Merge tag 'hyperv-next-signed-20210629' of...

Merge tag 'hyperv-next-signed-20210629' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux

Pull hyperv updates from Wei Liu:
 "Just a few minor enhancement patches and bug fixes"

* tag 'hyperv-next-signed-20210629' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux:
  PCI: hv: Add check for hyperv_initialized in init_hv_pci_drv()
  Drivers: hv: Move Hyper-V extended capability check to arch neutral code
  drivers: hv: Fix missing error code in vmbus_connect()
  x86/hyperv: fix logical processor creation
  hv_utils: Fix passing zero to 'PTR_ERR' warning
  scsi: storvsc: Use blk_mq_unique_tag() to generate requestIDs
  Drivers: hv: vmbus: Copy packets sent by Hyper-V out of the ring buffer
  hv_balloon: Remove redundant assignment to region_start

arch/x86/hyperv/hv_init.c

@@ -614,50 +614,3 @@ bool hv_is_isolation_supported(void)
 	return hv_get_isolation_type() != HV_ISOLATION_TYPE_NONE;
 }
 EXPORT_SYMBOL_GPL(hv_is_isolation_supported);
-
-/* Bit mask of the extended capability to query: see HV_EXT_CAPABILITY_xxx */
-bool hv_query_ext_cap(u64 cap_query)
-{
-	/*
-	 * The address of the 'hv_extended_cap' variable will be used as an
-	 * output parameter to the hypercall below and so it should be
-	 * compatible with 'virt_to_phys'. Which means, it's address should be
-	 * directly mapped. Use 'static' to keep it compatible; stack variables
-	 * can be virtually mapped, making them incompatible with
-	 * 'virt_to_phys'.
-	 * Hypercall input/output addresses should also be 8-byte aligned.
-	 */
-	static u64 hv_extended_cap __aligned(8);
-	static bool hv_extended_cap_queried;
-	u64 status;
-
-	/*
-	 * Querying extended capabilities is an extended hypercall. Check if the
-	 * partition supports extended hypercall, first.
-	 */
-	if (!(ms_hyperv.priv_high & HV_ENABLE_EXTENDED_HYPERCALLS))
-		return false;
-
-	/* Extended capabilities do not change at runtime. */
-	if (hv_extended_cap_queried)
-		return hv_extended_cap & cap_query;
-
-	status = hv_do_hypercall(HV_EXT_CALL_QUERY_CAPABILITIES, NULL,
-				 &hv_extended_cap);
-
-	/*
-	 * The query extended capabilities hypercall should not fail under
-	 * any normal circumstances. Avoid repeatedly making the hypercall, on
-	 * error.
-	 */
-	hv_extended_cap_queried = true;
-	status &= HV_HYPERCALL_RESULT_MASK;
-	if (status != HV_STATUS_SUCCESS) {
-		pr_err("Hyper-V: Extended query capabilities hypercall failed 0x%llx\n",
-		       status);
-		return false;
-	}
-
-	return hv_extended_cap & cap_query;
-}
-EXPORT_SYMBOL_GPL(hv_query_ext_cap);

arch/x86/kernel/cpu/mshyperv.c

@@ -236,7 +236,7 @@ static void __init hv_smp_prepare_cpus(unsigned int max_cpus)
 	for_each_present_cpu(i) {
 		if (i == 0)
 			continue;
-		ret = hv_call_add_logical_proc(numa_cpu_node(i), i, cpu_physical_id(i));
+		ret = hv_call_add_logical_proc(numa_cpu_node(i), i, i);
 		BUG_ON(ret);
 	}
 

drivers/Makefile

@@ -160,7 +160,7 @@ obj-$(CONFIG_SOUNDWIRE)		+= soundwire/
 
 # Virtualization drivers
 obj-$(CONFIG_VIRT_DRIVERS)	+= virt/
-obj-$(CONFIG_HYPERV)		+= hv/
+obj-$(subst m,y,$(CONFIG_HYPERV))	+= hv/
 
 obj-$(CONFIG_PM_DEVFREQ)	+= devfreq/
 obj-$(CONFIG_EXTCON)		+= extcon/

drivers/hv/Makefile

@@ -11,3 +11,6 @@ hv_vmbus-y := vmbus_drv.o \
 		 channel_mgmt.o ring_buffer.o hv_trace.o
 hv_vmbus-$(CONFIG_HYPERV_TESTING)	+= hv_debugfs.o
 hv_utils-y := hv_util.o hv_kvp.o hv_snapshot.o hv_fcopy.o hv_utils_transport.o
+
+# Code that must be built-in
+obj-$(subst m,y,$(CONFIG_HYPERV)) += hv_common.o

drivers/hv/channel.c

@@ -662,12 +662,15 @@ static int __vmbus_open(struct vmbus_channel *newchannel,
 	newchannel->onchannel_callback = onchannelcallback;
 	newchannel->channel_callback_context = context;
 
