diff options
| author | Subash Abhinov Kasiviswanathan <subashab@codeaurora.org> | 2016-11-14 12:40:34 -0700 |
|---|---|---|
| committer | Subash Abhinov Kasiviswanathan <subashab@codeaurora.org> | 2016-11-14 12:49:05 -0700 |
| commit | 2d9b2acc3938899ef0c0fe9e32aff2f8e882dd3c (patch) | |
| tree | 6881e469e9fbc910661089e08425ac3a1f3af496 /net/rmnet_data | |
| parent | 24d0c1f91eb2850889cb27ef6fdcadc0c8ff2834 (diff) | |
net: rmnet_data: Handle buffer bloat for TCP scenarios
After commit 605ad7f184b60cf ("tcp: refine TSO autosizing"), kernel
throttles uplink TCP data in case there is not sufficient amount
of socket buffer available due to delayed release of buffers
through TX completions in the physical net device.
Work around this by orphaning the socket buffer. This makes the
kernel assume that more packets can be sent in this scenario.
Out of band signaling and flow controlling at qdisc / HTB layer
should guarantee no issue for flow control.
Throughput difference for IPv4 TCP UL -
Before change : 143Mbps
After change : 146Mbps
CRs-Fixed: 1088104
Change-Id: I251ed7938c29e08954d4c81d3041cb235a39d266
Signed-off-by: Subash Abhinov Kasiviswanathan <subashab@codeaurora.org>
Diffstat (limited to 'net/rmnet_data')
| -rw-r--r-- | net/rmnet_data/rmnet_data_vnd.c | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/net/rmnet_data/rmnet_data_vnd.c b/net/rmnet_data/rmnet_data_vnd.c index 4e3a205551e0..2999e2c15fdb 100644 --- a/net/rmnet_data/rmnet_data_vnd.c +++ b/net/rmnet_data/rmnet_data_vnd.c @@ -180,6 +180,7 @@ static netdev_tx_t rmnet_vnd_start_xmit(struct sk_buff *skb, rmnet_vnd_add_qos_header(skb, dev, dev_conf->qos_version); + skb_orphan(skb); rmnet_egress_handler(skb, &dev_conf->local_ep); } else { dev->stats.tx_dropped++; |