* [PATCH 1/2] tun: reserves space for network in skb
From: Stephen Hemminger @ 2011-06-09 0:33 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev
In-Reply-To: <20110609003306.651532958@vyatta.com>
[-- Attachment #1: tun-tap-pad.patch --]
[-- Type: text/plain, Size: 1127 bytes --]
The tun driver allocates skb's to hold data from user and then passes
the data into the network stack as received data. Most network devices
allocate the receive skb with routines like dev_alloc_skb() that reserves
additional space for use by network protocol stack but tun does not.
Because of the lack of padding, when the packet is passed through bridge
netfilter a new skb has to be allocated.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
--- a/drivers/net/tun.c 2011-05-13 12:37:11.619318207 -0700
+++ b/drivers/net/tun.c 2011-05-13 16:43:25.643040523 -0700
@@ -584,7 +584,7 @@ static __inline__ ssize_t tun_get_user(s
{
struct tun_pi pi = { 0, cpu_to_be16(ETH_P_IP) };
struct sk_buff *skb;
- size_t len = count, align = 0;
+ size_t len = count, align = NET_SKB_PAD;
struct virtio_net_hdr gso = { 0 };
int offset = 0;
@@ -614,7 +614,7 @@ static __inline__ ssize_t tun_get_user(s
}
if ((tun->flags & TUN_TYPE_MASK) == TUN_TAP_DEV) {
- align = NET_IP_ALIGN;
+ align += NET_IP_ALIGN;
if (unlikely(len < ETH_HLEN ||
(gso.hdr_len && gso.hdr_len < ETH_HLEN)))
return -EINVAL;
^ permalink raw reply
* [PATCH 2/2] tun: dont force inline of functions
From: Stephen Hemminger @ 2011-06-09 0:33 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev
In-Reply-To: <20110609003306.651532958@vyatta.com>
[-- Attachment #1: tun-tap-noinline.patch --]
[-- Type: text/plain, Size: 1708 bytes --]
Current standard practice is to not mark most functions as inline
and let compiler decide instead.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
--- a/drivers/net/tun.c 2011-05-13 16:43:25.643040523 -0700
+++ b/drivers/net/tun.c 2011-05-13 16:47:06.041228664 -0700
@@ -550,9 +550,9 @@ static unsigned int tun_chr_poll(struct
/* prepad is the amount to reserve at front. len is length after that.
* linear is a hint as to how much to copy (usually headers). */
-static inline struct sk_buff *tun_alloc_skb(struct tun_struct *tun,
- size_t prepad, size_t len,
- size_t linear, int noblock)
+static struct sk_buff *tun_alloc_skb(struct tun_struct *tun,
+ size_t prepad, size_t len,
+ size_t linear, int noblock)
{
struct sock *sk = tun->socket.sk;
struct sk_buff *skb;
@@ -578,9 +578,9 @@ static inline struct sk_buff *tun_alloc_
}
/* Get packet from user space buffer */
-static __inline__ ssize_t tun_get_user(struct tun_struct *tun,
- const struct iovec *iv, size_t count,
- int noblock)
+static ssize_t tun_get_user(struct tun_struct *tun,
+ const struct iovec *iv, size_t count,
+ int noblock)
{
struct tun_pi pi = { 0, cpu_to_be16(ETH_P_IP) };
struct sk_buff *skb;
@@ -729,9 +729,9 @@ static ssize_t tun_chr_aio_write(struct
}
/* Put packet to the user space buffer */
-static __inline__ ssize_t tun_put_user(struct tun_struct *tun,
- struct sk_buff *skb,
- const struct iovec *iv, int len)
+static ssize_t tun_put_user(struct tun_struct *tun,
+ struct sk_buff *skb,
+ const struct iovec *iv, int len)
{
struct tun_pi pi = { 0, skb->protocol };
ssize_t total = 0;
^ permalink raw reply
* [PATCH 8/9] niu: support 64 bit stats interface
From: Stephen Hemminger @ 2011-06-09 0:54 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev
In-Reply-To: <20110609005356.160260858@vyatta.com>
[-- Attachment #1: niu-stats.patch --]
[-- Type: text/plain, Size: 2244 bytes --]
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
--- a/drivers/net/niu.c 2011-06-07 16:58:31.333079418 -0700
+++ b/drivers/net/niu.c 2011-06-07 17:29:17.234232746 -0700
@@ -6249,9 +6249,10 @@ static void niu_sync_mac_stats(struct ni
niu_sync_bmac_stats(np);
}
-static void niu_get_rx_stats(struct niu *np)
+static void niu_get_rx_stats(struct niu *np,
+ struct rtnl_link_stats64 *stats)
{
- unsigned long pkts, dropped, errors, bytes;
+ u64 pkts, dropped, errors, bytes;
struct rx_ring_info *rx_rings;
int i;
@@ -6273,15 +6274,16 @@ static void niu_get_rx_stats(struct niu
}
no_rings:
- np->dev->stats.rx_packets = pkts;
- np->dev->stats.rx_bytes = bytes;
- np->dev->stats.rx_dropped = dropped;
- np->dev->stats.rx_errors = errors;
+ stats->rx_packets = pkts;
+ stats->rx_bytes = bytes;
+ stats->rx_dropped = dropped;
+ stats->rx_errors = errors;
}
-static void niu_get_tx_stats(struct niu *np)
+static void niu_get_tx_stats(struct niu *np,
+ struct rtnl_link_stats64 *stats)
{
- unsigned long pkts, errors, bytes;
+ u64 pkts, errors, bytes;
struct tx_ring_info *tx_rings;
int i;
@@ -6300,20 +6302,22 @@ static void niu_get_tx_stats(struct niu
}
no_rings:
- np->dev->stats.tx_packets = pkts;
- np->dev->stats.tx_bytes = bytes;
- np->dev->stats.tx_errors = errors;
+ stats->tx_packets = pkts;
+ stats->tx_bytes = bytes;
+ stats->tx_errors = errors;
}
-static struct net_device_stats *niu_get_stats(struct net_device *dev)
+static struct rtnl_link_stats64 *niu_get_stats(struct net_device *dev,
+ struct rtnl_link_stats64 *stats)
{
struct niu *np = netdev_priv(dev);
if (netif_running(dev)) {
- niu_get_rx_stats(np);
- niu_get_tx_stats(np);
+ niu_get_rx_stats(np, stats);
+ niu_get_tx_stats(np, stats);
}
- return &dev->stats;
+
+ return stats;
}
static void niu_load_hash_xmac(struct niu *np, u16 *hash)
@@ -9711,7 +9715,7 @@ static const struct net_device_ops niu_n
.ndo_open = niu_open,
.ndo_stop = niu_close,
.ndo_start_xmit = niu_start_xmit,
- .ndo_get_stats = niu_get_stats,
+ .ndo_get_stats64 = niu_get_stats,
.ndo_set_multicast_list = niu_set_rx_mode,
.ndo_validate_addr = eth_validate_addr,
.ndo_set_mac_address = niu_set_mac_addr,
^ permalink raw reply
* [PATCH 4/9] ifb: convert to 64 bit stats
From: Stephen Hemminger @ 2011-06-09 0:54 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev
In-Reply-To: <20110609005356.160260858@vyatta.com>
[-- Attachment #1: ifb-stats64.patch --]
[-- Type: text/plain, Size: 2495 bytes --]
Convert input functional block device to use 64 bit stats.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
--- a/drivers/net/ifb.c 2011-06-07 16:58:31.317079332 -0700
+++ b/drivers/net/ifb.c 2011-06-07 17:29:02.958161955 -0700
@@ -42,7 +42,14 @@ struct ifb_private {
struct tasklet_struct ifb_tasklet;
int tasklet_pending;
struct sk_buff_head rq;
+ u64 rx_packets;
+ u64 rx_bytes;
+ u64 rx_dropped;
+
struct sk_buff_head tq;
+ u64 tx_packets;
+ u64 tx_bytes;
+ u64 tx_dropped;
};
static int numifbs = 2;
@@ -57,7 +64,6 @@ static void ri_tasklet(unsigned long dev
struct net_device *_dev = (struct net_device *)dev;
