static void hot_add_req(struct hv_dynmem_device *dm, struct dm_hot_add *msg)
+#ifdef CONFIG_MEMORY_HOTPLUG
+
+static void hv_bring_pgs_online(unsigned long start_pfn, unsigned long size)
 {
+	int i;
 
-	struct dm_hot_add_response resp;
+	for (i = 0; i < size; i++) {
+		struct page *pg;
+		pg = pfn_to_page(start_pfn + i);
+		__online_page_set_limits(pg);
+		__online_page_increment_counters(pg);
+		__online_page_free(pg);
+	}
+}
 
-	if (do_hot_add) {
+static void hv_mem_hot_add(unsigned long start, unsigned long size,
+				unsigned long pfn_count,
+				struct hv_hotadd_state *has)
+{
+	int ret = 0;
+	int i, nid, t;
+	unsigned long start_pfn;
+	unsigned long processed_pfn;
+	unsigned long total_pfn = pfn_count;
+
+	for (i = 0; i < (size/HA_CHUNK); i++) {
+		start_pfn = start + (i * HA_CHUNK);
+		has->ha_end_pfn +=  HA_CHUNK;
+
+		if (total_pfn > HA_CHUNK) {
+			processed_pfn = HA_CHUNK;
+			total_pfn -= HA_CHUNK;
+		} else {
+			processed_pfn = total_pfn;
+			total_pfn = 0;
+		}
+
+		has->covered_end_pfn +=  processed_pfn;
+
+		init_completion(&dm_device.ol_waitevent);
+		dm_device.ha_waiting = true;
+
+		nid = memory_add_physaddr_to_nid(PFN_PHYS(start_pfn));
+		ret = add_memory(nid, PFN_PHYS((start_pfn)),
+				(HA_CHUNK << PAGE_SHIFT));
 
-		pr_info("Memory hot add not supported\n");
+		if (ret) {
+			pr_info("hot_add memory failed error is %d\n", ret);
+			if (ret == -EEXIST) {
+				/*
+				 * This error indicates that the error
+				 * is not a transient failure. This is the
+				 * case where the guest's physical address map
+				 * precludes hot adding memory. Stop all further
+				 * memory hot-add.
+				 */
+				do_hot_add = false;
+			}
+			has->ha_end_pfn -= HA_CHUNK;
+			has->covered_end_pfn -=  processed_pfn;
+			break;
+		}
 
 		/*
-		 * Currently we do not support hot add.
-		 * Just fail the request.
+		 * Wait for the memory block to be onlined.
 		 */
+		t = wait_for_completion_timeout(&dm_device.ol_waitevent, 5*HZ);
+		if (t == 0) {
+			pr_info("hot_add memory timedout\n");
+			has->ha_end_pfn -= HA_CHUNK;
+			has->covered_end_pfn -=  processed_pfn;
+			break;
+		}
+
+	}
+
+	return;
+}
+
+static void hv_online_page(struct page *pg)
+{
+	struct list_head *cur;
+	struct hv_hotadd_state *has;
+	unsigned long cur_start_pgp;
+	unsigned long cur_end_pgp;
+
+	if (dm_device.ha_waiting) {
+		dm_device.ha_waiting = false;
+		complete(&dm_device.ol_waitevent);
+	}
+
+	list_for_each(cur, &dm_device.ha_region_list) {
+		has = list_entry(cur, struct hv_hotadd_state, list);
+		cur_start_pgp = (unsigned long)
+				pfn_to_page(has->covered_start_pfn);
+		cur_end_pgp = (unsigned long)pfn_to_page(has->covered_end_pfn);
+
+		if (((unsigned long)pg >= cur_start_pgp) &&
+			((unsigned long)pg < cur_end_pgp)) {
+			/*
+			 * This frame is currently backed; online the
+			 * page.
+			 */
+			__online_page_set_limits(pg);
+			__online_page_increment_counters(pg);
+			__online_page_free(pg);
+			has->covered_start_pfn++;
+		}
+	}
+}
+
+static bool pfn_covered(unsigned long start_pfn, unsigned long pfn_cnt)
+{
+	struct list_head *cur;
+	struct hv_hotadd_state *has;
+	unsigned long residual, new_inc;
+
+	if (list_empty(&dm_device.ha_region_list))
+		return false;
+
+	list_for_each(cur, &dm_device.ha_region_list) {
+		has = list_entry(cur, struct hv_hotadd_state, list);
+
+		/*
+		 * If the pfn range we are dealing with is not in the current
+		 * "hot add block", move on.
+		 */
+		if ((start_pfn >= has->end_pfn))
+			continue;
+		/*
+		 * If the current hot add-request extends beyond
+		 * our current limit; extend it.
+		 */
+		if ((start_pfn + pfn_cnt) > has->end_pfn) {
+			residual = (start_pfn + pfn_cnt - has->end_pfn);
+			/*
+			 * Extend the region by multiples of HA_CHUNK.
+			 */
+			new_inc = (residual / HA_CHUNK) * HA_CHUNK;
+			if (residual % HA_CHUNK)
+				new_inc += HA_CHUNK;
+
+			has->end_pfn += new_inc;
+		}
+
+		/*
+		 * If the current start pfn is not where the covered_end
+		 * is, update it.
+		 */
+
+		if (has->covered_end_pfn != start_pfn) {
+			has->covered_end_pfn = start_pfn;
+			has->covered_start_pfn = start_pfn;
+		}
+		return true;
+
 	}
 