-	err = hv_ringbuffer_init(&newchannel->outbound, page, send_pages);
+	if (!newchannel->max_pkt_size)
+		newchannel->max_pkt_size = VMBUS_DEFAULT_MAX_PKT_SIZE;
+
+	err = hv_ringbuffer_init(&newchannel->outbound, page, send_pages, 0);
 	if (err)
 		goto error_clean_ring;
 
-	err = hv_ringbuffer_init(&newchannel->inbound,
-				 &page[send_pages], recv_pages);
+	err = hv_ringbuffer_init(&newchannel->inbound, &page[send_pages],
+				 recv_pages, newchannel->max_pkt_size);
 	if (err)
 		goto error_clean_ring;
 
@@ -1186,15 +1189,14 @@ EXPORT_SYMBOL_GPL(vmbus_recvpacket_raw);
  * vmbus_next_request_id - Returns a new request id. It is also
  * the index at which the guest memory address is stored.
  * Uses a spin lock to avoid race conditions.
- * @rqstor: Pointer to the requestor struct
+ * @channel: Pointer to the VMbus channel struct
  * @rqst_add: Guest memory address to be stored in the array
  */
-u64 vmbus_next_request_id(struct vmbus_requestor *rqstor, u64 rqst_addr)
+u64 vmbus_next_request_id(struct vmbus_channel *channel, u64 rqst_addr)
 {
+	struct vmbus_requestor *rqstor = &channel->requestor;
 	unsigned long flags;
 	u64 current_id;
-	const struct vmbus_channel *channel =
-		container_of(rqstor, const struct vmbus_channel, requestor);
 
 	/* Check rqstor has been initialized */
 	if (!channel->rqstor_size)
@@ -1228,16 +1230,15 @@ EXPORT_SYMBOL_GPL(vmbus_next_request_id);
 /*
  * vmbus_request_addr - Returns the memory address stored at @trans_id
  * in @rqstor. Uses a spin lock to avoid race conditions.
- * @rqstor: Pointer to the requestor struct
+ * @channel: Pointer to the VMbus channel struct
  * @trans_id: Request id sent back from Hyper-V. Becomes the requestor's
  * next request id.
  */
-u64 vmbus_request_addr(struct vmbus_requestor *rqstor, u64 trans_id)
+u64 vmbus_request_addr(struct vmbus_channel *channel, u64 trans_id)
 {
+	struct vmbus_requestor *rqstor = &channel->requestor;
 	unsigned long flags;
 	u64 req_addr;
-	const struct vmbus_channel *channel =
-		container_of(rqstor, const struct vmbus_channel, requestor);
 
 	/* Check rqstor has been initialized */
 	if (!channel->rqstor_size)

drivers/hv/connection.c

@@ -232,8 +232,10 @@ int vmbus_connect(void)
 	 */
 
 	for (i = 0; ; i++) {
-		if (i == ARRAY_SIZE(vmbus_versions))
+		if (i == ARRAY_SIZE(vmbus_versions)) {
+			ret = -EDOM;
 			goto cleanup;
+		}
 
 		version = vmbus_versions[i];
 		if (version > max_version)

drivers/hv/hv_balloon.c

@@ -1010,7 +1010,6 @@ static void hot_add_req(struct work_struct *dummy)
 		 * 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;

drivers/hv/hv_common.c

+// SPDX-License-Identifier: GPL-2.0
+
+/*
+ * Architecture neutral utility routines for interacting with
+ * Hyper-V. This file is specifically for code that must be
+ * built-in to the kernel image when CONFIG_HYPERV is set
+ * (vs. being in a module) because it is called from architecture
+ * specific code under arch/.
+ *
+ * Copyright (C) 2021, Microsoft, Inc.
+ *
+ * Author : Michael Kelley <mikelley@microsoft.com>
+ */
+
+#include <linux/types.h>
+#include <linux/export.h>
+#include <linux/bitfield.h>
+#include <asm/hyperv-tlfs.h>
+#include <asm/mshyperv.h>
+
+
+/* Bit mask of the extended capability to query: see HV_EXT_CAPABILITY_xxx */
+bool hv_query_ext_cap(u64 cap_query)
+{
+	/*
+	 * The address of the 'hv_extended_cap' variable will be used as an
+	 * output parameter to the hypercall below and so it should be
+	 * compatible with 'virt_to_phys'. Which means, it's address should be
+	 * directly mapped. Use 'static' to keep it compatible; stack variables
+	 * can be virtually mapped, making them incompatible with
+	 * 'virt_to_phys'.
+	 * Hypercall input/output addresses should also be 8-byte aligned.
+	 */
+	static u64 hv_extended_cap __aligned(8);
+	static bool hv_extended_cap_queried;
+	u64 status;
+
+	/*
+	 * Querying extended capabilities is an extended hypercall. Check if the
+	 * partition supports extended hypercall, first.
+	 */
+	if (!(ms_hyperv.priv_high & HV_ENABLE_EXTENDED_HYPERCALLS))
+		return false;
+
+	/* Extended capabilities do not change at runtime. */
+	if (hv_extended_cap_queried)
+		return hv_extended_cap & cap_query;
+
+	status = hv_do_hypercall(HV_EXT_CALL_QUERY_CAPABILITIES, NULL,
+				 &hv_extended_cap);
+
+	/*
+	 * The query extended capabilities hypercall should not fail under
+	 * any normal circumstances. Avoid repeatedly making the hypercall, on
+	 * error.
+	 */
+	hv_extended_cap_queried = true;
+	if (!hv_result_success(status)) {
+		pr_err("Hyper-V: Extended query capabilities hypercall failed 0x%llx\n",
+		       status);
+		return false;
+	}
+
+	return hv_extended_cap & cap_query;
+}
+EXPORT_SYMBOL_GPL(hv_query_ext_cap);