struct ifb_private *dp = netdev_priv(_dev);
- struct net_device_stats *stats = &_dev->stats;
struct netdev_queue *txq;
struct sk_buff *skb;
@@ -77,15 +83,16 @@ static void ri_tasklet(unsigned long dev
skb->tc_verd = 0;
skb->tc_verd = SET_TC_NCLS(skb->tc_verd);
- stats->tx_packets++;
- stats->tx_bytes +=skb->len;
+
+ dp->tx_packets++;
+ dp->tx_bytes +=skb->len;
rcu_read_lock();
skb->dev = dev_get_by_index_rcu(&init_net, skb->skb_iif);
if (!skb->dev) {
rcu_read_unlock();
dev_kfree_skb(skb);
- stats->tx_dropped++;
+ dp->tx_dropped++;
if (skb_queue_len(&dp->tq) != 0)
goto resched;
break;
@@ -120,9 +127,26 @@ resched:
}
+static struct rtnl_link_stats64 *ifb_stats64(struct net_device *dev,
+ struct rtnl_link_stats64 *stats)
+{
+ struct ifb_private *dp = netdev_priv(dev);
+
+ stats->rx_packets = dp->rx_packets;
+ stats->rx_bytes = dp->rx_bytes;
+ stats->rx_dropped = dp->rx_dropped;
+ stats->tx_packets = dp->tx_packets;
+ stats->tx_bytes = dp->tx_bytes;
+ stats->tx_dropped = dp->tx_dropped;
+
+ return stats;
+}
+
+
static const struct net_device_ops ifb_netdev_ops = {
.ndo_open = ifb_open,
.ndo_stop = ifb_close,
+ .ndo_get_stats64 = ifb_stats64,
.ndo_start_xmit = ifb_xmit,
.ndo_validate_addr = eth_validate_addr,
};
@@ -153,15 +177,14 @@ static void ifb_setup(struct net_device
static netdev_tx_t ifb_xmit(struct sk_buff *skb, struct net_device *dev)
{
struct ifb_private *dp = netdev_priv(dev);
- struct net_device_stats *stats = &dev->stats;
u32 from = G_TC_FROM(skb->tc_verd);
- stats->rx_packets++;
- stats->rx_bytes+=skb->len;
+ dp->rx_packets++;
+ dp->rx_bytes+=skb->len;
if (!(from & (AT_INGRESS|AT_EGRESS)) || !skb->skb_iif) {
dev_kfree_skb(skb);
- stats->rx_dropped++;
+ dp->rx_dropped++;
return NETDEV_TX_OK;
}
^ permalink raw reply
* [PATCH 3/9] veth: convert to 64 bit statistics
From: Stephen Hemminger @ 2011-06-09 0:53 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev
In-Reply-To: <20110609005356.160260858@vyatta.com>
[-- Attachment #1: veth-stats64.patch --]
[-- Type: text/plain, Size: 2294 bytes --]
Not much change, device was already keeping per cpu statistics.
Use recent 64 statistics interface.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
--- a/drivers/net/veth.c 2011-06-07 20:04:42.221679879 +0900
+++ b/drivers/net/veth.c 2011-06-07 20:10:04.259276778 +0900
@@ -24,12 +24,12 @@
#define MAX_MTU 65535 /* Max L3 MTU (arbitrary) */
struct veth_net_stats {
- unsigned long rx_packets;
- unsigned long tx_packets;
- unsigned long rx_bytes;
- unsigned long tx_bytes;
- unsigned long tx_dropped;
- unsigned long rx_dropped;
+ u64 rx_packets;
+ u64 tx_packets;
+ u64 rx_bytes;
+ u64 tx_bytes;
+ u64 tx_dropped;
+ u64 rx_dropped;
};
struct veth_priv {
@@ -159,32 +159,27 @@ rx_drop:
* general routines
*/
-static struct net_device_stats *veth_get_stats(struct net_device *dev)
+static struct rtnl_link_stats64 *veth_get_stats64(struct net_device *dev,
+ struct rtnl_link_stats64 *tot)
{
struct veth_priv *priv;
int cpu;
- struct veth_net_stats *stats, total = {0};
+ struct veth_net_stats *stats;
priv = netdev_priv(dev);
for_each_possible_cpu(cpu) {
stats = per_cpu_ptr(priv->stats, cpu);
- total.rx_packets += stats->rx_packets;
- total.tx_packets += stats->tx_packets;
- total.rx_bytes += stats->rx_bytes;
- total.tx_bytes += stats->tx_bytes;
- total.tx_dropped += stats->tx_dropped;
- total.rx_dropped += stats->rx_dropped;
+ tot->rx_packets += stats->rx_packets;
+ tot->tx_packets += stats->tx_packets;
+ tot->rx_bytes += stats->rx_bytes;
+ tot->tx_bytes += stats->tx_bytes;
+ tot->tx_dropped += stats->tx_dropped;
+ tot->rx_dropped += stats->rx_dropped;
}
- dev->stats.rx_packets = total.rx_packets;
- dev->stats.tx_packets = total.tx_packets;
- dev->stats.rx_bytes = total.rx_bytes;
- dev->stats.tx_bytes = total.tx_bytes;
- dev->stats.tx_dropped = total.tx_dropped;
- dev->stats.rx_dropped = total.rx_dropped;
- return &dev->stats;
+ return tot;
}
static int veth_open(struct net_device *dev)
@@ -254,7 +249,7 @@ static const struct net_device_ops veth_
.ndo_stop = veth_close,
.ndo_start_xmit = veth_xmit,
.ndo_change_mtu = veth_change_mtu,
- .ndo_get_stats = veth_get_stats,
+ .ndo_get_stats64 = veth_get_stats64,
.ndo_set_mac_address = eth_mac_addr,
};
^ permalink raw reply
* [PATCH 5/9] netxen: convert to 64 bit statistics
From: Stephen Hemminger @ 2011-06-09 0:54 UTC (permalink / raw)
To: David S. Miller, Amit Kumar Salecha; +Cc: netdev
In-Reply-To: <20110609005356.160260858@vyatta.com>
[-- Attachment #1: netxen-getstats64.patch --]
[-- Type: text/plain, Size: 1752 bytes --]
Change to 64 bit statistics interface, driver was already maintaining 64 bit
value.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
--- a/drivers/net/netxen/netxen_nic_main.c 2011-06-07 20:33:48.150337471 +0900
+++ b/drivers/net/netxen/netxen_nic_main.c 2011-06-07 20:57:29.912628921 +0900
@@ -92,7 +92,8 @@ static irqreturn_t netxen_msi_intr(int i
static irqreturn_t netxen_msix_intr(int irq, void *data);
static void netxen_config_indev_addr(struct net_device *dev, unsigned long);
-static struct net_device_stats *netxen_nic_get_stats(struct net_device *netdev);
+static struct rtnl_link_stats64 *netxen_nic_get_stats(struct net_device *dev,
+ struct rtnl_link_stats64 *stats);
static int netxen_nic_set_mac(struct net_device *netdev, void *p);
/* PCI Device ID Table */
@@ -520,7 +521,7 @@ static const struct net_device_ops netxe
.ndo_open = netxen_nic_open,
.ndo_stop = netxen_nic_close,
.ndo_start_xmit = netxen_nic_xmit_frame,
- .ndo_get_stats = netxen_nic_get_stats,
+ .ndo_get_stats64 = netxen_nic_get_stats,
.ndo_validate_addr = eth_validate_addr,
.ndo_set_multicast_list = netxen_set_multicast_list,
.ndo_set_mac_address = netxen_nic_set_mac,
@@ -2110,10 +2111,10 @@ request_reset:
clear_bit(__NX_RESETTING, &adapter->state);
}
-static struct net_device_stats *netxen_nic_get_stats(struct net_device *netdev)
+static struct rtnl_link_stats64 *netxen_nic_get_stats(struct net_device *netdev,
+ struct rtnl_link_stats64 *stats)
{
struct netxen_adapter *adapter = netdev_priv(netdev);
- struct net_device_stats *stats = &netdev->stats;
stats->rx_packets = adapter->stats.rx_pkts + adapter->stats.lro_pkts;
stats->tx_packets = adapter->stats.xmitfinished;
^ permalink raw reply
* [PATCH 9/9] s2io: implement 64 bit stats
From: Stephen Hemminger @ 2011-06-09 0:54 UTC (permalink / raw)
To: David S. Miller, Jon Mason; +Cc: netdev
In-Reply-To: <20110609005356.160260858@vyatta.com>
[-- Attachment #1: s2io-stats.patch --]
[-- Type: text/plain, Size: 4085 bytes --]
Convert s2io driver to 64 bit statistics interface.