+	return false;
+}
+
+static unsigned long handle_pg_range(unsigned long pg_start,
+					unsigned long pg_count)
+{
+	unsigned long start_pfn = pg_start;
+	unsigned long pfn_cnt = pg_count;
+	unsigned long size;
+	struct list_head *cur;
+	struct hv_hotadd_state *has;
+	unsigned long pgs_ol = 0;
+	unsigned long old_covered_state;
+
+	if (list_empty(&dm_device.ha_region_list))
+		return 0;
+
+	list_for_each(cur, &dm_device.ha_region_list) {
+		has = list_entry(cur, struct hv_hotadd_state, list);
+
+		/*
+		 * If the pfn range we are dealing with is not in the current
+		 * "hot add block", move on.
+		 */
+		if ((start_pfn >= has->end_pfn))
+			continue;
+
+		old_covered_state = has->covered_end_pfn;
+
+		if (start_pfn < has->ha_end_pfn) {
+			/*
+			 * This is the case where we are backing pages
+			 * in an already hot added region. Bring
+			 * these pages online first.
+			 */
+			pgs_ol = has->ha_end_pfn - start_pfn;
+			if (pgs_ol > pfn_cnt)
+				pgs_ol = pfn_cnt;
+			hv_bring_pgs_online(start_pfn, pgs_ol);
+			has->covered_end_pfn +=  pgs_ol;
+			has->covered_start_pfn +=  pgs_ol;
+			pfn_cnt -= pgs_ol;
+		}
+
+		if ((has->ha_end_pfn < has->end_pfn) && (pfn_cnt > 0)) {
+			/*
+			 * We have some residual hot add range
+			 * that needs to be hot added; hot add
+			 * it now. Hot add a multiple of
+			 * of HA_CHUNK that fully covers the pages
+			 * we have.
+			 */
+			size = (has->end_pfn - has->ha_end_pfn);
+			if (pfn_cnt <= size) {
+				size = ((pfn_cnt / HA_CHUNK) * HA_CHUNK);
+				if (pfn_cnt % HA_CHUNK)
+					size += HA_CHUNK;
+			} else {
+				pfn_cnt = size;
+			}
+			hv_mem_hot_add(has->ha_end_pfn, size, pfn_cnt, has);
+		}
+		/*
+		 * If we managed to online any pages that were given to us,
+		 * we declare success.
+		 */
+		return has->covered_end_pfn - old_covered_state;
+
+	}
+
+	return 0;
+}
+
+static unsigned long process_hot_add(unsigned long pg_start,
+					unsigned long pfn_cnt,
+					unsigned long rg_start,
+					unsigned long rg_size)
+{
+	struct hv_hotadd_state *ha_region = NULL;
+
+	if (pfn_cnt == 0)
+		return 0;
+
+	if (!dm_device.host_specified_ha_region)
+		if (pfn_covered(pg_start, pfn_cnt))
+			goto do_pg_range;
+
+	/*
+	 * If the host has specified a hot-add range; deal with it first.
+	 */
+
+	if (rg_size != 0) {
+		ha_region = kzalloc(sizeof(struct hv_hotadd_state), GFP_KERNEL);
+		if (!ha_region)
+			return 0;
+
+		INIT_LIST_HEAD(&ha_region->list);
+
+		list_add_tail(&ha_region->list, &dm_device.ha_region_list);
+		ha_region->start_pfn = rg_start;
+		ha_region->ha_end_pfn = rg_start;
+		ha_region->covered_start_pfn = pg_start;
+		ha_region->covered_end_pfn = pg_start;
+		ha_region->end_pfn = rg_start + rg_size;
+	}
+
+do_pg_range:
+	/*
+	 * Process the page range specified; bringing them
+	 * online if possible.
+	 */
+	return handle_pg_range(pg_start, pfn_cnt);
+}
+
+#endif
+
+static void hot_add_req(struct work_struct *dummy)
+{
+	struct dm_hot_add_response resp;
+#ifdef CONFIG_MEMORY_HOTPLUG
+	unsigned long pg_start, pfn_cnt;
+	unsigned long rg_start, rg_sz;
+#endif
+	struct hv_dynmem_device *dm = &dm_device;
+
 	memset(&resp, 0, sizeof(struct dm_hot_add_response));
 	resp.hdr.type = DM_MEM_HOT_ADD_RESPONSE;
 	resp.hdr.size = sizeof(struct dm_hot_add_response);
 	resp.hdr.trans_id = atomic_inc_return(&trans_id);
 