drivers/hv/hv_fcopy.c

@@ -349,6 +349,7 @@ int hv_fcopy_init(struct hv_util_service *srv)
 {
 	recv_buffer = srv->recv_buffer;
 	fcopy_transaction.recv_channel = srv->channel;
+	fcopy_transaction.recv_channel->max_pkt_size = HV_HYP_PAGE_SIZE * 2;
 
 	/*
 	 * When this driver loads, the user level daemon that

drivers/hv/hv_kvp.c

@@ -757,6 +757,7 @@ hv_kvp_init(struct hv_util_service *srv)
 {
 	recv_buffer = srv->recv_buffer;
 	kvp_transaction.recv_channel = srv->channel;
+	kvp_transaction.recv_channel->max_pkt_size = HV_HYP_PAGE_SIZE * 4;
 
 	/*
 	 * When this driver loads, the user level daemon that

drivers/hv/hv_util.c

@@ -750,8 +750,8 @@ static int hv_timesync_init(struct hv_util_service *srv)
 	 */
 	hv_ptp_clock = ptp_clock_register(&ptp_hyperv_info, NULL);
 	if (IS_ERR_OR_NULL(hv_ptp_clock)) {
-		pr_err("cannot register PTP clock: %ld\n",
-		       PTR_ERR(hv_ptp_clock));
+		pr_err("cannot register PTP clock: %d\n",
+		       PTR_ERR_OR_ZERO(hv_ptp_clock));
 		hv_ptp_clock = NULL;
 	}
 

drivers/hv/hyperv_vmbus.h

@@ -174,7 +174,7 @@ extern int hv_synic_cleanup(unsigned int cpu);
 void hv_ringbuffer_pre_init(struct vmbus_channel *channel);
 
 int hv_ringbuffer_init(struct hv_ring_buffer_info *ring_info,
-		       struct page *pages, u32 pagecnt);
+		       struct page *pages, u32 pagecnt, u32 max_pkt_size);
 
 void hv_ringbuffer_cleanup(struct hv_ring_buffer_info *ring_info);
 

drivers/hv/ring_buffer.c

@@ -181,7 +181,7 @@ void hv_ringbuffer_pre_init(struct vmbus_channel *channel)
 
 /* Initialize the ring buffer. */
 int hv_ringbuffer_init(struct hv_ring_buffer_info *ring_info,
-		       struct page *pages, u32 page_cnt)
+		       struct page *pages, u32 page_cnt, u32 max_pkt_size)
 {
 	int i;
 	struct page **pages_wraparound;
@@ -223,6 +223,14 @@ int hv_ringbuffer_init(struct hv_ring_buffer_info *ring_info,
 		sizeof(struct hv_ring_buffer);
 	ring_info->priv_read_index = 0;
 
+	/* Initialize buffer that holds copies of incoming packets */
+	if (max_pkt_size) {
+		ring_info->pkt_buffer = kzalloc(max_pkt_size, GFP_KERNEL);
+		if (!ring_info->pkt_buffer)
+			return -ENOMEM;
+		ring_info->pkt_buffer_size = max_pkt_size;
+	}
+
 	spin_lock_init(&ring_info->ring_lock);
 
 	return 0;
@@ -235,6 +243,9 @@ void hv_ringbuffer_cleanup(struct hv_ring_buffer_info *ring_info)
 	vunmap(ring_info->ring_buffer);
 	ring_info->ring_buffer = NULL;
 	mutex_unlock(&ring_info->ring_buffer_mutex);
+
+	kfree(ring_info->pkt_buffer);
+	ring_info->pkt_buffer_size = 0;
 }
 