This driver keeps it private set of counters so change those
to the 64 bit netlink type.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
--- a/drivers/net/s2io.c 2011-06-07 17:35:37.372117743 -0700
+++ b/drivers/net/s2io.c 2011-06-07 17:44:14.102680070 -0700
@@ -4897,7 +4897,8 @@ static void s2io_updt_stats(struct s2io_
* Return value:
* pointer to the updated net_device_stats structure.
*/
-static struct net_device_stats *s2io_get_stats(struct net_device *dev)
+static struct rtnl_link_stats64 *s2io_get_stats(struct net_device *dev,
+ struct rtnl_link_stats64 *net_stats)
{
struct s2io_nic *sp = netdev_priv(dev);
struct mac_info *mac_control = &sp->mac_control;
@@ -4916,40 +4917,32 @@ static struct net_device_stats *s2io_get
*/
delta = ((u64) le32_to_cpu(stats->rmac_vld_frms_oflow) << 32 |
le32_to_cpu(stats->rmac_vld_frms)) - sp->stats.rx_packets;
- sp->stats.rx_packets += delta;
- dev->stats.rx_packets += delta;
+ sp->stats.tx_packets += delta;
delta = ((u64) le32_to_cpu(stats->tmac_frms_oflow) << 32 |
le32_to_cpu(stats->tmac_frms)) - sp->stats.tx_packets;
sp->stats.tx_packets += delta;
- dev->stats.tx_packets += delta;
delta = ((u64) le32_to_cpu(stats->rmac_data_octets_oflow) << 32 |
le32_to_cpu(stats->rmac_data_octets)) - sp->stats.rx_bytes;
sp->stats.rx_bytes += delta;
- dev->stats.rx_bytes += delta;
delta = ((u64) le32_to_cpu(stats->tmac_data_octets_oflow) << 32 |
le32_to_cpu(stats->tmac_data_octets)) - sp->stats.tx_bytes;
sp->stats.tx_bytes += delta;
- dev->stats.tx_bytes += delta;
delta = le64_to_cpu(stats->rmac_drop_frms) - sp->stats.rx_errors;
sp->stats.rx_errors += delta;
- dev->stats.rx_errors += delta;
delta = ((u64) le32_to_cpu(stats->tmac_any_err_frms_oflow) << 32 |
le32_to_cpu(stats->tmac_any_err_frms)) - sp->stats.tx_errors;
sp->stats.tx_errors += delta;
- dev->stats.tx_errors += delta;
delta = le64_to_cpu(stats->rmac_drop_frms) - sp->stats.rx_dropped;
sp->stats.rx_dropped += delta;
- dev->stats.rx_dropped += delta;
delta = le64_to_cpu(stats->tmac_drop_frms) - sp->stats.tx_dropped;
sp->stats.tx_dropped += delta;
- dev->stats.tx_dropped += delta;
/* The adapter MAC interprets pause frames as multicast packets, but
* does not pass them up. This erroneously increases the multicast
@@ -4961,19 +4954,16 @@ static struct net_device_stats *s2io_get
delta -= le64_to_cpu(stats->rmac_pause_ctrl_frms);
delta -= sp->stats.multicast;
sp->stats.multicast += delta;
- dev->stats.multicast += delta;
delta = ((u64) le32_to_cpu(stats->rmac_usized_frms_oflow) << 32 |
le32_to_cpu(stats->rmac_usized_frms)) +
le64_to_cpu(stats->rmac_long_frms) - sp->stats.rx_length_errors;
sp->stats.rx_length_errors += delta;
- dev->stats.rx_length_errors += delta;
delta = le64_to_cpu(stats->rmac_fcs_err_frms) - sp->stats.rx_crc_errors;
sp->stats.rx_crc_errors += delta;
- dev->stats.rx_crc_errors += delta;
- return &dev->stats;
+ return &sp->stats;
}
/**
@@ -7447,7 +7437,7 @@ static int rx_osm_handler(struct ring_in
if (err_mask != 0x5) {
DBG_PRINT(ERR_DBG, "%s: Rx error Value: 0x%x\n",
dev->name, err_mask);
- dev->stats.rx_crc_errors++;
+ sp->stats.rx_crc_errors++;
swstats->mem_freed
+= skb->truesize;
dev_kfree_skb(skb);
@@ -7731,7 +7721,7 @@ static int rts_ds_steer(struct s2io_nic
static const struct net_device_ops s2io_netdev_ops = {
.ndo_open = s2io_open,
.ndo_stop = s2io_close,
- .ndo_get_stats = s2io_get_stats,
+ .ndo_get_stats64 = s2io_get_stats,
.ndo_start_xmit = s2io_xmit,
.ndo_validate_addr = eth_validate_addr,
.ndo_set_multicast_list = s2io_set_multicast,
--- a/drivers/net/s2io.h 2011-06-07 17:38:50.921077503 -0700
+++ b/drivers/net/s2io.h 2011-06-07 17:43:59.478607560 -0700
@@ -871,7 +871,7 @@ struct s2io_nic {
struct mac_addr def_mac_addr[256];
- struct net_device_stats stats;
+ struct rtnl_link_stats64 stats;
int high_dma_flag;
int device_enabled_once;
^ permalink raw reply
* [PATCH 7/9] myricom: update to 64 bit stats
From: Stephen Hemminger @ 2011-06-09 0:54 UTC (permalink / raw)
To: David S. Miller, Andrew Gallatin, Brice Goglin; +Cc: netdev
In-Reply-To: <20110609005356.160260858@vyatta.com>
[-- Attachment #1: myricom-stats.patch --]
[-- Type: text/plain, Size: 2184 bytes --]
Driver was already keeping 64 bit counters, just not using the new interface.
Ps: IMHO drivers should not be duplicating network device
stats into ethtool stats. It is useless duplication.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
--- a/drivers/net/myri10ge/myri10ge.c 2011-06-07 16:58:31.333079418 -0700
+++ b/drivers/net/myri10ge/myri10ge.c 2011-06-07 17:29:13.298213223 -0700
@@ -377,7 +377,8 @@ static inline void put_be32(__be32 val,
__raw_writel((__force __u32) val, (__force void __iomem *)p);
}
-static struct net_device_stats *myri10ge_get_stats(struct net_device *dev);
+static struct rtnl_link_stats64 *myri10ge_get_stats(struct net_device *dev,
+ struct rtnl_link_stats64 *stats);
static void set_fw_name(struct myri10ge_priv *mgp, char *name, bool allocated)
{
@@ -1831,13 +1832,14 @@ myri10ge_get_ethtool_stats(struct net_de
{
struct myri10ge_priv *mgp = netdev_priv(netdev);
struct myri10ge_slice_state *ss;
+ struct rtnl_link_stats64 link_stats;
int slice;
int i;
/* force stats update */
- (void)myri10ge_get_stats(netdev);
+ (void)myri10ge_get_stats(netdev, &link_stats);
for (i = 0; i < MYRI10GE_NET_STATS_LEN; i++)
- data[i] = ((unsigned long *)&netdev->stats)[i];
+ data[i] = ((u64 *)&link_stats)[i];
data[i++] = (unsigned int)mgp->tx_boundary;
data[i++] = (unsigned int)mgp->wc_enabled;
@@ -2976,11 +2978,11 @@ drop:
return NETDEV_TX_OK;
}
-static struct net_device_stats *myri10ge_get_stats(struct net_device *dev)
+static struct rtnl_link_stats64 *myri10ge_get_stats(struct net_device *dev,
+ struct rtnl_link_stats64 *stats)
{
struct myri10ge_priv *mgp = netdev_priv(dev);
struct myri10ge_slice_netstats *slice_stats;
- struct net_device_stats *stats = &dev->stats;
int i;
spin_lock(&mgp->stats_lock);
@@ -3790,7 +3792,7 @@ static const struct net_device_ops myri1
.ndo_open = myri10ge_open,
.ndo_stop = myri10ge_close,
.ndo_start_xmit = myri10ge_xmit,
- .ndo_get_stats = myri10ge_get_stats,
+ .ndo_get_stats64 = myri10ge_get_stats,
.ndo_validate_addr = eth_validate_addr,
.ndo_change_mtu = myri10ge_change_mtu,
.ndo_fix_features = myri10ge_fix_features,
^ permalink raw reply
* [PATCH 1/9] vmxnet3: convert to 64 bit stats interface
From: Stephen Hemminger @ 2011-06-09 0:53 UTC (permalink / raw)
To: David S. Miller, Shreyas Bhatewara; +Cc: netdev, pv-drivers
In-Reply-To: <20110609005356.160260858@vyatta.com>
[-- Attachment #1: vmxnet3-stats64.patch --]
[-- Type: text/plain, Size: 4534 bytes --]
Convert vmxnet3 driver to 64 bit statistics interface.
