Skip to contentSkip to conversation

hv_sock: Hardening changes

PATCH Applied Andrea Parri (Microsoft) 2 participants 7 messages Apr 28, 2022

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

v2 cover hv_sock: Hardening changes

Changes since v1[1]:
- Expand commit message of patch #2
- Coding style changes

Changes since RFC[2]:
- Massage changelogs, fix typo
- Drop "hv_sock: Initialize send_buf in hvs_stream_enqueue()"
- Remove style/newline change
- Remove/"inline" hv_pkt_iter_first_raw()

Thanks,
Andrea

[1] https://lkml.kernel.org/r/[email protected]
[2] https://lkml.kernel.org/r/[email protected]

Andrea Parri (Microsoft) (5):
hv_sock: Check hv_pkt_iter_first_raw()'s return value
hv_sock: Copy packets sent by Hyper-V out of the ring buffer
hv_sock: Add validation for untrusted Hyper-V values
Drivers: hv: vmbus: Accept hv_sock offers in isolated guests
Drivers: hv: vmbus: Refactor the ring-buffer iterator functions

 drivers/hv/channel_mgmt.c        |  8 ++++--
 drivers/hv/ring_buffer.c         | 32 ++++++---------------
 include/linux/hyperv.h           | 48 ++++++++++----------------------
 net/vmw_vsock/hyperv_transport.c | 21 +++++++++++---
 4 files changed, 47 insertions(+), 62 deletions(-)
Signature · 1 line
2.25.1
#1 lore raw

PATCH v2 1/5 hv_sock: Check hv_pkt_iter_first_raw()'s return value

The function returns NULL if the ring buffer doesn't contain enough readable bytes to constitute a packet descriptor. The ring buffer's write_index is in memory which is shared with the Hyper-V host, an erroneous or malicious host could thus change its value and overturn the result of hvs_stream_has_data().

---
 net/vmw_vsock/hyperv_transport.c | 2 ++
 1 file changed, 2 insertions(+)
net/vmw_vsock/hyperv_transport.c +2 -0
index e111e13b66604..943352530936e 100644
@@ -603,6 +603,8 @@ static ssize_t hvs_stream_dequeue(struct vsock_sock *vsk, struct msghdr *msg,
 
 	if (need_refill) {
 		hvs->recv_desc = hv_pkt_iter_first_raw(hvs->chan);
+		if (!hvs->recv_desc)
+			return -ENOBUFS;
 		ret = hvs_update_recv_data(hvs);
 		if (ret)
 			return ret;
Signature · 1 line
2.25.1
#2 lore raw

PATCH v2 2/5 hv_sock: Copy packets sent by Hyper-V out of the ring buffer

In reply to Andrea Parri (Microsoft)

Pointers to VMbus packets sent by Hyper-V are used by the hv_sock driver within the guest VM. Hyper-V can send packets with erroneous values or modify packet fields after they are processed by the guest. To defend against these scenarios, copy the incoming packet after validating its length and offset fields using hv_pkt_iter_{first,next}(). Use HVS_PKT_LEN(HVS_MTU_SIZE) to initialize the buffer which holds the copies of the incoming packets. In this way, the packet can no longer be modified by the host.

---
 net/vmw_vsock/hyperv_transport.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
net/vmw_vsock/hyperv_transport.c +7 -2
index 943352530936e..8c37d07017fc4 100644
@@ -78,6 +78,9 @@ struct hvs_send_buf {
 					 ALIGN((payload_len), 8) + \
 					 VMBUS_PKT_TRAILER_SIZE)
 
+/* Upper bound on the size of a VMbus packet for hv_sock */
+#define HVS_MAX_PKT_SIZE	HVS_PKT_LEN(HVS_MTU_SIZE)
+
 union hvs_service_id {
 	guid_t	srv_id;
 
@@ -378,6 +381,8 @@ static void hvs_open_connection(struct vmbus_channel *chan)
 		rcvbuf = ALIGN(rcvbuf, HV_HYP_PAGE_SIZE);
 	}
 
+	chan->max_pkt_size = HVS_MAX_PKT_SIZE;
+
 	ret = vmbus_open(chan, sndbuf, rcvbuf, NULL, 0, hvs_channel_cb,
 			 conn_from_host ? new : sk);
 	if (ret != 0) {
@@ -602,7 +607,7 @@ 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_raw(hvs->chan);
+		hvs->recv_desc = hv_pkt_iter_first(hvs->chan);
 		if (!hvs->recv_desc)
 			return -ENOBUFS;
 		ret = hvs_update_recv_data(hvs);
@@ -618,7 +623,7 @@ 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_raw(hvs->chan, hvs->recv_desc);
+		hvs->recv_desc = hv_pkt_iter_next(hvs->chan, hvs->recv_desc);
 		if (hvs->recv_desc) {
 			ret = hvs_update_recv_data(hvs);
 			if (ret)
Signature · 1 line
2.25.1
#3 lore raw

PATCH v2 3/5 hv_sock: Add validation for untrusted Hyper-V values

In reply to Andrea Parri (Microsoft)

For additional robustness in the face of Hyper-V errors or malicious behavior, validate all values that originate from packets that Hyper-V has sent to the guest in the host-to-guest ring buffer. Ensure that invalid values cannot cause data being copied out of the bounds of the source buffer in hvs_stream_dequeue().