 /* Write to the ring buffer. */
@@ -301,10 +312,12 @@ int hv_ringbuffer_write(struct vmbus_channel *channel,
 	 */
 
 	if (desc->flags == VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED) {
-		rqst_id = vmbus_next_request_id(&channel->requestor, requestid);
-		if (rqst_id == VMBUS_RQST_ERROR) {
-			spin_unlock_irqrestore(&outring_info->ring_lock, flags);
-			return -EAGAIN;
+		if (channel->next_request_id_callback != NULL) {
+			rqst_id = channel->next_request_id_callback(channel, requestid);
+			if (rqst_id == VMBUS_RQST_ERROR) {
+				spin_unlock_irqrestore(&outring_info->ring_lock, flags);
+				return -EAGAIN;
+			}
 		}
 	}
 	desc = hv_get_ring_buffer(outring_info) + old_write;
@@ -332,7 +345,8 @@ int hv_ringbuffer_write(struct vmbus_channel *channel,
 	if (channel->rescind) {
 		if (rqst_id != VMBUS_NO_RQSTOR) {
 			/* Reclaim request ID to avoid leak of IDs */
-			vmbus_request_addr(&channel->requestor, rqst_id);
+			if (channel->request_addr_callback != NULL)
+				channel->request_addr_callback(channel, rqst_id);
 		}
 		return -ENODEV;
 	}
@@ -375,7 +389,7 @@ int hv_ringbuffer_read(struct vmbus_channel *channel,
 	memcpy(buffer, (const char *)desc + offset, packetlen);
 
 	/* Advance ring index to next packet descriptor */
-	__hv_pkt_iter_next(channel, desc);
+	__hv_pkt_iter_next(channel, desc, true);
 
 	/* Notify host of update */
 	hv_pkt_iter_close(channel);
@@ -401,6 +415,22 @@ static u32 hv_pkt_iter_avail(const struct hv_ring_buffer_info *rbi)
 		return (rbi->ring_datasize - priv_read_loc) + write_loc;
 }
 
+/*
+ * Get first vmbus packet without copying it out of the ring buffer
+ */
+struct vmpacket_descriptor *hv_pkt_iter_first_raw(struct vmbus_channel *channel)
+{
+	struct hv_ring_buffer_info *rbi = &channel->inbound;
+
+	hv_debug_delay_test(channel, MESSAGE_DELAY);
+
+	if (hv_pkt_iter_avail(rbi) < sizeof(struct vmpacket_descriptor))
+		return NULL;
+
+	return (struct vmpacket_descriptor *)(hv_get_ring_buffer(rbi) + rbi->priv_read_index);
+}
+EXPORT_SYMBOL_GPL(hv_pkt_iter_first_raw);
+
 /*
  * Get first vmbus packet from ring buffer after read_index
  *
@@ -409,17 +439,49 @@ static u32 hv_pkt_iter_avail(const struct hv_ring_buffer_info *rbi)
 struct vmpacket_descriptor *hv_pkt_iter_first(struct vmbus_channel *channel)
 {
 	struct hv_ring_buffer_info *rbi = &channel->inbound;
-	struct vmpacket_descriptor *desc;
+	struct vmpacket_descriptor *desc, *desc_copy;
+	u32 bytes_avail, pkt_len, pkt_offset;
 
-	hv_debug_delay_test(channel, MESSAGE_DELAY);
-	if (hv_pkt_iter_avail(rbi) < sizeof(struct vmpacket_descriptor))
+	desc = hv_pkt_iter_first_raw(channel);
+	if (!desc)
 		return NULL;
 
-	desc = hv_get_ring_buffer(rbi) + rbi->priv_read_index;
-	if (desc)
-		prefetch((char *)desc + (desc->len8 << 3));
+	bytes_avail = min(rbi->pkt_buffer_size, hv_pkt_iter_avail(rbi));
+
+	/*
+	 * Ensure the compiler does not use references to incoming Hyper-V values (which
+	 * could change at any moment) when reading local variables later in the code
+	 */
+	pkt_len = READ_ONCE(desc->len8) << 3;
+	pkt_offset = READ_ONCE(desc->offset8) << 3;
+
+	/*
+	 * If pkt_len is invalid, set it to the smaller of hv_pkt_iter_avail() and
+	 * rbi->pkt_buffer_size
+	 */
+	if (pkt_len < sizeof(struct vmpacket_descriptor) || pkt_len > bytes_avail)
+		pkt_len = bytes_avail;
+
+	/*
+	 * If pkt_offset is invalid, arbitrarily set it to
+	 * the size of vmpacket_descriptor
+	 */
+	if (pkt_offset < sizeof(struct vmpacket_descriptor) || pkt_offset > pkt_len)
+		pkt_offset = sizeof(struct vmpacket_descriptor);
+
+	/* Copy the Hyper-V packet out of the ring buffer */
+	desc_copy = (struct vmpacket_descriptor *)rbi->pkt_buffer;
+	memcpy(desc_copy, desc, pkt_len);
+
+	/*
+	 * Hyper-V could still change len8 and offset8 after the earlier read.
+	 * Ensure that desc_copy has legal values for len8 and offset8 that
+	 * are consistent with the copy we just made
+	 */
+	desc_copy->len8 = pkt_len >> 3;
+	desc_copy->offset8 = pkt_offset >> 3;
 