This driver was already counting packet per queue in a 64 bit value so not
a huge change. Eliminate unused old net_device_stats structure.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
--- a/drivers/net/vmxnet3/vmxnet3_drv.c 2011-06-04 22:01:31.932389206 +0900
+++ b/drivers/net/vmxnet3/vmxnet3_drv.c 2011-06-07 19:50:23.189420166 +0900
@@ -2864,7 +2864,7 @@ vmxnet3_probe_device(struct pci_dev *pde
.ndo_set_mac_address = vmxnet3_set_mac_addr,
.ndo_change_mtu = vmxnet3_change_mtu,
.ndo_set_features = vmxnet3_set_features,
- .ndo_get_stats = vmxnet3_get_stats,
+ .ndo_get_stats64 = vmxnet3_get_stats64,
.ndo_tx_timeout = vmxnet3_tx_timeout,
.ndo_set_multicast_list = vmxnet3_set_mc,
.ndo_vlan_rx_register = vmxnet3_vlan_rx_register,
--- a/drivers/net/vmxnet3/vmxnet3_ethtool.c 2011-06-04 21:54:27.110282630 +0900
+++ b/drivers/net/vmxnet3/vmxnet3_ethtool.c 2011-06-07 20:13:34.492319266 +0900
@@ -113,15 +113,15 @@ vmxnet3_global_stats[] = {
};
-struct net_device_stats *
-vmxnet3_get_stats(struct net_device *netdev)
+struct rtnl_link_stats64 *
+vmxnet3_get_stats64(struct net_device *netdev,
+ struct rtnl_link_stats64 *stats)
{
struct vmxnet3_adapter *adapter;
struct vmxnet3_tq_driver_stats *drvTxStats;
struct vmxnet3_rq_driver_stats *drvRxStats;
struct UPT1_TxStats *devTxStats;
struct UPT1_RxStats *devRxStats;
- struct net_device_stats *net_stats = &netdev->stats;
unsigned long flags;
int i;
@@ -132,36 +132,36 @@ vmxnet3_get_stats(struct net_device *net
VMXNET3_WRITE_BAR1_REG(adapter, VMXNET3_REG_CMD, VMXNET3_CMD_GET_STATS);
spin_unlock_irqrestore(&adapter->cmd_lock, flags);
- memset(net_stats, 0, sizeof(*net_stats));
for (i = 0; i < adapter->num_tx_queues; i++) {
devTxStats = &adapter->tqd_start[i].stats;
drvTxStats = &adapter->tx_queue[i].stats;
- net_stats->tx_packets += devTxStats->ucastPktsTxOK +
- devTxStats->mcastPktsTxOK +
- devTxStats->bcastPktsTxOK;
- net_stats->tx_bytes += devTxStats->ucastBytesTxOK +
- devTxStats->mcastBytesTxOK +
- devTxStats->bcastBytesTxOK;
- net_stats->tx_errors += devTxStats->pktsTxError;
- net_stats->tx_dropped += drvTxStats->drop_total;
+ stats->tx_packets += devTxStats->ucastPktsTxOK +
+ devTxStats->mcastPktsTxOK +
+ devTxStats->bcastPktsTxOK;
+ stats->tx_bytes += devTxStats->ucastBytesTxOK +
+ devTxStats->mcastBytesTxOK +
+ devTxStats->bcastBytesTxOK;
+ stats->tx_errors += devTxStats->pktsTxError;
+ stats->tx_dropped += drvTxStats->drop_total;
}
for (i = 0; i < adapter->num_rx_queues; i++) {
devRxStats = &adapter->rqd_start[i].stats;
drvRxStats = &adapter->rx_queue[i].stats;
- net_stats->rx_packets += devRxStats->ucastPktsRxOK +
- devRxStats->mcastPktsRxOK +
- devRxStats->bcastPktsRxOK;
-
- net_stats->rx_bytes += devRxStats->ucastBytesRxOK +
- devRxStats->mcastBytesRxOK +
- devRxStats->bcastBytesRxOK;
-
- net_stats->rx_errors += devRxStats->pktsRxError;
- net_stats->rx_dropped += drvRxStats->drop_total;
- net_stats->multicast += devRxStats->mcastPktsRxOK;
+ stats->rx_packets += devRxStats->ucastPktsRxOK +
+ devRxStats->mcastPktsRxOK +
+ devRxStats->bcastPktsRxOK;
+
+ stats->rx_bytes += devRxStats->ucastBytesRxOK +
+ devRxStats->mcastBytesRxOK +
+ devRxStats->bcastBytesRxOK;
+
+ stats->rx_errors += devRxStats->pktsRxError;
+ stats->rx_dropped += drvRxStats->drop_total;
+ stats->multicast += devRxStats->mcastPktsRxOK;
}
- return net_stats;
+
+ return stats;
}
static int
--- a/drivers/net/vmxnet3/vmxnet3_int.h 2011-06-04 22:02:03.048543506 +0900
+++ b/drivers/net/vmxnet3/vmxnet3_int.h 2011-06-07 19:54:22.890608779 +0900
@@ -323,7 +323,6 @@ struct vmxnet3_adapter {
struct Vmxnet3_TxQueueDesc *tqd_start; /* all tx queue desc */
struct Vmxnet3_RxQueueDesc *rqd_start; /* all rx queue desc */
struct net_device *netdev;
- struct net_device_stats net_stats;
struct pci_dev *pdev;
u8 __iomem *hw_addr0; /* for BAR 0 */
@@ -407,7 +406,9 @@ vmxnet3_create_queues(struct vmxnet3_ada
u32 tx_ring_size, u32 rx_ring_size, u32 rx_ring2_size);
extern void vmxnet3_set_ethtool_ops(struct net_device *netdev);
-extern struct net_device_stats *vmxnet3_get_stats(struct net_device *netdev);
+
+extern struct rtnl_link_stats64 *
+vmxnet3_get_stats64(struct net_device *dev, struct rtnl_link_stats64 *stats);
extern char vmxnet3_driver_name[];
#endif
^ permalink raw reply
* [PATCH 6/9] enic: update to support 64 bit stats
From: Stephen Hemminger @ 2011-06-09 0:54 UTC (permalink / raw)
To: David S. Miller, Christian Benvenuti, Vasanthy Kolluri,
Roopa Prabhu, David Wang
Cc: netdev
In-Reply-To: <20110609005356.160260858@vyatta.com>
[-- Attachment #1: enic-stats64.patch --]
[-- Type: text/plain, Size: 1503 bytes --]
The device driver already uses 64 bit statistics, it just
doesn't use the 64 bit interface.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
--- a/drivers/net/enic/enic_main.c 2011-06-07 16:58:31.317079332 -0700
+++ b/drivers/net/enic/enic_main.c 2011-06-07 17:29:09.670195233 -0700
@@ -801,10 +801,10 @@ static netdev_tx_t enic_hard_start_xmit(
}
/* dev_base_lock rwlock held, nominally process context */
-static struct net_device_stats *enic_get_stats(struct net_device *netdev)
+static struct rtnl_link_stats64 *enic_get_stats(struct net_device *netdev,
+ struct rtnl_link_stats64 *net_stats)
{
struct enic *enic = netdev_priv(netdev);
- struct net_device_stats *net_stats = &netdev->stats;
struct vnic_stats *stats;
enic_dev_stats_dump(enic, &stats);
@@ -2117,7 +2117,7 @@ static const struct net_device_ops enic_
.ndo_open = enic_open,
.ndo_stop = enic_stop,
.ndo_start_xmit = enic_hard_start_xmit,
- .ndo_get_stats = enic_get_stats,
+ .ndo_get_stats64 = enic_get_stats,
.ndo_validate_addr = eth_validate_addr,
.ndo_set_rx_mode = enic_set_rx_mode,
.ndo_set_multicast_list = enic_set_rx_mode,
@@ -2139,7 +2139,7 @@ static const struct net_device_ops enic_
.ndo_open = enic_open,
.ndo_stop = enic_stop,
.ndo_start_xmit = enic_hard_start_xmit,
- .ndo_get_stats = enic_get_stats,
+ .ndo_get_stats64 = enic_get_stats,
.ndo_validate_addr = eth_validate_addr,
.ndo_set_mac_address = enic_set_mac_address,
.ndo_set_rx_mode = enic_set_rx_mode,
^ permalink raw reply
* Re: [RFC] breakage in sysfs_readdir() and s_instances abuse in sysfs
From: Al Viro @ 2011-06-09 1:26 UTC (permalink / raw)
To: Eric W. Biederman; +Cc: linux-fsdevel, Linus Torvalds, netdev, Linux Containers
In-Reply-To: <20110607225902.GS11521@ZenIV.linux.org.uk>
On Tue, Jun 07, 2011 at 11:59:02PM +0100, Al Viro wrote:
> Completely untested patch follows:
FWIW, modulo a couple of brainos it seems to work here without blowing
everything up. Actual freeing of struct net is delayed until after
umount of sysfs instances refering it, shutdown is *not* delayed, sysfs
entries are removed nicely as they ought to, no objects from other net_ns
are picked by readdir or lookup...
