[PATCH net-next,1/6] hyperv: Fix the max_xfer_size in RNDIS initialization

[PATCH net-next,1/6] hyperv: Fix the max_xfer_size in RNDIS initialization

[Haiyang Zhang]

According to RNDIS specs, Windows sets this size to
0x4000. I use the same value here.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

---
 drivers/net/hyperv/rndis_filter.c |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)

diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c
index 06f8601..1337b64 100644
--- a/drivers/net/hyperv/rndis_filter.c
+++ b/drivers/net/hyperv/rndis_filter.c
@@ -684,8 +684,7 @@ static int rndis_filter_init_device(struct rndis_device *dev)
 	init = &request->request_msg.msg.init_req;
 	init->major_ver = RNDIS_MAJOR_VERSION;
 	init->minor_ver = RNDIS_MINOR_VERSION;
-	/* FIXME: Use 1536 - rounded ethernet frame size */
-	init->max_xfer_size = 2048;
+	init->max_xfer_size = 0x4000;
 
 	dev->state = RNDIS_DEV_INITIALIZING;
 
-- 
1.7.4.1

[PATCH net-next,2/6] hyperv: Fix the missing return value in rndis_filter_set_packet_filter()

[Haiyang Zhang]

Return ETIMEDOUT when the reply message is not received in time.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

---
 drivers/net/hyperv/rndis_filter.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c
index 1337b64..617eb2e 100644
--- a/drivers/net/hyperv/rndis_filter.c
+++ b/drivers/net/hyperv/rndis_filter.c
@@ -647,6 +647,7 @@ int rndis_filter_set_packet_filter(struct rndis_device *dev, u32 new_filter)
 	if (t == 0) {
 		netdev_err(ndev,
 			"timeout before we got a set response...\n");
+		ret = -ETIMEDOUT;
 		/*
 		 * We can't deallocate the request since we may still receive a
 		 * send completion for it.
-- 
1.7.4.1

Re: [PATCH net-next,2/6] hyperv: Fix the missing return value in rndis_filter_set_packet_filter()

[David Miller]

From: Haiyang Zhang <haiyangz@microsoft.com>
Date: Tue,  2 Oct 2012 08:30:20 -0700

> Return ETIMEDOUT when the reply message is not received in time.
> 
> Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
> Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

Applied.

[PATCH net-next, 3/6] hyperv: Fix page buffer handling in rndis_filter_send_request()

[Haiyang Zhang]

To prevent possible data corruption in RNDIS requests, add another
page buffer if the request message crossed page boundary.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

---
 drivers/net/hyperv/rndis_filter.c |   15 ++++++++++++++-
 1 files changed, 14 insertions(+), 1 deletions(-)

diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c
index 617eb2e..f25f41e 100644
--- a/drivers/net/hyperv/rndis_filter.c
+++ b/drivers/net/hyperv/rndis_filter.c
@@ -45,7 +45,8 @@ struct rndis_request {
 
 	/* Simplify allocation by having a netvsc packet inline */
 	struct hv_netvsc_packet	pkt;
-	struct hv_page_buffer buf;
+	/* Set 2 pages for rndis requests crossing page boundary */
+	struct hv_page_buffer buf[2];
 
 	struct rndis_message request_msg;
 	/*
@@ -227,6 +228,18 @@ static int rndis_filter_send_request(struct rndis_device *dev,
 	packet->page_buf[0].offset =
 		(unsigned long)&req->request_msg & (PAGE_SIZE - 1);
 
+	/* Add one page_buf when request_msg crossing page boundary */
+	if (packet->page_buf[0].offset + packet->page_buf[0].len > PAGE_SIZE) {
+		packet->page_buf_cnt++;
+		packet->page_buf[0].len = PAGE_SIZE -
+			packet->page_buf[0].offset;
+		packet->page_buf[1].pfn = virt_to_phys((void *)&req->request_msg
+			+ packet->page_buf[0].len) >> PAGE_SHIFT;
+		packet->page_buf[1].offset = 0;
+		packet->page_buf[1].len = req->request_msg.msg_len -
+			packet->page_buf[0].len;
+	}
+
 	packet->completion.send.send_completion_ctx = req;/* packet; */
 	packet->completion.send.send_completion =
 		rndis_filter_send_request_completion;
-- 
1.7.4.1

Re: [PATCH net-next,3/6] hyperv: Fix page buffer handling in rndis_filter_send_request()

[David Miller]

From: Haiyang Zhang <haiyangz@microsoft.com>
Date: Tue,  2 Oct 2012 08:30:21 -0700

> To prevent possible data corruption in RNDIS requests, add another
> page buffer if the request message crossed page boundary.
> 
> Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
> Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

Applied.