-	resp.page_count = 0;
-	resp.result = 0;
+#ifdef CONFIG_MEMORY_HOTPLUG
+	pg_start = dm->ha_wrk.ha_page_range.finfo.start_page;
+	pfn_cnt = dm->ha_wrk.ha_page_range.finfo.page_cnt;
+
+	rg_start = dm->ha_wrk.ha_region_range.finfo.start_page;
+	rg_sz = dm->ha_wrk.ha_region_range.finfo.page_cnt;
+
+	if ((rg_start == 0) && (!dm->host_specified_ha_region)) {
+		unsigned long region_size;
+		unsigned long region_start;
+
+		/*
+		 * The host has not specified the hot-add region.
+		 * Based on the hot-add page range being specified,
+		 * compute a hot-add region that can cover the pages
+		 * that need to be hot-added while ensuring the alignment
+		 * and size requirements of Linux as it relates to hot-add.
+		 */
+		region_start = pg_start;
+		region_size = (pfn_cnt / HA_CHUNK) * HA_CHUNK;
+		if (pfn_cnt % HA_CHUNK)
+			region_size += HA_CHUNK;
+
+		region_start = (pg_start / HA_CHUNK) * HA_CHUNK;
+
+		rg_start = region_start;
+		rg_sz = region_size;
+	}
+
+	if (do_hot_add)
+		resp.page_count = process_hot_add(pg_start, pfn_cnt,
+						rg_start, rg_sz);
+#endif
+	/*
+	 * The result field of the response structure has the
+	 * following semantics:
+	 *
+	 * 1. If all or some pages hot-added: Guest should return success.
+	 *
+	 * 2. If no pages could be hot-added:
+	 *
+	 * If the guest returns success, then the host
+	 * will not attempt any further hot-add operations. This
+	 * signifies a permanent failure.
+	 *
+	 * If the guest returns failure, then this failure will be
+	 * treated as a transient failure and the host may retry the
+	 * hot-add operation after some delay.
+	 */
+	if (resp.page_count > 0)
+		resp.result = 1;
+	else if (!do_hot_add)
+		resp.result = 1;
+	else
+		resp.result = 0;
+
+	if (!do_hot_add || (resp.page_count == 0))
+		pr_info("Memory hot add failed\n");
 
 	dm->state = DM_INITIALIZED;
 	vmbus_sendpacket(dm->dev->channel, &resp,
 			sizeof(struct dm_hot_add_response),
 			(unsigned long)NULL,
 			VM_PKT_DATA_INBAND, 0);
-
 }
 
 static void process_info(struct hv_dynmem_device *dm, struct dm_info_msg *msg)
@@ -523,7 +912,7 @@ static void process_info(struct hv_dynmem_device *dm, struct dm_info_msg *msg)
 	}
 }
 