If somebody has objections - yell. If I don't hear anything, the following
goes to -next:
commit 72d4e98002b45598bb88af74eeac20874f2789be
Author: Al Viro <viro@zeniv.linux.org.uk>
Date: Wed Jun 8 21:13:01 2011 -0400
Delay struct net freeing while there's a sysfs instance refering to it
* new refcount in struct net, controlling actual freeing of the memory
* new method in kobj_ns_type_operations (->put_ns())
* ->current_ns() semantics change - it's supposed to be followed by
corresponding ->put_ns(). For struct net in case of CONFIG_NET_NS it bumps
the new refcount; net_put_ns() decrements it and calls net_free() if the
last reference has been dropped.
* old net_free() callers call net_put_ns() instead.
* sysfs_exit_ns() is gone, along with a large part of callchain
leading to it; now that the references stored in ->ns[...] stay valid we
do not need to hunt them down and replace them with NULL. That fixes
problems in sysfs_lookup() and sysfs_readdir(), along with getting rid
of sb->s_instances abuse.
Note that struct net *shutdown* logics has not changed - net_cleanup()
is called exactly when it used to be called. The only thing postponed by
having a sysfs instance refering to that struct net is actual freeing of
memory occupied by struct net.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
diff --git a/fs/sysfs/mount.c b/fs/sysfs/mount.c
index 2668957..1e7a2ee 100644
--- a/fs/sysfs/mount.c
+++ b/fs/sysfs/mount.c
@@ -111,8 +111,11 @@ static struct dentry *sysfs_mount(struct file_system_type *fs_type,
info->ns[type] = kobj_ns_current(type);
sb = sget(fs_type, sysfs_test_super, sysfs_set_super, info);
- if (IS_ERR(sb) || sb->s_fs_info != info)
+ if (IS_ERR(sb) || sb->s_fs_info != info) {
+ for (type = KOBJ_NS_TYPE_NONE; type < KOBJ_NS_TYPES; type++)
+ kobj_ns_put(type, info->ns[type]);
kfree(info);
+ }
if (IS_ERR(sb))
return ERR_CAST(sb);
if (!sb->s_root) {
@@ -131,11 +134,14 @@ static struct dentry *sysfs_mount(struct file_system_type *fs_type,
static void sysfs_kill_sb(struct super_block *sb)
{
struct sysfs_super_info *info = sysfs_info(sb);
+ int type;
/* Remove the superblock from fs_supers/s_instances
* so we can't find it, before freeing sysfs_super_info.
*/
kill_anon_super(sb);
+ for (type = KOBJ_NS_TYPE_NONE; type < KOBJ_NS_TYPES; type++)
+ kobj_ns_put(type, info->ns[type]);
kfree(info);
}
@@ -145,28 +151,6 @@ static struct file_system_type sysfs_fs_type = {
.kill_sb = sysfs_kill_sb,
};
-void sysfs_exit_ns(enum kobj_ns_type type, const void *ns)
-{
- struct super_block *sb;
-
- mutex_lock(&sysfs_mutex);
- spin_lock(&sb_lock);
- list_for_each_entry(sb, &sysfs_fs_type.fs_supers, s_instances) {
- struct sysfs_super_info *info = sysfs_info(sb);
- /*
- * If we see a superblock on the fs_supers/s_instances
- * list the unmount has not completed and sb->s_fs_info
- * points to a valid struct sysfs_super_info.
- */
- /* Ignore superblocks with the wrong ns */
- if (info->ns[type] != ns)
- continue;
- info->ns[type] = NULL;
- }
- spin_unlock(&sb_lock);
- mutex_unlock(&sysfs_mutex);
-}
-
int __init sysfs_init(void)
{
int err = -ENOMEM;
diff --git a/fs/sysfs/sysfs.h b/fs/sysfs/sysfs.h
index 3d28af3..2ed2404 100644
--- a/fs/sysfs/sysfs.h
+++ b/fs/sysfs/sysfs.h
@@ -136,7 +136,7 @@ struct sysfs_addrm_cxt {
* instance).
*/
struct sysfs_super_info {
- const void *ns[KOBJ_NS_TYPES];
+ void *ns[KOBJ_NS_TYPES];
};
#define sysfs_info(SB) ((struct sysfs_super_info *)(SB->s_fs_info))
extern struct sysfs_dirent sysfs_root;
diff --git a/include/linux/kobject_ns.h b/include/linux/kobject_ns.h
index 82cb5bf..5fa481c 100644
--- a/include/linux/kobject_ns.h
+++ b/include/linux/kobject_ns.h
@@ -38,9 +38,10 @@ enum kobj_ns_type {
*/
struct kobj_ns_type_operations {
enum kobj_ns_type type;
- const void *(*current_ns)(void);
+ void *(*current_ns)(void);
const void *(*netlink_ns)(struct sock *sk);
const void *(*initial_ns)(void);
+ void (*put_ns)(void *);
};
int kobj_ns_type_register(const struct kobj_ns_type_operations *ops);
@@ -48,9 +49,9 @@ int kobj_ns_type_registered(enum kobj_ns_type type);
const struct kobj_ns_type_operations *kobj_child_ns_ops(struct kobject *parent);
const struct kobj_ns_type_operations *kobj_ns_ops(struct kobject *kobj);
-const void *kobj_ns_current(enum kobj_ns_type type);
+void *kobj_ns_current(enum kobj_ns_type type);
const void *kobj_ns_netlink(enum kobj_ns_type type, struct sock *sk);
const void *kobj_ns_initial(enum kobj_ns_type type);
-void kobj_ns_exit(enum kobj_ns_type type, const void *ns);
+void kobj_ns_put(enum kobj_ns_type type, void *ns);
#endif /* _LINUX_KOBJECT_NS_H */
diff --git a/include/linux/sysfs.h b/include/linux/sysfs.h
index c3acda6..e2696d7 100644
--- a/include/linux/sysfs.h
+++ b/include/linux/sysfs.h
@@ -177,9 +177,6 @@ struct sysfs_dirent *sysfs_get_dirent(struct sysfs_dirent *parent_sd,
struct sysfs_dirent *sysfs_get(struct sysfs_dirent *sd);
void sysfs_put(struct sysfs_dirent *sd);
-/* Called to clear a ns tag when it is no longer valid */
-void sysfs_exit_ns(enum kobj_ns_type type, const void *tag);
-
int __must_check sysfs_init(void);
#else /* CONFIG_SYSFS */
@@ -338,10 +335,6 @@ static inline void sysfs_put(struct sysfs_dirent *sd)
{
}
-static inline void sysfs_exit_ns(int type, const void *tag)
-{
-}
-
static inline int __must_check sysfs_init(void)
{
return 0;
diff --git a/include/net/net_namespace.h b/include/net/net_namespace.h
index 2bf9ed9..415af64 100644
--- a/include/net/net_namespace.h
+++ b/include/net/net_namespace.h
@@ -35,8 +35,11 @@ struct netns_ipvs;
#define NETDEV_HASHENTRIES (1 << NETDEV_HASHBITS)
struct net {
+ atomic_t passive; /* To decided when the network
+ * namespace should be freed.