[PATCH net-next,4/6] hyperv: Remove extra allocated space for recv_pkt_list elements

[Haiyang Zhang]

The receive code path doesn't use the page buffer, so remove the
extra allocated space here.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

---
 drivers/net/hyperv/hyperv_net.h |    2 --
 drivers/net/hyperv/netvsc.c     |    4 +---
 2 files changed, 1 insertions(+), 5 deletions(-)

diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h
index 95ceb35..d58f28c 100644
--- a/drivers/net/hyperv/hyperv_net.h
+++ b/drivers/net/hyperv/hyperv_net.h
@@ -465,8 +465,6 @@ struct nvsp_message {
 
 #define NETVSC_RECEIVE_BUFFER_ID		0xcafe
 
-#define NETVSC_RECEIVE_SG_COUNT			1
-
 /* Preallocated receive packets */
 #define NETVSC_RECEIVE_PACKETLIST_COUNT		256
 
diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c
index 4a1a5f5..d9c4c03 100644
--- a/drivers/net/hyperv/netvsc.c
+++ b/drivers/net/hyperv/netvsc.c
@@ -904,9 +904,7 @@ int netvsc_device_add(struct hv_device *device, void *additional_info)
 	INIT_LIST_HEAD(&net_device->recv_pkt_list);
 
 	for (i = 0; i < NETVSC_RECEIVE_PACKETLIST_COUNT; i++) {
-		packet = kzalloc(sizeof(struct hv_netvsc_packet) +
-				 (NETVSC_RECEIVE_SG_COUNT *
-				  sizeof(struct hv_page_buffer)), GFP_KERNEL);
+		packet = kzalloc(sizeof(struct hv_netvsc_packet), GFP_KERNEL);
 		if (!packet)
 			break;
 
-- 
1.7.4.1

Re: [PATCH net-next,4/6] hyperv: Remove extra allocated space for recv_pkt_list elements

[David Miller]

From: Haiyang Zhang <haiyangz@microsoft.com>
Date: Tue,  2 Oct 2012 08:30:22 -0700

> The receive code path doesn't use the page buffer, so remove the
> extra allocated space here.
> 
> Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
> Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

Applied.

[PATCH net-next,5/6] hyperv: Report actual status in receive completion packet

[Haiyang Zhang]

The existing code always reports NVSP_STAT_SUCCESS. This patch adds the
mechanism to report failure when it happens.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

---
 drivers/net/hyperv/hyperv_net.h   |    2 ++
 drivers/net/hyperv/netvsc.c       |   18 ++++++++++++------
 drivers/net/hyperv/netvsc_drv.c   |    2 ++
 drivers/net/hyperv/rndis_filter.c |   19 ++++++++++++++-----
 4 files changed, 30 insertions(+), 11 deletions(-)

diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h
index d58f28c..5fd6f46 100644
--- a/drivers/net/hyperv/hyperv_net.h
+++ b/drivers/net/hyperv/hyperv_net.h
@@ -35,6 +35,7 @@ struct hv_netvsc_packet;
 /* Represent the xfer page packet which contains 1 or more netvsc packet */
 struct xferpage_packet {
 	struct list_head list_ent;
+	u32 status;
 
 	/* # of netvsc packets this xfer packet contains */
 	u32 count;
@@ -47,6 +48,7 @@ struct xferpage_packet {
 struct hv_netvsc_packet {
 	/* Bookkeeping stuff */
 	struct list_head list_ent;
+	u32 status;
 
 	struct hv_device *device;
 	bool is_data_pkt;
diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c
index d9c4c03..1cd7748 100644
--- a/drivers/net/hyperv/netvsc.c
+++ b/drivers/net/hyperv/netvsc.c
@@ -558,7 +558,7 @@ int netvsc_send(struct hv_device *device,
 }
 
 static void netvsc_send_recv_completion(struct hv_device *device,
-					u64 transaction_id)
+					u64 transaction_id, u32 status)
 {
 	struct nvsp_message recvcompMessage;
 	int retries = 0;
@@ -571,9 +571,7 @@ static void netvsc_send_recv_completion(struct hv_device *device,
 	recvcompMessage.hdr.msg_type =
 				NVSP_MSG1_TYPE_SEND_RNDIS_PKT_COMPLETE;
 
-	/* FIXME: Pass in the status */
-	recvcompMessage.msg.v1_msg.send_rndis_pkt_complete.status =
-		NVSP_STAT_SUCCESS;
+	recvcompMessage.msg.v1_msg.send_rndis_pkt_complete.status = status;
 
 retry_send_cmplt:
 	/* Send the completion */
@@ -613,6 +611,7 @@ static void netvsc_receive_completion(void *context)
 	bool fsend_receive_comp = false;
 	unsigned long flags;
 	struct net_device *ndev;
+	u32 status = NVSP_STAT_NONE;
 
 	/*
 	 * Even though it seems logical to do a GetOutboundNetDevice() here to
@@ -627,6 +626,9 @@ static void netvsc_receive_completion(void *context)
 	/* Overloading use of the lock. */
 	spin_lock_irqsave(&net_device->recv_pkt_list_lock, flags);
 
+	if (packet->status != NVSP_STAT_SUCCESS)
+		packet->xfer_page_pkt->status = NVSP_STAT_FAIL;
+
 	packet->xfer_page_pkt->count--;
 
 	/*
@@ -636,6 +638,7 @@ static void netvsc_receive_completion(void *context)
 	if (packet->xfer_page_pkt->count == 0) {
 		fsend_receive_comp = true;
 		transaction_id = packet->completion.recv.recv_completion_tid;
+		status = packet->xfer_page_pkt->status;
 		list_add_tail(&packet->xfer_page_pkt->list_ent,
 			      &net_device->recv_pkt_list);
 
@@ -647,7 +650,7 @@ static void netvsc_receive_completion(void *context)
 
 	/* Send a receive completion for the xfer page packet */
 	if (fsend_receive_comp)
-		netvsc_send_recv_completion(device, transaction_id);
+		netvsc_send_recv_completion(device, transaction_id, status);
 
 }
 
@@ -736,7 +739,8 @@ static void netvsc_receive(struct hv_device *device,
 				       flags);
 
 		netvsc_send_recv_completion(device,
-					    vmxferpage_packet->d.trans_id);
+					    vmxferpage_packet->d.trans_id,
+					    NVSP_STAT_FAIL);
 
 		return;
 	}
@@ -744,6 +748,7 @@ static void netvsc_receive(struct hv_device *device,
 	/* Remove the 1st packet to represent the xfer page packet itself */
 	xferpage_packet = (struct xferpage_packet *)listHead.next;
 	list_del(&xferpage_packet->list_ent);
+	xferpage_packet->status = NVSP_STAT_SUCCESS;
 
 	/* This is how much we can satisfy */
 	xferpage_packet->count = count - 1;
@@ -760,6 +765,7 @@ static void netvsc_receive(struct hv_device *device,
 		list_del(&netvsc_packet->list_ent);
 
 		/* Initialize the netvsc packet */
+		netvsc_packet->status = NVSP_STAT_SUCCESS;
 		netvsc_packet->xfer_page_pkt = xferpage_packet;
 		netvsc_packet->completion.recv.recv_completion =
 					netvsc_receive_completion;
diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c
index e91111a..f825a62 100644
--- a/drivers/net/hyperv/netvsc_drv.c
+++ b/drivers/net/hyperv/netvsc_drv.c
@@ -265,6 +265,7 @@ int netvsc_recv_callback(struct hv_device *device_obj,
 	if (!net) {
 		netdev_err(net, "got receive callback but net device"
 			" not initialized yet\n");
+		packet->status = NVSP_STAT_FAIL;
 		return 0;
 	}
 
@@ -272,6 +273,7 @@ int netvsc_recv_callback(struct hv_device *device_obj,
 	skb = netdev_alloc_skb_ip_align(net, packet->total_data_buflen);
 	if (unlikely(!skb)) {
 		++net->stats.rx_dropped;
+		packet->status = NVSP_STAT_FAIL;
 		return 0;
 	}
 
diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c
index f25f41e..e7e12cf 100644
--- a/drivers/net/hyperv/rndis_filter.c
+++ b/drivers/net/hyperv/rndis_filter.c
@@ -411,9 +411,12 @@ int rndis_filter_receive(struct hv_device *dev,
 	struct rndis_device *rndis_dev;
 	struct rndis_message *rndis_msg;
 	struct net_device *ndev;
+	int ret = 0;
 
-	if (!net_dev)
-		return -EINVAL;
+	if (!net_dev) {
+		ret = -EINVAL;
+		goto exit;
+	}
 
 	ndev = net_dev->ndev;
 
@@ -421,14 +424,16 @@ int rndis_filter_receive(struct hv_device *dev,
 	if (!net_dev->extension) {
 		netdev_err(ndev, "got rndis message but no rndis device - "
 			  "dropping this message!\n");
-		return -ENODEV;
+		ret = -ENODEV;
+		goto exit;
 	}
 
 	rndis_dev = (struct rndis_device *)net_dev->extension;
 	if (rndis_dev->state == RNDIS_DEV_UNINITIALIZED) {
 		netdev_err(ndev, "got rndis message but rndis device "
 			   "uninitialized...dropping this message!\n");
-		return -ENODEV;
+		ret = -ENODEV;
+		goto exit;
 	}
 
 	rndis_msg = pkt->data;
@@ -460,7 +465,11 @@ int rndis_filter_receive(struct hv_device *dev,
 		break;
 	}
 
-	return 0;
+exit:
+	if (ret != 0)
+		pkt->status = NVSP_STAT_FAIL;
+
+	return ret;
 }
 
 static int rndis_filter_query_device(struct rndis_device *dev, u32 oid,
-- 
1.7.4.1

Re: [PATCH net-next,5/6] hyperv: Report actual status in receive completion packet

[David Miller]

From: Haiyang Zhang <haiyangz@microsoft.com>
Date: Tue,  2 Oct 2012 08:30:23 -0700

> The existing code always reports NVSP_STAT_SUCCESS. This patch adds the
> mechanism to report failure when it happens.
> 
> Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
> Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

Applied.

[PATCH net-next,6/6] hyperv: Add buffer for extended info after the RNDIS response message.

[Haiyang Zhang]

In some response messages, there may be some extended info after the
message.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

---
 drivers/net/hyperv/rndis_filter.c |   22 ++++++++++++----------
 1 files changed, 12 insertions(+), 10 deletions(-)

diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c
index e7e12cf..928148c 100644
--- a/drivers/net/hyperv/rndis_filter.c
+++ b/drivers/net/hyperv/rndis_filter.c
@@ -32,16 +32,19 @@
 #include "hyperv_net.h"
 
 
+#define RNDIS_EXT_LEN 100
 struct rndis_request {
 	struct list_head list_ent;
 	struct completion  wait_event;
 
+	struct rndis_message response_msg;
 	/*
-	 * FIXME: We assumed a fixed size response here. If we do ever need to
-	 * handle a bigger response, we can either define a max response
-	 * message or add a response buffer variable above this field
+	 * The buffer for extended info after the RNDIS response message. It's
+	 * referenced based on the data offset in the RNDIS message. Its size
+	 * is enough for current needs, and should be sufficient for the near
+	 * future.
 	 */
-	struct rndis_message response_msg;
+	u8 response_ext[RNDIS_EXT_LEN];
 
 	/* Simplify allocation by having a netvsc packet inline */
 	struct hv_netvsc_packet	pkt;
@@ -50,12 +53,10 @@ struct rndis_request {
 
 	struct rndis_message request_msg;
 	/*
-	 * The buffer for the extended info after the RNDIS message. It's
-	 * referenced based on the data offset in the RNDIS message. Its size
-	 * is enough for current needs, and should be sufficient for the near
-	 * future.
+	 * The buffer for the extended info after the RNDIS request message.
+	 * It is referenced and sized in a similar way as response_ext.
 	 */
-	u8 ext[100];
+	u8 request_ext[RNDIS_EXT_LEN];
 };
 
 static void rndis_filter_send_completion(void *ctx);
@@ -274,7 +275,8 @@ static void rndis_filter_receive_response(struct rndis_device *dev,
 	spin_unlock_irqrestore(&dev->request_lock, flags);
 
 	if (found) {
-		if (resp->msg_len <= sizeof(struct rndis_message)) {
+		if (resp->msg_len <=
+		    sizeof(struct rndis_message) + RNDIS_EXT_LEN) {
 			memcpy(&request->response_msg, resp,
 			       resp->msg_len);
 		} else {
-- 
1.7.4.1

Re: [PATCH net-next,6/6] hyperv: Add buffer for extended info after the RNDIS response message.

[David Miller]

From: Haiyang Zhang <haiyangz@microsoft.com>
Date: Tue,  2 Oct 2012 08:30:24 -0700

> In some response messages, there may be some extended info after the
> message.
> 
> Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
> Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

Applied.

Re: [PATCH net-next,1/6] hyperv: Fix the max_xfer_size in RNDIS initialization

[David Miller]

From: Haiyang Zhang <haiyangz@microsoft.com>
Date: Tue,  2 Oct 2012 08:30:19 -0700

> According to RNDIS specs, Windows sets this size to
> 0x4000. I use the same value here.
> 
> Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
> Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>

Applied.