---
 include/linux/hyperv.h           |  5 +++++
 net/vmw_vsock/hyperv_transport.c | 10 ++++++++--
 2 files changed, 13 insertions(+), 2 deletions(-)
include/linux/hyperv.h +5 -0
index 460a716f47486..a01c9fd0a3348 100644
@@ -1696,6 +1696,11 @@ static inline u32 hv_pkt_datalen(const struct vmpacket_descriptor *desc)
 	return (desc->len8 << 3) - (desc->offset8 << 3);
 }
 
+/* Get packet length associated with descriptor */
+static inline u32 hv_pkt_len(const struct vmpacket_descriptor *desc)
+{
+	return desc->len8 << 3;
+}
 
 struct vmpacket_descriptor *
 hv_pkt_iter_first_raw(struct vmbus_channel *channel);
net/vmw_vsock/hyperv_transport.c +8 -2
index 8c37d07017fc4..fd98229e3db30 100644
@@ -577,12 +577,18 @@ static bool hvs_dgram_allow(u32 cid, u32 port)
 static int hvs_update_recv_data(struct hvsock *hvs)
 {
 	struct hvs_recv_buf *recv_buf;
-	u32 payload_len;
+	u32 pkt_len, payload_len;
+
+	pkt_len = hv_pkt_len(hvs->recv_desc);
+
+	if (pkt_len < HVS_HEADER_LEN)
+		return -EIO;
 
 	recv_buf = (struct hvs_recv_buf *)(hvs->recv_desc + 1);
 	payload_len = recv_buf->hdr.data_size;
 
-	if (payload_len > HVS_MTU_SIZE)
+	if (payload_len > pkt_len - HVS_HEADER_LEN ||
+	    payload_len > HVS_MTU_SIZE)
 		return -EIO;
 
 	if (payload_len == 0)
Signature · 1 line
2.25.1
#4 lore raw

PATCH v2 4/5 Drivers: hv: vmbus: Accept hv_sock offers in isolated guests

In reply to Andrea Parri (Microsoft)

So that isolated guests can communicate with the host via hv_sock channels.

---
 drivers/hv/channel_mgmt.c | 8 ++++++--
 include/linux/hyperv.h    | 8 ++++++--
 2 files changed, 12 insertions(+), 4 deletions(-)
drivers/hv/channel_mgmt.c +6 -2
index 50fea31f019f6..0db1b775e0139 100644
@@ -976,13 +976,17 @@ find_primary_channel_by_offer(const struct vmbus_channel_offer_channel *offer)
 	return channel;
 }
 
-static bool vmbus_is_valid_device(const guid_t *guid)
+static bool vmbus_is_valid_offer(const struct vmbus_channel_offer_channel *offer)
 {
+	const guid_t *guid = &offer->offer.if_type;
 	u16 i;
 
 	if (!hv_is_isolation_supported())
 		return true;
 
+	if (is_hvsock_offer(offer))
+		return true;
+
 	for (i = 0; i < ARRAY_SIZE(vmbus_devs); i++) {
 		if (guid_equal(guid, &vmbus_devs[i].guid))
 			return vmbus_devs[i].allowed_in_isolated;
@@ -1004,7 +1008,7 @@ static void vmbus_onoffer(struct vmbus_channel_message_header *hdr)
 
 	trace_vmbus_onoffer(offer);
 
-	if (!vmbus_is_valid_device(&offer->offer.if_type)) {
+	if (!vmbus_is_valid_offer(offer)) {
 		pr_err_ratelimited("Invalid offer %d from the host supporting isolation\n",
 				   offer->child_relid);
 		atomic_dec(&vmbus_connection.offer_in_progress);
include/linux/hyperv.h +6 -2
index a01c9fd0a3348..b028905d8334e 100644
@@ -1064,10 +1064,14 @@ u64 vmbus_request_addr_match(struct vmbus_channel *channel, u64 trans_id,
 			     u64 rqst_addr);
 u64 vmbus_request_addr(struct vmbus_channel *channel, u64 trans_id);
 
+static inline bool is_hvsock_offer(const struct vmbus_channel_offer_channel *o)
+{
+	return !!(o->offer.chn_flags & VMBUS_CHANNEL_TLNPI_PROVIDER_OFFER);
+}
+
 static inline bool is_hvsock_channel(const struct vmbus_channel *c)
 {
-	return !!(c->offermsg.offer.chn_flags &
-		  VMBUS_CHANNEL_TLNPI_PROVIDER_OFFER);
+	return is_hvsock_offer(&c->offermsg);
 }
 
 static inline bool is_sub_channel(const struct vmbus_channel *c)
Signature · 1 line
2.25.1
#5 lore raw

PATCH v2 5/5 Drivers: hv: vmbus: Refactor the ring-buffer iterator functions

In reply to Andrea Parri (Microsoft)

With no users of hv_pkt_iter_next_raw() and no "external" users of hv_pkt_iter_first_raw(), the iterator functions can be refactored and simplified to remove some indirection/code.