+ */
atomic_t count; /* To decided when the network
- * namespace should be freed.
+ * namespace should be shut down.
*/
#ifdef NETNS_REFCNT_DEBUG
atomic_t use_count; /* To track references we
@@ -154,6 +157,9 @@ int net_eq(const struct net *net1, const struct net *net2)
{
return net1 == net2;
}
+
+extern void net_put_ns(void *);
+
#else
static inline struct net *get_net(struct net *net)
@@ -175,6 +181,8 @@ int net_eq(const struct net *net1, const struct net *net2)
{
return 1;
}
+
+#define net_put_ns NULL
#endif
diff --git a/lib/kobject.c b/lib/kobject.c
index 82dc34c..43116b2 100644
--- a/lib/kobject.c
+++ b/lib/kobject.c
@@ -948,9 +948,9 @@ const struct kobj_ns_type_operations *kobj_ns_ops(struct kobject *kobj)
}
-const void *kobj_ns_current(enum kobj_ns_type type)
+void *kobj_ns_current(enum kobj_ns_type type)
{
- const void *ns = NULL;
+ void *ns = NULL;
spin_lock(&kobj_ns_type_lock);
if ((type > KOBJ_NS_TYPE_NONE) && (type < KOBJ_NS_TYPES) &&
@@ -987,23 +987,15 @@ const void *kobj_ns_initial(enum kobj_ns_type type)
return ns;
}
-/*
- * kobj_ns_exit - invalidate a namespace tag
- *
- * @type: the namespace type (i.e. KOBJ_NS_TYPE_NET)
- * @ns: the actual namespace being invalidated
- *
- * This is called when a tag is no longer valid. For instance,
- * when a network namespace exits, it uses this helper to
- * make sure no sb's sysfs_info points to the now-invalidated
- * netns.
- */
-void kobj_ns_exit(enum kobj_ns_type type, const void *ns)
+void kobj_ns_put(enum kobj_ns_type type, void *ns)
{
- sysfs_exit_ns(type, ns);
+ spin_lock(&kobj_ns_type_lock);
+ if ((type > KOBJ_NS_TYPE_NONE) && (type < KOBJ_NS_TYPES) &&
+ kobj_ns_ops_tbl[type] && kobj_ns_ops_tbl[type]->put_ns)
+ kobj_ns_ops_tbl[type]->put_ns(ns);
+ spin_unlock(&kobj_ns_type_lock);
}
-
EXPORT_SYMBOL(kobject_get);
EXPORT_SYMBOL(kobject_put);
EXPORT_SYMBOL(kobject_del);
diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c
index 11b98bc..0c799f1 100644
--- a/net/core/net-sysfs.c
+++ b/net/core/net-sysfs.c
@@ -1179,9 +1179,14 @@ static void remove_queue_kobjects(struct net_device *net)
#endif
}
-static const void *net_current_ns(void)
+static void *net_current_ns(void)
{
- return current->nsproxy->net_ns;
+ struct net *ns = current->nsproxy->net_ns;
+#ifdef CONFIG_NET_NS
+ if (ns)
+ atomic_inc(&ns->passive);
+#endif
+ return ns;
}
static const void *net_initial_ns(void)
@@ -1199,19 +1204,10 @@ struct kobj_ns_type_operations net_ns_type_operations = {
.current_ns = net_current_ns,
.netlink_ns = net_netlink_ns,
.initial_ns = net_initial_ns,
+ .put_ns = net_put_ns,
};
EXPORT_SYMBOL_GPL(net_ns_type_operations);
-static void net_kobj_ns_exit(struct net *net)
-{
- kobj_ns_exit(KOBJ_NS_TYPE_NET, net);
-}
-
-static struct pernet_operations kobj_net_ops = {
- .exit = net_kobj_ns_exit,
-};
-
-
#ifdef CONFIG_HOTPLUG
static int netdev_uevent(struct device *d, struct kobj_uevent_env *env)
{
@@ -1339,6 +1335,5 @@ EXPORT_SYMBOL(netdev_class_remove_file);
int netdev_kobject_init(void)
{
kobj_ns_type_register(&net_ns_type_operations);
- register_pernet_subsys(&kobj_net_ops);
return class_register(&net_class);
}
diff --git a/net/core/net_namespace.c b/net/core/net_namespace.c
index 6c6b86d..19feb20 100644
--- a/net/core/net_namespace.c
+++ b/net/core/net_namespace.c
@@ -128,6 +128,7 @@ static __net_init int setup_net(struct net *net)
LIST_HEAD(net_exit_list);
atomic_set(&net->count, 1);
+ atomic_set(&net->passive, 1);
#ifdef NETNS_REFCNT_DEBUG
atomic_set(&net->use_count, 0);
@@ -210,6 +211,13 @@ static void net_free(struct net *net)
kmem_cache_free(net_cachep, net);
}
+void net_put_ns(void *p)
+{
+ struct net *ns = p;
+ if (ns && atomic_dec_and_test(&ns->passive))
+ net_free(ns);
+}
+
struct net *copy_net_ns(unsigned long flags, struct net *old_net)
{
struct net *net;
@@ -230,7 +238,7 @@ struct net *copy_net_ns(unsigned long flags, struct net *old_net)
}
mutex_unlock(&net_mutex);
if (rv < 0) {
- net_free(net);
+ net_put_ns(net);
return ERR_PTR(rv);
}
return net;
@@ -286,7 +294,7 @@ static void cleanup_net(struct work_struct *work)
/* Finally it is safe to free my network namespace structure */
list_for_each_entry_safe(net, tmp, &net_exit_list, exit_list) {
list_del_init(&net->exit_list);
- net_free(net);
+ net_put_ns(net);
}
}
static DECLARE_WORK(net_cleanup_work, cleanup_net);
^ permalink raw reply related
* Re: [PATCH 9/9] s2io: implement 64 bit stats
From: Ben Hutchings @ 2011-06-09 1:52 UTC (permalink / raw)
To: Stephen Hemminger; +Cc: David S. Miller, Jon Mason, netdev
In-Reply-To: <20110609005417.817071451@vyatta.com>
On Wed, 2011-06-08 at 17:54 -0700, Stephen Hemminger wrote:
> plain text document attachment (s2io-stats.patch)
> Convert s2io driver to 64 bit statistics interface.
> This driver keeps it private set of counters so change those
> to the 64 bit netlink type.
>
> Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
>
>
> --- a/drivers/net/s2io.c 2011-06-07 17:35:37.372117743 -0700
> +++ b/drivers/net/s2io.c 2011-06-07 17:44:14.102680070 -0700
> @@ -4897,7 +4897,8 @@ static void s2io_updt_stats(struct s2io_
> * Return value:
> * pointer to the updated net_device_stats structure.