-	return desc;
+	return desc_copy;
 }
 EXPORT_SYMBOL_GPL(hv_pkt_iter_first);
 
@@ -431,7 +493,8 @@ EXPORT_SYMBOL_GPL(hv_pkt_iter_first);
  */
 struct vmpacket_descriptor *
 __hv_pkt_iter_next(struct vmbus_channel *channel,
-		   const struct vmpacket_descriptor *desc)
+		   const struct vmpacket_descriptor *desc,
+		   bool copy)
 {
 	struct hv_ring_buffer_info *rbi = &channel->inbound;
 	u32 packetlen = desc->len8 << 3;
@@ -444,7 +507,7 @@ __hv_pkt_iter_next(struct vmbus_channel *channel,
 		rbi->priv_read_index -= dsize;
 
 	/* more data? */
-	return hv_pkt_iter_first(channel);
+	return copy ? hv_pkt_iter_first(channel) : hv_pkt_iter_first_raw(channel);
 }
 EXPORT_SYMBOL_GPL(__hv_pkt_iter_next);
 

drivers/net/hyperv/hyperv_net.h

@@ -895,9 +895,16 @@ static inline u32 netvsc_rqstor_size(unsigned long ringbytes)
 		ringbytes / NETVSC_MIN_IN_MSG_SIZE;
 }
 
+/* XFER PAGE packets can specify a maximum of 375 ranges for NDIS >= 6.0
+ * and a maximum of 64 ranges for NDIS < 6.0 with no RSC; with RSC, this
+ * limit is raised to 562 (= NVSP_RSC_MAX).
+ */
+#define NETVSC_MAX_XFER_PAGE_RANGES NVSP_RSC_MAX
 #define NETVSC_XFER_HEADER_SIZE(rng_cnt) \
 		(offsetof(struct vmtransfer_page_packet_header, ranges) + \
 		(rng_cnt) * sizeof(struct vmtransfer_page_range))
+#define NETVSC_MAX_PKT_SIZE (NETVSC_XFER_HEADER_SIZE(NETVSC_MAX_XFER_PAGE_RANGES) + \
+		sizeof(struct nvsp_message) + (sizeof(u32) * VRSS_SEND_TAB_SIZE))
 
 struct multi_send_data {
 	struct sk_buff *skb; /* skb containing the pkt */

drivers/net/hyperv/netvsc.c

@@ -757,7 +757,7 @@ static void netvsc_send_tx_complete(struct net_device *ndev,
 	int queue_sends;
 	u64 cmd_rqst;
 
-	cmd_rqst = vmbus_request_addr(&channel->requestor, (u64)desc->trans_id);
+	cmd_rqst = channel->request_addr_callback(channel, (u64)desc->trans_id);
 	if (cmd_rqst == VMBUS_RQST_ERROR) {
 		netdev_err(ndev, "Incorrect transaction id\n");
 		return;
@@ -817,8 +817,8 @@ static void netvsc_send_completion(struct net_device *ndev,
 
 	/* First check if this is a VMBUS completion without data payload */
 	if (!msglen) {
-		cmd_rqst = vmbus_request_addr(&incoming_channel->requestor,
-					      (u64)desc->trans_id);
+		cmd_rqst = incoming_channel->request_addr_callback(incoming_channel,
+								   (u64)desc->trans_id);
 		if (cmd_rqst == VMBUS_RQST_ERROR) {
 			netdev_err(ndev, "Invalid transaction id\n");
 			return;
@@ -1649,7 +1649,11 @@ struct netvsc_device *netvsc_device_add(struct hv_device *device,
 		       netvsc_poll, NAPI_POLL_WEIGHT);
 
 	/* Open the channel */
+	device->channel->next_request_id_callback = vmbus_next_request_id;
+	device->channel->request_addr_callback = vmbus_request_addr;
 	device->channel->rqstor_size = netvsc_rqstor_size(netvsc_ring_bytes);
+	device->channel->max_pkt_size = NETVSC_MAX_PKT_SIZE;
+
 	ret = vmbus_open(device->channel, netvsc_ring_bytes,
 			 netvsc_ring_bytes,  NULL, 0,
 			 netvsc_channel_cb, net_device->chan_table);

drivers/net/hyperv/rndis_filter.c

@@ -1259,7 +1259,11 @@ static void netvsc_sc_open(struct vmbus_channel *new_sc)
 	/* Set the channel before opening.*/
 	nvchan->channel = new_sc;
 
