Hi, all
After this patch is applied, the TCP && UDP tests are made.
The TCP bandwidth is 939 Mbits/sec. The UDP bandwidth is 806 Mbits/sec.
So I think this patch can work well.
host1 <-----> host2
host1: forcedeth NIC
IP: 1.1.1.107
iperf -s
host2: forcedeth NIC
IP:1.1.1.105
iperf -c 1.1.1.107
The TCP Bandwidth is as below:
------------------------------------------------------------
Client connecting to 1.1.1.107, TCP port 5001
TCP window size: 85.0 KByte (default)
------------------------------------------------------------
[ 3] local 1.1.1.105 port 46092 connected with 1.1.1.107 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.0 sec 1.09 GBytes 939 Mbits/sec
The UDP is as below:
iperf -c 1.1.1.107 -u -b 1000m
------------------------------------------------------------
Client connecting to 1.1.1.107, UDP port 5001
Sending 1470 byte datagrams
UDP buffer size: 208 KByte (default)
------------------------------------------------------------
[ 3] local 1.1.1.105 port 47265 connected with 1.1.1.107 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.0 sec 964 MBytes 809 Mbits/sec
[ 3] Sent 687990 datagrams
[ 3] Server Report:
[ 3] 0.0-10.0 sec 960 MBytes 806 Mbits/sec 0.019 ms 2942/687989
(0.43%)
[ 3] 0.0-10.0 sec 1 datagrams received out-of-order
Zhu Yanjun
On 2017/9/8 20:28, Zhu Yanjun wrote:
The variable tx_stop is used to indicate the tx queue state: started
or stopped. In fact, the inline function netif_queue_stopped can do
the same work. So replace the variable tx_stop with the
function netif_queue_stopped.
Signed-off-by: Zhu Yanjun <yanjun....@oracle.com>
---
drivers/net/ethernet/nvidia/forcedeth.c | 13 ++++---------
1 file changed, 4 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/nvidia/forcedeth.c
b/drivers/net/ethernet/nvidia/forcedeth.c
index 994a83a..e6e0de4 100644
--- a/drivers/net/ethernet/nvidia/forcedeth.c
+++ b/drivers/net/ethernet/nvidia/forcedeth.c
@@ -834,7 +834,6 @@ struct fe_priv {
u32 tx_pkts_in_progress;
struct nv_skb_map *tx_change_owner;
struct nv_skb_map *tx_end_flip;
- int tx_stop;
/* TX software stats */
struct u64_stats_sync swstats_tx_syncp;
@@ -1939,7 +1938,6 @@ static void nv_init_tx(struct net_device *dev)
np->tx_pkts_in_progress = 0;
np->tx_change_owner = NULL;
np->tx_end_flip = NULL;
- np->tx_stop = 0;
for (i = 0; i < np->tx_ring_size; i++) {
if (!nv_optimized(np)) {
@@ -2211,7 +2209,6 @@ static netdev_tx_t nv_start_xmit(struct sk_buff *skb,
struct net_device *dev)
empty_slots = nv_get_empty_tx_slots(np);
if (unlikely(empty_slots <= entries)) {
netif_stop_queue(dev);
- np->tx_stop = 1;
spin_unlock_irqrestore(&np->lock, flags);
return NETDEV_TX_BUSY;
}
@@ -2359,7 +2356,6 @@ static netdev_tx_t nv_start_xmit_optimized(struct sk_buff
*skb,
empty_slots = nv_get_empty_tx_slots(np);
if (unlikely(empty_slots <= entries)) {
netif_stop_queue(dev);
- np->tx_stop = 1;
spin_unlock_irqrestore(&np->lock, flags);
return NETDEV_TX_BUSY;
}
@@ -2583,8 +2579,8 @@ static int nv_tx_done(struct net_device *dev, int limit)
netdev_completed_queue(np->dev, tx_work, bytes_compl);
- if (unlikely((np->tx_stop == 1) && (np->get_tx.orig != orig_get_tx))) {
- np->tx_stop = 0;
+ if (unlikely(netif_queue_stopped(dev) &&
+ (np->get_tx.orig != orig_get_tx))) {
netif_wake_queue(dev);
}
return tx_work;
@@ -2637,8 +2633,8 @@ static int nv_tx_done_optimized(struct net_device *dev,
int limit)
netdev_completed_queue(np->dev, tx_work, bytes_cleaned);
- if (unlikely((np->tx_stop == 1) && (np->get_tx.ex != orig_get_tx))) {
- np->tx_stop = 0;
+ if (unlikely(netif_queue_stopped(dev) &&
+ (np->get_tx.ex != orig_get_tx))) {
netif_wake_queue(dev);
}
return tx_work;
@@ -2724,7 +2720,6 @@ static void nv_tx_timeout(struct net_device *dev)
/* 2) complete any outstanding tx and do not give HW any limited tx
pkts */
saved_tx_limit = np->tx_limit;
np->tx_limit = 0; /* prevent giving HW any limited pkts */
- np->tx_stop = 0; /* prevent waking tx queue */
if (!nv_optimized(np))
nv_tx_done(dev, np->tx_ring_size);
else