> */
> -static struct net_device_stats *s2io_get_stats(struct net_device *dev)
> +static struct rtnl_link_stats64 *s2io_get_stats(struct net_device *dev,
> + struct rtnl_link_stats64 *net_stats)
> {
> struct s2io_nic *sp = netdev_priv(dev);
> struct mac_info *mac_control = &sp->mac_control;
> @@ -4916,40 +4917,32 @@ static struct net_device_stats *s2io_get
> */
> delta = ((u64) le32_to_cpu(stats->rmac_vld_frms_oflow) << 32 |
> le32_to_cpu(stats->rmac_vld_frms)) - sp->stats.rx_packets;
> - sp->stats.rx_packets += delta;
> - dev->stats.rx_packets += delta;
> + sp->stats.tx_packets += delta;
This is now adding to tx_packets, not rx_packets.
> delta = ((u64) le32_to_cpu(stats->tmac_frms_oflow) << 32 |
> le32_to_cpu(stats->tmac_frms)) - sp->stats.tx_packets;
> sp->stats.tx_packets += delta;
> - dev->stats.tx_packets += delta;
>
> delta = ((u64) le32_to_cpu(stats->rmac_data_octets_oflow) << 32 |
> le32_to_cpu(stats->rmac_data_octets)) - sp->stats.rx_bytes;
> sp->stats.rx_bytes += delta;
> - dev->stats.rx_bytes += delta;
[...]
It seems to me that the delta calculations and sp->stats are no longer
necessary. These can be just:
net_stats->rx_packets = ((u64) le32_to_cpu(stats->rmac_vld_frms_oflow) << 32 |
le32_to_cpu(stats->rmac_vld_frms));
net_stats->tx_packets = ((u64) le32_to_cpu(stats->tmac_frms_oflow) << 32 |
le32_to_cpu(stats->tmac_frms));
net_stats->rx_bytes = ((u64) le32_to_cpu(stats->rmac_data_octets_oflow) << 32 |
le32_to_cpu(stats->rmac_data_octets));
...
[...]
> @@ -7447,7 +7437,7 @@ static int rx_osm_handler(struct ring_in
> if (err_mask != 0x5) {
> DBG_PRINT(ERR_DBG, "%s: Rx error Value: 0x%x\n",
> dev->name, err_mask);
> - dev->stats.rx_crc_errors++;
> + sp->stats.rx_crc_errors++;
[...]
This won't be atomic on 32-bit systems. If this isn't double-counting
CRC errors (I suspect it is) then there needs to be a separate unsigned
long counter that's folded in by s2io_get_stats().
Ben.
--
Ben Hutchings, Senior Software Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply
* Re: [PATCH 2/9] xen: convert to 64 bit stats interface
From: Ben Hutchings @ 2011-06-09 1:56 UTC (permalink / raw)
To: Stephen Hemminger; +Cc: David S. Miller, Jeremy Fitzhardinge, netdev, xen-devel
In-Reply-To: <20110609005417.385577995@vyatta.com>
On Wed, 2011-06-08 at 17:53 -0700, Stephen Hemminger wrote:
> Convert xen driver to 64 bit statistics interface.
> This driver was already counting packet per queue in a 64 bit value so not
> a huge change.
[...]
I think this driver will need to use u64_stats_sync.
Ben.
--
Ben Hutchings, Senior Software Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply
* Re: [PATCH 3/9] veth: convert to 64 bit statistics
From: Ben Hutchings @ 2011-06-09 2:00 UTC (permalink / raw)
To: Stephen Hemminger; +Cc: David S. Miller, netdev
In-Reply-To: <20110609005417.449670103@vyatta.com>
On Wed, 2011-06-08 at 17:53 -0700, Stephen Hemminger wrote:
> Not much change, device was already keeping per cpu statistics.
> Use recent 64 statistics interface.
[...]
It's also going to need to use u64_stats_sync functions.
Ben.
--
Ben Hutchings, Senior Software Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply
* [Patch] netpoll: prevent netpoll setup on slave devices
From: Amerigo Wang @ 2011-06-09 2:42 UTC (permalink / raw)
To: netdev; +Cc: WANG Cong, Neil Horman
In commit 8d8fc29d02a33e4bd5f4fa47823c1fd386346093
(netpoll: disable netpoll when enslave a device), we automatically
disable netpoll when the underlying device is being enslaved,
we also need to prevent people from setuping netpoll on
devices that are already enslaved.
Signed-off-by: WANG Cong <amwang@redhat.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
---
diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index 2d7d6d4..42ea4b0 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -792,6 +792,12 @@ int netpoll_setup(struct netpoll *np)
return -ENODEV;
}
+ if (ndev->master) {
+ printk(KERN_ERR "%s: %s is a slave device, aborting.\n",
+ np->name, np->dev_name);
+ return -EBUSY;
+ }
+
if (!netif_running(ndev)) {
unsigned long atmost, atleast;
^ permalink raw reply related
* [PATCH net-next 0/2] bonding: delete two unused variables
From: Weiping Pan @ 2011-06-09 2:51 UTC (permalink / raw)
To: fubar, andy; +Cc: netdev, linux-kernel, Weiping Pan
Delete two unused variables in bonding.
Weiping Pan (2):
bonding: delete unused ad_timer
bonding: delete unused arp_mon_pt
drivers/net/bonding/bond_3ad.h | 1 -
drivers/net/bonding/bonding.h | 1 -
2 files changed, 0 insertions(+), 2 deletions(-)
--
1.7.4.4
^ permalink raw reply
* [PATCH net-next 1/2] bonding: delete unused ad_timer
From: Weiping Pan @ 2011-06-09 2:51 UTC (permalink / raw)
To: fubar, andy; +Cc: netdev, linux-kernel, Weiping Pan
In-Reply-To: <cover.1307587586.git.panweiping3@gmail.com>
Now we use agg_select_timer and ad_work.
Signed-off-by: Weiping Pan <panweiping3@gmail.com>
---
drivers/net/bonding/bond_3ad.h | 1 -
1 files changed, 0 insertions(+), 1 deletions(-)
diff --git a/drivers/net/bonding/bond_3ad.h b/drivers/net/bonding/bond_3ad.h
index 0ee3f16..fb67185 100644
--- a/drivers/net/bonding/bond_3ad.h
+++ b/drivers/net/bonding/bond_3ad.h
@@ -257,7 +257,6 @@ struct ad_bond_info {
int lacp_fast; /* whether fast periodic tx should be
* requested
*/
- struct timer_list ad_timer;
};
struct ad_slave_info {
--
1.7.4.4
^ permalink raw reply related
* [PATCH net-next 2/2] bonding: delete unused arp_mon_pt
From: Weiping Pan @ 2011-06-09 2:51 UTC (permalink / raw)
To: fubar, andy; +Cc: netdev, linux-kernel, Weiping Pan
In-Reply-To: <cover.1307587586.git.panweiping3@gmail.com>
Now all received packets are handled by bond_handle_frame,
arp_mon_pt isn't used any more.
Signed-off-by: Weiping Pan <panweiping3@gmail.com>
---
drivers/net/bonding/bonding.h | 1 -
1 files changed, 0 insertions(+), 1 deletions(-)
diff --git a/drivers/net/bonding/bonding.h b/drivers/net/bonding/bonding.h
index ea1d005..382903f 100644
--- a/drivers/net/bonding/bonding.h
+++ b/drivers/net/bonding/bonding.h
@@ -240,7 +240,6 @@ struct bonding {
struct bond_params params;
struct list_head vlan_list;
struct vlan_group *vlgrp;
- struct packet_type arp_mon_pt;
struct workqueue_struct *wq;
struct delayed_work mii_work;
struct delayed_work arp_work;
--
1.7.4.4
^ permalink raw reply related
* Re: [PATCH net-next 0/2] bonding: delete two unused variables
From: Américo Wang @ 2011-06-09 3:19 UTC (permalink / raw)
To: Weiping Pan; +Cc: fubar, andy, netdev, linux-kernel
In-Reply-To: <cover.1307587586.git.panweiping3@gmail.com>
On Thu, Jun 9, 2011 at 10:51 AM, Weiping Pan <panweiping3@gmail.com> wrote:
> Delete two unused variables in bonding.