-unsigned long compute_balloon_floor(void)
+static unsigned long compute_balloon_floor(void)
 {
 	unsigned long min_pages;
 #define MB2PAGES(mb) ((mb) << (20 - PAGE_SHIFT))
@@ -644,6 +1033,14 @@ static int  alloc_balloon_pages(struct hv_dynmem_device *dm, int num_pages,
 
 		dm->num_pages_ballooned += alloc_unit;
 
+		/*
+		 * If we allocatted 2M pages; split them so we
+		 * can free them in any order we get.
+		 */
+
+		if (alloc_unit != 1)
+			split_page(pg, get_order(alloc_unit << PAGE_SHIFT));
+
 		bl_resp->range_count++;
 		bl_resp->range_array[i].finfo.start_page =
 			page_to_pfn(pg);
@@ -657,9 +1054,9 @@ static int  alloc_balloon_pages(struct hv_dynmem_device *dm, int num_pages,
 
 
 
-static void balloon_up(struct hv_dynmem_device *dm, struct dm_balloon *req)
+static void balloon_up(struct work_struct *dummy)
 {
-	int num_pages = req->num_pages;
+	int num_pages = dm_device.balloon_wrk.num_pages;
 	int num_ballooned = 0;
 	struct dm_balloon_response *bl_resp;
 	int alloc_unit;
@@ -670,9 +1067,10 @@ static void balloon_up(struct hv_dynmem_device *dm, struct dm_balloon *req)
 
 
 	/*
-	 * Currently, we only support 4k allocations.
+	 * We will attempt 2M allocations. However, if we fail to
+	 * allocate 2M chunks, we will go back to 4k allocations.
 	 */
-	alloc_unit = 1;
+	alloc_unit = 512;
 
 	while (!done) {
 		bl_resp = (struct dm_balloon_response *)send_buffer;
@@ -684,14 +1082,19 @@ static void balloon_up(struct hv_dynmem_device *dm, struct dm_balloon *req)
 
 
 		num_pages -= num_ballooned;
-		num_ballooned = alloc_balloon_pages(dm, num_pages,
+		num_ballooned = alloc_balloon_pages(&dm_device, num_pages,
 						bl_resp, alloc_unit,
 						 &alloc_error);
 
+		if ((alloc_error) && (alloc_unit != 1)) {
+			alloc_unit = 1;
+			continue;
+		}
+
 		if ((alloc_error) || (num_ballooned == num_pages)) {
 			bl_resp->more_pages = 0;
 			done = true;
-			dm->state = DM_INITIALIZED;
+			dm_device.state = DM_INITIALIZED;
 		}
 
 		/*
@@ -719,7 +1122,7 @@ static void balloon_up(struct hv_dynmem_device *dm, struct dm_balloon *req)
 			pr_info("Balloon response failed\n");
 
 			for (i = 0; i < bl_resp->range_count; i++)
-				free_balloon_pages(dm,
+				free_balloon_pages(&dm_device,
 						 &bl_resp->range_array[i]);
 
 			done = true;
@@ -761,7 +1164,6 @@ static int dm_thread_func(void *dm_dev)
 {
 	struct hv_dynmem_device *dm = dm_dev;
 	int t;
-	unsigned long  scan_start;
 
 	while (!kthread_should_stop()) {
 		t = wait_for_completion_timeout(&dm_device.config_event, 1*HZ);
@@ -773,22 +1175,6 @@ static int dm_thread_func(void *dm_dev)
 		if (t == 0)
 			post_status(dm);
 
-		scan_start = jiffies;
-		switch (dm->state) {
-		case DM_BALLOON_UP:
-			balloon_up(dm, (struct dm_balloon *)recv_buffer);
-			break;
-
-		case DM_HOT_ADD:
-			hot_add_req(dm, (struct dm_hot_add *)recv_buffer);
-			break;
-		default:
-			break;
-		}
-
-		if (!time_in_range(jiffies, scan_start, scan_start + HZ))
-			post_status(dm);
-
 	}
 
 	return 0;
@@ -861,6 +1247,10 @@ static void balloon_onchannelcallback(void *context)
 	struct dm_message *dm_msg;
 	struct dm_header *dm_hdr;
 	struct hv_dynmem_device *dm = hv_get_drvdata(dev);
+	struct dm_balloon *bal_msg;
+	struct dm_hot_add *ha_msg;
+	union dm_mem_page_range *ha_pg_range;
+	union dm_mem_page_range *ha_region;
 
 	memset(recv_buffer, 0, sizeof(recv_buffer));
 	vmbus_recvpacket(dev->channel, recv_buffer,
@@ -882,8 +1272,12 @@ static void balloon_onchannelcallback(void *context)
 			break;
 
 		case DM_BALLOON_REQUEST:
+			if (dm->state == DM_BALLOON_UP)
+				pr_warn("Currently ballooning\n");
+			bal_msg = (struct dm_balloon *)recv_buffer;
 			dm->state = DM_BALLOON_UP;
-			complete(&dm->config_event);
+			dm_device.balloon_wrk.num_pages = bal_msg->num_pages;
+			schedule_work(&dm_device.balloon_wrk.wrk);
 			break;
 
 		case DM_UNBALLOON_REQUEST:
@@ -893,8 +1287,31 @@ static void balloon_onchannelcallback(void *context)
 			break;
 
 		case DM_MEM_HOT_ADD_REQUEST:
+			if (dm->state == DM_HOT_ADD)
+				pr_warn("Currently hot-adding\n");
 			dm->state = DM_HOT_ADD;
-			complete(&dm->config_event);
+			ha_msg = (struct dm_hot_add *)recv_buffer;
+			if (ha_msg->hdr.size == sizeof(struct dm_hot_add)) {
+				/*
+				 * This is a normal hot-add request specifying
+				 * hot-add memory.
+				 */
+				ha_pg_range = &ha_msg->range;
+				dm->ha_wrk.ha_page_range = *ha_pg_range;
+				dm->ha_wrk.ha_region_range.page_range = 0;
+			} else {
+				/*
+				 * Host is specifying that we first hot-add
+				 * a region and then partially populate this
+				 * region.
+				 */
+				dm->host_specified_ha_region = true;
+				ha_pg_range = &ha_msg->range;
+				ha_region = &ha_pg_range[1];
+				dm->ha_wrk.ha_page_range = *ha_pg_range;
+				dm->ha_wrk.ha_region_range = *ha_region;
+			}
+			schedule_work(&dm_device.ha_wrk.wrk);
 			break;
 
 		case DM_INFO_MESSAGE:
@@ -937,6 +1354,10 @@ static int balloon_probe(struct hv_device *dev,
 	dm_device.next_version = DYNMEM_PROTOCOL_VERSION_WIN7;
 	init_completion(&dm_device.host_event);
 	init_completion(&dm_device.config_event);
+	INIT_LIST_HEAD(&dm_device.ha_region_list);
+	INIT_WORK(&dm_device.balloon_wrk.wrk, balloon_up);
+	INIT_WORK(&dm_device.ha_wrk.wrk, hot_add_req);
+	dm_device.host_specified_ha_region = false;
 
 	dm_device.thread =
 		 kthread_run(dm_thread_func, &dm_device, "hv_balloon");
@@ -945,6 +1366,10 @@ static int balloon_probe(struct hv_device *dev,
 		goto probe_error1;
 	}
 
+#ifdef CONFIG_MEMORY_HOTPLUG
+	set_online_page_callback(&hv_online_page);
+#endif
+
 	hv_set_drvdata(dev, &dm_device);
 	/*
 	 * Initiate the hand shake with the host and negotiate
@@ -962,8 +1387,7 @@ static int balloon_probe(struct hv_device *dev,
 	ret = vmbus_sendpacket(dev->channel, &version_req,
 				sizeof(struct dm_version_request),
 				(unsigned long)NULL,
-				VM_PKT_DATA_INBAND,
-				VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED);
+				VM_PKT_DATA_INBAND, 0);
 	if (ret)
 		goto probe_error2;
 
@@ -990,13 +1414,13 @@ static int balloon_probe(struct hv_device *dev,
 	cap_msg.hdr.trans_id = atomic_inc_return(&trans_id);
 
 	cap_msg.caps.cap_bits.balloon = 1;
+	cap_msg.caps.cap_bits.hot_add = 1;
+
 	/*
-	 * While we currently don't support hot-add,
-	 * we still advertise this capability since the
-	 * host requires that guests partcipating in the
-	 * dynamic memory protocol support hot add.
+	 * Specify our alignment requirements as it relates
+	 * memory hot-add. Specify 128MB alignment.
 	 */
-	cap_msg.caps.cap_bits.hot_add = 1;
+	cap_msg.caps.cap_bits.hot_add_alignment = 7;
 
 	/*
 	 * Currently the host does not use these
@@ -1009,8 +1433,7 @@ static int balloon_probe(struct hv_device *dev,
 	ret = vmbus_sendpacket(dev->channel, &cap_msg,
 				sizeof(struct dm_capabilities),
 				(unsigned long)NULL,
-				VM_PKT_DATA_INBAND,
-				VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED);
+				VM_PKT_DATA_INBAND, 0);
 	if (ret)
 		goto probe_error2;
 
@@ -1034,6 +1457,9 @@ static int balloon_probe(struct hv_device *dev,
 	return 0;
 
 probe_error2:
+#ifdef CONFIG_MEMORY_HOTPLUG
+	restore_online_page_callback(&hv_online_page);
+#endif
 	kthread_stop(dm_device.thread);
 
 probe_error1:
@@ -1046,13 +1472,26 @@ probe_error0:
 static int balloon_remove(struct hv_device *dev)
 {
 	struct hv_dynmem_device *dm = hv_get_drvdata(dev);
+	struct list_head *cur, *tmp;
+	struct hv_hotadd_state *has;
 
 	if (dm->num_pages_ballooned != 0)
 		pr_warn("Ballooned pages: %d\n", dm->num_pages_ballooned);
 
+	cancel_work_sync(&dm->balloon_wrk.wrk);
+	cancel_work_sync(&dm->ha_wrk.wrk);
+
 	vmbus_close(dev->channel);
 	kthread_stop(dm->thread);
 	kfree(send_buffer);
+#ifdef CONFIG_MEMORY_HOTPLUG
+	restore_online_page_callback(&hv_online_page);
+#endif
+	list_for_each_safe(cur, tmp, &dm->ha_region_list) {
+		has = list_entry(cur, struct hv_hotadd_state, list);
+		list_del(&has->list);
+		kfree(has);
+	}
 
 	return 0;
 }
@@ -1079,14 +1518,7 @@ static int __init init_balloon_drv(void)
 	return vmbus_driver_register(&balloon_drv);
 }
 
-static void exit_balloon_drv(void)
-{
-
-	vmbus_driver_unregister(&balloon_drv);
-}
-
 module_init(init_balloon_drv);
-module_exit(exit_balloon_drv);
 
 MODULE_DESCRIPTION("Hyper-V Balloon");
 MODULE_VERSION(HV_DRV_VERSION);
diff --git a/drivers/hv/hv_snapshot.c b/drivers/hv/hv_snapshot.c
new file mode 100644
index 000000000000..8ad5653ce447
--- /dev/null
+++ b/drivers/hv/hv_snapshot.c
@@ -0,0 +1,287 @@
+/*
+ * An implementation of host initiated guest snapshot.
+ *
+ *
+ * Copyright (C) 2013, Microsoft, Inc.
+ * Author : K. Y. Srinivasan <kys@microsoft.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ * NON INFRINGEMENT.  See the GNU General Public License for more
+ * details.
+ *
+ */
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/net.h>
+#include <linux/nls.h>
+#include <linux/connector.h>
+#include <linux/workqueue.h>
+#include <linux/hyperv.h>
+
+
+
+/*
+ * Global state maintained for transaction that is being processed.
+ * Note that only one transaction can be active at any point in time.
+ *
+ * This state is set when we receive a request from the host; we
+ * cleanup this state when the transaction is completed - when we respond
+ * to the host with the key value.
+ */
+
+static struct {
+	bool active; /* transaction status - active or not */
+	int recv_len; /* number of bytes received. */
+	struct vmbus_channel *recv_channel; /* chn we got the request */
+	u64 recv_req_id; /* request ID. */
+	struct hv_vss_msg  *msg; /* current message */
+} vss_transaction;
+
+
+static void vss_respond_to_host(int error);
+
+static struct cb_id vss_id = { CN_VSS_IDX, CN_VSS_VAL };
+static const char vss_name[] = "vss_kernel_module";
+static __u8 *recv_buffer;
+
+static void vss_send_op(struct work_struct *dummy);
+static DECLARE_WORK(vss_send_op_work, vss_send_op);
+
+/*
+ * Callback when data is received from user mode.
+ */
+
+static void
+vss_cn_callback(struct cn_msg *msg, struct netlink_skb_parms *nsp)
+{
+	struct hv_vss_msg *vss_msg;
+
+	vss_msg = (struct hv_vss_msg *)msg->data;
+
+	if (vss_msg->vss_hdr.operation == VSS_OP_REGISTER) {
+		pr_info("VSS daemon registered\n");
+		vss_transaction.active = false;
+		if (vss_transaction.recv_channel != NULL)
+			hv_vss_onchannelcallback(vss_transaction.recv_channel);
+		return;
+
+	}
+	vss_respond_to_host(vss_msg->error);
+}
+
+
+static void vss_send_op(struct work_struct *dummy)
+{
+	int op = vss_transaction.msg->vss_hdr.operation;
+	struct cn_msg *msg;
+	struct hv_vss_msg *vss_msg;
+
+	msg = kzalloc(sizeof(*msg) + sizeof(*vss_msg), GFP_ATOMIC);
+	if (!msg)
+		return;
+
+	vss_msg = (struct hv_vss_msg *)msg->data;
+
+	msg->id.idx =  CN_VSS_IDX;
+	msg->id.val = CN_VSS_VAL;
+
+	vss_msg->vss_hdr.operation = op;
+	msg->len = sizeof(struct hv_vss_msg);
+
+	cn_netlink_send(msg, 0, GFP_ATOMIC);
+	kfree(msg);
+
+	return;
+}
+
+/*
+ * Send a response back to the host.
+ */
+
+static void
+vss_respond_to_host(int error)
+{
+	struct icmsg_hdr *icmsghdrp;
+	u32	buf_len;
+	struct vmbus_channel *channel;
+	u64	req_id;
+
+	/*
+	 * If a transaction is not active; log and return.
+	 */
+
+	if (!vss_transaction.active) {
+		/*
+		 * This is a spurious call!
+		 */
+		pr_warn("VSS: Transaction not active\n");
+		return;
+	}
+	/*
+	 * Copy the global state for completing the transaction. Note that
+	 * only one transaction can be active at a time.
+	 */
+
+	buf_len = vss_transaction.recv_len;
+	channel = vss_transaction.recv_channel;
+	req_id = vss_transaction.recv_req_id;
+	vss_transaction.active = false;
+
+	icmsghdrp = (struct icmsg_hdr *)
+			&recv_buffer[sizeof(struct vmbuspipe_hdr)];
+
+	if (channel->onchannel_callback == NULL)
+		/*
+		 * We have raced with util driver being unloaded;
+		 * silently return.
+		 */
+		return;
+
+	icmsghdrp->status = error;
+
+	icmsghdrp->icflags = ICMSGHDRFLAG_TRANSACTION | ICMSGHDRFLAG_RESPONSE;
+
+	vmbus_sendpacket(channel, recv_buffer, buf_len, req_id,
+				VM_PKT_DATA_INBAND, 0);
+
+}
+
+/*
+ * This callback is invoked when we get a VSS message from the host.
+ * The host ensures that only one VSS transaction can be active at a time.
+ */
+
+void hv_vss_onchannelcallback(void *context)
+{
+	struct vmbus_channel *channel = context;
+	u32 recvlen;