+	new_sc->next_request_id_callback = vmbus_next_request_id;
+	new_sc->request_addr_callback = vmbus_request_addr;
 	new_sc->rqstor_size = netvsc_rqstor_size(netvsc_ring_bytes);
+	new_sc->max_pkt_size = NETVSC_MAX_PKT_SIZE;
+
 	ret = vmbus_open(new_sc, netvsc_ring_bytes,
 			 netvsc_ring_bytes, NULL, 0,
 			 netvsc_channel_cb, nvchan);

drivers/pci/controller/pci-hyperv.c

@@ -3476,6 +3476,9 @@ static void __exit exit_hv_pci_drv(void)
 
 static int __init init_hv_pci_drv(void)
 {
+	if (!hv_is_hyperv_initialized())
+		return -ENODEV;
+
 	/* Set the invalid domain number's bit, so it will not be used */
 	set_bit(HVPCI_DOM_INVALID, hvpci_dom_map);
 

drivers/scsi/storvsc_drv.c

@@ -406,6 +406,14 @@ static void storvsc_on_channel_callback(void *context);
 #define STORVSC_IDE_MAX_TARGETS				1
 #define STORVSC_IDE_MAX_CHANNELS			1
 
+/*
+ * Upper bound on the size of a storvsc packet. vmscsi_size_delta is not
+ * included in the calculation because it is set after STORVSC_MAX_PKT_SIZE
+ * is used in storvsc_connect_to_vsp
+ */
+#define STORVSC_MAX_PKT_SIZE (sizeof(struct vmpacket_descriptor) +\
+			      sizeof(struct vstor_packet))
+
 struct storvsc_cmd_request {
 	struct scsi_cmnd *cmd;
 
@@ -688,6 +696,23 @@ static void storvsc_change_target_cpu(struct vmbus_channel *channel, u32 old,
 	spin_unlock_irqrestore(&stor_device->lock, flags);
 }
 
+static u64 storvsc_next_request_id(struct vmbus_channel *channel, u64 rqst_addr)
+{
+	struct storvsc_cmd_request *request =
+		(struct storvsc_cmd_request *)(unsigned long)rqst_addr;
+
+	if (rqst_addr == VMBUS_RQST_INIT)
+		return VMBUS_RQST_INIT;
+	if (rqst_addr == VMBUS_RQST_RESET)
+		return VMBUS_RQST_RESET;
+
+	/*
+	 * Cannot return an ID of 0, which is reserved for an unsolicited
+	 * message from Hyper-V.
+	 */
+	return (u64)blk_mq_unique_tag(request->cmd->request) + 1;
+}
+
 static void handle_sc_creation(struct vmbus_channel *new_sc)
 {
 	struct hv_device *device = new_sc->primary_channel->device_obj;
@@ -701,12 +726,9 @@ static void handle_sc_creation(struct vmbus_channel *new_sc)
 		return;
 
 	memset(&props, 0, sizeof(struct vmstorage_channel_properties));
+	new_sc->max_pkt_size = STORVSC_MAX_PKT_SIZE;
 
-	/*
-	 * The size of vmbus_requestor is an upper bound on the number of requests
-	 * that can be in-progress at any one time across all channels.
-	 */
-	new_sc->rqstor_size = scsi_driver.can_queue;
+	new_sc->next_request_id_callback = storvsc_next_request_id;
 
 	ret = vmbus_open(new_sc,
 			 storvsc_ringbuffer_size,
@@ -773,7 +795,7 @@ static void  handle_multichannel_storage(struct hv_device *device, int max_chns)
 	ret = vmbus_sendpacket(device->channel, vstor_packet,
 			       (sizeof(struct vstor_packet) -
 			       stor_device->vmscsi_size_delta),
-			       (unsigned long)request,
+			       VMBUS_RQST_INIT,
 			       VM_PKT_DATA_INBAND,
 			       VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED);
 
@@ -842,7 +864,7 @@ static int storvsc_execute_vstor_op(struct hv_device *device,
 	ret = vmbus_sendpacket(device->channel, vstor_packet,
 			       (sizeof(struct vstor_packet) -
 			       stor_device->vmscsi_size_delta),
-			       (unsigned long)request,
+			       VMBUS_RQST_INIT,
 			       VM_PKT_DATA_INBAND,
 			       VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED);
 	if (ret != 0)
@@ -1244,6 +1266,7 @@ static void storvsc_on_channel_callback(void *context)
 	const struct vmpacket_descriptor *desc;
 	struct hv_device *device;
 	struct storvsc_device *stor_device;
+	struct Scsi_Host *shost;
 
 	if (channel->primary_channel != NULL)
 		device = channel->primary_channel->device_obj;
@@ -1254,20 +1277,12 @@ static void storvsc_on_channel_callback(void *context)
 	if (!stor_device)
 		return;
 
-	foreach_vmbus_pkt(desc, channel) {
-		void *packet = hv_pkt_data(desc);
-		struct storvsc_cmd_request *request;
-		u64 cmd_rqst;
-
-		cmd_rqst = vmbus_request_addr(&channel->requestor,
-					      desc->trans_id);
-		if (cmd_rqst == VMBUS_RQST_ERROR) {
-			dev_err(&device->device,
-				"Incorrect transaction id\n");
-			continue;
-		}
+	shost = stor_device->host;
 
