[PATCH 6/6] vhost_net: don't poll on -EFAULT

From: Jason Wang <jasowang@redhat.com>
To: netdev@vger.kernel.org, xma@us.ibm.com, davem@davemloft.net,
	linux-kernel@vger.kernel.org, mst@redhat.com
Cc: ebiederm@xmission.com
Subject: [PATCH 6/6] vhost_net: don't poll on -EFAULT
Date: Mon, 16 Apr 2012 14:08:33 +0800	[thread overview]
Message-ID: <20120416060833.14140.28139.stgit@intel-e5620-16-2.englab.nay.redhat.com> (raw)
In-Reply-To: <20120416060749.14140.19433.stgit@intel-e5620-16-2.englab.nay.redhat.com>

Currently, we restart tx polling unconditionally when sendmsg()
fails. This would cause unnecessary wakeups of vhost wokers as it's
only needed when the socket send buffer were exceeded. Fix this by
restart the tx polling only when sendmsg() returns value other than
-EFAULT.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/vhost/net.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
index 29abd65..035fa95 100644
--- a/drivers/vhost/net.c
+++ b/drivers/vhost/net.c
@@ -262,7 +262,8 @@ static void handle_tx(struct vhost_net *net)
 					UIO_MAXIOV;
 			}
 			vhost_discard_vq_desc(vq, 1);
-			tx_poll_start(net, sock);
+			if (err != -EFAULT)
+				tx_poll_start(net, sock);
 			break;
 		}
 		if (err != len)