---
 drivers/hv/ring_buffer.c | 32 +++++++++-----------------------
 include/linux/hyperv.h   | 35 ++++-------------------------------
 2 files changed, 13 insertions(+), 54 deletions(-)
drivers/hv/ring_buffer.c +9 -23
index e101b11f95e5d..59a4aa86d1f35 100644
@@ -429,7 +429,7 @@ 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, true);
+	__hv_pkt_iter_next(channel, desc);
 
 	/* Notify host of update */
 	hv_pkt_iter_close(channel);
@@ -464,22 +464,6 @@ 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
  *
@@ -491,11 +475,14 @@ struct vmpacket_descriptor *hv_pkt_iter_first(struct vmbus_channel *channel)
 	struct vmpacket_descriptor *desc, *desc_copy;
 	u32 bytes_avail, pkt_len, pkt_offset;
 
-	desc = hv_pkt_iter_first_raw(channel);
-	if (!desc)
+	hv_debug_delay_test(channel, MESSAGE_DELAY);
+
+	bytes_avail = hv_pkt_iter_avail(rbi);
+	if (bytes_avail < sizeof(struct vmpacket_descriptor))
 		return NULL;
+	bytes_avail = min(rbi->pkt_buffer_size, bytes_avail);
 
-	bytes_avail = min(rbi->pkt_buffer_size, hv_pkt_iter_avail(rbi));
+	desc = (struct vmpacket_descriptor *)(hv_get_ring_buffer(rbi) + rbi->priv_read_index);
 
 	/*
 	 * Ensure the compiler does not use references to incoming Hyper-V values (which
@@ -542,8 +529,7 @@ EXPORT_SYMBOL_GPL(hv_pkt_iter_first);
  */
 struct vmpacket_descriptor *
 __hv_pkt_iter_next(struct vmbus_channel *channel,
-		   const struct vmpacket_descriptor *desc,
-		   bool copy)
+		   const struct vmpacket_descriptor *desc)
 {
 	struct hv_ring_buffer_info *rbi = &channel->inbound;
 	u32 packetlen = desc->len8 << 3;
@@ -556,7 +542,7 @@ __hv_pkt_iter_next(struct vmbus_channel *channel,
 		rbi->priv_read_index -= dsize;
 
 	/* more data? */
-	return copy ? hv_pkt_iter_first(channel) : hv_pkt_iter_first_raw(channel);
+	return hv_pkt_iter_first(channel);
 }
 EXPORT_SYMBOL_GPL(__hv_pkt_iter_next);
 
include/linux/hyperv.h +4 -31
index b028905d8334e..c440c45887c26 100644
@@ -1706,55 +1706,28 @@ static inline u32 hv_pkt_len(const struct vmpacket_descriptor *desc)
 	return desc->len8 << 3;
 }
 
-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,
-		   bool copy);
+		   const struct vmpacket_descriptor *pkt);
 
 void hv_pkt_iter_close(struct vmbus_channel *channel);
 
 static inline struct vmpacket_descriptor *
-hv_pkt_iter_next_pkt(struct vmbus_channel *channel,
-		     const struct vmpacket_descriptor *pkt,
-		     bool copy)
+hv_pkt_iter_next(struct vmbus_channel *channel,
+		 const struct vmpacket_descriptor *pkt)
 {
 	struct vmpacket_descriptor *nxt;
 
-	nxt = __hv_pkt_iter_next(channel, pkt, copy);
+	nxt = __hv_pkt_iter_next(channel, pkt);
 	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))
Signature · 1 line
2.25.1
maintainer Applied
#6 lore raw
In reply to Andrea Parri (Microsoft)
Quoted 7 lines · Andrea Parri (Microsoft)

Andrea Parri (Microsoft) (5):
hv_sock: Check hv_pkt_iter_first_raw()'s return value
hv_sock: Copy packets sent by Hyper-V out of the ring buffer
hv_sock: Add validation for untrusted Hyper-V values
Drivers: hv: vmbus: Accept hv_sock offers in isolated guests
Drivers: hv: vmbus: Refactor the ring-buffer iterator functions

Applied to hyperv-next. Thanks.