>
> Weiping Pan (2):
> bonding: delete unused ad_timer
> bonding: delete unused arp_mon_pt
>
Both look good to me,
Reviewed-by: WANG Cong <xiyou.wangcong@gmail.com>
Thanks.
^ permalink raw reply
* Re: [PATCH 3/9] veth: convert to 64 bit statistics
From: Stephen Hemminger @ 2011-06-09 3:22 UTC (permalink / raw)
To: Ben Hutchings; +Cc: David S. Miller, netdev
In-Reply-To: <1307584841.22348.528.camel@localhost>
On Thu, 09 Jun 2011 03:00:41 +0100
Ben Hutchings <bhutchings@solarflare.com> wrote:
> On Wed, 2011-06-08 at 17:53 -0700, Stephen Hemminger wrote:
> > Not much change, device was already keeping per cpu statistics.
> > Use recent 64 statistics interface.
> [...]
>
> It's also going to need to use u64_stats_sync functions.
>
> Ben.
>
No veth is doing per-cpu update therefore the new code has the same guarantee
as the old code.
^ permalink raw reply
* Re: [PATCH 4/9] ifb: convert to 64 bit stats
From: Eric Dumazet @ 2011-06-09 3:29 UTC (permalink / raw)
To: Stephen Hemminger; +Cc: David S. Miller, netdev
In-Reply-To: <20110609005417.507470402@vyatta.com>
Le mercredi 08 juin 2011 à 17:54 -0700, Stephen Hemminger a écrit :
> pièce jointe document texte brut (ifb-stats64.patch)
> Convert input functional block device to use 64 bit stats.
>
> Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
>
>
> --- a/drivers/net/ifb.c 2011-06-07 16:58:31.317079332 -0700
> +++ b/drivers/net/ifb.c 2011-06-07 17:29:02.958161955 -0700
> @@ -42,7 +42,14 @@ struct ifb_private {
> struct tasklet_struct ifb_tasklet;
> int tasklet_pending;
> struct sk_buff_head rq;
> + u64 rx_packets;
> + u64 rx_bytes;
> + u64 rx_dropped;
> +
> struct sk_buff_head tq;
> + u64 tx_packets;
> + u64 tx_bytes;
> + u64 tx_dropped;
> };
>
Hi Stephen
This needs special synchronization on 32bit arches, as Ben pointed out
for other drivers.
^ permalink raw reply
* Re: [PATCH 4/9] ifb: convert to 64 bit stats
From: Stephen Hemminger @ 2011-06-09 3:32 UTC (permalink / raw)
To: Eric Dumazet; +Cc: David S. Miller, netdev
In-Reply-To: <1307590146.3980.28.camel@edumazet-laptop>
On Thu, 09 Jun 2011 05:29:06 +0200
Eric Dumazet <eric.dumazet@gmail.com> wrote:
> Le mercredi 08 juin 2011 à 17:54 -0700, Stephen Hemminger a écrit :
> > pièce jointe document texte brut (ifb-stats64.patch)
> > Convert input functional block device to use 64 bit stats.
> >
> > Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
> >
> >
> > --- a/drivers/net/ifb.c 2011-06-07 16:58:31.317079332 -0700
> > +++ b/drivers/net/ifb.c 2011-06-07 17:29:02.958161955 -0700
> > @@ -42,7 +42,14 @@ struct ifb_private {
> > struct tasklet_struct ifb_tasklet;
> > int tasklet_pending;
> > struct sk_buff_head rq;
> > + u64 rx_packets;
> > + u64 rx_bytes;
> > + u64 rx_dropped;
> > +
> > struct sk_buff_head tq;
> > + u64 tx_packets;
> > + u64 tx_bytes;
> > + u64 tx_dropped;
> > };
> >
>
> Hi Stephen
>
> This needs special synchronization on 32bit arches, as Ben pointed out
> for other drivers.
IFB is running in single thread context. Therefore just locking would
be enough
^ permalink raw reply
* Re: [PATCH 4/9] ifb: convert to 64 bit stats
From: Eric Dumazet @ 2011-06-09 3:37 UTC (permalink / raw)
To: Stephen Hemminger; +Cc: David S. Miller, netdev
In-Reply-To: <20110608203235.2a2be3e4@nehalam.ftrdhcpuser.net>
Le mercredi 08 juin 2011 à 20:32 -0700, Stephen Hemminger a écrit :
> IFB is running in single thread context. Therefore just locking would
> be enough
Yep, that is provided by include/linux/u64_stats_sync.h for free.
As a bonus, no locking on 64bit arches and a clean interface.
^ permalink raw reply
* Re: [PATCH] RFC2988bis + taking RTT sample from 3WHS for the passive open side
From: Jerry Chu @ 2011-06-09 3:39 UTC (permalink / raw)
To: Eric Dumazet
Cc: David Miller, Hagen Paul Pfeifer, tsunanet,
netdev@vger.kernel.org
In-Reply-To: <1307576361.3980.26.camel@edumazet-laptop>
On Wed, Jun 8, 2011 at 4:39 PM, Eric Dumazet <eric.dumazet@gmail.com> wrote:
> Le mercredi 08 juin 2011 à 15:26 -0700, Jerry Chu a écrit :
>> Eric,
>>
>> It just occurred to me now that initRTO is being reduced, both TCP_SYN_RETRIES
>> and TCP_SYNACK_RETRIES should be bumped up a bit, to e.g., 7 (?) to meet the
>> 3 minutes R2 requirement per RFC1122.
>>
>> If you agree, I will submit another patch.
>>
>
> Good catch, but no RFC lowered yet this 3 minutes requirement ?
Not that I know of. BTW, I'm not as nearly concerned about RFC compliance
as what's the "right thing" to do, e.g., if apps have some expectation on
how long, e.g., a connect() call will persistent before ETIMEOUT...
With exponential backoff, initRTO of 3secs and 5 retries will last 189 secs
before ETIMEOUT. Setting initRTO to 1sec reduces this time to 63 secs.
Guess it's still more than one minute so it's not too bad. Now I don't think
it's a good idea to raise retries to 7 or 247 secs, which seems too long (more
than 4 minutes) just to be RFC compliant. Raising retries to 6 or allowing
127 secs might be more reasonable though.
Thought?
Jerry
^ permalink raw reply
* Re: [PATCH net-next-2.6] inetpeer: remove unused list
From: Eric Dumazet @ 2011-06-09 3:47 UTC (permalink / raw)
To: David Miller; +Cc: tim.c.chen, andi, netdev
In-Reply-To: <20110608.170557.2273990802892743973.davem@davemloft.net>
Le mercredi 08 juin 2011 à 17:05 -0700, David Miller a écrit :
> From: Eric Dumazet <eric.dumazet@gmail.com>
> Date: Thu, 09 Jun 2011 01:35:34 +0200
>
> > Andi Kleen and Tim Chen reported huge contention on inetpeer
> > unused_peers.lock, on memcached workload on a 40 core machine, with
> > disabled route cache.
> >
> > It appears we constantly flip peers refcnt between 0 and 1 values, and
> > we must insert/remove peers from unused_peers.list, holding a contended
> > spinlock.
> >
> > Remove this list completely and perform a garbage collection on-the-fly,
> > at lookup time, using the expired nodes we met during the tree
> > traversal.
> >
> > This removes a lot of code, makes locking more standard, and obsoletes
> > two sysctls (inet_peer_gc_mintime and inet_peer_gc_maxtime). This also
> > removes two pointers in inet_peer structure.
> >
> > There is still a false sharing effect because refcnt is in first cache
> > line of object [were the links and keys used by lookups are located], we
> > might move it at the end of inet_peer structure to let this first cache
> > line mostly read by cpus.
> >
> > Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
>
> Didn't expect you to implement this so fast :-)
>
> Applied, thanks!
Thanks David
I would be glad to hear about numbers from Andi and Tim, I am very
jealous of their 80 Threads machine ;)
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox