-----Original Message-----
From: Ben Hutchings [mailto:ben@decadent.org.uk]
Sent: Tuesday, July 24, 2012 9:11 PM
To: KY Srinivasan
Cc: gregkh@linuxfoundation.org; linux-kernel@vger.kernel.org;
devel@linuxdriverproject.org; virtualization@lists.osdl.org; olaf@aepfle.de;
apw@canonical.com; netdev@vger.kernel.org
Subject: Re: [PATCH 03/17] Drivers: hv: kvp: Cleanup error handling in KVP

On Tue, 2012-07-24 at 09:01 -0700, K. Y. Srinivasan wrote:

quoted

In preparation to implementing IP injection, cleanup the way we propagate
and handle errors both in the driver as well as in the user level daemon.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
---
 drivers/hv/hv_kvp.c      |  112 +++++++++++++++++++++++++++++++++++++-

--------

quoted

 include/linux/hyperv.h   |   17 +++++---
 tools/hv/hv_kvp_daemon.c |   70 +++++++++++++++-------------
 3 files changed, 138 insertions(+), 61 deletions(-)

diff --git a/drivers/hv/hv_kvp.c b/drivers/hv/hv_kvp.c
index 0012eed..9b7fc4a 100644
--- a/drivers/hv/hv_kvp.c
+++ b/drivers/hv/hv_kvp.c

[...]

quoted

@@ -109,27 +154,52 @@ kvp_cn_callback(struct cn_msg *msg, struct

netlink_skb_parms *nsp)

quoted

 {
 	struct hv_kvp_msg *message;
 	struct hv_kvp_msg_enumerate *data;
+	int	error = 0;

 	message = (struct hv_kvp_msg *)msg->data;
-	switch (message->kvp_hdr.operation) {
+
+	/*
+	 * If we are negotiating the version information
+	 * with the daemon; handle that first.
+	 */
+
+	if (in_hand_shake) {
+		if (kvp_handle_handshake(message))
+			in_hand_shake = false;
+		return;
+	}
+
+	/*
+	 * Based on the version of the daemon, we propagate errors from the
+	 * daemon differently.
+	 */
+
+	data = &message->body.kvp_enum_data;
+
+	switch (dm_reg_value) {
 	case KVP_OP_REGISTER:
-		pr_info("KVP: user-mode registering done.\n");
-		kvp_register();
-		kvp_transaction.active = false;
-		hv_kvp_onchannelcallback(kvp_transaction.kvp_context);
+		/*
+		 * Null string is used to pass back error condition.
+		 */
+		if (!strlen(data->data.key))

Do we know that the key is null-terminated here?  Shouldn't we just
check whether data->data.key[0] == 0?

Yes, currently we do return null string to indicate error.

quoted

+			error = HV_S_CONT;
 		break;

-	default:
-		data = &message->body.kvp_enum_data;
+	case KVP_OP_REGISTER1:
 		/*
-		 * Complete the transaction by forwarding the key value
-		 * to the host. But first, cancel the timeout.
+		 * We use the message header information from
+		 * the user level daemon to transmit errors.
 		 */
-		if (cancel_delayed_work_sync(&kvp_work))
-			kvp_respond_to_host(data->data.key,
-					 data->data.value,
-					!strlen(data->data.key));
+		error = *((int *)(&message->kvp_hdr.operation));

[...]

What's with the casting (repeated in many other places)?  Wouldn't it be
better to redefine struct hv_kvp_msg to start with something like:

	union {
		struct hv_kvp_hdr	request;
		int			error;
	} kvp_hdr;

Agreed; will do.

Ben.

--
Ben Hutchings
If more than one person is responsible for a bug, no one is at fault.