+	u64 requestid;
+	struct hv_vss_msg *vss_msg;
+
+
+	struct icmsg_hdr *icmsghdrp;
+	struct icmsg_negotiate *negop = NULL;
+
+	if (vss_transaction.active) {
+		/*
+		 * We will defer processing this callback once
+		 * the current transaction is complete.
+		 */
+		vss_transaction.recv_channel = channel;
+		return;
+	}
+
+	vmbus_recvpacket(channel, recv_buffer, PAGE_SIZE * 2, &recvlen,
+			 &requestid);
+
+	if (recvlen > 0) {
+		icmsghdrp = (struct icmsg_hdr *)&recv_buffer[
+			sizeof(struct vmbuspipe_hdr)];
+
+		if (icmsghdrp->icmsgtype == ICMSGTYPE_NEGOTIATE) {
+			vmbus_prep_negotiate_resp(icmsghdrp, negop,
+				 recv_buffer, MAX_SRV_VER, MAX_SRV_VER);
+			/*
+			 * We currently negotiate the highest number the
+			 * host has presented. If this version is not
+			 * atleast 5.0, reject.
+			 */
+			negop = (struct icmsg_negotiate *)&recv_buffer[
+				sizeof(struct vmbuspipe_hdr) +
+				sizeof(struct icmsg_hdr)];
+
+			if (negop->icversion_data[1].major < 5)
+				negop->icframe_vercnt = 0;
+		} else {
+			vss_msg = (struct hv_vss_msg *)&recv_buffer[
+				sizeof(struct vmbuspipe_hdr) +
+				sizeof(struct icmsg_hdr)];
+
+			/*
+			 * Stash away this global state for completing the
+			 * transaction; note transactions are serialized.
+			 */
+
+			vss_transaction.recv_len = recvlen;
+			vss_transaction.recv_channel = channel;
+			vss_transaction.recv_req_id = requestid;
+			vss_transaction.active = true;
+			vss_transaction.msg = (struct hv_vss_msg *)vss_msg;
+
+			switch (vss_msg->vss_hdr.operation) {
+				/*
+				 * Initiate a "freeze/thaw"
+				 * operation in the guest.
+				 * We respond to the host once
+				 * the operation is complete.
+				 *
+				 * We send the message to the
+				 * user space daemon and the
+				 * operation is performed in
+				 * the daemon.
+				 */
+			case VSS_OP_FREEZE:
+			case VSS_OP_THAW:
+				schedule_work(&vss_send_op_work);
+				return;
+
+			case VSS_OP_HOT_BACKUP:
+				vss_msg->vss_cf.flags =
+					 VSS_HBU_NO_AUTO_RECOVERY;
+				vss_respond_to_host(0);
+				return;
+
+			case VSS_OP_GET_DM_INFO:
+				vss_msg->dm_info.flags = 0;
+				vss_respond_to_host(0);
+				return;
+
+			default:
+				vss_respond_to_host(0);
+				return;
+
+			}
+
+		}
+
+		icmsghdrp->icflags = ICMSGHDRFLAG_TRANSACTION
+			| ICMSGHDRFLAG_RESPONSE;
+
+		vmbus_sendpacket(channel, recv_buffer,
+				       recvlen, requestid,
+				       VM_PKT_DATA_INBAND, 0);
+	}
+
+}
+
+int
+hv_vss_init(struct hv_util_service *srv)
+{
+	int err;
+
+	err = cn_add_callback(&vss_id, vss_name, vss_cn_callback);
+	if (err)
+		return err;
+	recv_buffer = srv->recv_buffer;
+
+	/*
+	 * When this driver loads, the user level daemon that
+	 * processes the host requests may not yet be running.
+	 * Defer processing channel callbacks until the daemon
+	 * has registered.
+	 */
+	vss_transaction.active = true;
+	return 0;
+}
+
+void hv_vss_deinit(void)
+{
+	cn_del_callback(&vss_id);
+	cancel_work_sync(&vss_send_op_work);
+}
diff --git a/drivers/hv/hv_util.c b/drivers/hv/hv_util.c
index 1d4cbd8e8261..2f561c5dfe24 100644
--- a/drivers/hv/hv_util.c
+++ b/drivers/hv/hv_util.c
@@ -49,6 +49,12 @@ static struct hv_util_service util_kvp = {
 	.util_deinit = hv_kvp_deinit,
 };
 
+static struct hv_util_service util_vss = {
+	.util_cb = hv_vss_onchannelcallback,
+	.util_init = hv_vss_init,
+	.util_deinit = hv_vss_deinit,
+};
+
 static void perform_shutdown(struct work_struct *dummy)
 {
 	orderly_poweroff(true);
@@ -339,6 +345,10 @@ static const struct hv_vmbus_device_id id_table[] = {
 	{ HV_KVP_GUID,
 	  .driver_data = (unsigned long)&util_kvp
 	},
+	/* VSS GUID */
+	{ HV_VSS_GUID,
+	  .driver_data = (unsigned long)&util_vss
+	},
 	{ },
 };
 
diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c
index cafa72ffdc30..d6fbb5772b8d 100644
--- a/drivers/hv/ring_buffer.c
+++ b/drivers/hv/ring_buffer.c
@@ -71,6 +71,7 @@ u32 hv_end_read(struct hv_ring_buffer_info *rbi)
 
 static bool hv_need_to_signal(u32 old_write, struct hv_ring_buffer_info *rbi)
 {
+	smp_mb();
 	if (rbi->ring_buffer->interrupt_mask)
 		return false;