-		request = (struct storvsc_cmd_request *)(unsigned long)cmd_rqst;
+	foreach_vmbus_pkt(desc, channel) {
+		struct vstor_packet *packet = hv_pkt_data(desc);
+		struct storvsc_cmd_request *request = NULL;
+		u64 rqst_id = desc->trans_id;
 
 		if (hv_pkt_datalen(desc) < sizeof(struct vstor_packet) -
 				stor_device->vmscsi_size_delta) {
@@ -1275,14 +1290,44 @@ static void storvsc_on_channel_callback(void *context)
 			continue;
 		}
 
-		if (request == &stor_device->init_request ||
-		    request == &stor_device->reset_request) {
-			memcpy(&request->vstor_packet, packet,
-			       (sizeof(struct vstor_packet) - stor_device->vmscsi_size_delta));
-			complete(&request->wait_event);
+		if (rqst_id == VMBUS_RQST_INIT) {
+			request = &stor_device->init_request;
+		} else if (rqst_id == VMBUS_RQST_RESET) {
+			request = &stor_device->reset_request;
 		} else {
+			/* Hyper-V can send an unsolicited message with ID of 0 */
+			if (rqst_id == 0) {
+				/*
+				 * storvsc_on_receive() looks at the vstor_packet in the message
+				 * from the ring buffer.  If the operation in the vstor_packet is
+				 * COMPLETE_IO, then we call storvsc_on_io_completion(), and
+				 * dereference the guest memory address.  Make sure we don't call
+				 * storvsc_on_io_completion() with a guest memory address that is
+				 * zero if Hyper-V were to construct and send such a bogus packet.
+				 */
+				if (packet->operation == VSTOR_OPERATION_COMPLETE_IO) {
+					dev_err(&device->device, "Invalid packet with ID of 0\n");
+					continue;
+				}
+			} else {
+				struct scsi_cmnd *scmnd;
+
+				/* Transaction 'rqst_id' corresponds to tag 'rqst_id - 1' */
+				scmnd = scsi_host_find_tag(shost, rqst_id - 1);
+				if (scmnd == NULL) {
+					dev_err(&device->device, "Incorrect transaction ID\n");
+					continue;
+				}
+				request = (struct storvsc_cmd_request *)scsi_cmd_priv(scmnd);
+			}
+
 			storvsc_on_receive(stor_device, packet, request);
+			continue;
 		}
+
+		memcpy(&request->vstor_packet, packet,
+		       (sizeof(struct vstor_packet) - stor_device->vmscsi_size_delta));
+		complete(&request->wait_event);
 	}
 }
 
@@ -1294,11 +1339,8 @@ static int storvsc_connect_to_vsp(struct hv_device *device, u32 ring_size,
 
 	memset(&props, 0, sizeof(struct vmstorage_channel_properties));
 
-	/*
-	 * The size of vmbus_requestor is an upper bound on the number of requests
-	 * that can be in-progress at any one time across all channels.
-	 */
-	device->channel->rqstor_size = scsi_driver.can_queue;
+	device->channel->max_pkt_size = STORVSC_MAX_PKT_SIZE;
+	device->channel->next_request_id_callback = storvsc_next_request_id;
 
 	ret = vmbus_open(device->channel,
 			 ring_size,
@@ -1624,7 +1666,7 @@ static int storvsc_host_reset_handler(struct scsi_cmnd *scmnd)
 	ret = vmbus_sendpacket(device->channel, vstor_packet,
 			       (sizeof(struct vstor_packet) -
 				stor_device->vmscsi_size_delta),
-			       (unsigned long)&stor_device->reset_request,
+			       VMBUS_RQST_RESET,
 			       VM_PKT_DATA_INBAND,
 			       VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED);
 	if (ret != 0)

include/linux/hyperv.h

@@ -181,6 +181,10 @@ struct hv_ring_buffer_info {
 	 * being freed while the ring buffer is being accessed.
 	 */
 	struct mutex ring_buffer_mutex;
+
+	/* Buffer that holds a copy of an incoming host packet */
+	void *pkt_buffer;
+	u32 pkt_buffer_size;
 };
 
 
@@ -790,7 +794,11 @@ struct vmbus_requestor {
 
 #define VMBUS_NO_RQSTOR U64_MAX
 #define VMBUS_RQST_ERROR (U64_MAX - 1)
+/* NetVSC-specific */
 #define VMBUS_RQST_ID_NO_RESPONSE (U64_MAX - 2)
+/* StorVSC-specific */
+#define VMBUS_RQST_INIT (U64_MAX - 2)
+#define VMBUS_RQST_RESET (U64_MAX - 3)
 
 struct vmbus_device {
 	u16  dev_type;
@@ -799,6 +807,8 @@ struct vmbus_device {
 	bool allowed_in_isolated;
 };
 
+#define VMBUS_DEFAULT_MAX_PKT_SIZE 4096
+
 struct vmbus_channel {
 	struct list_head listentry;
 
@@ -1018,13 +1028,21 @@ struct vmbus_channel {
 	u32 fuzz_testing_interrupt_delay;
 	u32 fuzz_testing_message_delay;
 
+	/* callback to generate a request ID from a request address */
+	u64 (*next_request_id_callback)(struct vmbus_channel *channel, u64 rqst_addr);
+	/* callback to retrieve a request address from a request ID */
+	u64 (*request_addr_callback)(struct vmbus_channel *channel, u64 rqst_id);
+
 	/* request/transaction ids for VMBus */
 	struct vmbus_requestor requestor;
 	u32 rqstor_size;
+
+	/* The max size of a packet on this channel */
+	u32 max_pkt_size;
 };
 
-u64 vmbus_next_request_id(struct vmbus_requestor *rqstor, u64 rqst_addr);
-u64 vmbus_request_addr(struct vmbus_requestor *rqstor, u64 trans_id);
+u64 vmbus_next_request_id(struct vmbus_channel *channel, u64 rqst_addr);
+u64 vmbus_request_addr(struct vmbus_channel *channel, u64 trans_id);
 
 static inline bool is_hvsock_channel(const struct vmbus_channel *c)
 {
@@ -1662,32 +1680,55 @@ static inline u32 hv_pkt_datalen(const struct vmpacket_descriptor *desc)
 }
 
 
+struct vmpacket_descriptor *
+hv_pkt_iter_first_raw(struct vmbus_channel *channel);
+
 struct vmpacket_descriptor *
 hv_pkt_iter_first(struct vmbus_channel *channel);
 
 struct vmpacket_descriptor *
 __hv_pkt_iter_next(struct vmbus_channel *channel,
-		   const struct vmpacket_descriptor *pkt);
+		   const struct vmpacket_descriptor *pkt,
+		   bool copy);
 
 void hv_pkt_iter_close(struct vmbus_channel *channel);
 
-/*
- * Get next packet descriptor from iterator
- * If at end of list, return NULL and update host.
- */
 static inline struct vmpacket_descriptor *
-hv_pkt_iter_next(struct vmbus_channel *channel,
-		 const struct vmpacket_descriptor *pkt)
+hv_pkt_iter_next_pkt(struct vmbus_channel *channel,
+		     const struct vmpacket_descriptor *pkt,
+		     bool copy)
 {
 	struct vmpacket_descriptor *nxt;
 
-	nxt = __hv_pkt_iter_next(channel, pkt);
+	nxt = __hv_pkt_iter_next(channel, pkt, copy);
 	if (!nxt)
 		hv_pkt_iter_close(channel);
 
 	return nxt;
 }
 
+/*
+ * Get next packet descriptor without copying it out of the ring buffer
+ * If at end of list, return NULL and update host.
+ */
+static inline struct vmpacket_descriptor *
+hv_pkt_iter_next_raw(struct vmbus_channel *channel,
+		     const struct vmpacket_descriptor *pkt)
+{
+	return hv_pkt_iter_next_pkt(channel, pkt, false);
+}
+
+/*
+ * Get next packet descriptor from iterator
+ * If at end of list, return NULL and update host.
+ */
+static inline struct vmpacket_descriptor *
+hv_pkt_iter_next(struct vmbus_channel *channel,
+		 const struct vmpacket_descriptor *pkt)
+{
+	return hv_pkt_iter_next_pkt(channel, pkt, true);
+}
+
 #define foreach_vmbus_pkt(pkt, channel) \
 	for (pkt = hv_pkt_iter_first(channel); pkt; \
 	    pkt = hv_pkt_iter_next(channel, pkt))

net/vmw_vsock/hyperv_transport.c

@@ -596,7 +596,7 @@ static ssize_t hvs_stream_dequeue(struct vsock_sock *vsk, struct msghdr *msg,
 		return -EOPNOTSUPP;
 
 	if (need_refill) {
-		hvs->recv_desc = hv_pkt_iter_first(hvs->chan);
+		hvs->recv_desc = hv_pkt_iter_first_raw(hvs->chan);
 		ret = hvs_update_recv_data(hvs);
 		if (ret)
 			return ret;
@@ -610,7 +610,7 @@ static ssize_t hvs_stream_dequeue(struct vsock_sock *vsk, struct msghdr *msg,
 
 	hvs->recv_data_len -= to_read;
 	if (hvs->recv_data_len == 0) {
-		hvs->recv_desc = hv_pkt_iter_next(hvs->chan, hvs->recv_desc);
+		hvs->recv_desc = hv_pkt_iter_next_raw(hvs->chan, hvs->recv_desc);
 		if (hvs->recv_desc) {
 			ret = hvs_update_recv_data(hvs);
 			if (ret)