2019-05-27 14:55:01 +08:00
|
|
|
/* SPDX-License-Identifier: GPL-2.0-or-later */
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
* INET An implementation of the TCP/IP protocol suite for the LINUX
|
|
|
|
* operating system. INET is implemented using the BSD Socket
|
|
|
|
* interface as the means of communication with the user level.
|
|
|
|
*
|
|
|
|
* Definitions for the Interfaces handler.
|
|
|
|
*
|
|
|
|
* Version: @(#)dev.h 1.0.10 08/12/93
|
|
|
|
*
|
2005-05-06 07:16:16 +08:00
|
|
|
* Authors: Ross Biro
|
2005-04-17 06:20:36 +08:00
|
|
|
* Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
|
|
|
|
* Corey Minyard <wf-rch!minyard@relay.EU.net>
|
|
|
|
* Donald J. Becker, <becker@cesdis.gsfc.nasa.gov>
|
2008-10-14 10:01:08 +08:00
|
|
|
* Alan Cox, <alan@lxorguk.ukuu.org.uk>
|
2005-04-17 06:20:36 +08:00
|
|
|
* Bjorn Ekwall. <bj0rn@blox.se>
|
|
|
|
* Pekka Riikonen <priikone@poseidon.pspt.fi>
|
|
|
|
*
|
|
|
|
* Moved to /usr/include/linux for NET3
|
|
|
|
*/
|
|
|
|
#ifndef _LINUX_NETDEVICE_H
|
|
|
|
#define _LINUX_NETDEVICE_H
|
|
|
|
|
2006-12-04 12:15:30 +08:00
|
|
|
#include <linux/timer.h>
|
2011-11-24 09:12:59 +08:00
|
|
|
#include <linux/bug.h>
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
#include <linux/delay.h>
|
2011-07-27 07:09:06 +08:00
|
|
|
#include <linux/atomic.h>
|
2014-10-08 23:19:27 +08:00
|
|
|
#include <linux/prefetch.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
#include <asm/cache.h>
|
|
|
|
#include <asm/byteorder.h>
|
2022-03-11 13:14:20 +08:00
|
|
|
#include <asm/local.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
#include <linux/percpu.h>
|
2009-05-07 07:52:51 +08:00
|
|
|
#include <linux/rculist.h>
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
#include <linux/workqueue.h>
|
2011-11-29 00:33:09 +08:00
|
|
|
#include <linux/dynamic_queue_limits.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2007-09-12 20:57:09 +08:00
|
|
|
#include <net/net_namespace.h>
|
2008-11-25 17:02:08 +08:00
|
|
|
#ifdef CONFIG_DCB
|
2008-11-21 12:52:10 +08:00
|
|
|
#include <net/dcbnl.h>
|
|
|
|
#endif
|
2011-11-22 13:10:51 +08:00
|
|
|
#include <net/netprio_cgroup.h>
|
2018-01-03 18:26:09 +08:00
|
|
|
#include <net/xdp.h>
|
2007-09-12 20:57:09 +08:00
|
|
|
|
2011-11-15 23:29:55 +08:00
|
|
|
#include <linux/netdev_features.h>
|
2012-04-15 14:43:56 +08:00
|
|
|
#include <linux/neighbour.h>
|
2012-10-13 17:46:48 +08:00
|
|
|
#include <uapi/linux/netdevice.h>
|
2015-02-03 22:48:29 +08:00
|
|
|
#include <uapi/linux/if_bonding.h>
|
2016-02-17 13:16:15 +08:00
|
|
|
#include <uapi/linux/pkt_cls.h>
|
2016-08-10 17:05:15 +08:00
|
|
|
#include <linux/hashtable.h>
|
net-next: When a bond have a massive amount of VLANs with IPv6 addresses, performance of changing link state, attaching a VRF, changing an IPv6 address, etc. go down dramtically.
The source of most of the slow down is the `dev_addr_lists.c` module,
which mainatins a linked list of HW addresses.
When using IPv6, this list grows for each IPv6 address added on a
VLAN, since each IPv6 address has a multicast HW address associated with
it.
When performing any modification to the involved links, this list is
traversed many times, often for nothing, all while holding the RTNL
lock.
Instead, this patch adds an auxilliary rbtree which cuts down
traversal time significantly.
Performance can be seen with the following script:
#!/bin/bash
ip netns del test || true 2>/dev/null
ip netns add test
echo 1 | ip netns exec test tee /proc/sys/net/ipv6/conf/all/keep_addr_on_down > /dev/null
set -e
ip -n test link add foo type veth peer name bar
ip -n test link add b1 type bond
ip -n test link add florp type vrf table 10
ip -n test link set bar master b1
ip -n test link set foo up
ip -n test link set bar up
ip -n test link set b1 up
ip -n test link set florp up
VLAN_COUNT=1500
BASE_DEV=b1
echo Creating vlans
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test link add link $BASE_DEV name foo.\$i type vlan id \$i; done"
echo Bringing them up
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test link set foo.\$i up; done"
echo Assiging IPv6 Addresses
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test address add dev foo.\$i 2000::\$i/64; done"
echo Attaching to VRF
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test link set foo.\$i master florp; done"
On an Intel(R) Xeon(R) CPU E5-2650 v3 @ 2.30GHz machine, the performance
before the patch is (truncated):
Creating vlans
real 108.35
Bringing them up
real 4.96
Assiging IPv6 Addresses
real 19.22
Attaching to VRF
real 458.84
After the patch:
Creating vlans
real 5.59
Bringing them up
real 5.07
Assiging IPv6 Addresses
real 5.64
Attaching to VRF
real 25.37
Cc: David S. Miller <davem@davemloft.net>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Lu Wei <luwei32@huawei.com>
Cc: Xiongfeng Wang <wangxiongfeng2@huawei.com>
Cc: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Gilad Naaman <gnaaman@drivenets.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-19 15:17:27 +08:00
|
|
|
#include <linux/rbtree.h>
|
2021-12-10 15:44:21 +08:00
|
|
|
#include <net/net_trackers.h>
|
2022-05-10 11:57:37 +08:00
|
|
|
#include <net/net_debug.h>
|
2011-11-15 23:29:55 +08:00
|
|
|
|
2005-06-23 13:05:31 +08:00
|
|
|
struct netpoll_info;
|
2012-01-31 00:46:54 +08:00
|
|
|
struct device;
|
2020-11-21 06:50:52 +08:00
|
|
|
struct ethtool_ops;
|
2010-07-17 16:49:36 +08:00
|
|
|
struct phy_device;
|
2017-09-30 05:19:20 +08:00
|
|
|
struct dsa_port;
|
2020-05-19 21:03:13 +08:00
|
|
|
struct ip_tunnel_parm;
|
2020-03-25 20:52:31 +08:00
|
|
|
struct macsec_context;
|
|
|
|
struct macsec_ops;
|
2022-04-07 05:37:54 +08:00
|
|
|
struct netdev_name_node;
|
|
|
|
struct sd_flow_limit;
|
2018-03-29 06:44:16 +08:00
|
|
|
struct sfp_bus;
|
2007-04-24 03:20:05 +08:00
|
|
|
/* 802.11 specific */
|
|
|
|
struct wireless_dev;
|
2014-11-02 13:44:54 +08:00
|
|
|
/* 802.15.4 specific */
|
|
|
|
struct wpan_dev;
|
2015-04-22 18:14:37 +08:00
|
|
|
struct mpls_dev;
|
2016-06-17 03:21:00 +08:00
|
|
|
/* UDP Tunnel offloads */
|
|
|
|
struct udp_tunnel_info;
|
2020-07-10 08:42:46 +08:00
|
|
|
struct udp_tunnel_nic_info;
|
|
|
|
struct udp_tunnel_nic;
|
2016-07-20 03:16:48 +08:00
|
|
|
struct bpf_prog;
|
2017-07-18 00:27:07 +08:00
|
|
|
struct xdp_buff;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2020-09-10 01:37:51 +08:00
|
|
|
void synchronize_net(void);
|
2013-09-27 05:48:15 +08:00
|
|
|
void netdev_set_default_ethtool_ops(struct net_device *dev,
|
|
|
|
const struct ethtool_ops *ops);
|
2013-01-11 07:19:10 +08:00
|
|
|
|
2009-11-15 15:20:12 +08:00
|
|
|
/* Backlog congestion levels */
|
|
|
|
#define NET_RX_SUCCESS 0 /* keep 'em coming, baby */
|
|
|
|
#define NET_RX_DROP 1 /* packet dropped */
|
|
|
|
|
2020-02-15 18:50:21 +08:00
|
|
|
#define MAX_NEST_DEV 8
|
|
|
|
|
2009-11-10 14:14:14 +08:00
|
|
|
/*
|
|
|
|
* Transmit return codes: transmit return codes originate from three different
|
|
|
|
* namespaces:
|
|
|
|
*
|
|
|
|
* - qdisc return codes
|
|
|
|
* - driver transmit return codes
|
|
|
|
* - errno values
|
|
|
|
*
|
|
|
|
* Drivers are allowed to return any one of those in their hard_start_xmit()
|
|
|
|
* function. Real network devices commonly used with qdiscs should only return
|
|
|
|
* the driver transmit return codes though - when qdiscs are used, the actual
|
|
|
|
* transmission happens asynchronously, so the value is not propagated to
|
2016-03-24 02:47:23 +08:00
|
|
|
* higher layers. Virtual network devices transmit synchronously; in this case
|
|
|
|
* the driver transmit return codes are consumed by dev_queue_xmit(), and all
|
2009-11-10 14:14:14 +08:00
|
|
|
* others are propagated to higher layers.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* qdisc ->enqueue() return codes. */
|
|
|
|
#define NET_XMIT_SUCCESS 0x00
|
2009-11-15 15:20:12 +08:00
|
|
|
#define NET_XMIT_DROP 0x01 /* skb dropped */
|
|
|
|
#define NET_XMIT_CN 0x02 /* congestion notification */
|
|
|
|
#define NET_XMIT_MASK 0x0f /* qdisc flags in net/sch_generic.h */
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2006-11-14 21:21:36 +08:00
|
|
|
/* NET_XMIT_CN is special. It does not guarantee that this packet is lost. It
|
|
|
|
* indicates that the device will soon be dropping packets, or already drops
|
|
|
|
* some packets of the same priority; prompting us to send less aggressively. */
|
2009-11-10 14:14:14 +08:00
|
|
|
#define net_xmit_eval(e) ((e) == NET_XMIT_CN ? 0 : (e))
|
2005-04-17 06:20:36 +08:00
|
|
|
#define net_xmit_errno(e) ((e) != NET_XMIT_CN ? -ENOBUFS : 0)
|
|
|
|
|
2009-09-01 03:50:40 +08:00
|
|
|
/* Driver transmit return codes */
|
2009-11-15 15:20:12 +08:00
|
|
|
#define NETDEV_TX_MASK 0xf0
|
2009-11-10 14:14:14 +08:00
|
|
|
|
2009-09-01 03:50:40 +08:00
|
|
|
enum netdev_tx {
|
2009-11-10 14:14:14 +08:00
|
|
|
__NETDEV_TX_MIN = INT_MIN, /* make sure enum is signed */
|
2009-11-15 15:20:12 +08:00
|
|
|
NETDEV_TX_OK = 0x00, /* driver took care of packet */
|
|
|
|
NETDEV_TX_BUSY = 0x10, /* driver tx path was busy*/
|
2009-09-01 03:50:40 +08:00
|
|
|
};
|
|
|
|
typedef enum netdev_tx netdev_tx_t;
|
|
|
|
|
2009-11-15 15:20:12 +08:00
|
|
|
/*
|
|
|
|
* Current order: NETDEV_TX_MASK > NET_XMIT_MASK >= 0 is significant;
|
|
|
|
* hard_start_xmit() return < NET_XMIT_MASK means skb was consumed.
|
|
|
|
*/
|
|
|
|
static inline bool dev_xmit_complete(int rc)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* Positive cases with an skb consumed by a driver:
|
|
|
|
* - successful transmission (rc == NETDEV_TX_OK)
|
|
|
|
* - error while transmitting (rc < 0)
|
|
|
|
* - error while queueing to a different device (rc & NET_XMIT_MASK)
|
|
|
|
*/
|
|
|
|
if (likely(rc < NET_XMIT_MASK))
|
|
|
|
return true;
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
2016-03-24 02:47:23 +08:00
|
|
|
* Compute the worst-case header length according to the protocols
|
2005-04-17 06:20:36 +08:00
|
|
|
* used.
|
|
|
|
*/
|
2009-02-06 13:26:19 +08:00
|
|
|
|
2015-12-02 08:43:10 +08:00
|
|
|
#if defined(CONFIG_HYPERV_NET)
|
|
|
|
# define LL_MAX_HEADER 128
|
|
|
|
#elif defined(CONFIG_WLAN) || IS_ENABLED(CONFIG_AX25)
|
2008-05-13 11:17:33 +08:00
|
|
|
# if defined(CONFIG_MAC80211_MESH)
|
|
|
|
# define LL_MAX_HEADER 128
|
|
|
|
# else
|
|
|
|
# define LL_MAX_HEADER 96
|
|
|
|
# endif
|
2005-04-17 06:20:36 +08:00
|
|
|
#else
|
2008-05-13 11:17:33 +08:00
|
|
|
# define LL_MAX_HEADER 32
|
2005-04-17 06:20:36 +08:00
|
|
|
#endif
|
|
|
|
|
2011-11-25 22:40:26 +08:00
|
|
|
#if !IS_ENABLED(CONFIG_NET_IPIP) && !IS_ENABLED(CONFIG_NET_IPGRE) && \
|
|
|
|
!IS_ENABLED(CONFIG_IPV6_SIT) && !IS_ENABLED(CONFIG_IPV6_TUNNEL)
|
2005-04-17 06:20:36 +08:00
|
|
|
#define MAX_HEADER LL_MAX_HEADER
|
|
|
|
#else
|
|
|
|
#define MAX_HEADER (LL_MAX_HEADER + 48)
|
|
|
|
#endif
|
|
|
|
|
|
|
|
/*
|
2010-06-08 15:19:54 +08:00
|
|
|
* Old network device statistics. Fields are native words
|
|
|
|
* (unsigned long) so they can be read and written atomically.
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
2009-02-06 13:26:19 +08:00
|
|
|
|
2009-11-05 01:50:58 +08:00
|
|
|
struct net_device_stats {
|
2010-07-09 17:11:52 +08:00
|
|
|
unsigned long rx_packets;
|
|
|
|
unsigned long tx_packets;
|
|
|
|
unsigned long rx_bytes;
|
|
|
|
unsigned long tx_bytes;
|
|
|
|
unsigned long rx_errors;
|
|
|
|
unsigned long tx_errors;
|
|
|
|
unsigned long rx_dropped;
|
|
|
|
unsigned long tx_dropped;
|
|
|
|
unsigned long multicast;
|
2005-04-17 06:20:36 +08:00
|
|
|
unsigned long collisions;
|
|
|
|
unsigned long rx_length_errors;
|
2010-07-09 17:11:52 +08:00
|
|
|
unsigned long rx_over_errors;
|
|
|
|
unsigned long rx_crc_errors;
|
|
|
|
unsigned long rx_frame_errors;
|
|
|
|
unsigned long rx_fifo_errors;
|
|
|
|
unsigned long rx_missed_errors;
|
2005-04-17 06:20:36 +08:00
|
|
|
unsigned long tx_aborted_errors;
|
|
|
|
unsigned long tx_carrier_errors;
|
|
|
|
unsigned long tx_fifo_errors;
|
|
|
|
unsigned long tx_heartbeat_errors;
|
|
|
|
unsigned long tx_window_errors;
|
|
|
|
unsigned long rx_compressed;
|
|
|
|
unsigned long tx_compressed;
|
|
|
|
};
|
|
|
|
|
2022-03-11 13:14:20 +08:00
|
|
|
/* per-cpu stats, allocated on demand.
|
|
|
|
* Try to fit them in a single cache line, for dev_get_stats() sake.
|
|
|
|
*/
|
|
|
|
struct net_device_core_stats {
|
net: Use this_cpu_inc() to increment net->core_stats
The macro dev_core_stats_##FIELD##_inc() disables preemption and invokes
netdev_core_stats_alloc() to return a per-CPU pointer.
netdev_core_stats_alloc() will allocate memory on its first invocation
which breaks on PREEMPT_RT because it requires non-atomic context for
memory allocation.
This can be avoided by enabling preemption in netdev_core_stats_alloc()
assuming the caller always disables preemption.
It might be better to replace local_inc() with this_cpu_inc() now that
dev_core_stats_##FIELD##_inc() gained a preempt-disable section and does
not rely on already disabled preemption. This results in less
instructions on x86-64:
local_inc:
| incl %gs:__preempt_count(%rip) # __preempt_count
| movq 488(%rdi), %rax # _1->core_stats, _22
| testq %rax, %rax # _22
| je .L585 #,
| add %gs:this_cpu_off(%rip), %rax # this_cpu_off, tcp_ptr__
| .L586:
| testq %rax, %rax # _27
| je .L587 #,
| incq (%rax) # _6->a.counter
| .L587:
| decl %gs:__preempt_count(%rip) # __preempt_count
this_cpu_inc(), this patch:
| movq 488(%rdi), %rax # _1->core_stats, _5
| testq %rax, %rax # _5
| je .L591 #,
| .L585:
| incq %gs:(%rax) # _18->rx_dropped
Use unsigned long as type for the counter. Use this_cpu_inc() to
increment the counter. Use a plain read of the counter.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/YmbO0pxgtKpCw4SY@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-04-26 00:39:46 +08:00
|
|
|
unsigned long rx_dropped;
|
|
|
|
unsigned long tx_dropped;
|
|
|
|
unsigned long rx_nohandler;
|
2022-04-29 04:01:50 +08:00
|
|
|
unsigned long rx_otherhost_dropped;
|
net: Use this_cpu_inc() to increment net->core_stats
The macro dev_core_stats_##FIELD##_inc() disables preemption and invokes
netdev_core_stats_alloc() to return a per-CPU pointer.
netdev_core_stats_alloc() will allocate memory on its first invocation
which breaks on PREEMPT_RT because it requires non-atomic context for
memory allocation.
This can be avoided by enabling preemption in netdev_core_stats_alloc()
assuming the caller always disables preemption.
It might be better to replace local_inc() with this_cpu_inc() now that
dev_core_stats_##FIELD##_inc() gained a preempt-disable section and does
not rely on already disabled preemption. This results in less
instructions on x86-64:
local_inc:
| incl %gs:__preempt_count(%rip) # __preempt_count
| movq 488(%rdi), %rax # _1->core_stats, _22
| testq %rax, %rax # _22
| je .L585 #,
| add %gs:this_cpu_off(%rip), %rax # this_cpu_off, tcp_ptr__
| .L586:
| testq %rax, %rax # _27
| je .L587 #,
| incq (%rax) # _6->a.counter
| .L587:
| decl %gs:__preempt_count(%rip) # __preempt_count
this_cpu_inc(), this patch:
| movq 488(%rdi), %rax # _1->core_stats, _5
| testq %rax, %rax # _5
| je .L591 #,
| .L585:
| incq %gs:(%rax) # _18->rx_dropped
Use unsigned long as type for the counter. Use this_cpu_inc() to
increment the counter. Use a plain read of the counter.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/YmbO0pxgtKpCw4SY@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-04-26 00:39:46 +08:00
|
|
|
} __aligned(4 * sizeof(unsigned long));
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
#include <linux/cache.h>
|
|
|
|
#include <linux/skbuff.h>
|
|
|
|
|
2011-11-17 11:13:26 +08:00
|
|
|
#ifdef CONFIG_RPS
|
2012-02-24 15:31:31 +08:00
|
|
|
#include <linux/static_key.h>
|
2019-03-22 23:56:38 +08:00
|
|
|
extern struct static_key_false rps_needed;
|
|
|
|
extern struct static_key_false rfs_needed;
|
2011-11-17 11:13:26 +08:00
|
|
|
#endif
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
struct neighbour;
|
|
|
|
struct neigh_parms;
|
|
|
|
struct sk_buff;
|
|
|
|
|
2009-05-05 10:48:28 +08:00
|
|
|
struct netdev_hw_addr {
|
|
|
|
struct list_head list;
|
net-next: When a bond have a massive amount of VLANs with IPv6 addresses, performance of changing link state, attaching a VRF, changing an IPv6 address, etc. go down dramtically.
The source of most of the slow down is the `dev_addr_lists.c` module,
which mainatins a linked list of HW addresses.
When using IPv6, this list grows for each IPv6 address added on a
VLAN, since each IPv6 address has a multicast HW address associated with
it.
When performing any modification to the involved links, this list is
traversed many times, often for nothing, all while holding the RTNL
lock.
Instead, this patch adds an auxilliary rbtree which cuts down
traversal time significantly.
Performance can be seen with the following script:
#!/bin/bash
ip netns del test || true 2>/dev/null
ip netns add test
echo 1 | ip netns exec test tee /proc/sys/net/ipv6/conf/all/keep_addr_on_down > /dev/null
set -e
ip -n test link add foo type veth peer name bar
ip -n test link add b1 type bond
ip -n test link add florp type vrf table 10
ip -n test link set bar master b1
ip -n test link set foo up
ip -n test link set bar up
ip -n test link set b1 up
ip -n test link set florp up
VLAN_COUNT=1500
BASE_DEV=b1
echo Creating vlans
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test link add link $BASE_DEV name foo.\$i type vlan id \$i; done"
echo Bringing them up
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test link set foo.\$i up; done"
echo Assiging IPv6 Addresses
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test address add dev foo.\$i 2000::\$i/64; done"
echo Attaching to VRF
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test link set foo.\$i master florp; done"
On an Intel(R) Xeon(R) CPU E5-2650 v3 @ 2.30GHz machine, the performance
before the patch is (truncated):
Creating vlans
real 108.35
Bringing them up
real 4.96
Assiging IPv6 Addresses
real 19.22
Attaching to VRF
real 458.84
After the patch:
Creating vlans
real 5.59
Bringing them up
real 5.07
Assiging IPv6 Addresses
real 5.64
Attaching to VRF
real 25.37
Cc: David S. Miller <davem@davemloft.net>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Lu Wei <luwei32@huawei.com>
Cc: Xiongfeng Wang <wangxiongfeng2@huawei.com>
Cc: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Gilad Naaman <gnaaman@drivenets.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-19 15:17:27 +08:00
|
|
|
struct rb_node node;
|
2009-05-05 10:48:28 +08:00
|
|
|
unsigned char addr[MAX_ADDR_LEN];
|
|
|
|
unsigned char type;
|
2009-05-23 07:22:17 +08:00
|
|
|
#define NETDEV_HW_ADDR_T_LAN 1
|
|
|
|
#define NETDEV_HW_ADDR_T_SAN 2
|
2020-10-02 01:12:50 +08:00
|
|
|
#define NETDEV_HW_ADDR_T_UNICAST 3
|
|
|
|
#define NETDEV_HW_ADDR_T_MULTICAST 4
|
2010-04-02 05:22:57 +08:00
|
|
|
bool global_use;
|
2013-04-15 17:54:25 +08:00
|
|
|
int sync_cnt;
|
2010-09-20 02:24:02 +08:00
|
|
|
int refcount;
|
2013-04-03 05:10:07 +08:00
|
|
|
int synced;
|
2009-05-05 10:48:28 +08:00
|
|
|
struct rcu_head rcu_head;
|
|
|
|
};
|
|
|
|
|
2009-06-17 09:12:19 +08:00
|
|
|
struct netdev_hw_addr_list {
|
|
|
|
struct list_head list;
|
|
|
|
int count;
|
net-next: When a bond have a massive amount of VLANs with IPv6 addresses, performance of changing link state, attaching a VRF, changing an IPv6 address, etc. go down dramtically.
The source of most of the slow down is the `dev_addr_lists.c` module,
which mainatins a linked list of HW addresses.
When using IPv6, this list grows for each IPv6 address added on a
VLAN, since each IPv6 address has a multicast HW address associated with
it.
When performing any modification to the involved links, this list is
traversed many times, often for nothing, all while holding the RTNL
lock.
Instead, this patch adds an auxilliary rbtree which cuts down
traversal time significantly.
Performance can be seen with the following script:
#!/bin/bash
ip netns del test || true 2>/dev/null
ip netns add test
echo 1 | ip netns exec test tee /proc/sys/net/ipv6/conf/all/keep_addr_on_down > /dev/null
set -e
ip -n test link add foo type veth peer name bar
ip -n test link add b1 type bond
ip -n test link add florp type vrf table 10
ip -n test link set bar master b1
ip -n test link set foo up
ip -n test link set bar up
ip -n test link set b1 up
ip -n test link set florp up
VLAN_COUNT=1500
BASE_DEV=b1
echo Creating vlans
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test link add link $BASE_DEV name foo.\$i type vlan id \$i; done"
echo Bringing them up
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test link set foo.\$i up; done"
echo Assiging IPv6 Addresses
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test address add dev foo.\$i 2000::\$i/64; done"
echo Attaching to VRF
ip netns exec test time -p bash -c "for i in \$(seq 1 $VLAN_COUNT);
do ip -n test link set foo.\$i master florp; done"
On an Intel(R) Xeon(R) CPU E5-2650 v3 @ 2.30GHz machine, the performance
before the patch is (truncated):
Creating vlans
real 108.35
Bringing them up
real 4.96
Assiging IPv6 Addresses
real 19.22
Attaching to VRF
real 458.84
After the patch:
Creating vlans
real 5.59
Bringing them up
real 5.07
Assiging IPv6 Addresses
real 5.64
Attaching to VRF
real 25.37
Cc: David S. Miller <davem@davemloft.net>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Lu Wei <luwei32@huawei.com>
Cc: Xiongfeng Wang <wangxiongfeng2@huawei.com>
Cc: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Gilad Naaman <gnaaman@drivenets.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-19 15:17:27 +08:00
|
|
|
|
|
|
|
/* Auxiliary tree for faster lookup on addition and deletion */
|
|
|
|
struct rb_root tree;
|
2009-06-17 09:12:19 +08:00
|
|
|
};
|
|
|
|
|
2010-04-02 05:22:57 +08:00
|
|
|
#define netdev_hw_addr_list_count(l) ((l)->count)
|
|
|
|
#define netdev_hw_addr_list_empty(l) (netdev_hw_addr_list_count(l) == 0)
|
|
|
|
#define netdev_hw_addr_list_for_each(ha, l) \
|
|
|
|
list_for_each_entry(ha, &(l)->list, list)
|
2010-01-26 05:36:10 +08:00
|
|
|
|
2010-04-02 05:22:57 +08:00
|
|
|
#define netdev_uc_count(dev) netdev_hw_addr_list_count(&(dev)->uc)
|
|
|
|
#define netdev_uc_empty(dev) netdev_hw_addr_list_empty(&(dev)->uc)
|
|
|
|
#define netdev_for_each_uc_addr(ha, dev) \
|
|
|
|
netdev_hw_addr_list_for_each(ha, &(dev)->uc)
|
2022-09-11 09:06:57 +08:00
|
|
|
#define netdev_for_each_synced_uc_addr(_ha, _dev) \
|
|
|
|
netdev_for_each_uc_addr((_ha), (_dev)) \
|
|
|
|
if ((_ha)->sync_cnt)
|
2010-02-05 02:22:25 +08:00
|
|
|
|
2010-04-02 05:22:57 +08:00
|
|
|
#define netdev_mc_count(dev) netdev_hw_addr_list_count(&(dev)->mc)
|
|
|
|
#define netdev_mc_empty(dev) netdev_hw_addr_list_empty(&(dev)->mc)
|
2010-04-08 07:40:09 +08:00
|
|
|
#define netdev_for_each_mc_addr(ha, dev) \
|
2010-04-02 05:22:57 +08:00
|
|
|
netdev_hw_addr_list_for_each(ha, &(dev)->mc)
|
2022-09-11 09:06:57 +08:00
|
|
|
#define netdev_for_each_synced_mc_addr(_ha, _dev) \
|
|
|
|
netdev_for_each_mc_addr((_ha), (_dev)) \
|
|
|
|
if ((_ha)->sync_cnt)
|
2010-02-05 02:22:25 +08:00
|
|
|
|
2009-11-05 01:50:58 +08:00
|
|
|
struct hh_cache {
|
2017-04-10 16:11:17 +08:00
|
|
|
unsigned int hh_len;
|
2006-12-08 07:08:17 +08:00
|
|
|
seqlock_t hh_lock;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
/* cached hardware header; allow for machine alignment needs. */
|
|
|
|
#define HH_DATA_MOD 16
|
|
|
|
#define HH_DATA_OFF(__len) \
|
2005-06-03 07:48:05 +08:00
|
|
|
(HH_DATA_MOD - (((__len - 1) & (HH_DATA_MOD - 1)) + 1))
|
2005-04-17 06:20:36 +08:00
|
|
|
#define HH_DATA_ALIGN(__len) \
|
|
|
|
(((__len)+(HH_DATA_MOD-1))&~(HH_DATA_MOD - 1))
|
|
|
|
unsigned long hh_data[HH_DATA_ALIGN(LL_MAX_HEADER) / sizeof(long)];
|
|
|
|
};
|
|
|
|
|
2016-03-24 02:47:23 +08:00
|
|
|
/* Reserve HH_DATA_MOD byte-aligned hard_header_len, but at least that much.
|
2005-04-17 06:20:36 +08:00
|
|
|
* Alternative is:
|
|
|
|
* dev->hard_header_len ? (dev->hard_header_len +
|
|
|
|
* (HH_DATA_MOD - 1)) & ~(HH_DATA_MOD - 1) : 0
|
|
|
|
*
|
|
|
|
* We could use other alignment values, but we must maintain the
|
|
|
|
* relationship HH alignment <= LL alignment.
|
|
|
|
*/
|
|
|
|
#define LL_RESERVED_SPACE(dev) \
|
2008-05-13 11:48:31 +08:00
|
|
|
((((dev)->hard_header_len+(dev)->needed_headroom)&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
|
2005-04-17 06:20:36 +08:00
|
|
|
#define LL_RESERVED_SPACE_EXTRA(dev,extra) \
|
2008-05-13 11:48:31 +08:00
|
|
|
((((dev)->hard_header_len+(dev)->needed_headroom+(extra))&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2007-10-09 16:40:57 +08:00
|
|
|
struct header_ops {
|
|
|
|
int (*create) (struct sk_buff *skb, struct net_device *dev,
|
|
|
|
unsigned short type, const void *daddr,
|
2012-04-15 13:58:06 +08:00
|
|
|
const void *saddr, unsigned int len);
|
2007-10-09 16:40:57 +08:00
|
|
|
int (*parse)(const struct sk_buff *skb, unsigned char *haddr);
|
2011-07-13 14:28:12 +08:00
|
|
|
int (*cache)(const struct neighbour *neigh, struct hh_cache *hh, __be16 type);
|
2007-10-09 16:40:57 +08:00
|
|
|
void (*cache_update)(struct hh_cache *hh,
|
|
|
|
const struct net_device *dev,
|
|
|
|
const unsigned char *haddr);
|
2016-03-10 10:58:32 +08:00
|
|
|
bool (*validate)(const char *ll_header, unsigned int len);
|
2019-02-21 20:39:58 +08:00
|
|
|
__be16 (*parse_protocol)(const struct sk_buff *skb);
|
2007-10-09 16:40:57 +08:00
|
|
|
};
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/* These flag bits are private to the generic network queueing
|
2016-03-24 02:47:23 +08:00
|
|
|
* layer; they may not be explicitly referenced by any other
|
2005-04-17 06:20:36 +08:00
|
|
|
* code.
|
|
|
|
*/
|
|
|
|
|
2009-11-05 01:50:58 +08:00
|
|
|
enum netdev_state_t {
|
2005-04-17 06:20:36 +08:00
|
|
|
__LINK_STATE_START,
|
|
|
|
__LINK_STATE_PRESENT,
|
|
|
|
__LINK_STATE_NOCARRIER,
|
2006-03-21 09:09:11 +08:00
|
|
|
__LINK_STATE_LINKWATCH_PENDING,
|
|
|
|
__LINK_STATE_DORMANT,
|
2020-04-20 06:11:50 +08:00
|
|
|
__LINK_STATE_TESTING,
|
2005-04-17 06:20:36 +08:00
|
|
|
};
|
|
|
|
|
2018-07-05 14:34:32 +08:00
|
|
|
struct gro_list {
|
|
|
|
struct list_head list;
|
|
|
|
int count;
|
|
|
|
};
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/*
|
net: convert gro_count to bitmask
gro_hash size is 192 bytes, and uses 3 cache lines, if there is few
flows, gro_hash may be not fully used, so it is unnecessary to iterate
all gro_hash in napi_gro_flush(), to occupy unnecessary cacheline.
convert gro_count to a bitmask, and rename it as gro_bitmask, each bit
represents a element of gro_hash, only flush a gro_hash element if the
related bit is set, to speed up napi_gro_flush().
and update gro_bitmask only if it will be changed, to reduce cache
update
Suggested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Cc: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-13 14:41:36 +08:00
|
|
|
* size of gro hash buckets, must less than bit number of
|
|
|
|
* napi_struct::gro_bitmask
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
*/
|
2018-06-24 13:14:02 +08:00
|
|
|
#define GRO_HASH_BUCKETS 8
|
net: convert gro_count to bitmask
gro_hash size is 192 bytes, and uses 3 cache lines, if there is few
flows, gro_hash may be not fully used, so it is unnecessary to iterate
all gro_hash in napi_gro_flush(), to occupy unnecessary cacheline.
convert gro_count to a bitmask, and rename it as gro_bitmask, each bit
represents a element of gro_hash, only flush a gro_hash element if the
related bit is set, to speed up napi_gro_flush().
and update gro_bitmask only if it will be changed, to reduce cache
update
Suggested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Cc: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-13 14:41:36 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Structure for NAPI scheduling similar to tasklet but with weighting
|
|
|
|
*/
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
struct napi_struct {
|
|
|
|
/* The poll_list must only be managed by the entity which
|
|
|
|
* changes the state of the NAPI_STATE_SCHED bit. This means
|
|
|
|
* whoever atomically sets that bit can add this napi_struct
|
2016-03-24 02:47:23 +08:00
|
|
|
* to the per-CPU poll_list, and whoever clears that bit
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
* can remove from the list right before clearing the bit.
|
|
|
|
*/
|
|
|
|
struct list_head poll_list;
|
|
|
|
|
|
|
|
unsigned long state;
|
|
|
|
int weight;
|
net: napi: add hard irqs deferral feature
Back in commit 3b47d30396ba ("net: gro: add a per device gro flush timer")
we added the ability to arm one high resolution timer, that we used
to keep not-complete packets in GRO engine a bit longer, hoping that further
frames might be added to them.
Since then, we added the napi_complete_done() interface, and commit
364b6055738b ("net: busy-poll: return busypolling status to drivers")
allowed drivers to avoid re-arming NIC interrupts if we made a promise
that their NAPI poll() handler would be called in the near future.
This infrastructure can be leveraged, thanks to a new device parameter,
which allows to arm the napi hrtimer, instead of re-arming the device
hard IRQ.
We have noticed that on some servers with 32 RX queues or more, the chit-chat
between the NIC and the host caused by IRQ delivery and re-arming could hurt
throughput by ~20% on 100Gbit NIC.
In contrast, hrtimers are using local (percpu) resources and might have lower
cost.
The new tunable, named napi_defer_hard_irqs, is placed in the same hierarchy
than gro_flush_timeout (/sys/class/net/ethX/)
By default, both gro_flush_timeout and napi_defer_hard_irqs are zero.
This patch does not change the prior behavior of gro_flush_timeout
if used alone : NIC hard irqs should be rearmed as before.
One concrete usage can be :
echo 20000 >/sys/class/net/eth1/gro_flush_timeout
echo 10 >/sys/class/net/eth1/napi_defer_hard_irqs
If at least one packet is retired, then we will reset napi counter
to 10 (napi_defer_hard_irqs), ensuring at least 10 periodic scans
of the queue.
On busy queues, this should avoid NIC hard IRQ, while before this patch IRQ
avoidance was only possible if napi->poll() was exhausting its budget
and not call napi_complete_done().
This feature also can be used to work around some non-optimal NIC irq
coalescing strategies.
Having the ability to insert XX usec delays between each napi->poll()
can increase cache efficiency, since we increase batch sizes.
It also keeps serving cpus not idle too long, reducing tail latencies.
Co-developed-by: Luigi Rizzo <lrizzo@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-04-23 00:13:27 +08:00
|
|
|
int defer_hard_irqs_count;
|
net: convert gro_count to bitmask
gro_hash size is 192 bytes, and uses 3 cache lines, if there is few
flows, gro_hash may be not fully used, so it is unnecessary to iterate
all gro_hash in napi_gro_flush(), to occupy unnecessary cacheline.
convert gro_count to a bitmask, and rename it as gro_bitmask, each bit
represents a element of gro_hash, only flush a gro_hash element if the
related bit is set, to speed up napi_gro_flush().
and update gro_bitmask only if it will be changed, to reduce cache
update
Suggested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Cc: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-13 14:41:36 +08:00
|
|
|
unsigned long gro_bitmask;
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
int (*poll)(struct napi_struct *, int);
|
|
|
|
#ifdef CONFIG_NETPOLL
|
|
|
|
int poll_owner;
|
|
|
|
#endif
|
2009-01-05 08:13:40 +08:00
|
|
|
struct net_device *dev;
|
2018-07-05 14:34:32 +08:00
|
|
|
struct gro_list gro_hash[GRO_HASH_BUCKETS];
|
2009-01-05 08:13:40 +08:00
|
|
|
struct sk_buff *skb;
|
net: use listified RX for handling GRO_NORMAL skbs
When GRO decides not to coalesce a packet, in napi_frags_finish(), instead
of passing it to the stack immediately, place it on a list in the napi
struct. Then, at flush time (napi_complete_done(), napi_poll(), or
napi_busy_loop()), call netif_receive_skb_list_internal() on the list.
We'd like to do that in napi_gro_flush(), but it's not called if
!napi->gro_bitmask, so we have to do it in the callers instead. (There are
a handful of drivers that call napi_gro_flush() themselves, but it's not
clear why, or whether this will affect them.)
Because a full 64 packets is an inefficiently large batch, also consume the
list whenever it exceeds gro_normal_batch, a new net/core sysctl that
defaults to 8.
Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-08-06 21:53:55 +08:00
|
|
|
struct list_head rx_list; /* Pending GRO_NORMAL skbs */
|
|
|
|
int rx_count; /* length of rx_list */
|
net: gro: add a per device gro flush timer
Tuning coalescing parameters on NIC can be really hard.
Servers can handle both bulk and RPC like traffic, with conflicting
goals : bulk flows want as big GRO packets as possible, RPC want minimal
latencies.
To reach big GRO packets on 10Gbe NIC, one can use :
ethtool -C eth0 rx-usecs 4 rx-frames 44
But this penalizes rpc sessions, with an increase of latencies, up to
50% in some cases, as NICs generally do not force an interrupt when
a packet with TCP Push flag is received.
Some NICs do not have an absolute timer, only a timer rearmed for every
incoming packet.
This patch uses a different strategy : Let GRO stack decides what do do,
based on traffic pattern.
Packets with Push flag wont be delayed.
Packets without Push flag might be held in GRO engine, if we keep
receiving data.
This new mechanism is off by default, and shall be enabled by setting
/sys/class/net/ethX/gro_flush_timeout to a value in nanosecond.
To fully enable this mechanism, drivers should use napi_complete_done()
instead of napi_complete().
Tested:
Ran 200 netperf TCP_STREAM from A to B (10Gbe mlx4 link, 8 RX queues)
Without this feature, we send back about 305,000 ACK per second.
GRO aggregation ratio is low (811/305 = 2.65 segments per GRO packet)
Setting a timer of 2000 nsec is enough to increase GRO packet sizes
and reduce number of ACK packets. (811/19.2 = 42)
Receiver performs less calls to upper stacks, less wakes up.
This also reduces cpu usage on the sender, as it receives less ACK
packets.
Note that reducing number of wakes up increases cpu efficiency, but can
decrease QPS, as applications wont have the chance to warmup cpu caches
doing a partial read of RPC requests/answers if they fit in one skb.
B:~# sar -n DEV 1 10 | grep eth0 | tail -1
Average: eth0 811269.80 305732.30 1199462.57 19705.72 0.00
0.00 0.50
B:~# echo 2000 >/sys/class/net/eth0/gro_flush_timeout
B:~# sar -n DEV 1 10 | grep eth0 | tail -1
Average: eth0 811577.30 19230.80 1199916.51 1239.80 0.00
0.00 0.50
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-11-07 13:09:44 +08:00
|
|
|
struct hrtimer timer;
|
2012-09-26 15:07:47 +08:00
|
|
|
struct list_head dev_list;
|
2013-06-10 16:39:41 +08:00
|
|
|
struct hlist_node napi_hash_node;
|
|
|
|
unsigned int napi_id;
|
2021-02-09 03:34:09 +08:00
|
|
|
struct task_struct *thread;
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
};
|
|
|
|
|
2009-11-05 01:50:58 +08:00
|
|
|
enum {
|
net: Introduce preferred busy-polling
The existing busy-polling mode, enabled by the SO_BUSY_POLL socket
option or system-wide using the /proc/sys/net/core/busy_read knob, is
an opportunistic. That means that if the NAPI context is not
scheduled, it will poll it. If, after busy-polling, the budget is
exceeded the busy-polling logic will schedule the NAPI onto the
regular softirq handling.
One implication of the behavior above is that a busy/heavy loaded NAPI
context will never enter/allow for busy-polling. Some applications
prefer that most NAPI processing would be done by busy-polling.
This series adds a new socket option, SO_PREFER_BUSY_POLL, that works
in concert with the napi_defer_hard_irqs and gro_flush_timeout
knobs. The napi_defer_hard_irqs and gro_flush_timeout knobs were
introduced in commit 6f8b12d661d0 ("net: napi: add hard irqs deferral
feature"), and allows for a user to defer interrupts to be enabled and
instead schedule the NAPI context from a watchdog timer. When a user
enables the SO_PREFER_BUSY_POLL, again with the other knobs enabled,
and the NAPI context is being processed by a softirq, the softirq NAPI
processing will exit early to allow the busy-polling to be performed.
If the application stops performing busy-polling via a system call,
the watchdog timer defined by gro_flush_timeout will timeout, and
regular softirq handling will resume.
In summary; Heavy traffic applications that prefer busy-polling over
softirq processing should use this option.
Example usage:
$ echo 2 | sudo tee /sys/class/net/ens785f1/napi_defer_hard_irqs
$ echo 200000 | sudo tee /sys/class/net/ens785f1/gro_flush_timeout
Note that the timeout should be larger than the userspace processing
window, otherwise the watchdog will timeout and fall back to regular
softirq processing.
Enable the SO_BUSY_POLL/SO_PREFER_BUSY_POLL options on your socket.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/bpf/20201130185205.196029-2-bjorn.topel@gmail.com
2020-12-01 02:51:56 +08:00
|
|
|
NAPI_STATE_SCHED, /* Poll is scheduled */
|
|
|
|
NAPI_STATE_MISSED, /* reschedule a napi */
|
|
|
|
NAPI_STATE_DISABLE, /* Disable pending */
|
|
|
|
NAPI_STATE_NPSVC, /* Netpoll - don't dequeue from poll_list */
|
|
|
|
NAPI_STATE_LISTED, /* NAPI added to system lists */
|
|
|
|
NAPI_STATE_NO_BUSY_POLL, /* Do not add in napi_hash, no busy polling */
|
|
|
|
NAPI_STATE_IN_BUSY_POLL, /* sk_busy_loop() owns this NAPI */
|
|
|
|
NAPI_STATE_PREFER_BUSY_POLL, /* prefer busy-polling over softirq processing*/
|
2021-02-09 03:34:09 +08:00
|
|
|
NAPI_STATE_THREADED, /* The poll is performed inside its own thread*/
|
2021-03-17 06:36:47 +08:00
|
|
|
NAPI_STATE_SCHED_THREADED, /* Napi is currently scheduled in threaded mode */
|
2016-11-16 02:15:11 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
enum {
|
net: Introduce preferred busy-polling
The existing busy-polling mode, enabled by the SO_BUSY_POLL socket
option or system-wide using the /proc/sys/net/core/busy_read knob, is
an opportunistic. That means that if the NAPI context is not
scheduled, it will poll it. If, after busy-polling, the budget is
exceeded the busy-polling logic will schedule the NAPI onto the
regular softirq handling.
One implication of the behavior above is that a busy/heavy loaded NAPI
context will never enter/allow for busy-polling. Some applications
prefer that most NAPI processing would be done by busy-polling.
This series adds a new socket option, SO_PREFER_BUSY_POLL, that works
in concert with the napi_defer_hard_irqs and gro_flush_timeout
knobs. The napi_defer_hard_irqs and gro_flush_timeout knobs were
introduced in commit 6f8b12d661d0 ("net: napi: add hard irqs deferral
feature"), and allows for a user to defer interrupts to be enabled and
instead schedule the NAPI context from a watchdog timer. When a user
enables the SO_PREFER_BUSY_POLL, again with the other knobs enabled,
and the NAPI context is being processed by a softirq, the softirq NAPI
processing will exit early to allow the busy-polling to be performed.
If the application stops performing busy-polling via a system call,
the watchdog timer defined by gro_flush_timeout will timeout, and
regular softirq handling will resume.
In summary; Heavy traffic applications that prefer busy-polling over
softirq processing should use this option.
Example usage:
$ echo 2 | sudo tee /sys/class/net/ens785f1/napi_defer_hard_irqs
$ echo 200000 | sudo tee /sys/class/net/ens785f1/gro_flush_timeout
Note that the timeout should be larger than the userspace processing
window, otherwise the watchdog will timeout and fall back to regular
softirq processing.
Enable the SO_BUSY_POLL/SO_PREFER_BUSY_POLL options on your socket.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/bpf/20201130185205.196029-2-bjorn.topel@gmail.com
2020-12-01 02:51:56 +08:00
|
|
|
NAPIF_STATE_SCHED = BIT(NAPI_STATE_SCHED),
|
|
|
|
NAPIF_STATE_MISSED = BIT(NAPI_STATE_MISSED),
|
|
|
|
NAPIF_STATE_DISABLE = BIT(NAPI_STATE_DISABLE),
|
|
|
|
NAPIF_STATE_NPSVC = BIT(NAPI_STATE_NPSVC),
|
|
|
|
NAPIF_STATE_LISTED = BIT(NAPI_STATE_LISTED),
|
|
|
|
NAPIF_STATE_NO_BUSY_POLL = BIT(NAPI_STATE_NO_BUSY_POLL),
|
|
|
|
NAPIF_STATE_IN_BUSY_POLL = BIT(NAPI_STATE_IN_BUSY_POLL),
|
|
|
|
NAPIF_STATE_PREFER_BUSY_POLL = BIT(NAPI_STATE_PREFER_BUSY_POLL),
|
2021-02-09 03:34:09 +08:00
|
|
|
NAPIF_STATE_THREADED = BIT(NAPI_STATE_THREADED),
|
2021-03-17 06:36:47 +08:00
|
|
|
NAPIF_STATE_SCHED_THREADED = BIT(NAPI_STATE_SCHED_THREADED),
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
};
|
|
|
|
|
2009-10-29 15:17:09 +08:00
|
|
|
enum gro_result {
|
2009-03-17 01:50:02 +08:00
|
|
|
GRO_MERGED,
|
|
|
|
GRO_MERGED_FREE,
|
|
|
|
GRO_HELD,
|
|
|
|
GRO_NORMAL,
|
2017-02-15 16:39:44 +08:00
|
|
|
GRO_CONSUMED,
|
2009-03-17 01:50:02 +08:00
|
|
|
};
|
2009-10-29 15:17:09 +08:00
|
|
|
typedef enum gro_result gro_result_t;
|
2009-03-17 01:50:02 +08:00
|
|
|
|
2011-03-12 11:14:39 +08:00
|
|
|
/*
|
|
|
|
* enum rx_handler_result - Possible return values for rx_handlers.
|
|
|
|
* @RX_HANDLER_CONSUMED: skb was consumed by rx_handler, do not process it
|
|
|
|
* further.
|
|
|
|
* @RX_HANDLER_ANOTHER: Do another round in receive path. This is indicated in
|
|
|
|
* case skb->dev was changed by rx_handler.
|
|
|
|
* @RX_HANDLER_EXACT: Force exact delivery, no wildcard.
|
2016-03-24 02:47:23 +08:00
|
|
|
* @RX_HANDLER_PASS: Do nothing, pass the skb as if no rx_handler was called.
|
2011-03-12 11:14:39 +08:00
|
|
|
*
|
|
|
|
* rx_handlers are functions called from inside __netif_receive_skb(), to do
|
|
|
|
* special processing of the skb, prior to delivery to protocol handlers.
|
|
|
|
*
|
|
|
|
* Currently, a net_device can only have a single rx_handler registered. Trying
|
|
|
|
* to register a second rx_handler will return -EBUSY.
|
|
|
|
*
|
|
|
|
* To register a rx_handler on a net_device, use netdev_rx_handler_register().
|
|
|
|
* To unregister a rx_handler on a net_device, use
|
|
|
|
* netdev_rx_handler_unregister().
|
|
|
|
*
|
|
|
|
* Upon return, rx_handler is expected to tell __netif_receive_skb() what to
|
|
|
|
* do with the skb.
|
|
|
|
*
|
2016-03-24 02:47:23 +08:00
|
|
|
* If the rx_handler consumed the skb in some way, it should return
|
2011-03-12 11:14:39 +08:00
|
|
|
* RX_HANDLER_CONSUMED. This is appropriate when the rx_handler arranged for
|
2016-03-24 02:47:23 +08:00
|
|
|
* the skb to be delivered in some other way.
|
2011-03-12 11:14:39 +08:00
|
|
|
*
|
|
|
|
* If the rx_handler changed skb->dev, to divert the skb to another
|
|
|
|
* net_device, it should return RX_HANDLER_ANOTHER. The rx_handler for the
|
|
|
|
* new device will be called if it exists.
|
|
|
|
*
|
2016-03-24 02:47:23 +08:00
|
|
|
* If the rx_handler decides the skb should be ignored, it should return
|
2011-03-12 11:14:39 +08:00
|
|
|
* RX_HANDLER_EXACT. The skb will only be delivered to protocol handlers that
|
2012-09-20 09:47:58 +08:00
|
|
|
* are registered on exact device (ptype->dev == skb->dev).
|
2011-03-12 11:14:39 +08:00
|
|
|
*
|
2016-03-24 02:47:23 +08:00
|
|
|
* If the rx_handler didn't change skb->dev, but wants the skb to be normally
|
2011-03-12 11:14:39 +08:00
|
|
|
* delivered, it should return RX_HANDLER_PASS.
|
|
|
|
*
|
|
|
|
* A device without a registered rx_handler will behave as if rx_handler
|
|
|
|
* returned RX_HANDLER_PASS.
|
|
|
|
*/
|
|
|
|
|
|
|
|
enum rx_handler_result {
|
|
|
|
RX_HANDLER_CONSUMED,
|
|
|
|
RX_HANDLER_ANOTHER,
|
|
|
|
RX_HANDLER_EXACT,
|
|
|
|
RX_HANDLER_PASS,
|
|
|
|
};
|
|
|
|
typedef enum rx_handler_result rx_handler_result_t;
|
|
|
|
typedef rx_handler_result_t rx_handler_func_t(struct sk_buff **pskb);
|
2010-06-02 05:52:08 +08:00
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
void __napi_schedule(struct napi_struct *n);
|
2014-10-29 09:05:13 +08:00
|
|
|
void __napi_schedule_irqoff(struct napi_struct *n);
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool napi_disable_pending(struct napi_struct *n)
|
2008-01-08 12:35:07 +08:00
|
|
|
{
|
|
|
|
return test_bit(NAPI_STATE_DISABLE, &n->state);
|
|
|
|
}
|
|
|
|
|
net: Introduce preferred busy-polling
The existing busy-polling mode, enabled by the SO_BUSY_POLL socket
option or system-wide using the /proc/sys/net/core/busy_read knob, is
an opportunistic. That means that if the NAPI context is not
scheduled, it will poll it. If, after busy-polling, the budget is
exceeded the busy-polling logic will schedule the NAPI onto the
regular softirq handling.
One implication of the behavior above is that a busy/heavy loaded NAPI
context will never enter/allow for busy-polling. Some applications
prefer that most NAPI processing would be done by busy-polling.
This series adds a new socket option, SO_PREFER_BUSY_POLL, that works
in concert with the napi_defer_hard_irqs and gro_flush_timeout
knobs. The napi_defer_hard_irqs and gro_flush_timeout knobs were
introduced in commit 6f8b12d661d0 ("net: napi: add hard irqs deferral
feature"), and allows for a user to defer interrupts to be enabled and
instead schedule the NAPI context from a watchdog timer. When a user
enables the SO_PREFER_BUSY_POLL, again with the other knobs enabled,
and the NAPI context is being processed by a softirq, the softirq NAPI
processing will exit early to allow the busy-polling to be performed.
If the application stops performing busy-polling via a system call,
the watchdog timer defined by gro_flush_timeout will timeout, and
regular softirq handling will resume.
In summary; Heavy traffic applications that prefer busy-polling over
softirq processing should use this option.
Example usage:
$ echo 2 | sudo tee /sys/class/net/ens785f1/napi_defer_hard_irqs
$ echo 200000 | sudo tee /sys/class/net/ens785f1/gro_flush_timeout
Note that the timeout should be larger than the userspace processing
window, otherwise the watchdog will timeout and fall back to regular
softirq processing.
Enable the SO_BUSY_POLL/SO_PREFER_BUSY_POLL options on your socket.
Signed-off-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/bpf/20201130185205.196029-2-bjorn.topel@gmail.com
2020-12-01 02:51:56 +08:00
|
|
|
static inline bool napi_prefer_busy_poll(struct napi_struct *n)
|
|
|
|
{
|
|
|
|
return test_bit(NAPI_STATE_PREFER_BUSY_POLL, &n->state);
|
|
|
|
}
|
|
|
|
|
net: solve a NAPI race
While playing with mlx4 hardware timestamping of RX packets, I found
that some packets were received by TCP stack with a ~200 ms delay...
Since the timestamp was provided by the NIC, and my probe was added
in tcp_v4_rcv() while in BH handler, I was confident it was not
a sender issue, or a drop in the network.
This would happen with a very low probability, but hurting RPC
workloads.
A NAPI driver normally arms the IRQ after the napi_complete_done(),
after NAPI_STATE_SCHED is cleared, so that the hard irq handler can grab
it.
Problem is that if another point in the stack grabs NAPI_STATE_SCHED bit
while IRQ are not disabled, we might have later an IRQ firing and
finding this bit set, right before napi_complete_done() clears it.
This can happen with busy polling users, or if gro_flush_timeout is
used. But some other uses of napi_schedule() in drivers can cause this
as well.
thread 1 thread 2 (could be on same cpu, or not)
// busy polling or napi_watchdog()
napi_schedule();
...
napi->poll()
device polling:
read 2 packets from ring buffer
Additional 3rd packet is
available.
device hard irq
// does nothing because
NAPI_STATE_SCHED bit is owned by thread 1
napi_schedule();
napi_complete_done(napi, 2);
rearm_irq();
Note that rearm_irq() will not force the device to send an additional
IRQ for the packet it already signaled (3rd packet in my example)
This patch adds a new NAPI_STATE_MISSED bit, that napi_schedule_prep()
can set if it could not grab NAPI_STATE_SCHED
Then napi_complete_done() properly reschedules the napi to make sure
we do not miss something.
Since we manipulate multiple bits at once, use cmpxchg() like in
sk_busy_loop() to provide proper transactions.
In v2, I changed napi_watchdog() to use a relaxed variant of
napi_schedule_prep() : No need to set NAPI_STATE_MISSED from this point.
In v3, I added more details in the changelog and clears
NAPI_STATE_MISSED in busy_poll_stop()
In v4, I added the ideas given by Alexander Duyck in v3 review
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Alexander Duyck <alexander.duyck@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-03-01 02:34:50 +08:00
|
|
|
bool napi_schedule_prep(struct napi_struct *n);
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
|
|
|
|
/**
|
|
|
|
* napi_schedule - schedule NAPI poll
|
2016-03-24 02:47:23 +08:00
|
|
|
* @n: NAPI context
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
*
|
|
|
|
* Schedule NAPI poll routine to be called if it is not already
|
|
|
|
* running.
|
|
|
|
*/
|
|
|
|
static inline void napi_schedule(struct napi_struct *n)
|
|
|
|
{
|
|
|
|
if (napi_schedule_prep(n))
|
|
|
|
__napi_schedule(n);
|
|
|
|
}
|
|
|
|
|
2014-10-29 09:05:13 +08:00
|
|
|
/**
|
|
|
|
* napi_schedule_irqoff - schedule NAPI poll
|
2016-03-24 02:47:23 +08:00
|
|
|
* @n: NAPI context
|
2014-10-29 09:05:13 +08:00
|
|
|
*
|
|
|
|
* Variant of napi_schedule(), assuming hard irqs are masked.
|
|
|
|
*/
|
|
|
|
static inline void napi_schedule_irqoff(struct napi_struct *n)
|
|
|
|
{
|
|
|
|
if (napi_schedule_prep(n))
|
|
|
|
__napi_schedule_irqoff(n);
|
|
|
|
}
|
|
|
|
|
2007-10-10 06:47:37 +08:00
|
|
|
/* Try to reschedule poll. Called by dev->poll() after napi_complete(). */
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool napi_reschedule(struct napi_struct *napi)
|
2007-10-10 06:47:37 +08:00
|
|
|
{
|
|
|
|
if (napi_schedule_prep(napi)) {
|
|
|
|
__napi_schedule(napi);
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
return true;
|
2007-10-10 06:47:37 +08:00
|
|
|
}
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
return false;
|
2007-10-10 06:47:37 +08:00
|
|
|
}
|
|
|
|
|
2016-11-16 02:15:13 +08:00
|
|
|
bool napi_complete_done(struct napi_struct *n, int work_done);
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* napi_complete - NAPI processing complete
|
2016-03-24 02:47:23 +08:00
|
|
|
* @n: NAPI context
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
*
|
|
|
|
* Mark NAPI processing as complete.
|
net: gro: add a per device gro flush timer
Tuning coalescing parameters on NIC can be really hard.
Servers can handle both bulk and RPC like traffic, with conflicting
goals : bulk flows want as big GRO packets as possible, RPC want minimal
latencies.
To reach big GRO packets on 10Gbe NIC, one can use :
ethtool -C eth0 rx-usecs 4 rx-frames 44
But this penalizes rpc sessions, with an increase of latencies, up to
50% in some cases, as NICs generally do not force an interrupt when
a packet with TCP Push flag is received.
Some NICs do not have an absolute timer, only a timer rearmed for every
incoming packet.
This patch uses a different strategy : Let GRO stack decides what do do,
based on traffic pattern.
Packets with Push flag wont be delayed.
Packets without Push flag might be held in GRO engine, if we keep
receiving data.
This new mechanism is off by default, and shall be enabled by setting
/sys/class/net/ethX/gro_flush_timeout to a value in nanosecond.
To fully enable this mechanism, drivers should use napi_complete_done()
instead of napi_complete().
Tested:
Ran 200 netperf TCP_STREAM from A to B (10Gbe mlx4 link, 8 RX queues)
Without this feature, we send back about 305,000 ACK per second.
GRO aggregation ratio is low (811/305 = 2.65 segments per GRO packet)
Setting a timer of 2000 nsec is enough to increase GRO packet sizes
and reduce number of ACK packets. (811/19.2 = 42)
Receiver performs less calls to upper stacks, less wakes up.
This also reduces cpu usage on the sender, as it receives less ACK
packets.
Note that reducing number of wakes up increases cpu efficiency, but can
decrease QPS, as applications wont have the chance to warmup cpu caches
doing a partial read of RPC requests/answers if they fit in one skb.
B:~# sar -n DEV 1 10 | grep eth0 | tail -1
Average: eth0 811269.80 305732.30 1199462.57 19705.72 0.00
0.00 0.50
B:~# echo 2000 >/sys/class/net/eth0/gro_flush_timeout
B:~# sar -n DEV 1 10 | grep eth0 | tail -1
Average: eth0 811577.30 19230.80 1199916.51 1239.80 0.00
0.00 0.50
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-11-07 13:09:44 +08:00
|
|
|
* Consider using napi_complete_done() instead.
|
2016-11-16 02:15:13 +08:00
|
|
|
* Return false if device should avoid rearming interrupts.
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
*/
|
2016-11-16 02:15:13 +08:00
|
|
|
static inline bool napi_complete(struct napi_struct *n)
|
net: gro: add a per device gro flush timer
Tuning coalescing parameters on NIC can be really hard.
Servers can handle both bulk and RPC like traffic, with conflicting
goals : bulk flows want as big GRO packets as possible, RPC want minimal
latencies.
To reach big GRO packets on 10Gbe NIC, one can use :
ethtool -C eth0 rx-usecs 4 rx-frames 44
But this penalizes rpc sessions, with an increase of latencies, up to
50% in some cases, as NICs generally do not force an interrupt when
a packet with TCP Push flag is received.
Some NICs do not have an absolute timer, only a timer rearmed for every
incoming packet.
This patch uses a different strategy : Let GRO stack decides what do do,
based on traffic pattern.
Packets with Push flag wont be delayed.
Packets without Push flag might be held in GRO engine, if we keep
receiving data.
This new mechanism is off by default, and shall be enabled by setting
/sys/class/net/ethX/gro_flush_timeout to a value in nanosecond.
To fully enable this mechanism, drivers should use napi_complete_done()
instead of napi_complete().
Tested:
Ran 200 netperf TCP_STREAM from A to B (10Gbe mlx4 link, 8 RX queues)
Without this feature, we send back about 305,000 ACK per second.
GRO aggregation ratio is low (811/305 = 2.65 segments per GRO packet)
Setting a timer of 2000 nsec is enough to increase GRO packet sizes
and reduce number of ACK packets. (811/19.2 = 42)
Receiver performs less calls to upper stacks, less wakes up.
This also reduces cpu usage on the sender, as it receives less ACK
packets.
Note that reducing number of wakes up increases cpu efficiency, but can
decrease QPS, as applications wont have the chance to warmup cpu caches
doing a partial read of RPC requests/answers if they fit in one skb.
B:~# sar -n DEV 1 10 | grep eth0 | tail -1
Average: eth0 811269.80 305732.30 1199462.57 19705.72 0.00
0.00 0.50
B:~# echo 2000 >/sys/class/net/eth0/gro_flush_timeout
B:~# sar -n DEV 1 10 | grep eth0 | tail -1
Average: eth0 811577.30 19230.80 1199916.51 1239.80 0.00
0.00 0.50
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-11-07 13:09:44 +08:00
|
|
|
{
|
|
|
|
return napi_complete_done(n, 0);
|
|
|
|
}
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
|
2021-02-09 03:34:10 +08:00
|
|
|
int dev_set_threaded(struct net_device *dev, bool threaded);
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* napi_disable - prevent NAPI from scheduling
|
2016-03-24 02:47:23 +08:00
|
|
|
* @n: NAPI context
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
*
|
|
|
|
* Stop NAPI from being scheduled on this context.
|
|
|
|
* Waits till any outstanding processing completes.
|
|
|
|
*/
|
net: gro: add a per device gro flush timer
Tuning coalescing parameters on NIC can be really hard.
Servers can handle both bulk and RPC like traffic, with conflicting
goals : bulk flows want as big GRO packets as possible, RPC want minimal
latencies.
To reach big GRO packets on 10Gbe NIC, one can use :
ethtool -C eth0 rx-usecs 4 rx-frames 44
But this penalizes rpc sessions, with an increase of latencies, up to
50% in some cases, as NICs generally do not force an interrupt when
a packet with TCP Push flag is received.
Some NICs do not have an absolute timer, only a timer rearmed for every
incoming packet.
This patch uses a different strategy : Let GRO stack decides what do do,
based on traffic pattern.
Packets with Push flag wont be delayed.
Packets without Push flag might be held in GRO engine, if we keep
receiving data.
This new mechanism is off by default, and shall be enabled by setting
/sys/class/net/ethX/gro_flush_timeout to a value in nanosecond.
To fully enable this mechanism, drivers should use napi_complete_done()
instead of napi_complete().
Tested:
Ran 200 netperf TCP_STREAM from A to B (10Gbe mlx4 link, 8 RX queues)
Without this feature, we send back about 305,000 ACK per second.
GRO aggregation ratio is low (811/305 = 2.65 segments per GRO packet)
Setting a timer of 2000 nsec is enough to increase GRO packet sizes
and reduce number of ACK packets. (811/19.2 = 42)
Receiver performs less calls to upper stacks, less wakes up.
This also reduces cpu usage on the sender, as it receives less ACK
packets.
Note that reducing number of wakes up increases cpu efficiency, but can
decrease QPS, as applications wont have the chance to warmup cpu caches
doing a partial read of RPC requests/answers if they fit in one skb.
B:~# sar -n DEV 1 10 | grep eth0 | tail -1
Average: eth0 811269.80 305732.30 1199462.57 19705.72 0.00
0.00 0.50
B:~# echo 2000 >/sys/class/net/eth0/gro_flush_timeout
B:~# sar -n DEV 1 10 | grep eth0 | tail -1
Average: eth0 811577.30 19230.80 1199916.51 1239.80 0.00
0.00 0.50
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-11-07 13:09:44 +08:00
|
|
|
void napi_disable(struct napi_struct *n);
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
|
2021-02-09 03:34:09 +08:00
|
|
|
void napi_enable(struct napi_struct *n);
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
|
2007-10-18 04:26:41 +08:00
|
|
|
/**
|
|
|
|
* napi_synchronize - wait until NAPI is not running
|
2016-03-24 02:47:23 +08:00
|
|
|
* @n: NAPI context
|
2007-10-18 04:26:41 +08:00
|
|
|
*
|
|
|
|
* Wait until NAPI is done being scheduled on this context.
|
|
|
|
* Waits till any outstanding processing completes but
|
|
|
|
* does not disable future activations.
|
|
|
|
*/
|
|
|
|
static inline void napi_synchronize(const struct napi_struct *n)
|
|
|
|
{
|
2016-01-22 18:43:44 +08:00
|
|
|
if (IS_ENABLED(CONFIG_SMP))
|
|
|
|
while (test_bit(NAPI_STATE_SCHED, &n->state))
|
|
|
|
msleep(1);
|
|
|
|
else
|
|
|
|
barrier();
|
2007-10-18 04:26:41 +08:00
|
|
|
}
|
|
|
|
|
2018-08-28 20:44:28 +08:00
|
|
|
/**
|
|
|
|
* napi_if_scheduled_mark_missed - if napi is running, set the
|
|
|
|
* NAPIF_STATE_MISSED
|
|
|
|
* @n: NAPI context
|
|
|
|
*
|
|
|
|
* If napi is running, set the NAPIF_STATE_MISSED, and return true if
|
|
|
|
* NAPI is scheduled.
|
|
|
|
**/
|
|
|
|
static inline bool napi_if_scheduled_mark_missed(struct napi_struct *n)
|
|
|
|
{
|
|
|
|
unsigned long val, new;
|
|
|
|
|
2022-08-22 22:32:43 +08:00
|
|
|
val = READ_ONCE(n->state);
|
2018-08-28 20:44:28 +08:00
|
|
|
do {
|
|
|
|
if (val & NAPIF_STATE_DISABLE)
|
|
|
|
return true;
|
|
|
|
|
|
|
|
if (!(val & NAPIF_STATE_SCHED))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
new = val | NAPIF_STATE_MISSED;
|
2022-08-22 22:32:43 +08:00
|
|
|
} while (!try_cmpxchg(&n->state, &val, new));
|
2018-08-28 20:44:28 +08:00
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2009-11-05 01:50:58 +08:00
|
|
|
enum netdev_queue_state_t {
|
2011-11-29 00:32:44 +08:00
|
|
|
__QUEUE_STATE_DRV_XOFF,
|
|
|
|
__QUEUE_STATE_STACK_XOFF,
|
2008-08-01 07:58:50 +08:00
|
|
|
__QUEUE_STATE_FROZEN,
|
2008-07-09 14:14:46 +08:00
|
|
|
};
|
2014-04-03 02:52:57 +08:00
|
|
|
|
|
|
|
#define QUEUE_STATE_DRV_XOFF (1 << __QUEUE_STATE_DRV_XOFF)
|
|
|
|
#define QUEUE_STATE_STACK_XOFF (1 << __QUEUE_STATE_STACK_XOFF)
|
|
|
|
#define QUEUE_STATE_FROZEN (1 << __QUEUE_STATE_FROZEN)
|
|
|
|
|
|
|
|
#define QUEUE_STATE_ANY_XOFF (QUEUE_STATE_DRV_XOFF | QUEUE_STATE_STACK_XOFF)
|
|
|
|
#define QUEUE_STATE_ANY_XOFF_OR_FROZEN (QUEUE_STATE_ANY_XOFF | \
|
|
|
|
QUEUE_STATE_FROZEN)
|
|
|
|
#define QUEUE_STATE_DRV_XOFF_OR_FROZEN (QUEUE_STATE_DRV_XOFF | \
|
|
|
|
QUEUE_STATE_FROZEN)
|
|
|
|
|
2011-11-29 00:32:44 +08:00
|
|
|
/*
|
|
|
|
* __QUEUE_STATE_DRV_XOFF is used by drivers to stop the transmit queue. The
|
|
|
|
* netif_tx_* functions below are used to manipulate this flag. The
|
|
|
|
* __QUEUE_STATE_STACK_XOFF flag is used by the stack to stop the transmit
|
|
|
|
* queue independently. The netif_xmit_*stopped functions below are called
|
|
|
|
* to check if the queue has been stopped by the driver or stack (either
|
|
|
|
* of the XOFF bits are set in the state). Drivers should not need to call
|
|
|
|
* netif_xmit*stopped functions, they should only be using netif_tx_*.
|
|
|
|
*/
|
2008-07-09 14:14:46 +08:00
|
|
|
|
2008-07-09 07:55:56 +08:00
|
|
|
struct netdev_queue {
|
2009-04-28 19:43:42 +08:00
|
|
|
/*
|
2016-03-24 02:47:23 +08:00
|
|
|
* read-mostly part
|
2009-04-28 19:43:42 +08:00
|
|
|
*/
|
2008-07-09 07:55:56 +08:00
|
|
|
struct net_device *dev;
|
2021-12-05 12:21:59 +08:00
|
|
|
netdevice_tracker dev_tracker;
|
|
|
|
|
2014-09-13 11:04:52 +08:00
|
|
|
struct Qdisc __rcu *qdisc;
|
2008-07-09 08:42:10 +08:00
|
|
|
struct Qdisc *qdisc_sleeping;
|
2011-11-16 20:15:10 +08:00
|
|
|
#ifdef CONFIG_SYSFS
|
2010-11-21 21:17:27 +08:00
|
|
|
struct kobject kobj;
|
|
|
|
#endif
|
2010-11-29 16:14:37 +08:00
|
|
|
#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
|
|
|
|
int numa_node;
|
|
|
|
#endif
|
2016-05-03 09:29:09 +08:00
|
|
|
unsigned long tx_maxrate;
|
|
|
|
/*
|
|
|
|
* Number of TX timeouts for this queue
|
|
|
|
* (/sys/class/net/DEV/Q/trans_timeout)
|
|
|
|
*/
|
2021-11-17 11:29:21 +08:00
|
|
|
atomic_long_t trans_timeout;
|
2018-07-10 00:19:38 +08:00
|
|
|
|
|
|
|
/* Subordinate device that the queue has been assigned to */
|
|
|
|
struct net_device *sb_dev;
|
2018-10-01 20:51:33 +08:00
|
|
|
#ifdef CONFIG_XDP_SOCKETS
|
2020-08-28 16:26:15 +08:00
|
|
|
struct xsk_buff_pool *pool;
|
2018-10-01 20:51:33 +08:00
|
|
|
#endif
|
2009-04-28 19:43:42 +08:00
|
|
|
/*
|
2016-03-24 02:47:23 +08:00
|
|
|
* write-mostly part
|
2009-04-28 19:43:42 +08:00
|
|
|
*/
|
|
|
|
spinlock_t _xmit_lock ____cacheline_aligned_in_smp;
|
|
|
|
int xmit_lock_owner;
|
2009-05-18 11:55:16 +08:00
|
|
|
/*
|
2016-05-03 22:33:14 +08:00
|
|
|
* Time (in jiffies) of last Tx
|
2009-05-18 11:55:16 +08:00
|
|
|
*/
|
|
|
|
unsigned long trans_start;
|
2011-11-16 20:15:10 +08:00
|
|
|
|
2011-11-29 00:33:09 +08:00
|
|
|
unsigned long state;
|
|
|
|
|
|
|
|
#ifdef CONFIG_BQL
|
|
|
|
struct dql dql;
|
|
|
|
#endif
|
2008-07-17 15:34:19 +08:00
|
|
|
} ____cacheline_aligned_in_smp;
|
2008-07-09 07:55:56 +08:00
|
|
|
|
net: do not create fallback tunnels for non-default namespaces
fallback tunnels (like tunl0, gre0, gretap0, erspan0, sit0,
ip6tnl0, ip6gre0) are automatically created when the corresponding
module is loaded.
These tunnels are also automatically created when a new network
namespace is created, at a great cost.
In many cases, netns are used for isolation purposes, and these
extra network devices are a waste of resources. We are using
thousands of netns per host, and hit the netns creation/delete
bottleneck a lot. (Many thanks to Kirill for recent work on this)
Add a new sysctl so that we can opt-out from this automatic creation.
Note that these tunnels are still created for the initial namespace,
to be the least intrusive for typical setups.
Tested:
lpk43:~# cat add_del_unshare.sh
for i in `seq 1 40`
do
(for j in `seq 1 100` ; do unshare -n /bin/true >/dev/null ; done) &
done
wait
lpk43:~# echo 0 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m37.521s
user 0m0.886s
sys 7m7.084s
lpk43:~# echo 1 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m4.761s
user 0m0.851s
sys 1m8.343s
lpk43:~#
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-03-09 04:51:41 +08:00
|
|
|
extern int sysctl_fb_tunnels_only_for_init_net;
|
2019-01-18 15:27:11 +08:00
|
|
|
extern int sysctl_devconf_inherit_init_net;
|
net: do not create fallback tunnels for non-default namespaces
fallback tunnels (like tunl0, gre0, gretap0, erspan0, sit0,
ip6tnl0, ip6gre0) are automatically created when the corresponding
module is loaded.
These tunnels are also automatically created when a new network
namespace is created, at a great cost.
In many cases, netns are used for isolation purposes, and these
extra network devices are a waste of resources. We are using
thousands of netns per host, and hit the netns creation/delete
bottleneck a lot. (Many thanks to Kirill for recent work on this)
Add a new sysctl so that we can opt-out from this automatic creation.
Note that these tunnels are still created for the initial namespace,
to be the least intrusive for typical setups.
Tested:
lpk43:~# cat add_del_unshare.sh
for i in `seq 1 40`
do
(for j in `seq 1 100` ; do unshare -n /bin/true >/dev/null ; done) &
done
wait
lpk43:~# echo 0 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m37.521s
user 0m0.886s
sys 7m7.084s
lpk43:~# echo 1 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m4.761s
user 0m0.851s
sys 1m8.343s
lpk43:~#
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-03-09 04:51:41 +08:00
|
|
|
|
2020-08-27 00:05:35 +08:00
|
|
|
/*
|
|
|
|
* sysctl_fb_tunnels_only_for_init_net == 0 : For all netns
|
|
|
|
* == 1 : For initns only
|
|
|
|
* == 2 : For none.
|
|
|
|
*/
|
net: do not create fallback tunnels for non-default namespaces
fallback tunnels (like tunl0, gre0, gretap0, erspan0, sit0,
ip6tnl0, ip6gre0) are automatically created when the corresponding
module is loaded.
These tunnels are also automatically created when a new network
namespace is created, at a great cost.
In many cases, netns are used for isolation purposes, and these
extra network devices are a waste of resources. We are using
thousands of netns per host, and hit the netns creation/delete
bottleneck a lot. (Many thanks to Kirill for recent work on this)
Add a new sysctl so that we can opt-out from this automatic creation.
Note that these tunnels are still created for the initial namespace,
to be the least intrusive for typical setups.
Tested:
lpk43:~# cat add_del_unshare.sh
for i in `seq 1 40`
do
(for j in `seq 1 100` ; do unshare -n /bin/true >/dev/null ; done) &
done
wait
lpk43:~# echo 0 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m37.521s
user 0m0.886s
sys 7m7.084s
lpk43:~# echo 1 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m4.761s
user 0m0.851s
sys 1m8.343s
lpk43:~#
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-03-09 04:51:41 +08:00
|
|
|
static inline bool net_has_fallback_tunnels(const struct net *net)
|
|
|
|
{
|
2022-08-24 01:46:56 +08:00
|
|
|
#if IS_ENABLED(CONFIG_SYSCTL)
|
|
|
|
int fb_tunnels_only_for_init_net = READ_ONCE(sysctl_fb_tunnels_only_for_init_net);
|
|
|
|
|
|
|
|
return !fb_tunnels_only_for_init_net ||
|
|
|
|
(net_eq(net, &init_net) && fb_tunnels_only_for_init_net == 1);
|
|
|
|
#else
|
|
|
|
return true;
|
|
|
|
#endif
|
net: do not create fallback tunnels for non-default namespaces
fallback tunnels (like tunl0, gre0, gretap0, erspan0, sit0,
ip6tnl0, ip6gre0) are automatically created when the corresponding
module is loaded.
These tunnels are also automatically created when a new network
namespace is created, at a great cost.
In many cases, netns are used for isolation purposes, and these
extra network devices are a waste of resources. We are using
thousands of netns per host, and hit the netns creation/delete
bottleneck a lot. (Many thanks to Kirill for recent work on this)
Add a new sysctl so that we can opt-out from this automatic creation.
Note that these tunnels are still created for the initial namespace,
to be the least intrusive for typical setups.
Tested:
lpk43:~# cat add_del_unshare.sh
for i in `seq 1 40`
do
(for j in `seq 1 100` ; do unshare -n /bin/true >/dev/null ; done) &
done
wait
lpk43:~# echo 0 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m37.521s
user 0m0.886s
sys 7m7.084s
lpk43:~# echo 1 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m4.761s
user 0m0.851s
sys 1m8.343s
lpk43:~#
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-03-09 04:51:41 +08:00
|
|
|
}
|
|
|
|
|
2022-08-24 01:46:57 +08:00
|
|
|
static inline int net_inherit_devconf(void)
|
|
|
|
{
|
|
|
|
#if IS_ENABLED(CONFIG_SYSCTL)
|
|
|
|
return READ_ONCE(sysctl_devconf_inherit_init_net);
|
|
|
|
#else
|
|
|
|
return 0;
|
|
|
|
#endif
|
net: do not create fallback tunnels for non-default namespaces
fallback tunnels (like tunl0, gre0, gretap0, erspan0, sit0,
ip6tnl0, ip6gre0) are automatically created when the corresponding
module is loaded.
These tunnels are also automatically created when a new network
namespace is created, at a great cost.
In many cases, netns are used for isolation purposes, and these
extra network devices are a waste of resources. We are using
thousands of netns per host, and hit the netns creation/delete
bottleneck a lot. (Many thanks to Kirill for recent work on this)
Add a new sysctl so that we can opt-out from this automatic creation.
Note that these tunnels are still created for the initial namespace,
to be the least intrusive for typical setups.
Tested:
lpk43:~# cat add_del_unshare.sh
for i in `seq 1 40`
do
(for j in `seq 1 100` ; do unshare -n /bin/true >/dev/null ; done) &
done
wait
lpk43:~# echo 0 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m37.521s
user 0m0.886s
sys 7m7.084s
lpk43:~# echo 1 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m4.761s
user 0m0.851s
sys 1m8.343s
lpk43:~#
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-03-09 04:51:41 +08:00
|
|
|
}
|
|
|
|
|
2010-11-29 16:14:37 +08:00
|
|
|
static inline int netdev_queue_numa_node_read(const struct netdev_queue *q)
|
|
|
|
{
|
|
|
|
#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
|
|
|
|
return q->numa_node;
|
|
|
|
#else
|
2010-12-14 11:09:15 +08:00
|
|
|
return NUMA_NO_NODE;
|
2010-11-29 16:14:37 +08:00
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void netdev_queue_numa_node_write(struct netdev_queue *q, int node)
|
|
|
|
{
|
|
|
|
#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
|
|
|
|
q->numa_node = node;
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2010-03-25 03:13:54 +08:00
|
|
|
#ifdef CONFIG_RPS
|
2010-03-16 16:03:29 +08:00
|
|
|
/*
|
|
|
|
* This structure holds an RPS map which can be of variable length. The
|
|
|
|
* map is an array of CPUs.
|
|
|
|
*/
|
|
|
|
struct rps_map {
|
|
|
|
unsigned int len;
|
|
|
|
struct rcu_head rcu;
|
2020-03-02 19:59:33 +08:00
|
|
|
u16 cpus[];
|
2010-03-16 16:03:29 +08:00
|
|
|
};
|
2011-12-24 14:56:49 +08:00
|
|
|
#define RPS_MAP_SIZE(_num) (sizeof(struct rps_map) + ((_num) * sizeof(u16)))
|
2010-03-16 16:03:29 +08:00
|
|
|
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
/*
|
2011-01-19 19:03:53 +08:00
|
|
|
* The rps_dev_flow structure contains the mapping of a flow to a CPU, the
|
|
|
|
* tail pointer for that CPU's input queue at the time of last enqueue, and
|
|
|
|
* a hardware filter index.
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
*/
|
|
|
|
struct rps_dev_flow {
|
|
|
|
u16 cpu;
|
2011-01-19 19:03:53 +08:00
|
|
|
u16 filter;
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
unsigned int last_qtail;
|
|
|
|
};
|
2011-01-19 19:03:53 +08:00
|
|
|
#define RPS_NO_FILTER 0xffff
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The rps_dev_flow_table structure contains a table of flow mappings.
|
|
|
|
*/
|
|
|
|
struct rps_dev_flow_table {
|
|
|
|
unsigned int mask;
|
|
|
|
struct rcu_head rcu;
|
2020-03-02 19:59:33 +08:00
|
|
|
struct rps_dev_flow flows[];
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
};
|
|
|
|
#define RPS_DEV_FLOW_TABLE_SIZE(_num) (sizeof(struct rps_dev_flow_table) + \
|
2011-12-24 14:56:49 +08:00
|
|
|
((_num) * sizeof(struct rps_dev_flow)))
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The rps_sock_flow_table contains mappings of flows to the last CPU
|
|
|
|
* on which they were processed by the application (set in recvmsg).
|
2016-03-24 02:47:23 +08:00
|
|
|
* Each entry is a 32bit value. Upper part is the high-order bits
|
|
|
|
* of flow hash, lower part is CPU number.
|
net: rfs: add hash collision detection
Receive Flow Steering is a nice solution but suffers from
hash collisions when a mix of connected and unconnected traffic
is received on the host, when flow hash table is populated.
Also, clearing flow in inet_release() makes RFS not very good
for short lived flows, as many packets can follow close().
(FIN , ACK packets, ...)
This patch extends the information stored into global hash table
to not only include cpu number, but upper part of the hash value.
I use a 32bit value, and dynamically split it in two parts.
For host with less than 64 possible cpus, this gives 6 bits for the
cpu number, and 26 (32-6) bits for the upper part of the hash.
Since hash bucket selection use low order bits of the hash, we have
a full hash match, if /proc/sys/net/core/rps_sock_flow_entries is big
enough.
If the hash found in flow table does not match, we fallback to RPS (if
it is enabled for the rxqueue).
This means that a packet for an non connected flow can avoid the
IPI through a unrelated/victim CPU.
This also means we no longer have to clear the table at socket
close time, and this helps short lived flows performance.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-02-07 04:59:01 +08:00
|
|
|
* rps_cpu_mask is used to partition the space, depending on number of
|
2016-03-24 02:47:23 +08:00
|
|
|
* possible CPUs : rps_cpu_mask = roundup_pow_of_two(nr_cpu_ids) - 1
|
|
|
|
* For example, if 64 CPUs are possible, rps_cpu_mask = 0x3f,
|
net: rfs: add hash collision detection
Receive Flow Steering is a nice solution but suffers from
hash collisions when a mix of connected and unconnected traffic
is received on the host, when flow hash table is populated.
Also, clearing flow in inet_release() makes RFS not very good
for short lived flows, as many packets can follow close().
(FIN , ACK packets, ...)
This patch extends the information stored into global hash table
to not only include cpu number, but upper part of the hash value.
I use a 32bit value, and dynamically split it in two parts.
For host with less than 64 possible cpus, this gives 6 bits for the
cpu number, and 26 (32-6) bits for the upper part of the hash.
Since hash bucket selection use low order bits of the hash, we have
a full hash match, if /proc/sys/net/core/rps_sock_flow_entries is big
enough.
If the hash found in flow table does not match, we fallback to RPS (if
it is enabled for the rxqueue).
This means that a packet for an non connected flow can avoid the
IPI through a unrelated/victim CPU.
This also means we no longer have to clear the table at socket
close time, and this helps short lived flows performance.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-02-07 04:59:01 +08:00
|
|
|
* meaning we use 32-6=26 bits for the hash.
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
*/
|
|
|
|
struct rps_sock_flow_table {
|
net: rfs: add hash collision detection
Receive Flow Steering is a nice solution but suffers from
hash collisions when a mix of connected and unconnected traffic
is received on the host, when flow hash table is populated.
Also, clearing flow in inet_release() makes RFS not very good
for short lived flows, as many packets can follow close().
(FIN , ACK packets, ...)
This patch extends the information stored into global hash table
to not only include cpu number, but upper part of the hash value.
I use a 32bit value, and dynamically split it in two parts.
For host with less than 64 possible cpus, this gives 6 bits for the
cpu number, and 26 (32-6) bits for the upper part of the hash.
Since hash bucket selection use low order bits of the hash, we have
a full hash match, if /proc/sys/net/core/rps_sock_flow_entries is big
enough.
If the hash found in flow table does not match, we fallback to RPS (if
it is enabled for the rxqueue).
This means that a packet for an non connected flow can avoid the
IPI through a unrelated/victim CPU.
This also means we no longer have to clear the table at socket
close time, and this helps short lived flows performance.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-02-07 04:59:01 +08:00
|
|
|
u32 mask;
|
2015-02-09 12:39:13 +08:00
|
|
|
|
2020-03-02 19:59:33 +08:00
|
|
|
u32 ents[] ____cacheline_aligned_in_smp;
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
};
|
net: rfs: add hash collision detection
Receive Flow Steering is a nice solution but suffers from
hash collisions when a mix of connected and unconnected traffic
is received on the host, when flow hash table is populated.
Also, clearing flow in inet_release() makes RFS not very good
for short lived flows, as many packets can follow close().
(FIN , ACK packets, ...)
This patch extends the information stored into global hash table
to not only include cpu number, but upper part of the hash value.
I use a 32bit value, and dynamically split it in two parts.
For host with less than 64 possible cpus, this gives 6 bits for the
cpu number, and 26 (32-6) bits for the upper part of the hash.
Since hash bucket selection use low order bits of the hash, we have
a full hash match, if /proc/sys/net/core/rps_sock_flow_entries is big
enough.
If the hash found in flow table does not match, we fallback to RPS (if
it is enabled for the rxqueue).
This means that a packet for an non connected flow can avoid the
IPI through a unrelated/victim CPU.
This also means we no longer have to clear the table at socket
close time, and this helps short lived flows performance.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-02-07 04:59:01 +08:00
|
|
|
#define RPS_SOCK_FLOW_TABLE_SIZE(_num) (offsetof(struct rps_sock_flow_table, ents[_num]))
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
|
|
|
|
#define RPS_NO_CPU 0xffff
|
|
|
|
|
net: rfs: add hash collision detection
Receive Flow Steering is a nice solution but suffers from
hash collisions when a mix of connected and unconnected traffic
is received on the host, when flow hash table is populated.
Also, clearing flow in inet_release() makes RFS not very good
for short lived flows, as many packets can follow close().
(FIN , ACK packets, ...)
This patch extends the information stored into global hash table
to not only include cpu number, but upper part of the hash value.
I use a 32bit value, and dynamically split it in two parts.
For host with less than 64 possible cpus, this gives 6 bits for the
cpu number, and 26 (32-6) bits for the upper part of the hash.
Since hash bucket selection use low order bits of the hash, we have
a full hash match, if /proc/sys/net/core/rps_sock_flow_entries is big
enough.
If the hash found in flow table does not match, we fallback to RPS (if
it is enabled for the rxqueue).
This means that a packet for an non connected flow can avoid the
IPI through a unrelated/victim CPU.
This also means we no longer have to clear the table at socket
close time, and this helps short lived flows performance.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-02-07 04:59:01 +08:00
|
|
|
extern u32 rps_cpu_mask;
|
|
|
|
extern struct rps_sock_flow_table __rcu *rps_sock_flow_table;
|
|
|
|
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
static inline void rps_record_sock_flow(struct rps_sock_flow_table *table,
|
|
|
|
u32 hash)
|
|
|
|
{
|
|
|
|
if (table && hash) {
|
net: rfs: add hash collision detection
Receive Flow Steering is a nice solution but suffers from
hash collisions when a mix of connected and unconnected traffic
is received on the host, when flow hash table is populated.
Also, clearing flow in inet_release() makes RFS not very good
for short lived flows, as many packets can follow close().
(FIN , ACK packets, ...)
This patch extends the information stored into global hash table
to not only include cpu number, but upper part of the hash value.
I use a 32bit value, and dynamically split it in two parts.
For host with less than 64 possible cpus, this gives 6 bits for the
cpu number, and 26 (32-6) bits for the upper part of the hash.
Since hash bucket selection use low order bits of the hash, we have
a full hash match, if /proc/sys/net/core/rps_sock_flow_entries is big
enough.
If the hash found in flow table does not match, we fallback to RPS (if
it is enabled for the rxqueue).
This means that a packet for an non connected flow can avoid the
IPI through a unrelated/victim CPU.
This also means we no longer have to clear the table at socket
close time, and this helps short lived flows performance.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-02-07 04:59:01 +08:00
|
|
|
unsigned int index = hash & table->mask;
|
|
|
|
u32 val = hash & ~rps_cpu_mask;
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
|
2016-03-24 02:47:23 +08:00
|
|
|
/* We only give a hint, preemption can change CPU under us */
|
net: rfs: add hash collision detection
Receive Flow Steering is a nice solution but suffers from
hash collisions when a mix of connected and unconnected traffic
is received on the host, when flow hash table is populated.
Also, clearing flow in inet_release() makes RFS not very good
for short lived flows, as many packets can follow close().
(FIN , ACK packets, ...)
This patch extends the information stored into global hash table
to not only include cpu number, but upper part of the hash value.
I use a 32bit value, and dynamically split it in two parts.
For host with less than 64 possible cpus, this gives 6 bits for the
cpu number, and 26 (32-6) bits for the upper part of the hash.
Since hash bucket selection use low order bits of the hash, we have
a full hash match, if /proc/sys/net/core/rps_sock_flow_entries is big
enough.
If the hash found in flow table does not match, we fallback to RPS (if
it is enabled for the rxqueue).
This means that a packet for an non connected flow can avoid the
IPI through a unrelated/victim CPU.
This also means we no longer have to clear the table at socket
close time, and this helps short lived flows performance.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-02-07 04:59:01 +08:00
|
|
|
val |= raw_smp_processor_id();
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
|
net: rfs: add hash collision detection
Receive Flow Steering is a nice solution but suffers from
hash collisions when a mix of connected and unconnected traffic
is received on the host, when flow hash table is populated.
Also, clearing flow in inet_release() makes RFS not very good
for short lived flows, as many packets can follow close().
(FIN , ACK packets, ...)
This patch extends the information stored into global hash table
to not only include cpu number, but upper part of the hash value.
I use a 32bit value, and dynamically split it in two parts.
For host with less than 64 possible cpus, this gives 6 bits for the
cpu number, and 26 (32-6) bits for the upper part of the hash.
Since hash bucket selection use low order bits of the hash, we have
a full hash match, if /proc/sys/net/core/rps_sock_flow_entries is big
enough.
If the hash found in flow table does not match, we fallback to RPS (if
it is enabled for the rxqueue).
This means that a packet for an non connected flow can avoid the
IPI through a unrelated/victim CPU.
This also means we no longer have to clear the table at socket
close time, and this helps short lived flows performance.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-02-07 04:59:01 +08:00
|
|
|
if (table->ents[index] != val)
|
|
|
|
table->ents[index] = val;
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-01-19 19:03:53 +08:00
|
|
|
#ifdef CONFIG_RFS_ACCEL
|
2013-09-27 05:48:15 +08:00
|
|
|
bool rps_may_expire_flow(struct net_device *dev, u16 rxq_index, u32 flow_id,
|
|
|
|
u16 filter_id);
|
2011-01-19 19:03:53 +08:00
|
|
|
#endif
|
2014-01-17 14:23:28 +08:00
|
|
|
#endif /* CONFIG_RPS */
|
2011-01-19 19:03:53 +08:00
|
|
|
|
2010-03-16 16:03:29 +08:00
|
|
|
/* This structure contains an instance of an RX queue. */
|
|
|
|
struct netdev_rx_queue {
|
netdevice: move xdp_rxq within netdev_rx_queue
Both struct netdev_rx_queue and struct xdp_rxq_info are cacheline
aligned. This causes extra padding before and after the xdp_rxq
member. Move the member upfront, so that it's naturally aligned.
Before:
/* size: 256, cachelines: 4, members: 6 */
/* sum members: 160, holes: 1, sum holes: 40 */
/* padding: 56 */
/* paddings: 1, sum paddings: 36 */
/* forced alignments: 1, forced holes: 1, sum forced holes: 40 */
After:
/* size: 192, cachelines: 3, members: 6 */
/* padding: 32 */
/* paddings: 1, sum paddings: 36 */
/* forced alignments: 1 */
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Link: https://lore.kernel.org/r/20210823180135.1153608-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-08-24 02:01:35 +08:00
|
|
|
struct xdp_rxq_info xdp_rxq;
|
2014-01-17 14:23:28 +08:00
|
|
|
#ifdef CONFIG_RPS
|
2010-10-25 11:02:02 +08:00
|
|
|
struct rps_map __rcu *rps_map;
|
|
|
|
struct rps_dev_flow_table __rcu *rps_flow_table;
|
2014-01-17 14:23:28 +08:00
|
|
|
#endif
|
2010-10-25 11:02:02 +08:00
|
|
|
struct kobject kobj;
|
2010-11-09 18:47:38 +08:00
|
|
|
struct net_device *dev;
|
2021-12-05 12:21:58 +08:00
|
|
|
netdevice_tracker dev_tracker;
|
|
|
|
|
2018-10-01 20:51:33 +08:00
|
|
|
#ifdef CONFIG_XDP_SOCKETS
|
2020-08-28 16:26:15 +08:00
|
|
|
struct xsk_buff_pool *pool;
|
2018-10-01 20:51:33 +08:00
|
|
|
#endif
|
2010-03-16 16:03:29 +08:00
|
|
|
} ____cacheline_aligned_in_smp;
|
2014-01-17 14:23:28 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* RX queue sysfs structures and functions.
|
|
|
|
*/
|
|
|
|
struct rx_queue_attribute {
|
|
|
|
struct attribute attr;
|
2017-08-19 04:46:24 +08:00
|
|
|
ssize_t (*show)(struct netdev_rx_queue *queue, char *buf);
|
2014-01-17 14:23:28 +08:00
|
|
|
ssize_t (*store)(struct netdev_rx_queue *queue,
|
2017-08-19 04:46:24 +08:00
|
|
|
const char *buf, size_t len);
|
2014-01-17 14:23:28 +08:00
|
|
|
};
|
2008-11-20 13:32:24 +08:00
|
|
|
|
2021-03-19 02:37:46 +08:00
|
|
|
/* XPS map type and offset of the xps map within net_device->xps_maps[]. */
|
|
|
|
enum xps_map_type {
|
|
|
|
XPS_CPUS = 0,
|
|
|
|
XPS_RXQS,
|
|
|
|
XPS_MAPS_MAX,
|
|
|
|
};
|
|
|
|
|
2010-11-26 16:36:09 +08:00
|
|
|
#ifdef CONFIG_XPS
|
|
|
|
/*
|
|
|
|
* This structure holds an XPS map which can be of variable length. The
|
|
|
|
* map is an array of queues.
|
|
|
|
*/
|
|
|
|
struct xps_map {
|
|
|
|
unsigned int len;
|
|
|
|
unsigned int alloc_len;
|
|
|
|
struct rcu_head rcu;
|
2020-03-02 19:59:33 +08:00
|
|
|
u16 queues[];
|
2010-11-26 16:36:09 +08:00
|
|
|
};
|
2011-12-24 14:56:49 +08:00
|
|
|
#define XPS_MAP_SIZE(_num) (sizeof(struct xps_map) + ((_num) * sizeof(u16)))
|
2015-10-25 17:00:32 +08:00
|
|
|
#define XPS_MIN_MAP_ALLOC ((L1_CACHE_ALIGN(offsetof(struct xps_map, queues[1])) \
|
|
|
|
- sizeof(struct xps_map)) / sizeof(u16))
|
2010-11-26 16:36:09 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* This structure holds all XPS maps for device. Maps are indexed by CPU.
|
2021-03-19 02:37:43 +08:00
|
|
|
*
|
2021-03-19 02:37:44 +08:00
|
|
|
* We keep track of the number of cpus/rxqs used when the struct is allocated,
|
|
|
|
* in nr_ids. This will help not accessing out-of-bound memory.
|
|
|
|
*
|
2021-03-19 02:37:43 +08:00
|
|
|
* We keep track of the number of traffic classes used when the struct is
|
|
|
|
* allocated, in num_tc. This will be used to navigate the maps, to ensure we're
|
|
|
|
* not crossing its upper bound, as the original dev->num_tc can be updated in
|
|
|
|
* the meantime.
|
2010-11-26 16:36:09 +08:00
|
|
|
*/
|
|
|
|
struct xps_dev_maps {
|
|
|
|
struct rcu_head rcu;
|
2021-03-19 02:37:44 +08:00
|
|
|
unsigned int nr_ids;
|
2021-03-19 02:37:43 +08:00
|
|
|
s16 num_tc;
|
2020-03-02 19:59:33 +08:00
|
|
|
struct xps_map __rcu *attr_map[]; /* Either CPUs map or RXQs map */
|
2010-11-26 16:36:09 +08:00
|
|
|
};
|
2018-06-30 12:26:41 +08:00
|
|
|
|
|
|
|
#define XPS_CPU_DEV_MAPS_SIZE(_tcs) (sizeof(struct xps_dev_maps) + \
|
2016-10-28 23:50:13 +08:00
|
|
|
(nr_cpu_ids * (_tcs) * sizeof(struct xps_map *)))
|
2018-06-30 12:26:41 +08:00
|
|
|
|
|
|
|
#define XPS_RXQ_DEV_MAPS_SIZE(_tcs, _rxqs) (sizeof(struct xps_dev_maps) +\
|
|
|
|
(_rxqs * (_tcs) * sizeof(struct xps_map *)))
|
|
|
|
|
2010-11-26 16:36:09 +08:00
|
|
|
#endif /* CONFIG_XPS */
|
|
|
|
|
2011-01-17 16:06:04 +08:00
|
|
|
#define TC_MAX_QUEUE 16
|
|
|
|
#define TC_BITMASK 15
|
|
|
|
/* HW offloaded queuing disciplines txq count and offset maps */
|
|
|
|
struct netdev_tc_txq {
|
|
|
|
u16 count;
|
|
|
|
u16 offset;
|
|
|
|
};
|
|
|
|
|
2012-01-05 04:23:39 +08:00
|
|
|
#if defined(CONFIG_FCOE) || defined(CONFIG_FCOE_MODULE)
|
|
|
|
/*
|
|
|
|
* This structure is to hold information about the device
|
|
|
|
* configured to run FCoE protocol stack.
|
|
|
|
*/
|
|
|
|
struct netdev_fcoe_hbainfo {
|
|
|
|
char manufacturer[64];
|
|
|
|
char serial_number[64];
|
|
|
|
char hardware_version[64];
|
|
|
|
char driver_version[64];
|
|
|
|
char optionrom_version[64];
|
|
|
|
char firmware_version[64];
|
|
|
|
char model[256];
|
|
|
|
char model_description[256];
|
|
|
|
};
|
|
|
|
#endif
|
|
|
|
|
2014-11-28 21:34:16 +08:00
|
|
|
#define MAX_PHYS_ITEM_ID_LEN 32
|
2013-07-30 00:16:49 +08:00
|
|
|
|
2014-11-28 21:34:16 +08:00
|
|
|
/* This structure holds a unique identifier to identify some
|
|
|
|
* physical item (port for example) used by a netdevice.
|
2013-07-30 00:16:49 +08:00
|
|
|
*/
|
2014-11-28 21:34:16 +08:00
|
|
|
struct netdev_phys_item_id {
|
|
|
|
unsigned char id[MAX_PHYS_ITEM_ID_LEN];
|
2013-07-30 00:16:49 +08:00
|
|
|
unsigned char id_len;
|
|
|
|
};
|
|
|
|
|
2015-07-19 09:24:49 +08:00
|
|
|
static inline bool netdev_phys_item_id_same(struct netdev_phys_item_id *a,
|
|
|
|
struct netdev_phys_item_id *b)
|
|
|
|
{
|
|
|
|
return a->id_len == b->id_len &&
|
|
|
|
memcmp(a->id, b->id, a->id_len) == 0;
|
|
|
|
}
|
|
|
|
|
2014-02-16 22:55:20 +08:00
|
|
|
typedef u16 (*select_queue_fallback_t)(struct net_device *dev,
|
2018-07-10 00:20:04 +08:00
|
|
|
struct sk_buff *skb,
|
|
|
|
struct net_device *sb_dev);
|
2014-02-16 22:55:20 +08:00
|
|
|
|
2021-03-24 09:30:32 +08:00
|
|
|
enum net_device_path_type {
|
|
|
|
DEV_PATH_ETHERNET = 0,
|
2021-03-24 09:30:33 +08:00
|
|
|
DEV_PATH_VLAN,
|
2021-03-24 09:30:34 +08:00
|
|
|
DEV_PATH_BRIDGE,
|
2021-03-24 09:30:36 +08:00
|
|
|
DEV_PATH_PPPOE,
|
2021-03-24 09:30:37 +08:00
|
|
|
DEV_PATH_DSA,
|
2022-04-06 03:57:48 +08:00
|
|
|
DEV_PATH_MTK_WDMA,
|
2021-03-24 09:30:32 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
struct net_device_path {
|
|
|
|
enum net_device_path_type type;
|
|
|
|
const struct net_device *dev;
|
2021-03-24 09:30:33 +08:00
|
|
|
union {
|
|
|
|
struct {
|
|
|
|
u16 id;
|
|
|
|
__be16 proto;
|
2021-03-24 09:30:36 +08:00
|
|
|
u8 h_dest[ETH_ALEN];
|
2021-03-24 09:30:33 +08:00
|
|
|
} encap;
|
2021-03-24 09:30:35 +08:00
|
|
|
struct {
|
|
|
|
enum {
|
|
|
|
DEV_PATH_BR_VLAN_KEEP,
|
|
|
|
DEV_PATH_BR_VLAN_TAG,
|
|
|
|
DEV_PATH_BR_VLAN_UNTAG,
|
2021-03-24 09:30:48 +08:00
|
|
|
DEV_PATH_BR_VLAN_UNTAG_HW,
|
2021-03-24 09:30:35 +08:00
|
|
|
} vlan_mode;
|
|
|
|
u16 vlan_id;
|
|
|
|
__be16 vlan_proto;
|
|
|
|
} bridge;
|
2021-03-24 09:30:37 +08:00
|
|
|
struct {
|
|
|
|
int port;
|
|
|
|
u16 proto;
|
|
|
|
} dsa;
|
2022-04-06 03:57:48 +08:00
|
|
|
struct {
|
|
|
|
u8 wdma_idx;
|
|
|
|
u8 queue;
|
|
|
|
u16 wcid;
|
|
|
|
u8 bss;
|
|
|
|
} mtk_wdma;
|
2021-03-24 09:30:33 +08:00
|
|
|
};
|
2021-03-24 09:30:32 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
#define NET_DEVICE_PATH_STACK_MAX 5
|
2021-03-24 09:30:35 +08:00
|
|
|
#define NET_DEVICE_PATH_VLAN_MAX 2
|
2021-03-24 09:30:32 +08:00
|
|
|
|
|
|
|
struct net_device_path_stack {
|
|
|
|
int num_paths;
|
|
|
|
struct net_device_path path[NET_DEVICE_PATH_STACK_MAX];
|
|
|
|
};
|
|
|
|
|
|
|
|
struct net_device_path_ctx {
|
|
|
|
const struct net_device *dev;
|
2022-05-09 20:26:15 +08:00
|
|
|
u8 daddr[ETH_ALEN];
|
2021-03-24 09:30:35 +08:00
|
|
|
|
|
|
|
int num_vlans;
|
|
|
|
struct {
|
|
|
|
u16 id;
|
|
|
|
__be16 proto;
|
|
|
|
} vlan[NET_DEVICE_PATH_VLAN_MAX];
|
2021-03-24 09:30:32 +08:00
|
|
|
};
|
|
|
|
|
2017-08-07 16:15:17 +08:00
|
|
|
enum tc_setup_type {
|
2022-09-28 17:51:57 +08:00
|
|
|
TC_QUERY_CAPS,
|
2017-11-06 14:23:42 +08:00
|
|
|
TC_SETUP_QDISC_MQPRIO,
|
2016-02-17 13:17:09 +08:00
|
|
|
TC_SETUP_CLSU32,
|
2016-03-08 18:42:29 +08:00
|
|
|
TC_SETUP_CLSFLOWER,
|
2017-08-07 16:15:18 +08:00
|
|
|
TC_SETUP_CLSMATCHALL,
|
2016-09-21 18:43:53 +08:00
|
|
|
TC_SETUP_CLSBPF,
|
2017-10-19 21:50:29 +08:00
|
|
|
TC_SETUP_BLOCK,
|
2017-11-06 14:23:43 +08:00
|
|
|
TC_SETUP_QDISC_CBS,
|
2017-11-06 14:23:41 +08:00
|
|
|
TC_SETUP_QDISC_RED,
|
2018-01-14 19:33:15 +08:00
|
|
|
TC_SETUP_QDISC_PRIO,
|
2018-05-26 12:53:35 +08:00
|
|
|
TC_SETUP_QDISC_MQ,
|
net/sched: Introduce the ETF Qdisc
The ETF (Earliest TxTime First) qdisc uses the information added
earlier in this series (the socket option SO_TXTIME and the new
role of sk_buff->tstamp) to schedule packets transmission based
on absolute time.
For some workloads, just bandwidth enforcement is not enough, and
precise control of the transmission of packets is necessary.
Example:
$ tc qdisc replace dev enp2s0 parent root handle 100 mqprio num_tc 3 \
map 2 2 1 0 2 2 2 2 2 2 2 2 2 2 2 2 queues 1@0 1@1 2@2 hw 0
$ tc qdisc add dev enp2s0 parent 100:1 etf delta 100000 \
clockid CLOCK_TAI
In this example, the Qdisc will provide SW best-effort for the control
of the transmission time to the network adapter, the time stamp in the
socket will be in reference to the clockid CLOCK_TAI and packets
will leave the qdisc "delta" (100000) nanoseconds before its transmission
time.
The ETF qdisc will buffer packets sorted by their txtime. It will drop
packets on enqueue() if their skbuff clockid does not match the clock
reference of the Qdisc. Moreover, on dequeue(), a packet will be dropped
if it expires while being enqueued.
The qdisc also supports the SO_TXTIME deadline mode. For this mode, it
will dequeue a packet as soon as possible and change the skb timestamp
to 'now' during etf_dequeue().
Note that both the qdisc's and the SO_TXTIME ABIs allow for a clockid
to be configured, but it's been decided that usage of CLOCK_TAI should
be enforced until we decide to allow for other clockids to be used.
The rationale here is that PTP times are usually in the TAI scale, thus
no other clocks should be necessary. For now, the qdisc will return
EINVAL if any clocks other than CLOCK_TAI are used.
Signed-off-by: Jesus Sanchez-Palencia <jesus.sanchez-palencia@intel.com>
Signed-off-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-04 06:42:53 +08:00
|
|
|
TC_SETUP_QDISC_ETF,
|
2018-11-13 06:58:10 +08:00
|
|
|
TC_SETUP_ROOT_QDISC,
|
2018-11-20 07:21:42 +08:00
|
|
|
TC_SETUP_QDISC_GRED,
|
taprio: Add support for hardware offloading
This allows taprio to offload the schedule enforcement to capable
network cards, resulting in more precise windows and less CPU usage.
The gate mask acts on traffic classes (groups of queues of same
priority), as specified in IEEE 802.1Q-2018, and following the existing
taprio and mqprio semantics.
It is up to the driver to perform conversion between tc and individual
netdev queues if for some reason it needs to make that distinction.
Full offload is requested from the network interface by specifying
"flags 2" in the tc qdisc creation command, which in turn corresponds to
the TCA_TAPRIO_ATTR_FLAG_FULL_OFFLOAD bit.
The important detail here is the clockid which is implicitly /dev/ptpN
for full offload, and hence not configurable.
A reference counting API is added to support the use case where Ethernet
drivers need to keep the taprio offload structure locally (i.e. they are
a multi-port switch driver, and configuring a port depends on the
settings of other ports as well). The refcount_t variable is kept in a
private structure (__tc_taprio_qopt_offload) and not exposed to drivers.
In the future, the private structure might also be expanded with a
backpointer to taprio_sched *q, to implement the notification system
described in the patch (of when admin became oper, or an error occurred,
etc, so the offload can be monitored with 'tc qdisc show').
Signed-off-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Signed-off-by: Voon Weifeng <weifeng.voon@intel.com>
Signed-off-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-09-15 09:59:58 +08:00
|
|
|
TC_SETUP_QDISC_TAPRIO,
|
2019-11-12 07:29:56 +08:00
|
|
|
TC_SETUP_FT,
|
2019-12-18 22:55:15 +08:00
|
|
|
TC_SETUP_QDISC_ETS,
|
2020-01-24 21:23:06 +08:00
|
|
|
TC_SETUP_QDISC_TBF,
|
2020-03-05 15:16:40 +08:00
|
|
|
TC_SETUP_QDISC_FIFO,
|
sch_htb: Hierarchical QoS hardware offload
HTB doesn't scale well because of contention on a single lock, and it
also consumes CPU. This patch adds support for offloading HTB to
hardware that supports hierarchical rate limiting.
In the offload mode, HTB passes control commands to the driver using
ndo_setup_tc. The driver has to replicate the whole hierarchy of classes
and their settings (rate, ceil) in the NIC. Every modification of the
HTB tree caused by the admin results in ndo_setup_tc being called.
After this setup, the HTB algorithm is done completely in the NIC. An SQ
(send queue) is created for every leaf class and attached to the
hierarchy, so that the NIC can calculate and obey aggregated rate
limits, too. In the future, it can be changed, so that multiple SQs will
back a single leaf class.
ndo_select_queue is responsible for selecting the right queue that
serves the traffic class of each packet.
The data path works as follows: a packet is classified by clsact, the
driver selects a hardware queue according to its class, and the packet
is enqueued into this queue's qdisc.
This solution addresses two main problems of scaling HTB:
1. Contention by flow classification. Currently the filters are attached
to the HTB instance as follows:
# tc filter add dev eth0 parent 1:0 protocol ip flower dst_port 80
classid 1:10
It's possible to move classification to clsact egress hook, which is
thread-safe and lock-free:
# tc filter add dev eth0 egress protocol ip flower dst_port 80
action skbedit priority 1:10
This way classification still happens in software, but the lock
contention is eliminated, and it happens before selecting the TX queue,
allowing the driver to translate the class to the corresponding hardware
queue in ndo_select_queue.
Note that this is already compatible with non-offloaded HTB and doesn't
require changes to the kernel nor iproute2.
2. Contention by handling packets. HTB is not multi-queue, it attaches
to a whole net device, and handling of all packets takes the same lock.
When HTB is offloaded, it registers itself as a multi-queue qdisc,
similarly to mq: HTB is attached to the netdev, and each queue has its
own qdisc.
Some features of HTB may be not supported by some particular hardware,
for example, the maximum number of classes may be limited, the
granularity of rate and ceil parameters may be different, etc. - so, the
offload is not enabled by default, a new parameter is used to enable it:
# tc qdisc replace dev eth0 root handle 1: htb offload
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-19 20:08:13 +08:00
|
|
|
TC_SETUP_QDISC_HTB,
|
2021-12-18 02:16:22 +08:00
|
|
|
TC_SETUP_ACT,
|
2016-02-17 13:16:43 +08:00
|
|
|
};
|
|
|
|
|
2017-11-04 04:56:16 +08:00
|
|
|
/* These structures hold the attributes of bpf state that are being passed
|
|
|
|
* to the netdevice through the bpf op.
|
2016-07-20 03:16:48 +08:00
|
|
|
*/
|
2017-11-04 04:56:16 +08:00
|
|
|
enum bpf_netdev_command {
|
2016-07-20 03:16:48 +08:00
|
|
|
/* Set or clear a bpf program used in the earliest stages of packet
|
|
|
|
* rx. The prog will have been loaded as BPF_PROG_TYPE_XDP. The callee
|
|
|
|
* is responsible for calling bpf_prog_put on any old progs that are
|
|
|
|
* stored. In case of error, the callee need not release the new prog
|
|
|
|
* reference, but on success it takes ownership and must bpf_prog_put
|
|
|
|
* when it is no longer used.
|
|
|
|
*/
|
|
|
|
XDP_SETUP_PROG,
|
2017-06-22 09:25:04 +08:00
|
|
|
XDP_SETUP_PROG_HW,
|
2017-11-04 04:56:17 +08:00
|
|
|
/* BPF program for offload callbacks, invoked at program load time. */
|
2018-01-12 12:29:09 +08:00
|
|
|
BPF_OFFLOAD_MAP_ALLOC,
|
|
|
|
BPF_OFFLOAD_MAP_FREE,
|
2020-08-28 16:26:15 +08:00
|
|
|
XDP_SETUP_XSK_POOL,
|
2016-07-20 03:16:48 +08:00
|
|
|
};
|
|
|
|
|
2017-12-28 10:39:05 +08:00
|
|
|
struct bpf_prog_offload_ops;
|
2017-05-01 12:46:46 +08:00
|
|
|
struct netlink_ext_ack;
|
2018-06-04 20:05:53 +08:00
|
|
|
struct xdp_umem;
|
xdp: Move devmap bulk queue into struct net_device
Commit 96360004b862 ("xdp: Make devmap flush_list common for all map
instances"), changed devmap flushing to be a global operation instead of a
per-map operation. However, the queue structure used for bulking was still
allocated as part of the containing map.
This patch moves the devmap bulk queue into struct net_device. The
motivation for this is reusing it for the non-map variant of XDP_REDIRECT,
which will be changed in a subsequent commit. To avoid other fields of
struct net_device moving to different cache lines, we also move a couple of
other members around.
We defer the actual allocation of the bulk queue structure until the
NETDEV_REGISTER notification devmap.c. This makes it possible to check for
ndo_xdp_xmit support before allocating the structure, which is not possible
at the time struct net_device is allocated. However, we keep the freeing in
free_netdev() to avoid adding another RCU callback on NETDEV_UNREGISTER.
Because of this change, we lose the reference back to the map that
originated the redirect, so change the tracepoint to always return 0 as the
map ID and index. Otherwise no functional change is intended with this
patch.
After this patch, the relevant part of struct net_device looks like this,
according to pahole:
/* --- cacheline 14 boundary (896 bytes) --- */
struct netdev_queue * _tx __attribute__((__aligned__(64))); /* 896 8 */
unsigned int num_tx_queues; /* 904 4 */
unsigned int real_num_tx_queues; /* 908 4 */
struct Qdisc * qdisc; /* 912 8 */
unsigned int tx_queue_len; /* 920 4 */
spinlock_t tx_global_lock; /* 924 4 */
struct xdp_dev_bulk_queue * xdp_bulkq; /* 928 8 */
struct xps_dev_maps * xps_cpus_map; /* 936 8 */
struct xps_dev_maps * xps_rxqs_map; /* 944 8 */
struct mini_Qdisc * miniq_egress; /* 952 8 */
/* --- cacheline 15 boundary (960 bytes) --- */
struct hlist_head qdisc_hash[16]; /* 960 128 */
/* --- cacheline 17 boundary (1088 bytes) --- */
struct timer_list watchdog_timer; /* 1088 40 */
/* XXX last struct has 4 bytes of padding */
int watchdog_timeo; /* 1128 4 */
/* XXX 4 bytes hole, try to pack */
struct list_head todo_list; /* 1136 16 */
/* --- cacheline 18 boundary (1152 bytes) --- */
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Björn Töpel <bjorn.topel@intel.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/157918768397.1458396.12673224324627072349.stgit@toke.dk
2020-01-16 23:14:44 +08:00
|
|
|
struct xdp_dev_bulk_queue;
|
2020-07-22 14:45:57 +08:00
|
|
|
struct bpf_xdp_link;
|
2017-05-01 12:46:46 +08:00
|
|
|
|
2020-07-22 14:45:55 +08:00
|
|
|
enum bpf_xdp_mode {
|
|
|
|
XDP_MODE_SKB = 0,
|
|
|
|
XDP_MODE_DRV = 1,
|
|
|
|
XDP_MODE_HW = 2,
|
|
|
|
__MAX_XDP_MODE
|
|
|
|
};
|
|
|
|
|
|
|
|
struct bpf_xdp_entity {
|
|
|
|
struct bpf_prog *prog;
|
2020-07-22 14:45:57 +08:00
|
|
|
struct bpf_xdp_link *link;
|
2020-07-22 14:45:55 +08:00
|
|
|
};
|
2017-05-01 12:46:46 +08:00
|
|
|
|
2017-11-04 04:56:16 +08:00
|
|
|
struct netdev_bpf {
|
|
|
|
enum bpf_netdev_command command;
|
2016-07-20 03:16:48 +08:00
|
|
|
union {
|
|
|
|
/* XDP_SETUP_PROG */
|
2017-05-01 12:46:46 +08:00
|
|
|
struct {
|
2017-06-22 09:25:03 +08:00
|
|
|
u32 flags;
|
2017-05-01 12:46:46 +08:00
|
|
|
struct bpf_prog *prog;
|
|
|
|
struct netlink_ext_ack *extack;
|
|
|
|
};
|
2018-01-12 12:29:09 +08:00
|
|
|
/* BPF_OFFLOAD_MAP_ALLOC, BPF_OFFLOAD_MAP_FREE */
|
|
|
|
struct {
|
|
|
|
struct bpf_offloaded_map *offmap;
|
|
|
|
};
|
2020-08-28 16:26:15 +08:00
|
|
|
/* XDP_SETUP_XSK_POOL */
|
2018-06-04 20:05:53 +08:00
|
|
|
struct {
|
2020-08-28 16:26:15 +08:00
|
|
|
struct xsk_buff_pool *pool;
|
2019-02-14 01:07:29 +08:00
|
|
|
u16 queue_id;
|
2018-06-04 20:05:53 +08:00
|
|
|
} xsk;
|
2016-07-20 03:16:48 +08:00
|
|
|
};
|
|
|
|
};
|
2016-02-17 13:16:43 +08:00
|
|
|
|
2019-08-14 15:27:16 +08:00
|
|
|
/* Flags for ndo_xsk_wakeup. */
|
|
|
|
#define XDP_WAKEUP_RX (1 << 0)
|
|
|
|
#define XDP_WAKEUP_TX (1 << 1)
|
|
|
|
|
2017-04-14 16:06:10 +08:00
|
|
|
#ifdef CONFIG_XFRM_OFFLOAD
|
|
|
|
struct xfrmdev_ops {
|
|
|
|
int (*xdo_dev_state_add) (struct xfrm_state *x);
|
|
|
|
void (*xdo_dev_state_delete) (struct xfrm_state *x);
|
|
|
|
void (*xdo_dev_state_free) (struct xfrm_state *x);
|
|
|
|
bool (*xdo_dev_offload_ok) (struct sk_buff *skb,
|
|
|
|
struct xfrm_state *x);
|
2018-01-14 17:39:10 +08:00
|
|
|
void (*xdo_dev_state_advance_esn) (struct xfrm_state *x);
|
2017-04-14 16:06:10 +08:00
|
|
|
};
|
|
|
|
#endif
|
|
|
|
|
2017-10-03 05:50:05 +08:00
|
|
|
struct dev_ifalias {
|
|
|
|
struct rcu_head rcuhead;
|
|
|
|
char ifalias[];
|
|
|
|
};
|
|
|
|
|
2019-02-26 11:34:03 +08:00
|
|
|
struct devlink;
|
2019-04-26 03:32:03 +08:00
|
|
|
struct tlsdev_ops;
|
2019-02-26 11:34:03 +08:00
|
|
|
|
2020-01-25 19:17:08 +08:00
|
|
|
struct netdev_net_notifier {
|
|
|
|
struct list_head list;
|
|
|
|
struct notifier_block *nb;
|
|
|
|
};
|
|
|
|
|
2008-11-20 13:32:24 +08:00
|
|
|
/*
|
|
|
|
* This structure defines the management hooks for network devices.
|
2008-11-21 12:14:53 +08:00
|
|
|
* The following hooks can be defined; unless noted otherwise, they are
|
|
|
|
* optional and can be filled with a null pointer.
|
2008-11-20 13:32:24 +08:00
|
|
|
*
|
|
|
|
* int (*ndo_init)(struct net_device *dev);
|
2016-03-24 02:47:23 +08:00
|
|
|
* This function is called once when a network device is registered.
|
|
|
|
* The network device can use this for any late stage initialization
|
|
|
|
* or semantic validation. It can fail with an error code which will
|
|
|
|
* be propagated back to register_netdev.
|
2008-11-20 13:32:24 +08:00
|
|
|
*
|
|
|
|
* void (*ndo_uninit)(struct net_device *dev);
|
|
|
|
* This function is called when device is unregistered or when registration
|
|
|
|
* fails. It is not called if init fails.
|
|
|
|
*
|
|
|
|
* int (*ndo_open)(struct net_device *dev);
|
2016-03-24 02:47:23 +08:00
|
|
|
* This function is called when a network device transitions to the up
|
2008-11-20 13:32:24 +08:00
|
|
|
* state.
|
|
|
|
*
|
|
|
|
* int (*ndo_stop)(struct net_device *dev);
|
2016-03-24 02:47:23 +08:00
|
|
|
* This function is called when a network device transitions to the down
|
2008-11-20 13:32:24 +08:00
|
|
|
* state.
|
|
|
|
*
|
2009-09-01 03:50:40 +08:00
|
|
|
* netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
|
|
|
|
* struct net_device *dev);
|
2008-11-21 12:14:53 +08:00
|
|
|
* Called when a packet needs to be transmitted.
|
2015-04-03 19:47:17 +08:00
|
|
|
* Returns NETDEV_TX_OK. Can return NETDEV_TX_BUSY, but you should stop
|
|
|
|
* the queue before that can happen; it's for obsolete devices and weird
|
|
|
|
* corner cases, but the stack really does a non-trivial amount
|
|
|
|
* of useless work if you return NETDEV_TX_BUSY.
|
2016-03-24 02:47:23 +08:00
|
|
|
* Required; cannot be NULL.
|
2008-11-21 12:14:53 +08:00
|
|
|
*
|
2017-02-01 08:03:13 +08:00
|
|
|
* netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
|
|
|
|
* struct net_device *dev
|
|
|
|
* netdev_features_t features);
|
|
|
|
* Called by core transmit path to determine if device is capable of
|
|
|
|
* performing offload operations on a given packet. This is to give
|
|
|
|
* the device an opportunity to implement any restrictions that cannot
|
|
|
|
* be otherwise expressed by feature flags. The check is called with
|
|
|
|
* the set of features that the stack has calculated and it returns
|
|
|
|
* those the driver believes to be appropriate.
|
2016-01-06 22:53:50 +08:00
|
|
|
*
|
2014-01-10 16:18:26 +08:00
|
|
|
* u16 (*ndo_select_queue)(struct net_device *dev, struct sk_buff *skb,
|
2019-03-20 18:02:06 +08:00
|
|
|
* struct net_device *sb_dev);
|
2016-03-24 02:47:23 +08:00
|
|
|
* Called to decide which queue to use when device supports multiple
|
2008-11-21 12:14:53 +08:00
|
|
|
* transmit queues.
|
|
|
|
*
|
2008-11-20 13:32:24 +08:00
|
|
|
* void (*ndo_change_rx_flags)(struct net_device *dev, int flags);
|
|
|
|
* This function is called to allow device receiver to make
|
2016-03-24 02:47:23 +08:00
|
|
|
* changes to configuration when multicast or promiscuous is enabled.
|
2008-11-20 13:32:24 +08:00
|
|
|
*
|
|
|
|
* void (*ndo_set_rx_mode)(struct net_device *dev);
|
|
|
|
* This function is called device changes address list filtering.
|
2011-08-16 14:29:00 +08:00
|
|
|
* If driver handles unicast address filtering, it should set
|
2016-03-24 02:47:23 +08:00
|
|
|
* IFF_UNICAST_FLT in its priv_flags.
|
2008-11-20 13:32:24 +08:00
|
|
|
*
|
|
|
|
* int (*ndo_set_mac_address)(struct net_device *dev, void *addr);
|
|
|
|
* This function is called when the Media Access Control address
|
2009-04-27 20:45:54 +08:00
|
|
|
* needs to be changed. If this interface is not defined, the
|
2016-03-24 02:47:23 +08:00
|
|
|
* MAC address can not be changed.
|
2008-11-20 13:32:24 +08:00
|
|
|
*
|
|
|
|
* int (*ndo_validate_addr)(struct net_device *dev);
|
|
|
|
* Test if Media Access Control address is valid for the device.
|
|
|
|
*
|
|
|
|
* int (*ndo_do_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
|
2021-07-27 21:45:17 +08:00
|
|
|
* Old-style ioctl entry point. This is used internally by the
|
|
|
|
* appletalk and ieee802154 subsystems but is no longer called by
|
|
|
|
* the device ioctl handler.
|
|
|
|
*
|
|
|
|
* int (*ndo_siocbond)(struct net_device *dev, struct ifreq *ifr, int cmd);
|
|
|
|
* Used by the bonding driver for its device specific ioctls:
|
|
|
|
* SIOCBONDENSLAVE, SIOCBONDRELEASE, SIOCBONDSETHWADDR, SIOCBONDCHANGEACTIVE,
|
|
|
|
* SIOCBONDSLAVEINFOQUERY, and SIOCBONDINFOQUERY
|
2008-11-20 13:32:24 +08:00
|
|
|
*
|
2021-07-27 21:45:13 +08:00
|
|
|
* * int (*ndo_eth_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
|
|
|
|
* Called for ethernet specific ioctls: SIOCGMIIPHY, SIOCGMIIREG,
|
|
|
|
* SIOCSMIIREG, SIOCSHWTSTAMP and SIOCGHWTSTAMP.
|
|
|
|
*
|
2008-11-20 13:32:24 +08:00
|
|
|
* int (*ndo_set_config)(struct net_device *dev, struct ifmap *map);
|
|
|
|
* Used to set network devices bus interface parameters. This interface
|
2016-03-24 02:47:23 +08:00
|
|
|
* is retained for legacy reasons; new devices should use the bus
|
2008-11-20 13:32:24 +08:00
|
|
|
* interface (PCI) for low level management.
|
|
|
|
*
|
|
|
|
* int (*ndo_change_mtu)(struct net_device *dev, int new_mtu);
|
|
|
|
* Called when a user wants to change the Maximum Transfer Unit
|
2017-06-14 15:15:24 +08:00
|
|
|
* of a device.
|
2008-11-20 13:32:24 +08:00
|
|
|
*
|
netdev: pass the stuck queue to the timeout handler
This allows incrementing the correct timeout statistic without any mess.
Down the road, devices can learn to reset just the specific queue.
The patch was generated with the following script:
use strict;
use warnings;
our $^I = '.bak';
my @work = (
["arch/m68k/emu/nfeth.c", "nfeth_tx_timeout"],
["arch/um/drivers/net_kern.c", "uml_net_tx_timeout"],
["arch/um/drivers/vector_kern.c", "vector_net_tx_timeout"],
["arch/xtensa/platforms/iss/network.c", "iss_net_tx_timeout"],
["drivers/char/pcmcia/synclink_cs.c", "hdlcdev_tx_timeout"],
["drivers/infiniband/ulp/ipoib/ipoib_main.c", "ipoib_timeout"],
["drivers/infiniband/ulp/ipoib/ipoib_main.c", "ipoib_timeout"],
["drivers/message/fusion/mptlan.c", "mpt_lan_tx_timeout"],
["drivers/misc/sgi-xp/xpnet.c", "xpnet_dev_tx_timeout"],
["drivers/net/appletalk/cops.c", "cops_timeout"],
["drivers/net/arcnet/arcdevice.h", "arcnet_timeout"],
["drivers/net/arcnet/arcnet.c", "arcnet_timeout"],
["drivers/net/arcnet/com20020.c", "arcnet_timeout"],
["drivers/net/ethernet/3com/3c509.c", "el3_tx_timeout"],
["drivers/net/ethernet/3com/3c515.c", "corkscrew_timeout"],
["drivers/net/ethernet/3com/3c574_cs.c", "el3_tx_timeout"],
["drivers/net/ethernet/3com/3c589_cs.c", "el3_tx_timeout"],
["drivers/net/ethernet/3com/3c59x.c", "vortex_tx_timeout"],
["drivers/net/ethernet/3com/3c59x.c", "vortex_tx_timeout"],
["drivers/net/ethernet/3com/typhoon.c", "typhoon_tx_timeout"],
["drivers/net/ethernet/8390/8390.h", "ei_tx_timeout"],
["drivers/net/ethernet/8390/8390.h", "eip_tx_timeout"],
["drivers/net/ethernet/8390/8390.c", "ei_tx_timeout"],
["drivers/net/ethernet/8390/8390p.c", "eip_tx_timeout"],
["drivers/net/ethernet/8390/ax88796.c", "ax_ei_tx_timeout"],
["drivers/net/ethernet/8390/axnet_cs.c", "axnet_tx_timeout"],
["drivers/net/ethernet/8390/etherh.c", "__ei_tx_timeout"],
["drivers/net/ethernet/8390/hydra.c", "__ei_tx_timeout"],
["drivers/net/ethernet/8390/mac8390.c", "__ei_tx_timeout"],
["drivers/net/ethernet/8390/mcf8390.c", "__ei_tx_timeout"],
["drivers/net/ethernet/8390/lib8390.c", "__ei_tx_timeout"],
["drivers/net/ethernet/8390/ne2k-pci.c", "ei_tx_timeout"],
["drivers/net/ethernet/8390/pcnet_cs.c", "ei_tx_timeout"],
["drivers/net/ethernet/8390/smc-ultra.c", "ei_tx_timeout"],
["drivers/net/ethernet/8390/wd.c", "ei_tx_timeout"],
["drivers/net/ethernet/8390/zorro8390.c", "__ei_tx_timeout"],
["drivers/net/ethernet/adaptec/starfire.c", "tx_timeout"],
["drivers/net/ethernet/agere/et131x.c", "et131x_tx_timeout"],
["drivers/net/ethernet/allwinner/sun4i-emac.c", "emac_timeout"],
["drivers/net/ethernet/alteon/acenic.c", "ace_watchdog"],
["drivers/net/ethernet/amazon/ena/ena_netdev.c", "ena_tx_timeout"],
["drivers/net/ethernet/amd/7990.h", "lance_tx_timeout"],
["drivers/net/ethernet/amd/7990.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/a2065.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/am79c961a.c", "am79c961_timeout"],
["drivers/net/ethernet/amd/amd8111e.c", "amd8111e_tx_timeout"],
["drivers/net/ethernet/amd/ariadne.c", "ariadne_tx_timeout"],
["drivers/net/ethernet/amd/atarilance.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/au1000_eth.c", "au1000_tx_timeout"],
["drivers/net/ethernet/amd/declance.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/lance.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/mvme147.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/ni65.c", "ni65_timeout"],
["drivers/net/ethernet/amd/nmclan_cs.c", "mace_tx_timeout"],
["drivers/net/ethernet/amd/pcnet32.c", "pcnet32_tx_timeout"],
["drivers/net/ethernet/amd/sunlance.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/xgbe/xgbe-drv.c", "xgbe_tx_timeout"],
["drivers/net/ethernet/apm/xgene-v2/main.c", "xge_timeout"],
["drivers/net/ethernet/apm/xgene/xgene_enet_main.c", "xgene_enet_timeout"],
["drivers/net/ethernet/apple/macmace.c", "mace_tx_timeout"],
["drivers/net/ethernet/atheros/ag71xx.c", "ag71xx_tx_timeout"],
["drivers/net/ethernet/atheros/alx/main.c", "alx_tx_timeout"],
["drivers/net/ethernet/atheros/atl1c/atl1c_main.c", "atl1c_tx_timeout"],
["drivers/net/ethernet/atheros/atl1e/atl1e_main.c", "atl1e_tx_timeout"],
["drivers/net/ethernet/atheros/atlx/atl.c", "atlx_tx_timeout"],
["drivers/net/ethernet/atheros/atlx/atl1.c", "atlx_tx_timeout"],
["drivers/net/ethernet/atheros/atlx/atl2.c", "atl2_tx_timeout"],
["drivers/net/ethernet/broadcom/b44.c", "b44_tx_timeout"],
["drivers/net/ethernet/broadcom/bcmsysport.c", "bcm_sysport_tx_timeout"],
["drivers/net/ethernet/broadcom/bnx2.c", "bnx2_tx_timeout"],
["drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h", "bnx2x_tx_timeout"],
["drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c", "bnx2x_tx_timeout"],
["drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c", "bnx2x_tx_timeout"],
["drivers/net/ethernet/broadcom/bnxt/bnxt.c", "bnxt_tx_timeout"],
["drivers/net/ethernet/broadcom/genet/bcmgenet.c", "bcmgenet_timeout"],
["drivers/net/ethernet/broadcom/sb1250-mac.c", "sbmac_tx_timeout"],
["drivers/net/ethernet/broadcom/tg3.c", "tg3_tx_timeout"],
["drivers/net/ethernet/calxeda/xgmac.c", "xgmac_tx_timeout"],
["drivers/net/ethernet/cavium/liquidio/lio_main.c", "liquidio_tx_timeout"],
["drivers/net/ethernet/cavium/liquidio/lio_vf_main.c", "liquidio_tx_timeout"],
["drivers/net/ethernet/cavium/liquidio/lio_vf_rep.c", "lio_vf_rep_tx_timeout"],
["drivers/net/ethernet/cavium/thunder/nicvf_main.c", "nicvf_tx_timeout"],
["drivers/net/ethernet/cirrus/cs89x0.c", "net_timeout"],
["drivers/net/ethernet/cisco/enic/enic_main.c", "enic_tx_timeout"],
["drivers/net/ethernet/cisco/enic/enic_main.c", "enic_tx_timeout"],
["drivers/net/ethernet/cortina/gemini.c", "gmac_tx_timeout"],
["drivers/net/ethernet/davicom/dm9000.c", "dm9000_timeout"],
["drivers/net/ethernet/dec/tulip/de2104x.c", "de_tx_timeout"],
["drivers/net/ethernet/dec/tulip/tulip_core.c", "tulip_tx_timeout"],
["drivers/net/ethernet/dec/tulip/winbond-840.c", "tx_timeout"],
["drivers/net/ethernet/dlink/dl2k.c", "rio_tx_timeout"],
["drivers/net/ethernet/dlink/sundance.c", "tx_timeout"],
["drivers/net/ethernet/emulex/benet/be_main.c", "be_tx_timeout"],
["drivers/net/ethernet/ethoc.c", "ethoc_tx_timeout"],
["drivers/net/ethernet/faraday/ftgmac100.c", "ftgmac100_tx_timeout"],
["drivers/net/ethernet/fealnx.c", "fealnx_tx_timeout"],
["drivers/net/ethernet/freescale/dpaa/dpaa_eth.c", "dpaa_tx_timeout"],
["drivers/net/ethernet/freescale/fec_main.c", "fec_timeout"],
["drivers/net/ethernet/freescale/fec_mpc52xx.c", "mpc52xx_fec_tx_timeout"],
["drivers/net/ethernet/freescale/fs_enet/fs_enet-main.c", "fs_timeout"],
["drivers/net/ethernet/freescale/gianfar.c", "gfar_timeout"],
["drivers/net/ethernet/freescale/ucc_geth.c", "ucc_geth_timeout"],
["drivers/net/ethernet/fujitsu/fmvj18x_cs.c", "fjn_tx_timeout"],
["drivers/net/ethernet/google/gve/gve_main.c", "gve_tx_timeout"],
["drivers/net/ethernet/hisilicon/hip04_eth.c", "hip04_timeout"],
["drivers/net/ethernet/hisilicon/hix5hd2_gmac.c", "hix5hd2_net_timeout"],
["drivers/net/ethernet/hisilicon/hns/hns_enet.c", "hns_nic_net_timeout"],
["drivers/net/ethernet/hisilicon/hns3/hns3_enet.c", "hns3_nic_net_timeout"],
["drivers/net/ethernet/huawei/hinic/hinic_main.c", "hinic_tx_timeout"],
["drivers/net/ethernet/i825xx/82596.c", "i596_tx_timeout"],
["drivers/net/ethernet/i825xx/ether1.c", "ether1_timeout"],
["drivers/net/ethernet/i825xx/lib82596.c", "i596_tx_timeout"],
["drivers/net/ethernet/i825xx/sun3_82586.c", "sun3_82586_timeout"],
["drivers/net/ethernet/ibm/ehea/ehea_main.c", "ehea_tx_watchdog"],
["drivers/net/ethernet/ibm/emac/core.c", "emac_tx_timeout"],
["drivers/net/ethernet/ibm/emac/core.c", "emac_tx_timeout"],
["drivers/net/ethernet/ibm/ibmvnic.c", "ibmvnic_tx_timeout"],
["drivers/net/ethernet/intel/e100.c", "e100_tx_timeout"],
["drivers/net/ethernet/intel/e1000/e1000_main.c", "e1000_tx_timeout"],
["drivers/net/ethernet/intel/e1000e/netdev.c", "e1000_tx_timeout"],
["drivers/net/ethernet/intel/fm10k/fm10k_netdev.c", "fm10k_tx_timeout"],
["drivers/net/ethernet/intel/i40e/i40e_main.c", "i40e_tx_timeout"],
["drivers/net/ethernet/intel/iavf/iavf_main.c", "iavf_tx_timeout"],
["drivers/net/ethernet/intel/ice/ice_main.c", "ice_tx_timeout"],
["drivers/net/ethernet/intel/ice/ice_main.c", "ice_tx_timeout"],
["drivers/net/ethernet/intel/igb/igb_main.c", "igb_tx_timeout"],
["drivers/net/ethernet/intel/igbvf/netdev.c", "igbvf_tx_timeout"],
["drivers/net/ethernet/intel/ixgb/ixgb_main.c", "ixgb_tx_timeout"],
["drivers/net/ethernet/intel/ixgbe/ixgbe_debugfs.c", "adapter->netdev->netdev_ops->ndo_tx_timeout(adapter->netdev);"],
["drivers/net/ethernet/intel/ixgbe/ixgbe_main.c", "ixgbe_tx_timeout"],
["drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c", "ixgbevf_tx_timeout"],
["drivers/net/ethernet/jme.c", "jme_tx_timeout"],
["drivers/net/ethernet/korina.c", "korina_tx_timeout"],
["drivers/net/ethernet/lantiq_etop.c", "ltq_etop_tx_timeout"],
["drivers/net/ethernet/marvell/mv643xx_eth.c", "mv643xx_eth_tx_timeout"],
["drivers/net/ethernet/marvell/pxa168_eth.c", "pxa168_eth_tx_timeout"],
["drivers/net/ethernet/marvell/skge.c", "skge_tx_timeout"],
["drivers/net/ethernet/marvell/sky2.c", "sky2_tx_timeout"],
["drivers/net/ethernet/marvell/sky2.c", "sky2_tx_timeout"],
["drivers/net/ethernet/mediatek/mtk_eth_soc.c", "mtk_tx_timeout"],
["drivers/net/ethernet/mellanox/mlx4/en_netdev.c", "mlx4_en_tx_timeout"],
["drivers/net/ethernet/mellanox/mlx4/en_netdev.c", "mlx4_en_tx_timeout"],
["drivers/net/ethernet/mellanox/mlx5/core/en_main.c", "mlx5e_tx_timeout"],
["drivers/net/ethernet/micrel/ks8842.c", "ks8842_tx_timeout"],
["drivers/net/ethernet/micrel/ksz884x.c", "netdev_tx_timeout"],
["drivers/net/ethernet/microchip/enc28j60.c", "enc28j60_tx_timeout"],
["drivers/net/ethernet/microchip/encx24j600.c", "encx24j600_tx_timeout"],
["drivers/net/ethernet/natsemi/sonic.h", "sonic_tx_timeout"],
["drivers/net/ethernet/natsemi/sonic.c", "sonic_tx_timeout"],
["drivers/net/ethernet/natsemi/jazzsonic.c", "sonic_tx_timeout"],
["drivers/net/ethernet/natsemi/macsonic.c", "sonic_tx_timeout"],
["drivers/net/ethernet/natsemi/natsemi.c", "ns_tx_timeout"],
["drivers/net/ethernet/natsemi/ns83820.c", "ns83820_tx_timeout"],
["drivers/net/ethernet/natsemi/xtsonic.c", "sonic_tx_timeout"],
["drivers/net/ethernet/neterion/s2io.h", "s2io_tx_watchdog"],
["drivers/net/ethernet/neterion/s2io.c", "s2io_tx_watchdog"],
["drivers/net/ethernet/neterion/vxge/vxge-main.c", "vxge_tx_watchdog"],
["drivers/net/ethernet/netronome/nfp/nfp_net_common.c", "nfp_net_tx_timeout"],
["drivers/net/ethernet/nvidia/forcedeth.c", "nv_tx_timeout"],
["drivers/net/ethernet/nvidia/forcedeth.c", "nv_tx_timeout"],
["drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c", "pch_gbe_tx_timeout"],
["drivers/net/ethernet/packetengines/hamachi.c", "hamachi_tx_timeout"],
["drivers/net/ethernet/packetengines/yellowfin.c", "yellowfin_tx_timeout"],
["drivers/net/ethernet/pensando/ionic/ionic_lif.c", "ionic_tx_timeout"],
["drivers/net/ethernet/qlogic/netxen/netxen_nic_main.c", "netxen_tx_timeout"],
["drivers/net/ethernet/qlogic/qla3xxx.c", "ql3xxx_tx_timeout"],
["drivers/net/ethernet/qlogic/qlcnic/qlcnic_main.c", "qlcnic_tx_timeout"],
["drivers/net/ethernet/qualcomm/emac/emac.c", "emac_tx_timeout"],
["drivers/net/ethernet/qualcomm/qca_spi.c", "qcaspi_netdev_tx_timeout"],
["drivers/net/ethernet/qualcomm/qca_uart.c", "qcauart_netdev_tx_timeout"],
["drivers/net/ethernet/rdc/r6040.c", "r6040_tx_timeout"],
["drivers/net/ethernet/realtek/8139cp.c", "cp_tx_timeout"],
["drivers/net/ethernet/realtek/8139too.c", "rtl8139_tx_timeout"],
["drivers/net/ethernet/realtek/atp.c", "tx_timeout"],
["drivers/net/ethernet/realtek/r8169_main.c", "rtl8169_tx_timeout"],
["drivers/net/ethernet/renesas/ravb_main.c", "ravb_tx_timeout"],
["drivers/net/ethernet/renesas/sh_eth.c", "sh_eth_tx_timeout"],
["drivers/net/ethernet/renesas/sh_eth.c", "sh_eth_tx_timeout"],
["drivers/net/ethernet/samsung/sxgbe/sxgbe_main.c", "sxgbe_tx_timeout"],
["drivers/net/ethernet/seeq/ether3.c", "ether3_timeout"],
["drivers/net/ethernet/seeq/sgiseeq.c", "timeout"],
["drivers/net/ethernet/sfc/efx.c", "efx_watchdog"],
["drivers/net/ethernet/sfc/falcon/efx.c", "ef4_watchdog"],
["drivers/net/ethernet/sgi/ioc3-eth.c", "ioc3_timeout"],
["drivers/net/ethernet/sgi/meth.c", "meth_tx_timeout"],
["drivers/net/ethernet/silan/sc92031.c", "sc92031_tx_timeout"],
["drivers/net/ethernet/sis/sis190.c", "sis190_tx_timeout"],
["drivers/net/ethernet/sis/sis900.c", "sis900_tx_timeout"],
["drivers/net/ethernet/smsc/epic100.c", "epic_tx_timeout"],
["drivers/net/ethernet/smsc/smc911x.c", "smc911x_timeout"],
["drivers/net/ethernet/smsc/smc9194.c", "smc_timeout"],
["drivers/net/ethernet/smsc/smc91c92_cs.c", "smc_tx_timeout"],
["drivers/net/ethernet/smsc/smc91x.c", "smc_timeout"],
["drivers/net/ethernet/stmicro/stmmac/stmmac_main.c", "stmmac_tx_timeout"],
["drivers/net/ethernet/sun/cassini.c", "cas_tx_timeout"],
["drivers/net/ethernet/sun/ldmvsw.c", "sunvnet_tx_timeout_common"],
["drivers/net/ethernet/sun/niu.c", "niu_tx_timeout"],
["drivers/net/ethernet/sun/sunbmac.c", "bigmac_tx_timeout"],
["drivers/net/ethernet/sun/sungem.c", "gem_tx_timeout"],
["drivers/net/ethernet/sun/sunhme.c", "happy_meal_tx_timeout"],
["drivers/net/ethernet/sun/sunqe.c", "qe_tx_timeout"],
["drivers/net/ethernet/sun/sunvnet.c", "sunvnet_tx_timeout_common"],
["drivers/net/ethernet/sun/sunvnet_common.c", "sunvnet_tx_timeout_common"],
["drivers/net/ethernet/sun/sunvnet_common.h", "sunvnet_tx_timeout_common"],
["drivers/net/ethernet/synopsys/dwc-xlgmac-net.c", "xlgmac_tx_timeout"],
["drivers/net/ethernet/ti/cpmac.c", "cpmac_tx_timeout"],
["drivers/net/ethernet/ti/cpsw.c", "cpsw_ndo_tx_timeout"],
["drivers/net/ethernet/ti/cpsw_priv.c", "cpsw_ndo_tx_timeout"],
["drivers/net/ethernet/ti/cpsw_priv.h", "cpsw_ndo_tx_timeout"],
["drivers/net/ethernet/ti/davinci_emac.c", "emac_dev_tx_timeout"],
["drivers/net/ethernet/ti/netcp_core.c", "netcp_ndo_tx_timeout"],
["drivers/net/ethernet/ti/tlan.c", "tlan_tx_timeout"],
["drivers/net/ethernet/toshiba/ps3_gelic_net.h", "gelic_net_tx_timeout"],
["drivers/net/ethernet/toshiba/ps3_gelic_net.c", "gelic_net_tx_timeout"],
["drivers/net/ethernet/toshiba/ps3_gelic_wireless.c", "gelic_net_tx_timeout"],
["drivers/net/ethernet/toshiba/spider_net.c", "spider_net_tx_timeout"],
["drivers/net/ethernet/toshiba/tc35815.c", "tc35815_tx_timeout"],
["drivers/net/ethernet/via/via-rhine.c", "rhine_tx_timeout"],
["drivers/net/ethernet/wiznet/w5100.c", "w5100_tx_timeout"],
["drivers/net/ethernet/wiznet/w5300.c", "w5300_tx_timeout"],
["drivers/net/ethernet/xilinx/xilinx_emaclite.c", "xemaclite_tx_timeout"],
["drivers/net/ethernet/xircom/xirc2ps_cs.c", "xirc_tx_timeout"],
["drivers/net/fjes/fjes_main.c", "fjes_tx_retry"],
["drivers/net/slip/slip.c", "sl_tx_timeout"],
["include/linux/usb/usbnet.h", "usbnet_tx_timeout"],
["drivers/net/usb/aqc111.c", "usbnet_tx_timeout"],
["drivers/net/usb/asix_devices.c", "usbnet_tx_timeout"],
["drivers/net/usb/asix_devices.c", "usbnet_tx_timeout"],
["drivers/net/usb/asix_devices.c", "usbnet_tx_timeout"],
["drivers/net/usb/ax88172a.c", "usbnet_tx_timeout"],
["drivers/net/usb/ax88179_178a.c", "usbnet_tx_timeout"],
["drivers/net/usb/catc.c", "catc_tx_timeout"],
["drivers/net/usb/cdc_mbim.c", "usbnet_tx_timeout"],
["drivers/net/usb/cdc_ncm.c", "usbnet_tx_timeout"],
["drivers/net/usb/dm9601.c", "usbnet_tx_timeout"],
["drivers/net/usb/hso.c", "hso_net_tx_timeout"],
["drivers/net/usb/int51x1.c", "usbnet_tx_timeout"],
["drivers/net/usb/ipheth.c", "ipheth_tx_timeout"],
["drivers/net/usb/kaweth.c", "kaweth_tx_timeout"],
["drivers/net/usb/lan78xx.c", "lan78xx_tx_timeout"],
["drivers/net/usb/mcs7830.c", "usbnet_tx_timeout"],
["drivers/net/usb/pegasus.c", "pegasus_tx_timeout"],
["drivers/net/usb/qmi_wwan.c", "usbnet_tx_timeout"],
["drivers/net/usb/r8152.c", "rtl8152_tx_timeout"],
["drivers/net/usb/rndis_host.c", "usbnet_tx_timeout"],
["drivers/net/usb/rtl8150.c", "rtl8150_tx_timeout"],
["drivers/net/usb/sierra_net.c", "usbnet_tx_timeout"],
["drivers/net/usb/smsc75xx.c", "usbnet_tx_timeout"],
["drivers/net/usb/smsc95xx.c", "usbnet_tx_timeout"],
["drivers/net/usb/sr9700.c", "usbnet_tx_timeout"],
["drivers/net/usb/sr9800.c", "usbnet_tx_timeout"],
["drivers/net/usb/usbnet.c", "usbnet_tx_timeout"],
["drivers/net/vmxnet3/vmxnet3_drv.c", "vmxnet3_tx_timeout"],
["drivers/net/wan/cosa.c", "cosa_net_timeout"],
["drivers/net/wan/farsync.c", "fst_tx_timeout"],
["drivers/net/wan/fsl_ucc_hdlc.c", "uhdlc_tx_timeout"],
["drivers/net/wan/lmc/lmc_main.c", "lmc_driver_timeout"],
["drivers/net/wan/x25_asy.c", "x25_asy_timeout"],
["drivers/net/wimax/i2400m/netdev.c", "i2400m_tx_timeout"],
["drivers/net/wireless/intel/ipw2x00/ipw2100.c", "ipw2100_tx_timeout"],
["drivers/net/wireless/intersil/hostap/hostap_main.c", "prism2_tx_timeout"],
["drivers/net/wireless/intersil/hostap/hostap_main.c", "prism2_tx_timeout"],
["drivers/net/wireless/intersil/hostap/hostap_main.c", "prism2_tx_timeout"],
["drivers/net/wireless/intersil/orinoco/main.c", "orinoco_tx_timeout"],
["drivers/net/wireless/intersil/orinoco/orinoco_usb.c", "orinoco_tx_timeout"],
["drivers/net/wireless/intersil/orinoco/orinoco.h", "orinoco_tx_timeout"],
["drivers/net/wireless/intersil/prism54/islpci_dev.c", "islpci_eth_tx_timeout"],
["drivers/net/wireless/intersil/prism54/islpci_eth.c", "islpci_eth_tx_timeout"],
["drivers/net/wireless/intersil/prism54/islpci_eth.h", "islpci_eth_tx_timeout"],
["drivers/net/wireless/marvell/mwifiex/main.c", "mwifiex_tx_timeout"],
["drivers/net/wireless/quantenna/qtnfmac/core.c", "qtnf_netdev_tx_timeout"],
["drivers/net/wireless/quantenna/qtnfmac/core.h", "qtnf_netdev_tx_timeout"],
["drivers/net/wireless/rndis_wlan.c", "usbnet_tx_timeout"],
["drivers/net/wireless/wl3501_cs.c", "wl3501_tx_timeout"],
["drivers/net/wireless/zydas/zd1201.c", "zd1201_tx_timeout"],
["drivers/s390/net/qeth_core.h", "qeth_tx_timeout"],
["drivers/s390/net/qeth_core_main.c", "qeth_tx_timeout"],
["drivers/s390/net/qeth_l2_main.c", "qeth_tx_timeout"],
["drivers/s390/net/qeth_l2_main.c", "qeth_tx_timeout"],
["drivers/s390/net/qeth_l3_main.c", "qeth_tx_timeout"],
["drivers/s390/net/qeth_l3_main.c", "qeth_tx_timeout"],
["drivers/staging/ks7010/ks_wlan_net.c", "ks_wlan_tx_timeout"],
["drivers/staging/qlge/qlge_main.c", "qlge_tx_timeout"],
["drivers/staging/rtl8192e/rtl8192e/rtl_core.c", "_rtl92e_tx_timeout"],
["drivers/staging/rtl8192u/r8192U_core.c", "tx_timeout"],
["drivers/staging/unisys/visornic/visornic_main.c", "visornic_xmit_timeout"],
["drivers/staging/wlan-ng/p80211netdev.c", "p80211knetdev_tx_timeout"],
["drivers/tty/n_gsm.c", "gsm_mux_net_tx_timeout"],
["drivers/tty/synclink.c", "hdlcdev_tx_timeout"],
["drivers/tty/synclink_gt.c", "hdlcdev_tx_timeout"],
["drivers/tty/synclinkmp.c", "hdlcdev_tx_timeout"],
["net/atm/lec.c", "lec_tx_timeout"],
["net/bluetooth/bnep/netdev.c", "bnep_net_timeout"]
);
for my $p (@work) {
my @pair = @$p;
my $file = $pair[0];
my $func = $pair[1];
print STDERR $file , ": ", $func,"\n";
our @ARGV = ($file);
while (<ARGV>) {
if (m/($func\s*\(struct\s+net_device\s+\*[A-Za-z_]?[A-Za-z-0-9_]*)(\))/) {
print STDERR "found $1+$2 in $file\n";
}
if (s/($func\s*\(struct\s+net_device\s+\*[A-Za-z_]?[A-Za-z-0-9_]*)(\))/$1, unsigned int txqueue$2/) {
print STDERR "$func found in $file\n";
}
print;
}
}
where the list of files and functions is simply from:
git grep ndo_tx_timeout, with manual addition of headers
in the rare cases where the function is from a header,
then manually changing the few places which actually
call ndo_tx_timeout.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Heiner Kallweit <hkallweit1@gmail.com>
Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Acked-by: Shannon Nelson <snelson@pensando.io>
Reviewed-by: Martin Habets <mhabets@solarflare.com>
changes from v9:
fixup a forward declaration
changes from v9:
more leftovers from v3 change
changes from v8:
fix up a missing direct call to timeout
rebased on net-next
changes from v7:
fixup leftovers from v3 change
changes from v6:
fix typo in rtl driver
changes from v5:
add missing files (allow any net device argument name)
changes from v4:
add a missing driver header
changes from v3:
change queue # to unsigned
Changes from v2:
added headers
Changes from v1:
Fix errors found by kbuild:
generalize the pattern a bit, to pick up
a couple of instances missed by the previous
version.
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-10 22:23:51 +08:00
|
|
|
* void (*ndo_tx_timeout)(struct net_device *dev, unsigned int txqueue);
|
2016-03-24 02:47:23 +08:00
|
|
|
* Callback used when the transmitter has not made any progress
|
2008-11-20 13:32:24 +08:00
|
|
|
* for dev->watchdog ticks.
|
|
|
|
*
|
2017-01-07 11:12:52 +08:00
|
|
|
* void (*ndo_get_stats64)(struct net_device *dev,
|
|
|
|
* struct rtnl_link_stats64 *storage);
|
2009-10-08 04:53:11 +08:00
|
|
|
* struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
|
2008-11-20 13:32:24 +08:00
|
|
|
* Called when a user wants to get the network device usage
|
2010-06-08 15:19:54 +08:00
|
|
|
* statistics. Drivers must do one of the following:
|
2010-07-09 17:11:52 +08:00
|
|
|
* 1. Define @ndo_get_stats64 to fill in a zero-initialised
|
|
|
|
* rtnl_link_stats64 structure passed by the caller.
|
2010-06-16 06:08:48 +08:00
|
|
|
* 2. Define @ndo_get_stats to update a net_device_stats structure
|
2010-06-08 15:19:54 +08:00
|
|
|
* (which should normally be dev->stats) and return a pointer to
|
|
|
|
* it. The structure may be changed asynchronously only if each
|
|
|
|
* field is written atomically.
|
|
|
|
* 3. Update dev->stats asynchronously and atomically, and define
|
|
|
|
* neither operation.
|
2008-11-20 13:32:24 +08:00
|
|
|
*
|
2016-11-23 05:09:54 +08:00
|
|
|
* bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id)
|
2016-09-16 21:05:36 +08:00
|
|
|
* Return true if this device supports offload stats of this attr_id.
|
|
|
|
*
|
|
|
|
* int (*ndo_get_offload_stats)(int attr_id, const struct net_device *dev,
|
|
|
|
* void *attr_data)
|
|
|
|
* Get statistics for offload operations by attr_id. Write it into the
|
|
|
|
* attr_data pointer.
|
|
|
|
*
|
2015-01-12 17:16:25 +08:00
|
|
|
* int (*ndo_vlan_rx_add_vid)(struct net_device *dev, __be16 proto, u16 vid);
|
2016-03-24 02:47:23 +08:00
|
|
|
* If device supports VLAN filtering this function is called when a
|
2013-04-19 10:04:28 +08:00
|
|
|
* VLAN id is registered.
|
2008-11-20 13:32:24 +08:00
|
|
|
*
|
2015-01-12 17:16:25 +08:00
|
|
|
* int (*ndo_vlan_rx_kill_vid)(struct net_device *dev, __be16 proto, u16 vid);
|
2016-03-24 02:47:23 +08:00
|
|
|
* If device supports VLAN filtering this function is called when a
|
2013-04-19 10:04:28 +08:00
|
|
|
* VLAN id is unregistered.
|
2008-11-20 13:32:24 +08:00
|
|
|
*
|
|
|
|
* void (*ndo_poll_controller)(struct net_device *dev);
|
2010-02-10 09:43:46 +08:00
|
|
|
*
|
|
|
|
* SR-IOV management functions.
|
|
|
|
* int (*ndo_set_vf_mac)(struct net_device *dev, int vf, u8* mac);
|
2016-09-22 17:11:15 +08:00
|
|
|
* int (*ndo_set_vf_vlan)(struct net_device *dev, int vf, u16 vlan,
|
|
|
|
* u8 qos, __be16 proto);
|
net-next:v4: Add support to configure SR-IOV VF minimum and maximum Tx rate through ip tool.
o min_tx_rate puts lower limit on the VF bandwidth. VF is guaranteed
to have a bandwidth of at least this value.
max_tx_rate puts cap on the VF bandwidth. VF can have a bandwidth
of up to this value.
o A new handler set_vf_rate for attr IFLA_VF_RATE has been introduced
which takes 4 arguments:
netdev, VF number, min_tx_rate, max_tx_rate
o ndo_set_vf_rate replaces ndo_set_vf_tx_rate handler.
o Drivers that currently implement ndo_set_vf_tx_rate should now call
ndo_set_vf_rate instead and reject attempt to set a minimum bandwidth
greater than 0 for IFLA_VF_TX_RATE when IFLA_VF_RATE is not yet
implemented by driver.
o If user enters only one of either min_tx_rate or max_tx_rate, then,
userland should read back the other value from driver and set both
for IFLA_VF_RATE.
Drivers that have not yet implemented IFLA_VF_RATE should always
return min_tx_rate as 0 when read from ip tool.
o If both IFLA_VF_TX_RATE and IFLA_VF_RATE options are specified, then
IFLA_VF_RATE should override.
o Idea is to have consistent display of rate values to user.
o Usage example: -
./ip link set p4p1 vf 0 rate 900
./ip link show p4p1
32: p4p1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode
DEFAULT qlen 1000
link/ether 00:0e:1e:08:b0:f0 brd ff:ff:ff:ff:ff:ff
vf 0 MAC 3e:a0:ca:bd:ae:5a, tx rate 900 (Mbps), max_tx_rate 900Mbps
vf 1 MAC f6:c6:7c:3f:3d:6c
vf 2 MAC 56:32:43:98:d7:71
vf 3 MAC d6:be:c3:b5:85:ff
vf 4 MAC ee:a9:9a:1e:19:14
vf 5 MAC 4a:d0:4c:07:52:18
vf 6 MAC 3a:76:44:93:62:f9
vf 7 MAC 82:e9:e7:e3:15:1a
./ip link set p4p1 vf 0 max_tx_rate 300 min_tx_rate 200
./ip link show p4p1
32: p4p1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode
DEFAULT qlen 1000
link/ether 00:0e:1e:08:b0:f0 brd ff:ff:ff:ff:ff:ff
vf 0 MAC 3e:a0:ca:bd:ae:5a, tx rate 300 (Mbps), max_tx_rate 300Mbps,
min_tx_rate 200Mbps
vf 1 MAC f6:c6:7c:3f:3d:6c
vf 2 MAC 56:32:43:98:d7:71
vf 3 MAC d6:be:c3:b5:85:ff
vf 4 MAC ee:a9:9a:1e:19:14
vf 5 MAC 4a:d0:4c:07:52:18
vf 6 MAC 3a:76:44:93:62:f9
vf 7 MAC 82:e9:e7:e3:15:1a
./ip link set p4p1 vf 0 max_tx_rate 600 rate 300
./ip link show p4p1
32: p4p1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode
DEFAULT qlen 1000
link/ether 00:0e:1e:08:b0:f brd ff:ff:ff:ff:ff:ff
vf 0 MAC 3e:a0:ca:bd:ae:5, tx rate 600 (Mbps), max_tx_rate 600Mbps,
min_tx_rate 200Mbps
vf 1 MAC f6:c6:7c:3f:3d:6c
vf 2 MAC 56:32:43:98:d7:71
vf 3 MAC d6:be:c3:b5:85:ff
vf 4 MAC ee:a9:9a:1e:19:14
vf 5 MAC 4a:d0:4c:07:52:18
vf 6 MAC 3a:76:44:93:62:f9
vf 7 MAC 82:e9:e7:e3:15:1a
Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-05-22 21:59:05 +08:00
|
|
|
* int (*ndo_set_vf_rate)(struct net_device *dev, int vf, int min_tx_rate,
|
|
|
|
* int max_tx_rate);
|
2011-10-08 11:05:24 +08:00
|
|
|
* int (*ndo_set_vf_spoofchk)(struct net_device *dev, int vf, bool setting);
|
2015-08-28 14:57:55 +08:00
|
|
|
* int (*ndo_set_vf_trust)(struct net_device *dev, int vf, bool setting);
|
2010-02-10 09:43:46 +08:00
|
|
|
* int (*ndo_get_vf_config)(struct net_device *dev,
|
|
|
|
* int vf, struct ifla_vf_info *ivf);
|
2013-06-13 18:19:10 +08:00
|
|
|
* int (*ndo_set_vf_link_state)(struct net_device *dev, int vf, int link_state);
|
net: Add netlink support for virtual port management (was iovnl)
Add new netdev ops ndo_{set|get}_vf_port to allow setting of
port-profile on a netdev interface. Extends netlink socket RTM_SETLINK/
RTM_GETLINK with two new sub msgs called IFLA_VF_PORTS and IFLA_PORT_SELF
(added to end of IFLA_cmd list). These are both nested atrtibutes
using this layout:
[IFLA_NUM_VF]
[IFLA_VF_PORTS]
[IFLA_VF_PORT]
[IFLA_PORT_*], ...
[IFLA_VF_PORT]
[IFLA_PORT_*], ...
...
[IFLA_PORT_SELF]
[IFLA_PORT_*], ...
These attributes are design to be set and get symmetrically. VF_PORTS
is a list of VF_PORTs, one for each VF, when dealing with an SR-IOV
device. PORT_SELF is for the PF of the SR-IOV device, in case it wants
to also have a port-profile, or for the case where the VF==PF, like in
enic patch 2/2 of this patch set.
A port-profile is used to configure/enable the external switch virtual port
backing the netdev interface, not to configure the host-facing side of the
netdev. A port-profile is an identifier known to the switch. How port-
profiles are installed on the switch or how available port-profiles are
made know to the host is outside the scope of this patch.
There are two types of port-profiles specs in the netlink msg. The first spec
is for 802.1Qbg (pre-)standard, VDP protocol. The second spec is for devices
that run a similar protocol as VDP but in firmware, thus hiding the protocol
details. In either case, the specs have much in common and makes sense to
define the netlink msg as the union of the two specs. For example, both specs
have a notition of associating/deassociating a port-profile. And both specs
require some information from the hypervisor manager, such as client port
instance ID.
The general flow is the port-profile is applied to a host netdev interface
using RTM_SETLINK, the receiver of the RTM_SETLINK msg communicates with the
switch, and the switch virtual port backing the host netdev interface is
configured/enabled based on the settings defined by the port-profile. What
those settings comprise, and how those settings are managed is again
outside the scope of this patch, since this patch only deals with the
first step in the flow.
Signed-off-by: Scott Feldman <scofeldm@cisco.com>
Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-05-18 13:49:55 +08:00
|
|
|
* int (*ndo_set_vf_port)(struct net_device *dev, int vf,
|
|
|
|
* struct nlattr *port[]);
|
2015-03-31 02:35:23 +08:00
|
|
|
*
|
|
|
|
* Enable or disable the VF ability to query its RSS Redirection Table and
|
|
|
|
* Hash Key. This is needed since on some devices VF share this information
|
2016-03-24 02:47:23 +08:00
|
|
|
* with PF and querying it may introduce a theoretical security risk.
|
2015-03-31 02:35:23 +08:00
|
|
|
* int (*ndo_set_vf_rss_query_en)(struct net_device *dev, int vf, bool setting);
|
net: Add netlink support for virtual port management (was iovnl)
Add new netdev ops ndo_{set|get}_vf_port to allow setting of
port-profile on a netdev interface. Extends netlink socket RTM_SETLINK/
RTM_GETLINK with two new sub msgs called IFLA_VF_PORTS and IFLA_PORT_SELF
(added to end of IFLA_cmd list). These are both nested atrtibutes
using this layout:
[IFLA_NUM_VF]
[IFLA_VF_PORTS]
[IFLA_VF_PORT]
[IFLA_PORT_*], ...
[IFLA_VF_PORT]
[IFLA_PORT_*], ...
...
[IFLA_PORT_SELF]
[IFLA_PORT_*], ...
These attributes are design to be set and get symmetrically. VF_PORTS
is a list of VF_PORTs, one for each VF, when dealing with an SR-IOV
device. PORT_SELF is for the PF of the SR-IOV device, in case it wants
to also have a port-profile, or for the case where the VF==PF, like in
enic patch 2/2 of this patch set.
A port-profile is used to configure/enable the external switch virtual port
backing the netdev interface, not to configure the host-facing side of the
netdev. A port-profile is an identifier known to the switch. How port-
profiles are installed on the switch or how available port-profiles are
made know to the host is outside the scope of this patch.
There are two types of port-profiles specs in the netlink msg. The first spec
is for 802.1Qbg (pre-)standard, VDP protocol. The second spec is for devices
that run a similar protocol as VDP but in firmware, thus hiding the protocol
details. In either case, the specs have much in common and makes sense to
define the netlink msg as the union of the two specs. For example, both specs
have a notition of associating/deassociating a port-profile. And both specs
require some information from the hypervisor manager, such as client port
instance ID.
The general flow is the port-profile is applied to a host netdev interface
using RTM_SETLINK, the receiver of the RTM_SETLINK msg communicates with the
switch, and the switch virtual port backing the host netdev interface is
configured/enabled based on the settings defined by the port-profile. What
those settings comprise, and how those settings are managed is again
outside the scope of this patch, since this patch only deals with the
first step in the flow.
Signed-off-by: Scott Feldman <scofeldm@cisco.com>
Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-05-18 13:49:55 +08:00
|
|
|
* int (*ndo_get_vf_port)(struct net_device *dev, int vf, struct sk_buff *skb);
|
2017-08-07 16:15:17 +08:00
|
|
|
* int (*ndo_setup_tc)(struct net_device *dev, enum tc_setup_type type,
|
2017-08-07 16:15:32 +08:00
|
|
|
* void *type_data);
|
2017-01-27 06:44:17 +08:00
|
|
|
* Called to setup any 'tc' scheduler, classifier or action on @dev.
|
|
|
|
* This is always called from the stack with the rtnl lock held and netif
|
|
|
|
* tx queues stopped. This allows the netdevice to perform queue
|
|
|
|
* management safely.
|
2011-01-19 19:03:53 +08:00
|
|
|
*
|
2011-03-09 16:48:03 +08:00
|
|
|
* Fiber Channel over Ethernet (FCoE) offload functions.
|
|
|
|
* int (*ndo_fcoe_enable)(struct net_device *dev);
|
|
|
|
* Called when the FCoE protocol stack wants to start using LLD for FCoE
|
|
|
|
* so the underlying device can perform whatever needed configuration or
|
|
|
|
* initialization to support acceleration of FCoE traffic.
|
|
|
|
*
|
|
|
|
* int (*ndo_fcoe_disable)(struct net_device *dev);
|
|
|
|
* Called when the FCoE protocol stack wants to stop using LLD for FCoE
|
|
|
|
* so the underlying device can perform whatever needed clean-ups to
|
|
|
|
* stop supporting acceleration of FCoE traffic.
|
|
|
|
*
|
|
|
|
* int (*ndo_fcoe_ddp_setup)(struct net_device *dev, u16 xid,
|
|
|
|
* struct scatterlist *sgl, unsigned int sgc);
|
|
|
|
* Called when the FCoE Initiator wants to initialize an I/O that
|
|
|
|
* is a possible candidate for Direct Data Placement (DDP). The LLD can
|
|
|
|
* perform necessary setup and returns 1 to indicate the device is set up
|
|
|
|
* successfully to perform DDP on this I/O, otherwise this returns 0.
|
|
|
|
*
|
|
|
|
* int (*ndo_fcoe_ddp_done)(struct net_device *dev, u16 xid);
|
|
|
|
* Called when the FCoE Initiator/Target is done with the DDPed I/O as
|
|
|
|
* indicated by the FC exchange id 'xid', so the underlying device can
|
|
|
|
* clean up and reuse resources for later DDP requests.
|
|
|
|
*
|
|
|
|
* int (*ndo_fcoe_ddp_target)(struct net_device *dev, u16 xid,
|
|
|
|
* struct scatterlist *sgl, unsigned int sgc);
|
|
|
|
* Called when the FCoE Target wants to initialize an I/O that
|
|
|
|
* is a possible candidate for Direct Data Placement (DDP). The LLD can
|
|
|
|
* perform necessary setup and returns 1 to indicate the device is set up
|
|
|
|
* successfully to perform DDP on this I/O, otherwise this returns 0.
|
|
|
|
*
|
2012-01-05 04:23:39 +08:00
|
|
|
* int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
|
|
|
|
* struct netdev_fcoe_hbainfo *hbainfo);
|
|
|
|
* Called when the FCoE Protocol stack wants information on the underlying
|
|
|
|
* device. This information is utilized by the FCoE protocol stack to
|
|
|
|
* register attributes with Fiber Channel management service as per the
|
|
|
|
* FC-GS Fabric Device Management Information(FDMI) specification.
|
|
|
|
*
|
2011-03-09 16:48:03 +08:00
|
|
|
* int (*ndo_fcoe_get_wwn)(struct net_device *dev, u64 *wwn, int type);
|
|
|
|
* Called when the underlying device wants to override default World Wide
|
|
|
|
* Name (WWN) generation mechanism in FCoE protocol stack to pass its own
|
|
|
|
* World Wide Port Name (WWPN) or World Wide Node Name (WWNN) to the FCoE
|
|
|
|
* protocol stack to use.
|
|
|
|
*
|
2011-01-19 19:03:53 +08:00
|
|
|
* RFS acceleration.
|
|
|
|
* int (*ndo_rx_flow_steer)(struct net_device *dev, const struct sk_buff *skb,
|
|
|
|
* u16 rxq_index, u32 flow_id);
|
|
|
|
* Set hardware filter for RFS. rxq_index is the target queue index;
|
|
|
|
* flow_id is a flow ID to be passed to rps_may_expire_flow() later.
|
|
|
|
* Return the filter ID on success, or a negative error code.
|
2011-02-13 18:15:37 +08:00
|
|
|
*
|
2013-01-04 06:49:02 +08:00
|
|
|
* Slave management functions (for bridge, bonding, etc).
|
2011-02-13 18:15:37 +08:00
|
|
|
* int (*ndo_add_slave)(struct net_device *dev, struct net_device *slave_dev);
|
|
|
|
* Called to make another netdev an underling.
|
|
|
|
*
|
|
|
|
* int (*ndo_del_slave)(struct net_device *dev, struct net_device *slave_dev);
|
|
|
|
* Called to release previously enslaved netdev.
|
2011-02-16 00:59:17 +08:00
|
|
|
*
|
2020-05-01 03:21:31 +08:00
|
|
|
* struct net_device *(*ndo_get_xmit_slave)(struct net_device *dev,
|
|
|
|
* struct sk_buff *skb,
|
|
|
|
* bool all_slaves);
|
|
|
|
* Get the xmit slave of master device. If all_slaves is true, function
|
|
|
|
* assume all the slaves can transmit.
|
|
|
|
*
|
2011-02-16 00:59:17 +08:00
|
|
|
* Feature/offload setting functions.
|
2017-02-01 08:03:13 +08:00
|
|
|
* netdev_features_t (*ndo_fix_features)(struct net_device *dev,
|
|
|
|
* netdev_features_t features);
|
|
|
|
* Adjusts the requested feature flags according to device-specific
|
|
|
|
* constraints, and returns the resulting flags. Must not modify
|
|
|
|
* the device state.
|
|
|
|
*
|
2011-11-15 23:29:55 +08:00
|
|
|
* int (*ndo_set_features)(struct net_device *dev, netdev_features_t features);
|
2011-02-16 00:59:17 +08:00
|
|
|
* Called to update device configuration to new features. Passed
|
|
|
|
* feature set might be less than what was returned by ndo_fix_features()).
|
|
|
|
* Must return >0 or -errno if it changed dev->features itself.
|
|
|
|
*
|
2012-10-01 20:32:33 +08:00
|
|
|
* int (*ndo_fdb_add)(struct ndmsg *ndm, struct nlattr *tb[],
|
|
|
|
* struct net_device *dev,
|
2019-01-17 07:06:50 +08:00
|
|
|
* const unsigned char *addr, u16 vid, u16 flags,
|
|
|
|
* struct netlink_ext_ack *extack);
|
2012-04-15 14:43:56 +08:00
|
|
|
* Adds an FDB entry to dev for addr.
|
2013-02-13 20:00:18 +08:00
|
|
|
* int (*ndo_fdb_del)(struct ndmsg *ndm, struct nlattr *tb[],
|
|
|
|
* struct net_device *dev,
|
2014-11-28 21:34:15 +08:00
|
|
|
* const unsigned char *addr, u16 vid)
|
2012-04-15 14:43:56 +08:00
|
|
|
* Deletes the FDB entry from dev coresponding to addr.
|
2022-04-13 18:51:56 +08:00
|
|
|
* int (*ndo_fdb_del_bulk)(struct ndmsg *ndm, struct nlattr *tb[],
|
|
|
|
* struct net_device *dev,
|
|
|
|
* u16 vid,
|
|
|
|
* struct netlink_ext_ack *extack);
|
2012-04-15 14:43:56 +08:00
|
|
|
* int (*ndo_fdb_dump)(struct sk_buff *skb, struct netlink_callback *cb,
|
2014-07-10 19:01:58 +08:00
|
|
|
* struct net_device *dev, struct net_device *filter_dev,
|
2016-08-31 12:56:45 +08:00
|
|
|
* int *idx)
|
2012-04-15 14:43:56 +08:00
|
|
|
* Used to add FDB entries to dump requests. Implementers should add
|
|
|
|
* entries to skb and update idx with the number of entries.
|
2012-10-24 16:12:57 +08:00
|
|
|
*
|
2015-03-17 18:16:00 +08:00
|
|
|
* int (*ndo_bridge_setlink)(struct net_device *dev, struct nlmsghdr *nlh,
|
2018-12-13 01:02:48 +08:00
|
|
|
* u16 flags, struct netlink_ext_ack *extack)
|
2012-10-24 16:12:57 +08:00
|
|
|
* int (*ndo_bridge_getlink)(struct sk_buff *skb, u32 pid, u32 seq,
|
2015-04-29 00:33:49 +08:00
|
|
|
* struct net_device *dev, u32 filter_mask,
|
|
|
|
* int nlflags)
|
2015-03-17 18:16:00 +08:00
|
|
|
* int (*ndo_bridge_dellink)(struct net_device *dev, struct nlmsghdr *nlh,
|
|
|
|
* u16 flags);
|
2012-12-28 07:49:37 +08:00
|
|
|
*
|
|
|
|
* int (*ndo_change_carrier)(struct net_device *dev, bool new_carrier);
|
|
|
|
* Called to change device carrier. Soft-devices (like dummy, team, etc)
|
|
|
|
* which do not represent real hardware may define this to allow their
|
|
|
|
* userspace components to manage their virtual carrier state. Devices
|
|
|
|
* that determine carrier state from physical hardware properties (eg
|
|
|
|
* network cables) or protocol-dependent mechanisms (eg
|
|
|
|
* USB_CDC_NOTIFY_NETWORK_CONNECTION) should NOT implement this function.
|
2013-07-30 00:16:49 +08:00
|
|
|
*
|
|
|
|
* int (*ndo_get_phys_port_id)(struct net_device *dev,
|
2014-11-28 21:34:16 +08:00
|
|
|
* struct netdev_phys_item_id *ppid);
|
2013-07-30 00:16:49 +08:00
|
|
|
* Called to get ID of physical port of this device. If driver does
|
|
|
|
* not implement this, it is assumed that the hw is not able to have
|
|
|
|
* multiple net devices on single physical port.
|
2013-09-04 17:13:38 +08:00
|
|
|
*
|
2019-02-07 01:45:35 +08:00
|
|
|
* int (*ndo_get_port_parent_id)(struct net_device *dev,
|
|
|
|
* struct netdev_phys_item_id *ppid)
|
|
|
|
* Called to get the parent ID of the physical port of this device.
|
|
|
|
*
|
2013-11-07 01:54:46 +08:00
|
|
|
* void* (*ndo_dfwd_add_station)(struct net_device *pdev,
|
|
|
|
* struct net_device *dev)
|
|
|
|
* Called by upper layer devices to accelerate switching or other
|
|
|
|
* station functionality into hardware. 'pdev is the lowerdev
|
|
|
|
* to use for the offload and 'dev' is the net device that will
|
|
|
|
* back the offload. Returns a pointer to the private structure
|
|
|
|
* the upper layer will maintain.
|
|
|
|
* void (*ndo_dfwd_del_station)(struct net_device *pdev, void *priv)
|
|
|
|
* Called by upper layer device to delete the station created
|
|
|
|
* by 'ndo_dfwd_add_station'. 'pdev' is the net device backing
|
|
|
|
* the station and priv is the structure returned by the add
|
|
|
|
* operation.
|
2015-03-18 20:57:33 +08:00
|
|
|
* int (*ndo_set_tx_maxrate)(struct net_device *dev,
|
|
|
|
* int queue_index, u32 maxrate);
|
|
|
|
* Called when a user wants to set a max-rate limitation of specific
|
|
|
|
* TX queue.
|
2015-04-02 23:07:00 +08:00
|
|
|
* int (*ndo_get_iflink)(const struct net_device *dev);
|
|
|
|
* Called to get the iflink value of this device.
|
2015-10-23 09:17:16 +08:00
|
|
|
* int (*ndo_fill_metadata_dst)(struct net_device *dev, struct sk_buff *skb);
|
|
|
|
* This function is used to get egress tunnel information for given skb.
|
|
|
|
* This is useful for retrieving outer tunnel header parameters while
|
|
|
|
* sampling packet.
|
2016-02-26 17:45:37 +08:00
|
|
|
* void (*ndo_set_rx_headroom)(struct net_device *dev, int needed_headroom);
|
|
|
|
* This function is used to specify the headroom that the skb must
|
|
|
|
* consider when allocation skb during packet reception. Setting
|
|
|
|
* appropriate rx headroom value allows avoiding skb head copy on
|
2016-03-24 02:47:23 +08:00
|
|
|
* forward. Setting a negative value resets the rx headroom to the
|
2016-02-26 17:45:37 +08:00
|
|
|
* default value.
|
2017-11-04 04:56:16 +08:00
|
|
|
* int (*ndo_bpf)(struct net_device *dev, struct netdev_bpf *bpf);
|
2016-07-20 03:16:48 +08:00
|
|
|
* This function is used to set or query state related to XDP on the
|
2017-11-04 04:56:16 +08:00
|
|
|
* netdevice and manage BPF offload. See definition of
|
|
|
|
* enum bpf_netdev_command for details.
|
2018-05-31 16:59:47 +08:00
|
|
|
* int (*ndo_xdp_xmit)(struct net_device *dev, int n, struct xdp_frame **xdp,
|
|
|
|
* u32 flags);
|
xdp: change ndo_xdp_xmit API to support bulking
This patch change the API for ndo_xdp_xmit to support bulking
xdp_frames.
When kernel is compiled with CONFIG_RETPOLINE, XDP sees a huge slowdown.
Most of the slowdown is caused by DMA API indirect function calls, but
also the net_device->ndo_xdp_xmit() call.
Benchmarked patch with CONFIG_RETPOLINE, using xdp_redirect_map with
single flow/core test (CPU E5-1650 v4 @ 3.60GHz), showed
performance improved:
for driver ixgbe: 6,042,682 pps -> 6,853,768 pps = +811,086 pps
for driver i40e : 6,187,169 pps -> 6,724,519 pps = +537,350 pps
With frames avail as a bulk inside the driver ndo_xdp_xmit call,
further optimizations are possible, like bulk DMA-mapping for TX.
Testing without CONFIG_RETPOLINE show the same performance for
physical NIC drivers.
The virtual NIC driver tun sees a huge performance boost, as it can
avoid doing per frame producer locking, but instead amortize the
locking cost over the bulk.
V2: Fix compile errors reported by kbuild test robot <lkp@intel.com>
V4: Isolated ndo, driver changes and callers.
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
2018-05-24 22:46:12 +08:00
|
|
|
* This function is used to submit @n XDP packets for transmit on a
|
|
|
|
* netdevice. Returns number of frames successfully transmitted, frames
|
|
|
|
* that got dropped are freed/returned via xdp_return_frame().
|
|
|
|
* Returns negative number, means general error invoking ndo, meaning
|
|
|
|
* no frames were xmit'ed and core-caller will free all frames.
|
2021-07-31 13:57:33 +08:00
|
|
|
* struct net_device *(*ndo_xdp_get_xmit_slave)(struct net_device *dev,
|
|
|
|
* struct xdp_buff *xdp);
|
|
|
|
* Get the xmit slave of master device based on the xdp_buff.
|
2019-08-14 15:27:16 +08:00
|
|
|
* int (*ndo_xsk_wakeup)(struct net_device *dev, u32 queue_id, u32 flags);
|
|
|
|
* This function is used to wake up the softirq, ksoftirqd or kthread
|
|
|
|
* responsible for sending and/or receiving packets on a specific
|
|
|
|
* queue id bound to an AF_XDP socket. The flags field specifies if
|
|
|
|
* only RX, only Tx, or both should be woken up using the flags
|
|
|
|
* XDP_WAKEUP_RX and XDP_WAKEUP_TX.
|
2020-05-19 21:03:13 +08:00
|
|
|
* int (*ndo_tunnel_ctl)(struct net_device *dev, struct ip_tunnel_parm *p,
|
|
|
|
* int cmd);
|
|
|
|
* Add, change, delete or get information on an IPv4 tunnel.
|
2020-10-11 07:40:02 +08:00
|
|
|
* struct net_device *(*ndo_get_peer_dev)(struct net_device *dev);
|
|
|
|
* If a device is paired with a peer device, return the peer instance.
|
|
|
|
* The caller must be under RCU read context.
|
2021-03-24 09:30:32 +08:00
|
|
|
* int (*ndo_fill_forward_path)(struct net_device_path_ctx *ctx, struct net_device_path *path);
|
|
|
|
* Get the forwarding path to reach the real device from the HW destination address
|
2022-05-07 04:01:40 +08:00
|
|
|
* ktime_t (*ndo_get_tstamp)(struct net_device *dev,
|
|
|
|
* const struct skb_shared_hwtstamps *hwtstamps,
|
|
|
|
* bool cycles);
|
|
|
|
* Get hardware timestamp based on normal/adjustable time or free running
|
|
|
|
* cycle counter. This function is required if physical clock supports a
|
|
|
|
* free running cycle counter.
|
2008-11-20 13:32:24 +08:00
|
|
|
*/
|
|
|
|
struct net_device_ops {
|
|
|
|
int (*ndo_init)(struct net_device *dev);
|
|
|
|
void (*ndo_uninit)(struct net_device *dev);
|
|
|
|
int (*ndo_open)(struct net_device *dev);
|
|
|
|
int (*ndo_stop)(struct net_device *dev);
|
2016-01-06 22:53:50 +08:00
|
|
|
netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
|
|
|
|
struct net_device *dev);
|
|
|
|
netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
|
|
|
|
struct net_device *dev,
|
|
|
|
netdev_features_t features);
|
2008-11-21 12:14:53 +08:00
|
|
|
u16 (*ndo_select_queue)(struct net_device *dev,
|
2014-01-10 16:18:26 +08:00
|
|
|
struct sk_buff *skb,
|
2019-03-20 18:02:06 +08:00
|
|
|
struct net_device *sb_dev);
|
2008-11-20 13:32:24 +08:00
|
|
|
void (*ndo_change_rx_flags)(struct net_device *dev,
|
|
|
|
int flags);
|
|
|
|
void (*ndo_set_rx_mode)(struct net_device *dev);
|
|
|
|
int (*ndo_set_mac_address)(struct net_device *dev,
|
|
|
|
void *addr);
|
|
|
|
int (*ndo_validate_addr)(struct net_device *dev);
|
|
|
|
int (*ndo_do_ioctl)(struct net_device *dev,
|
|
|
|
struct ifreq *ifr, int cmd);
|
2021-07-27 21:45:13 +08:00
|
|
|
int (*ndo_eth_ioctl)(struct net_device *dev,
|
|
|
|
struct ifreq *ifr, int cmd);
|
2021-07-27 21:45:17 +08:00
|
|
|
int (*ndo_siocbond)(struct net_device *dev,
|
|
|
|
struct ifreq *ifr, int cmd);
|
2021-07-27 21:45:14 +08:00
|
|
|
int (*ndo_siocwandev)(struct net_device *dev,
|
|
|
|
struct if_settings *ifs);
|
2021-07-27 21:44:47 +08:00
|
|
|
int (*ndo_siocdevprivate)(struct net_device *dev,
|
|
|
|
struct ifreq *ifr,
|
|
|
|
void __user *data, int cmd);
|
2008-11-20 13:32:24 +08:00
|
|
|
int (*ndo_set_config)(struct net_device *dev,
|
|
|
|
struct ifmap *map);
|
2008-11-21 12:14:53 +08:00
|
|
|
int (*ndo_change_mtu)(struct net_device *dev,
|
|
|
|
int new_mtu);
|
|
|
|
int (*ndo_neigh_setup)(struct net_device *dev,
|
|
|
|
struct neigh_parms *);
|
netdev: pass the stuck queue to the timeout handler
This allows incrementing the correct timeout statistic without any mess.
Down the road, devices can learn to reset just the specific queue.
The patch was generated with the following script:
use strict;
use warnings;
our $^I = '.bak';
my @work = (
["arch/m68k/emu/nfeth.c", "nfeth_tx_timeout"],
["arch/um/drivers/net_kern.c", "uml_net_tx_timeout"],
["arch/um/drivers/vector_kern.c", "vector_net_tx_timeout"],
["arch/xtensa/platforms/iss/network.c", "iss_net_tx_timeout"],
["drivers/char/pcmcia/synclink_cs.c", "hdlcdev_tx_timeout"],
["drivers/infiniband/ulp/ipoib/ipoib_main.c", "ipoib_timeout"],
["drivers/infiniband/ulp/ipoib/ipoib_main.c", "ipoib_timeout"],
["drivers/message/fusion/mptlan.c", "mpt_lan_tx_timeout"],
["drivers/misc/sgi-xp/xpnet.c", "xpnet_dev_tx_timeout"],
["drivers/net/appletalk/cops.c", "cops_timeout"],
["drivers/net/arcnet/arcdevice.h", "arcnet_timeout"],
["drivers/net/arcnet/arcnet.c", "arcnet_timeout"],
["drivers/net/arcnet/com20020.c", "arcnet_timeout"],
["drivers/net/ethernet/3com/3c509.c", "el3_tx_timeout"],
["drivers/net/ethernet/3com/3c515.c", "corkscrew_timeout"],
["drivers/net/ethernet/3com/3c574_cs.c", "el3_tx_timeout"],
["drivers/net/ethernet/3com/3c589_cs.c", "el3_tx_timeout"],
["drivers/net/ethernet/3com/3c59x.c", "vortex_tx_timeout"],
["drivers/net/ethernet/3com/3c59x.c", "vortex_tx_timeout"],
["drivers/net/ethernet/3com/typhoon.c", "typhoon_tx_timeout"],
["drivers/net/ethernet/8390/8390.h", "ei_tx_timeout"],
["drivers/net/ethernet/8390/8390.h", "eip_tx_timeout"],
["drivers/net/ethernet/8390/8390.c", "ei_tx_timeout"],
["drivers/net/ethernet/8390/8390p.c", "eip_tx_timeout"],
["drivers/net/ethernet/8390/ax88796.c", "ax_ei_tx_timeout"],
["drivers/net/ethernet/8390/axnet_cs.c", "axnet_tx_timeout"],
["drivers/net/ethernet/8390/etherh.c", "__ei_tx_timeout"],
["drivers/net/ethernet/8390/hydra.c", "__ei_tx_timeout"],
["drivers/net/ethernet/8390/mac8390.c", "__ei_tx_timeout"],
["drivers/net/ethernet/8390/mcf8390.c", "__ei_tx_timeout"],
["drivers/net/ethernet/8390/lib8390.c", "__ei_tx_timeout"],
["drivers/net/ethernet/8390/ne2k-pci.c", "ei_tx_timeout"],
["drivers/net/ethernet/8390/pcnet_cs.c", "ei_tx_timeout"],
["drivers/net/ethernet/8390/smc-ultra.c", "ei_tx_timeout"],
["drivers/net/ethernet/8390/wd.c", "ei_tx_timeout"],
["drivers/net/ethernet/8390/zorro8390.c", "__ei_tx_timeout"],
["drivers/net/ethernet/adaptec/starfire.c", "tx_timeout"],
["drivers/net/ethernet/agere/et131x.c", "et131x_tx_timeout"],
["drivers/net/ethernet/allwinner/sun4i-emac.c", "emac_timeout"],
["drivers/net/ethernet/alteon/acenic.c", "ace_watchdog"],
["drivers/net/ethernet/amazon/ena/ena_netdev.c", "ena_tx_timeout"],
["drivers/net/ethernet/amd/7990.h", "lance_tx_timeout"],
["drivers/net/ethernet/amd/7990.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/a2065.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/am79c961a.c", "am79c961_timeout"],
["drivers/net/ethernet/amd/amd8111e.c", "amd8111e_tx_timeout"],
["drivers/net/ethernet/amd/ariadne.c", "ariadne_tx_timeout"],
["drivers/net/ethernet/amd/atarilance.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/au1000_eth.c", "au1000_tx_timeout"],
["drivers/net/ethernet/amd/declance.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/lance.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/mvme147.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/ni65.c", "ni65_timeout"],
["drivers/net/ethernet/amd/nmclan_cs.c", "mace_tx_timeout"],
["drivers/net/ethernet/amd/pcnet32.c", "pcnet32_tx_timeout"],
["drivers/net/ethernet/amd/sunlance.c", "lance_tx_timeout"],
["drivers/net/ethernet/amd/xgbe/xgbe-drv.c", "xgbe_tx_timeout"],
["drivers/net/ethernet/apm/xgene-v2/main.c", "xge_timeout"],
["drivers/net/ethernet/apm/xgene/xgene_enet_main.c", "xgene_enet_timeout"],
["drivers/net/ethernet/apple/macmace.c", "mace_tx_timeout"],
["drivers/net/ethernet/atheros/ag71xx.c", "ag71xx_tx_timeout"],
["drivers/net/ethernet/atheros/alx/main.c", "alx_tx_timeout"],
["drivers/net/ethernet/atheros/atl1c/atl1c_main.c", "atl1c_tx_timeout"],
["drivers/net/ethernet/atheros/atl1e/atl1e_main.c", "atl1e_tx_timeout"],
["drivers/net/ethernet/atheros/atlx/atl.c", "atlx_tx_timeout"],
["drivers/net/ethernet/atheros/atlx/atl1.c", "atlx_tx_timeout"],
["drivers/net/ethernet/atheros/atlx/atl2.c", "atl2_tx_timeout"],
["drivers/net/ethernet/broadcom/b44.c", "b44_tx_timeout"],
["drivers/net/ethernet/broadcom/bcmsysport.c", "bcm_sysport_tx_timeout"],
["drivers/net/ethernet/broadcom/bnx2.c", "bnx2_tx_timeout"],
["drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h", "bnx2x_tx_timeout"],
["drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c", "bnx2x_tx_timeout"],
["drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c", "bnx2x_tx_timeout"],
["drivers/net/ethernet/broadcom/bnxt/bnxt.c", "bnxt_tx_timeout"],
["drivers/net/ethernet/broadcom/genet/bcmgenet.c", "bcmgenet_timeout"],
["drivers/net/ethernet/broadcom/sb1250-mac.c", "sbmac_tx_timeout"],
["drivers/net/ethernet/broadcom/tg3.c", "tg3_tx_timeout"],
["drivers/net/ethernet/calxeda/xgmac.c", "xgmac_tx_timeout"],
["drivers/net/ethernet/cavium/liquidio/lio_main.c", "liquidio_tx_timeout"],
["drivers/net/ethernet/cavium/liquidio/lio_vf_main.c", "liquidio_tx_timeout"],
["drivers/net/ethernet/cavium/liquidio/lio_vf_rep.c", "lio_vf_rep_tx_timeout"],
["drivers/net/ethernet/cavium/thunder/nicvf_main.c", "nicvf_tx_timeout"],
["drivers/net/ethernet/cirrus/cs89x0.c", "net_timeout"],
["drivers/net/ethernet/cisco/enic/enic_main.c", "enic_tx_timeout"],
["drivers/net/ethernet/cisco/enic/enic_main.c", "enic_tx_timeout"],
["drivers/net/ethernet/cortina/gemini.c", "gmac_tx_timeout"],
["drivers/net/ethernet/davicom/dm9000.c", "dm9000_timeout"],
["drivers/net/ethernet/dec/tulip/de2104x.c", "de_tx_timeout"],
["drivers/net/ethernet/dec/tulip/tulip_core.c", "tulip_tx_timeout"],
["drivers/net/ethernet/dec/tulip/winbond-840.c", "tx_timeout"],
["drivers/net/ethernet/dlink/dl2k.c", "rio_tx_timeout"],
["drivers/net/ethernet/dlink/sundance.c", "tx_timeout"],
["drivers/net/ethernet/emulex/benet/be_main.c", "be_tx_timeout"],
["drivers/net/ethernet/ethoc.c", "ethoc_tx_timeout"],
["drivers/net/ethernet/faraday/ftgmac100.c", "ftgmac100_tx_timeout"],
["drivers/net/ethernet/fealnx.c", "fealnx_tx_timeout"],
["drivers/net/ethernet/freescale/dpaa/dpaa_eth.c", "dpaa_tx_timeout"],
["drivers/net/ethernet/freescale/fec_main.c", "fec_timeout"],
["drivers/net/ethernet/freescale/fec_mpc52xx.c", "mpc52xx_fec_tx_timeout"],
["drivers/net/ethernet/freescale/fs_enet/fs_enet-main.c", "fs_timeout"],
["drivers/net/ethernet/freescale/gianfar.c", "gfar_timeout"],
["drivers/net/ethernet/freescale/ucc_geth.c", "ucc_geth_timeout"],
["drivers/net/ethernet/fujitsu/fmvj18x_cs.c", "fjn_tx_timeout"],
["drivers/net/ethernet/google/gve/gve_main.c", "gve_tx_timeout"],
["drivers/net/ethernet/hisilicon/hip04_eth.c", "hip04_timeout"],
["drivers/net/ethernet/hisilicon/hix5hd2_gmac.c", "hix5hd2_net_timeout"],
["drivers/net/ethernet/hisilicon/hns/hns_enet.c", "hns_nic_net_timeout"],
["drivers/net/ethernet/hisilicon/hns3/hns3_enet.c", "hns3_nic_net_timeout"],
["drivers/net/ethernet/huawei/hinic/hinic_main.c", "hinic_tx_timeout"],
["drivers/net/ethernet/i825xx/82596.c", "i596_tx_timeout"],
["drivers/net/ethernet/i825xx/ether1.c", "ether1_timeout"],
["drivers/net/ethernet/i825xx/lib82596.c", "i596_tx_timeout"],
["drivers/net/ethernet/i825xx/sun3_82586.c", "sun3_82586_timeout"],
["drivers/net/ethernet/ibm/ehea/ehea_main.c", "ehea_tx_watchdog"],
["drivers/net/ethernet/ibm/emac/core.c", "emac_tx_timeout"],
["drivers/net/ethernet/ibm/emac/core.c", "emac_tx_timeout"],
["drivers/net/ethernet/ibm/ibmvnic.c", "ibmvnic_tx_timeout"],
["drivers/net/ethernet/intel/e100.c", "e100_tx_timeout"],
["drivers/net/ethernet/intel/e1000/e1000_main.c", "e1000_tx_timeout"],
["drivers/net/ethernet/intel/e1000e/netdev.c", "e1000_tx_timeout"],
["drivers/net/ethernet/intel/fm10k/fm10k_netdev.c", "fm10k_tx_timeout"],
["drivers/net/ethernet/intel/i40e/i40e_main.c", "i40e_tx_timeout"],
["drivers/net/ethernet/intel/iavf/iavf_main.c", "iavf_tx_timeout"],
["drivers/net/ethernet/intel/ice/ice_main.c", "ice_tx_timeout"],
["drivers/net/ethernet/intel/ice/ice_main.c", "ice_tx_timeout"],
["drivers/net/ethernet/intel/igb/igb_main.c", "igb_tx_timeout"],
["drivers/net/ethernet/intel/igbvf/netdev.c", "igbvf_tx_timeout"],
["drivers/net/ethernet/intel/ixgb/ixgb_main.c", "ixgb_tx_timeout"],
["drivers/net/ethernet/intel/ixgbe/ixgbe_debugfs.c", "adapter->netdev->netdev_ops->ndo_tx_timeout(adapter->netdev);"],
["drivers/net/ethernet/intel/ixgbe/ixgbe_main.c", "ixgbe_tx_timeout"],
["drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c", "ixgbevf_tx_timeout"],
["drivers/net/ethernet/jme.c", "jme_tx_timeout"],
["drivers/net/ethernet/korina.c", "korina_tx_timeout"],
["drivers/net/ethernet/lantiq_etop.c", "ltq_etop_tx_timeout"],
["drivers/net/ethernet/marvell/mv643xx_eth.c", "mv643xx_eth_tx_timeout"],
["drivers/net/ethernet/marvell/pxa168_eth.c", "pxa168_eth_tx_timeout"],
["drivers/net/ethernet/marvell/skge.c", "skge_tx_timeout"],
["drivers/net/ethernet/marvell/sky2.c", "sky2_tx_timeout"],
["drivers/net/ethernet/marvell/sky2.c", "sky2_tx_timeout"],
["drivers/net/ethernet/mediatek/mtk_eth_soc.c", "mtk_tx_timeout"],
["drivers/net/ethernet/mellanox/mlx4/en_netdev.c", "mlx4_en_tx_timeout"],
["drivers/net/ethernet/mellanox/mlx4/en_netdev.c", "mlx4_en_tx_timeout"],
["drivers/net/ethernet/mellanox/mlx5/core/en_main.c", "mlx5e_tx_timeout"],
["drivers/net/ethernet/micrel/ks8842.c", "ks8842_tx_timeout"],
["drivers/net/ethernet/micrel/ksz884x.c", "netdev_tx_timeout"],
["drivers/net/ethernet/microchip/enc28j60.c", "enc28j60_tx_timeout"],
["drivers/net/ethernet/microchip/encx24j600.c", "encx24j600_tx_timeout"],
["drivers/net/ethernet/natsemi/sonic.h", "sonic_tx_timeout"],
["drivers/net/ethernet/natsemi/sonic.c", "sonic_tx_timeout"],
["drivers/net/ethernet/natsemi/jazzsonic.c", "sonic_tx_timeout"],
["drivers/net/ethernet/natsemi/macsonic.c", "sonic_tx_timeout"],
["drivers/net/ethernet/natsemi/natsemi.c", "ns_tx_timeout"],
["drivers/net/ethernet/natsemi/ns83820.c", "ns83820_tx_timeout"],
["drivers/net/ethernet/natsemi/xtsonic.c", "sonic_tx_timeout"],
["drivers/net/ethernet/neterion/s2io.h", "s2io_tx_watchdog"],
["drivers/net/ethernet/neterion/s2io.c", "s2io_tx_watchdog"],
["drivers/net/ethernet/neterion/vxge/vxge-main.c", "vxge_tx_watchdog"],
["drivers/net/ethernet/netronome/nfp/nfp_net_common.c", "nfp_net_tx_timeout"],
["drivers/net/ethernet/nvidia/forcedeth.c", "nv_tx_timeout"],
["drivers/net/ethernet/nvidia/forcedeth.c", "nv_tx_timeout"],
["drivers/net/ethernet/oki-semi/pch_gbe/pch_gbe_main.c", "pch_gbe_tx_timeout"],
["drivers/net/ethernet/packetengines/hamachi.c", "hamachi_tx_timeout"],
["drivers/net/ethernet/packetengines/yellowfin.c", "yellowfin_tx_timeout"],
["drivers/net/ethernet/pensando/ionic/ionic_lif.c", "ionic_tx_timeout"],
["drivers/net/ethernet/qlogic/netxen/netxen_nic_main.c", "netxen_tx_timeout"],
["drivers/net/ethernet/qlogic/qla3xxx.c", "ql3xxx_tx_timeout"],
["drivers/net/ethernet/qlogic/qlcnic/qlcnic_main.c", "qlcnic_tx_timeout"],
["drivers/net/ethernet/qualcomm/emac/emac.c", "emac_tx_timeout"],
["drivers/net/ethernet/qualcomm/qca_spi.c", "qcaspi_netdev_tx_timeout"],
["drivers/net/ethernet/qualcomm/qca_uart.c", "qcauart_netdev_tx_timeout"],
["drivers/net/ethernet/rdc/r6040.c", "r6040_tx_timeout"],
["drivers/net/ethernet/realtek/8139cp.c", "cp_tx_timeout"],
["drivers/net/ethernet/realtek/8139too.c", "rtl8139_tx_timeout"],
["drivers/net/ethernet/realtek/atp.c", "tx_timeout"],
["drivers/net/ethernet/realtek/r8169_main.c", "rtl8169_tx_timeout"],
["drivers/net/ethernet/renesas/ravb_main.c", "ravb_tx_timeout"],
["drivers/net/ethernet/renesas/sh_eth.c", "sh_eth_tx_timeout"],
["drivers/net/ethernet/renesas/sh_eth.c", "sh_eth_tx_timeout"],
["drivers/net/ethernet/samsung/sxgbe/sxgbe_main.c", "sxgbe_tx_timeout"],
["drivers/net/ethernet/seeq/ether3.c", "ether3_timeout"],
["drivers/net/ethernet/seeq/sgiseeq.c", "timeout"],
["drivers/net/ethernet/sfc/efx.c", "efx_watchdog"],
["drivers/net/ethernet/sfc/falcon/efx.c", "ef4_watchdog"],
["drivers/net/ethernet/sgi/ioc3-eth.c", "ioc3_timeout"],
["drivers/net/ethernet/sgi/meth.c", "meth_tx_timeout"],
["drivers/net/ethernet/silan/sc92031.c", "sc92031_tx_timeout"],
["drivers/net/ethernet/sis/sis190.c", "sis190_tx_timeout"],
["drivers/net/ethernet/sis/sis900.c", "sis900_tx_timeout"],
["drivers/net/ethernet/smsc/epic100.c", "epic_tx_timeout"],
["drivers/net/ethernet/smsc/smc911x.c", "smc911x_timeout"],
["drivers/net/ethernet/smsc/smc9194.c", "smc_timeout"],
["drivers/net/ethernet/smsc/smc91c92_cs.c", "smc_tx_timeout"],
["drivers/net/ethernet/smsc/smc91x.c", "smc_timeout"],
["drivers/net/ethernet/stmicro/stmmac/stmmac_main.c", "stmmac_tx_timeout"],
["drivers/net/ethernet/sun/cassini.c", "cas_tx_timeout"],
["drivers/net/ethernet/sun/ldmvsw.c", "sunvnet_tx_timeout_common"],
["drivers/net/ethernet/sun/niu.c", "niu_tx_timeout"],
["drivers/net/ethernet/sun/sunbmac.c", "bigmac_tx_timeout"],
["drivers/net/ethernet/sun/sungem.c", "gem_tx_timeout"],
["drivers/net/ethernet/sun/sunhme.c", "happy_meal_tx_timeout"],
["drivers/net/ethernet/sun/sunqe.c", "qe_tx_timeout"],
["drivers/net/ethernet/sun/sunvnet.c", "sunvnet_tx_timeout_common"],
["drivers/net/ethernet/sun/sunvnet_common.c", "sunvnet_tx_timeout_common"],
["drivers/net/ethernet/sun/sunvnet_common.h", "sunvnet_tx_timeout_common"],
["drivers/net/ethernet/synopsys/dwc-xlgmac-net.c", "xlgmac_tx_timeout"],
["drivers/net/ethernet/ti/cpmac.c", "cpmac_tx_timeout"],
["drivers/net/ethernet/ti/cpsw.c", "cpsw_ndo_tx_timeout"],
["drivers/net/ethernet/ti/cpsw_priv.c", "cpsw_ndo_tx_timeout"],
["drivers/net/ethernet/ti/cpsw_priv.h", "cpsw_ndo_tx_timeout"],
["drivers/net/ethernet/ti/davinci_emac.c", "emac_dev_tx_timeout"],
["drivers/net/ethernet/ti/netcp_core.c", "netcp_ndo_tx_timeout"],
["drivers/net/ethernet/ti/tlan.c", "tlan_tx_timeout"],
["drivers/net/ethernet/toshiba/ps3_gelic_net.h", "gelic_net_tx_timeout"],
["drivers/net/ethernet/toshiba/ps3_gelic_net.c", "gelic_net_tx_timeout"],
["drivers/net/ethernet/toshiba/ps3_gelic_wireless.c", "gelic_net_tx_timeout"],
["drivers/net/ethernet/toshiba/spider_net.c", "spider_net_tx_timeout"],
["drivers/net/ethernet/toshiba/tc35815.c", "tc35815_tx_timeout"],
["drivers/net/ethernet/via/via-rhine.c", "rhine_tx_timeout"],
["drivers/net/ethernet/wiznet/w5100.c", "w5100_tx_timeout"],
["drivers/net/ethernet/wiznet/w5300.c", "w5300_tx_timeout"],
["drivers/net/ethernet/xilinx/xilinx_emaclite.c", "xemaclite_tx_timeout"],
["drivers/net/ethernet/xircom/xirc2ps_cs.c", "xirc_tx_timeout"],
["drivers/net/fjes/fjes_main.c", "fjes_tx_retry"],
["drivers/net/slip/slip.c", "sl_tx_timeout"],
["include/linux/usb/usbnet.h", "usbnet_tx_timeout"],
["drivers/net/usb/aqc111.c", "usbnet_tx_timeout"],
["drivers/net/usb/asix_devices.c", "usbnet_tx_timeout"],
["drivers/net/usb/asix_devices.c", "usbnet_tx_timeout"],
["drivers/net/usb/asix_devices.c", "usbnet_tx_timeout"],
["drivers/net/usb/ax88172a.c", "usbnet_tx_timeout"],
["drivers/net/usb/ax88179_178a.c", "usbnet_tx_timeout"],
["drivers/net/usb/catc.c", "catc_tx_timeout"],
["drivers/net/usb/cdc_mbim.c", "usbnet_tx_timeout"],
["drivers/net/usb/cdc_ncm.c", "usbnet_tx_timeout"],
["drivers/net/usb/dm9601.c", "usbnet_tx_timeout"],
["drivers/net/usb/hso.c", "hso_net_tx_timeout"],
["drivers/net/usb/int51x1.c", "usbnet_tx_timeout"],
["drivers/net/usb/ipheth.c", "ipheth_tx_timeout"],
["drivers/net/usb/kaweth.c", "kaweth_tx_timeout"],
["drivers/net/usb/lan78xx.c", "lan78xx_tx_timeout"],
["drivers/net/usb/mcs7830.c", "usbnet_tx_timeout"],
["drivers/net/usb/pegasus.c", "pegasus_tx_timeout"],
["drivers/net/usb/qmi_wwan.c", "usbnet_tx_timeout"],
["drivers/net/usb/r8152.c", "rtl8152_tx_timeout"],
["drivers/net/usb/rndis_host.c", "usbnet_tx_timeout"],
["drivers/net/usb/rtl8150.c", "rtl8150_tx_timeout"],
["drivers/net/usb/sierra_net.c", "usbnet_tx_timeout"],
["drivers/net/usb/smsc75xx.c", "usbnet_tx_timeout"],
["drivers/net/usb/smsc95xx.c", "usbnet_tx_timeout"],
["drivers/net/usb/sr9700.c", "usbnet_tx_timeout"],
["drivers/net/usb/sr9800.c", "usbnet_tx_timeout"],
["drivers/net/usb/usbnet.c", "usbnet_tx_timeout"],
["drivers/net/vmxnet3/vmxnet3_drv.c", "vmxnet3_tx_timeout"],
["drivers/net/wan/cosa.c", "cosa_net_timeout"],
["drivers/net/wan/farsync.c", "fst_tx_timeout"],
["drivers/net/wan/fsl_ucc_hdlc.c", "uhdlc_tx_timeout"],
["drivers/net/wan/lmc/lmc_main.c", "lmc_driver_timeout"],
["drivers/net/wan/x25_asy.c", "x25_asy_timeout"],
["drivers/net/wimax/i2400m/netdev.c", "i2400m_tx_timeout"],
["drivers/net/wireless/intel/ipw2x00/ipw2100.c", "ipw2100_tx_timeout"],
["drivers/net/wireless/intersil/hostap/hostap_main.c", "prism2_tx_timeout"],
["drivers/net/wireless/intersil/hostap/hostap_main.c", "prism2_tx_timeout"],
["drivers/net/wireless/intersil/hostap/hostap_main.c", "prism2_tx_timeout"],
["drivers/net/wireless/intersil/orinoco/main.c", "orinoco_tx_timeout"],
["drivers/net/wireless/intersil/orinoco/orinoco_usb.c", "orinoco_tx_timeout"],
["drivers/net/wireless/intersil/orinoco/orinoco.h", "orinoco_tx_timeout"],
["drivers/net/wireless/intersil/prism54/islpci_dev.c", "islpci_eth_tx_timeout"],
["drivers/net/wireless/intersil/prism54/islpci_eth.c", "islpci_eth_tx_timeout"],
["drivers/net/wireless/intersil/prism54/islpci_eth.h", "islpci_eth_tx_timeout"],
["drivers/net/wireless/marvell/mwifiex/main.c", "mwifiex_tx_timeout"],
["drivers/net/wireless/quantenna/qtnfmac/core.c", "qtnf_netdev_tx_timeout"],
["drivers/net/wireless/quantenna/qtnfmac/core.h", "qtnf_netdev_tx_timeout"],
["drivers/net/wireless/rndis_wlan.c", "usbnet_tx_timeout"],
["drivers/net/wireless/wl3501_cs.c", "wl3501_tx_timeout"],
["drivers/net/wireless/zydas/zd1201.c", "zd1201_tx_timeout"],
["drivers/s390/net/qeth_core.h", "qeth_tx_timeout"],
["drivers/s390/net/qeth_core_main.c", "qeth_tx_timeout"],
["drivers/s390/net/qeth_l2_main.c", "qeth_tx_timeout"],
["drivers/s390/net/qeth_l2_main.c", "qeth_tx_timeout"],
["drivers/s390/net/qeth_l3_main.c", "qeth_tx_timeout"],
["drivers/s390/net/qeth_l3_main.c", "qeth_tx_timeout"],
["drivers/staging/ks7010/ks_wlan_net.c", "ks_wlan_tx_timeout"],
["drivers/staging/qlge/qlge_main.c", "qlge_tx_timeout"],
["drivers/staging/rtl8192e/rtl8192e/rtl_core.c", "_rtl92e_tx_timeout"],
["drivers/staging/rtl8192u/r8192U_core.c", "tx_timeout"],
["drivers/staging/unisys/visornic/visornic_main.c", "visornic_xmit_timeout"],
["drivers/staging/wlan-ng/p80211netdev.c", "p80211knetdev_tx_timeout"],
["drivers/tty/n_gsm.c", "gsm_mux_net_tx_timeout"],
["drivers/tty/synclink.c", "hdlcdev_tx_timeout"],
["drivers/tty/synclink_gt.c", "hdlcdev_tx_timeout"],
["drivers/tty/synclinkmp.c", "hdlcdev_tx_timeout"],
["net/atm/lec.c", "lec_tx_timeout"],
["net/bluetooth/bnep/netdev.c", "bnep_net_timeout"]
);
for my $p (@work) {
my @pair = @$p;
my $file = $pair[0];
my $func = $pair[1];
print STDERR $file , ": ", $func,"\n";
our @ARGV = ($file);
while (<ARGV>) {
if (m/($func\s*\(struct\s+net_device\s+\*[A-Za-z_]?[A-Za-z-0-9_]*)(\))/) {
print STDERR "found $1+$2 in $file\n";
}
if (s/($func\s*\(struct\s+net_device\s+\*[A-Za-z_]?[A-Za-z-0-9_]*)(\))/$1, unsigned int txqueue$2/) {
print STDERR "$func found in $file\n";
}
print;
}
}
where the list of files and functions is simply from:
git grep ndo_tx_timeout, with manual addition of headers
in the rare cases where the function is from a header,
then manually changing the few places which actually
call ndo_tx_timeout.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Heiner Kallweit <hkallweit1@gmail.com>
Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Acked-by: Shannon Nelson <snelson@pensando.io>
Reviewed-by: Martin Habets <mhabets@solarflare.com>
changes from v9:
fixup a forward declaration
changes from v9:
more leftovers from v3 change
changes from v8:
fix up a missing direct call to timeout
rebased on net-next
changes from v7:
fixup leftovers from v3 change
changes from v6:
fix typo in rtl driver
changes from v5:
add missing files (allow any net device argument name)
changes from v4:
add a missing driver header
changes from v3:
change queue # to unsigned
Changes from v2:
added headers
Changes from v1:
Fix errors found by kbuild:
generalize the pattern a bit, to pick up
a couple of instances missed by the previous
version.
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-12-10 22:23:51 +08:00
|
|
|
void (*ndo_tx_timeout) (struct net_device *dev,
|
|
|
|
unsigned int txqueue);
|
2008-11-20 13:32:24 +08:00
|
|
|
|
2017-01-07 11:12:52 +08:00
|
|
|
void (*ndo_get_stats64)(struct net_device *dev,
|
|
|
|
struct rtnl_link_stats64 *storage);
|
2016-11-23 05:09:54 +08:00
|
|
|
bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id);
|
2016-09-16 21:05:36 +08:00
|
|
|
int (*ndo_get_offload_stats)(int attr_id,
|
|
|
|
const struct net_device *dev,
|
|
|
|
void *attr_data);
|
2008-11-20 13:32:24 +08:00
|
|
|
struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
|
|
|
|
|
2011-12-09 08:52:37 +08:00
|
|
|
int (*ndo_vlan_rx_add_vid)(struct net_device *dev,
|
2013-04-19 10:04:28 +08:00
|
|
|
__be16 proto, u16 vid);
|
2011-12-09 08:52:37 +08:00
|
|
|
int (*ndo_vlan_rx_kill_vid)(struct net_device *dev,
|
2013-04-19 10:04:28 +08:00
|
|
|
__be16 proto, u16 vid);
|
2008-11-20 13:32:24 +08:00
|
|
|
#ifdef CONFIG_NET_POLL_CONTROLLER
|
|
|
|
void (*ndo_poll_controller)(struct net_device *dev);
|
2010-06-11 00:12:47 +08:00
|
|
|
int (*ndo_netpoll_setup)(struct net_device *dev,
|
2014-03-28 06:36:38 +08:00
|
|
|
struct netpoll_info *info);
|
2010-05-06 15:47:21 +08:00
|
|
|
void (*ndo_netpoll_cleanup)(struct net_device *dev);
|
2008-11-20 13:32:24 +08:00
|
|
|
#endif
|
2010-02-10 09:43:46 +08:00
|
|
|
int (*ndo_set_vf_mac)(struct net_device *dev,
|
|
|
|
int queue, u8 *mac);
|
|
|
|
int (*ndo_set_vf_vlan)(struct net_device *dev,
|
2016-09-22 17:11:15 +08:00
|
|
|
int queue, u16 vlan,
|
|
|
|
u8 qos, __be16 proto);
|
net-next:v4: Add support to configure SR-IOV VF minimum and maximum Tx rate through ip tool.
o min_tx_rate puts lower limit on the VF bandwidth. VF is guaranteed
to have a bandwidth of at least this value.
max_tx_rate puts cap on the VF bandwidth. VF can have a bandwidth
of up to this value.
o A new handler set_vf_rate for attr IFLA_VF_RATE has been introduced
which takes 4 arguments:
netdev, VF number, min_tx_rate, max_tx_rate
o ndo_set_vf_rate replaces ndo_set_vf_tx_rate handler.
o Drivers that currently implement ndo_set_vf_tx_rate should now call
ndo_set_vf_rate instead and reject attempt to set a minimum bandwidth
greater than 0 for IFLA_VF_TX_RATE when IFLA_VF_RATE is not yet
implemented by driver.
o If user enters only one of either min_tx_rate or max_tx_rate, then,
userland should read back the other value from driver and set both
for IFLA_VF_RATE.
Drivers that have not yet implemented IFLA_VF_RATE should always
return min_tx_rate as 0 when read from ip tool.
o If both IFLA_VF_TX_RATE and IFLA_VF_RATE options are specified, then
IFLA_VF_RATE should override.
o Idea is to have consistent display of rate values to user.
o Usage example: -
./ip link set p4p1 vf 0 rate 900
./ip link show p4p1
32: p4p1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode
DEFAULT qlen 1000
link/ether 00:0e:1e:08:b0:f0 brd ff:ff:ff:ff:ff:ff
vf 0 MAC 3e:a0:ca:bd:ae:5a, tx rate 900 (Mbps), max_tx_rate 900Mbps
vf 1 MAC f6:c6:7c:3f:3d:6c
vf 2 MAC 56:32:43:98:d7:71
vf 3 MAC d6:be:c3:b5:85:ff
vf 4 MAC ee:a9:9a:1e:19:14
vf 5 MAC 4a:d0:4c:07:52:18
vf 6 MAC 3a:76:44:93:62:f9
vf 7 MAC 82:e9:e7:e3:15:1a
./ip link set p4p1 vf 0 max_tx_rate 300 min_tx_rate 200
./ip link show p4p1
32: p4p1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode
DEFAULT qlen 1000
link/ether 00:0e:1e:08:b0:f0 brd ff:ff:ff:ff:ff:ff
vf 0 MAC 3e:a0:ca:bd:ae:5a, tx rate 300 (Mbps), max_tx_rate 300Mbps,
min_tx_rate 200Mbps
vf 1 MAC f6:c6:7c:3f:3d:6c
vf 2 MAC 56:32:43:98:d7:71
vf 3 MAC d6:be:c3:b5:85:ff
vf 4 MAC ee:a9:9a:1e:19:14
vf 5 MAC 4a:d0:4c:07:52:18
vf 6 MAC 3a:76:44:93:62:f9
vf 7 MAC 82:e9:e7:e3:15:1a
./ip link set p4p1 vf 0 max_tx_rate 600 rate 300
./ip link show p4p1
32: p4p1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode
DEFAULT qlen 1000
link/ether 00:0e:1e:08:b0:f brd ff:ff:ff:ff:ff:ff
vf 0 MAC 3e:a0:ca:bd:ae:5, tx rate 600 (Mbps), max_tx_rate 600Mbps,
min_tx_rate 200Mbps
vf 1 MAC f6:c6:7c:3f:3d:6c
vf 2 MAC 56:32:43:98:d7:71
vf 3 MAC d6:be:c3:b5:85:ff
vf 4 MAC ee:a9:9a:1e:19:14
vf 5 MAC 4a:d0:4c:07:52:18
vf 6 MAC 3a:76:44:93:62:f9
vf 7 MAC 82:e9:e7:e3:15:1a
Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-05-22 21:59:05 +08:00
|
|
|
int (*ndo_set_vf_rate)(struct net_device *dev,
|
|
|
|
int vf, int min_tx_rate,
|
|
|
|
int max_tx_rate);
|
2011-10-08 11:05:24 +08:00
|
|
|
int (*ndo_set_vf_spoofchk)(struct net_device *dev,
|
|
|
|
int vf, bool setting);
|
2015-08-28 14:57:55 +08:00
|
|
|
int (*ndo_set_vf_trust)(struct net_device *dev,
|
|
|
|
int vf, bool setting);
|
2010-02-10 09:43:46 +08:00
|
|
|
int (*ndo_get_vf_config)(struct net_device *dev,
|
|
|
|
int vf,
|
|
|
|
struct ifla_vf_info *ivf);
|
2013-06-13 18:19:10 +08:00
|
|
|
int (*ndo_set_vf_link_state)(struct net_device *dev,
|
|
|
|
int vf, int link_state);
|
2015-06-15 22:59:07 +08:00
|
|
|
int (*ndo_get_vf_stats)(struct net_device *dev,
|
|
|
|
int vf,
|
|
|
|
struct ifla_vf_stats
|
|
|
|
*vf_stats);
|
net: Add netlink support for virtual port management (was iovnl)
Add new netdev ops ndo_{set|get}_vf_port to allow setting of
port-profile on a netdev interface. Extends netlink socket RTM_SETLINK/
RTM_GETLINK with two new sub msgs called IFLA_VF_PORTS and IFLA_PORT_SELF
(added to end of IFLA_cmd list). These are both nested atrtibutes
using this layout:
[IFLA_NUM_VF]
[IFLA_VF_PORTS]
[IFLA_VF_PORT]
[IFLA_PORT_*], ...
[IFLA_VF_PORT]
[IFLA_PORT_*], ...
...
[IFLA_PORT_SELF]
[IFLA_PORT_*], ...
These attributes are design to be set and get symmetrically. VF_PORTS
is a list of VF_PORTs, one for each VF, when dealing with an SR-IOV
device. PORT_SELF is for the PF of the SR-IOV device, in case it wants
to also have a port-profile, or for the case where the VF==PF, like in
enic patch 2/2 of this patch set.
A port-profile is used to configure/enable the external switch virtual port
backing the netdev interface, not to configure the host-facing side of the
netdev. A port-profile is an identifier known to the switch. How port-
profiles are installed on the switch or how available port-profiles are
made know to the host is outside the scope of this patch.
There are two types of port-profiles specs in the netlink msg. The first spec
is for 802.1Qbg (pre-)standard, VDP protocol. The second spec is for devices
that run a similar protocol as VDP but in firmware, thus hiding the protocol
details. In either case, the specs have much in common and makes sense to
define the netlink msg as the union of the two specs. For example, both specs
have a notition of associating/deassociating a port-profile. And both specs
require some information from the hypervisor manager, such as client port
instance ID.
The general flow is the port-profile is applied to a host netdev interface
using RTM_SETLINK, the receiver of the RTM_SETLINK msg communicates with the
switch, and the switch virtual port backing the host netdev interface is
configured/enabled based on the settings defined by the port-profile. What
those settings comprise, and how those settings are managed is again
outside the scope of this patch, since this patch only deals with the
first step in the flow.
Signed-off-by: Scott Feldman <scofeldm@cisco.com>
Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-05-18 13:49:55 +08:00
|
|
|
int (*ndo_set_vf_port)(struct net_device *dev,
|
|
|
|
int vf,
|
|
|
|
struct nlattr *port[]);
|
|
|
|
int (*ndo_get_vf_port)(struct net_device *dev,
|
|
|
|
int vf, struct sk_buff *skb);
|
2019-11-06 21:30:07 +08:00
|
|
|
int (*ndo_get_vf_guid)(struct net_device *dev,
|
|
|
|
int vf,
|
|
|
|
struct ifla_vf_guid *node_guid,
|
|
|
|
struct ifla_vf_guid *port_guid);
|
2016-03-12 04:58:34 +08:00
|
|
|
int (*ndo_set_vf_guid)(struct net_device *dev,
|
|
|
|
int vf, u64 guid,
|
|
|
|
int guid_type);
|
2015-03-31 02:35:23 +08:00
|
|
|
int (*ndo_set_vf_rss_query_en)(
|
|
|
|
struct net_device *dev,
|
|
|
|
int vf, bool setting);
|
2016-02-17 13:16:43 +08:00
|
|
|
int (*ndo_setup_tc)(struct net_device *dev,
|
2017-08-07 16:15:17 +08:00
|
|
|
enum tc_setup_type type,
|
2017-08-07 16:15:32 +08:00
|
|
|
void *type_data);
|
2011-11-25 22:40:26 +08:00
|
|
|
#if IS_ENABLED(CONFIG_FCOE)
|
2009-08-31 20:31:36 +08:00
|
|
|
int (*ndo_fcoe_enable)(struct net_device *dev);
|
|
|
|
int (*ndo_fcoe_disable)(struct net_device *dev);
|
2009-02-28 06:06:59 +08:00
|
|
|
int (*ndo_fcoe_ddp_setup)(struct net_device *dev,
|
|
|
|
u16 xid,
|
|
|
|
struct scatterlist *sgl,
|
|
|
|
unsigned int sgc);
|
|
|
|
int (*ndo_fcoe_ddp_done)(struct net_device *dev,
|
|
|
|
u16 xid);
|
2011-02-01 15:22:06 +08:00
|
|
|
int (*ndo_fcoe_ddp_target)(struct net_device *dev,
|
|
|
|
u16 xid,
|
|
|
|
struct scatterlist *sgl,
|
|
|
|
unsigned int sgc);
|
2012-01-05 04:23:39 +08:00
|
|
|
int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
|
|
|
|
struct netdev_fcoe_hbainfo *hbainfo);
|
2011-08-26 17:45:41 +08:00
|
|
|
#endif
|
|
|
|
|
2011-11-25 22:40:26 +08:00
|
|
|
#if IS_ENABLED(CONFIG_LIBFCOE)
|
2009-10-29 02:24:35 +08:00
|
|
|
#define NETDEV_FCOE_WWNN 0
|
|
|
|
#define NETDEV_FCOE_WWPN 1
|
|
|
|
int (*ndo_fcoe_get_wwn)(struct net_device *dev,
|
|
|
|
u64 *wwn, int type);
|
2009-02-28 06:06:59 +08:00
|
|
|
#endif
|
2011-08-26 17:45:41 +08:00
|
|
|
|
2011-01-19 19:03:53 +08:00
|
|
|
#ifdef CONFIG_RFS_ACCEL
|
|
|
|
int (*ndo_rx_flow_steer)(struct net_device *dev,
|
|
|
|
const struct sk_buff *skb,
|
|
|
|
u16 rxq_index,
|
|
|
|
u32 flow_id);
|
|
|
|
#endif
|
2011-02-13 18:15:37 +08:00
|
|
|
int (*ndo_add_slave)(struct net_device *dev,
|
2017-10-05 08:48:46 +08:00
|
|
|
struct net_device *slave_dev,
|
|
|
|
struct netlink_ext_ack *extack);
|
2011-02-13 18:15:37 +08:00
|
|
|
int (*ndo_del_slave)(struct net_device *dev,
|
|
|
|
struct net_device *slave_dev);
|
2020-05-01 03:21:31 +08:00
|
|
|
struct net_device* (*ndo_get_xmit_slave)(struct net_device *dev,
|
|
|
|
struct sk_buff *skb,
|
|
|
|
bool all_slaves);
|
2021-01-17 22:59:42 +08:00
|
|
|
struct net_device* (*ndo_sk_get_lower_dev)(struct net_device *dev,
|
|
|
|
struct sock *sk);
|
2011-11-15 23:29:55 +08:00
|
|
|
netdev_features_t (*ndo_fix_features)(struct net_device *dev,
|
|
|
|
netdev_features_t features);
|
2011-02-16 00:59:17 +08:00
|
|
|
int (*ndo_set_features)(struct net_device *dev,
|
2011-11-15 23:29:55 +08:00
|
|
|
netdev_features_t features);
|
2016-07-05 17:27:37 +08:00
|
|
|
int (*ndo_neigh_construct)(struct net_device *dev,
|
|
|
|
struct neighbour *n);
|
|
|
|
void (*ndo_neigh_destroy)(struct net_device *dev,
|
|
|
|
struct neighbour *n);
|
2012-04-15 14:43:56 +08:00
|
|
|
|
|
|
|
int (*ndo_fdb_add)(struct ndmsg *ndm,
|
2012-10-01 20:32:33 +08:00
|
|
|
struct nlattr *tb[],
|
2012-04-15 14:43:56 +08:00
|
|
|
struct net_device *dev,
|
2012-09-17 18:03:26 +08:00
|
|
|
const unsigned char *addr,
|
2014-11-28 21:34:15 +08:00
|
|
|
u16 vid,
|
2019-01-17 07:06:50 +08:00
|
|
|
u16 flags,
|
|
|
|
struct netlink_ext_ack *extack);
|
2012-04-15 14:43:56 +08:00
|
|
|
int (*ndo_fdb_del)(struct ndmsg *ndm,
|
2013-02-13 20:00:18 +08:00
|
|
|
struct nlattr *tb[],
|
2012-04-15 14:43:56 +08:00
|
|
|
struct net_device *dev,
|
2014-11-28 21:34:15 +08:00
|
|
|
const unsigned char *addr,
|
2022-05-05 23:09:57 +08:00
|
|
|
u16 vid, struct netlink_ext_ack *extack);
|
2022-04-13 18:51:56 +08:00
|
|
|
int (*ndo_fdb_del_bulk)(struct ndmsg *ndm,
|
|
|
|
struct nlattr *tb[],
|
|
|
|
struct net_device *dev,
|
|
|
|
u16 vid,
|
|
|
|
struct netlink_ext_ack *extack);
|
2012-04-15 14:43:56 +08:00
|
|
|
int (*ndo_fdb_dump)(struct sk_buff *skb,
|
|
|
|
struct netlink_callback *cb,
|
|
|
|
struct net_device *dev,
|
2014-07-10 19:01:58 +08:00
|
|
|
struct net_device *filter_dev,
|
2016-08-31 12:56:45 +08:00
|
|
|
int *idx);
|
2018-12-16 14:35:08 +08:00
|
|
|
int (*ndo_fdb_get)(struct sk_buff *skb,
|
|
|
|
struct nlattr *tb[],
|
|
|
|
struct net_device *dev,
|
|
|
|
const unsigned char *addr,
|
|
|
|
u16 vid, u32 portid, u32 seq,
|
|
|
|
struct netlink_ext_ack *extack);
|
2012-10-24 16:12:57 +08:00
|
|
|
int (*ndo_bridge_setlink)(struct net_device *dev,
|
2015-01-30 14:40:12 +08:00
|
|
|
struct nlmsghdr *nlh,
|
2018-12-13 01:02:48 +08:00
|
|
|
u16 flags,
|
|
|
|
struct netlink_ext_ack *extack);
|
2012-10-24 16:12:57 +08:00
|
|
|
int (*ndo_bridge_getlink)(struct sk_buff *skb,
|
|
|
|
u32 pid, u32 seq,
|
2013-02-13 20:00:13 +08:00
|
|
|
struct net_device *dev,
|
2015-04-29 00:33:49 +08:00
|
|
|
u32 filter_mask,
|
|
|
|
int nlflags);
|
2013-02-13 20:00:12 +08:00
|
|
|
int (*ndo_bridge_dellink)(struct net_device *dev,
|
2015-01-30 14:40:12 +08:00
|
|
|
struct nlmsghdr *nlh,
|
|
|
|
u16 flags);
|
2012-12-28 07:49:37 +08:00
|
|
|
int (*ndo_change_carrier)(struct net_device *dev,
|
|
|
|
bool new_carrier);
|
2013-07-30 00:16:49 +08:00
|
|
|
int (*ndo_get_phys_port_id)(struct net_device *dev,
|
2014-11-28 21:34:16 +08:00
|
|
|
struct netdev_phys_item_id *ppid);
|
2019-02-07 01:45:35 +08:00
|
|
|
int (*ndo_get_port_parent_id)(struct net_device *dev,
|
|
|
|
struct netdev_phys_item_id *ppid);
|
2015-03-18 10:23:15 +08:00
|
|
|
int (*ndo_get_phys_port_name)(struct net_device *dev,
|
|
|
|
char *name, size_t len);
|
2013-11-07 01:54:46 +08:00
|
|
|
void* (*ndo_dfwd_add_station)(struct net_device *pdev,
|
|
|
|
struct net_device *dev);
|
|
|
|
void (*ndo_dfwd_del_station)(struct net_device *pdev,
|
|
|
|
void *priv);
|
|
|
|
|
2015-03-18 20:57:33 +08:00
|
|
|
int (*ndo_set_tx_maxrate)(struct net_device *dev,
|
|
|
|
int queue_index,
|
|
|
|
u32 maxrate);
|
2015-04-02 23:07:00 +08:00
|
|
|
int (*ndo_get_iflink)(const struct net_device *dev);
|
2015-10-23 09:17:16 +08:00
|
|
|
int (*ndo_fill_metadata_dst)(struct net_device *dev,
|
|
|
|
struct sk_buff *skb);
|
2016-02-26 17:45:37 +08:00
|
|
|
void (*ndo_set_rx_headroom)(struct net_device *dev,
|
|
|
|
int needed_headroom);
|
2017-11-04 04:56:16 +08:00
|
|
|
int (*ndo_bpf)(struct net_device *dev,
|
|
|
|
struct netdev_bpf *bpf);
|
xdp: change ndo_xdp_xmit API to support bulking
This patch change the API for ndo_xdp_xmit to support bulking
xdp_frames.
When kernel is compiled with CONFIG_RETPOLINE, XDP sees a huge slowdown.
Most of the slowdown is caused by DMA API indirect function calls, but
also the net_device->ndo_xdp_xmit() call.
Benchmarked patch with CONFIG_RETPOLINE, using xdp_redirect_map with
single flow/core test (CPU E5-1650 v4 @ 3.60GHz), showed
performance improved:
for driver ixgbe: 6,042,682 pps -> 6,853,768 pps = +811,086 pps
for driver i40e : 6,187,169 pps -> 6,724,519 pps = +537,350 pps
With frames avail as a bulk inside the driver ndo_xdp_xmit call,
further optimizations are possible, like bulk DMA-mapping for TX.
Testing without CONFIG_RETPOLINE show the same performance for
physical NIC drivers.
The virtual NIC driver tun sees a huge performance boost, as it can
avoid doing per frame producer locking, but instead amortize the
locking cost over the bulk.
V2: Fix compile errors reported by kbuild test robot <lkp@intel.com>
V4: Isolated ndo, driver changes and callers.
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
2018-05-24 22:46:12 +08:00
|
|
|
int (*ndo_xdp_xmit)(struct net_device *dev, int n,
|
2018-05-31 16:59:47 +08:00
|
|
|
struct xdp_frame **xdp,
|
|
|
|
u32 flags);
|
2021-07-31 13:57:33 +08:00
|
|
|
struct net_device * (*ndo_xdp_get_xmit_slave)(struct net_device *dev,
|
|
|
|
struct xdp_buff *xdp);
|
2019-08-14 15:27:16 +08:00
|
|
|
int (*ndo_xsk_wakeup)(struct net_device *dev,
|
|
|
|
u32 queue_id, u32 flags);
|
2020-05-19 21:03:13 +08:00
|
|
|
int (*ndo_tunnel_ctl)(struct net_device *dev,
|
|
|
|
struct ip_tunnel_parm *p, int cmd);
|
2020-10-11 07:40:02 +08:00
|
|
|
struct net_device * (*ndo_get_peer_dev)(struct net_device *dev);
|
2021-03-24 09:30:32 +08:00
|
|
|
int (*ndo_fill_forward_path)(struct net_device_path_ctx *ctx,
|
|
|
|
struct net_device_path *path);
|
2022-05-07 04:01:40 +08:00
|
|
|
ktime_t (*ndo_get_tstamp)(struct net_device *dev,
|
|
|
|
const struct skb_shared_hwtstamps *hwtstamps,
|
|
|
|
bool cycles);
|
2008-11-20 13:32:24 +08:00
|
|
|
};
|
|
|
|
|
2014-02-26 09:15:13 +08:00
|
|
|
/**
|
2020-11-16 18:17:59 +08:00
|
|
|
* enum netdev_priv_flags - &struct net_device priv_flags
|
2014-02-26 09:15:13 +08:00
|
|
|
*
|
|
|
|
* These are the &struct net_device, they are only set internally
|
|
|
|
* by drivers and used in the kernel. These flags are invisible to
|
2016-03-24 02:47:23 +08:00
|
|
|
* userspace; this means that the order of these flags can change
|
2014-02-26 09:15:13 +08:00
|
|
|
* during any kernel release.
|
|
|
|
*
|
|
|
|
* You should have a pretty good reason to be extending these flags.
|
|
|
|
*
|
|
|
|
* @IFF_802_1Q_VLAN: 802.1Q VLAN device
|
|
|
|
* @IFF_EBRIDGE: Ethernet bridging device
|
|
|
|
* @IFF_BONDING: bonding master or slave
|
|
|
|
* @IFF_ISATAP: ISATAP interface (RFC4214)
|
|
|
|
* @IFF_WAN_HDLC: WAN HDLC device
|
|
|
|
* @IFF_XMIT_DST_RELEASE: dev_hard_start_xmit() is allowed to
|
|
|
|
* release skb->dst
|
|
|
|
* @IFF_DONT_BRIDGE: disallow bridging this ether dev
|
|
|
|
* @IFF_DISABLE_NETPOLL: disable netpoll at run-time
|
|
|
|
* @IFF_MACVLAN_PORT: device used as macvlan port
|
|
|
|
* @IFF_BRIDGE_PORT: device used as bridge port
|
|
|
|
* @IFF_OVS_DATAPATH: device used as Open vSwitch datapath port
|
|
|
|
* @IFF_TX_SKB_SHARING: The interface supports sharing skbs on transmit
|
|
|
|
* @IFF_UNICAST_FLT: Supports unicast filtering
|
|
|
|
* @IFF_TEAM_PORT: device used as team port
|
|
|
|
* @IFF_SUPP_NOFCS: device supports sending custom FCS
|
|
|
|
* @IFF_LIVE_ADDR_CHANGE: device supports hardware address
|
|
|
|
* change when it's running
|
|
|
|
* @IFF_MACVLAN: Macvlan device
|
2016-03-22 04:58:28 +08:00
|
|
|
* @IFF_XMIT_DST_RELEASE_PERM: IFF_XMIT_DST_RELEASE not taking into account
|
|
|
|
* underlying stacked devices
|
2015-09-30 11:07:10 +08:00
|
|
|
* @IFF_L3MDEV_MASTER: device is an L3 master device
|
2015-08-14 01:01:06 +08:00
|
|
|
* @IFF_NO_QUEUE: device can run without qdisc attached
|
2015-08-27 15:31:20 +08:00
|
|
|
* @IFF_OPENVSWITCH: device is a Open vSwitch master
|
2015-10-05 23:51:24 +08:00
|
|
|
* @IFF_L3MDEV_SLAVE: device is enslaved to an L3 master device
|
2015-12-03 19:12:06 +08:00
|
|
|
* @IFF_TEAM: device is a team device
|
2016-02-09 08:05:03 +08:00
|
|
|
* @IFF_RXFH_CONFIGURED: device has had Rx Flow indirection table configured
|
2016-02-26 17:45:37 +08:00
|
|
|
* @IFF_PHONY_HEADROOM: the headroom value is controlled by an external
|
|
|
|
* entity (i.e. the master device for bridged veth)
|
2016-03-12 01:07:32 +08:00
|
|
|
* @IFF_MACSEC: device is a MACsec device
|
2018-03-09 17:39:24 +08:00
|
|
|
* @IFF_NO_RX_HANDLER: device doesn't support the rx_handler hook
|
2018-05-25 00:55:13 +08:00
|
|
|
* @IFF_FAILOVER: device is a failover master device
|
|
|
|
* @IFF_FAILOVER_SLAVE: device is lower dev of a failover master device
|
ipvlan, l3mdev: fix broken l3s mode wrt local routes
While implementing ipvlan l3 and l3s mode for kubernetes CNI plugin,
I ran into the issue that while l3 mode is working fine, l3s mode
does not have any connectivity to kube-apiserver and hence all pods
end up in Error state as well. The ipvlan master device sits on
top of a bond device and hostns traffic to kube-apiserver (also running
in hostns) is DNATed from 10.152.183.1:443 to 139.178.29.207:37573
where the latter is the address of the bond0. While in l3 mode, a
curl to https://10.152.183.1:443 or to https://139.178.29.207:37573
works fine from hostns, neither of them do in case of l3s. In the
latter only a curl to https://127.0.0.1:37573 appeared to work where
for local addresses of bond0 I saw kernel suddenly starting to emit
ARP requests to query HW address of bond0 which remained unanswered
and neighbor entries in INCOMPLETE state. These ARP requests only
happen while in l3s.
Debugging this further, I found the issue is that l3s mode is piggy-
backing on l3 master device, and in this case local routes are using
l3mdev_master_dev_rcu(dev) instead of net->loopback_dev as per commit
f5a0aab84b74 ("net: ipv4: dst for local input routes should use l3mdev
if relevant") and 5f02ce24c269 ("net: l3mdev: Allow the l3mdev to be
a loopback"). I found that reverting them back into using the
net->loopback_dev fixed ipvlan l3s connectivity and got everything
working for the CNI.
Now judging from 4fbae7d83c98 ("ipvlan: Introduce l3s mode") and the
l3mdev paper in [0] the only sole reason why ipvlan l3s is relying
on l3 master device is to get the l3mdev_ip_rcv() receive hook for
setting the dst entry of the input route without adding its own
ipvlan specific hacks into the receive path, however, any l3 domain
semantics beyond just that are breaking l3s operation. Note that
ipvlan also has the ability to dynamically switch its internal
operation from l3 to l3s for all ports via ipvlan_set_port_mode()
at runtime. In any case, l3 vs l3s soley distinguishes itself by
'de-confusing' netfilter through switching skb->dev to ipvlan slave
device late in NF_INET_LOCAL_IN before handing the skb to L4.
Minimal fix taken here is to add a IFF_L3MDEV_RX_HANDLER flag which,
if set from ipvlan setup, gets us only the wanted l3mdev_l3_rcv() hook
without any additional l3mdev semantics on top. This should also have
minimal impact since dev->priv_flags is already hot in cache. With
this set, l3s mode is working fine and I also get things like
masquerading pod traffic on the ipvlan master properly working.
[0] https://netdevconf.org/1.2/papers/ahern-what-is-l3mdev-paper.pdf
Fixes: f5a0aab84b74 ("net: ipv4: dst for local input routes should use l3mdev if relevant")
Fixes: 5f02ce24c269 ("net: l3mdev: Allow the l3mdev to be a loopback")
Fixes: 4fbae7d83c98 ("ipvlan: Introduce l3s mode")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Mahesh Bandewar <maheshb@google.com>
Cc: David Ahern <dsa@cumulusnetworks.com>
Cc: Florian Westphal <fw@strlen.de>
Cc: Martynas Pumputis <m@lambda.lt>
Acked-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-01-30 19:49:48 +08:00
|
|
|
* @IFF_L3MDEV_RX_HANDLER: only invoke the rx handler of L3 master device
|
net: Add priv_flags for allow tx skb without linear
In some cases, we hope to construct skb directly based on the existing
memory without copying data. In this case, the page will be placed
directly in the skb, and the linear space of skb is empty. But
unfortunately, many the network card does not support this operation.
For example Mellanox Technologies MT27710 Family [ConnectX-4 Lx] will
get the following error message:
mlx5_core 0000:3b:00.1 eth1: Error cqe on cqn 0x817, ci 0x8,
qn 0x1dbb, opcode 0xd, syndrome 0x1, vendor syndrome 0x68
00000000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000030: 00 00 00 00 60 10 68 01 0a 00 1d bb 00 0f 9f d2
WQE DUMP: WQ size 1024 WQ cur size 0, WQE index 0xf, len: 64
00000000: 00 00 0f 0a 00 1d bb 03 00 00 00 08 00 00 00 00
00000010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000020: 00 00 00 2b 00 08 00 00 00 00 00 05 9e e3 08 00
00000030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
mlx5_core 0000:3b:00.1 eth1: ERR CQE on SQ: 0x1dbb
So a priv_flag is added here to indicate whether the network card
supports this feature.
Suggested-by: Alexander Lobakin <alobakin@pm.me>
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Signed-off-by: Alexander Lobakin <alobakin@pm.me>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20210218204908.5455-3-alobakin@pm.me
2021-02-19 04:50:02 +08:00
|
|
|
* @IFF_TX_SKB_NO_LINEAR: device/driver is capable of xmitting frames with
|
|
|
|
* skb_headlen(skb) == 0 (data starts from frag0)
|
2021-11-23 09:24:47 +08:00
|
|
|
* @IFF_CHANGE_PROTO_DOWN: device supports setting carrier via IFLA_PROTO_DOWN
|
2014-02-26 09:15:13 +08:00
|
|
|
*/
|
|
|
|
enum netdev_priv_flags {
|
|
|
|
IFF_802_1Q_VLAN = 1<<0,
|
|
|
|
IFF_EBRIDGE = 1<<1,
|
2015-08-27 15:31:21 +08:00
|
|
|
IFF_BONDING = 1<<2,
|
|
|
|
IFF_ISATAP = 1<<3,
|
|
|
|
IFF_WAN_HDLC = 1<<4,
|
|
|
|
IFF_XMIT_DST_RELEASE = 1<<5,
|
|
|
|
IFF_DONT_BRIDGE = 1<<6,
|
|
|
|
IFF_DISABLE_NETPOLL = 1<<7,
|
|
|
|
IFF_MACVLAN_PORT = 1<<8,
|
|
|
|
IFF_BRIDGE_PORT = 1<<9,
|
|
|
|
IFF_OVS_DATAPATH = 1<<10,
|
|
|
|
IFF_TX_SKB_SHARING = 1<<11,
|
|
|
|
IFF_UNICAST_FLT = 1<<12,
|
|
|
|
IFF_TEAM_PORT = 1<<13,
|
|
|
|
IFF_SUPP_NOFCS = 1<<14,
|
|
|
|
IFF_LIVE_ADDR_CHANGE = 1<<15,
|
|
|
|
IFF_MACVLAN = 1<<16,
|
|
|
|
IFF_XMIT_DST_RELEASE_PERM = 1<<17,
|
2018-03-06 17:56:31 +08:00
|
|
|
IFF_L3MDEV_MASTER = 1<<18,
|
|
|
|
IFF_NO_QUEUE = 1<<19,
|
|
|
|
IFF_OPENVSWITCH = 1<<20,
|
|
|
|
IFF_L3MDEV_SLAVE = 1<<21,
|
|
|
|
IFF_TEAM = 1<<22,
|
|
|
|
IFF_RXFH_CONFIGURED = 1<<23,
|
|
|
|
IFF_PHONY_HEADROOM = 1<<24,
|
|
|
|
IFF_MACSEC = 1<<25,
|
2018-03-09 17:39:24 +08:00
|
|
|
IFF_NO_RX_HANDLER = 1<<26,
|
2018-05-25 00:55:13 +08:00
|
|
|
IFF_FAILOVER = 1<<27,
|
|
|
|
IFF_FAILOVER_SLAVE = 1<<28,
|
ipvlan, l3mdev: fix broken l3s mode wrt local routes
While implementing ipvlan l3 and l3s mode for kubernetes CNI plugin,
I ran into the issue that while l3 mode is working fine, l3s mode
does not have any connectivity to kube-apiserver and hence all pods
end up in Error state as well. The ipvlan master device sits on
top of a bond device and hostns traffic to kube-apiserver (also running
in hostns) is DNATed from 10.152.183.1:443 to 139.178.29.207:37573
where the latter is the address of the bond0. While in l3 mode, a
curl to https://10.152.183.1:443 or to https://139.178.29.207:37573
works fine from hostns, neither of them do in case of l3s. In the
latter only a curl to https://127.0.0.1:37573 appeared to work where
for local addresses of bond0 I saw kernel suddenly starting to emit
ARP requests to query HW address of bond0 which remained unanswered
and neighbor entries in INCOMPLETE state. These ARP requests only
happen while in l3s.
Debugging this further, I found the issue is that l3s mode is piggy-
backing on l3 master device, and in this case local routes are using
l3mdev_master_dev_rcu(dev) instead of net->loopback_dev as per commit
f5a0aab84b74 ("net: ipv4: dst for local input routes should use l3mdev
if relevant") and 5f02ce24c269 ("net: l3mdev: Allow the l3mdev to be
a loopback"). I found that reverting them back into using the
net->loopback_dev fixed ipvlan l3s connectivity and got everything
working for the CNI.
Now judging from 4fbae7d83c98 ("ipvlan: Introduce l3s mode") and the
l3mdev paper in [0] the only sole reason why ipvlan l3s is relying
on l3 master device is to get the l3mdev_ip_rcv() receive hook for
setting the dst entry of the input route without adding its own
ipvlan specific hacks into the receive path, however, any l3 domain
semantics beyond just that are breaking l3s operation. Note that
ipvlan also has the ability to dynamically switch its internal
operation from l3 to l3s for all ports via ipvlan_set_port_mode()
at runtime. In any case, l3 vs l3s soley distinguishes itself by
'de-confusing' netfilter through switching skb->dev to ipvlan slave
device late in NF_INET_LOCAL_IN before handing the skb to L4.
Minimal fix taken here is to add a IFF_L3MDEV_RX_HANDLER flag which,
if set from ipvlan setup, gets us only the wanted l3mdev_l3_rcv() hook
without any additional l3mdev semantics on top. This should also have
minimal impact since dev->priv_flags is already hot in cache. With
this set, l3s mode is working fine and I also get things like
masquerading pod traffic on the ipvlan master properly working.
[0] https://netdevconf.org/1.2/papers/ahern-what-is-l3mdev-paper.pdf
Fixes: f5a0aab84b74 ("net: ipv4: dst for local input routes should use l3mdev if relevant")
Fixes: 5f02ce24c269 ("net: l3mdev: Allow the l3mdev to be a loopback")
Fixes: 4fbae7d83c98 ("ipvlan: Introduce l3s mode")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Mahesh Bandewar <maheshb@google.com>
Cc: David Ahern <dsa@cumulusnetworks.com>
Cc: Florian Westphal <fw@strlen.de>
Cc: Martynas Pumputis <m@lambda.lt>
Acked-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-01-30 19:49:48 +08:00
|
|
|
IFF_L3MDEV_RX_HANDLER = 1<<29,
|
2022-11-08 01:42:42 +08:00
|
|
|
/* was IFF_LIVE_RENAME_OK */
|
2022-06-23 21:32:32 +08:00
|
|
|
IFF_TX_SKB_NO_LINEAR = BIT_ULL(31),
|
2021-11-23 09:24:47 +08:00
|
|
|
IFF_CHANGE_PROTO_DOWN = BIT_ULL(32),
|
2014-02-26 09:15:13 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
#define IFF_802_1Q_VLAN IFF_802_1Q_VLAN
|
|
|
|
#define IFF_EBRIDGE IFF_EBRIDGE
|
|
|
|
#define IFF_BONDING IFF_BONDING
|
|
|
|
#define IFF_ISATAP IFF_ISATAP
|
|
|
|
#define IFF_WAN_HDLC IFF_WAN_HDLC
|
|
|
|
#define IFF_XMIT_DST_RELEASE IFF_XMIT_DST_RELEASE
|
|
|
|
#define IFF_DONT_BRIDGE IFF_DONT_BRIDGE
|
|
|
|
#define IFF_DISABLE_NETPOLL IFF_DISABLE_NETPOLL
|
|
|
|
#define IFF_MACVLAN_PORT IFF_MACVLAN_PORT
|
|
|
|
#define IFF_BRIDGE_PORT IFF_BRIDGE_PORT
|
|
|
|
#define IFF_OVS_DATAPATH IFF_OVS_DATAPATH
|
|
|
|
#define IFF_TX_SKB_SHARING IFF_TX_SKB_SHARING
|
|
|
|
#define IFF_UNICAST_FLT IFF_UNICAST_FLT
|
|
|
|
#define IFF_TEAM_PORT IFF_TEAM_PORT
|
|
|
|
#define IFF_SUPP_NOFCS IFF_SUPP_NOFCS
|
|
|
|
#define IFF_LIVE_ADDR_CHANGE IFF_LIVE_ADDR_CHANGE
|
|
|
|
#define IFF_MACVLAN IFF_MACVLAN
|
2014-10-06 09:38:35 +08:00
|
|
|
#define IFF_XMIT_DST_RELEASE_PERM IFF_XMIT_DST_RELEASE_PERM
|
2015-09-30 11:07:10 +08:00
|
|
|
#define IFF_L3MDEV_MASTER IFF_L3MDEV_MASTER
|
2015-08-14 01:01:06 +08:00
|
|
|
#define IFF_NO_QUEUE IFF_NO_QUEUE
|
2015-08-27 15:31:20 +08:00
|
|
|
#define IFF_OPENVSWITCH IFF_OPENVSWITCH
|
2015-11-04 21:59:06 +08:00
|
|
|
#define IFF_L3MDEV_SLAVE IFF_L3MDEV_SLAVE
|
2015-12-03 19:12:06 +08:00
|
|
|
#define IFF_TEAM IFF_TEAM
|
2016-02-09 08:05:03 +08:00
|
|
|
#define IFF_RXFH_CONFIGURED IFF_RXFH_CONFIGURED
|
2021-02-19 04:49:41 +08:00
|
|
|
#define IFF_PHONY_HEADROOM IFF_PHONY_HEADROOM
|
2016-03-12 01:07:32 +08:00
|
|
|
#define IFF_MACSEC IFF_MACSEC
|
2018-03-09 17:39:24 +08:00
|
|
|
#define IFF_NO_RX_HANDLER IFF_NO_RX_HANDLER
|
2018-05-25 00:55:13 +08:00
|
|
|
#define IFF_FAILOVER IFF_FAILOVER
|
|
|
|
#define IFF_FAILOVER_SLAVE IFF_FAILOVER_SLAVE
|
ipvlan, l3mdev: fix broken l3s mode wrt local routes
While implementing ipvlan l3 and l3s mode for kubernetes CNI plugin,
I ran into the issue that while l3 mode is working fine, l3s mode
does not have any connectivity to kube-apiserver and hence all pods
end up in Error state as well. The ipvlan master device sits on
top of a bond device and hostns traffic to kube-apiserver (also running
in hostns) is DNATed from 10.152.183.1:443 to 139.178.29.207:37573
where the latter is the address of the bond0. While in l3 mode, a
curl to https://10.152.183.1:443 or to https://139.178.29.207:37573
works fine from hostns, neither of them do in case of l3s. In the
latter only a curl to https://127.0.0.1:37573 appeared to work where
for local addresses of bond0 I saw kernel suddenly starting to emit
ARP requests to query HW address of bond0 which remained unanswered
and neighbor entries in INCOMPLETE state. These ARP requests only
happen while in l3s.
Debugging this further, I found the issue is that l3s mode is piggy-
backing on l3 master device, and in this case local routes are using
l3mdev_master_dev_rcu(dev) instead of net->loopback_dev as per commit
f5a0aab84b74 ("net: ipv4: dst for local input routes should use l3mdev
if relevant") and 5f02ce24c269 ("net: l3mdev: Allow the l3mdev to be
a loopback"). I found that reverting them back into using the
net->loopback_dev fixed ipvlan l3s connectivity and got everything
working for the CNI.
Now judging from 4fbae7d83c98 ("ipvlan: Introduce l3s mode") and the
l3mdev paper in [0] the only sole reason why ipvlan l3s is relying
on l3 master device is to get the l3mdev_ip_rcv() receive hook for
setting the dst entry of the input route without adding its own
ipvlan specific hacks into the receive path, however, any l3 domain
semantics beyond just that are breaking l3s operation. Note that
ipvlan also has the ability to dynamically switch its internal
operation from l3 to l3s for all ports via ipvlan_set_port_mode()
at runtime. In any case, l3 vs l3s soley distinguishes itself by
'de-confusing' netfilter through switching skb->dev to ipvlan slave
device late in NF_INET_LOCAL_IN before handing the skb to L4.
Minimal fix taken here is to add a IFF_L3MDEV_RX_HANDLER flag which,
if set from ipvlan setup, gets us only the wanted l3mdev_l3_rcv() hook
without any additional l3mdev semantics on top. This should also have
minimal impact since dev->priv_flags is already hot in cache. With
this set, l3s mode is working fine and I also get things like
masquerading pod traffic on the ipvlan master properly working.
[0] https://netdevconf.org/1.2/papers/ahern-what-is-l3mdev-paper.pdf
Fixes: f5a0aab84b74 ("net: ipv4: dst for local input routes should use l3mdev if relevant")
Fixes: 5f02ce24c269 ("net: l3mdev: Allow the l3mdev to be a loopback")
Fixes: 4fbae7d83c98 ("ipvlan: Introduce l3s mode")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Mahesh Bandewar <maheshb@google.com>
Cc: David Ahern <dsa@cumulusnetworks.com>
Cc: Florian Westphal <fw@strlen.de>
Cc: Martynas Pumputis <m@lambda.lt>
Acked-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-01-30 19:49:48 +08:00
|
|
|
#define IFF_L3MDEV_RX_HANDLER IFF_L3MDEV_RX_HANDLER
|
net: Add priv_flags for allow tx skb without linear
In some cases, we hope to construct skb directly based on the existing
memory without copying data. In this case, the page will be placed
directly in the skb, and the linear space of skb is empty. But
unfortunately, many the network card does not support this operation.
For example Mellanox Technologies MT27710 Family [ConnectX-4 Lx] will
get the following error message:
mlx5_core 0000:3b:00.1 eth1: Error cqe on cqn 0x817, ci 0x8,
qn 0x1dbb, opcode 0xd, syndrome 0x1, vendor syndrome 0x68
00000000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000030: 00 00 00 00 60 10 68 01 0a 00 1d bb 00 0f 9f d2
WQE DUMP: WQ size 1024 WQ cur size 0, WQE index 0xf, len: 64
00000000: 00 00 0f 0a 00 1d bb 03 00 00 00 08 00 00 00 00
00000010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00000020: 00 00 00 2b 00 08 00 00 00 00 00 05 9e e3 08 00
00000030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
mlx5_core 0000:3b:00.1 eth1: ERR CQE on SQ: 0x1dbb
So a priv_flag is added here to indicate whether the network card
supports this feature.
Suggested-by: Alexander Lobakin <alobakin@pm.me>
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Signed-off-by: Alexander Lobakin <alobakin@pm.me>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20210218204908.5455-3-alobakin@pm.me
2021-02-19 04:50:02 +08:00
|
|
|
#define IFF_TX_SKB_NO_LINEAR IFF_TX_SKB_NO_LINEAR
|
2014-02-26 09:15:13 +08:00
|
|
|
|
2021-02-23 15:01:26 +08:00
|
|
|
/* Specifies the type of the struct net_device::ml_priv pointer */
|
|
|
|
enum netdev_ml_priv_type {
|
|
|
|
ML_PRIV_NONE,
|
|
|
|
ML_PRIV_CAN,
|
|
|
|
};
|
|
|
|
|
2014-07-31 02:27:36 +08:00
|
|
|
/**
|
|
|
|
* struct net_device - The DEVICE structure.
|
2017-05-12 20:35:46 +08:00
|
|
|
*
|
|
|
|
* Actually, this whole structure is a big mistake. It mixes I/O
|
|
|
|
* data with strictly "high-level" data, and it has to know about
|
|
|
|
* almost every data structure used in the INET module.
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
|
|
|
* @name: This is the first field of the "visible" part of this structure
|
|
|
|
* (i.e. as seen by users in the "Space.c" file). It is the name
|
2017-05-12 20:35:46 +08:00
|
|
|
* of the interface.
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
2019-09-30 17:48:15 +08:00
|
|
|
* @name_node: Name hashlist node
|
2014-07-31 02:27:36 +08:00
|
|
|
* @ifalias: SNMP alias
|
|
|
|
* @mem_end: Shared memory end
|
|
|
|
* @mem_start: Shared memory start
|
|
|
|
* @base_addr: Device I/O address
|
|
|
|
* @irq: Device IRQ number
|
|
|
|
*
|
|
|
|
* @state: Generic network queuing layer state, see netdev_state_t
|
|
|
|
* @dev_list: The global list of network devices
|
2016-03-24 02:47:23 +08:00
|
|
|
* @napi_list: List entry used for polling NAPI devices
|
|
|
|
* @unreg_list: List entry when we are unregistering the
|
|
|
|
* device; see the function unregister_netdev
|
|
|
|
* @close_list: List entry used when we are closing the device
|
2016-03-22 05:08:28 +08:00
|
|
|
* @ptype_all: Device-specific packet handlers for all protocols
|
|
|
|
* @ptype_specific: Device-specific, protocol-specific packet handlers
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
|
|
|
* @adj_list: Directly linked devices, like slaves for bonding
|
|
|
|
* @features: Currently active device features
|
|
|
|
* @hw_features: User-changeable features
|
|
|
|
*
|
|
|
|
* @wanted_features: User-requested features
|
|
|
|
* @vlan_features: Mask of features inheritable by VLAN devices
|
|
|
|
*
|
|
|
|
* @hw_enc_features: Mask of features inherited by encapsulating devices
|
|
|
|
* This field indicates what encapsulation
|
|
|
|
* offloads the hardware is capable of doing,
|
|
|
|
* and drivers will need to set them appropriately.
|
|
|
|
*
|
|
|
|
* @mpls_features: Mask of features inheritable by MPLS
|
2020-02-13 14:28:20 +08:00
|
|
|
* @gso_partial_features: value(s) from NETIF_F_GSO\*
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
|
|
|
* @ifindex: interface index
|
2016-03-24 02:47:23 +08:00
|
|
|
* @group: The group the device belongs to
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
|
|
|
* @stats: Statistics struct, which was left as a legacy, use
|
|
|
|
* rtnl_link_stats64 instead
|
|
|
|
*
|
2022-03-11 13:14:20 +08:00
|
|
|
* @core_stats: core networking counters,
|
2014-07-31 02:27:36 +08:00
|
|
|
* do not use this in drivers
|
2018-01-23 11:14:25 +08:00
|
|
|
* @carrier_up_count: Number of times the carrier has been up
|
|
|
|
* @carrier_down_count: Number of times the carrier has been down
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
|
|
|
* @wireless_handlers: List of functions to handle Wireless Extensions,
|
|
|
|
* instead of ioctl,
|
|
|
|
* see <net/iw_handler.h> for details.
|
|
|
|
* @wireless_data: Instance data managed by the core of wireless extensions
|
|
|
|
*
|
|
|
|
* @netdev_ops: Includes several pointers to callbacks,
|
|
|
|
* if one wants to override the ndo_*() functions
|
|
|
|
* @ethtool_ops: Management operations
|
2020-02-13 14:28:20 +08:00
|
|
|
* @l3mdev_ops: Layer 3 master device operations
|
2016-06-16 03:20:23 +08:00
|
|
|
* @ndisc_ops: Includes callbacks for different IPv6 neighbour
|
|
|
|
* discovery handling. Necessary for e.g. 6LoWPAN.
|
2020-02-13 14:28:20 +08:00
|
|
|
* @xfrmdev_ops: Transformation offload operations
|
|
|
|
* @tlsdev_ops: Transport Layer Security offload operations
|
2015-03-02 14:11:09 +08:00
|
|
|
* @header_ops: Includes callbacks for creating,parsing,caching,etc
|
2014-07-31 02:27:36 +08:00
|
|
|
* of Layer 2 headers.
|
|
|
|
*
|
|
|
|
* @flags: Interface flags (a la BSD)
|
|
|
|
* @priv_flags: Like 'flags' but invisible to userspace,
|
|
|
|
* see if.h for the definitions
|
|
|
|
* @gflags: Global flags ( kept as legacy )
|
|
|
|
* @padded: How much padding added by alloc_netdev()
|
|
|
|
* @operstate: RFC2863 operstate
|
|
|
|
* @link_mode: Mapping policy to operstate
|
|
|
|
* @if_port: Selectable AUI, TP, ...
|
|
|
|
* @dma: DMA channel
|
|
|
|
* @mtu: Interface MTU value
|
net: centralize net_device min/max MTU checking
While looking into an MTU issue with sfc, I started noticing that almost
every NIC driver with an ndo_change_mtu function implemented almost
exactly the same range checks, and in many cases, that was the only
practical thing their ndo_change_mtu function was doing. Quite a few
drivers have either 68, 64, 60 or 46 as their minimum MTU value checked,
and then various sizes from 1500 to 65535 for their maximum MTU value. We
can remove a whole lot of redundant code here if we simple store min_mtu
and max_mtu in net_device, and check against those in net/core/dev.c's
dev_set_mtu().
In theory, there should be zero functional change with this patch, it just
puts the infrastructure in place. Subsequent patches will attempt to start
using said infrastructure, with theoretically zero change in
functionality.
CC: netdev@vger.kernel.org
Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-10-08 10:04:33 +08:00
|
|
|
* @min_mtu: Interface Minimum MTU value
|
|
|
|
* @max_mtu: Interface Maximum MTU value
|
2014-07-31 02:27:36 +08:00
|
|
|
* @type: Interface hardware type
|
2016-03-10 10:58:32 +08:00
|
|
|
* @hard_header_len: Maximum hardware header length.
|
2017-02-08 04:57:20 +08:00
|
|
|
* @min_header_len: Minimum hardware header length
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
|
|
|
* @needed_headroom: Extra headroom the hardware may need, but not in all
|
|
|
|
* cases can this be guaranteed
|
|
|
|
* @needed_tailroom: Extra tailroom the hardware may need, but not in all
|
|
|
|
* cases can this be guaranteed. Some cases also use
|
|
|
|
* LL_MAX_HEADER instead to allocate the skb
|
|
|
|
*
|
|
|
|
* interface address info:
|
|
|
|
*
|
|
|
|
* @perm_addr: Permanent hw address
|
|
|
|
* @addr_assign_type: Hw address assignment type
|
|
|
|
* @addr_len: Hardware address length
|
2019-10-22 02:47:50 +08:00
|
|
|
* @upper_level: Maximum depth level of upper devices.
|
|
|
|
* @lower_level: Maximum depth level of lower devices.
|
2016-06-16 03:20:16 +08:00
|
|
|
* @neigh_priv_len: Used in neigh_alloc()
|
2014-07-31 02:27:36 +08:00
|
|
|
* @dev_id: Used to differentiate devices that share
|
|
|
|
* the same link layer address
|
|
|
|
* @dev_port: Used to differentiate devices that share
|
|
|
|
* the same function
|
|
|
|
* @addr_list_lock: XXX: need comments on this one
|
2020-02-13 14:28:20 +08:00
|
|
|
* @name_assign_type: network interface name assignment type
|
2016-03-24 02:47:23 +08:00
|
|
|
* @uc_promisc: Counter that indicates promiscuous mode
|
2014-07-31 02:27:36 +08:00
|
|
|
* has been enabled due to the need to listen to
|
|
|
|
* additional unicast addresses in a device that
|
|
|
|
* does not implement ndo_set_rx_mode()
|
2015-04-10 21:52:38 +08:00
|
|
|
* @uc: unicast mac addresses
|
|
|
|
* @mc: multicast mac addresses
|
|
|
|
* @dev_addrs: list of device hw addresses
|
|
|
|
* @queues_kset: Group of all Kobjects in the Tx and RX queues
|
2016-03-24 02:47:23 +08:00
|
|
|
* @promiscuity: Number of times the NIC is told to work in
|
|
|
|
* promiscuous mode; if it becomes 0 the NIC will
|
|
|
|
* exit promiscuous mode
|
2014-07-31 02:27:36 +08:00
|
|
|
* @allmulti: Counter, enables or disables allmulticast mode
|
|
|
|
*
|
|
|
|
* @vlan_info: VLAN info
|
|
|
|
* @dsa_ptr: dsa specific data
|
|
|
|
* @tipc_ptr: TIPC specific data
|
|
|
|
* @atalk_ptr: AppleTalk link
|
|
|
|
* @ip_ptr: IPv4 specific data
|
|
|
|
* @ip6_ptr: IPv6 specific data
|
|
|
|
* @ax25_ptr: AX.25 specific data
|
|
|
|
* @ieee80211_ptr: IEEE 802.11 specific data, assign before registering
|
2020-02-13 14:28:20 +08:00
|
|
|
* @ieee802154_ptr: IEEE 802.15.4 low-rate Wireless Personal Area Network
|
|
|
|
* device struct
|
|
|
|
* @mpls_ptr: mpls_dev struct pointer
|
2021-07-29 10:20:44 +08:00
|
|
|
* @mctp_ptr: MCTP specific data
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
|
|
|
* @dev_addr: Hw address (before bcast,
|
|
|
|
* because most packets are unicast)
|
|
|
|
*
|
|
|
|
* @_rx: Array of RX queues
|
|
|
|
* @num_rx_queues: Number of RX queues
|
|
|
|
* allocated at register_netdev() time
|
|
|
|
* @real_num_rx_queues: Number of RX queues currently active in device
|
2020-02-13 14:28:20 +08:00
|
|
|
* @xdp_prog: XDP sockets filter program pointer
|
|
|
|
* @gro_flush_timeout: timeout for GRO layer in NAPI
|
2020-06-23 15:08:59 +08:00
|
|
|
* @napi_defer_hard_irqs: If not zero, provides a counter that would
|
|
|
|
* allow to avoid NIC hard IRQ, on busy queues.
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
|
|
|
* @rx_handler: handler for received packets
|
|
|
|
* @rx_handler_data: XXX: need comments on this one
|
2017-11-03 18:46:25 +08:00
|
|
|
* @miniq_ingress: ingress/clsact qdisc specific data for
|
|
|
|
* ingress processing
|
2014-07-31 02:27:36 +08:00
|
|
|
* @ingress_queue: XXX: need comments on this one
|
2019-11-20 19:33:59 +08:00
|
|
|
* @nf_hooks_ingress: netfilter hooks executed for ingress packets
|
2014-07-31 02:27:36 +08:00
|
|
|
* @broadcast: hw bcast address
|
|
|
|
*
|
2015-04-10 21:52:38 +08:00
|
|
|
* @rx_cpu_rmap: CPU reverse-mapping for RX completion interrupts,
|
|
|
|
* indexed by RX queue number. Assigned by driver.
|
|
|
|
* This must only be set if the ndo_rx_flow_steer
|
|
|
|
* operation is defined
|
|
|
|
* @index_hlist: Device index hash chain
|
|
|
|
*
|
2014-07-31 02:27:36 +08:00
|
|
|
* @_tx: Array of TX queues
|
|
|
|
* @num_tx_queues: Number of TX queues allocated at alloc_netdev_mq() time
|
|
|
|
* @real_num_tx_queues: Number of TX queues currently active in device
|
|
|
|
* @qdisc: Root qdisc from userspace point of view
|
|
|
|
* @tx_queue_len: Max frames per queue allowed
|
|
|
|
* @tx_global_lock: XXX: need comments on this one
|
2020-02-13 14:28:20 +08:00
|
|
|
* @xdp_bulkq: XDP device bulk queue
|
2021-03-19 02:37:46 +08:00
|
|
|
* @xps_maps: all CPUs/RXQs maps for XPS device
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
|
|
|
* @xps_maps: XXX: need comments on this one
|
2017-11-03 18:46:25 +08:00
|
|
|
* @miniq_egress: clsact qdisc specific data for
|
|
|
|
* egress processing
|
netfilter: Introduce egress hook
Support classifying packets with netfilter on egress to satisfy user
requirements such as:
* outbound security policies for containers (Laura)
* filtering and mangling intra-node Direct Server Return (DSR) traffic
on a load balancer (Laura)
* filtering locally generated traffic coming in through AF_PACKET,
such as local ARP traffic generated for clustering purposes or DHCP
(Laura; the AF_PACKET plumbing is contained in a follow-up commit)
* L2 filtering from ingress and egress for AVB (Audio Video Bridging)
and gPTP with nftables (Pablo)
* in the future: in-kernel NAT64/NAT46 (Pablo)
The egress hook introduced herein complements the ingress hook added by
commit e687ad60af09 ("netfilter: add netfilter ingress hook after
handle_ing() under unique static key"). A patch for nftables to hook up
egress rules from user space has been submitted separately, so users may
immediately take advantage of the feature.
Alternatively or in addition to netfilter, packets can be classified
with traffic control (tc). On ingress, packets are classified first by
tc, then by netfilter. On egress, the order is reversed for symmetry.
Conceptually, tc and netfilter can be thought of as layers, with
netfilter layered above tc.
Traffic control is capable of redirecting packets to another interface
(man 8 tc-mirred). E.g., an ingress packet may be redirected from the
host namespace to a container via a veth connection:
tc ingress (host) -> tc egress (veth host) -> tc ingress (veth container)
In this case, netfilter egress classifying is not performed when leaving
the host namespace! That's because the packet is still on the tc layer.
If tc redirects the packet to a physical interface in the host namespace
such that it leaves the system, the packet is never subjected to
netfilter egress classifying. That is only logical since it hasn't
passed through netfilter ingress classifying either.
Packets can alternatively be redirected at the netfilter layer using
nft fwd. Such a packet *is* subjected to netfilter egress classifying
since it has reached the netfilter layer.
Internally, the skb->nf_skip_egress flag controls whether netfilter is
invoked on egress by __dev_queue_xmit(). Because __dev_queue_xmit() may
be called recursively by tunnel drivers such as vxlan, the flag is
reverted to false after sch_handle_egress(). This ensures that
netfilter is applied both on the overlay and underlying network.
Interaction between tc and netfilter is possible by setting and querying
skb->mark.
If netfilter egress classifying is not enabled on any interface, it is
patched out of the data path by way of a static_key and doesn't make a
performance difference that is discernible from noise:
Before: 1537 1538 1538 1537 1538 1537 Mb/sec
After: 1536 1534 1539 1539 1539 1540 Mb/sec
Before + tc accept: 1418 1418 1418 1419 1419 1418 Mb/sec
After + tc accept: 1419 1424 1418 1419 1422 1420 Mb/sec
Before + tc drop: 1620 1619 1619 1619 1620 1620 Mb/sec
After + tc drop: 1616 1624 1625 1624 1622 1619 Mb/sec
When netfilter egress classifying is enabled on at least one interface,
a minimal performance penalty is incurred for every egress packet, even
if the interface it's transmitted over doesn't have any netfilter egress
rules configured. That is caused by checking dev->nf_hooks_egress
against NULL.
Measurements were performed on a Core i7-3615QM. Commands to reproduce:
ip link add dev foo type dummy
ip link set dev foo up
modprobe pktgen
echo "add_device foo" > /proc/net/pktgen/kpktgend_3
samples/pktgen/pktgen_bench_xmit_mode_queue_xmit.sh -i foo -n 400000000 -m "11:11:11:11:11:11" -d 1.1.1.1
Accept all traffic with tc:
tc qdisc add dev foo clsact
tc filter add dev foo egress bpf da bytecode '1,6 0 0 0,'
Drop all traffic with tc:
tc qdisc add dev foo clsact
tc filter add dev foo egress bpf da bytecode '1,6 0 0 2,'
Apply this patch when measuring packet drops to avoid errors in dmesg:
https://lore.kernel.org/netdev/a73dda33-57f4-95d8-ea51-ed483abd6a7a@iogearbox.net/
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Cc: Laura García Liébana <nevola@gmail.com>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Thomas Graf <tgraf@suug.ch>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2021-10-09 04:06:03 +08:00
|
|
|
* @nf_hooks_egress: netfilter hooks executed for egress packets
|
2020-02-13 14:28:20 +08:00
|
|
|
* @qdisc_hash: qdisc hash table
|
2014-07-31 02:27:36 +08:00
|
|
|
* @watchdog_timeo: Represents the timeout that is used by
|
2016-03-24 02:47:23 +08:00
|
|
|
* the watchdog (see dev_watchdog())
|
2014-07-31 02:27:36 +08:00
|
|
|
* @watchdog_timer: List of timers
|
|
|
|
*
|
2020-09-07 11:31:16 +08:00
|
|
|
* @proto_down_reason: reason a netdev interface is held down
|
2014-07-31 02:27:36 +08:00
|
|
|
* @pcpu_refcnt: Number of references to this device
|
2021-03-23 02:21:45 +08:00
|
|
|
* @dev_refcnt: Number of references to this device
|
2021-12-05 12:21:57 +08:00
|
|
|
* @refcnt_tracker: Tracker directory for tracked references to this device
|
2014-07-31 02:27:36 +08:00
|
|
|
* @todo_list: Delayed register/unregister
|
|
|
|
* @link_watch_list: XXX: need comments on this one
|
|
|
|
*
|
|
|
|
* @reg_state: Register/unregister state machine
|
|
|
|
* @dismantle: Device is going to be freed
|
|
|
|
* @rtnl_link_state: This enum represents the phases of creating
|
|
|
|
* a new link
|
|
|
|
*
|
net: Fix inconsistent teardown and release of private netdev state.
Network devices can allocate reasources and private memory using
netdev_ops->ndo_init(). However, the release of these resources
can occur in one of two different places.
Either netdev_ops->ndo_uninit() or netdev->destructor().
The decision of which operation frees the resources depends upon
whether it is necessary for all netdev refs to be released before it
is safe to perform the freeing.
netdev_ops->ndo_uninit() presumably can occur right after the
NETDEV_UNREGISTER notifier completes and the unicast and multicast
address lists are flushed.
netdev->destructor(), on the other hand, does not run until the
netdev references all go away.
Further complicating the situation is that netdev->destructor()
almost universally does also a free_netdev().
This creates a problem for the logic in register_netdevice().
Because all callers of register_netdevice() manage the freeing
of the netdev, and invoke free_netdev(dev) if register_netdevice()
fails.
If netdev_ops->ndo_init() succeeds, but something else fails inside
of register_netdevice(), it does call ndo_ops->ndo_uninit(). But
it is not able to invoke netdev->destructor().
This is because netdev->destructor() will do a free_netdev() and
then the caller of register_netdevice() will do the same.
However, this means that the resources that would normally be released
by netdev->destructor() will not be.
Over the years drivers have added local hacks to deal with this, by
invoking their destructor parts by hand when register_netdevice()
fails.
Many drivers do not try to deal with this, and instead we have leaks.
Let's close this hole by formalizing the distinction between what
private things need to be freed up by netdev->destructor() and whether
the driver needs unregister_netdevice() to perform the free_netdev().
netdev->priv_destructor() performs all actions to free up the private
resources that used to be freed by netdev->destructor(), except for
free_netdev().
netdev->needs_free_netdev is a boolean that indicates whether
free_netdev() should be done at the end of unregister_netdevice().
Now, register_netdevice() can sanely release all resources after
ndo_ops->ndo_init() succeeds, by invoking both ndo_ops->ndo_uninit()
and netdev->priv_destructor().
And at the end of unregister_netdevice(), we invoke
netdev->priv_destructor() and optionally call free_netdev().
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-05-09 00:52:56 +08:00
|
|
|
* @needs_free_netdev: Should unregister perform free_netdev?
|
|
|
|
* @priv_destructor: Called from unregister
|
2014-07-31 02:27:36 +08:00
|
|
|
* @npinfo: XXX: need comments on this one
|
|
|
|
* @nd_net: Network namespace this network device is inside
|
|
|
|
*
|
|
|
|
* @ml_priv: Mid-layer private
|
2021-02-23 15:01:26 +08:00
|
|
|
* @ml_priv_type: Mid-layer private type
|
2014-07-31 02:27:36 +08:00
|
|
|
* @lstats: Loopback statistics
|
|
|
|
* @tstats: Tunnel statistics
|
|
|
|
* @dstats: Dummy statistics
|
|
|
|
* @vstats: Virtual ethernet statistics
|
|
|
|
*
|
|
|
|
* @garp_port: GARP
|
|
|
|
* @mrp_port: MRP
|
|
|
|
*
|
2022-03-22 13:10:53 +08:00
|
|
|
* @dm_private: Drop monitor private
|
|
|
|
*
|
2014-07-31 02:27:36 +08:00
|
|
|
* @dev: Class/net/name entry
|
|
|
|
* @sysfs_groups: Space for optional device, statistics and wireless
|
|
|
|
* sysfs groups
|
|
|
|
*
|
|
|
|
* @sysfs_rx_queue_group: Space for optional per-rx queue attributes
|
|
|
|
* @rtnl_link_ops: Rtnl_link_ops
|
|
|
|
*
|
|
|
|
* @gso_max_size: Maximum size of generic segmentation offload
|
2022-05-06 10:51:32 +08:00
|
|
|
* @tso_max_size: Device (as in HW) limit on the max TSO request size
|
2014-07-31 02:27:36 +08:00
|
|
|
* @gso_max_segs: Maximum number of segments that can be passed to the
|
|
|
|
* NIC for GSO
|
2022-05-06 10:51:32 +08:00
|
|
|
* @tso_max_segs: Device (as in HW) limit on the max TSO segment count
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
|
|
|
* @dcbnl_ops: Data Center Bridging netlink ops
|
|
|
|
* @num_tc: Number of traffic classes in the net device
|
|
|
|
* @tc_to_txq: XXX: need comments on this one
|
2016-11-22 10:28:36 +08:00
|
|
|
* @prio_tc_map: XXX: need comments on this one
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
|
|
|
* @fcoe_ddp_xid: Max exchange id for FCoE LRO by ddp
|
|
|
|
*
|
|
|
|
* @priomap: XXX: need comments on this one
|
|
|
|
* @phydev: Physical device may attach itself
|
|
|
|
* for hardware timestamping
|
2018-03-29 06:44:16 +08:00
|
|
|
* @sfp_bus: attached &struct sfp_bus structure.
|
2020-05-03 13:22:19 +08:00
|
|
|
*
|
|
|
|
* @qdisc_tx_busylock: lockdep class annotating Qdisc->busylock spinlock
|
2014-07-31 02:27:36 +08:00
|
|
|
*
|
2015-07-15 04:43:19 +08:00
|
|
|
* @proto_down: protocol port state information can be sent to the
|
|
|
|
* switch driver and used to set the phys state of the
|
|
|
|
* switch port.
|
|
|
|
*
|
2018-09-25 03:58:59 +08:00
|
|
|
* @wol_enabled: Wake-on-LAN is enabled
|
|
|
|
*
|
2021-02-09 03:34:09 +08:00
|
|
|
* @threaded: napi threaded mode is enabled
|
|
|
|
*
|
2020-01-25 19:17:08 +08:00
|
|
|
* @net_notifier_list: List of per-net netdev notifier block
|
|
|
|
* that follow this device when it is moved
|
|
|
|
* to another network namespace.
|
|
|
|
*
|
2020-03-25 20:52:31 +08:00
|
|
|
* @macsec_ops: MACsec offloading ops
|
|
|
|
*
|
2020-07-10 08:42:46 +08:00
|
|
|
* @udp_tunnel_nic_info: static structure describing the UDP tunnel
|
|
|
|
* offload capabilities of the device
|
|
|
|
* @udp_tunnel_nic: UDP tunnel offload state
|
2020-09-07 11:32:30 +08:00
|
|
|
* @xdp_state: stores info on attached XDP BPF programs
|
2020-07-10 08:42:46 +08:00
|
|
|
*
|
2021-12-16 23:19:16 +08:00
|
|
|
* @nested_level: Used as a parameter of spin_lock_nested() of
|
2020-10-02 13:49:45 +08:00
|
|
|
* dev->addr_list_lock.
|
|
|
|
* @unlink_list: As netif_addr_lock() can be called recursively,
|
|
|
|
* keep a list of interfaces to be deleted.
|
2022-01-05 18:48:38 +08:00
|
|
|
* @gro_max_size: Maximum size of aggregated packet in generic
|
|
|
|
* receive offload (GRO)
|
2020-10-02 13:49:45 +08:00
|
|
|
*
|
2021-11-19 22:21:53 +08:00
|
|
|
* @dev_addr_shadow: Copy of @dev_addr to catch direct writes.
|
2021-12-05 12:22:14 +08:00
|
|
|
* @linkwatch_dev_tracker: refcount tracker used by linkwatch.
|
2021-12-07 09:30:30 +08:00
|
|
|
* @watchdog_dev_tracker: refcount tracker used by watchdog.
|
2022-02-08 02:41:07 +08:00
|
|
|
* @dev_registered_tracker: tracker for reference held while
|
|
|
|
* registered
|
net: dev: Add hardware stats support
Offloading switch device drivers may be able to collect statistics of the
traffic taking place in the HW datapath that pertains to a certain soft
netdevice, such as VLAN. Add the necessary infrastructure to allow exposing
these statistics to the offloaded netdevice in question. The API was shaped
by the following considerations:
- Collection of HW statistics is not free: there may be a finite number of
counters, and the act of counting may have a performance impact. It is
therefore necessary to allow toggling whether HW counting should be done
for any particular SW netdevice.
- As the drivers are loaded and removed, a particular device may get
offloaded and unoffloaded again. At the same time, the statistics values
need to stay monotonic (modulo the eventual 64-bit wraparound),
increasing only to reflect traffic measured in the device.
To that end, the netdevice keeps around a lazily-allocated copy of struct
rtnl_link_stats64. Device drivers then contribute to the values kept
therein at various points. Even as the driver goes away, the struct stays
around to maintain the statistics values.
- Different HW devices may be able to count different things. The
motivation behind this patch in particular is exposure of HW counters on
Nvidia Spectrum switches, where the only practical approach to counting
traffic on offloaded soft netdevices currently is to use router interface
counters, and count L3 traffic. Correspondingly that is the statistics
suite added in this patch.
Other devices may be able to measure different kinds of traffic, and for
that reason, the APIs are built to allow uniform access to different
statistics suites.
- Because soft netdevices and offloading drivers are only loosely bound, a
netdevice uses a notifier chain to communicate with the drivers. Several
new notifiers, NETDEV_OFFLOAD_XSTATS_*, have been added to carry messages
to the offloading drivers.
- Devices can have various conditions for when a particular counter is
available. As the device is configured and reconfigured, the device
offload may become or cease being suitable for counter binding. A
netdevice can use a notifier type NETDEV_OFFLOAD_XSTATS_REPORT_USED to
ping offloading drivers and determine whether anyone currently implements
a given statistics suite. This information can then be propagated to user
space.
When the driver decides to unoffload a netdevice, it can use a
newly-added function, netdev_offload_xstats_report_delta(), to record
outstanding collected statistics, before destroying the HW counter.
This patch adds a helper, call_netdevice_notifiers_info_robust(), for
dispatching a notifier with the possibility of unwind when one of the
consumers bails. Given the wish to eventually get rid of the global
notifier block altogether, this helper only invokes the per-netns notifier
block.
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 00:31:20 +08:00
|
|
|
* @offload_xstats_l3: L3 HW stats for this netdevice.
|
2021-11-19 22:21:53 +08:00
|
|
|
*
|
2022-11-03 00:02:03 +08:00
|
|
|
* @devlink_port: Pointer to related devlink port structure.
|
|
|
|
* Assigned by a driver before netdev registration using
|
|
|
|
* SET_NETDEV_DEVLINK_PORT macro. This pointer is static
|
|
|
|
* during the time netdevice is registered.
|
|
|
|
*
|
2005-04-17 06:20:36 +08:00
|
|
|
* FIXME: cleanup struct net_device such that network protocol info
|
|
|
|
* moves out.
|
|
|
|
*/
|
|
|
|
|
2009-11-05 01:50:58 +08:00
|
|
|
struct net_device {
|
2005-04-17 06:20:36 +08:00
|
|
|
char name[IFNAMSIZ];
|
2019-09-30 17:48:15 +08:00
|
|
|
struct netdev_name_node *name_node;
|
2017-10-03 05:50:05 +08:00
|
|
|
struct dev_ifalias __rcu *ifalias;
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
* I/O specific fields
|
|
|
|
* FIXME: Merge these and struct ifmap into one
|
|
|
|
*/
|
2014-07-31 02:27:36 +08:00
|
|
|
unsigned long mem_end;
|
|
|
|
unsigned long mem_start;
|
|
|
|
unsigned long base_addr;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
/*
|
2014-07-31 02:27:36 +08:00
|
|
|
* Some hardware also needs these fields (state,dev_list,
|
|
|
|
* napi_list,unreg_list,close_list) but they are not
|
2005-04-17 06:20:36 +08:00
|
|
|
* part of the usual set specified in Space.c.
|
|
|
|
*/
|
|
|
|
|
|
|
|
unsigned long state;
|
|
|
|
|
2007-05-04 06:13:45 +08:00
|
|
|
struct list_head dev_list;
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
struct list_head napi_list;
|
2009-10-27 15:03:04 +08:00
|
|
|
struct list_head unreg_list;
|
2013-10-06 10:26:05 +08:00
|
|
|
struct list_head close_list;
|
2015-01-28 03:35:48 +08:00
|
|
|
struct list_head ptype_all;
|
|
|
|
struct list_head ptype_specific;
|
net: add adj_list to save only neighbours
Currently, we distinguish neighbours (first-level linked devices) from
non-neighbours by the neighbour bool in the netdev_adjacent. This could be
quite time-consuming in case we would like to traverse *only* through
neighbours - cause we'd have to traverse through all devices and check for
this flag, and in a (quite common) scenario where we have lots of vlans on
top of bridge, which is on top of a bond - the bonding would have to go
through all those vlans to get its upper neighbour linked devices.
This situation is really unpleasant, cause there are already a lot of cases
when a device with slaves needs to go through them in hot path.
To fix this, introduce a new upper/lower device lists structure -
adj_list, which contains only the neighbours. It works always in
pair with the all_adj_list structure (renamed from upper/lower_dev_list),
i.e. both of them contain the same links, only that all_adj_list contains
also non-neighbour device links. It's really a small change visible,
currently, only for __netdev_adjacent_dev_insert/remove(), and doesn't
change the main linked logic at all.
Also, add some comments a fix a name collision in
netdev_for_each_upper_dev_rcu() and rework the naming by the following
rules:
netdev_(all_)(upper|lower)_*
If "all_" is present, then we work with the whole list of upper/lower
devices, otherwise - only with direct neighbours. Uninline functions - to
get better stack traces.
CC: "David S. Miller" <davem@davemloft.net>
CC: Eric Dumazet <edumazet@google.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Cong Wang <amwang@redhat.com>
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2013-09-25 15:20:07 +08:00
|
|
|
|
|
|
|
struct {
|
|
|
|
struct list_head upper;
|
|
|
|
struct list_head lower;
|
|
|
|
} adj_list;
|
|
|
|
|
2021-01-27 01:39:39 +08:00
|
|
|
/* Read-mostly cache-line for fast-path access */
|
|
|
|
unsigned int flags;
|
2021-11-23 09:24:47 +08:00
|
|
|
unsigned long long priv_flags;
|
2021-01-27 01:39:39 +08:00
|
|
|
const struct net_device_ops *netdev_ops;
|
|
|
|
int ifindex;
|
|
|
|
unsigned short gflags;
|
|
|
|
unsigned short hard_header_len;
|
|
|
|
|
|
|
|
/* Note : dev->mtu is often read without holding a lock.
|
|
|
|
* Writers usually hold RTNL.
|
|
|
|
* It is recommended to use READ_ONCE() to annotate the reads,
|
|
|
|
* and to use WRITE_ONCE() to annotate the writes.
|
|
|
|
*/
|
|
|
|
unsigned int mtu;
|
|
|
|
unsigned short needed_headroom;
|
|
|
|
unsigned short needed_tailroom;
|
|
|
|
|
2011-11-15 23:29:55 +08:00
|
|
|
netdev_features_t features;
|
|
|
|
netdev_features_t hw_features;
|
|
|
|
netdev_features_t wanted_features;
|
|
|
|
netdev_features_t vlan_features;
|
2012-12-07 22:14:14 +08:00
|
|
|
netdev_features_t hw_enc_features;
|
2013-05-24 05:02:52 +08:00
|
|
|
netdev_features_t mpls_features;
|
2016-04-11 09:45:03 +08:00
|
|
|
netdev_features_t gso_partial_features;
|
2011-01-25 07:32:47 +08:00
|
|
|
|
2021-01-27 01:39:39 +08:00
|
|
|
unsigned int min_mtu;
|
|
|
|
unsigned int max_mtu;
|
|
|
|
unsigned short type;
|
|
|
|
unsigned char min_header_len;
|
|
|
|
unsigned char name_assign_type;
|
|
|
|
|
2015-04-02 23:07:09 +08:00
|
|
|
int group;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2021-01-27 01:39:39 +08:00
|
|
|
struct net_device_stats stats; /* not used by modern drivers */
|
2014-03-27 23:45:56 +08:00
|
|
|
|
2022-03-11 13:14:20 +08:00
|
|
|
struct net_device_core_stats __percpu *core_stats;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2018-01-19 01:59:13 +08:00
|
|
|
/* Stats to monitor link on/off, flapping */
|
|
|
|
atomic_t carrier_up_count;
|
|
|
|
atomic_t carrier_down_count;
|
|
|
|
|
2007-04-27 11:48:23 +08:00
|
|
|
#ifdef CONFIG_WIRELESS_EXT
|
2016-03-24 02:47:23 +08:00
|
|
|
const struct iw_handler_def *wireless_handlers;
|
|
|
|
struct iw_public_data *wireless_data;
|
2007-04-27 11:48:23 +08:00
|
|
|
#endif
|
2006-09-09 02:16:13 +08:00
|
|
|
const struct ethtool_ops *ethtool_ops;
|
2015-09-30 11:07:11 +08:00
|
|
|
#ifdef CONFIG_NET_L3_MASTER_DEV
|
|
|
|
const struct l3mdev_ops *l3mdev_ops;
|
|
|
|
#endif
|
2016-06-16 03:20:23 +08:00
|
|
|
#if IS_ENABLED(CONFIG_IPV6)
|
|
|
|
const struct ndisc_ops *ndisc_ops;
|
|
|
|
#endif
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2017-12-20 07:35:49 +08:00
|
|
|
#ifdef CONFIG_XFRM_OFFLOAD
|
2017-04-14 16:06:10 +08:00
|
|
|
const struct xfrmdev_ops *xfrmdev_ops;
|
|
|
|
#endif
|
|
|
|
|
2018-04-30 15:16:13 +08:00
|
|
|
#if IS_ENABLED(CONFIG_TLS_DEVICE)
|
|
|
|
const struct tlsdev_ops *tlsdev_ops;
|
|
|
|
#endif
|
|
|
|
|
2007-10-09 16:40:57 +08:00
|
|
|
const struct header_ops *header_ops;
|
|
|
|
|
2014-07-31 02:27:36 +08:00
|
|
|
unsigned char operstate;
|
|
|
|
unsigned char link_mode;
|
2006-03-21 09:09:11 +08:00
|
|
|
|
2014-07-31 02:27:36 +08:00
|
|
|
unsigned char if_port;
|
|
|
|
unsigned char dma;
|
2011-05-10 01:42:46 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/* Interface address info. */
|
2014-07-31 02:27:36 +08:00
|
|
|
unsigned char perm_addr[MAX_ADDR_LEN];
|
|
|
|
unsigned char addr_assign_type;
|
|
|
|
unsigned char addr_len;
|
2019-10-22 02:47:50 +08:00
|
|
|
unsigned char upper_level;
|
|
|
|
unsigned char lower_level;
|
net: core: add nested_level variable in net_device
This patch is to add a new variable 'nested_level' into the net_device
structure.
This variable will be used as a parameter of spin_lock_nested() of
dev->addr_list_lock.
netif_addr_lock() can be called recursively so spin_lock_nested() is
used instead of spin_lock() and dev->lower_level is used as a parameter
of spin_lock_nested().
But, dev->lower_level value can be updated while it is being used.
So, lockdep would warn a possible deadlock scenario.
When a stacked interface is deleted, netif_{uc | mc}_sync() is
called recursively.
So, spin_lock_nested() is called recursively too.
At this moment, the dev->lower_level variable is used as a parameter of it.
dev->lower_level value is updated when interfaces are being unlinked/linked
immediately.
Thus, After unlinking, dev->lower_level shouldn't be a parameter of
spin_lock_nested().
A (macvlan)
|
B (vlan)
|
C (bridge)
|
D (macvlan)
|
E (vlan)
|
F (bridge)
A->lower_level : 6
B->lower_level : 5
C->lower_level : 4
D->lower_level : 3
E->lower_level : 2
F->lower_level : 1
When an interface 'A' is removed, it releases resources.
At this moment, netif_addr_lock() would be called.
Then, netdev_upper_dev_unlink() is called recursively.
Then dev->lower_level is updated.
There is no problem.
But, when the bridge module is removed, 'C' and 'F' interfaces
are removed at once.
If 'F' is removed first, a lower_level value is like below.
A->lower_level : 5
B->lower_level : 4
C->lower_level : 3
D->lower_level : 2
E->lower_level : 1
F->lower_level : 1
Then, 'C' is removed. at this moment, netif_addr_lock() is called
recursively.
The ordering is like this.
C(3)->D(2)->E(1)->F(1)
At this moment, the lower_level value of 'E' and 'F' are the same.
So, lockdep warns a possible deadlock scenario.
In order to avoid this problem, a new variable 'nested_level' is added.
This value is the same as dev->lower_level - 1.
But this value is updated in rtnl_unlock().
So, this variable can be used as a parameter of spin_lock_nested() safely
in the rtnl context.
Test commands:
ip link add br0 type bridge vlan_filtering 1
ip link add vlan1 link br0 type vlan id 10
ip link add macvlan2 link vlan1 type macvlan
ip link add br3 type bridge vlan_filtering 1
ip link set macvlan2 master br3
ip link add vlan4 link br3 type vlan id 10
ip link add macvlan5 link vlan4 type macvlan
ip link add br6 type bridge vlan_filtering 1
ip link set macvlan5 master br6
ip link add vlan7 link br6 type vlan id 10
ip link add macvlan8 link vlan7 type macvlan
ip link set br0 up
ip link set vlan1 up
ip link set macvlan2 up
ip link set br3 up
ip link set vlan4 up
ip link set macvlan5 up
ip link set br6 up
ip link set vlan7 up
ip link set macvlan8 up
modprobe -rv bridge
Splat looks like:
[ 36.057436][ T744] WARNING: possible recursive locking detected
[ 36.058848][ T744] 5.9.0-rc6+ #728 Not tainted
[ 36.059959][ T744] --------------------------------------------
[ 36.061391][ T744] ip/744 is trying to acquire lock:
[ 36.062590][ T744] ffff8c4767509280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_set_rx_mode+0x19/0x30
[ 36.064922][ T744]
[ 36.064922][ T744] but task is already holding lock:
[ 36.066626][ T744] ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.068851][ T744]
[ 36.068851][ T744] other info that might help us debug this:
[ 36.070731][ T744] Possible unsafe locking scenario:
[ 36.070731][ T744]
[ 36.072497][ T744] CPU0
[ 36.073238][ T744] ----
[ 36.074007][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.075290][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.076590][ T744]
[ 36.076590][ T744] *** DEADLOCK ***
[ 36.076590][ T744]
[ 36.078515][ T744] May be due to missing lock nesting notation
[ 36.078515][ T744]
[ 36.080491][ T744] 3 locks held by ip/744:
[ 36.081471][ T744] #0: ffffffff98571df0 (rtnl_mutex){+.+.}-{3:3}, at: rtnetlink_rcv_msg+0x236/0x490
[ 36.083614][ T744] #1: ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.085942][ T744] #2: ffff8c476c8da280 (&bridge_netdev_addr_lock_key/4){+...}-{2:2}, at: dev_uc_sync+0x39/0x80
[ 36.088400][ T744]
[ 36.088400][ T744] stack backtrace:
[ 36.089772][ T744] CPU: 6 PID: 744 Comm: ip Not tainted 5.9.0-rc6+ #728
[ 36.091364][ T744] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
[ 36.093630][ T744] Call Trace:
[ 36.094416][ T744] dump_stack+0x77/0x9b
[ 36.095385][ T744] __lock_acquire+0xbc3/0x1f40
[ 36.096522][ T744] lock_acquire+0xb4/0x3b0
[ 36.097540][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.098657][ T744] ? rtmsg_ifinfo+0x1f/0x30
[ 36.099711][ T744] ? __dev_notify_flags+0xa5/0xf0
[ 36.100874][ T744] ? rtnl_is_locked+0x11/0x20
[ 36.101967][ T744] ? __dev_set_promiscuity+0x7b/0x1a0
[ 36.103230][ T744] _raw_spin_lock_bh+0x38/0x70
[ 36.104348][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.105461][ T744] dev_set_rx_mode+0x19/0x30
[ 36.106532][ T744] dev_set_promiscuity+0x36/0x50
[ 36.107692][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.108929][ T744] dev_set_promiscuity+0x1e/0x50
[ 36.110093][ T744] br_port_set_promisc+0x1f/0x40 [bridge]
[ 36.111415][ T744] br_manage_promisc+0x8b/0xe0 [bridge]
[ 36.112728][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.113967][ T744] ? __hw_addr_sync_one+0x23/0x50
[ 36.115135][ T744] __dev_set_rx_mode+0x68/0x90
[ 36.116249][ T744] dev_uc_sync+0x70/0x80
[ 36.117244][ T744] dev_uc_add+0x50/0x60
[ 36.118223][ T744] macvlan_open+0x18e/0x1f0 [macvlan]
[ 36.119470][ T744] __dev_open+0xd6/0x170
[ 36.120470][ T744] __dev_change_flags+0x181/0x1d0
[ 36.121644][ T744] dev_change_flags+0x23/0x60
[ 36.122741][ T744] do_setlink+0x30a/0x11e0
[ 36.123778][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.124929][ T744] ? __nla_validate_parse.part.6+0x45/0x8e0
[ 36.126309][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.127457][ T744] __rtnl_newlink+0x546/0x8e0
[ 36.128560][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.129623][ T744] ? deactivate_slab.isra.85+0x6a1/0x850
[ 36.130946][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.132102][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.133176][ T744] ? is_bpf_text_address+0x5/0xe0
[ 36.134364][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.135445][ T744] ? rcu_read_lock_sched_held+0x32/0x60
[ 36.136771][ T744] ? kmem_cache_alloc_trace+0x2d8/0x380
[ 36.138070][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.139164][ T744] rtnl_newlink+0x47/0x70
[ ... ]
Fixes: 845e0ebb4408 ("net: change addr_list_lock back to static key")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-26 02:13:29 +08:00
|
|
|
|
2013-12-12 17:15:59 +08:00
|
|
|
unsigned short neigh_priv_len;
|
2014-07-31 02:27:36 +08:00
|
|
|
unsigned short dev_id;
|
|
|
|
unsigned short dev_port;
|
2021-01-27 01:39:39 +08:00
|
|
|
unsigned short padded;
|
|
|
|
|
2009-05-23 07:22:17 +08:00
|
|
|
spinlock_t addr_list_lock;
|
2021-01-27 01:39:39 +08:00
|
|
|
int irq;
|
net: core: add nested_level variable in net_device
This patch is to add a new variable 'nested_level' into the net_device
structure.
This variable will be used as a parameter of spin_lock_nested() of
dev->addr_list_lock.
netif_addr_lock() can be called recursively so spin_lock_nested() is
used instead of spin_lock() and dev->lower_level is used as a parameter
of spin_lock_nested().
But, dev->lower_level value can be updated while it is being used.
So, lockdep would warn a possible deadlock scenario.
When a stacked interface is deleted, netif_{uc | mc}_sync() is
called recursively.
So, spin_lock_nested() is called recursively too.
At this moment, the dev->lower_level variable is used as a parameter of it.
dev->lower_level value is updated when interfaces are being unlinked/linked
immediately.
Thus, After unlinking, dev->lower_level shouldn't be a parameter of
spin_lock_nested().
A (macvlan)
|
B (vlan)
|
C (bridge)
|
D (macvlan)
|
E (vlan)
|
F (bridge)
A->lower_level : 6
B->lower_level : 5
C->lower_level : 4
D->lower_level : 3
E->lower_level : 2
F->lower_level : 1
When an interface 'A' is removed, it releases resources.
At this moment, netif_addr_lock() would be called.
Then, netdev_upper_dev_unlink() is called recursively.
Then dev->lower_level is updated.
There is no problem.
But, when the bridge module is removed, 'C' and 'F' interfaces
are removed at once.
If 'F' is removed first, a lower_level value is like below.
A->lower_level : 5
B->lower_level : 4
C->lower_level : 3
D->lower_level : 2
E->lower_level : 1
F->lower_level : 1
Then, 'C' is removed. at this moment, netif_addr_lock() is called
recursively.
The ordering is like this.
C(3)->D(2)->E(1)->F(1)
At this moment, the lower_level value of 'E' and 'F' are the same.
So, lockdep warns a possible deadlock scenario.
In order to avoid this problem, a new variable 'nested_level' is added.
This value is the same as dev->lower_level - 1.
But this value is updated in rtnl_unlock().
So, this variable can be used as a parameter of spin_lock_nested() safely
in the rtnl context.
Test commands:
ip link add br0 type bridge vlan_filtering 1
ip link add vlan1 link br0 type vlan id 10
ip link add macvlan2 link vlan1 type macvlan
ip link add br3 type bridge vlan_filtering 1
ip link set macvlan2 master br3
ip link add vlan4 link br3 type vlan id 10
ip link add macvlan5 link vlan4 type macvlan
ip link add br6 type bridge vlan_filtering 1
ip link set macvlan5 master br6
ip link add vlan7 link br6 type vlan id 10
ip link add macvlan8 link vlan7 type macvlan
ip link set br0 up
ip link set vlan1 up
ip link set macvlan2 up
ip link set br3 up
ip link set vlan4 up
ip link set macvlan5 up
ip link set br6 up
ip link set vlan7 up
ip link set macvlan8 up
modprobe -rv bridge
Splat looks like:
[ 36.057436][ T744] WARNING: possible recursive locking detected
[ 36.058848][ T744] 5.9.0-rc6+ #728 Not tainted
[ 36.059959][ T744] --------------------------------------------
[ 36.061391][ T744] ip/744 is trying to acquire lock:
[ 36.062590][ T744] ffff8c4767509280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_set_rx_mode+0x19/0x30
[ 36.064922][ T744]
[ 36.064922][ T744] but task is already holding lock:
[ 36.066626][ T744] ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.068851][ T744]
[ 36.068851][ T744] other info that might help us debug this:
[ 36.070731][ T744] Possible unsafe locking scenario:
[ 36.070731][ T744]
[ 36.072497][ T744] CPU0
[ 36.073238][ T744] ----
[ 36.074007][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.075290][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.076590][ T744]
[ 36.076590][ T744] *** DEADLOCK ***
[ 36.076590][ T744]
[ 36.078515][ T744] May be due to missing lock nesting notation
[ 36.078515][ T744]
[ 36.080491][ T744] 3 locks held by ip/744:
[ 36.081471][ T744] #0: ffffffff98571df0 (rtnl_mutex){+.+.}-{3:3}, at: rtnetlink_rcv_msg+0x236/0x490
[ 36.083614][ T744] #1: ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.085942][ T744] #2: ffff8c476c8da280 (&bridge_netdev_addr_lock_key/4){+...}-{2:2}, at: dev_uc_sync+0x39/0x80
[ 36.088400][ T744]
[ 36.088400][ T744] stack backtrace:
[ 36.089772][ T744] CPU: 6 PID: 744 Comm: ip Not tainted 5.9.0-rc6+ #728
[ 36.091364][ T744] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
[ 36.093630][ T744] Call Trace:
[ 36.094416][ T744] dump_stack+0x77/0x9b
[ 36.095385][ T744] __lock_acquire+0xbc3/0x1f40
[ 36.096522][ T744] lock_acquire+0xb4/0x3b0
[ 36.097540][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.098657][ T744] ? rtmsg_ifinfo+0x1f/0x30
[ 36.099711][ T744] ? __dev_notify_flags+0xa5/0xf0
[ 36.100874][ T744] ? rtnl_is_locked+0x11/0x20
[ 36.101967][ T744] ? __dev_set_promiscuity+0x7b/0x1a0
[ 36.103230][ T744] _raw_spin_lock_bh+0x38/0x70
[ 36.104348][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.105461][ T744] dev_set_rx_mode+0x19/0x30
[ 36.106532][ T744] dev_set_promiscuity+0x36/0x50
[ 36.107692][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.108929][ T744] dev_set_promiscuity+0x1e/0x50
[ 36.110093][ T744] br_port_set_promisc+0x1f/0x40 [bridge]
[ 36.111415][ T744] br_manage_promisc+0x8b/0xe0 [bridge]
[ 36.112728][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.113967][ T744] ? __hw_addr_sync_one+0x23/0x50
[ 36.115135][ T744] __dev_set_rx_mode+0x68/0x90
[ 36.116249][ T744] dev_uc_sync+0x70/0x80
[ 36.117244][ T744] dev_uc_add+0x50/0x60
[ 36.118223][ T744] macvlan_open+0x18e/0x1f0 [macvlan]
[ 36.119470][ T744] __dev_open+0xd6/0x170
[ 36.120470][ T744] __dev_change_flags+0x181/0x1d0
[ 36.121644][ T744] dev_change_flags+0x23/0x60
[ 36.122741][ T744] do_setlink+0x30a/0x11e0
[ 36.123778][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.124929][ T744] ? __nla_validate_parse.part.6+0x45/0x8e0
[ 36.126309][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.127457][ T744] __rtnl_newlink+0x546/0x8e0
[ 36.128560][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.129623][ T744] ? deactivate_slab.isra.85+0x6a1/0x850
[ 36.130946][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.132102][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.133176][ T744] ? is_bpf_text_address+0x5/0xe0
[ 36.134364][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.135445][ T744] ? rcu_read_lock_sched_held+0x32/0x60
[ 36.136771][ T744] ? kmem_cache_alloc_trace+0x2d8/0x380
[ 36.138070][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.139164][ T744] rtnl_newlink+0x47/0x70
[ ... ]
Fixes: 845e0ebb4408 ("net: change addr_list_lock back to static key")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-26 02:13:29 +08:00
|
|
|
|
2014-07-31 02:27:36 +08:00
|
|
|
struct netdev_hw_addr_list uc;
|
|
|
|
struct netdev_hw_addr_list mc;
|
|
|
|
struct netdev_hw_addr_list dev_addrs;
|
|
|
|
|
net: reorder some fields of net_device
As time passed, some fields were added in net_device, and not
at sensible offsets.
Lets reorder some fields to reduce number of cache lines in RX path.
Fields not used in data path should be moved out of this critical cache
line.
In particular, move broadcast[] to the end of the rx section,
as it is less used, and ethernet uses only the beginning of the 32bytes
field.
Before patch :
offsetof(struct net_device,dev_addr)=0x258
offsetof(struct net_device,rx_handler)=0x2b8
offsetof(struct net_device,ingress_queue)=0x2c8
offsetof(struct net_device,broadcast)=0x278
After :
offsetof(struct net_device,dev_addr)=0x280
offsetof(struct net_device,rx_handler)=0x298
offsetof(struct net_device,ingress_queue)=0x2a8
offsetof(struct net_device,broadcast)=0x2b0
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2013-03-30 14:31:03 +08:00
|
|
|
#ifdef CONFIG_SYSFS
|
|
|
|
struct kset *queues_kset;
|
net: core: add nested_level variable in net_device
This patch is to add a new variable 'nested_level' into the net_device
structure.
This variable will be used as a parameter of spin_lock_nested() of
dev->addr_list_lock.
netif_addr_lock() can be called recursively so spin_lock_nested() is
used instead of spin_lock() and dev->lower_level is used as a parameter
of spin_lock_nested().
But, dev->lower_level value can be updated while it is being used.
So, lockdep would warn a possible deadlock scenario.
When a stacked interface is deleted, netif_{uc | mc}_sync() is
called recursively.
So, spin_lock_nested() is called recursively too.
At this moment, the dev->lower_level variable is used as a parameter of it.
dev->lower_level value is updated when interfaces are being unlinked/linked
immediately.
Thus, After unlinking, dev->lower_level shouldn't be a parameter of
spin_lock_nested().
A (macvlan)
|
B (vlan)
|
C (bridge)
|
D (macvlan)
|
E (vlan)
|
F (bridge)
A->lower_level : 6
B->lower_level : 5
C->lower_level : 4
D->lower_level : 3
E->lower_level : 2
F->lower_level : 1
When an interface 'A' is removed, it releases resources.
At this moment, netif_addr_lock() would be called.
Then, netdev_upper_dev_unlink() is called recursively.
Then dev->lower_level is updated.
There is no problem.
But, when the bridge module is removed, 'C' and 'F' interfaces
are removed at once.
If 'F' is removed first, a lower_level value is like below.
A->lower_level : 5
B->lower_level : 4
C->lower_level : 3
D->lower_level : 2
E->lower_level : 1
F->lower_level : 1
Then, 'C' is removed. at this moment, netif_addr_lock() is called
recursively.
The ordering is like this.
C(3)->D(2)->E(1)->F(1)
At this moment, the lower_level value of 'E' and 'F' are the same.
So, lockdep warns a possible deadlock scenario.
In order to avoid this problem, a new variable 'nested_level' is added.
This value is the same as dev->lower_level - 1.
But this value is updated in rtnl_unlock().
So, this variable can be used as a parameter of spin_lock_nested() safely
in the rtnl context.
Test commands:
ip link add br0 type bridge vlan_filtering 1
ip link add vlan1 link br0 type vlan id 10
ip link add macvlan2 link vlan1 type macvlan
ip link add br3 type bridge vlan_filtering 1
ip link set macvlan2 master br3
ip link add vlan4 link br3 type vlan id 10
ip link add macvlan5 link vlan4 type macvlan
ip link add br6 type bridge vlan_filtering 1
ip link set macvlan5 master br6
ip link add vlan7 link br6 type vlan id 10
ip link add macvlan8 link vlan7 type macvlan
ip link set br0 up
ip link set vlan1 up
ip link set macvlan2 up
ip link set br3 up
ip link set vlan4 up
ip link set macvlan5 up
ip link set br6 up
ip link set vlan7 up
ip link set macvlan8 up
modprobe -rv bridge
Splat looks like:
[ 36.057436][ T744] WARNING: possible recursive locking detected
[ 36.058848][ T744] 5.9.0-rc6+ #728 Not tainted
[ 36.059959][ T744] --------------------------------------------
[ 36.061391][ T744] ip/744 is trying to acquire lock:
[ 36.062590][ T744] ffff8c4767509280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_set_rx_mode+0x19/0x30
[ 36.064922][ T744]
[ 36.064922][ T744] but task is already holding lock:
[ 36.066626][ T744] ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.068851][ T744]
[ 36.068851][ T744] other info that might help us debug this:
[ 36.070731][ T744] Possible unsafe locking scenario:
[ 36.070731][ T744]
[ 36.072497][ T744] CPU0
[ 36.073238][ T744] ----
[ 36.074007][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.075290][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.076590][ T744]
[ 36.076590][ T744] *** DEADLOCK ***
[ 36.076590][ T744]
[ 36.078515][ T744] May be due to missing lock nesting notation
[ 36.078515][ T744]
[ 36.080491][ T744] 3 locks held by ip/744:
[ 36.081471][ T744] #0: ffffffff98571df0 (rtnl_mutex){+.+.}-{3:3}, at: rtnetlink_rcv_msg+0x236/0x490
[ 36.083614][ T744] #1: ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.085942][ T744] #2: ffff8c476c8da280 (&bridge_netdev_addr_lock_key/4){+...}-{2:2}, at: dev_uc_sync+0x39/0x80
[ 36.088400][ T744]
[ 36.088400][ T744] stack backtrace:
[ 36.089772][ T744] CPU: 6 PID: 744 Comm: ip Not tainted 5.9.0-rc6+ #728
[ 36.091364][ T744] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
[ 36.093630][ T744] Call Trace:
[ 36.094416][ T744] dump_stack+0x77/0x9b
[ 36.095385][ T744] __lock_acquire+0xbc3/0x1f40
[ 36.096522][ T744] lock_acquire+0xb4/0x3b0
[ 36.097540][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.098657][ T744] ? rtmsg_ifinfo+0x1f/0x30
[ 36.099711][ T744] ? __dev_notify_flags+0xa5/0xf0
[ 36.100874][ T744] ? rtnl_is_locked+0x11/0x20
[ 36.101967][ T744] ? __dev_set_promiscuity+0x7b/0x1a0
[ 36.103230][ T744] _raw_spin_lock_bh+0x38/0x70
[ 36.104348][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.105461][ T744] dev_set_rx_mode+0x19/0x30
[ 36.106532][ T744] dev_set_promiscuity+0x36/0x50
[ 36.107692][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.108929][ T744] dev_set_promiscuity+0x1e/0x50
[ 36.110093][ T744] br_port_set_promisc+0x1f/0x40 [bridge]
[ 36.111415][ T744] br_manage_promisc+0x8b/0xe0 [bridge]
[ 36.112728][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.113967][ T744] ? __hw_addr_sync_one+0x23/0x50
[ 36.115135][ T744] __dev_set_rx_mode+0x68/0x90
[ 36.116249][ T744] dev_uc_sync+0x70/0x80
[ 36.117244][ T744] dev_uc_add+0x50/0x60
[ 36.118223][ T744] macvlan_open+0x18e/0x1f0 [macvlan]
[ 36.119470][ T744] __dev_open+0xd6/0x170
[ 36.120470][ T744] __dev_change_flags+0x181/0x1d0
[ 36.121644][ T744] dev_change_flags+0x23/0x60
[ 36.122741][ T744] do_setlink+0x30a/0x11e0
[ 36.123778][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.124929][ T744] ? __nla_validate_parse.part.6+0x45/0x8e0
[ 36.126309][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.127457][ T744] __rtnl_newlink+0x546/0x8e0
[ 36.128560][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.129623][ T744] ? deactivate_slab.isra.85+0x6a1/0x850
[ 36.130946][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.132102][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.133176][ T744] ? is_bpf_text_address+0x5/0xe0
[ 36.134364][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.135445][ T744] ? rcu_read_lock_sched_held+0x32/0x60
[ 36.136771][ T744] ? kmem_cache_alloc_trace+0x2d8/0x380
[ 36.138070][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.139164][ T744] rtnl_newlink+0x47/0x70
[ ... ]
Fixes: 845e0ebb4408 ("net: change addr_list_lock back to static key")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-26 02:13:29 +08:00
|
|
|
#endif
|
|
|
|
#ifdef CONFIG_LOCKDEP
|
|
|
|
struct list_head unlink_list;
|
net: reorder some fields of net_device
As time passed, some fields were added in net_device, and not
at sensible offsets.
Lets reorder some fields to reduce number of cache lines in RX path.
Fields not used in data path should be moved out of this critical cache
line.
In particular, move broadcast[] to the end of the rx section,
as it is less used, and ethernet uses only the beginning of the 32bytes
field.
Before patch :
offsetof(struct net_device,dev_addr)=0x258
offsetof(struct net_device,rx_handler)=0x2b8
offsetof(struct net_device,ingress_queue)=0x2c8
offsetof(struct net_device,broadcast)=0x278
After :
offsetof(struct net_device,dev_addr)=0x280
offsetof(struct net_device,rx_handler)=0x298
offsetof(struct net_device,ingress_queue)=0x2a8
offsetof(struct net_device,broadcast)=0x2b0
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2013-03-30 14:31:03 +08:00
|
|
|
#endif
|
2008-06-18 12:12:48 +08:00
|
|
|
unsigned int promiscuity;
|
|
|
|
unsigned int allmulti;
|
net: core: add nested_level variable in net_device
This patch is to add a new variable 'nested_level' into the net_device
structure.
This variable will be used as a parameter of spin_lock_nested() of
dev->addr_list_lock.
netif_addr_lock() can be called recursively so spin_lock_nested() is
used instead of spin_lock() and dev->lower_level is used as a parameter
of spin_lock_nested().
But, dev->lower_level value can be updated while it is being used.
So, lockdep would warn a possible deadlock scenario.
When a stacked interface is deleted, netif_{uc | mc}_sync() is
called recursively.
So, spin_lock_nested() is called recursively too.
At this moment, the dev->lower_level variable is used as a parameter of it.
dev->lower_level value is updated when interfaces are being unlinked/linked
immediately.
Thus, After unlinking, dev->lower_level shouldn't be a parameter of
spin_lock_nested().
A (macvlan)
|
B (vlan)
|
C (bridge)
|
D (macvlan)
|
E (vlan)
|
F (bridge)
A->lower_level : 6
B->lower_level : 5
C->lower_level : 4
D->lower_level : 3
E->lower_level : 2
F->lower_level : 1
When an interface 'A' is removed, it releases resources.
At this moment, netif_addr_lock() would be called.
Then, netdev_upper_dev_unlink() is called recursively.
Then dev->lower_level is updated.
There is no problem.
But, when the bridge module is removed, 'C' and 'F' interfaces
are removed at once.
If 'F' is removed first, a lower_level value is like below.
A->lower_level : 5
B->lower_level : 4
C->lower_level : 3
D->lower_level : 2
E->lower_level : 1
F->lower_level : 1
Then, 'C' is removed. at this moment, netif_addr_lock() is called
recursively.
The ordering is like this.
C(3)->D(2)->E(1)->F(1)
At this moment, the lower_level value of 'E' and 'F' are the same.
So, lockdep warns a possible deadlock scenario.
In order to avoid this problem, a new variable 'nested_level' is added.
This value is the same as dev->lower_level - 1.
But this value is updated in rtnl_unlock().
So, this variable can be used as a parameter of spin_lock_nested() safely
in the rtnl context.
Test commands:
ip link add br0 type bridge vlan_filtering 1
ip link add vlan1 link br0 type vlan id 10
ip link add macvlan2 link vlan1 type macvlan
ip link add br3 type bridge vlan_filtering 1
ip link set macvlan2 master br3
ip link add vlan4 link br3 type vlan id 10
ip link add macvlan5 link vlan4 type macvlan
ip link add br6 type bridge vlan_filtering 1
ip link set macvlan5 master br6
ip link add vlan7 link br6 type vlan id 10
ip link add macvlan8 link vlan7 type macvlan
ip link set br0 up
ip link set vlan1 up
ip link set macvlan2 up
ip link set br3 up
ip link set vlan4 up
ip link set macvlan5 up
ip link set br6 up
ip link set vlan7 up
ip link set macvlan8 up
modprobe -rv bridge
Splat looks like:
[ 36.057436][ T744] WARNING: possible recursive locking detected
[ 36.058848][ T744] 5.9.0-rc6+ #728 Not tainted
[ 36.059959][ T744] --------------------------------------------
[ 36.061391][ T744] ip/744 is trying to acquire lock:
[ 36.062590][ T744] ffff8c4767509280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_set_rx_mode+0x19/0x30
[ 36.064922][ T744]
[ 36.064922][ T744] but task is already holding lock:
[ 36.066626][ T744] ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.068851][ T744]
[ 36.068851][ T744] other info that might help us debug this:
[ 36.070731][ T744] Possible unsafe locking scenario:
[ 36.070731][ T744]
[ 36.072497][ T744] CPU0
[ 36.073238][ T744] ----
[ 36.074007][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.075290][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.076590][ T744]
[ 36.076590][ T744] *** DEADLOCK ***
[ 36.076590][ T744]
[ 36.078515][ T744] May be due to missing lock nesting notation
[ 36.078515][ T744]
[ 36.080491][ T744] 3 locks held by ip/744:
[ 36.081471][ T744] #0: ffffffff98571df0 (rtnl_mutex){+.+.}-{3:3}, at: rtnetlink_rcv_msg+0x236/0x490
[ 36.083614][ T744] #1: ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.085942][ T744] #2: ffff8c476c8da280 (&bridge_netdev_addr_lock_key/4){+...}-{2:2}, at: dev_uc_sync+0x39/0x80
[ 36.088400][ T744]
[ 36.088400][ T744] stack backtrace:
[ 36.089772][ T744] CPU: 6 PID: 744 Comm: ip Not tainted 5.9.0-rc6+ #728
[ 36.091364][ T744] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
[ 36.093630][ T744] Call Trace:
[ 36.094416][ T744] dump_stack+0x77/0x9b
[ 36.095385][ T744] __lock_acquire+0xbc3/0x1f40
[ 36.096522][ T744] lock_acquire+0xb4/0x3b0
[ 36.097540][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.098657][ T744] ? rtmsg_ifinfo+0x1f/0x30
[ 36.099711][ T744] ? __dev_notify_flags+0xa5/0xf0
[ 36.100874][ T744] ? rtnl_is_locked+0x11/0x20
[ 36.101967][ T744] ? __dev_set_promiscuity+0x7b/0x1a0
[ 36.103230][ T744] _raw_spin_lock_bh+0x38/0x70
[ 36.104348][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.105461][ T744] dev_set_rx_mode+0x19/0x30
[ 36.106532][ T744] dev_set_promiscuity+0x36/0x50
[ 36.107692][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.108929][ T744] dev_set_promiscuity+0x1e/0x50
[ 36.110093][ T744] br_port_set_promisc+0x1f/0x40 [bridge]
[ 36.111415][ T744] br_manage_promisc+0x8b/0xe0 [bridge]
[ 36.112728][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.113967][ T744] ? __hw_addr_sync_one+0x23/0x50
[ 36.115135][ T744] __dev_set_rx_mode+0x68/0x90
[ 36.116249][ T744] dev_uc_sync+0x70/0x80
[ 36.117244][ T744] dev_uc_add+0x50/0x60
[ 36.118223][ T744] macvlan_open+0x18e/0x1f0 [macvlan]
[ 36.119470][ T744] __dev_open+0xd6/0x170
[ 36.120470][ T744] __dev_change_flags+0x181/0x1d0
[ 36.121644][ T744] dev_change_flags+0x23/0x60
[ 36.122741][ T744] do_setlink+0x30a/0x11e0
[ 36.123778][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.124929][ T744] ? __nla_validate_parse.part.6+0x45/0x8e0
[ 36.126309][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.127457][ T744] __rtnl_newlink+0x546/0x8e0
[ 36.128560][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.129623][ T744] ? deactivate_slab.isra.85+0x6a1/0x850
[ 36.130946][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.132102][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.133176][ T744] ? is_bpf_text_address+0x5/0xe0
[ 36.134364][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.135445][ T744] ? rcu_read_lock_sched_held+0x32/0x60
[ 36.136771][ T744] ? kmem_cache_alloc_trace+0x2d8/0x380
[ 36.138070][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.139164][ T744] rtnl_newlink+0x47/0x70
[ ... ]
Fixes: 845e0ebb4408 ("net: change addr_list_lock back to static key")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-26 02:13:29 +08:00
|
|
|
bool uc_promisc;
|
|
|
|
#ifdef CONFIG_LOCKDEP
|
|
|
|
unsigned char nested_level;
|
|
|
|
#endif
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
2016-03-24 02:47:23 +08:00
|
|
|
/* Protocol-specific pointers */
|
2010-10-20 21:56:05 +08:00
|
|
|
|
2022-05-20 04:20:54 +08:00
|
|
|
struct in_device __rcu *ip_ptr;
|
|
|
|
struct inet6_dev __rcu *ip6_ptr;
|
2011-11-25 22:40:26 +08:00
|
|
|
#if IS_ENABLED(CONFIG_VLAN_8021Q)
|
2014-07-31 02:27:36 +08:00
|
|
|
struct vlan_info __rcu *vlan_info;
|
2010-10-20 21:56:05 +08:00
|
|
|
#endif
|
2011-11-25 22:38:38 +08:00
|
|
|
#if IS_ENABLED(CONFIG_NET_DSA)
|
2017-09-30 05:19:20 +08:00
|
|
|
struct dsa_port *dsa_ptr;
|
tipc: remove TIPC usage of field af_packet_priv in struct net_device
TIPC is currently using the field 'af_packet_priv' in struct net_device
as a handle to find the bearer instance associated to the given network
device. But, by doing so it is blocking other networking cleanups, such
as the one discussed here:
http://patchwork.ozlabs.org/patch/178044/
This commit removes this usage from TIPC. Instead, we introduce a new
field, 'tipc_ptr', to the net_device structure, to serve this purpose.
When TIPC bearer is enabled, the bearer object is associated to
'tipc_ptr'. When a TIPC packet arrives in the recv_msg() upcall
from a networking device, the bearer object can now be obtained from
'tipc_ptr'. When a bearer is disabled, the bearer object is detached
from its underlying network device by setting 'tipc_ptr' to NULL.
Additionally, an RCU lock is used to protect the new pointer.
Henceforth, the existing tipc_net_lock is used in write mode to
serialize write accesses to this pointer, while the new RCU lock is
applied on the read side to ensure that the pointer is 100% valid
within its wrapped area for all readers.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Cc: Patrick McHardy <kaber@trash.net>
Reviewed-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2013-12-11 12:45:41 +08:00
|
|
|
#endif
|
|
|
|
#if IS_ENABLED(CONFIG_TIPC)
|
2014-07-31 02:27:36 +08:00
|
|
|
struct tipc_bearer __rcu *tipc_ptr;
|
net: Distributed Switch Architecture protocol support
Distributed Switch Architecture is a protocol for managing hardware
switch chips. It consists of a set of MII management registers and
commands to configure the switch, and an ethernet header format to
signal which of the ports of the switch a packet was received from
or is intended to be sent to.
The switches that this driver supports are typically embedded in
access points and routers, and a typical setup with a DSA switch
looks something like this:
+-----------+ +-----------+
| | RGMII | |
| +-------+ +------ 1000baseT MDI ("WAN")
| | | 6-port +------ 1000baseT MDI ("LAN1")
| CPU | | ethernet +------ 1000baseT MDI ("LAN2")
| |MIImgmt| switch +------ 1000baseT MDI ("LAN3")
| +-------+ w/5 PHYs +------ 1000baseT MDI ("LAN4")
| | | |
+-----------+ +-----------+
The switch driver presents each port on the switch as a separate
network interface to Linux, polls the switch to maintain software
link state of those ports, forwards MII management interface
accesses to those network interfaces (e.g. as done by ethtool) to
the switch, and exposes the switch's hardware statistics counters
via the appropriate Linux kernel interfaces.
This initial patch supports the MII management interface register
layout of the Marvell 88E6123, 88E6161 and 88E6165 switch chips, and
supports the "Ethertype DSA" packet tagging format.
(There is no officially registered ethertype for the Ethertype DSA
packet format, so we just grab a random one. The ethertype to use
is programmed into the switch, and the switch driver uses the value
of ETH_P_EDSA for this, so this define can be changed at any time in
the future if the one we chose is allocated to another protocol or
if Ethertype DSA gets its own officially registered ethertype, and
everything will continue to work.)
Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Tested-by: Nicolas Pitre <nico@marvell.com>
Tested-by: Byron Bradley <byron.bbradley@gmail.com>
Tested-by: Tim Ellis <tim.ellis@mac.com>
Tested-by: Peter van Valderen <linux@ddcrew.com>
Tested-by: Dirk Teurlings <dirk@upexia.nl>
Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-07 21:44:02 +08:00
|
|
|
#endif
|
2021-12-29 19:36:20 +08:00
|
|
|
#if IS_ENABLED(CONFIG_ATALK)
|
2014-07-31 02:27:36 +08:00
|
|
|
void *atalk_ptr;
|
2018-02-14 00:52:02 +08:00
|
|
|
#endif
|
2018-02-14 00:52:01 +08:00
|
|
|
#if IS_ENABLED(CONFIG_AX25)
|
2014-07-31 02:27:36 +08:00
|
|
|
void *ax25_ptr;
|
2018-02-14 00:52:01 +08:00
|
|
|
#endif
|
2022-05-20 04:20:54 +08:00
|
|
|
#if IS_ENABLED(CONFIG_CFG80211)
|
2014-07-31 02:27:36 +08:00
|
|
|
struct wireless_dev *ieee80211_ptr;
|
2022-05-20 04:20:54 +08:00
|
|
|
#endif
|
|
|
|
#if IS_ENABLED(CONFIG_IEEE802154) || IS_ENABLED(CONFIG_6LOWPAN)
|
2014-11-02 13:44:54 +08:00
|
|
|
struct wpan_dev *ieee802154_ptr;
|
2022-05-20 04:20:54 +08:00
|
|
|
#endif
|
2015-04-22 18:14:37 +08:00
|
|
|
#if IS_ENABLED(CONFIG_MPLS_ROUTING)
|
|
|
|
struct mpls_dev __rcu *mpls_ptr;
|
|
|
|
#endif
|
2021-07-29 10:20:44 +08:00
|
|
|
#if IS_ENABLED(CONFIG_MCTP)
|
|
|
|
struct mctp_dev __rcu *mctp_ptr;
|
|
|
|
#endif
|
2005-04-17 06:20:36 +08:00
|
|
|
|
[NET]: Reorder some hot fields of struct net_device
Place them on separate cache lines in SMP to lower memory bouncing
between multiple CPU accessing the device.
- One part is mostly used on receive path (including
eth_type_trans()) (poll_list, poll, quota, weight, last_rx,
dev_addr, broadcast)
- One part is mostly used on queue transmit path (qdisc)
(queue_lock, qdisc, qdisc_sleeping, qdisc_list, tx_queue_len)
- One part is mostly used on xmit path (device)
(xmit_lock, xmit_lock_owner, priv, hard_start_xmit, trans_start)
'features' is placed outside of these hot points, in a location that
may be shared by all cpus (because mostly read)
name_hlist is moved close to name[IFNAMSIZ] to speedup __dev_get_by_name()
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-09-28 06:23:16 +08:00
|
|
|
/*
|
2010-09-16 10:58:13 +08:00
|
|
|
* Cache lines mostly used on receive path (including eth_type_trans())
|
[NET]: Reorder some hot fields of struct net_device
Place them on separate cache lines in SMP to lower memory bouncing
between multiple CPU accessing the device.
- One part is mostly used on receive path (including
eth_type_trans()) (poll_list, poll, quota, weight, last_rx,
dev_addr, broadcast)
- One part is mostly used on queue transmit path (qdisc)
(queue_lock, qdisc, qdisc_sleeping, qdisc_list, tx_queue_len)
- One part is mostly used on xmit path (device)
(xmit_lock, xmit_lock_owner, priv, hard_start_xmit, trans_start)
'features' is placed outside of these hot points, in a location that
may be shared by all cpus (because mostly read)
name_hlist is moved close to name[IFNAMSIZ] to speedup __dev_get_by_name()
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-09-28 06:23:16 +08:00
|
|
|
*/
|
|
|
|
/* Interface address info used in eth_type_trans() */
|
2021-11-19 22:21:51 +08:00
|
|
|
const unsigned char *dev_addr;
|
2009-05-05 10:48:28 +08:00
|
|
|
|
2010-03-16 16:03:29 +08:00
|
|
|
struct netdev_rx_queue *_rx;
|
|
|
|
unsigned int num_rx_queues;
|
2010-09-27 16:24:33 +08:00
|
|
|
unsigned int real_num_rx_queues;
|
2010-03-16 16:03:29 +08:00
|
|
|
|
2017-04-26 02:36:52 +08:00
|
|
|
struct bpf_prog __rcu *xdp_prog;
|
net: gro: add a per device gro flush timer
Tuning coalescing parameters on NIC can be really hard.
Servers can handle both bulk and RPC like traffic, with conflicting
goals : bulk flows want as big GRO packets as possible, RPC want minimal
latencies.
To reach big GRO packets on 10Gbe NIC, one can use :
ethtool -C eth0 rx-usecs 4 rx-frames 44
But this penalizes rpc sessions, with an increase of latencies, up to
50% in some cases, as NICs generally do not force an interrupt when
a packet with TCP Push flag is received.
Some NICs do not have an absolute timer, only a timer rearmed for every
incoming packet.
This patch uses a different strategy : Let GRO stack decides what do do,
based on traffic pattern.
Packets with Push flag wont be delayed.
Packets without Push flag might be held in GRO engine, if we keep
receiving data.
This new mechanism is off by default, and shall be enabled by setting
/sys/class/net/ethX/gro_flush_timeout to a value in nanosecond.
To fully enable this mechanism, drivers should use napi_complete_done()
instead of napi_complete().
Tested:
Ran 200 netperf TCP_STREAM from A to B (10Gbe mlx4 link, 8 RX queues)
Without this feature, we send back about 305,000 ACK per second.
GRO aggregation ratio is low (811/305 = 2.65 segments per GRO packet)
Setting a timer of 2000 nsec is enough to increase GRO packet sizes
and reduce number of ACK packets. (811/19.2 = 42)
Receiver performs less calls to upper stacks, less wakes up.
This also reduces cpu usage on the sender, as it receives less ACK
packets.
Note that reducing number of wakes up increases cpu efficiency, but can
decrease QPS, as applications wont have the chance to warmup cpu caches
doing a partial read of RPC requests/answers if they fit in one skb.
B:~# sar -n DEV 1 10 | grep eth0 | tail -1
Average: eth0 811269.80 305732.30 1199462.57 19705.72 0.00
0.00 0.50
B:~# echo 2000 >/sys/class/net/eth0/gro_flush_timeout
B:~# sar -n DEV 1 10 | grep eth0 | tail -1
Average: eth0 811577.30 19230.80 1199916.51 1239.80 0.00
0.00 0.50
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-11-07 13:09:44 +08:00
|
|
|
unsigned long gro_flush_timeout;
|
net: napi: add hard irqs deferral feature
Back in commit 3b47d30396ba ("net: gro: add a per device gro flush timer")
we added the ability to arm one high resolution timer, that we used
to keep not-complete packets in GRO engine a bit longer, hoping that further
frames might be added to them.
Since then, we added the napi_complete_done() interface, and commit
364b6055738b ("net: busy-poll: return busypolling status to drivers")
allowed drivers to avoid re-arming NIC interrupts if we made a promise
that their NAPI poll() handler would be called in the near future.
This infrastructure can be leveraged, thanks to a new device parameter,
which allows to arm the napi hrtimer, instead of re-arming the device
hard IRQ.
We have noticed that on some servers with 32 RX queues or more, the chit-chat
between the NIC and the host caused by IRQ delivery and re-arming could hurt
throughput by ~20% on 100Gbit NIC.
In contrast, hrtimers are using local (percpu) resources and might have lower
cost.
The new tunable, named napi_defer_hard_irqs, is placed in the same hierarchy
than gro_flush_timeout (/sys/class/net/ethX/)
By default, both gro_flush_timeout and napi_defer_hard_irqs are zero.
This patch does not change the prior behavior of gro_flush_timeout
if used alone : NIC hard irqs should be rearmed as before.
One concrete usage can be :
echo 20000 >/sys/class/net/eth1/gro_flush_timeout
echo 10 >/sys/class/net/eth1/napi_defer_hard_irqs
If at least one packet is retired, then we will reset napi counter
to 10 (napi_defer_hard_irqs), ensuring at least 10 periodic scans
of the queue.
On busy queues, this should avoid NIC hard IRQ, while before this patch IRQ
avoidance was only possible if napi->poll() was exhausting its budget
and not call napi_complete_done().
This feature also can be used to work around some non-optimal NIC irq
coalescing strategies.
Having the ability to insert XX usec delays between each napi->poll()
can increase cache efficiency, since we increase batch sizes.
It also keeps serving cpus not idle too long, reducing tail latencies.
Co-developed-by: Luigi Rizzo <lrizzo@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-04-23 00:13:27 +08:00
|
|
|
int napi_defer_hard_irqs;
|
2022-05-14 02:34:03 +08:00
|
|
|
#define GRO_LEGACY_MAX_SIZE 65536u
|
|
|
|
/* TCP minimal MSS is 8 (TCP_MIN_GSO_SIZE),
|
|
|
|
* and shinfo->gso_segs is a 16bit field.
|
|
|
|
*/
|
|
|
|
#define GRO_MAX_SIZE (8 * 65535u)
|
2022-01-05 18:48:38 +08:00
|
|
|
unsigned int gro_max_size;
|
2010-11-15 14:38:12 +08:00
|
|
|
rx_handler_func_t __rcu *rx_handler;
|
|
|
|
void __rcu *rx_handler_data;
|
2008-07-17 15:34:19 +08:00
|
|
|
|
2015-05-12 01:28:49 +08:00
|
|
|
#ifdef CONFIG_NET_CLS_ACT
|
2017-11-03 18:46:25 +08:00
|
|
|
struct mini_Qdisc __rcu *miniq_ingress;
|
net: sched: further simplify handle_ing
Ingress qdisc has no other purpose than calling into tc_classify()
that executes attached classifier(s) and action(s).
It has a 1:1 relationship to dev->ingress_queue. After having commit
087c1a601ad7 ("net: sched: run ingress qdisc without locks") removed
the central ingress lock, one major contention point is gone.
The extra indirection layers however, are not necessary for calling
into ingress qdisc. pktgen calling locally into netif_receive_skb()
with a dummy u32, single CPU result on a Supermicro X10SLM-F, Xeon
E3-1240: before ~21,1 Mpps, after patch ~22,9 Mpps.
We can redirect the private classifier list to the netdev directly,
without changing any classifier API bits (!) and execute on that from
handle_ing() side. The __QDISC_STATE_DEACTIVATE test can be removed,
ingress qdisc doesn't have a queue and thus dev_deactivate_queue()
is also not applicable, ingress_cl_list provides similar behaviour.
In other words, ingress qdisc acts like TCQ_F_BUILTIN qdisc.
One next possible step is the removal of the dev's ingress (dummy)
netdev_queue, and to only have the list member in the netdevice
itself.
Note, the filter chain is RCU protected and individual filter elements
are being kfree'd by sched subsystem after RCU grace period. RCU read
lock is being held by __netif_receive_skb_core().
Joint work with Alexei Starovoitov.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@plumgrid.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-10 04:51:32 +08:00
|
|
|
#endif
|
2010-10-02 14:11:55 +08:00
|
|
|
struct netdev_queue __rcu *ingress_queue;
|
netfilter: add netfilter ingress hook after handle_ing() under unique static key
This patch adds the Netfilter ingress hook just after the existing tc ingress
hook, that seems to be the consensus solution for this.
Note that the Netfilter hook resides under the global static key that enables
ingress filtering. Nonetheless, Netfilter still also has its own static key for
minimal impact on the existing handle_ing().
* Without this patch:
Result: OK: 6216490(c6216338+d152) usec, 100000000 (60byte,0frags)
16086246pps 7721Mb/sec (7721398080bps) errors: 100000000
42.46% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core
25.92% kpktgend_0 [kernel.kallsyms] [k] kfree_skb
7.81% kpktgend_0 [pktgen] [k] pktgen_thread_worker
5.62% kpktgend_0 [kernel.kallsyms] [k] ip_rcv
2.70% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal
2.34% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk
1.44% kpktgend_0 [kernel.kallsyms] [k] __build_skb
* With this patch:
Result: OK: 6214833(c6214731+d101) usec, 100000000 (60byte,0frags)
16090536pps 7723Mb/sec (7723457280bps) errors: 100000000
41.23% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core
26.57% kpktgend_0 [kernel.kallsyms] [k] kfree_skb
7.72% kpktgend_0 [pktgen] [k] pktgen_thread_worker
5.55% kpktgend_0 [kernel.kallsyms] [k] ip_rcv
2.78% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal
2.06% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk
1.43% kpktgend_0 [kernel.kallsyms] [k] __build_skb
* Without this patch + tc ingress:
tc filter add dev eth4 parent ffff: protocol ip prio 1 \
u32 match ip dst 4.3.2.1/32
Result: OK: 9269001(c9268821+d179) usec, 100000000 (60byte,0frags)
10788648pps 5178Mb/sec (5178551040bps) errors: 100000000
40.99% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core
17.50% kpktgend_0 [kernel.kallsyms] [k] kfree_skb
11.77% kpktgend_0 [cls_u32] [k] u32_classify
5.62% kpktgend_0 [kernel.kallsyms] [k] tc_classify_compat
5.18% kpktgend_0 [pktgen] [k] pktgen_thread_worker
3.23% kpktgend_0 [kernel.kallsyms] [k] tc_classify
2.97% kpktgend_0 [kernel.kallsyms] [k] ip_rcv
1.83% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal
1.50% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk
0.99% kpktgend_0 [kernel.kallsyms] [k] __build_skb
* With this patch + tc ingress:
tc filter add dev eth4 parent ffff: protocol ip prio 1 \
u32 match ip dst 4.3.2.1/32
Result: OK: 9308218(c9308091+d126) usec, 100000000 (60byte,0frags)
10743194pps 5156Mb/sec (5156733120bps) errors: 100000000
42.01% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core
17.78% kpktgend_0 [kernel.kallsyms] [k] kfree_skb
11.70% kpktgend_0 [cls_u32] [k] u32_classify
5.46% kpktgend_0 [kernel.kallsyms] [k] tc_classify_compat
5.16% kpktgend_0 [pktgen] [k] pktgen_thread_worker
2.98% kpktgend_0 [kernel.kallsyms] [k] ip_rcv
2.84% kpktgend_0 [kernel.kallsyms] [k] tc_classify
1.96% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal
1.57% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk
Note that the results are very similar before and after.
I can see gcc gets the code under the ingress static key out of the hot path.
Then, on that cold branch, it generates the code to accomodate the netfilter
ingress static key. My explanation for this is that this reduces the pressure
on the instruction cache for non-users as the new code is out of the hot path,
and it comes with minimal impact for tc ingress users.
Using gcc version 4.8.4 on:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 8
[...]
L1d cache: 16K
L1i cache: 64K
L2 cache: 2048K
L3 cache: 8192K
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Acked-by: Alexei Starovoitov <ast@plumgrid.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-14 00:19:38 +08:00
|
|
|
#ifdef CONFIG_NETFILTER_INGRESS
|
2017-08-24 06:08:32 +08:00
|
|
|
struct nf_hook_entries __rcu *nf_hooks_ingress;
|
netfilter: add netfilter ingress hook after handle_ing() under unique static key
This patch adds the Netfilter ingress hook just after the existing tc ingress
hook, that seems to be the consensus solution for this.
Note that the Netfilter hook resides under the global static key that enables
ingress filtering. Nonetheless, Netfilter still also has its own static key for
minimal impact on the existing handle_ing().
* Without this patch:
Result: OK: 6216490(c6216338+d152) usec, 100000000 (60byte,0frags)
16086246pps 7721Mb/sec (7721398080bps) errors: 100000000
42.46% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core
25.92% kpktgend_0 [kernel.kallsyms] [k] kfree_skb
7.81% kpktgend_0 [pktgen] [k] pktgen_thread_worker
5.62% kpktgend_0 [kernel.kallsyms] [k] ip_rcv
2.70% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal
2.34% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk
1.44% kpktgend_0 [kernel.kallsyms] [k] __build_skb
* With this patch:
Result: OK: 6214833(c6214731+d101) usec, 100000000 (60byte,0frags)
16090536pps 7723Mb/sec (7723457280bps) errors: 100000000
41.23% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core
26.57% kpktgend_0 [kernel.kallsyms] [k] kfree_skb
7.72% kpktgend_0 [pktgen] [k] pktgen_thread_worker
5.55% kpktgend_0 [kernel.kallsyms] [k] ip_rcv
2.78% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal
2.06% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk
1.43% kpktgend_0 [kernel.kallsyms] [k] __build_skb
* Without this patch + tc ingress:
tc filter add dev eth4 parent ffff: protocol ip prio 1 \
u32 match ip dst 4.3.2.1/32
Result: OK: 9269001(c9268821+d179) usec, 100000000 (60byte,0frags)
10788648pps 5178Mb/sec (5178551040bps) errors: 100000000
40.99% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core
17.50% kpktgend_0 [kernel.kallsyms] [k] kfree_skb
11.77% kpktgend_0 [cls_u32] [k] u32_classify
5.62% kpktgend_0 [kernel.kallsyms] [k] tc_classify_compat
5.18% kpktgend_0 [pktgen] [k] pktgen_thread_worker
3.23% kpktgend_0 [kernel.kallsyms] [k] tc_classify
2.97% kpktgend_0 [kernel.kallsyms] [k] ip_rcv
1.83% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal
1.50% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk
0.99% kpktgend_0 [kernel.kallsyms] [k] __build_skb
* With this patch + tc ingress:
tc filter add dev eth4 parent ffff: protocol ip prio 1 \
u32 match ip dst 4.3.2.1/32
Result: OK: 9308218(c9308091+d126) usec, 100000000 (60byte,0frags)
10743194pps 5156Mb/sec (5156733120bps) errors: 100000000
42.01% kpktgend_0 [kernel.kallsyms] [k] __netif_receive_skb_core
17.78% kpktgend_0 [kernel.kallsyms] [k] kfree_skb
11.70% kpktgend_0 [cls_u32] [k] u32_classify
5.46% kpktgend_0 [kernel.kallsyms] [k] tc_classify_compat
5.16% kpktgend_0 [pktgen] [k] pktgen_thread_worker
2.98% kpktgend_0 [kernel.kallsyms] [k] ip_rcv
2.84% kpktgend_0 [kernel.kallsyms] [k] tc_classify
1.96% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_internal
1.57% kpktgend_0 [kernel.kallsyms] [k] netif_receive_skb_sk
Note that the results are very similar before and after.
I can see gcc gets the code under the ingress static key out of the hot path.
Then, on that cold branch, it generates the code to accomodate the netfilter
ingress static key. My explanation for this is that this reduces the pressure
on the instruction cache for non-users as the new code is out of the hot path,
and it comes with minimal impact for tc ingress users.
Using gcc version 4.8.4 on:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 8
[...]
L1d cache: 16K
L1i cache: 64K
L2 cache: 2048K
L3 cache: 8192K
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Acked-by: Alexei Starovoitov <ast@plumgrid.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-14 00:19:38 +08:00
|
|
|
#endif
|
net: sched: further simplify handle_ing
Ingress qdisc has no other purpose than calling into tc_classify()
that executes attached classifier(s) and action(s).
It has a 1:1 relationship to dev->ingress_queue. After having commit
087c1a601ad7 ("net: sched: run ingress qdisc without locks") removed
the central ingress lock, one major contention point is gone.
The extra indirection layers however, are not necessary for calling
into ingress qdisc. pktgen calling locally into netif_receive_skb()
with a dummy u32, single CPU result on a Supermicro X10SLM-F, Xeon
E3-1240: before ~21,1 Mpps, after patch ~22,9 Mpps.
We can redirect the private classifier list to the netdev directly,
without changing any classifier API bits (!) and execute on that from
handle_ing() side. The __QDISC_STATE_DEACTIVATE test can be removed,
ingress qdisc doesn't have a queue and thus dev_deactivate_queue()
is also not applicable, ingress_cl_list provides similar behaviour.
In other words, ingress qdisc acts like TCQ_F_BUILTIN qdisc.
One next possible step is the removal of the dev's ingress (dummy)
netdev_queue, and to only have the list member in the netdevice
itself.
Note, the filter chain is RCU protected and individual filter elements
are being kfree'd by sched subsystem after RCU grace period. RCU read
lock is being held by __netif_receive_skb_core().
Joint work with Alexei Starovoitov.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@plumgrid.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-10 04:51:32 +08:00
|
|
|
|
2014-07-31 02:27:36 +08:00
|
|
|
unsigned char broadcast[MAX_ADDR_LEN];
|
2015-04-10 21:52:38 +08:00
|
|
|
#ifdef CONFIG_RFS_ACCEL
|
|
|
|
struct cpu_rmap *rx_cpu_rmap;
|
|
|
|
#endif
|
|
|
|
struct hlist_node index_hlist;
|
2010-09-16 10:58:13 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Cache lines mostly used on transmit path
|
|
|
|
*/
|
2008-07-17 15:34:19 +08:00
|
|
|
struct netdev_queue *_tx ____cacheline_aligned_in_smp;
|
|
|
|
unsigned int num_tx_queues;
|
2008-07-17 16:56:23 +08:00
|
|
|
unsigned int real_num_tx_queues;
|
2022-02-12 04:06:23 +08:00
|
|
|
struct Qdisc __rcu *qdisc;
|
2017-05-17 18:30:44 +08:00
|
|
|
unsigned int tx_queue_len;
|
2008-08-01 07:58:50 +08:00
|
|
|
spinlock_t tx_global_lock;
|
xdp: Move devmap bulk queue into struct net_device
Commit 96360004b862 ("xdp: Make devmap flush_list common for all map
instances"), changed devmap flushing to be a global operation instead of a
per-map operation. However, the queue structure used for bulking was still
allocated as part of the containing map.
This patch moves the devmap bulk queue into struct net_device. The
motivation for this is reusing it for the non-map variant of XDP_REDIRECT,
which will be changed in a subsequent commit. To avoid other fields of
struct net_device moving to different cache lines, we also move a couple of
other members around.
We defer the actual allocation of the bulk queue structure until the
NETDEV_REGISTER notification devmap.c. This makes it possible to check for
ndo_xdp_xmit support before allocating the structure, which is not possible
at the time struct net_device is allocated. However, we keep the freeing in
free_netdev() to avoid adding another RCU callback on NETDEV_UNREGISTER.
Because of this change, we lose the reference back to the map that
originated the redirect, so change the tracepoint to always return 0 as the
map ID and index. Otherwise no functional change is intended with this
patch.
After this patch, the relevant part of struct net_device looks like this,
according to pahole:
/* --- cacheline 14 boundary (896 bytes) --- */
struct netdev_queue * _tx __attribute__((__aligned__(64))); /* 896 8 */
unsigned int num_tx_queues; /* 904 4 */
unsigned int real_num_tx_queues; /* 908 4 */
struct Qdisc * qdisc; /* 912 8 */
unsigned int tx_queue_len; /* 920 4 */
spinlock_t tx_global_lock; /* 924 4 */
struct xdp_dev_bulk_queue * xdp_bulkq; /* 928 8 */
struct xps_dev_maps * xps_cpus_map; /* 936 8 */
struct xps_dev_maps * xps_rxqs_map; /* 944 8 */
struct mini_Qdisc * miniq_egress; /* 952 8 */
/* --- cacheline 15 boundary (960 bytes) --- */
struct hlist_head qdisc_hash[16]; /* 960 128 */
/* --- cacheline 17 boundary (1088 bytes) --- */
struct timer_list watchdog_timer; /* 1088 40 */
/* XXX last struct has 4 bytes of padding */
int watchdog_timeo; /* 1128 4 */
/* XXX 4 bytes hole, try to pack */
struct list_head todo_list; /* 1136 16 */
/* --- cacheline 18 boundary (1152 bytes) --- */
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Björn Töpel <bjorn.topel@intel.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/157918768397.1458396.12673224324627072349.stgit@toke.dk
2020-01-16 23:14:44 +08:00
|
|
|
|
|
|
|
struct xdp_dev_bulk_queue __percpu *xdp_bulkq;
|
2010-09-16 10:58:13 +08:00
|
|
|
|
2010-11-26 16:36:09 +08:00
|
|
|
#ifdef CONFIG_XPS
|
2021-03-19 02:37:46 +08:00
|
|
|
struct xps_dev_maps __rcu *xps_maps[XPS_MAPS_MAX];
|
2010-11-26 16:36:09 +08:00
|
|
|
#endif
|
net, sched: add clsact qdisc
This work adds a generalization of the ingress qdisc as a qdisc holding
only classifiers. The clsact qdisc works on ingress, but also on egress.
In both cases, it's execution happens without taking the qdisc lock, and
the main difference for the egress part compared to prior version of [1]
is that this can be applied with _any_ underlying real egress qdisc (also
classless ones).
Besides solving the use-case of [1], that is, allowing for more programmability
on assigning skb->priority for the mqprio case that is supported by most
popular 10G+ NICs, it also opens up a lot more flexibility for other tc
applications. The main work on classification can already be done at clsact
egress time if the use-case allows and state stored for later retrieval
f.e. again in skb->priority with major/minors (which is checked by most
classful qdiscs before consulting tc_classify()) and/or in other skb fields
like skb->tc_index for some light-weight post-processing to get to the
eventual classid in case of a classful qdisc. Another use case is that
the clsact egress part allows to have a central egress counterpart to
the ingress classifiers, so that classifiers can easily share state (e.g.
in cls_bpf via eBPF maps) for ingress and egress.
Currently, default setups like mq + pfifo_fast would require for this to
use, for example, prio qdisc instead (to get a tc_classify() run) and to
duplicate the egress classifier for each queue. With clsact, it allows
for leaving the setup as is, it can additionally assign skb->priority to
put the skb in one of pfifo_fast's bands and it can share state with maps.
Moreover, we can access the skb's dst entry (f.e. to retrieve tclassid)
w/o the need to perform a skb_dst_force() to hold on to it any longer. In
lwt case, we can also use this facility to setup dst metadata via cls_bpf
(bpf_skb_set_tunnel_key()) without needing a real egress qdisc just for
that (case of IFF_NO_QUEUE devices, for example).
The realization can be done without any changes to the scheduler core
framework. All it takes is that we have two a-priori defined minors/child
classes, where we can mux between ingress and egress classifier list
(dev->ingress_cl_list and dev->egress_cl_list, latter stored close to
dev->_tx to avoid extra cacheline miss for moderate loads). The egress
part is a bit similar modelled to handle_ing() and patched to a noop in
case the functionality is not used. Both handlers are now called
sch_handle_ingress() and sch_handle_egress(), code sharing among the two
doesn't seem practical as there are various minor differences in both
paths, so that making them conditional in a single handler would rather
slow things down.
Full compatibility to ingress qdisc is provided as well. Since both
piggyback on TC_H_CLSACT, only one of them (ingress/clsact) can exist
per netdevice, and thus ingress qdisc specific behaviour can be retained
for user space. This means, either a user does 'tc qdisc add dev foo ingress'
and configures ingress qdisc as usual, or the 'tc qdisc add dev foo clsact'
alternative, where both, ingress and egress classifier can be configured
as in the below example. ingress qdisc supports attaching classifier to any
minor number whereas clsact has two fixed minors for muxing between the
lists, therefore to not break user space setups, they are better done as
two separate qdiscs.
I decided to extend the sch_ingress module with clsact functionality so
that commonly used code can be reused, the module is being aliased with
sch_clsact so that it can be auto-loaded properly. Alternative would have been
to add a flag when initializing ingress to alter its behaviour plus aliasing
to a different name (as it's more than just ingress). However, the first would
end up, based on the flag, choosing the new/old behaviour by calling different
function implementations to handle each anyway, the latter would require to
register ingress qdisc once again under different alias. So, this really begs
to provide a minimal, cleaner approach to have Qdisc_ops and Qdisc_class_ops
by its own that share callbacks used by both.
Example, adding qdisc:
# tc qdisc add dev foo clsact
# tc qdisc show dev foo
qdisc mq 0: root
qdisc pfifo_fast 0: parent :1 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: parent :2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: parent :3 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: parent :4 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc clsact ffff: parent ffff:fff1
Adding filters (deleting, etc works analogous by specifying ingress/egress):
# tc filter add dev foo ingress bpf da obj bar.o sec ingress
# tc filter add dev foo egress bpf da obj bar.o sec egress
# tc filter show dev foo ingress
filter protocol all pref 49152 bpf
filter protocol all pref 49152 bpf handle 0x1 bar.o:[ingress] direct-action
# tc filter show dev foo egress
filter protocol all pref 49152 bpf
filter protocol all pref 49152 bpf handle 0x1 bar.o:[egress] direct-action
A 'tc filter show dev foo' or 'tc filter show dev foo parent ffff:' will
show an empty list for clsact. Either using the parent names (ingress/egress)
or specifying the full major/minor will then show the related filter lists.
Prior work on a mqprio prequeue() facility [1] was done mainly by John Fastabend.
[1] http://patchwork.ozlabs.org/patch/512949/
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-01-08 05:29:47 +08:00
|
|
|
#ifdef CONFIG_NET_CLS_ACT
|
2017-11-03 18:46:25 +08:00
|
|
|
struct mini_Qdisc __rcu *miniq_egress;
|
net, sched: add clsact qdisc
This work adds a generalization of the ingress qdisc as a qdisc holding
only classifiers. The clsact qdisc works on ingress, but also on egress.
In both cases, it's execution happens without taking the qdisc lock, and
the main difference for the egress part compared to prior version of [1]
is that this can be applied with _any_ underlying real egress qdisc (also
classless ones).
Besides solving the use-case of [1], that is, allowing for more programmability
on assigning skb->priority for the mqprio case that is supported by most
popular 10G+ NICs, it also opens up a lot more flexibility for other tc
applications. The main work on classification can already be done at clsact
egress time if the use-case allows and state stored for later retrieval
f.e. again in skb->priority with major/minors (which is checked by most
classful qdiscs before consulting tc_classify()) and/or in other skb fields
like skb->tc_index for some light-weight post-processing to get to the
eventual classid in case of a classful qdisc. Another use case is that
the clsact egress part allows to have a central egress counterpart to
the ingress classifiers, so that classifiers can easily share state (e.g.
in cls_bpf via eBPF maps) for ingress and egress.
Currently, default setups like mq + pfifo_fast would require for this to
use, for example, prio qdisc instead (to get a tc_classify() run) and to
duplicate the egress classifier for each queue. With clsact, it allows
for leaving the setup as is, it can additionally assign skb->priority to
put the skb in one of pfifo_fast's bands and it can share state with maps.
Moreover, we can access the skb's dst entry (f.e. to retrieve tclassid)
w/o the need to perform a skb_dst_force() to hold on to it any longer. In
lwt case, we can also use this facility to setup dst metadata via cls_bpf
(bpf_skb_set_tunnel_key()) without needing a real egress qdisc just for
that (case of IFF_NO_QUEUE devices, for example).
The realization can be done without any changes to the scheduler core
framework. All it takes is that we have two a-priori defined minors/child
classes, where we can mux between ingress and egress classifier list
(dev->ingress_cl_list and dev->egress_cl_list, latter stored close to
dev->_tx to avoid extra cacheline miss for moderate loads). The egress
part is a bit similar modelled to handle_ing() and patched to a noop in
case the functionality is not used. Both handlers are now called
sch_handle_ingress() and sch_handle_egress(), code sharing among the two
doesn't seem practical as there are various minor differences in both
paths, so that making them conditional in a single handler would rather
slow things down.
Full compatibility to ingress qdisc is provided as well. Since both
piggyback on TC_H_CLSACT, only one of them (ingress/clsact) can exist
per netdevice, and thus ingress qdisc specific behaviour can be retained
for user space. This means, either a user does 'tc qdisc add dev foo ingress'
and configures ingress qdisc as usual, or the 'tc qdisc add dev foo clsact'
alternative, where both, ingress and egress classifier can be configured
as in the below example. ingress qdisc supports attaching classifier to any
minor number whereas clsact has two fixed minors for muxing between the
lists, therefore to not break user space setups, they are better done as
two separate qdiscs.
I decided to extend the sch_ingress module with clsact functionality so
that commonly used code can be reused, the module is being aliased with
sch_clsact so that it can be auto-loaded properly. Alternative would have been
to add a flag when initializing ingress to alter its behaviour plus aliasing
to a different name (as it's more than just ingress). However, the first would
end up, based on the flag, choosing the new/old behaviour by calling different
function implementations to handle each anyway, the latter would require to
register ingress qdisc once again under different alias. So, this really begs
to provide a minimal, cleaner approach to have Qdisc_ops and Qdisc_class_ops
by its own that share callbacks used by both.
Example, adding qdisc:
# tc qdisc add dev foo clsact
# tc qdisc show dev foo
qdisc mq 0: root
qdisc pfifo_fast 0: parent :1 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: parent :2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: parent :3 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: parent :4 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc clsact ffff: parent ffff:fff1
Adding filters (deleting, etc works analogous by specifying ingress/egress):
# tc filter add dev foo ingress bpf da obj bar.o sec ingress
# tc filter add dev foo egress bpf da obj bar.o sec egress
# tc filter show dev foo ingress
filter protocol all pref 49152 bpf
filter protocol all pref 49152 bpf handle 0x1 bar.o:[ingress] direct-action
# tc filter show dev foo egress
filter protocol all pref 49152 bpf
filter protocol all pref 49152 bpf handle 0x1 bar.o:[egress] direct-action
A 'tc filter show dev foo' or 'tc filter show dev foo parent ffff:' will
show an empty list for clsact. Either using the parent names (ingress/egress)
or specifying the full major/minor will then show the related filter lists.
Prior work on a mqprio prequeue() facility [1] was done mainly by John Fastabend.
[1] http://patchwork.ozlabs.org/patch/512949/
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-01-08 05:29:47 +08:00
|
|
|
#endif
|
netfilter: Introduce egress hook
Support classifying packets with netfilter on egress to satisfy user
requirements such as:
* outbound security policies for containers (Laura)
* filtering and mangling intra-node Direct Server Return (DSR) traffic
on a load balancer (Laura)
* filtering locally generated traffic coming in through AF_PACKET,
such as local ARP traffic generated for clustering purposes or DHCP
(Laura; the AF_PACKET plumbing is contained in a follow-up commit)
* L2 filtering from ingress and egress for AVB (Audio Video Bridging)
and gPTP with nftables (Pablo)
* in the future: in-kernel NAT64/NAT46 (Pablo)
The egress hook introduced herein complements the ingress hook added by
commit e687ad60af09 ("netfilter: add netfilter ingress hook after
handle_ing() under unique static key"). A patch for nftables to hook up
egress rules from user space has been submitted separately, so users may
immediately take advantage of the feature.
Alternatively or in addition to netfilter, packets can be classified
with traffic control (tc). On ingress, packets are classified first by
tc, then by netfilter. On egress, the order is reversed for symmetry.
Conceptually, tc and netfilter can be thought of as layers, with
netfilter layered above tc.
Traffic control is capable of redirecting packets to another interface
(man 8 tc-mirred). E.g., an ingress packet may be redirected from the
host namespace to a container via a veth connection:
tc ingress (host) -> tc egress (veth host) -> tc ingress (veth container)
In this case, netfilter egress classifying is not performed when leaving
the host namespace! That's because the packet is still on the tc layer.
If tc redirects the packet to a physical interface in the host namespace
such that it leaves the system, the packet is never subjected to
netfilter egress classifying. That is only logical since it hasn't
passed through netfilter ingress classifying either.
Packets can alternatively be redirected at the netfilter layer using
nft fwd. Such a packet *is* subjected to netfilter egress classifying
since it has reached the netfilter layer.
Internally, the skb->nf_skip_egress flag controls whether netfilter is
invoked on egress by __dev_queue_xmit(). Because __dev_queue_xmit() may
be called recursively by tunnel drivers such as vxlan, the flag is
reverted to false after sch_handle_egress(). This ensures that
netfilter is applied both on the overlay and underlying network.
Interaction between tc and netfilter is possible by setting and querying
skb->mark.
If netfilter egress classifying is not enabled on any interface, it is
patched out of the data path by way of a static_key and doesn't make a
performance difference that is discernible from noise:
Before: 1537 1538 1538 1537 1538 1537 Mb/sec
After: 1536 1534 1539 1539 1539 1540 Mb/sec
Before + tc accept: 1418 1418 1418 1419 1419 1418 Mb/sec
After + tc accept: 1419 1424 1418 1419 1422 1420 Mb/sec
Before + tc drop: 1620 1619 1619 1619 1620 1620 Mb/sec
After + tc drop: 1616 1624 1625 1624 1622 1619 Mb/sec
When netfilter egress classifying is enabled on at least one interface,
a minimal performance penalty is incurred for every egress packet, even
if the interface it's transmitted over doesn't have any netfilter egress
rules configured. That is caused by checking dev->nf_hooks_egress
against NULL.
Measurements were performed on a Core i7-3615QM. Commands to reproduce:
ip link add dev foo type dummy
ip link set dev foo up
modprobe pktgen
echo "add_device foo" > /proc/net/pktgen/kpktgend_3
samples/pktgen/pktgen_bench_xmit_mode_queue_xmit.sh -i foo -n 400000000 -m "11:11:11:11:11:11" -d 1.1.1.1
Accept all traffic with tc:
tc qdisc add dev foo clsact
tc filter add dev foo egress bpf da bytecode '1,6 0 0 0,'
Drop all traffic with tc:
tc qdisc add dev foo clsact
tc filter add dev foo egress bpf da bytecode '1,6 0 0 2,'
Apply this patch when measuring packet drops to avoid errors in dmesg:
https://lore.kernel.org/netdev/a73dda33-57f4-95d8-ea51-ed483abd6a7a@iogearbox.net/
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Cc: Laura García Liébana <nevola@gmail.com>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Thomas Graf <tgraf@suug.ch>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2021-10-09 04:06:03 +08:00
|
|
|
#ifdef CONFIG_NETFILTER_EGRESS
|
|
|
|
struct nf_hook_entries __rcu *nf_hooks_egress;
|
|
|
|
#endif
|
2015-07-19 09:24:48 +08:00
|
|
|
|
xdp: Move devmap bulk queue into struct net_device
Commit 96360004b862 ("xdp: Make devmap flush_list common for all map
instances"), changed devmap flushing to be a global operation instead of a
per-map operation. However, the queue structure used for bulking was still
allocated as part of the containing map.
This patch moves the devmap bulk queue into struct net_device. The
motivation for this is reusing it for the non-map variant of XDP_REDIRECT,
which will be changed in a subsequent commit. To avoid other fields of
struct net_device moving to different cache lines, we also move a couple of
other members around.
We defer the actual allocation of the bulk queue structure until the
NETDEV_REGISTER notification devmap.c. This makes it possible to check for
ndo_xdp_xmit support before allocating the structure, which is not possible
at the time struct net_device is allocated. However, we keep the freeing in
free_netdev() to avoid adding another RCU callback on NETDEV_UNREGISTER.
Because of this change, we lose the reference back to the map that
originated the redirect, so change the tracepoint to always return 0 as the
map ID and index. Otherwise no functional change is intended with this
patch.
After this patch, the relevant part of struct net_device looks like this,
according to pahole:
/* --- cacheline 14 boundary (896 bytes) --- */
struct netdev_queue * _tx __attribute__((__aligned__(64))); /* 896 8 */
unsigned int num_tx_queues; /* 904 4 */
unsigned int real_num_tx_queues; /* 908 4 */
struct Qdisc * qdisc; /* 912 8 */
unsigned int tx_queue_len; /* 920 4 */
spinlock_t tx_global_lock; /* 924 4 */
struct xdp_dev_bulk_queue * xdp_bulkq; /* 928 8 */
struct xps_dev_maps * xps_cpus_map; /* 936 8 */
struct xps_dev_maps * xps_rxqs_map; /* 944 8 */
struct mini_Qdisc * miniq_egress; /* 952 8 */
/* --- cacheline 15 boundary (960 bytes) --- */
struct hlist_head qdisc_hash[16]; /* 960 128 */
/* --- cacheline 17 boundary (1088 bytes) --- */
struct timer_list watchdog_timer; /* 1088 40 */
/* XXX last struct has 4 bytes of padding */
int watchdog_timeo; /* 1128 4 */
/* XXX 4 bytes hole, try to pack */
struct list_head todo_list; /* 1136 16 */
/* --- cacheline 18 boundary (1152 bytes) --- */
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Björn Töpel <bjorn.topel@intel.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/157918768397.1458396.12673224324627072349.stgit@toke.dk
2020-01-16 23:14:44 +08:00
|
|
|
#ifdef CONFIG_NET_SCHED
|
|
|
|
DECLARE_HASHTABLE (qdisc_hash, 4);
|
|
|
|
#endif
|
[NET]: Reorder some hot fields of struct net_device
Place them on separate cache lines in SMP to lower memory bouncing
between multiple CPU accessing the device.
- One part is mostly used on receive path (including
eth_type_trans()) (poll_list, poll, quota, weight, last_rx,
dev_addr, broadcast)
- One part is mostly used on queue transmit path (qdisc)
(queue_lock, qdisc, qdisc_sleeping, qdisc_list, tx_queue_len)
- One part is mostly used on xmit path (device)
(xmit_lock, xmit_lock_owner, priv, hard_start_xmit, trans_start)
'features' is placed outside of these hot points, in a location that
may be shared by all cpus (because mostly read)
name_hlist is moved close to name[IFNAMSIZ] to speedup __dev_get_by_name()
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-09-28 06:23:16 +08:00
|
|
|
/* These may be needed for future network-power-down code. */
|
|
|
|
struct timer_list watchdog_timer;
|
xdp: Move devmap bulk queue into struct net_device
Commit 96360004b862 ("xdp: Make devmap flush_list common for all map
instances"), changed devmap flushing to be a global operation instead of a
per-map operation. However, the queue structure used for bulking was still
allocated as part of the containing map.
This patch moves the devmap bulk queue into struct net_device. The
motivation for this is reusing it for the non-map variant of XDP_REDIRECT,
which will be changed in a subsequent commit. To avoid other fields of
struct net_device moving to different cache lines, we also move a couple of
other members around.
We defer the actual allocation of the bulk queue structure until the
NETDEV_REGISTER notification devmap.c. This makes it possible to check for
ndo_xdp_xmit support before allocating the structure, which is not possible
at the time struct net_device is allocated. However, we keep the freeing in
free_netdev() to avoid adding another RCU callback on NETDEV_UNREGISTER.
Because of this change, we lose the reference back to the map that
originated the redirect, so change the tracepoint to always return 0 as the
map ID and index. Otherwise no functional change is intended with this
patch.
After this patch, the relevant part of struct net_device looks like this,
according to pahole:
/* --- cacheline 14 boundary (896 bytes) --- */
struct netdev_queue * _tx __attribute__((__aligned__(64))); /* 896 8 */
unsigned int num_tx_queues; /* 904 4 */
unsigned int real_num_tx_queues; /* 908 4 */
struct Qdisc * qdisc; /* 912 8 */
unsigned int tx_queue_len; /* 920 4 */
spinlock_t tx_global_lock; /* 924 4 */
struct xdp_dev_bulk_queue * xdp_bulkq; /* 928 8 */
struct xps_dev_maps * xps_cpus_map; /* 936 8 */
struct xps_dev_maps * xps_rxqs_map; /* 944 8 */
struct mini_Qdisc * miniq_egress; /* 952 8 */
/* --- cacheline 15 boundary (960 bytes) --- */
struct hlist_head qdisc_hash[16]; /* 960 128 */
/* --- cacheline 17 boundary (1088 bytes) --- */
struct timer_list watchdog_timer; /* 1088 40 */
/* XXX last struct has 4 bytes of padding */
int watchdog_timeo; /* 1128 4 */
/* XXX 4 bytes hole, try to pack */
struct list_head todo_list; /* 1136 16 */
/* --- cacheline 18 boundary (1152 bytes) --- */
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Björn Töpel <bjorn.topel@intel.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/157918768397.1458396.12673224324627072349.stgit@toke.dk
2020-01-16 23:14:44 +08:00
|
|
|
int watchdog_timeo;
|
[NET]: Reorder some hot fields of struct net_device
Place them on separate cache lines in SMP to lower memory bouncing
between multiple CPU accessing the device.
- One part is mostly used on receive path (including
eth_type_trans()) (poll_list, poll, quota, weight, last_rx,
dev_addr, broadcast)
- One part is mostly used on queue transmit path (qdisc)
(queue_lock, qdisc, qdisc_sleeping, qdisc_list, tx_queue_len)
- One part is mostly used on xmit path (device)
(xmit_lock, xmit_lock_owner, priv, hard_start_xmit, trans_start)
'features' is placed outside of these hot points, in a location that
may be shared by all cpus (because mostly read)
name_hlist is moved close to name[IFNAMSIZ] to speedup __dev_get_by_name()
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-09-28 06:23:16 +08:00
|
|
|
|
2020-08-01 08:34:01 +08:00
|
|
|
u32 proto_down_reason;
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
struct list_head todo_list;
|
2021-03-20 01:39:33 +08:00
|
|
|
|
|
|
|
#ifdef CONFIG_PCPU_DEV_REFCNT
|
xdp: Move devmap bulk queue into struct net_device
Commit 96360004b862 ("xdp: Make devmap flush_list common for all map
instances"), changed devmap flushing to be a global operation instead of a
per-map operation. However, the queue structure used for bulking was still
allocated as part of the containing map.
This patch moves the devmap bulk queue into struct net_device. The
motivation for this is reusing it for the non-map variant of XDP_REDIRECT,
which will be changed in a subsequent commit. To avoid other fields of
struct net_device moving to different cache lines, we also move a couple of
other members around.
We defer the actual allocation of the bulk queue structure until the
NETDEV_REGISTER notification devmap.c. This makes it possible to check for
ndo_xdp_xmit support before allocating the structure, which is not possible
at the time struct net_device is allocated. However, we keep the freeing in
free_netdev() to avoid adding another RCU callback on NETDEV_UNREGISTER.
Because of this change, we lose the reference back to the map that
originated the redirect, so change the tracepoint to always return 0 as the
map ID and index. Otherwise no functional change is intended with this
patch.
After this patch, the relevant part of struct net_device looks like this,
according to pahole:
/* --- cacheline 14 boundary (896 bytes) --- */
struct netdev_queue * _tx __attribute__((__aligned__(64))); /* 896 8 */
unsigned int num_tx_queues; /* 904 4 */
unsigned int real_num_tx_queues; /* 908 4 */
struct Qdisc * qdisc; /* 912 8 */
unsigned int tx_queue_len; /* 920 4 */
spinlock_t tx_global_lock; /* 924 4 */
struct xdp_dev_bulk_queue * xdp_bulkq; /* 928 8 */
struct xps_dev_maps * xps_cpus_map; /* 936 8 */
struct xps_dev_maps * xps_rxqs_map; /* 944 8 */
struct mini_Qdisc * miniq_egress; /* 952 8 */
/* --- cacheline 15 boundary (960 bytes) --- */
struct hlist_head qdisc_hash[16]; /* 960 128 */
/* --- cacheline 17 boundary (1088 bytes) --- */
struct timer_list watchdog_timer; /* 1088 40 */
/* XXX last struct has 4 bytes of padding */
int watchdog_timeo; /* 1128 4 */
/* XXX 4 bytes hole, try to pack */
struct list_head todo_list; /* 1136 16 */
/* --- cacheline 18 boundary (1152 bytes) --- */
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Björn Töpel <bjorn.topel@intel.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/157918768397.1458396.12673224324627072349.stgit@toke.dk
2020-01-16 23:14:44 +08:00
|
|
|
int __percpu *pcpu_refcnt;
|
2021-03-20 01:39:33 +08:00
|
|
|
#else
|
|
|
|
refcount_t dev_refcnt;
|
|
|
|
#endif
|
2021-12-05 12:21:57 +08:00
|
|
|
struct ref_tracker_dir refcnt_tracker;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
linkwatch: linkwatch_forget_dev() to speedup device dismantle
Herbert Xu a écrit :
> On Tue, Nov 17, 2009 at 04:26:04AM -0800, David Miller wrote:
>> Really, the link watch stuff is just due for a redesign. I don't
>> think a simple hack is going to cut it this time, sorry Eric :-)
>
> I have no objections against any redesigns, but since the only
> caller of linkwatch_forget_dev runs in process context with the
> RTNL, it could also legally emit those events.
Thanks guys, here an updated version then, before linkwatch surgery ?
In this version, I force the event to be sent synchronously.
[PATCH net-next-2.6] linkwatch: linkwatch_forget_dev() to speedup device dismantle
time ip link del eth3.103 ; time ip link del eth3.104 ; time ip link del eth3.105
real 0m0.266s
user 0m0.000s
sys 0m0.001s
real 0m0.770s
user 0m0.000s
sys 0m0.000s
real 0m1.022s
user 0m0.000s
sys 0m0.000s
One problem of current schem in vlan dismantle phase is the
holding of device done by following chain :
vlan_dev_stop() ->
netif_carrier_off(dev) ->
linkwatch_fire_event(dev) ->
dev_hold() ...
And __linkwatch_run_queue() runs up to one second later...
A generic fix to this problem is to add a linkwatch_forget_dev() method
to unlink the device from the list of watched devices.
dev->link_watch_next becomes dev->link_watch_list (and use a bit more memory),
to be able to unlink device in O(1).
After patch :
time ip link del eth3.103 ; time ip link del eth3.104 ; time ip link del eth3.105
real 0m0.024s
user 0m0.000s
sys 0m0.000s
real 0m0.032s
user 0m0.000s
sys 0m0.001s
real 0m0.033s
user 0m0.000s
sys 0m0.000s
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-11-17 13:59:21 +08:00
|
|
|
struct list_head link_watch_list;
|
2007-05-09 09:34:17 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
enum { NETREG_UNINITIALIZED=0,
|
2006-05-11 04:21:17 +08:00
|
|
|
NETREG_REGISTERED, /* completed register_netdevice */
|
2005-04-17 06:20:36 +08:00
|
|
|
NETREG_UNREGISTERING, /* called unregister_netdevice */
|
|
|
|
NETREG_UNREGISTERED, /* completed unregister todo */
|
|
|
|
NETREG_RELEASED, /* called free_netdev */
|
2009-01-15 13:05:05 +08:00
|
|
|
NETREG_DUMMY, /* dummy device for NAPI poll */
|
2011-05-19 20:24:16 +08:00
|
|
|
} reg_state:8;
|
|
|
|
|
2014-07-31 02:27:36 +08:00
|
|
|
bool dismantle;
|
2010-02-26 14:34:51 +08:00
|
|
|
|
|
|
|
enum {
|
|
|
|
RTNL_LINK_INITIALIZED,
|
|
|
|
RTNL_LINK_INITIALIZING,
|
|
|
|
} rtnl_link_state:16;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
net: Fix inconsistent teardown and release of private netdev state.
Network devices can allocate reasources and private memory using
netdev_ops->ndo_init(). However, the release of these resources
can occur in one of two different places.
Either netdev_ops->ndo_uninit() or netdev->destructor().
The decision of which operation frees the resources depends upon
whether it is necessary for all netdev refs to be released before it
is safe to perform the freeing.
netdev_ops->ndo_uninit() presumably can occur right after the
NETDEV_UNREGISTER notifier completes and the unicast and multicast
address lists are flushed.
netdev->destructor(), on the other hand, does not run until the
netdev references all go away.
Further complicating the situation is that netdev->destructor()
almost universally does also a free_netdev().
This creates a problem for the logic in register_netdevice().
Because all callers of register_netdevice() manage the freeing
of the netdev, and invoke free_netdev(dev) if register_netdevice()
fails.
If netdev_ops->ndo_init() succeeds, but something else fails inside
of register_netdevice(), it does call ndo_ops->ndo_uninit(). But
it is not able to invoke netdev->destructor().
This is because netdev->destructor() will do a free_netdev() and
then the caller of register_netdevice() will do the same.
However, this means that the resources that would normally be released
by netdev->destructor() will not be.
Over the years drivers have added local hacks to deal with this, by
invoking their destructor parts by hand when register_netdevice()
fails.
Many drivers do not try to deal with this, and instead we have leaks.
Let's close this hole by formalizing the distinction between what
private things need to be freed up by netdev->destructor() and whether
the driver needs unregister_netdevice() to perform the free_netdev().
netdev->priv_destructor() performs all actions to free up the private
resources that used to be freed by netdev->destructor(), except for
free_netdev().
netdev->needs_free_netdev is a boolean that indicates whether
free_netdev() should be done at the end of unregister_netdevice().
Now, register_netdevice() can sanely release all resources after
ndo_ops->ndo_init() succeeds, by invoking both ndo_ops->ndo_uninit()
and netdev->priv_destructor().
And at the end of unregister_netdevice(), we invoke
netdev->priv_destructor() and optionally call free_netdev().
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-05-09 00:52:56 +08:00
|
|
|
bool needs_free_netdev;
|
|
|
|
void (*priv_destructor)(struct net_device *dev);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
#ifdef CONFIG_NETPOLL
|
2013-01-23 05:29:39 +08:00
|
|
|
struct netpoll_info __rcu *npinfo;
|
2005-04-17 06:20:36 +08:00
|
|
|
#endif
|
2008-07-15 18:03:33 +08:00
|
|
|
|
2015-03-12 12:06:44 +08:00
|
|
|
possible_net_t nd_net;
|
2007-09-12 17:56:32 +08:00
|
|
|
|
2008-05-12 18:29:11 +08:00
|
|
|
/* mid-layer private */
|
2021-02-23 15:01:26 +08:00
|
|
|
void *ml_priv;
|
|
|
|
enum netdev_ml_priv_type ml_priv_type;
|
|
|
|
|
2010-09-24 07:51:51 +08:00
|
|
|
union {
|
2014-07-31 02:27:36 +08:00
|
|
|
struct pcpu_lstats __percpu *lstats;
|
2014-01-04 13:57:59 +08:00
|
|
|
struct pcpu_sw_netstats __percpu *tstats;
|
2014-07-31 02:27:36 +08:00
|
|
|
struct pcpu_dstats __percpu *dstats;
|
2010-09-24 07:51:51 +08:00
|
|
|
};
|
2014-07-31 02:27:36 +08:00
|
|
|
|
2017-02-10 23:43:50 +08:00
|
|
|
#if IS_ENABLED(CONFIG_GARP)
|
2010-10-25 05:32:36 +08:00
|
|
|
struct garp_port __rcu *garp_port;
|
2017-02-10 23:43:50 +08:00
|
|
|
#endif
|
|
|
|
#if IS_ENABLED(CONFIG_MRP)
|
2013-02-09 01:17:06 +08:00
|
|
|
struct mrp_port __rcu *mrp_port;
|
2017-02-10 23:43:50 +08:00
|
|
|
#endif
|
2022-02-23 06:04:50 +08:00
|
|
|
#if IS_ENABLED(CONFIG_NET_DROP_MONITOR)
|
|
|
|
struct dm_hw_stat_delta __rcu *dm_private;
|
|
|
|
#endif
|
2016-03-24 02:47:23 +08:00
|
|
|
struct device dev;
|
2009-10-29 22:18:21 +08:00
|
|
|
const struct attribute_group *sysfs_groups[4];
|
2014-01-17 14:23:28 +08:00
|
|
|
const struct attribute_group *sysfs_rx_queue_group;
|
2007-06-14 03:03:51 +08:00
|
|
|
|
|
|
|
const struct rtnl_link_ops *rtnl_link_ops;
|
2007-07-07 04:36:20 +08:00
|
|
|
|
[NET]: Add per-connection option to set max TSO frame size
Update: My mailer ate one of Jarek's feedback mails... Fixed the
parameter in netif_set_gso_max_size() to be u32, not u16. Fixed the
whitespace issue due to a patch import botch. Changed the types from
u32 to unsigned int to be more consistent with other variables in the
area. Also brought the patch up to the latest net-2.6.26 tree.
Update: Made gso_max_size container 32 bits, not 16. Moved the
location of gso_max_size within netdev to be less hotpath. Made more
consistent names between the sock and netdev layers, and added a
define for the max GSO size.
Update: Respun for net-2.6.26 tree.
Update: changed max_gso_frame_size and sk_gso_max_size from signed to
unsigned - thanks Stephen!
This patch adds the ability for device drivers to control the size of
the TSO frames being sent to them, per TCP connection. By setting the
netdevice's gso_max_size value, the socket layer will set the GSO
frame size based on that value. This will propogate into the TCP
layer, and send TSO's of that size to the hardware.
This can be desirable to help tune the bursty nature of TSO on a
per-adapter basis, where one may have 1 GbE and 10 GbE devices
coexisting in a system, one running multiqueue and the other not, etc.
This can also be desirable for devices that cannot support full 64 KB
TSO's, but still want to benefit from some level of segmentation
offloading.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2008-03-21 18:43:19 +08:00
|
|
|
/* for setting kernel sock attribute on TCP connection setup */
|
2022-05-14 02:33:58 +08:00
|
|
|
#define GSO_MAX_SEGS 65535u
|
2022-05-14 02:33:57 +08:00
|
|
|
#define GSO_LEGACY_MAX_SIZE 65536u
|
2022-05-14 02:33:58 +08:00
|
|
|
/* TCP minimal MSS is 8 (TCP_MIN_GSO_SIZE),
|
|
|
|
* and shinfo->gso_segs is a 16bit field.
|
|
|
|
*/
|
|
|
|
#define GSO_MAX_SIZE (8 * GSO_MAX_SEGS)
|
2022-05-14 02:33:57 +08:00
|
|
|
|
[NET]: Add per-connection option to set max TSO frame size
Update: My mailer ate one of Jarek's feedback mails... Fixed the
parameter in netif_set_gso_max_size() to be u32, not u16. Fixed the
whitespace issue due to a patch import botch. Changed the types from
u32 to unsigned int to be more consistent with other variables in the
area. Also brought the patch up to the latest net-2.6.26 tree.
Update: Made gso_max_size container 32 bits, not 16. Moved the
location of gso_max_size within netdev to be less hotpath. Made more
consistent names between the sock and netdev layers, and added a
define for the max GSO size.
Update: Respun for net-2.6.26 tree.
Update: changed max_gso_frame_size and sk_gso_max_size from signed to
unsigned - thanks Stephen!
This patch adds the ability for device drivers to control the size of
the TSO frames being sent to them, per TCP connection. By setting the
netdevice's gso_max_size value, the socket layer will set the GSO
frame size based on that value. This will propogate into the TCP
layer, and send TSO's of that size to the hardware.
This can be desirable to help tune the bursty nature of TSO on a
per-adapter basis, where one may have 1 GbE and 10 GbE devices
coexisting in a system, one running multiqueue and the other not, etc.
This can also be desirable for devices that cannot support full 64 KB
TSO's, but still want to benefit from some level of segmentation
offloading.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2008-03-21 18:43:19 +08:00
|
|
|
unsigned int gso_max_size;
|
2022-05-06 10:51:32 +08:00
|
|
|
#define TSO_LEGACY_MAX_SIZE 65536
|
|
|
|
#define TSO_MAX_SIZE UINT_MAX
|
|
|
|
unsigned int tso_max_size;
|
2012-07-30 23:57:00 +08:00
|
|
|
u16 gso_max_segs;
|
2022-05-06 10:51:32 +08:00
|
|
|
#define TSO_MAX_SEGS U16_MAX
|
|
|
|
u16 tso_max_segs;
|
2016-04-10 02:29:58 +08:00
|
|
|
|
2008-11-25 17:02:08 +08:00
|
|
|
#ifdef CONFIG_DCB
|
2009-10-05 14:01:03 +08:00
|
|
|
const struct dcbnl_rtnl_ops *dcbnl_ops;
|
2008-11-21 12:52:10 +08:00
|
|
|
#endif
|
2018-07-10 00:19:38 +08:00
|
|
|
s16 num_tc;
|
2016-03-24 02:47:23 +08:00
|
|
|
struct netdev_tc_txq tc_to_txq[TC_MAX_QUEUE];
|
|
|
|
u8 prio_tc_map[TC_BITMASK + 1];
|
2008-11-21 12:52:10 +08:00
|
|
|
|
2011-11-25 22:40:26 +08:00
|
|
|
#if IS_ENABLED(CONFIG_FCOE)
|
2009-02-28 06:06:59 +08:00
|
|
|
unsigned int fcoe_ddp_xid;
|
2011-11-22 13:10:51 +08:00
|
|
|
#endif
|
2013-12-30 00:27:11 +08:00
|
|
|
#if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
|
2011-11-22 13:10:51 +08:00
|
|
|
struct netprio_map __rcu *priomap;
|
2009-02-28 06:06:59 +08:00
|
|
|
#endif
|
2016-03-24 02:47:23 +08:00
|
|
|
struct phy_device *phydev;
|
2018-03-29 06:44:16 +08:00
|
|
|
struct sfp_bus *sfp_bus;
|
2020-05-03 13:22:19 +08:00
|
|
|
struct lock_class_key *qdisc_tx_busylock;
|
2016-03-24 02:47:23 +08:00
|
|
|
bool proto_down;
|
2018-09-25 03:58:59 +08:00
|
|
|
unsigned wol_enabled:1;
|
2021-02-09 03:34:09 +08:00
|
|
|
unsigned threaded:1;
|
2020-01-25 19:17:08 +08:00
|
|
|
|
|
|
|
struct list_head net_notifier_list;
|
2020-03-25 20:52:31 +08:00
|
|
|
|
|
|
|
#if IS_ENABLED(CONFIG_MACSEC)
|
|
|
|
/* MACsec management functions */
|
|
|
|
const struct macsec_ops *macsec_ops;
|
|
|
|
#endif
|
2020-07-10 08:42:46 +08:00
|
|
|
const struct udp_tunnel_nic_info *udp_tunnel_nic_info;
|
|
|
|
struct udp_tunnel_nic *udp_tunnel_nic;
|
2020-07-22 14:45:55 +08:00
|
|
|
|
|
|
|
/* protected by rtnl_lock */
|
|
|
|
struct bpf_xdp_entity xdp_state[__MAX_XDP_MODE];
|
2021-11-19 22:21:53 +08:00
|
|
|
|
|
|
|
u8 dev_addr_shadow[MAX_ADDR_LEN];
|
2021-12-05 12:22:14 +08:00
|
|
|
netdevice_tracker linkwatch_dev_tracker;
|
2021-12-07 09:30:30 +08:00
|
|
|
netdevice_tracker watchdog_dev_tracker;
|
2022-02-08 02:41:07 +08:00
|
|
|
netdevice_tracker dev_registered_tracker;
|
net: dev: Add hardware stats support
Offloading switch device drivers may be able to collect statistics of the
traffic taking place in the HW datapath that pertains to a certain soft
netdevice, such as VLAN. Add the necessary infrastructure to allow exposing
these statistics to the offloaded netdevice in question. The API was shaped
by the following considerations:
- Collection of HW statistics is not free: there may be a finite number of
counters, and the act of counting may have a performance impact. It is
therefore necessary to allow toggling whether HW counting should be done
for any particular SW netdevice.
- As the drivers are loaded and removed, a particular device may get
offloaded and unoffloaded again. At the same time, the statistics values
need to stay monotonic (modulo the eventual 64-bit wraparound),
increasing only to reflect traffic measured in the device.
To that end, the netdevice keeps around a lazily-allocated copy of struct
rtnl_link_stats64. Device drivers then contribute to the values kept
therein at various points. Even as the driver goes away, the struct stays
around to maintain the statistics values.
- Different HW devices may be able to count different things. The
motivation behind this patch in particular is exposure of HW counters on
Nvidia Spectrum switches, where the only practical approach to counting
traffic on offloaded soft netdevices currently is to use router interface
counters, and count L3 traffic. Correspondingly that is the statistics
suite added in this patch.
Other devices may be able to measure different kinds of traffic, and for
that reason, the APIs are built to allow uniform access to different
statistics suites.
- Because soft netdevices and offloading drivers are only loosely bound, a
netdevice uses a notifier chain to communicate with the drivers. Several
new notifiers, NETDEV_OFFLOAD_XSTATS_*, have been added to carry messages
to the offloading drivers.
- Devices can have various conditions for when a particular counter is
available. As the device is configured and reconfigured, the device
offload may become or cease being suitable for counter binding. A
netdevice can use a notifier type NETDEV_OFFLOAD_XSTATS_REPORT_USED to
ping offloading drivers and determine whether anyone currently implements
a given statistics suite. This information can then be propagated to user
space.
When the driver decides to unoffload a netdevice, it can use a
newly-added function, netdev_offload_xstats_report_delta(), to record
outstanding collected statistics, before destroying the HW counter.
This patch adds a helper, call_netdevice_notifiers_info_robust(), for
dispatching a notifier with the possibility of unwind when one of the
consumers bails. Given the wish to eventually get rid of the global
notifier block altogether, this helper only invokes the per-netns notifier
block.
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 00:31:20 +08:00
|
|
|
struct rtnl_hw_stats64 *offload_xstats_l3;
|
2022-11-03 00:02:03 +08:00
|
|
|
|
|
|
|
struct devlink_port *devlink_port;
|
2005-04-17 06:20:36 +08:00
|
|
|
};
|
2002-04-10 03:14:34 +08:00
|
|
|
#define to_net_dev(d) container_of(d, struct net_device, dev)
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2022-11-03 00:02:03 +08:00
|
|
|
/*
|
|
|
|
* Driver should use this to assign devlink port instance to a netdevice
|
|
|
|
* before it registers the netdevice. Therefore devlink_port is static
|
|
|
|
* during the netdev lifetime after it is registered.
|
|
|
|
*/
|
|
|
|
#define SET_NETDEV_DEVLINK_PORT(dev, port) \
|
|
|
|
({ \
|
|
|
|
WARN_ON((dev)->reg_state != NETREG_UNINITIALIZED); \
|
|
|
|
((dev)->devlink_port = (port)); \
|
|
|
|
})
|
|
|
|
|
2017-04-19 03:36:58 +08:00
|
|
|
static inline bool netif_elide_gro(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
if (!(dev->features & NETIF_F_GRO) || dev->xdp_prog)
|
|
|
|
return true;
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
#define NETDEV_ALIGN 32
|
|
|
|
|
2011-01-17 16:06:04 +08:00
|
|
|
static inline
|
|
|
|
int netdev_get_prio_tc_map(const struct net_device *dev, u32 prio)
|
|
|
|
{
|
|
|
|
return dev->prio_tc_map[prio & TC_BITMASK];
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline
|
|
|
|
int netdev_set_prio_tc_map(struct net_device *dev, u8 prio, u8 tc)
|
|
|
|
{
|
|
|
|
if (tc >= dev->num_tc)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
dev->prio_tc_map[prio & TC_BITMASK] = tc & TC_BITMASK;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-10-28 23:43:49 +08:00
|
|
|
int netdev_txq_to_tc(struct net_device *dev, unsigned int txq);
|
2016-10-28 23:43:20 +08:00
|
|
|
void netdev_reset_tc(struct net_device *dev);
|
|
|
|
int netdev_set_tc_queue(struct net_device *dev, u8 tc, u16 count, u16 offset);
|
|
|
|
int netdev_set_num_tc(struct net_device *dev, u8 num_tc);
|
2011-01-17 16:06:04 +08:00
|
|
|
|
|
|
|
static inline
|
|
|
|
int netdev_get_num_tc(struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->num_tc;
|
|
|
|
}
|
|
|
|
|
2020-08-26 20:54:16 +08:00
|
|
|
static inline void net_prefetch(void *p)
|
|
|
|
{
|
|
|
|
prefetch(p);
|
|
|
|
#if L1_CACHE_BYTES < 128
|
|
|
|
prefetch((u8 *)p + L1_CACHE_BYTES);
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void net_prefetchw(void *p)
|
|
|
|
{
|
|
|
|
prefetchw(p);
|
|
|
|
#if L1_CACHE_BYTES < 128
|
|
|
|
prefetchw((u8 *)p + L1_CACHE_BYTES);
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2018-07-10 00:19:38 +08:00
|
|
|
void netdev_unbind_sb_channel(struct net_device *dev,
|
|
|
|
struct net_device *sb_dev);
|
|
|
|
int netdev_bind_sb_channel_queue(struct net_device *dev,
|
|
|
|
struct net_device *sb_dev,
|
|
|
|
u8 tc, u16 count, u16 offset);
|
|
|
|
int netdev_set_sb_channel(struct net_device *dev, u16 channel);
|
|
|
|
static inline int netdev_get_sb_channel(struct net_device *dev)
|
|
|
|
{
|
|
|
|
return max_t(int, -dev->num_tc, 0);
|
|
|
|
}
|
|
|
|
|
2008-07-17 15:34:19 +08:00
|
|
|
static inline
|
|
|
|
struct netdev_queue *netdev_get_tx_queue(const struct net_device *dev,
|
|
|
|
unsigned int index)
|
|
|
|
{
|
|
|
|
return &dev->_tx[index];
|
|
|
|
}
|
|
|
|
|
2014-08-27 17:11:27 +08:00
|
|
|
static inline struct netdev_queue *skb_get_tx_queue(const struct net_device *dev,
|
|
|
|
const struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
return netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
|
|
|
|
}
|
|
|
|
|
2008-07-17 15:34:19 +08:00
|
|
|
static inline void netdev_for_each_tx_queue(struct net_device *dev,
|
|
|
|
void (*f)(struct net_device *,
|
|
|
|
struct netdev_queue *,
|
|
|
|
void *),
|
|
|
|
void *arg)
|
|
|
|
{
|
|
|
|
unsigned int i;
|
|
|
|
|
|
|
|
for (i = 0; i < dev->num_tx_queues; i++)
|
|
|
|
f(dev, &dev->_tx[i], arg);
|
|
|
|
}
|
|
|
|
|
2020-05-03 13:22:19 +08:00
|
|
|
#define netdev_lockdep_set_classes(dev) \
|
|
|
|
{ \
|
|
|
|
static struct lock_class_key qdisc_tx_busylock_key; \
|
|
|
|
static struct lock_class_key qdisc_xmit_lock_key; \
|
2020-06-09 05:53:01 +08:00
|
|
|
static struct lock_class_key dev_addr_list_lock_key; \
|
2020-05-03 13:22:19 +08:00
|
|
|
unsigned int i; \
|
|
|
|
\
|
|
|
|
(dev)->qdisc_tx_busylock = &qdisc_tx_busylock_key; \
|
2020-06-09 05:53:01 +08:00
|
|
|
lockdep_set_class(&(dev)->addr_list_lock, \
|
|
|
|
&dev_addr_list_lock_key); \
|
2020-05-03 13:22:19 +08:00
|
|
|
for (i = 0; i < (dev)->num_tx_queues; i++) \
|
|
|
|
lockdep_set_class(&(dev)->_tx[i]._xmit_lock, \
|
|
|
|
&qdisc_xmit_lock_key); \
|
|
|
|
}
|
|
|
|
|
2019-03-20 18:02:05 +08:00
|
|
|
u16 netdev_pick_tx(struct net_device *dev, struct sk_buff *skb,
|
|
|
|
struct net_device *sb_dev);
|
2019-03-20 18:02:04 +08:00
|
|
|
struct netdev_queue *netdev_core_pick_tx(struct net_device *dev,
|
|
|
|
struct sk_buff *skb,
|
|
|
|
struct net_device *sb_dev);
|
2012-09-18 04:16:31 +08:00
|
|
|
|
2016-02-26 17:45:37 +08:00
|
|
|
/* returns the headroom that the master device needs to take in account
|
|
|
|
* when forwarding to this dev
|
|
|
|
*/
|
|
|
|
static inline unsigned netdev_get_fwd_headroom(struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_PHONY_HEADROOM ? 0 : dev->needed_headroom;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void netdev_set_rx_headroom(struct net_device *dev, int new_hr)
|
|
|
|
{
|
|
|
|
if (dev->netdev_ops->ndo_set_rx_headroom)
|
|
|
|
dev->netdev_ops->ndo_set_rx_headroom(dev, new_hr);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* set the device rx headroom to the dev's default */
|
|
|
|
static inline void netdev_reset_rx_headroom(struct net_device *dev)
|
|
|
|
{
|
|
|
|
netdev_set_rx_headroom(dev, -1);
|
|
|
|
}
|
|
|
|
|
2021-02-23 15:01:26 +08:00
|
|
|
static inline void *netdev_get_ml_priv(struct net_device *dev,
|
|
|
|
enum netdev_ml_priv_type type)
|
|
|
|
{
|
|
|
|
if (dev->ml_priv_type != type)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
return dev->ml_priv;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void netdev_set_ml_priv(struct net_device *dev,
|
|
|
|
void *ml_priv,
|
|
|
|
enum netdev_ml_priv_type type)
|
|
|
|
{
|
|
|
|
WARN(dev->ml_priv_type && dev->ml_priv_type != type,
|
|
|
|
"Overwriting already set ml_priv_type (%u) with different ml_priv_type (%u)!\n",
|
|
|
|
dev->ml_priv_type, type);
|
|
|
|
WARN(!dev->ml_priv_type && dev->ml_priv,
|
|
|
|
"Overwriting already set ml_priv and ml_priv_type is ML_PRIV_NONE!\n");
|
|
|
|
|
|
|
|
dev->ml_priv = ml_priv;
|
|
|
|
dev->ml_priv_type = type;
|
|
|
|
}
|
|
|
|
|
2008-03-25 20:47:49 +08:00
|
|
|
/*
|
|
|
|
* Net namespace inlines
|
|
|
|
*/
|
|
|
|
static inline
|
|
|
|
struct net *dev_net(const struct net_device *dev)
|
|
|
|
{
|
2010-06-01 14:51:19 +08:00
|
|
|
return read_pnet(&dev->nd_net);
|
2008-03-25 20:47:49 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline
|
2008-03-26 15:48:17 +08:00
|
|
|
void dev_net_set(struct net_device *dev, struct net *net)
|
2008-03-25 20:47:49 +08:00
|
|
|
{
|
2015-03-12 12:06:44 +08:00
|
|
|
write_pnet(&dev->nd_net, net);
|
2008-03-25 20:47:49 +08:00
|
|
|
}
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netdev_priv - access network device private data
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Get network device private data
|
|
|
|
*/
|
2007-06-14 03:03:21 +08:00
|
|
|
static inline void *netdev_priv(const struct net_device *dev)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
2009-05-27 12:42:37 +08:00
|
|
|
return (char *)dev + ALIGN(sizeof(struct net_device), NETDEV_ALIGN);
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Set the sysfs physical device reference for the network logical device
|
|
|
|
* if set prior to registration will cause a symlink during initialization.
|
|
|
|
*/
|
2002-04-10 03:14:34 +08:00
|
|
|
#define SET_NETDEV_DEV(net, pdev) ((net)->dev.parent = (pdev))
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2009-09-01 05:08:19 +08:00
|
|
|
/* Set the sysfs device type for the network logical device to allow
|
2013-10-12 07:29:46 +08:00
|
|
|
* fine-grained identification of different network device types. For
|
2016-03-24 02:47:23 +08:00
|
|
|
* example Ethernet, Wireless LAN, Bluetooth, WiMAX etc.
|
2009-09-01 05:08:19 +08:00
|
|
|
*/
|
|
|
|
#define SET_NETDEV_DEVTYPE(net, devtype) ((net)->dev.type = (devtype))
|
|
|
|
|
2013-03-05 23:57:22 +08:00
|
|
|
/* Default NAPI poll() weight
|
|
|
|
* Device drivers are strongly advised to not use bigger value
|
|
|
|
*/
|
|
|
|
#define NAPI_POLL_WEIGHT 64
|
|
|
|
|
2022-05-03 07:27:03 +08:00
|
|
|
void netif_napi_add_weight(struct net_device *dev, struct napi_struct *napi,
|
|
|
|
int (*poll)(struct napi_struct *, int), int weight);
|
|
|
|
|
2007-11-01 17:21:47 +08:00
|
|
|
/**
|
2022-05-03 07:27:03 +08:00
|
|
|
* netif_napi_add() - initialize a NAPI context
|
|
|
|
* @dev: network device
|
|
|
|
* @napi: NAPI context
|
|
|
|
* @poll: polling function
|
2007-11-01 17:21:47 +08:00
|
|
|
*
|
2016-03-24 02:47:23 +08:00
|
|
|
* netif_napi_add() must be used to initialize a NAPI context prior to calling
|
|
|
|
* *any* of the other NAPI-related functions.
|
2007-11-01 17:21:47 +08:00
|
|
|
*/
|
2022-05-03 07:27:03 +08:00
|
|
|
static inline void
|
|
|
|
netif_napi_add(struct net_device *dev, struct napi_struct *napi,
|
2022-09-27 21:27:53 +08:00
|
|
|
int (*poll)(struct napi_struct *, int))
|
2022-05-03 07:27:03 +08:00
|
|
|
{
|
2022-09-27 21:27:53 +08:00
|
|
|
netif_napi_add_weight(dev, napi, poll, NAPI_POLL_WEIGHT);
|
2022-05-03 07:27:03 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline void
|
|
|
|
netif_napi_add_tx_weight(struct net_device *dev,
|
|
|
|
struct napi_struct *napi,
|
|
|
|
int (*poll)(struct napi_struct *, int),
|
|
|
|
int weight)
|
|
|
|
{
|
|
|
|
set_bit(NAPI_STATE_NO_BUSY_POLL, &napi->state);
|
|
|
|
netif_napi_add_weight(dev, napi, poll, weight);
|
|
|
|
}
|
|
|
|
|
2015-11-18 22:31:00 +08:00
|
|
|
/**
|
2022-05-03 07:27:03 +08:00
|
|
|
* netif_napi_add_tx() - initialize a NAPI context to be used for Tx only
|
|
|
|
* @dev: network device
|
|
|
|
* @napi: NAPI context
|
|
|
|
* @poll: polling function
|
2015-11-18 22:31:00 +08:00
|
|
|
*
|
|
|
|
* This variant of netif_napi_add() should be used from drivers using NAPI
|
|
|
|
* to exclusively poll a TX queue.
|
|
|
|
* This will avoid we add it into napi_hash[], thus polluting this hash table.
|
|
|
|
*/
|
2022-05-03 07:27:03 +08:00
|
|
|
static inline void netif_napi_add_tx(struct net_device *dev,
|
2015-11-18 22:31:00 +08:00
|
|
|
struct napi_struct *napi,
|
2022-05-03 07:27:03 +08:00
|
|
|
int (*poll)(struct napi_struct *, int))
|
2015-11-18 22:31:00 +08:00
|
|
|
{
|
2022-05-03 07:27:03 +08:00
|
|
|
netif_napi_add_tx_weight(dev, napi, poll, NAPI_POLL_WEIGHT);
|
2015-11-18 22:31:00 +08:00
|
|
|
}
|
|
|
|
|
2020-09-10 01:37:51 +08:00
|
|
|
/**
|
|
|
|
* __netif_napi_del - remove a NAPI context
|
|
|
|
* @napi: NAPI context
|
|
|
|
*
|
|
|
|
* Warning: caller must observe RCU grace period before freeing memory
|
|
|
|
* containing @napi. Drivers might want to call this helper to combine
|
|
|
|
* all the needed RCU grace periods into a single one.
|
|
|
|
*/
|
|
|
|
void __netif_napi_del(struct napi_struct *napi);
|
|
|
|
|
2008-07-09 06:13:05 +08:00
|
|
|
/**
|
2016-03-24 02:47:23 +08:00
|
|
|
* netif_napi_del - remove a NAPI context
|
|
|
|
* @napi: NAPI context
|
2008-07-09 06:13:05 +08:00
|
|
|
*
|
2016-03-24 02:47:23 +08:00
|
|
|
* netif_napi_del() removes a NAPI context from the network device NAPI list
|
2008-07-09 06:13:05 +08:00
|
|
|
*/
|
2020-09-10 01:37:51 +08:00
|
|
|
static inline void netif_napi_del(struct napi_struct *napi)
|
|
|
|
{
|
|
|
|
__netif_napi_del(napi);
|
|
|
|
synchronize_net();
|
|
|
|
}
|
net: Add Generic Receive Offload infrastructure
This patch adds the top-level GRO (Generic Receive Offload) infrastructure.
This is pretty similar to LRO except that this is protocol-independent.
Instead of holding packets in an lro_mgr structure, they're now held in
napi_struct.
For drivers that intend to use this, they can set the NETIF_F_GRO bit and
call napi_gro_receive instead of netif_receive_skb or just call netif_rx.
The latter will call napi_receive_skb automatically. When napi_gro_receive
is used, the driver must either call napi_complete/napi_rx_complete, or
call napi_gro_flush in softirq context if the driver uses the primitives
__napi_complete/__napi_rx_complete.
Protocols will set the gro_receive and gro_complete function pointers in
order to participate in this scheme.
In addition to the packet, gro_receive will get a list of currently held
packets. Each packet in the list has a same_flow field which is non-zero
if it is a potential match for the new packet. For each packet that may
match, they also have a flush field which is non-zero if the held packet
must not be merged with the new packet.
Once gro_receive has determined that the new skb matches a held packet,
the held packet may be processed immediately if the new skb cannot be
merged with it. In this case gro_receive should return the pointer to
the existing skb in gro_list. Otherwise the new skb should be merged into
the existing packet and NULL should be returned, unless the new skb makes
it impossible for any further merges to be made (e.g., FIN packet) where
the merged skb should be returned.
Whenever the skb is merged into an existing entry, the gro_receive
function should set NAPI_GRO_CB(skb)->same_flow. Note that if an skb
merely matches an existing entry but can't be merged with it, then
this shouldn't be set.
If gro_receive finds it pointless to hold the new skb for future merging,
it should set NAPI_GRO_CB(skb)->flush.
Held packets will be flushed by napi_gro_flush which is called by
napi_complete and napi_rx_complete.
Currently held packets are stored in a singly liked list just like LRO.
The list is limited to a maximum of 8 entries. In future, this may be
expanded to use a hash table to allow more flows to be held for merging.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
2008-12-16 15:38:52 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
struct packet_type {
|
2005-08-10 10:34:12 +08:00
|
|
|
__be16 type; /* This is really htons(ether_type). */
|
2018-09-03 22:23:36 +08:00
|
|
|
bool ignore_outgoing;
|
2005-08-10 10:34:12 +08:00
|
|
|
struct net_device *dev; /* NULL is wildcarded here */
|
2021-12-14 23:09:33 +08:00
|
|
|
netdevice_tracker dev_tracker;
|
2005-08-10 10:34:12 +08:00
|
|
|
int (*func) (struct sk_buff *,
|
|
|
|
struct net_device *,
|
|
|
|
struct packet_type *,
|
|
|
|
struct net_device *);
|
2018-07-02 23:14:12 +08:00
|
|
|
void (*list_func) (struct list_head *,
|
|
|
|
struct packet_type *,
|
|
|
|
struct net_device *);
|
2012-08-17 06:02:58 +08:00
|
|
|
bool (*id_match)(struct packet_type *ptype,
|
|
|
|
struct sock *sk);
|
2022-01-19 03:20:13 +08:00
|
|
|
struct net *af_packet_net;
|
2005-04-17 06:20:36 +08:00
|
|
|
void *af_packet_priv;
|
|
|
|
struct list_head list;
|
|
|
|
};
|
|
|
|
|
2012-11-15 16:49:23 +08:00
|
|
|
struct offload_callbacks {
|
2006-06-28 04:22:38 +08:00
|
|
|
struct sk_buff *(*gso_segment)(struct sk_buff *skb,
|
2011-11-15 23:29:55 +08:00
|
|
|
netdev_features_t features);
|
2018-06-24 13:13:49 +08:00
|
|
|
struct sk_buff *(*gro_receive)(struct list_head *head,
|
|
|
|
struct sk_buff *skb);
|
net-gro: Prepare GRO stack for the upcoming tunneling support
This patch modifies the GRO stack to avoid the use of "network_header"
and associated macros like ip_hdr() and ipv6_hdr() in order to allow
an arbitary number of IP hdrs (v4 or v6) to be used in the
encapsulation chain. This lays the foundation for various IP
tunneling support (IP-in-IP, GRE, VXLAN, SIT,...) to be added later.
With this patch, the GRO stack traversing now is mostly based on
skb_gro_offset rather than special hdr offsets saved in skb (e.g.,
skb->network_header). As a result all but the top layer (i.e., the
the transport layer) must have hdrs of the same length in order for
a pkt to be considered for aggregation. Therefore when adding a new
encap layer (e.g., for tunneling), one must check and skip flows
(e.g., by setting NAPI_GRO_CB(p)->same_flow to 0) that have a
different hdr length.
Note that unlike the network header, the transport header can and
will continue to be set by the GRO code since there will be at
most one "transport layer" in the encap chain.
Signed-off-by: H.K. Jerry Chu <hkchu@google.com>
Suggested-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2013-12-12 12:53:45 +08:00
|
|
|
int (*gro_complete)(struct sk_buff *skb, int nhoff);
|
2012-11-15 16:49:23 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
struct packet_offload {
|
|
|
|
__be16 type; /* This is really htons(ether_type). */
|
2015-06-02 05:56:09 +08:00
|
|
|
u16 priority;
|
2012-11-15 16:49:23 +08:00
|
|
|
struct offload_callbacks callbacks;
|
|
|
|
struct list_head list;
|
2005-04-17 06:20:36 +08:00
|
|
|
};
|
|
|
|
|
2016-03-24 02:47:23 +08:00
|
|
|
/* often modified stats are per-CPU, other are shared (netdev->stats) */
|
2014-01-04 13:57:59 +08:00
|
|
|
struct pcpu_sw_netstats {
|
2022-06-08 23:46:37 +08:00
|
|
|
u64_stats_t rx_packets;
|
|
|
|
u64_stats_t rx_bytes;
|
|
|
|
u64_stats_t tx_packets;
|
|
|
|
u64_stats_t tx_bytes;
|
2014-01-04 13:57:59 +08:00
|
|
|
struct u64_stats_sync syncp;
|
2018-11-16 23:24:24 +08:00
|
|
|
} __aligned(4 * sizeof(u64));
|
2018-09-14 16:00:51 +08:00
|
|
|
|
|
|
|
struct pcpu_lstats {
|
2019-11-08 08:27:22 +08:00
|
|
|
u64_stats_t packets;
|
|
|
|
u64_stats_t bytes;
|
2018-09-14 16:00:51 +08:00
|
|
|
struct u64_stats_sync syncp;
|
2018-11-16 23:24:24 +08:00
|
|
|
} __aligned(2 * sizeof(u64));
|
2014-01-04 13:57:59 +08:00
|
|
|
|
2019-11-08 08:27:14 +08:00
|
|
|
void dev_lstats_read(struct net_device *dev, u64 *packets, u64 *bytes);
|
|
|
|
|
2020-10-06 04:34:18 +08:00
|
|
|
static inline void dev_sw_netstats_rx_add(struct net_device *dev, unsigned int len)
|
|
|
|
{
|
|
|
|
struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
|
|
|
|
|
|
|
|
u64_stats_update_begin(&tstats->syncp);
|
2022-06-08 23:46:37 +08:00
|
|
|
u64_stats_add(&tstats->rx_bytes, len);
|
|
|
|
u64_stats_inc(&tstats->rx_packets);
|
2020-10-06 04:34:18 +08:00
|
|
|
u64_stats_update_end(&tstats->syncp);
|
|
|
|
}
|
|
|
|
|
2020-10-30 01:29:59 +08:00
|
|
|
static inline void dev_sw_netstats_tx_add(struct net_device *dev,
|
|
|
|
unsigned int packets,
|
|
|
|
unsigned int len)
|
|
|
|
{
|
|
|
|
struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
|
|
|
|
|
|
|
|
u64_stats_update_begin(&tstats->syncp);
|
2022-06-08 23:46:37 +08:00
|
|
|
u64_stats_add(&tstats->tx_bytes, len);
|
|
|
|
u64_stats_add(&tstats->tx_packets, packets);
|
2020-10-30 01:29:59 +08:00
|
|
|
u64_stats_update_end(&tstats->syncp);
|
|
|
|
}
|
|
|
|
|
2019-11-08 08:27:15 +08:00
|
|
|
static inline void dev_lstats_add(struct net_device *dev, unsigned int len)
|
|
|
|
{
|
|
|
|
struct pcpu_lstats *lstats = this_cpu_ptr(dev->lstats);
|
|
|
|
|
|
|
|
u64_stats_update_begin(&lstats->syncp);
|
2019-11-08 08:27:22 +08:00
|
|
|
u64_stats_add(&lstats->bytes, len);
|
|
|
|
u64_stats_inc(&lstats->packets);
|
2019-11-08 08:27:15 +08:00
|
|
|
u64_stats_update_end(&lstats->syncp);
|
|
|
|
}
|
|
|
|
|
2015-11-10 21:31:18 +08:00
|
|
|
#define __netdev_alloc_pcpu_stats(type, gfp) \
|
|
|
|
({ \
|
|
|
|
typeof(type) __percpu *pcpu_stats = alloc_percpu_gfp(type, gfp);\
|
|
|
|
if (pcpu_stats) { \
|
|
|
|
int __cpu; \
|
|
|
|
for_each_possible_cpu(__cpu) { \
|
|
|
|
typeof(type) *stat; \
|
|
|
|
stat = per_cpu_ptr(pcpu_stats, __cpu); \
|
|
|
|
u64_stats_init(&stat->syncp); \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
pcpu_stats; \
|
2014-02-14 03:46:28 +08:00
|
|
|
})
|
|
|
|
|
2015-11-10 21:31:18 +08:00
|
|
|
#define netdev_alloc_pcpu_stats(type) \
|
2015-12-05 20:58:11 +08:00
|
|
|
__netdev_alloc_pcpu_stats(type, GFP_KERNEL)
|
2015-11-10 21:31:18 +08:00
|
|
|
|
2020-10-30 01:31:21 +08:00
|
|
|
#define devm_netdev_alloc_pcpu_stats(dev, type) \
|
|
|
|
({ \
|
|
|
|
typeof(type) __percpu *pcpu_stats = devm_alloc_percpu(dev, type);\
|
|
|
|
if (pcpu_stats) { \
|
|
|
|
int __cpu; \
|
|
|
|
for_each_possible_cpu(__cpu) { \
|
|
|
|
typeof(type) *stat; \
|
|
|
|
stat = per_cpu_ptr(pcpu_stats, __cpu); \
|
|
|
|
u64_stats_init(&stat->syncp); \
|
|
|
|
} \
|
|
|
|
} \
|
|
|
|
pcpu_stats; \
|
|
|
|
})
|
|
|
|
|
2015-12-03 19:12:12 +08:00
|
|
|
enum netdev_lag_tx_type {
|
|
|
|
NETDEV_LAG_TX_TYPE_UNKNOWN,
|
|
|
|
NETDEV_LAG_TX_TYPE_RANDOM,
|
|
|
|
NETDEV_LAG_TX_TYPE_BROADCAST,
|
|
|
|
NETDEV_LAG_TX_TYPE_ROUNDROBIN,
|
|
|
|
NETDEV_LAG_TX_TYPE_ACTIVEBACKUP,
|
|
|
|
NETDEV_LAG_TX_TYPE_HASH,
|
|
|
|
};
|
|
|
|
|
2018-05-24 10:22:52 +08:00
|
|
|
enum netdev_lag_hash {
|
|
|
|
NETDEV_LAG_HASH_NONE,
|
|
|
|
NETDEV_LAG_HASH_L2,
|
|
|
|
NETDEV_LAG_HASH_L34,
|
|
|
|
NETDEV_LAG_HASH_L23,
|
|
|
|
NETDEV_LAG_HASH_E23,
|
|
|
|
NETDEV_LAG_HASH_E34,
|
bonding: add a vlan+srcmac tx hashing option
This comes from an end-user request, where they're running multiple VMs on
hosts with bonded interfaces connected to some interest switch topologies,
where 802.3ad isn't an option. They're currently running a proprietary
solution that effectively achieves load-balancing of VMs and bandwidth
utilization improvements with a similar form of transmission algorithm.
Basically, each VM has it's own vlan, so it always sends its traffic out
the same interface, unless that interface fails. Traffic gets split
between the interfaces, maintaining a consistent path, with failover still
available if an interface goes down.
Unlike bond_eth_hash(), this hash function is using the full source MAC
address instead of just the last byte, as there are so few components to
the hash, and in the no-vlan case, we would be returning just the last
byte of the source MAC as the hash value. It's entirely possible to have
two NICs in a bond with the same last byte of their MAC, but not the same
MAC, so this adjustment should guarantee distinct hashes in all cases.
This has been rudimetarily tested to provide similar results to the
proprietary solution it is aiming to replace. A patch for iproute2 is also
posted, to properly support the new mode there as well.
Cc: Jay Vosburgh <j.vosburgh@gmail.com>
Cc: Veaceslav Falico <vfalico@gmail.com>
Cc: Andy Gospodarek <andy@greyhouse.net>
Cc: Thomas Davis <tadavis@lbl.gov>
Signed-off-by: Jarod Wilson <jarod@redhat.com>
Link: https://lore.kernel.org/r/20210119010927.1191922-1-jarod@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-19 09:09:27 +08:00
|
|
|
NETDEV_LAG_HASH_VLAN_SRCMAC,
|
2018-05-24 10:22:52 +08:00
|
|
|
NETDEV_LAG_HASH_UNKNOWN,
|
|
|
|
};
|
|
|
|
|
2015-12-03 19:12:12 +08:00
|
|
|
struct netdev_lag_upper_info {
|
|
|
|
enum netdev_lag_tx_type tx_type;
|
2018-05-24 10:22:52 +08:00
|
|
|
enum netdev_lag_hash hash_type;
|
2015-12-03 19:12:12 +08:00
|
|
|
};
|
|
|
|
|
2015-12-03 19:12:16 +08:00
|
|
|
struct netdev_lag_lower_state_info {
|
|
|
|
u8 link_up : 1,
|
|
|
|
tx_enabled : 1;
|
|
|
|
};
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
#include <linux/notifier.h>
|
|
|
|
|
2018-03-24 00:47:19 +08:00
|
|
|
/* netdevice notifier chain. Please remember to update netdev_cmd_to_name()
|
|
|
|
* and the rtnetlink notification exclusion list in rtnetlink_event() when
|
|
|
|
* adding new types.
|
2011-07-26 08:13:09 +08:00
|
|
|
*/
|
2018-03-24 00:47:19 +08:00
|
|
|
enum netdev_cmd {
|
|
|
|
NETDEV_UP = 1, /* For now you can't veto a device up/down */
|
|
|
|
NETDEV_DOWN,
|
|
|
|
NETDEV_REBOOT, /* Tell a protocol stack a network interface
|
2011-07-26 08:13:09 +08:00
|
|
|
detected a hardware crash and restarted
|
|
|
|
- we can use this eg to kick tcp sessions
|
|
|
|
once done */
|
2018-03-24 00:47:19 +08:00
|
|
|
NETDEV_CHANGE, /* Notify device state change */
|
|
|
|
NETDEV_REGISTER,
|
|
|
|
NETDEV_UNREGISTER,
|
|
|
|
NETDEV_CHANGEMTU, /* notify after mtu change happened */
|
2018-12-13 19:54:33 +08:00
|
|
|
NETDEV_CHANGEADDR, /* notify after the address change */
|
|
|
|
NETDEV_PRE_CHANGEADDR, /* notify before the address change */
|
2018-03-24 00:47:19 +08:00
|
|
|
NETDEV_GOING_DOWN,
|
|
|
|
NETDEV_CHANGENAME,
|
|
|
|
NETDEV_FEAT_CHANGE,
|
|
|
|
NETDEV_BONDING_FAILOVER,
|
|
|
|
NETDEV_PRE_UP,
|
|
|
|
NETDEV_PRE_TYPE_CHANGE,
|
|
|
|
NETDEV_POST_TYPE_CHANGE,
|
|
|
|
NETDEV_POST_INIT,
|
2022-11-03 00:02:03 +08:00
|
|
|
NETDEV_PRE_UNINIT,
|
2018-03-24 00:47:19 +08:00
|
|
|
NETDEV_RELEASE,
|
|
|
|
NETDEV_NOTIFY_PEERS,
|
|
|
|
NETDEV_JOIN,
|
|
|
|
NETDEV_CHANGEUPPER,
|
|
|
|
NETDEV_RESEND_IGMP,
|
|
|
|
NETDEV_PRECHANGEMTU, /* notify before mtu change happened */
|
|
|
|
NETDEV_CHANGEINFODATA,
|
|
|
|
NETDEV_BONDING_INFO,
|
|
|
|
NETDEV_PRECHANGEUPPER,
|
|
|
|
NETDEV_CHANGELOWERSTATE,
|
|
|
|
NETDEV_UDP_TUNNEL_PUSH_INFO,
|
|
|
|
NETDEV_UDP_TUNNEL_DROP_INFO,
|
|
|
|
NETDEV_CHANGE_TX_QUEUE_LEN,
|
2018-03-28 22:46:54 +08:00
|
|
|
NETDEV_CVLAN_FILTER_PUSH_INFO,
|
|
|
|
NETDEV_CVLAN_FILTER_DROP_INFO,
|
|
|
|
NETDEV_SVLAN_FILTER_PUSH_INFO,
|
|
|
|
NETDEV_SVLAN_FILTER_DROP_INFO,
|
net: dev: Add hardware stats support
Offloading switch device drivers may be able to collect statistics of the
traffic taking place in the HW datapath that pertains to a certain soft
netdevice, such as VLAN. Add the necessary infrastructure to allow exposing
these statistics to the offloaded netdevice in question. The API was shaped
by the following considerations:
- Collection of HW statistics is not free: there may be a finite number of
counters, and the act of counting may have a performance impact. It is
therefore necessary to allow toggling whether HW counting should be done
for any particular SW netdevice.
- As the drivers are loaded and removed, a particular device may get
offloaded and unoffloaded again. At the same time, the statistics values
need to stay monotonic (modulo the eventual 64-bit wraparound),
increasing only to reflect traffic measured in the device.
To that end, the netdevice keeps around a lazily-allocated copy of struct
rtnl_link_stats64. Device drivers then contribute to the values kept
therein at various points. Even as the driver goes away, the struct stays
around to maintain the statistics values.
- Different HW devices may be able to count different things. The
motivation behind this patch in particular is exposure of HW counters on
Nvidia Spectrum switches, where the only practical approach to counting
traffic on offloaded soft netdevices currently is to use router interface
counters, and count L3 traffic. Correspondingly that is the statistics
suite added in this patch.
Other devices may be able to measure different kinds of traffic, and for
that reason, the APIs are built to allow uniform access to different
statistics suites.
- Because soft netdevices and offloading drivers are only loosely bound, a
netdevice uses a notifier chain to communicate with the drivers. Several
new notifiers, NETDEV_OFFLOAD_XSTATS_*, have been added to carry messages
to the offloading drivers.
- Devices can have various conditions for when a particular counter is
available. As the device is configured and reconfigured, the device
offload may become or cease being suitable for counter binding. A
netdevice can use a notifier type NETDEV_OFFLOAD_XSTATS_REPORT_USED to
ping offloading drivers and determine whether anyone currently implements
a given statistics suite. This information can then be propagated to user
space.
When the driver decides to unoffload a netdevice, it can use a
newly-added function, netdev_offload_xstats_report_delta(), to record
outstanding collected statistics, before destroying the HW counter.
This patch adds a helper, call_netdevice_notifiers_info_robust(), for
dispatching a notifier with the possibility of unwind when one of the
consumers bails. Given the wish to eventually get rid of the global
notifier block altogether, this helper only invokes the per-netns notifier
block.
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 00:31:20 +08:00
|
|
|
NETDEV_OFFLOAD_XSTATS_ENABLE,
|
|
|
|
NETDEV_OFFLOAD_XSTATS_DISABLE,
|
|
|
|
NETDEV_OFFLOAD_XSTATS_REPORT_USED,
|
|
|
|
NETDEV_OFFLOAD_XSTATS_REPORT_DELTA,
|
2018-03-24 00:47:19 +08:00
|
|
|
};
|
|
|
|
const char *netdev_cmd_to_name(enum netdev_cmd cmd);
|
2011-07-26 08:13:09 +08:00
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
int register_netdevice_notifier(struct notifier_block *nb);
|
|
|
|
int unregister_netdevice_notifier(struct notifier_block *nb);
|
2019-09-30 16:15:10 +08:00
|
|
|
int register_netdevice_notifier_net(struct net *net, struct notifier_block *nb);
|
|
|
|
int unregister_netdevice_notifier_net(struct net *net,
|
|
|
|
struct notifier_block *nb);
|
2022-11-08 21:22:06 +08:00
|
|
|
void move_netdevice_notifier_net(struct net *src_net, struct net *dst_net,
|
|
|
|
struct notifier_block *nb);
|
2020-01-25 19:17:08 +08:00
|
|
|
int register_netdevice_notifier_dev_net(struct net_device *dev,
|
|
|
|
struct notifier_block *nb,
|
|
|
|
struct netdev_net_notifier *nn);
|
|
|
|
int unregister_netdevice_notifier_dev_net(struct net_device *dev,
|
|
|
|
struct notifier_block *nb,
|
|
|
|
struct netdev_net_notifier *nn);
|
2013-05-28 09:30:21 +08:00
|
|
|
|
|
|
|
struct netdev_notifier_info {
|
2017-10-05 08:48:45 +08:00
|
|
|
struct net_device *dev;
|
|
|
|
struct netlink_ext_ack *extack;
|
2013-05-28 09:30:21 +08:00
|
|
|
};
|
|
|
|
|
net: ipv4: update fnhe_pmtu when first hop's MTU changes
Since commit 5aad1de5ea2c ("ipv4: use separate genid for next hop
exceptions"), exceptions get deprecated separately from cached
routes. In particular, administrative changes don't clear PMTU anymore.
As Stefano described in commit e9fa1495d738 ("ipv6: Reflect MTU changes
on PMTU of exceptions for MTU-less routes"), the PMTU discovered before
the local MTU change can become stale:
- if the local MTU is now lower than the PMTU, that PMTU is now
incorrect
- if the local MTU was the lowest value in the path, and is increased,
we might discover a higher PMTU
Similarly to what commit e9fa1495d738 did for IPv6, update PMTU in those
cases.
If the exception was locked, the discovered PMTU was smaller than the
minimal accepted PMTU. In that case, if the new local MTU is smaller
than the current PMTU, let PMTU discovery figure out if locking of the
exception is still needed.
To do this, we need to know the old link MTU in the NETDEV_CHANGEMTU
notifier. By the time the notifier is called, dev->mtu has been
changed. This patch adds the old MTU as additional information in the
notifier structure, and a new call_netdevice_notifiers_u32() function.
Fixes: 5aad1de5ea2c ("ipv4: use separate genid for next hop exceptions")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-10-09 23:48:14 +08:00
|
|
|
struct netdev_notifier_info_ext {
|
|
|
|
struct netdev_notifier_info info; /* must be first */
|
|
|
|
union {
|
|
|
|
u32 mtu;
|
|
|
|
} ext;
|
|
|
|
};
|
|
|
|
|
2013-05-28 09:30:22 +08:00
|
|
|
struct netdev_notifier_change_info {
|
|
|
|
struct netdev_notifier_info info; /* must be first */
|
|
|
|
unsigned int flags_changed;
|
|
|
|
};
|
|
|
|
|
2015-08-27 15:31:18 +08:00
|
|
|
struct netdev_notifier_changeupper_info {
|
|
|
|
struct netdev_notifier_info info; /* must be first */
|
|
|
|
struct net_device *upper_dev; /* new upper dev */
|
|
|
|
bool master; /* is upper dev master */
|
2016-03-24 02:47:23 +08:00
|
|
|
bool linking; /* is the notification for link or unlink */
|
2015-12-03 19:12:11 +08:00
|
|
|
void *upper_info; /* upper dev info */
|
2015-08-27 15:31:18 +08:00
|
|
|
};
|
|
|
|
|
2015-12-03 19:12:15 +08:00
|
|
|
struct netdev_notifier_changelowerstate_info {
|
|
|
|
struct netdev_notifier_info info; /* must be first */
|
|
|
|
void *lower_state_info; /* is lower dev state */
|
|
|
|
};
|
|
|
|
|
2018-12-13 19:54:33 +08:00
|
|
|
struct netdev_notifier_pre_changeaddr_info {
|
|
|
|
struct netdev_notifier_info info; /* must be first */
|
|
|
|
const unsigned char *dev_addr;
|
|
|
|
};
|
|
|
|
|
net: dev: Add hardware stats support
Offloading switch device drivers may be able to collect statistics of the
traffic taking place in the HW datapath that pertains to a certain soft
netdevice, such as VLAN. Add the necessary infrastructure to allow exposing
these statistics to the offloaded netdevice in question. The API was shaped
by the following considerations:
- Collection of HW statistics is not free: there may be a finite number of
counters, and the act of counting may have a performance impact. It is
therefore necessary to allow toggling whether HW counting should be done
for any particular SW netdevice.
- As the drivers are loaded and removed, a particular device may get
offloaded and unoffloaded again. At the same time, the statistics values
need to stay monotonic (modulo the eventual 64-bit wraparound),
increasing only to reflect traffic measured in the device.
To that end, the netdevice keeps around a lazily-allocated copy of struct
rtnl_link_stats64. Device drivers then contribute to the values kept
therein at various points. Even as the driver goes away, the struct stays
around to maintain the statistics values.
- Different HW devices may be able to count different things. The
motivation behind this patch in particular is exposure of HW counters on
Nvidia Spectrum switches, where the only practical approach to counting
traffic on offloaded soft netdevices currently is to use router interface
counters, and count L3 traffic. Correspondingly that is the statistics
suite added in this patch.
Other devices may be able to measure different kinds of traffic, and for
that reason, the APIs are built to allow uniform access to different
statistics suites.
- Because soft netdevices and offloading drivers are only loosely bound, a
netdevice uses a notifier chain to communicate with the drivers. Several
new notifiers, NETDEV_OFFLOAD_XSTATS_*, have been added to carry messages
to the offloading drivers.
- Devices can have various conditions for when a particular counter is
available. As the device is configured and reconfigured, the device
offload may become or cease being suitable for counter binding. A
netdevice can use a notifier type NETDEV_OFFLOAD_XSTATS_REPORT_USED to
ping offloading drivers and determine whether anyone currently implements
a given statistics suite. This information can then be propagated to user
space.
When the driver decides to unoffload a netdevice, it can use a
newly-added function, netdev_offload_xstats_report_delta(), to record
outstanding collected statistics, before destroying the HW counter.
This patch adds a helper, call_netdevice_notifiers_info_robust(), for
dispatching a notifier with the possibility of unwind when one of the
consumers bails. Given the wish to eventually get rid of the global
notifier block altogether, this helper only invokes the per-netns notifier
block.
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 00:31:20 +08:00
|
|
|
enum netdev_offload_xstats_type {
|
|
|
|
NETDEV_OFFLOAD_XSTATS_TYPE_L3 = 1,
|
|
|
|
};
|
|
|
|
|
|
|
|
struct netdev_notifier_offload_xstats_info {
|
|
|
|
struct netdev_notifier_info info; /* must be first */
|
|
|
|
enum netdev_offload_xstats_type type;
|
|
|
|
|
|
|
|
union {
|
|
|
|
/* NETDEV_OFFLOAD_XSTATS_REPORT_DELTA */
|
|
|
|
struct netdev_notifier_offload_xstats_rd *report_delta;
|
|
|
|
/* NETDEV_OFFLOAD_XSTATS_REPORT_USED */
|
|
|
|
struct netdev_notifier_offload_xstats_ru *report_used;
|
|
|
|
};
|
|
|
|
};
|
|
|
|
|
|
|
|
int netdev_offload_xstats_enable(struct net_device *dev,
|
|
|
|
enum netdev_offload_xstats_type type,
|
|
|
|
struct netlink_ext_ack *extack);
|
|
|
|
int netdev_offload_xstats_disable(struct net_device *dev,
|
|
|
|
enum netdev_offload_xstats_type type);
|
|
|
|
bool netdev_offload_xstats_enabled(const struct net_device *dev,
|
|
|
|
enum netdev_offload_xstats_type type);
|
|
|
|
int netdev_offload_xstats_get(struct net_device *dev,
|
|
|
|
enum netdev_offload_xstats_type type,
|
|
|
|
struct rtnl_hw_stats64 *stats, bool *used,
|
|
|
|
struct netlink_ext_ack *extack);
|
|
|
|
void
|
|
|
|
netdev_offload_xstats_report_delta(struct netdev_notifier_offload_xstats_rd *rd,
|
|
|
|
const struct rtnl_hw_stats64 *stats);
|
|
|
|
void
|
|
|
|
netdev_offload_xstats_report_used(struct netdev_notifier_offload_xstats_ru *ru);
|
|
|
|
void netdev_offload_xstats_push_delta(struct net_device *dev,
|
|
|
|
enum netdev_offload_xstats_type type,
|
|
|
|
const struct rtnl_hw_stats64 *stats);
|
|
|
|
|
2013-05-29 11:30:50 +08:00
|
|
|
static inline void netdev_notifier_info_init(struct netdev_notifier_info *info,
|
|
|
|
struct net_device *dev)
|
|
|
|
{
|
|
|
|
info->dev = dev;
|
2017-10-05 08:48:45 +08:00
|
|
|
info->extack = NULL;
|
2013-05-29 11:30:50 +08:00
|
|
|
}
|
|
|
|
|
2013-05-28 09:30:21 +08:00
|
|
|
static inline struct net_device *
|
|
|
|
netdev_notifier_info_to_dev(const struct netdev_notifier_info *info)
|
|
|
|
{
|
|
|
|
return info->dev;
|
|
|
|
}
|
|
|
|
|
2017-10-05 08:48:45 +08:00
|
|
|
static inline struct netlink_ext_ack *
|
|
|
|
netdev_notifier_info_to_extack(const struct netdev_notifier_info *info)
|
|
|
|
{
|
|
|
|
return info->extack;
|
|
|
|
}
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
int call_netdevice_notifiers(unsigned long val, struct net_device *dev);
|
2011-07-26 08:13:09 +08:00
|
|
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
extern rwlock_t dev_base_lock; /* Device list lock */
|
|
|
|
|
2007-09-18 02:56:21 +08:00
|
|
|
#define for_each_netdev(net, d) \
|
|
|
|
list_for_each_entry(d, &(net)->dev_base_head, dev_list)
|
2009-11-30 06:25:26 +08:00
|
|
|
#define for_each_netdev_reverse(net, d) \
|
|
|
|
list_for_each_entry_reverse(d, &(net)->dev_base_head, dev_list)
|
2009-11-04 21:43:23 +08:00
|
|
|
#define for_each_netdev_rcu(net, d) \
|
|
|
|
list_for_each_entry_rcu(d, &(net)->dev_base_head, dev_list)
|
2007-09-18 02:56:21 +08:00
|
|
|
#define for_each_netdev_safe(net, d, n) \
|
|
|
|
list_for_each_entry_safe(d, n, &(net)->dev_base_head, dev_list)
|
|
|
|
#define for_each_netdev_continue(net, d) \
|
|
|
|
list_for_each_entry_continue(d, &(net)->dev_base_head, dev_list)
|
2019-09-30 16:15:09 +08:00
|
|
|
#define for_each_netdev_continue_reverse(net, d) \
|
|
|
|
list_for_each_entry_continue_reverse(d, &(net)->dev_base_head, \
|
|
|
|
dev_list)
|
2009-11-10 15:54:47 +08:00
|
|
|
#define for_each_netdev_continue_rcu(net, d) \
|
|
|
|
list_for_each_entry_continue_rcu(d, &(net)->dev_base_head, dev_list)
|
2013-03-12 10:49:01 +08:00
|
|
|
#define for_each_netdev_in_bond_rcu(bond, slave) \
|
|
|
|
for_each_netdev_rcu(&init_net, slave) \
|
2015-01-14 15:52:35 +08:00
|
|
|
if (netdev_master_upper_dev_get_rcu(slave) == (bond))
|
2007-09-18 02:56:21 +08:00
|
|
|
#define net_device_entry(lh) list_entry(lh, struct net_device, dev_list)
|
2007-05-04 06:13:45 +08:00
|
|
|
|
2007-09-12 20:57:09 +08:00
|
|
|
static inline struct net_device *next_net_device(struct net_device *dev)
|
|
|
|
{
|
|
|
|
struct list_head *lh;
|
|
|
|
struct net *net;
|
|
|
|
|
2008-03-25 20:47:49 +08:00
|
|
|
net = dev_net(dev);
|
2007-09-12 20:57:09 +08:00
|
|
|
lh = dev->dev_list.next;
|
|
|
|
return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
|
|
|
|
}
|
|
|
|
|
2009-11-12 01:34:30 +08:00
|
|
|
static inline struct net_device *next_net_device_rcu(struct net_device *dev)
|
|
|
|
{
|
|
|
|
struct list_head *lh;
|
|
|
|
struct net *net;
|
|
|
|
|
|
|
|
net = dev_net(dev);
|
2011-01-27 02:08:02 +08:00
|
|
|
lh = rcu_dereference(list_next_rcu(&dev->dev_list));
|
2009-11-12 01:34:30 +08:00
|
|
|
return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
|
|
|
|
}
|
|
|
|
|
2007-09-12 20:57:09 +08:00
|
|
|
static inline struct net_device *first_net_device(struct net *net)
|
|
|
|
{
|
|
|
|
return list_empty(&net->dev_base_head) ? NULL :
|
|
|
|
net_device_entry(net->dev_base_head.next);
|
|
|
|
}
|
2007-05-04 06:13:45 +08:00
|
|
|
|
2011-01-27 02:08:02 +08:00
|
|
|
static inline struct net_device *first_net_device_rcu(struct net *net)
|
|
|
|
{
|
|
|
|
struct list_head *lh = rcu_dereference(list_next_rcu(&net->dev_base_head));
|
|
|
|
|
|
|
|
return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
|
|
|
|
}
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
int netdev_boot_setup_check(struct net_device *dev);
|
|
|
|
struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
|
|
|
|
const char *hwaddr);
|
|
|
|
struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type);
|
|
|
|
void dev_add_pack(struct packet_type *pt);
|
|
|
|
void dev_remove_pack(struct packet_type *pt);
|
|
|
|
void __dev_remove_pack(struct packet_type *pt);
|
|
|
|
void dev_add_offload(struct packet_offload *po);
|
|
|
|
void dev_remove_offload(struct packet_offload *po);
|
|
|
|
|
2015-04-02 23:07:00 +08:00
|
|
|
int dev_get_iflink(const struct net_device *dev);
|
2015-10-23 09:17:16 +08:00
|
|
|
int dev_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb);
|
2021-03-24 09:30:32 +08:00
|
|
|
int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
|
|
|
|
struct net_device_path_stack *stack);
|
2014-09-12 06:35:09 +08:00
|
|
|
struct net_device *__dev_get_by_flags(struct net *net, unsigned short flags,
|
|
|
|
unsigned short mask);
|
2013-09-27 05:48:15 +08:00
|
|
|
struct net_device *dev_get_by_name(struct net *net, const char *name);
|
|
|
|
struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
|
|
|
|
struct net_device *__dev_get_by_name(struct net *net, const char *name);
|
2021-10-08 00:16:50 +08:00
|
|
|
bool netdev_name_in_use(struct net *net, const char *name);
|
2013-09-27 05:48:15 +08:00
|
|
|
int dev_alloc_name(struct net_device *dev, const char *name);
|
2018-12-07 01:05:36 +08:00
|
|
|
int dev_open(struct net_device *dev, struct netlink_ext_ack *extack);
|
2017-07-19 06:59:27 +08:00
|
|
|
void dev_close(struct net_device *dev);
|
|
|
|
void dev_close_many(struct list_head *head, bool unlink);
|
2013-09-27 05:48:15 +08:00
|
|
|
void dev_disable_lro(struct net_device *dev);
|
2015-09-16 09:04:18 +08:00
|
|
|
int dev_loopback_xmit(struct net *net, struct sock *sk, struct sk_buff *newskb);
|
2018-07-10 00:19:54 +08:00
|
|
|
u16 dev_pick_tx_zero(struct net_device *dev, struct sk_buff *skb,
|
2019-03-20 18:02:06 +08:00
|
|
|
struct net_device *sb_dev);
|
2018-07-10 00:19:54 +08:00
|
|
|
u16 dev_pick_tx_cpu_id(struct net_device *dev, struct sk_buff *skb,
|
2019-03-20 18:02:06 +08:00
|
|
|
struct net_device *sb_dev);
|
2020-11-24 01:56:00 +08:00
|
|
|
|
2022-04-28 18:58:46 +08:00
|
|
|
int __dev_queue_xmit(struct sk_buff *skb, struct net_device *sb_dev);
|
2020-11-24 01:56:00 +08:00
|
|
|
int __dev_direct_xmit(struct sk_buff *skb, u16 queue_id);
|
|
|
|
|
2022-04-28 18:58:46 +08:00
|
|
|
static inline int dev_queue_xmit(struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
return __dev_queue_xmit(skb, NULL);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int dev_queue_xmit_accel(struct sk_buff *skb,
|
|
|
|
struct net_device *sb_dev)
|
|
|
|
{
|
|
|
|
return __dev_queue_xmit(skb, sb_dev);
|
|
|
|
}
|
|
|
|
|
2020-11-24 01:56:00 +08:00
|
|
|
static inline int dev_direct_xmit(struct sk_buff *skb, u16 queue_id)
|
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
ret = __dev_direct_xmit(skb, queue_id);
|
|
|
|
if (!dev_xmit_complete(ret))
|
|
|
|
kfree_skb(skb);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
int register_netdevice(struct net_device *dev);
|
|
|
|
void unregister_netdevice_queue(struct net_device *dev, struct list_head *head);
|
|
|
|
void unregister_netdevice_many(struct list_head *head);
|
2009-10-27 15:03:04 +08:00
|
|
|
static inline void unregister_netdevice(struct net_device *dev)
|
|
|
|
{
|
|
|
|
unregister_netdevice_queue(dev, NULL);
|
|
|
|
}
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
int netdev_refcnt_read(const struct net_device *dev);
|
|
|
|
void free_netdev(struct net_device *dev);
|
2013-10-31 04:10:44 +08:00
|
|
|
void netdev_freemem(struct net_device *dev);
|
2013-09-27 05:48:15 +08:00
|
|
|
int init_dummy_netdev(struct net_device *dev);
|
2009-01-15 13:05:05 +08:00
|
|
|
|
2020-05-01 03:21:31 +08:00
|
|
|
struct net_device *netdev_get_xmit_slave(struct net_device *dev,
|
|
|
|
struct sk_buff *skb,
|
|
|
|
bool all_slaves);
|
2021-01-17 22:59:42 +08:00
|
|
|
struct net_device *netdev_sk_get_lowest_dev(struct net_device *dev,
|
|
|
|
struct sock *sk);
|
2013-09-27 05:48:15 +08:00
|
|
|
struct net_device *dev_get_by_index(struct net *net, int ifindex);
|
|
|
|
struct net_device *__dev_get_by_index(struct net *net, int ifindex);
|
|
|
|
struct net_device *dev_get_by_index_rcu(struct net *net, int ifindex);
|
2017-05-19 23:52:37 +08:00
|
|
|
struct net_device *dev_get_by_napi_id(unsigned int napi_id);
|
2013-09-27 05:48:15 +08:00
|
|
|
int dev_restart(struct net_device *dev);
|
2009-01-29 22:19:50 +08:00
|
|
|
|
2017-02-15 16:39:39 +08:00
|
|
|
|
2007-10-09 16:36:32 +08:00
|
|
|
static inline int dev_hard_header(struct sk_buff *skb, struct net_device *dev,
|
|
|
|
unsigned short type,
|
2007-10-09 16:40:57 +08:00
|
|
|
const void *daddr, const void *saddr,
|
2012-04-15 13:58:06 +08:00
|
|
|
unsigned int len)
|
2007-10-09 16:36:32 +08:00
|
|
|
{
|
2007-10-22 22:16:14 +08:00
|
|
|
if (!dev->header_ops || !dev->header_ops->create)
|
2007-10-09 16:36:32 +08:00
|
|
|
return 0;
|
2007-10-09 16:40:57 +08:00
|
|
|
|
|
|
|
return dev->header_ops->create(skb, dev, type, daddr, saddr, len);
|
2007-10-09 16:36:32 +08:00
|
|
|
}
|
|
|
|
|
2007-09-27 13:13:38 +08:00
|
|
|
static inline int dev_parse_header(const struct sk_buff *skb,
|
|
|
|
unsigned char *haddr)
|
|
|
|
{
|
|
|
|
const struct net_device *dev = skb->dev;
|
|
|
|
|
2007-10-18 20:09:28 +08:00
|
|
|
if (!dev->header_ops || !dev->header_ops->parse)
|
2007-09-27 13:13:38 +08:00
|
|
|
return 0;
|
2007-10-09 16:40:57 +08:00
|
|
|
return dev->header_ops->parse(skb, haddr);
|
2007-09-27 13:13:38 +08:00
|
|
|
}
|
|
|
|
|
2019-02-21 20:39:58 +08:00
|
|
|
static inline __be16 dev_parse_header_protocol(const struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
const struct net_device *dev = skb->dev;
|
|
|
|
|
|
|
|
if (!dev->header_ops || !dev->header_ops->parse_protocol)
|
|
|
|
return 0;
|
|
|
|
return dev->header_ops->parse_protocol(skb);
|
|
|
|
}
|
|
|
|
|
2016-03-10 10:58:32 +08:00
|
|
|
/* ll_header must have at least hard_header_len allocated */
|
|
|
|
static inline bool dev_validate_header(const struct net_device *dev,
|
|
|
|
char *ll_header, int len)
|
|
|
|
{
|
|
|
|
if (likely(len >= dev->hard_header_len))
|
|
|
|
return true;
|
2017-02-08 04:57:20 +08:00
|
|
|
if (len < dev->min_header_len)
|
|
|
|
return false;
|
2016-03-10 10:58:32 +08:00
|
|
|
|
|
|
|
if (capable(CAP_SYS_RAWIO)) {
|
|
|
|
memset(ll_header + len, 0, dev->hard_header_len - len);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (dev->header_ops && dev->header_ops->validate)
|
|
|
|
return dev->header_ops->validate(ll_header, len);
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2020-11-21 14:28:17 +08:00
|
|
|
static inline bool dev_has_header(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->header_ops && dev->header_ops->create;
|
|
|
|
}
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
2016-03-24 02:47:23 +08:00
|
|
|
* Incoming packets are placed on per-CPU queues
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
2009-11-05 01:50:58 +08:00
|
|
|
struct softnet_data {
|
2005-04-17 06:20:36 +08:00
|
|
|
struct list_head poll_list;
|
2010-04-28 06:07:33 +08:00
|
|
|
struct sk_buff_head process_queue;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2010-05-02 13:42:16 +08:00
|
|
|
/* stats */
|
2010-05-03 13:27:59 +08:00
|
|
|
unsigned int processed;
|
|
|
|
unsigned int time_squeeze;
|
2010-04-15 15:16:59 +08:00
|
|
|
#ifdef CONFIG_RPS
|
2010-04-19 13:07:33 +08:00
|
|
|
struct softnet_data *rps_ipi_list;
|
2014-11-02 22:00:12 +08:00
|
|
|
#endif
|
|
|
|
#ifdef CONFIG_NET_FLOW_LIMIT
|
|
|
|
struct sd_flow_limit __rcu *flow_limit;
|
|
|
|
#endif
|
|
|
|
struct Qdisc *output_queue;
|
|
|
|
struct Qdisc **output_queue_tailp;
|
|
|
|
struct sk_buff *completion_queue;
|
2017-12-20 17:41:36 +08:00
|
|
|
#ifdef CONFIG_XFRM_OFFLOAD
|
|
|
|
struct sk_buff_head xfrm_backlog;
|
|
|
|
#endif
|
2019-04-01 22:42:13 +08:00
|
|
|
/* written and read only by owning cpu: */
|
|
|
|
struct {
|
|
|
|
u16 recursion;
|
|
|
|
u8 more;
|
net: sched: use queue_mapping to pick tx queue
This patch fixes issue:
* If we install tc filters with act_skbedit in clsact hook.
It doesn't work, because netdev_core_pick_tx() overwrites
queue_mapping.
$ tc filter ... action skbedit queue_mapping 1
And this patch is useful:
* We can use FQ + EDT to implement efficient policies. Tx queues
are picked by xps, ndo_select_queue of netdev driver, or skb hash
in netdev_core_pick_tx(). In fact, the netdev driver, and skb
hash are _not_ under control. xps uses the CPUs map to select Tx
queues, but we can't figure out which task_struct of pod/containter
running on this cpu in most case. We can use clsact filters to classify
one pod/container traffic to one Tx queue. Why ?
In containter networking environment, there are two kinds of pod/
containter/net-namespace. One kind (e.g. P1, P2), the high throughput
is key in these applications. But avoid running out of network resource,
the outbound traffic of these pods is limited, using or sharing one
dedicated Tx queues assigned HTB/TBF/FQ Qdisc. Other kind of pods
(e.g. Pn), the low latency of data access is key. And the traffic is not
limited. Pods use or share other dedicated Tx queues assigned FIFO Qdisc.
This choice provides two benefits. First, contention on the HTB/FQ Qdisc
lock is significantly reduced since fewer CPUs contend for the same queue.
More importantly, Qdisc contention can be eliminated completely if each
CPU has its own FIFO Qdisc for the second kind of pods.
There must be a mechanism in place to support classifying traffic based on
pods/container to different Tx queues. Note that clsact is outside of Qdisc
while Qdisc can run a classifier to select a sub-queue under the lock.
In general recording the decision in the skb seems a little heavy handed.
This patch introduces a per-CPU variable, suggested by Eric.
The xmit.skip_txqueue flag is firstly cleared in __dev_queue_xmit().
- Tx Qdisc may install that skbedit actions, then xmit.skip_txqueue flag
is set in qdisc->enqueue() though tx queue has been selected in
netdev_tx_queue_mapping() or netdev_core_pick_tx(). That flag is cleared
firstly in __dev_queue_xmit(), is useful:
- Avoid picking Tx queue with netdev_tx_queue_mapping() in next netdev
in such case: eth0 macvlan - eth0.3 vlan - eth0 ixgbe-phy:
For example, eth0, macvlan in pod, which root Qdisc install skbedit
queue_mapping, send packets to eth0.3, vlan in host. In __dev_queue_xmit() of
eth0.3, clear the flag, does not select tx queue according to skb->queue_mapping
because there is no filters in clsact or tx Qdisc of this netdev.
Same action taked in eth0, ixgbe in Host.
- Avoid picking Tx queue for next packet. If we set xmit.skip_txqueue
in tx Qdisc (qdisc->enqueue()), the proper way to clear it is clearing it
in __dev_queue_xmit when processing next packets.
For performance reasons, use the static key. If user does not config the NET_EGRESS,
the patch will not be compiled.
+----+ +----+ +----+
| P1 | | P2 | | Pn |
+----+ +----+ +----+
| | |
+-----------+-----------+
|
| clsact/skbedit
| MQ
v
+-----------+-----------+
| q0 | q1 | qn
v v v
HTB/FQ HTB/FQ ... FIFO
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Jonathan Lemon <jonathan.lemon@gmail.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Alexander Lobakin <alobakin@pm.me>
Cc: Paolo Abeni <pabeni@redhat.com>
Cc: Talal Ahmad <talalahmad@google.com>
Cc: Kevin Hao <haokexin@gmail.com>
Cc: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Cc: Antoine Tenart <atenart@kernel.org>
Cc: Wei Wang <weiwan@google.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Suggested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2022-04-16 00:40:45 +08:00
|
|
|
#ifdef CONFIG_NET_EGRESS
|
|
|
|
u8 skip_txqueue;
|
|
|
|
#endif
|
2019-04-01 22:42:13 +08:00
|
|
|
} xmit;
|
2014-11-02 22:00:12 +08:00
|
|
|
#ifdef CONFIG_RPS
|
2016-04-27 06:30:07 +08:00
|
|
|
/* input_queue_head should be written by cpu owning this struct,
|
|
|
|
* and only read by other cpus. Worth using a cache line.
|
|
|
|
*/
|
|
|
|
unsigned int input_queue_head ____cacheline_aligned_in_smp;
|
|
|
|
|
|
|
|
/* Elements below can be accessed between CPUs for RPS/RFS */
|
smp: Avoid using two cache lines for struct call_single_data
struct call_single_data is used in IPIs to transfer information between
CPUs. Its size is bigger than sizeof(unsigned long) and less than
cache line size. Currently it is not allocated with any explicit alignment
requirements. This makes it possible for allocated call_single_data to
cross two cache lines, which results in double the number of the cache lines
that need to be transferred among CPUs.
This can be fixed by requiring call_single_data to be aligned with the
size of call_single_data. Currently the size of call_single_data is the
power of 2. If we add new fields to call_single_data, we may need to
add padding to make sure the size of new definition is the power of 2
as well.
Fortunately, this is enforced by GCC, which will report bad sizes.
To set alignment requirements of call_single_data to the size of
call_single_data, a struct definition and a typedef is used.
To test the effect of the patch, I used the vm-scalability multiple
thread swap test case (swap-w-seq-mt). The test will create multiple
threads and each thread will eat memory until all RAM and part of swap
is used, so that huge number of IPIs are triggered when unmapping
memory. In the test, the throughput of memory writing improves ~5%
compared with misaligned call_single_data, because of faster IPIs.
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Huang, Ying <ying.huang@intel.com>
[ Add call_single_data_t and align with size of call_single_data. ]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/87bmnqd6lz.fsf@yhuang-mobile.sh.intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-08-08 12:30:00 +08:00
|
|
|
call_single_data_t csd ____cacheline_aligned_in_smp;
|
2010-04-19 13:07:33 +08:00
|
|
|
struct softnet_data *rps_ipi_next;
|
|
|
|
unsigned int cpu;
|
2010-05-21 02:37:59 +08:00
|
|
|
unsigned int input_queue_tail;
|
2010-03-19 08:45:44 +08:00
|
|
|
#endif
|
2022-11-08 22:05:59 +08:00
|
|
|
unsigned int received_rps;
|
2012-04-15 13:58:06 +08:00
|
|
|
unsigned int dropped;
|
2010-03-16 16:03:29 +08:00
|
|
|
struct sk_buff_head input_pkt_queue;
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
struct napi_struct backlog;
|
2013-05-20 12:02:32 +08:00
|
|
|
|
net: generalize skb freeing deferral to per-cpu lists
Logic added in commit f35f821935d8 ("tcp: defer skb freeing after socket
lock is released") helped bulk TCP flows to move the cost of skbs
frees outside of critical section where socket lock was held.
But for RPC traffic, or hosts with RFS enabled, the solution is far from
being ideal.
For RPC traffic, recvmsg() has to return to user space right after
skb payload has been consumed, meaning that BH handler has no chance
to pick the skb before recvmsg() thread. This issue is more visible
with BIG TCP, as more RPC fit one skb.
For RFS, even if BH handler picks the skbs, they are still picked
from the cpu on which user thread is running.
Ideally, it is better to free the skbs (and associated page frags)
on the cpu that originally allocated them.
This patch removes the per socket anchor (sk->defer_list) and
instead uses a per-cpu list, which will hold more skbs per round.
This new per-cpu list is drained at the end of net_action_rx(),
after incoming packets have been processed, to lower latencies.
In normal conditions, skbs are added to the per-cpu list with
no further action. In the (unlikely) cases where the cpu does not
run net_action_rx() handler fast enough, we use an IPI to raise
NET_RX_SOFTIRQ on the remote cpu.
Also, we do not bother draining the per-cpu list from dev_cpu_dead()
This is because skbs in this list have no requirement on how fast
they should be freed.
Note that we can add in the future a small per-cpu cache
if we see any contention on sd->defer_lock.
Tested on a pair of hosts with 100Gbit NIC, RFS enabled,
and /proc/sys/net/ipv4/tcp_rmem[2] tuned to 16MB to work around
page recycling strategy used by NIC driver (its page pool capacity
being too small compared to number of skbs/pages held in sockets
receive queues)
Note that this tuning was only done to demonstrate worse
conditions for skb freeing for this particular test.
These conditions can happen in more general production workload.
10 runs of one TCP_STREAM flow
Before:
Average throughput: 49685 Mbit.
Kernel profiles on cpu running user thread recvmsg() show high cost for
skb freeing related functions (*)
57.81% [kernel] [k] copy_user_enhanced_fast_string
(*) 12.87% [kernel] [k] skb_release_data
(*) 4.25% [kernel] [k] __free_one_page
(*) 3.57% [kernel] [k] __list_del_entry_valid
1.85% [kernel] [k] __netif_receive_skb_core
1.60% [kernel] [k] __skb_datagram_iter
(*) 1.59% [kernel] [k] free_unref_page_commit
(*) 1.16% [kernel] [k] __slab_free
1.16% [kernel] [k] _copy_to_iter
(*) 1.01% [kernel] [k] kfree
(*) 0.88% [kernel] [k] free_unref_page
0.57% [kernel] [k] ip6_rcv_core
0.55% [kernel] [k] ip6t_do_table
0.54% [kernel] [k] flush_smp_call_function_queue
(*) 0.54% [kernel] [k] free_pcppages_bulk
0.51% [kernel] [k] llist_reverse_order
0.38% [kernel] [k] process_backlog
(*) 0.38% [kernel] [k] free_pcp_prepare
0.37% [kernel] [k] tcp_recvmsg_locked
(*) 0.37% [kernel] [k] __list_add_valid
0.34% [kernel] [k] sock_rfree
0.34% [kernel] [k] _raw_spin_lock_irq
(*) 0.33% [kernel] [k] __page_cache_release
0.33% [kernel] [k] tcp_v6_rcv
(*) 0.33% [kernel] [k] __put_page
(*) 0.29% [kernel] [k] __mod_zone_page_state
0.27% [kernel] [k] _raw_spin_lock
After patch:
Average throughput: 73076 Mbit.
Kernel profiles on cpu running user thread recvmsg() looks better:
81.35% [kernel] [k] copy_user_enhanced_fast_string
1.95% [kernel] [k] _copy_to_iter
1.95% [kernel] [k] __skb_datagram_iter
1.27% [kernel] [k] __netif_receive_skb_core
1.03% [kernel] [k] ip6t_do_table
0.60% [kernel] [k] sock_rfree
0.50% [kernel] [k] tcp_v6_rcv
0.47% [kernel] [k] ip6_rcv_core
0.45% [kernel] [k] read_tsc
0.44% [kernel] [k] _raw_spin_lock_irqsave
0.37% [kernel] [k] _raw_spin_lock
0.37% [kernel] [k] native_irq_return_iret
0.33% [kernel] [k] __inet6_lookup_established
0.31% [kernel] [k] ip6_protocol_deliver_rcu
0.29% [kernel] [k] tcp_rcv_established
0.29% [kernel] [k] llist_reverse_order
v2: kdoc issue (kernel bots)
do not defer if (alloc_cpu == smp_processor_id()) (Paolo)
replace the sk_buff_head with a single-linked list (Jakub)
add a READ_ONCE()/WRITE_ONCE() for the lockless read of sd->defer_list
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
Link: https://lore.kernel.org/r/20220422201237.416238-1-eric.dumazet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-04-23 04:12:37 +08:00
|
|
|
/* Another possibly contended cache line */
|
|
|
|
spinlock_t defer_lock ____cacheline_aligned_in_smp;
|
|
|
|
int defer_count;
|
2022-05-16 12:24:53 +08:00
|
|
|
int defer_ipi_scheduled;
|
net: generalize skb freeing deferral to per-cpu lists
Logic added in commit f35f821935d8 ("tcp: defer skb freeing after socket
lock is released") helped bulk TCP flows to move the cost of skbs
frees outside of critical section where socket lock was held.
But for RPC traffic, or hosts with RFS enabled, the solution is far from
being ideal.
For RPC traffic, recvmsg() has to return to user space right after
skb payload has been consumed, meaning that BH handler has no chance
to pick the skb before recvmsg() thread. This issue is more visible
with BIG TCP, as more RPC fit one skb.
For RFS, even if BH handler picks the skbs, they are still picked
from the cpu on which user thread is running.
Ideally, it is better to free the skbs (and associated page frags)
on the cpu that originally allocated them.
This patch removes the per socket anchor (sk->defer_list) and
instead uses a per-cpu list, which will hold more skbs per round.
This new per-cpu list is drained at the end of net_action_rx(),
after incoming packets have been processed, to lower latencies.
In normal conditions, skbs are added to the per-cpu list with
no further action. In the (unlikely) cases where the cpu does not
run net_action_rx() handler fast enough, we use an IPI to raise
NET_RX_SOFTIRQ on the remote cpu.
Also, we do not bother draining the per-cpu list from dev_cpu_dead()
This is because skbs in this list have no requirement on how fast
they should be freed.
Note that we can add in the future a small per-cpu cache
if we see any contention on sd->defer_lock.
Tested on a pair of hosts with 100Gbit NIC, RFS enabled,
and /proc/sys/net/ipv4/tcp_rmem[2] tuned to 16MB to work around
page recycling strategy used by NIC driver (its page pool capacity
being too small compared to number of skbs/pages held in sockets
receive queues)
Note that this tuning was only done to demonstrate worse
conditions for skb freeing for this particular test.
These conditions can happen in more general production workload.
10 runs of one TCP_STREAM flow
Before:
Average throughput: 49685 Mbit.
Kernel profiles on cpu running user thread recvmsg() show high cost for
skb freeing related functions (*)
57.81% [kernel] [k] copy_user_enhanced_fast_string
(*) 12.87% [kernel] [k] skb_release_data
(*) 4.25% [kernel] [k] __free_one_page
(*) 3.57% [kernel] [k] __list_del_entry_valid
1.85% [kernel] [k] __netif_receive_skb_core
1.60% [kernel] [k] __skb_datagram_iter
(*) 1.59% [kernel] [k] free_unref_page_commit
(*) 1.16% [kernel] [k] __slab_free
1.16% [kernel] [k] _copy_to_iter
(*) 1.01% [kernel] [k] kfree
(*) 0.88% [kernel] [k] free_unref_page
0.57% [kernel] [k] ip6_rcv_core
0.55% [kernel] [k] ip6t_do_table
0.54% [kernel] [k] flush_smp_call_function_queue
(*) 0.54% [kernel] [k] free_pcppages_bulk
0.51% [kernel] [k] llist_reverse_order
0.38% [kernel] [k] process_backlog
(*) 0.38% [kernel] [k] free_pcp_prepare
0.37% [kernel] [k] tcp_recvmsg_locked
(*) 0.37% [kernel] [k] __list_add_valid
0.34% [kernel] [k] sock_rfree
0.34% [kernel] [k] _raw_spin_lock_irq
(*) 0.33% [kernel] [k] __page_cache_release
0.33% [kernel] [k] tcp_v6_rcv
(*) 0.33% [kernel] [k] __put_page
(*) 0.29% [kernel] [k] __mod_zone_page_state
0.27% [kernel] [k] _raw_spin_lock
After patch:
Average throughput: 73076 Mbit.
Kernel profiles on cpu running user thread recvmsg() looks better:
81.35% [kernel] [k] copy_user_enhanced_fast_string
1.95% [kernel] [k] _copy_to_iter
1.95% [kernel] [k] __skb_datagram_iter
1.27% [kernel] [k] __netif_receive_skb_core
1.03% [kernel] [k] ip6t_do_table
0.60% [kernel] [k] sock_rfree
0.50% [kernel] [k] tcp_v6_rcv
0.47% [kernel] [k] ip6_rcv_core
0.45% [kernel] [k] read_tsc
0.44% [kernel] [k] _raw_spin_lock_irqsave
0.37% [kernel] [k] _raw_spin_lock
0.37% [kernel] [k] native_irq_return_iret
0.33% [kernel] [k] __inet6_lookup_established
0.31% [kernel] [k] ip6_protocol_deliver_rcu
0.29% [kernel] [k] tcp_rcv_established
0.29% [kernel] [k] llist_reverse_order
v2: kdoc issue (kernel bots)
do not defer if (alloc_cpu == smp_processor_id()) (Paolo)
replace the sk_buff_head with a single-linked list (Jakub)
add a READ_ONCE()/WRITE_ONCE() for the lockless read of sd->defer_list
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
Link: https://lore.kernel.org/r/20220422201237.416238-1-eric.dumazet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-04-23 04:12:37 +08:00
|
|
|
struct sk_buff *defer_list;
|
|
|
|
call_single_data_t defer_csd;
|
2005-04-17 06:20:36 +08:00
|
|
|
};
|
|
|
|
|
2010-05-21 02:37:59 +08:00
|
|
|
static inline void input_queue_head_incr(struct softnet_data *sd)
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
{
|
|
|
|
#ifdef CONFIG_RPS
|
2010-05-21 02:37:59 +08:00
|
|
|
sd->input_queue_head++;
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void input_queue_tail_incr_save(struct softnet_data *sd,
|
|
|
|
unsigned int *qtail)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_RPS
|
|
|
|
*qtail = ++sd->input_queue_tail;
|
rfs: Receive Flow Steering
This patch implements receive flow steering (RFS). RFS steers
received packets for layer 3 and 4 processing to the CPU where
the application for the corresponding flow is running. RFS is an
extension of Receive Packet Steering (RPS).
The basic idea of RFS is that when an application calls recvmsg
(or sendmsg) the application's running CPU is stored in a hash
table that is indexed by the connection's rxhash which is stored in
the socket structure. The rxhash is passed in skb's received on
the connection from netif_receive_skb. For each received packet,
the associated rxhash is used to look up the CPU in the hash table,
if a valid CPU is set then the packet is steered to that CPU using
the RPS mechanisms.
The convolution of the simple approach is that it would potentially
allow OOO packets. If threads are thrashing around CPUs or multiple
threads are trying to read from the same sockets, a quickly changing
CPU value in the hash table could cause rampant OOO packets--
we consider this a non-starter.
To avoid OOO packets, this solution implements two types of hash
tables: rps_sock_flow_table and rps_dev_flow_table.
rps_sock_table is a global hash table. Each entry is just a CPU
number and it is populated in recvmsg and sendmsg as described above.
This table contains the "desired" CPUs for flows.
rps_dev_flow_table is specific to each device queue. Each entry
contains a CPU and a tail queue counter. The CPU is the "current"
CPU for a matching flow. The tail queue counter holds the value
of a tail queue counter for the associated CPU's backlog queue at
the time of last enqueue for a flow matching the entry.
Each backlog queue has a queue head counter which is incremented
on dequeue, and so a queue tail counter is computed as queue head
count + queue length. When a packet is enqueued on a backlog queue,
the current value of the queue tail counter is saved in the hash
entry of the rps_dev_flow_table.
And now the trick: when selecting the CPU for RPS (get_rps_cpu)
the rps_sock_flow table and the rps_dev_flow table for the RX queue
are consulted. When the desired CPU for the flow (found in the
rps_sock_flow table) does not match the current CPU (found in the
rps_dev_flow table), the current CPU is changed to the desired CPU
if one of the following is true:
- The current CPU is unset (equal to RPS_NO_CPU)
- Current CPU is offline
- The current CPU's queue head counter >= queue tail counter in the
rps_dev_flow table. This checks if the queue tail has advanced
beyond the last packet that was enqueued using this table entry.
This guarantees that all packets queued using this entry have been
dequeued, thus preserving in order delivery.
Making each queue have its own rps_dev_flow table has two advantages:
1) the tail queue counters will be written on each receive, so
keeping the table local to interrupting CPU s good for locality. 2)
this allows lockless access to the table-- the CPU number and queue
tail counter need to be accessed together under mutual exclusion
from netif_receive_skb, we assume that this is only called from
device napi_poll which is non-reentrant.
This patch implements RFS for TCP and connected UDP sockets.
It should be usable for other flow oriented protocols.
There are two configuration parameters for RFS. The
"rps_flow_entries" kernel init parameter sets the number of
entries in the rps_sock_flow_table, the per rxqueue sysfs entry
"rps_flow_cnt" contains the number of entries in the rps_dev_flow
table for the rxqueue. Both are rounded to power of two.
The obvious benefit of RFS (over just RPS) is that it achieves
CPU locality between the receive processing for a flow and the
applications processing; this can result in increased performance
(higher pps, lower latency).
The benefits of RFS are dependent on cache hierarchy, application
load, and other factors. On simple benchmarks, we don't necessarily
see improvement and sometimes see degradation. However, for more
complex benchmarks and for applications where cache pressure is
much higher this technique seems to perform very well.
Below are some benchmark results which show the potential benfit of
this patch. The netperf test has 500 instances of netperf TCP_RR
test with 1 byte req. and resp. The RPC test is an request/response
test similar in structure to netperf RR test ith 100 threads on
each host, but does more work in userspace that netperf.
e1000e on 8 core Intel
No RFS or RPS 104K tps at 30% CPU
No RFS (best RPS config): 290K tps at 63% CPU
RFS 303K tps at 61% CPU
RPC test tps CPU% 50/90/99% usec latency Latency StdDev
No RFS/RPS 103K 48% 757/900/3185 4472.35
RPS only: 174K 73% 415/993/2468 491.66
RFS 223K 73% 379/651/1382 315.61
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2010-04-17 07:01:27 +08:00
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2010-03-16 16:03:29 +08:00
|
|
|
DECLARE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2019-04-01 22:42:13 +08:00
|
|
|
static inline int dev_recursion_level(void)
|
|
|
|
{
|
2019-04-03 14:28:35 +08:00
|
|
|
return this_cpu_read(softnet_data.xmit.recursion);
|
2019-04-01 22:42:13 +08:00
|
|
|
}
|
|
|
|
|
2020-06-16 23:52:05 +08:00
|
|
|
#define XMIT_RECURSION_LIMIT 8
|
2019-04-01 22:42:13 +08:00
|
|
|
static inline bool dev_xmit_recursion(void)
|
|
|
|
{
|
|
|
|
return unlikely(__this_cpu_read(softnet_data.xmit.recursion) >
|
|
|
|
XMIT_RECURSION_LIMIT);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void dev_xmit_recursion_inc(void)
|
|
|
|
{
|
|
|
|
__this_cpu_inc(softnet_data.xmit.recursion);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void dev_xmit_recursion_dec(void)
|
|
|
|
{
|
|
|
|
__this_cpu_dec(softnet_data.xmit.recursion);
|
|
|
|
}
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
void __netif_schedule(struct Qdisc *q);
|
2014-09-13 11:04:52 +08:00
|
|
|
void netif_schedule_queue(struct netdev_queue *txq);
|
2008-07-09 14:11:25 +08:00
|
|
|
|
2008-07-17 16:56:23 +08:00
|
|
|
static inline void netif_tx_schedule_all(struct net_device *dev)
|
|
|
|
{
|
|
|
|
unsigned int i;
|
|
|
|
|
|
|
|
for (i = 0; i < dev->num_tx_queues; i++)
|
|
|
|
netif_schedule_queue(netdev_get_tx_queue(dev, i));
|
|
|
|
}
|
|
|
|
|
net: force inlining of netif_tx_start/stop_queue, sock_hold, __sock_put
Sometimes gcc mysteriously doesn't inline
very small functions we expect to be inlined. See
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66122
Arguably, gcc should do better, but gcc people aren't willing
to invest time into it, asking to use __always_inline instead.
With this .config:
http://busybox.net/~vda/kernel_config_OPTIMIZE_INLINING_and_Os,
the following functions get deinlined many times.
netif_tx_stop_queue: 207 copies, 590 calls:
55 push %rbp
48 89 e5 mov %rsp,%rbp
f0 80 8f e0 01 00 00 01 lock orb $0x1,0x1e0(%rdi)
5d pop %rbp
c3 retq
netif_tx_start_queue: 47 copies, 111 calls
55 push %rbp
48 89 e5 mov %rsp,%rbp
f0 80 a7 e0 01 00 00 fe lock andb $0xfe,0x1e0(%rdi)
5d pop %rbp
c3 retq
sock_hold: 39 copies, 124 calls
55 push %rbp
48 89 e5 mov %rsp,%rbp
f0 ff 87 80 00 00 00 lock incl 0x80(%rdi)
5d pop %rbp
c3 retq
__sock_put: 6 copies, 13 calls
55 push %rbp
48 89 e5 mov %rsp,%rbp
f0 ff 8f 80 00 00 00 lock decl 0x80(%rdi)
5d pop %rbp
c3 retq
This patch fixes this via s/inline/__always_inline/.
Code size decrease after the patch is ~2.5k:
text data bss dec hex filename
56719876 56364551 36196352 149280779 8e5d80b vmlinux_before
56717440 56364551 36196352 149278343 8e5ce87 vmlinux
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
CC: David S. Miller <davem@davemloft.net>
CC: linux-kernel@vger.kernel.org
CC: netdev@vger.kernel.org
CC: netfilter-devel@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-04-08 23:51:54 +08:00
|
|
|
static __always_inline void netif_tx_start_queue(struct netdev_queue *dev_queue)
|
2008-07-23 05:09:06 +08:00
|
|
|
{
|
2011-11-29 00:32:44 +08:00
|
|
|
clear_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
|
2008-07-23 05:09:06 +08:00
|
|
|
}
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netif_start_queue - allow transmit
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Allow upper layers to call the device hard_start_xmit routine.
|
|
|
|
*/
|
2005-04-17 06:20:36 +08:00
|
|
|
static inline void netif_start_queue(struct net_device *dev)
|
|
|
|
{
|
2008-07-17 15:34:19 +08:00
|
|
|
netif_tx_start_queue(netdev_get_tx_queue(dev, 0));
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
2008-07-17 16:56:23 +08:00
|
|
|
static inline void netif_tx_start_all_queues(struct net_device *dev)
|
|
|
|
{
|
|
|
|
unsigned int i;
|
|
|
|
|
|
|
|
for (i = 0; i < dev->num_tx_queues; i++) {
|
|
|
|
struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
|
|
|
|
netif_tx_start_queue(txq);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-09-13 11:04:52 +08:00
|
|
|
void netif_tx_wake_queue(struct netdev_queue *dev_queue);
|
2008-07-09 14:14:46 +08:00
|
|
|
|
2008-07-23 05:09:06 +08:00
|
|
|
/**
|
|
|
|
* netif_wake_queue - restart transmit
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Allow upper layers to call the device hard_start_xmit routine.
|
|
|
|
* Used for flow control when transmit resources are available.
|
|
|
|
*/
|
2008-07-09 14:14:46 +08:00
|
|
|
static inline void netif_wake_queue(struct net_device *dev)
|
|
|
|
{
|
2008-07-17 15:34:19 +08:00
|
|
|
netif_tx_wake_queue(netdev_get_tx_queue(dev, 0));
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
2008-07-17 16:56:23 +08:00
|
|
|
static inline void netif_tx_wake_all_queues(struct net_device *dev)
|
|
|
|
{
|
|
|
|
unsigned int i;
|
|
|
|
|
|
|
|
for (i = 0; i < dev->num_tx_queues; i++) {
|
|
|
|
struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
|
|
|
|
netif_tx_wake_queue(txq);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
net: force inlining of netif_tx_start/stop_queue, sock_hold, __sock_put
Sometimes gcc mysteriously doesn't inline
very small functions we expect to be inlined. See
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66122
Arguably, gcc should do better, but gcc people aren't willing
to invest time into it, asking to use __always_inline instead.
With this .config:
http://busybox.net/~vda/kernel_config_OPTIMIZE_INLINING_and_Os,
the following functions get deinlined many times.
netif_tx_stop_queue: 207 copies, 590 calls:
55 push %rbp
48 89 e5 mov %rsp,%rbp
f0 80 8f e0 01 00 00 01 lock orb $0x1,0x1e0(%rdi)
5d pop %rbp
c3 retq
netif_tx_start_queue: 47 copies, 111 calls
55 push %rbp
48 89 e5 mov %rsp,%rbp
f0 80 a7 e0 01 00 00 fe lock andb $0xfe,0x1e0(%rdi)
5d pop %rbp
c3 retq
sock_hold: 39 copies, 124 calls
55 push %rbp
48 89 e5 mov %rsp,%rbp
f0 ff 87 80 00 00 00 lock incl 0x80(%rdi)
5d pop %rbp
c3 retq
__sock_put: 6 copies, 13 calls
55 push %rbp
48 89 e5 mov %rsp,%rbp
f0 ff 8f 80 00 00 00 lock decl 0x80(%rdi)
5d pop %rbp
c3 retq
This patch fixes this via s/inline/__always_inline/.
Code size decrease after the patch is ~2.5k:
text data bss dec hex filename
56719876 56364551 36196352 149280779 8e5d80b vmlinux_before
56717440 56364551 36196352 149278343 8e5ce87 vmlinux
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
CC: David S. Miller <davem@davemloft.net>
CC: linux-kernel@vger.kernel.org
CC: netdev@vger.kernel.org
CC: netfilter-devel@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-04-08 23:51:54 +08:00
|
|
|
static __always_inline void netif_tx_stop_queue(struct netdev_queue *dev_queue)
|
2008-07-23 05:09:06 +08:00
|
|
|
{
|
2011-11-29 00:32:44 +08:00
|
|
|
set_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
|
2008-07-23 05:09:06 +08:00
|
|
|
}
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netif_stop_queue - stop transmitted packets
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Stop upper layers calling the device hard_start_xmit routine.
|
|
|
|
* Used for flow control when transmit resources are unavailable.
|
|
|
|
*/
|
2005-04-17 06:20:36 +08:00
|
|
|
static inline void netif_stop_queue(struct net_device *dev)
|
|
|
|
{
|
2008-07-17 15:34:19 +08:00
|
|
|
netif_tx_stop_queue(netdev_get_tx_queue(dev, 0));
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
2015-05-12 03:17:53 +08:00
|
|
|
void netif_tx_stop_all_queues(struct net_device *dev);
|
2008-07-17 16:56:23 +08:00
|
|
|
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool netif_tx_queue_stopped(const struct netdev_queue *dev_queue)
|
2008-07-23 05:09:06 +08:00
|
|
|
{
|
2011-11-29 00:32:44 +08:00
|
|
|
return test_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
|
2008-07-23 05:09:06 +08:00
|
|
|
}
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netif_queue_stopped - test if transmit queue is flowblocked
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Test if transmit queue on device is currently unable to send.
|
|
|
|
*/
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool netif_queue_stopped(const struct net_device *dev)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
2008-07-17 15:34:19 +08:00
|
|
|
return netif_tx_queue_stopped(netdev_get_tx_queue(dev, 0));
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool netif_xmit_stopped(const struct netdev_queue *dev_queue)
|
2008-08-01 07:58:50 +08:00
|
|
|
{
|
2011-11-29 00:32:44 +08:00
|
|
|
return dev_queue->state & QUEUE_STATE_ANY_XOFF;
|
|
|
|
}
|
|
|
|
|
2014-04-03 02:52:57 +08:00
|
|
|
static inline bool
|
|
|
|
netif_xmit_frozen_or_stopped(const struct netdev_queue *dev_queue)
|
2011-11-29 00:32:44 +08:00
|
|
|
{
|
|
|
|
return dev_queue->state & QUEUE_STATE_ANY_XOFF_OR_FROZEN;
|
|
|
|
}
|
|
|
|
|
2014-04-03 02:52:57 +08:00
|
|
|
static inline bool
|
|
|
|
netif_xmit_frozen_or_drv_stopped(const struct netdev_queue *dev_queue)
|
|
|
|
{
|
|
|
|
return dev_queue->state & QUEUE_STATE_DRV_XOFF_OR_FROZEN;
|
|
|
|
}
|
|
|
|
|
2021-03-21 21:48:49 +08:00
|
|
|
/**
|
|
|
|
* netdev_queue_set_dql_min_limit - set dql minimum limit
|
|
|
|
* @dev_queue: pointer to transmit queue
|
|
|
|
* @min_limit: dql minimum limit
|
|
|
|
*
|
|
|
|
* Forces xmit_more() to return true until the minimum threshold
|
|
|
|
* defined by @min_limit is reached (or until the tx queue is
|
|
|
|
* empty). Warning: to be use with care, misuse will impact the
|
|
|
|
* latency.
|
|
|
|
*/
|
|
|
|
static inline void netdev_queue_set_dql_min_limit(struct netdev_queue *dev_queue,
|
|
|
|
unsigned int min_limit)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_BQL
|
|
|
|
dev_queue->dql.min_limit = min_limit;
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2014-10-08 23:19:27 +08:00
|
|
|
/**
|
|
|
|
* netdev_txq_bql_enqueue_prefetchw - prefetch bql data for write
|
|
|
|
* @dev_queue: pointer to transmit queue
|
|
|
|
*
|
|
|
|
* BQL enabled drivers might use this helper in their ndo_start_xmit(),
|
2016-03-24 02:47:23 +08:00
|
|
|
* to give appropriate hint to the CPU.
|
2014-10-08 23:19:27 +08:00
|
|
|
*/
|
|
|
|
static inline void netdev_txq_bql_enqueue_prefetchw(struct netdev_queue *dev_queue)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_BQL
|
|
|
|
prefetchw(&dev_queue->dql.num_queued);
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* netdev_txq_bql_complete_prefetchw - prefetch bql data for write
|
|
|
|
* @dev_queue: pointer to transmit queue
|
|
|
|
*
|
|
|
|
* BQL enabled drivers might use this helper in their TX completion path,
|
2016-03-24 02:47:23 +08:00
|
|
|
* to give appropriate hint to the CPU.
|
2014-10-08 23:19:27 +08:00
|
|
|
*/
|
|
|
|
static inline void netdev_txq_bql_complete_prefetchw(struct netdev_queue *dev_queue)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_BQL
|
|
|
|
prefetchw(&dev_queue->dql.limit);
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2022-09-01 02:44:27 +08:00
|
|
|
/**
|
|
|
|
* netdev_tx_sent_queue - report the number of bytes queued to a given tx queue
|
|
|
|
* @dev_queue: network device queue
|
|
|
|
* @bytes: number of bytes queued to the device queue
|
|
|
|
*
|
|
|
|
* Report the number of bytes queued for sending/completion to the network
|
|
|
|
* device hardware queue. @bytes should be a good approximation and should
|
|
|
|
* exactly match netdev_completed_queue() @bytes.
|
|
|
|
* This is typically called once per packet, from ndo_start_xmit().
|
|
|
|
*/
|
2011-11-29 00:32:52 +08:00
|
|
|
static inline void netdev_tx_sent_queue(struct netdev_queue *dev_queue,
|
|
|
|
unsigned int bytes)
|
|
|
|
{
|
2011-11-29 00:33:09 +08:00
|
|
|
#ifdef CONFIG_BQL
|
|
|
|
dql_queued(&dev_queue->dql, bytes);
|
2012-02-07 10:29:06 +08:00
|
|
|
|
|
|
|
if (likely(dql_avail(&dev_queue->dql) >= 0))
|
|
|
|
return;
|
|
|
|
|
|
|
|
set_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The XOFF flag must be set before checking the dql_avail below,
|
|
|
|
* because in netdev_tx_completed_queue we update the dql_completed
|
|
|
|
* before checking the XOFF flag.
|
|
|
|
*/
|
|
|
|
smp_mb();
|
|
|
|
|
|
|
|
/* check again in case another CPU has just made room avail */
|
|
|
|
if (unlikely(dql_avail(&dev_queue->dql) >= 0))
|
|
|
|
clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
|
2011-11-29 00:33:09 +08:00
|
|
|
#endif
|
2011-11-29 00:32:52 +08:00
|
|
|
}
|
|
|
|
|
net: bql: add __netdev_tx_sent_queue()
When qdisc_run() tries to use BQL budget to bulk-dequeue a batch
of packets, GSO can later transform this list in another list
of skbs, and each skb is sent to device ndo_start_xmit(),
one at a time, with skb->xmit_more being set to one but
for last skb.
Problem is that very often, BQL limit is hit in the middle of
the packet train, forcing dev_hard_start_xmit() to stop the
bulk send and requeue the end of the list.
BQL role is to avoid head of line blocking, making sure
a qdisc can deliver high priority packets before low priority ones.
But there is no way requeued packets can be bypassed by fresh
packets in the qdisc.
Aborting the bulk send increases TX softirqs, and hot cache
lines (after skb_segment()) are wasted.
Note that for TSO packets, we never split a packet in the middle
because of BQL limit being hit.
Drivers should be able to update BQL counters without
flipping/caring about BQL status, if the current skb
has xmit_more set.
Upper layers are ultimately responsible to stop sending another
packet train when BQL limit is hit.
Code template in a driver might look like the following :
send_doorbell = __netdev_tx_sent_queue(tx_queue, nr_bytes, skb->xmit_more);
Note that __netdev_tx_sent_queue() use is not mandatory,
since following patch will change dev_hard_start_xmit()
to not care about BQL status.
But it is highly recommended so that xmit_more full benefits
can be reached (less doorbells sent, and less atomic operations as well)
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-10-31 23:39:12 +08:00
|
|
|
/* Variant of netdev_tx_sent_queue() for drivers that are aware
|
|
|
|
* that they should not test BQL status themselves.
|
|
|
|
* We do want to change __QUEUE_STATE_STACK_XOFF only for the last
|
|
|
|
* skb of a batch.
|
|
|
|
* Returns true if the doorbell must be used to kick the NIC.
|
|
|
|
*/
|
|
|
|
static inline bool __netdev_tx_sent_queue(struct netdev_queue *dev_queue,
|
|
|
|
unsigned int bytes,
|
|
|
|
bool xmit_more)
|
|
|
|
{
|
|
|
|
if (xmit_more) {
|
|
|
|
#ifdef CONFIG_BQL
|
|
|
|
dql_queued(&dev_queue->dql, bytes);
|
|
|
|
#endif
|
|
|
|
return netif_tx_queue_stopped(dev_queue);
|
|
|
|
}
|
|
|
|
netdev_tx_sent_queue(dev_queue, bytes);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2013-09-06 23:58:00 +08:00
|
|
|
/**
|
2022-09-01 02:44:27 +08:00
|
|
|
* netdev_sent_queue - report the number of bytes queued to hardware
|
|
|
|
* @dev: network device
|
|
|
|
* @bytes: number of bytes queued to the hardware device queue
|
2013-09-06 23:58:00 +08:00
|
|
|
*
|
2022-09-01 02:44:27 +08:00
|
|
|
* Report the number of bytes queued for sending/completion to the network
|
|
|
|
* device hardware queue#0. @bytes should be a good approximation and should
|
|
|
|
* exactly match netdev_completed_queue() @bytes.
|
|
|
|
* This is typically called once per packet, from ndo_start_xmit().
|
2013-09-06 23:58:00 +08:00
|
|
|
*/
|
2011-11-29 00:32:52 +08:00
|
|
|
static inline void netdev_sent_queue(struct net_device *dev, unsigned int bytes)
|
|
|
|
{
|
|
|
|
netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes);
|
|
|
|
}
|
|
|
|
|
2018-11-25 21:30:29 +08:00
|
|
|
static inline bool __netdev_sent_queue(struct net_device *dev,
|
|
|
|
unsigned int bytes,
|
|
|
|
bool xmit_more)
|
|
|
|
{
|
|
|
|
return __netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes,
|
|
|
|
xmit_more);
|
|
|
|
}
|
|
|
|
|
2022-09-01 02:44:27 +08:00
|
|
|
/**
|
|
|
|
* netdev_tx_completed_queue - report number of packets/bytes at TX completion.
|
|
|
|
* @dev_queue: network device queue
|
|
|
|
* @pkts: number of packets (currently ignored)
|
|
|
|
* @bytes: number of bytes dequeued from the device queue
|
|
|
|
*
|
|
|
|
* Must be called at most once per TX completion round (and not per
|
|
|
|
* individual packet), so that BQL can adjust its limits appropriately.
|
|
|
|
*/
|
2011-11-29 00:32:52 +08:00
|
|
|
static inline void netdev_tx_completed_queue(struct netdev_queue *dev_queue,
|
2012-04-15 13:58:06 +08:00
|
|
|
unsigned int pkts, unsigned int bytes)
|
2011-11-29 00:32:52 +08:00
|
|
|
{
|
2011-11-29 00:33:09 +08:00
|
|
|
#ifdef CONFIG_BQL
|
2012-02-07 10:29:06 +08:00
|
|
|
if (unlikely(!bytes))
|
|
|
|
return;
|
|
|
|
|
|
|
|
dql_completed(&dev_queue->dql, bytes);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Without the memory barrier there is a small possiblity that
|
|
|
|
* netdev_tx_sent_queue will miss the update and cause the queue to
|
|
|
|
* be stopped forever
|
|
|
|
*/
|
|
|
|
smp_mb();
|
|
|
|
|
2019-08-22 14:58:16 +08:00
|
|
|
if (unlikely(dql_avail(&dev_queue->dql) < 0))
|
2012-02-07 10:29:06 +08:00
|
|
|
return;
|
|
|
|
|
|
|
|
if (test_and_clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state))
|
|
|
|
netif_schedule_queue(dev_queue);
|
2011-11-29 00:33:09 +08:00
|
|
|
#endif
|
2011-11-29 00:32:52 +08:00
|
|
|
}
|
|
|
|
|
2013-09-06 23:58:00 +08:00
|
|
|
/**
|
|
|
|
* netdev_completed_queue - report bytes and packets completed by device
|
|
|
|
* @dev: network device
|
|
|
|
* @pkts: actual number of packets sent over the medium
|
|
|
|
* @bytes: actual number of bytes sent over the medium
|
|
|
|
*
|
|
|
|
* Report the number of bytes and packets transmitted by the network device
|
|
|
|
* hardware queue over the physical medium, @bytes must exactly match the
|
|
|
|
* @bytes amount passed to netdev_sent_queue()
|
|
|
|
*/
|
2011-11-29 00:32:52 +08:00
|
|
|
static inline void netdev_completed_queue(struct net_device *dev,
|
2012-04-15 13:58:06 +08:00
|
|
|
unsigned int pkts, unsigned int bytes)
|
2011-11-29 00:32:52 +08:00
|
|
|
{
|
|
|
|
netdev_tx_completed_queue(netdev_get_tx_queue(dev, 0), pkts, bytes);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void netdev_tx_reset_queue(struct netdev_queue *q)
|
|
|
|
{
|
2011-11-29 00:33:09 +08:00
|
|
|
#ifdef CONFIG_BQL
|
2012-02-07 10:29:01 +08:00
|
|
|
clear_bit(__QUEUE_STATE_STACK_XOFF, &q->state);
|
2011-11-29 00:33:09 +08:00
|
|
|
dql_reset(&q->dql);
|
|
|
|
#endif
|
2011-11-29 00:32:52 +08:00
|
|
|
}
|
|
|
|
|
2013-09-06 23:58:00 +08:00
|
|
|
/**
|
|
|
|
* netdev_reset_queue - reset the packets and bytes count of a network device
|
|
|
|
* @dev_queue: network device
|
|
|
|
*
|
|
|
|
* Reset the bytes and packet count of a network device and clear the
|
|
|
|
* software flow control OFF bit for this network device
|
|
|
|
*/
|
2011-11-29 00:32:52 +08:00
|
|
|
static inline void netdev_reset_queue(struct net_device *dev_queue)
|
|
|
|
{
|
|
|
|
netdev_tx_reset_queue(netdev_get_tx_queue(dev_queue, 0));
|
2008-08-01 07:58:50 +08:00
|
|
|
}
|
|
|
|
|
2014-02-16 22:55:21 +08:00
|
|
|
/**
|
|
|
|
* netdev_cap_txqueue - check if selected tx queue exceeds device queues
|
|
|
|
* @dev: network device
|
|
|
|
* @queue_index: given tx queue index
|
|
|
|
*
|
|
|
|
* Returns 0 if given tx queue index >= number of device tx queues,
|
|
|
|
* otherwise returns the originally passed tx queue index.
|
|
|
|
*/
|
|
|
|
static inline u16 netdev_cap_txqueue(struct net_device *dev, u16 queue_index)
|
|
|
|
{
|
|
|
|
if (unlikely(queue_index >= dev->real_num_tx_queues)) {
|
|
|
|
net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
|
|
|
|
dev->name, queue_index,
|
|
|
|
dev->real_num_tx_queues);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
return queue_index;
|
|
|
|
}
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netif_running - test if up
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Test if the device has been brought up.
|
|
|
|
*/
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool netif_running(const struct net_device *dev)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
return test_bit(__LINK_STATE_START, &dev->state);
|
|
|
|
}
|
|
|
|
|
2007-07-07 04:36:20 +08:00
|
|
|
/*
|
2016-03-24 02:47:23 +08:00
|
|
|
* Routines to manage the subqueues on a device. We only need start,
|
2007-07-07 04:36:20 +08:00
|
|
|
* stop, and a check if it's stopped. All other device management is
|
|
|
|
* done at the overall netdevice level.
|
|
|
|
* Also test the device if we're multiqueue.
|
|
|
|
*/
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
|
|
|
|
/**
|
|
|
|
* netif_start_subqueue - allow sending packets on subqueue
|
|
|
|
* @dev: network device
|
|
|
|
* @queue_index: sub queue index
|
|
|
|
*
|
|
|
|
* Start individual transmit queue of a device with multiple transmit queues.
|
|
|
|
*/
|
2007-07-07 04:36:20 +08:00
|
|
|
static inline void netif_start_subqueue(struct net_device *dev, u16 queue_index)
|
|
|
|
{
|
2008-07-17 16:56:23 +08:00
|
|
|
struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
|
2009-08-30 04:21:21 +08:00
|
|
|
|
|
|
|
netif_tx_start_queue(txq);
|
2007-07-07 04:36:20 +08:00
|
|
|
}
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netif_stop_subqueue - stop sending packets on subqueue
|
|
|
|
* @dev: network device
|
|
|
|
* @queue_index: sub queue index
|
|
|
|
*
|
|
|
|
* Stop individual transmit queue of a device with multiple transmit queues.
|
|
|
|
*/
|
2007-07-07 04:36:20 +08:00
|
|
|
static inline void netif_stop_subqueue(struct net_device *dev, u16 queue_index)
|
|
|
|
{
|
2008-07-17 16:56:23 +08:00
|
|
|
struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
|
2009-08-30 04:21:21 +08:00
|
|
|
netif_tx_stop_queue(txq);
|
2007-07-07 04:36:20 +08:00
|
|
|
}
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
2020-11-16 18:17:59 +08:00
|
|
|
* __netif_subqueue_stopped - test status of subqueue
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
* @dev: network device
|
|
|
|
* @queue_index: sub queue index
|
|
|
|
*
|
|
|
|
* Check individual transmit queue of a device with multiple transmit queues.
|
|
|
|
*/
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool __netif_subqueue_stopped(const struct net_device *dev,
|
|
|
|
u16 queue_index)
|
2007-07-07 04:36:20 +08:00
|
|
|
{
|
2008-07-17 16:56:23 +08:00
|
|
|
struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
|
2009-08-30 04:21:21 +08:00
|
|
|
|
|
|
|
return netif_tx_queue_stopped(txq);
|
2007-07-07 04:36:20 +08:00
|
|
|
}
|
|
|
|
|
2020-11-16 18:17:59 +08:00
|
|
|
/**
|
|
|
|
* netif_subqueue_stopped - test status of subqueue
|
|
|
|
* @dev: network device
|
|
|
|
* @skb: sub queue buffer pointer
|
|
|
|
*
|
|
|
|
* Check individual transmit queue of a device with multiple transmit queues.
|
|
|
|
*/
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool netif_subqueue_stopped(const struct net_device *dev,
|
|
|
|
struct sk_buff *skb)
|
2007-10-22 08:01:56 +08:00
|
|
|
{
|
|
|
|
return __netif_subqueue_stopped(dev, skb_get_queue_mapping(skb));
|
|
|
|
}
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
|
2017-01-12 13:13:02 +08:00
|
|
|
/**
|
|
|
|
* netif_wake_subqueue - allow sending packets on subqueue
|
|
|
|
* @dev: network device
|
|
|
|
* @queue_index: sub queue index
|
|
|
|
*
|
|
|
|
* Resume individual transmit queue of a device with multiple transmit queues.
|
|
|
|
*/
|
|
|
|
static inline void netif_wake_subqueue(struct net_device *dev, u16 queue_index)
|
|
|
|
{
|
|
|
|
struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
|
|
|
|
|
|
|
|
netif_tx_wake_queue(txq);
|
|
|
|
}
|
2007-07-07 04:36:20 +08:00
|
|
|
|
2013-01-10 16:57:02 +08:00
|
|
|
#ifdef CONFIG_XPS
|
2013-10-09 11:07:53 +08:00
|
|
|
int netif_set_xps_queue(struct net_device *dev, const struct cpumask *mask,
|
2013-09-27 05:48:15 +08:00
|
|
|
u16 index);
|
2018-06-30 12:26:41 +08:00
|
|
|
int __netif_set_xps_queue(struct net_device *dev, const unsigned long *mask,
|
2021-03-19 02:37:46 +08:00
|
|
|
u16 index, enum xps_map_type type);
|
2018-06-30 12:26:41 +08:00
|
|
|
|
|
|
|
/**
|
|
|
|
* netif_attr_test_mask - Test a CPU or Rx queue set in a mask
|
|
|
|
* @j: CPU/Rx queue index
|
|
|
|
* @mask: bitmask of all cpus/rx queues
|
|
|
|
* @nr_bits: number of bits in the bitmask
|
|
|
|
*
|
|
|
|
* Test if a CPU or Rx queue index is set in a mask of all CPU/Rx queues.
|
|
|
|
*/
|
|
|
|
static inline bool netif_attr_test_mask(unsigned long j,
|
|
|
|
const unsigned long *mask,
|
|
|
|
unsigned int nr_bits)
|
|
|
|
{
|
|
|
|
cpu_max_bits_warn(j, nr_bits);
|
|
|
|
return test_bit(j, mask);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* netif_attr_test_online - Test for online CPU/Rx queue
|
|
|
|
* @j: CPU/Rx queue index
|
|
|
|
* @online_mask: bitmask for CPUs/Rx queues that are online
|
|
|
|
* @nr_bits: number of bits in the bitmask
|
|
|
|
*
|
|
|
|
* Returns true if a CPU/Rx queue is online.
|
|
|
|
*/
|
|
|
|
static inline bool netif_attr_test_online(unsigned long j,
|
|
|
|
const unsigned long *online_mask,
|
|
|
|
unsigned int nr_bits)
|
|
|
|
{
|
|
|
|
cpu_max_bits_warn(j, nr_bits);
|
|
|
|
|
|
|
|
if (online_mask)
|
|
|
|
return test_bit(j, online_mask);
|
|
|
|
|
|
|
|
return (j < nr_bits);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* netif_attrmask_next - get the next CPU/Rx queue in a cpu/Rx queues mask
|
|
|
|
* @n: CPU/Rx queue index
|
|
|
|
* @srcp: the cpumask/Rx queue mask pointer
|
|
|
|
* @nr_bits: number of bits in the bitmask
|
|
|
|
*
|
|
|
|
* Returns >= nr_bits if no further CPUs/Rx queues set.
|
|
|
|
*/
|
|
|
|
static inline unsigned int netif_attrmask_next(int n, const unsigned long *srcp,
|
|
|
|
unsigned int nr_bits)
|
|
|
|
{
|
2022-10-15 00:07:46 +08:00
|
|
|
/* -1 is a legal arg here. */
|
|
|
|
if (n != -1)
|
|
|
|
cpu_max_bits_warn(n, nr_bits);
|
2018-06-30 12:26:41 +08:00
|
|
|
|
|
|
|
if (srcp)
|
|
|
|
return find_next_bit(srcp, nr_bits, n + 1);
|
|
|
|
|
|
|
|
return n + 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2020-02-13 14:28:20 +08:00
|
|
|
* netif_attrmask_next_and - get the next CPU/Rx queue in \*src1p & \*src2p
|
2018-06-30 12:26:41 +08:00
|
|
|
* @n: CPU/Rx queue index
|
|
|
|
* @src1p: the first CPUs/Rx queues mask pointer
|
|
|
|
* @src2p: the second CPUs/Rx queues mask pointer
|
|
|
|
* @nr_bits: number of bits in the bitmask
|
|
|
|
*
|
|
|
|
* Returns >= nr_bits if no further CPUs/Rx queues set in both.
|
|
|
|
*/
|
|
|
|
static inline int netif_attrmask_next_and(int n, const unsigned long *src1p,
|
|
|
|
const unsigned long *src2p,
|
|
|
|
unsigned int nr_bits)
|
|
|
|
{
|
2022-10-15 00:07:46 +08:00
|
|
|
/* -1 is a legal arg here. */
|
|
|
|
if (n != -1)
|
|
|
|
cpu_max_bits_warn(n, nr_bits);
|
2018-06-30 12:26:41 +08:00
|
|
|
|
|
|
|
if (src1p && src2p)
|
|
|
|
return find_next_and_bit(src1p, src2p, nr_bits, n + 1);
|
|
|
|
else if (src1p)
|
|
|
|
return find_next_bit(src1p, nr_bits, n + 1);
|
|
|
|
else if (src2p)
|
|
|
|
return find_next_bit(src2p, nr_bits, n + 1);
|
|
|
|
|
|
|
|
return n + 1;
|
|
|
|
}
|
2013-01-10 16:57:02 +08:00
|
|
|
#else
|
|
|
|
static inline int netif_set_xps_queue(struct net_device *dev,
|
2013-10-02 14:14:06 +08:00
|
|
|
const struct cpumask *mask,
|
2013-01-10 16:57:02 +08:00
|
|
|
u16 index)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
2018-08-10 16:47:43 +08:00
|
|
|
|
|
|
|
static inline int __netif_set_xps_queue(struct net_device *dev,
|
|
|
|
const unsigned long *mask,
|
2021-03-19 02:37:46 +08:00
|
|
|
u16 index, enum xps_map_type type)
|
2018-08-10 16:47:43 +08:00
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
2013-01-10 16:57:02 +08:00
|
|
|
#endif
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netif_is_multiqueue - test if device has multiple transmit queues
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Check if device has multiple transmit queues
|
|
|
|
*/
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool netif_is_multiqueue(const struct net_device *dev)
|
2007-07-07 04:36:20 +08:00
|
|
|
{
|
2010-09-23 04:43:57 +08:00
|
|
|
return dev->num_tx_queues > 1;
|
2007-07-07 04:36:20 +08:00
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq);
|
2010-07-01 21:21:57 +08:00
|
|
|
|
2014-01-17 14:23:28 +08:00
|
|
|
#ifdef CONFIG_SYSFS
|
2013-09-27 05:48:15 +08:00
|
|
|
int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq);
|
2010-09-27 16:24:33 +08:00
|
|
|
#else
|
|
|
|
static inline int netif_set_real_num_rx_queues(struct net_device *dev,
|
2018-07-31 11:43:51 +08:00
|
|
|
unsigned int rxqs)
|
2010-09-27 16:24:33 +08:00
|
|
|
{
|
2018-07-31 11:43:51 +08:00
|
|
|
dev->real_num_rx_queues = rxqs;
|
2010-09-27 16:24:33 +08:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
#endif
|
2021-08-03 21:05:26 +08:00
|
|
|
int netif_set_real_num_queues(struct net_device *dev,
|
|
|
|
unsigned int txq, unsigned int rxq);
|
2010-09-27 16:24:33 +08:00
|
|
|
|
2018-01-31 19:58:56 +08:00
|
|
|
static inline struct netdev_rx_queue *
|
|
|
|
__netif_get_rx_queue(struct net_device *dev, unsigned int rxq)
|
|
|
|
{
|
|
|
|
return dev->_rx + rxq;
|
|
|
|
}
|
|
|
|
|
2014-01-17 14:23:28 +08:00
|
|
|
#ifdef CONFIG_SYSFS
|
|
|
|
static inline unsigned int get_netdev_rx_queue_index(
|
|
|
|
struct netdev_rx_queue *queue)
|
|
|
|
{
|
|
|
|
struct net_device *dev = queue->dev;
|
|
|
|
int index = queue - dev->_rx;
|
|
|
|
|
|
|
|
BUG_ON(index >= dev->num_rx_queues);
|
|
|
|
return index;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
int netif_get_num_default_rss_queues(void);
|
2012-07-01 11:18:50 +08:00
|
|
|
|
2013-12-05 20:45:08 +08:00
|
|
|
enum skb_free_reason {
|
|
|
|
SKB_REASON_CONSUMED,
|
|
|
|
SKB_REASON_DROPPED,
|
|
|
|
};
|
|
|
|
|
|
|
|
void __dev_kfree_skb_irq(struct sk_buff *skb, enum skb_free_reason reason);
|
|
|
|
void __dev_kfree_skb_any(struct sk_buff *skb, enum skb_free_reason reason);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2013-12-05 20:45:08 +08:00
|
|
|
/*
|
|
|
|
* It is not allowed to call kfree_skb() or consume_skb() from hardware
|
|
|
|
* interrupt context or with hardware interrupts being disabled.
|
2021-08-13 22:57:49 +08:00
|
|
|
* (in_hardirq() || irqs_disabled())
|
2013-12-05 20:45:08 +08:00
|
|
|
*
|
|
|
|
* We provide four helpers that can be used in following contexts :
|
|
|
|
*
|
|
|
|
* dev_kfree_skb_irq(skb) when caller drops a packet from irq context,
|
|
|
|
* replacing kfree_skb(skb)
|
|
|
|
*
|
|
|
|
* dev_consume_skb_irq(skb) when caller consumes a packet from irq context.
|
|
|
|
* Typically used in place of consume_skb(skb) in TX completion path
|
|
|
|
*
|
|
|
|
* dev_kfree_skb_any(skb) when caller doesn't know its current irq context,
|
|
|
|
* replacing kfree_skb(skb)
|
|
|
|
*
|
|
|
|
* dev_consume_skb_any(skb) when caller doesn't know its current irq context,
|
|
|
|
* and consumed a packet. Used in place of consume_skb(skb)
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
2013-12-05 20:45:08 +08:00
|
|
|
static inline void dev_kfree_skb_irq(struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
__dev_kfree_skb_irq(skb, SKB_REASON_DROPPED);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void dev_consume_skb_irq(struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
__dev_kfree_skb_irq(skb, SKB_REASON_CONSUMED);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void dev_kfree_skb_any(struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
__dev_kfree_skb_any(skb, SKB_REASON_DROPPED);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void dev_consume_skb_any(struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
__dev_kfree_skb_any(skb, SKB_REASON_CONSUMED);
|
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2021-07-02 19:18:21 +08:00
|
|
|
u32 bpf_prog_run_generic_xdp(struct sk_buff *skb, struct xdp_buff *xdp,
|
|
|
|
struct bpf_prog *xdp_prog);
|
2017-08-11 19:41:17 +08:00
|
|
|
void generic_xdp_tx(struct sk_buff *skb, struct bpf_prog *xdp_prog);
|
|
|
|
int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff *skb);
|
2013-09-27 05:48:15 +08:00
|
|
|
int netif_rx(struct sk_buff *skb);
|
2022-02-12 07:38:38 +08:00
|
|
|
int __netif_rx(struct sk_buff *skb);
|
|
|
|
|
2015-09-16 09:04:15 +08:00
|
|
|
int netif_receive_skb(struct sk_buff *skb);
|
bpf: cpumap xdp_buff to skb conversion and allocation
This patch makes cpumap functional, by adding SKB allocation and
invoking the network stack on the dequeuing CPU.
For constructing the SKB on the remote CPU, the xdp_buff in converted
into a struct xdp_pkt, and it mapped into the top headroom of the
packet, to avoid allocating separate mem. For now, struct xdp_pkt is
just a cpumap internal data structure, with info carried between
enqueue to dequeue.
If a driver doesn't have enough headroom it is simply dropped, with
return code -EOVERFLOW. This will be picked up the xdp tracepoint
infrastructure, to allow users to catch this.
V2: take into account xdp->data_meta
V4:
- Drop busypoll tricks, keeping it more simple.
- Skip RPS and Generic-XDP-recursive-reinjection, suggested by Alexei
V5: correct RCU read protection around __netif_receive_skb_core.
V6: Setting TASK_RUNNING vs TASK_INTERRUPTIBLE based on talk with Rik van Riel
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-16 18:19:39 +08:00
|
|
|
int netif_receive_skb_core(struct sk_buff *skb);
|
2021-11-16 01:05:54 +08:00
|
|
|
void netif_receive_skb_list_internal(struct list_head *head);
|
2018-07-02 23:12:45 +08:00
|
|
|
void netif_receive_skb_list(struct list_head *head);
|
2013-09-27 05:48:15 +08:00
|
|
|
gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb);
|
|
|
|
void napi_gro_flush(struct napi_struct *napi, bool flush_old);
|
|
|
|
struct sk_buff *napi_get_frags(struct napi_struct *napi);
|
2022-09-28 16:43:09 +08:00
|
|
|
void napi_get_frags_check(struct napi_struct *napi);
|
2013-09-27 05:48:15 +08:00
|
|
|
gro_result_t napi_gro_frags(struct napi_struct *napi);
|
net-gre-gro: Add GRE support to the GRO stack
This patch built on top of Commit 299603e8370a93dd5d8e8d800f0dff1ce2c53d36
("net-gro: Prepare GRO stack for the upcoming tunneling support") to add
the support of the standard GRE (RFC1701/RFC2784/RFC2890) to the GRO
stack. It also serves as an example for supporting other encapsulation
protocols in the GRO stack in the future.
The patch supports version 0 and all the flags (key, csum, seq#) but
will flush any pkt with the S (seq#) flag. This is because the S flag
is not support by GSO, and a GRO pkt may end up in the forwarding path,
thus requiring GSO support to break it up correctly.
Currently the "packet_offload" structure only contains L3 (ETH_P_IP/
ETH_P_IPV6) GRO offload support so the encapped pkts are limited to
IP pkts (i.e., w/o L2 hdr). But support for other protocol type can
be easily added, so is the support for GRE variations like NVGRE.
The patch also support csum offload. Specifically if the csum flag is on
and the h/w is capable of checksumming the payload (CHECKSUM_COMPLETE),
the code will take advantage of the csum computed by the h/w when
validating the GRE csum.
Note that commit 60769a5dcd8755715c7143b4571d5c44f01796f1 "ipv4: gre:
add GRO capability" already introduces GRO capability to IPv4 GRE
tunnels, using the gro_cells infrastructure. But GRO is done after
GRE hdr has been removed (i.e., decapped). The following patch applies
GRO when pkts first come in (before hitting the GRE tunnel code). There
is some performance advantage for applying GRO as early as possible.
Also this approach is transparent to other subsystem like Open vSwitch
where GRE decap is handled outside of the IP stack hence making it
harder for the gro_cells stuff to apply. On the other hand, some NICs
are still not capable of hashing on the inner hdr of a GRE pkt (RSS).
In that case the GRO processing of pkts from the same remote host will
all happen on the same CPU and the performance may be suboptimal.
I'm including some rough preliminary performance numbers below. Note
that the performance will be highly dependent on traffic load, mix as
usual. Moreover it also depends on NIC offload features hence the
following is by no means a comprehesive study. Local testing and tuning
will be needed to decide the best setting.
All tests spawned 50 copies of netperf TCP_STREAM and ran for 30 secs.
(super_netperf 50 -H 192.168.1.18 -l 30)
An IP GRE tunnel with only the key flag on (e.g., ip tunnel add gre1
mode gre local 10.246.17.18 remote 10.246.17.17 ttl 255 key 123)
is configured.
The GRO support for pkts AFTER decap are controlled through the device
feature of the GRE device (e.g., ethtool -K gre1 gro on/off).
1.1 ethtool -K gre1 gro off; ethtool -K eth0 gro off
thruput: 9.16Gbps
CPU utilization: 19%
1.2 ethtool -K gre1 gro on; ethtool -K eth0 gro off
thruput: 5.9Gbps
CPU utilization: 15%
1.3 ethtool -K gre1 gro off; ethtool -K eth0 gro on
thruput: 9.26Gbps
CPU utilization: 12-13%
1.4 ethtool -K gre1 gro on; ethtool -K eth0 gro on
thruput: 9.26Gbps
CPU utilization: 10%
The following tests were performed on a different NIC that is capable of
csum offload. I.e., the h/w is capable of computing IP payload csum
(CHECKSUM_COMPLETE).
2.1 ethtool -K gre1 gro on (hence will use gro_cells)
2.1.1 ethtool -K eth0 gro off; csum offload disabled
thruput: 8.53Gbps
CPU utilization: 9%
2.1.2 ethtool -K eth0 gro off; csum offload enabled
thruput: 8.97Gbps
CPU utilization: 7-8%
2.1.3 ethtool -K eth0 gro on; csum offload disabled
thruput: 8.83Gbps
CPU utilization: 5-6%
2.1.4 ethtool -K eth0 gro on; csum offload enabled
thruput: 8.98Gbps
CPU utilization: 5%
2.2 ethtool -K gre1 gro off
2.2.1 ethtool -K eth0 gro off; csum offload disabled
thruput: 5.93Gbps
CPU utilization: 9%
2.2.2 ethtool -K eth0 gro off; csum offload enabled
thruput: 5.62Gbps
CPU utilization: 8%
2.2.3 ethtool -K eth0 gro on; csum offload disabled
thruput: 7.69Gbps
CPU utilization: 8%
2.2.4 ethtool -K eth0 gro on; csum offload enabled
thruput: 8.96Gbps
CPU utilization: 5-6%
Signed-off-by: H.K. Jerry Chu <hkchu@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-01-08 02:23:19 +08:00
|
|
|
struct packet_offload *gro_find_receive_by_type(__be16 type);
|
|
|
|
struct packet_offload *gro_find_complete_by_type(__be16 type);
|
2009-04-16 17:02:07 +08:00
|
|
|
|
|
|
|
static inline void napi_free_frags(struct napi_struct *napi)
|
|
|
|
{
|
|
|
|
kfree_skb(napi->skb);
|
|
|
|
napi->skb = NULL;
|
|
|
|
}
|
|
|
|
|
2016-09-02 13:18:34 +08:00
|
|
|
bool netdev_is_rx_handler_busy(struct net_device *dev);
|
2013-09-27 05:48:15 +08:00
|
|
|
int netdev_rx_handler_register(struct net_device *dev,
|
|
|
|
rx_handler_func_t *rx_handler,
|
|
|
|
void *rx_handler_data);
|
|
|
|
void netdev_rx_handler_unregister(struct net_device *dev);
|
|
|
|
|
|
|
|
bool dev_valid_name(const char *name);
|
2021-08-27 03:46:01 +08:00
|
|
|
static inline bool is_socket_ioctl_cmd(unsigned int cmd)
|
|
|
|
{
|
|
|
|
return _IOC_TYPE(cmd) == SOCK_IOC_TYPE;
|
|
|
|
}
|
net: socket: rework compat_ifreq_ioctl()
compat_ifreq_ioctl() is one of the last users of copy_in_user() and
compat_alloc_user_space(), as it attempts to convert the 'struct ifreq'
arguments from 32-bit to 64-bit format as used by dev_ioctl() and a
couple of socket family specific interpretations.
The current implementation works correctly when calling dev_ioctl(),
inet_ioctl(), ieee802154_sock_ioctl(), atalk_ioctl(), qrtr_ioctl()
and packet_ioctl(). The ioctl handlers for x25, netrom, rose and x25 do
not interpret the arguments and only block the corresponding commands,
so they do not care.
For af_inet6 and af_decnet however, the compat conversion is slightly
incorrect, as it will copy more data than the native handler accesses,
both of them use a structure that is shorter than ifreq.
Replace the copy_in_user() conversion with a pair of accessor functions
to read and write the ifreq data in place with the correct length where
needed, while leaving the other ones to copy the (already compatible)
structures directly.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 22:29:03 +08:00
|
|
|
int get_user_ifreq(struct ifreq *ifr, void __user **ifrdata, void __user *arg);
|
|
|
|
int put_user_ifreq(struct ifreq *ifr, void __user *arg);
|
2017-10-06 00:59:44 +08:00
|
|
|
int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr,
|
2021-07-27 21:45:12 +08:00
|
|
|
void __user *data, bool *need_copyout);
|
2021-07-22 22:29:02 +08:00
|
|
|
int dev_ifconf(struct net *net, struct ifconf __user *ifc);
|
2021-07-27 21:45:12 +08:00
|
|
|
int dev_ethtool(struct net *net, struct ifreq *ifr, void __user *userdata);
|
2013-09-27 05:48:15 +08:00
|
|
|
unsigned int dev_get_flags(const struct net_device *);
|
2018-12-07 01:05:43 +08:00
|
|
|
int __dev_change_flags(struct net_device *dev, unsigned int flags,
|
|
|
|
struct netlink_ext_ack *extack);
|
2018-12-07 01:05:42 +08:00
|
|
|
int dev_change_flags(struct net_device *dev, unsigned int flags,
|
|
|
|
struct netlink_ext_ack *extack);
|
2013-09-27 05:48:15 +08:00
|
|
|
int dev_set_alias(struct net_device *, const char *, size_t);
|
2017-10-03 05:50:05 +08:00
|
|
|
int dev_get_alias(const struct net_device *, char *, size_t);
|
2021-04-07 14:40:51 +08:00
|
|
|
int __dev_change_net_namespace(struct net_device *dev, struct net *net,
|
|
|
|
const char *pat, int new_ifindex);
|
|
|
|
static inline
|
2021-04-05 15:12:23 +08:00
|
|
|
int dev_change_net_namespace(struct net_device *dev, struct net *net,
|
2021-04-07 14:40:51 +08:00
|
|
|
const char *pat)
|
|
|
|
{
|
|
|
|
return __dev_change_net_namespace(dev, net, pat, 0);
|
|
|
|
}
|
2017-07-07 06:01:57 +08:00
|
|
|
int __dev_set_mtu(struct net_device *, int);
|
2013-09-27 05:48:15 +08:00
|
|
|
int dev_set_mtu(struct net_device *, int);
|
2018-12-13 19:54:35 +08:00
|
|
|
int dev_pre_changeaddr_notify(struct net_device *dev, const char *addr,
|
|
|
|
struct netlink_ext_ack *extack);
|
2018-12-13 19:54:30 +08:00
|
|
|
int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa,
|
|
|
|
struct netlink_ext_ack *extack);
|
net: fix dev_ifsioc_locked() race condition
dev_ifsioc_locked() is called with only RCU read lock, so when
there is a parallel writer changing the mac address, it could
get a partially updated mac address, as shown below:
Thread 1 Thread 2
// eth_commit_mac_addr_change()
memcpy(dev->dev_addr, addr->sa_data, ETH_ALEN);
// dev_ifsioc_locked()
memcpy(ifr->ifr_hwaddr.sa_data,
dev->dev_addr,...);
Close this race condition by guarding them with a RW semaphore,
like netdev_get_name(). We can not use seqlock here as it does not
allow blocking. The writers already take RTNL anyway, so this does
not affect the slow path. To avoid bothering existing
dev_set_mac_address() callers in drivers, introduce a new wrapper
just for user-facing callers on ioctl and rtnetlink paths.
Note, bonding also changes slave mac addresses but that requires
a separate patch due to the complexity of bonding code.
Fixes: 3710becf8a58 ("net: RCU locking for simple ioctl()")
Reported-by: "Gong, Sishuai" <sishuai@purdue.edu>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 03:34:10 +08:00
|
|
|
int dev_set_mac_address_user(struct net_device *dev, struct sockaddr *sa,
|
|
|
|
struct netlink_ext_ack *extack);
|
|
|
|
int dev_get_mac_address(struct sockaddr *sa, struct net *net, char *dev_name);
|
2019-02-07 01:45:35 +08:00
|
|
|
int dev_get_port_parent_id(struct net_device *dev,
|
|
|
|
struct netdev_phys_item_id *ppid, bool recurse);
|
|
|
|
bool netdev_port_same_parent_id(struct net_device *a, struct net_device *b);
|
2017-12-20 17:41:36 +08:00
|
|
|
struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *dev, bool *again);
|
2014-08-31 10:22:20 +08:00
|
|
|
struct sk_buff *dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
|
|
|
|
struct netdev_queue *txq, int *ret);
|
xdp: refine xdp api with regards to generic xdp
While working on the iproute2 generic XDP frontend, I noticed that
as of right now it's possible to have native *and* generic XDP
programs loaded both at the same time for the case when a driver
supports native XDP.
The intended model for generic XDP from b5cdae3291f7 ("net: Generic
XDP") is, however, that only one out of the two can be present at
once which is also indicated as such in the XDP netlink dump part.
The main rationale for generic XDP is to ease accessibility (in
case a driver does not yet have XDP support) and to generically
provide a semantical model as an example for driver developers
wanting to add XDP support. The generic XDP option for an XDP
aware driver can still be useful for comparing and testing both
implementations.
However, it is not intended to have a second XDP processing stage
or layer with exactly the same functionality of the first native
stage. Only reason could be to have a partial fallback for future
XDP features that are not supported yet in the native implementation
and we probably also shouldn't strive for such fallback and instead
encourage native feature support in the first place. Given there's
currently no such fallback issue or use case, lets not go there yet
if we don't need to.
Therefore, change semantics for loading XDP and bail out if the
user tries to load a generic XDP program when a native one is
present and vice versa. Another alternative to bailing out would
be to handle the transition from one flavor to another gracefully,
but that would require to bring the device down, exchange both
types of programs, and bring it up again in order to avoid a tiny
window where a packet could hit both hooks. Given this complicates
the logic for just a debugging feature in the native case, I went
with the simpler variant.
For the dump, remove IFLA_XDP_FLAGS that was added with b5cdae3291f7
and reuse IFLA_XDP_ATTACHED for indicating the mode. Dumping all
or just a subset of flags that were used for loading the XDP prog
is suboptimal in the long run since not all flags are useful for
dumping and if we start to reuse the same flag definitions for
load and dump, then we'll waste bit space. What we really just
want is to dump the mode for now.
Current IFLA_XDP_ATTACHED semantics are: nothing was installed (0),
a program is running at the native driver layer (1). Thus, add a
mode that says that a program is running at generic XDP layer (2).
Applications will handle this fine in that older binaries will
just indicate that something is attached at XDP layer, effectively
this is similar to IFLA_XDP_FLAGS attr that we would have had
modulo the redundancy.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-05-12 07:04:46 +08:00
|
|
|
|
2020-07-22 14:45:57 +08:00
|
|
|
int bpf_xdp_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
|
2021-07-31 13:57:33 +08:00
|
|
|
u8 dev_xdp_prog_count(struct net_device *dev);
|
2020-07-22 14:45:55 +08:00
|
|
|
u32 dev_xdp_prog_id(struct net_device *dev, enum bpf_xdp_mode mode);
|
2020-07-22 14:45:57 +08:00
|
|
|
|
2014-04-17 13:45:03 +08:00
|
|
|
int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
|
2013-09-27 05:48:15 +08:00
|
|
|
int dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
|
2021-02-09 21:38:29 +08:00
|
|
|
int dev_forward_skb_nomtu(struct net_device *dev, struct sk_buff *skb);
|
2016-04-28 23:59:28 +08:00
|
|
|
bool is_skb_forwardable(const struct net_device *dev,
|
|
|
|
const struct sk_buff *skb);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2021-02-09 21:38:29 +08:00
|
|
|
static __always_inline bool __is_skb_forwardable(const struct net_device *dev,
|
|
|
|
const struct sk_buff *skb,
|
|
|
|
const bool check_mtu)
|
|
|
|
{
|
|
|
|
const u32 vlan_hdr_len = 4; /* VLAN_HLEN */
|
|
|
|
unsigned int len;
|
|
|
|
|
|
|
|
if (!(dev->flags & IFF_UP))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
if (!check_mtu)
|
|
|
|
return true;
|
|
|
|
|
|
|
|
len = dev->mtu + dev->hard_header_len + vlan_hdr_len;
|
|
|
|
if (skb->len <= len)
|
|
|
|
return true;
|
|
|
|
|
|
|
|
/* if TSO is enabled, we don't care about the length as the packet
|
|
|
|
* could be forwarded without being segmented before
|
|
|
|
*/
|
|
|
|
if (skb_is_gso(skb))
|
|
|
|
return true;
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
net: Use this_cpu_inc() to increment net->core_stats
The macro dev_core_stats_##FIELD##_inc() disables preemption and invokes
netdev_core_stats_alloc() to return a per-CPU pointer.
netdev_core_stats_alloc() will allocate memory on its first invocation
which breaks on PREEMPT_RT because it requires non-atomic context for
memory allocation.
This can be avoided by enabling preemption in netdev_core_stats_alloc()
assuming the caller always disables preemption.
It might be better to replace local_inc() with this_cpu_inc() now that
dev_core_stats_##FIELD##_inc() gained a preempt-disable section and does
not rely on already disabled preemption. This results in less
instructions on x86-64:
local_inc:
| incl %gs:__preempt_count(%rip) # __preempt_count
| movq 488(%rdi), %rax # _1->core_stats, _22
| testq %rax, %rax # _22
| je .L585 #,
| add %gs:this_cpu_off(%rip), %rax # this_cpu_off, tcp_ptr__
| .L586:
| testq %rax, %rax # _27
| je .L587 #,
| incq (%rax) # _6->a.counter
| .L587:
| decl %gs:__preempt_count(%rip) # __preempt_count
this_cpu_inc(), this patch:
| movq 488(%rdi), %rax # _1->core_stats, _5
| testq %rax, %rax # _5
| je .L591 #,
| .L585:
| incq %gs:(%rax) # _18->rx_dropped
Use unsigned long as type for the counter. Use this_cpu_inc() to
increment the counter. Use a plain read of the counter.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/YmbO0pxgtKpCw4SY@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-04-26 00:39:46 +08:00
|
|
|
struct net_device_core_stats __percpu *netdev_core_stats_alloc(struct net_device *dev);
|
2022-03-11 13:14:20 +08:00
|
|
|
|
net: Use this_cpu_inc() to increment net->core_stats
The macro dev_core_stats_##FIELD##_inc() disables preemption and invokes
netdev_core_stats_alloc() to return a per-CPU pointer.
netdev_core_stats_alloc() will allocate memory on its first invocation
which breaks on PREEMPT_RT because it requires non-atomic context for
memory allocation.
This can be avoided by enabling preemption in netdev_core_stats_alloc()
assuming the caller always disables preemption.
It might be better to replace local_inc() with this_cpu_inc() now that
dev_core_stats_##FIELD##_inc() gained a preempt-disable section and does
not rely on already disabled preemption. This results in less
instructions on x86-64:
local_inc:
| incl %gs:__preempt_count(%rip) # __preempt_count
| movq 488(%rdi), %rax # _1->core_stats, _22
| testq %rax, %rax # _22
| je .L585 #,
| add %gs:this_cpu_off(%rip), %rax # this_cpu_off, tcp_ptr__
| .L586:
| testq %rax, %rax # _27
| je .L587 #,
| incq (%rax) # _6->a.counter
| .L587:
| decl %gs:__preempt_count(%rip) # __preempt_count
this_cpu_inc(), this patch:
| movq 488(%rdi), %rax # _1->core_stats, _5
| testq %rax, %rax # _5
| je .L591 #,
| .L585:
| incq %gs:(%rax) # _18->rx_dropped
Use unsigned long as type for the counter. Use this_cpu_inc() to
increment the counter. Use a plain read of the counter.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/YmbO0pxgtKpCw4SY@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-04-26 00:39:46 +08:00
|
|
|
static inline struct net_device_core_stats __percpu *dev_core_stats(struct net_device *dev)
|
2022-03-11 13:14:20 +08:00
|
|
|
{
|
|
|
|
/* This READ_ONCE() pairs with the write in netdev_core_stats_alloc() */
|
|
|
|
struct net_device_core_stats __percpu *p = READ_ONCE(dev->core_stats);
|
|
|
|
|
|
|
|
if (likely(p))
|
net: Use this_cpu_inc() to increment net->core_stats
The macro dev_core_stats_##FIELD##_inc() disables preemption and invokes
netdev_core_stats_alloc() to return a per-CPU pointer.
netdev_core_stats_alloc() will allocate memory on its first invocation
which breaks on PREEMPT_RT because it requires non-atomic context for
memory allocation.
This can be avoided by enabling preemption in netdev_core_stats_alloc()
assuming the caller always disables preemption.
It might be better to replace local_inc() with this_cpu_inc() now that
dev_core_stats_##FIELD##_inc() gained a preempt-disable section and does
not rely on already disabled preemption. This results in less
instructions on x86-64:
local_inc:
| incl %gs:__preempt_count(%rip) # __preempt_count
| movq 488(%rdi), %rax # _1->core_stats, _22
| testq %rax, %rax # _22
| je .L585 #,
| add %gs:this_cpu_off(%rip), %rax # this_cpu_off, tcp_ptr__
| .L586:
| testq %rax, %rax # _27
| je .L587 #,
| incq (%rax) # _6->a.counter
| .L587:
| decl %gs:__preempt_count(%rip) # __preempt_count
this_cpu_inc(), this patch:
| movq 488(%rdi), %rax # _1->core_stats, _5
| testq %rax, %rax # _5
| je .L591 #,
| .L585:
| incq %gs:(%rax) # _18->rx_dropped
Use unsigned long as type for the counter. Use this_cpu_inc() to
increment the counter. Use a plain read of the counter.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/YmbO0pxgtKpCw4SY@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-04-26 00:39:46 +08:00
|
|
|
return p;
|
2022-03-11 13:14:20 +08:00
|
|
|
|
|
|
|
return netdev_core_stats_alloc(dev);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define DEV_CORE_STATS_INC(FIELD) \
|
|
|
|
static inline void dev_core_stats_##FIELD##_inc(struct net_device *dev) \
|
|
|
|
{ \
|
net: Use this_cpu_inc() to increment net->core_stats
The macro dev_core_stats_##FIELD##_inc() disables preemption and invokes
netdev_core_stats_alloc() to return a per-CPU pointer.
netdev_core_stats_alloc() will allocate memory on its first invocation
which breaks on PREEMPT_RT because it requires non-atomic context for
memory allocation.
This can be avoided by enabling preemption in netdev_core_stats_alloc()
assuming the caller always disables preemption.
It might be better to replace local_inc() with this_cpu_inc() now that
dev_core_stats_##FIELD##_inc() gained a preempt-disable section and does
not rely on already disabled preemption. This results in less
instructions on x86-64:
local_inc:
| incl %gs:__preempt_count(%rip) # __preempt_count
| movq 488(%rdi), %rax # _1->core_stats, _22
| testq %rax, %rax # _22
| je .L585 #,
| add %gs:this_cpu_off(%rip), %rax # this_cpu_off, tcp_ptr__
| .L586:
| testq %rax, %rax # _27
| je .L587 #,
| incq (%rax) # _6->a.counter
| .L587:
| decl %gs:__preempt_count(%rip) # __preempt_count
this_cpu_inc(), this patch:
| movq 488(%rdi), %rax # _1->core_stats, _5
| testq %rax, %rax # _5
| je .L591 #,
| .L585:
| incq %gs:(%rax) # _18->rx_dropped
Use unsigned long as type for the counter. Use this_cpu_inc() to
increment the counter. Use a plain read of the counter.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/YmbO0pxgtKpCw4SY@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-04-26 00:39:46 +08:00
|
|
|
struct net_device_core_stats __percpu *p; \
|
2022-03-13 05:45:05 +08:00
|
|
|
\
|
|
|
|
p = dev_core_stats(dev); \
|
2022-03-11 13:14:20 +08:00
|
|
|
if (p) \
|
net: Use this_cpu_inc() to increment net->core_stats
The macro dev_core_stats_##FIELD##_inc() disables preemption and invokes
netdev_core_stats_alloc() to return a per-CPU pointer.
netdev_core_stats_alloc() will allocate memory on its first invocation
which breaks on PREEMPT_RT because it requires non-atomic context for
memory allocation.
This can be avoided by enabling preemption in netdev_core_stats_alloc()
assuming the caller always disables preemption.
It might be better to replace local_inc() with this_cpu_inc() now that
dev_core_stats_##FIELD##_inc() gained a preempt-disable section and does
not rely on already disabled preemption. This results in less
instructions on x86-64:
local_inc:
| incl %gs:__preempt_count(%rip) # __preempt_count
| movq 488(%rdi), %rax # _1->core_stats, _22
| testq %rax, %rax # _22
| je .L585 #,
| add %gs:this_cpu_off(%rip), %rax # this_cpu_off, tcp_ptr__
| .L586:
| testq %rax, %rax # _27
| je .L587 #,
| incq (%rax) # _6->a.counter
| .L587:
| decl %gs:__preempt_count(%rip) # __preempt_count
this_cpu_inc(), this patch:
| movq 488(%rdi), %rax # _1->core_stats, _5
| testq %rax, %rax # _5
| je .L591 #,
| .L585:
| incq %gs:(%rax) # _18->rx_dropped
Use unsigned long as type for the counter. Use this_cpu_inc() to
increment the counter. Use a plain read of the counter.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/YmbO0pxgtKpCw4SY@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-04-26 00:39:46 +08:00
|
|
|
this_cpu_inc(p->FIELD); \
|
2022-03-11 13:14:20 +08:00
|
|
|
}
|
|
|
|
DEV_CORE_STATS_INC(rx_dropped)
|
|
|
|
DEV_CORE_STATS_INC(tx_dropped)
|
|
|
|
DEV_CORE_STATS_INC(rx_nohandler)
|
2022-04-07 01:26:00 +08:00
|
|
|
DEV_CORE_STATS_INC(rx_otherhost_dropped)
|
2022-03-11 13:14:20 +08:00
|
|
|
|
2016-11-10 07:36:33 +08:00
|
|
|
static __always_inline int ____dev_forward_skb(struct net_device *dev,
|
2021-02-09 21:38:29 +08:00
|
|
|
struct sk_buff *skb,
|
|
|
|
const bool check_mtu)
|
2016-11-10 07:36:33 +08:00
|
|
|
{
|
|
|
|
if (skb_orphan_frags(skb, GFP_ATOMIC) ||
|
2021-02-09 21:38:29 +08:00
|
|
|
unlikely(!__is_skb_forwardable(dev, skb, check_mtu))) {
|
2022-03-11 13:14:20 +08:00
|
|
|
dev_core_stats_rx_dropped_inc(dev);
|
2016-11-10 07:36:33 +08:00
|
|
|
kfree_skb(skb);
|
|
|
|
return NET_RX_DROP;
|
|
|
|
}
|
|
|
|
|
2021-06-24 16:05:05 +08:00
|
|
|
skb_scrub_packet(skb, !net_eq(dev_net(dev), dev_net(skb->dev)));
|
2016-11-10 07:36:33 +08:00
|
|
|
skb->priority = 0;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2018-10-10 06:57:49 +08:00
|
|
|
bool dev_nit_active(struct net_device *dev);
|
2016-05-11 02:19:50 +08:00
|
|
|
void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev);
|
|
|
|
|
2022-02-05 06:42:37 +08:00
|
|
|
static inline void __dev_put(struct net_device *dev)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
2021-08-05 19:54:34 +08:00
|
|
|
if (dev) {
|
2021-03-20 01:39:33 +08:00
|
|
|
#ifdef CONFIG_PCPU_DEV_REFCNT
|
2021-08-05 19:54:34 +08:00
|
|
|
this_cpu_dec(*dev->pcpu_refcnt);
|
2021-03-20 01:39:33 +08:00
|
|
|
#else
|
2021-08-05 19:54:34 +08:00
|
|
|
refcount_dec(&dev->dev_refcnt);
|
2021-03-20 01:39:33 +08:00
|
|
|
#endif
|
2021-08-05 19:54:34 +08:00
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
2022-02-05 06:42:37 +08:00
|
|
|
static inline void __dev_hold(struct net_device *dev)
|
2006-03-21 14:32:28 +08:00
|
|
|
{
|
2021-08-05 19:54:34 +08:00
|
|
|
if (dev) {
|
2021-03-20 01:39:33 +08:00
|
|
|
#ifdef CONFIG_PCPU_DEV_REFCNT
|
2021-08-05 19:54:34 +08:00
|
|
|
this_cpu_inc(*dev->pcpu_refcnt);
|
2021-03-20 01:39:33 +08:00
|
|
|
#else
|
2021-08-05 19:54:34 +08:00
|
|
|
refcount_inc(&dev->dev_refcnt);
|
2021-03-20 01:39:33 +08:00
|
|
|
#endif
|
2021-08-05 19:54:34 +08:00
|
|
|
}
|
2006-03-21 14:32:28 +08:00
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2022-02-05 06:42:37 +08:00
|
|
|
static inline void __netdev_tracker_alloc(struct net_device *dev,
|
|
|
|
netdevice_tracker *tracker,
|
|
|
|
gfp_t gfp)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_NET_DEV_REFCNT_TRACKER
|
|
|
|
ref_tracker_alloc(&dev->refcnt_tracker, tracker, gfp);
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
/* netdev_tracker_alloc() can upgrade a prior untracked reference
|
|
|
|
* taken by dev_get_by_name()/dev_get_by_index() to a tracked one.
|
|
|
|
*/
|
2021-12-05 12:21:57 +08:00
|
|
|
static inline void netdev_tracker_alloc(struct net_device *dev,
|
|
|
|
netdevice_tracker *tracker, gfp_t gfp)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_NET_DEV_REFCNT_TRACKER
|
2022-02-05 06:42:37 +08:00
|
|
|
refcount_dec(&dev->refcnt_tracker.no_tracker);
|
|
|
|
__netdev_tracker_alloc(dev, tracker, gfp);
|
2021-12-05 12:21:57 +08:00
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void netdev_tracker_free(struct net_device *dev,
|
|
|
|
netdevice_tracker *tracker)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_NET_DEV_REFCNT_TRACKER
|
|
|
|
ref_tracker_free(&dev->refcnt_tracker, tracker);
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2022-06-08 12:39:55 +08:00
|
|
|
static inline void netdev_hold(struct net_device *dev,
|
|
|
|
netdevice_tracker *tracker, gfp_t gfp)
|
2021-12-05 12:21:57 +08:00
|
|
|
{
|
|
|
|
if (dev) {
|
2022-02-05 06:42:37 +08:00
|
|
|
__dev_hold(dev);
|
|
|
|
__netdev_tracker_alloc(dev, tracker, gfp);
|
2021-12-05 12:21:57 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-06-08 12:39:55 +08:00
|
|
|
static inline void netdev_put(struct net_device *dev,
|
|
|
|
netdevice_tracker *tracker)
|
2021-12-05 12:21:57 +08:00
|
|
|
{
|
|
|
|
if (dev) {
|
|
|
|
netdev_tracker_free(dev, tracker);
|
2022-02-05 06:42:37 +08:00
|
|
|
__dev_put(dev);
|
2021-12-05 12:21:57 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-02-05 06:42:37 +08:00
|
|
|
/**
|
|
|
|
* dev_hold - get reference to device
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Hold reference to device to keep it from being freed.
|
2022-06-08 12:39:55 +08:00
|
|
|
* Try using netdev_hold() instead.
|
2022-02-05 06:42:37 +08:00
|
|
|
*/
|
|
|
|
static inline void dev_hold(struct net_device *dev)
|
|
|
|
{
|
2022-06-08 12:39:55 +08:00
|
|
|
netdev_hold(dev, NULL, GFP_ATOMIC);
|
2022-02-05 06:42:37 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* dev_put - release reference to device
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Release reference to device to allow it to be freed.
|
2022-06-08 12:39:55 +08:00
|
|
|
* Try using netdev_put() instead.
|
2022-02-05 06:42:37 +08:00
|
|
|
*/
|
|
|
|
static inline void dev_put(struct net_device *dev)
|
|
|
|
{
|
2022-06-08 12:39:55 +08:00
|
|
|
netdev_put(dev, NULL);
|
2022-02-05 06:42:37 +08:00
|
|
|
}
|
|
|
|
|
2022-06-08 12:39:55 +08:00
|
|
|
static inline void netdev_ref_replace(struct net_device *odev,
|
|
|
|
struct net_device *ndev,
|
|
|
|
netdevice_tracker *tracker,
|
|
|
|
gfp_t gfp)
|
2021-12-05 12:22:03 +08:00
|
|
|
{
|
|
|
|
if (odev)
|
2021-12-14 23:15:15 +08:00
|
|
|
netdev_tracker_free(odev, tracker);
|
|
|
|
|
2022-02-05 06:42:37 +08:00
|
|
|
__dev_hold(ndev);
|
|
|
|
__dev_put(odev);
|
2021-12-14 23:15:15 +08:00
|
|
|
|
2021-12-05 12:22:03 +08:00
|
|
|
if (ndev)
|
2022-02-05 06:42:37 +08:00
|
|
|
__netdev_tracker_alloc(ndev, tracker, gfp);
|
2021-12-05 12:22:03 +08:00
|
|
|
}
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/* Carrier loss detection, dial on demand. The functions netif_carrier_on
|
|
|
|
* and _off may be called from IRQ context, but it is caller
|
|
|
|
* who is responsible for serialization of these calls.
|
2006-03-21 09:09:11 +08:00
|
|
|
*
|
|
|
|
* The name carrier is inappropriate, these functions should really be
|
|
|
|
* called netif_lowerlayer_*() because they represent the state of any
|
|
|
|
* kind of lower layer not just hardware media.
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
2013-09-27 05:48:15 +08:00
|
|
|
void linkwatch_fire_event(struct net_device *dev);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netif_carrier_ok - test if carrier present
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Check if carrier is present on device
|
|
|
|
*/
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool netif_carrier_ok(const struct net_device *dev)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
return !test_bit(__LINK_STATE_NOCARRIER, &dev->state);
|
|
|
|
}
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
unsigned long dev_trans_start(struct net_device *dev);
|
2009-05-18 11:55:16 +08:00
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
void __netdev_watchdog_up(struct net_device *dev);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
void netif_carrier_on(struct net_device *dev);
|
|
|
|
void netif_carrier_off(struct net_device *dev);
|
2021-05-20 01:18:25 +08:00
|
|
|
void netif_carrier_event(struct net_device *dev);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netif_dormant_on - mark device as dormant.
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Mark device as dormant (as per RFC2863).
|
|
|
|
*
|
|
|
|
* The dormant state indicates that the relevant interface is not
|
|
|
|
* actually in a condition to pass packets (i.e., it is not 'up') but is
|
|
|
|
* in a "pending" state, waiting for some external event. For "on-
|
|
|
|
* demand" interfaces, this new state identifies the situation where the
|
|
|
|
* interface is waiting for events to place it in the up state.
|
|
|
|
*/
|
2006-03-21 09:09:11 +08:00
|
|
|
static inline void netif_dormant_on(struct net_device *dev)
|
|
|
|
{
|
|
|
|
if (!test_and_set_bit(__LINK_STATE_DORMANT, &dev->state))
|
|
|
|
linkwatch_fire_event(dev);
|
|
|
|
}
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netif_dormant_off - set device as not dormant.
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Device is not in dormant state.
|
|
|
|
*/
|
2006-03-21 09:09:11 +08:00
|
|
|
static inline void netif_dormant_off(struct net_device *dev)
|
|
|
|
{
|
|
|
|
if (test_and_clear_bit(__LINK_STATE_DORMANT, &dev->state))
|
|
|
|
linkwatch_fire_event(dev);
|
|
|
|
}
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
2017-04-26 11:05:12 +08:00
|
|
|
* netif_dormant - test if device is dormant
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
* @dev: network device
|
|
|
|
*
|
2017-04-26 11:05:12 +08:00
|
|
|
* Check if device is dormant.
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
*/
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool netif_dormant(const struct net_device *dev)
|
2006-03-21 09:09:11 +08:00
|
|
|
{
|
|
|
|
return test_bit(__LINK_STATE_DORMANT, &dev->state);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2020-04-20 06:11:50 +08:00
|
|
|
/**
|
|
|
|
* netif_testing_on - mark device as under test.
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Mark device as under test (as per RFC2863).
|
|
|
|
*
|
|
|
|
* The testing state indicates that some test(s) must be performed on
|
|
|
|
* the interface. After completion, of the test, the interface state
|
|
|
|
* will change to up, dormant, or down, as appropriate.
|
|
|
|
*/
|
|
|
|
static inline void netif_testing_on(struct net_device *dev)
|
|
|
|
{
|
|
|
|
if (!test_and_set_bit(__LINK_STATE_TESTING, &dev->state))
|
|
|
|
linkwatch_fire_event(dev);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* netif_testing_off - set device as not under test.
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Device is not in testing state.
|
|
|
|
*/
|
|
|
|
static inline void netif_testing_off(struct net_device *dev)
|
|
|
|
{
|
|
|
|
if (test_and_clear_bit(__LINK_STATE_TESTING, &dev->state))
|
|
|
|
linkwatch_fire_event(dev);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* netif_testing - test if device is under test
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Check if device is under test
|
|
|
|
*/
|
|
|
|
static inline bool netif_testing(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return test_bit(__LINK_STATE_TESTING, &dev->state);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netif_oper_up - test if device is operational
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Check if carrier is operational
|
|
|
|
*/
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool netif_oper_up(const struct net_device *dev)
|
2009-11-05 01:50:58 +08:00
|
|
|
{
|
2006-03-21 09:09:11 +08:00
|
|
|
return (dev->operstate == IF_OPER_UP ||
|
|
|
|
dev->operstate == IF_OPER_UNKNOWN /* backward compat */);
|
|
|
|
}
|
|
|
|
|
[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
|
|
|
/**
|
|
|
|
* netif_device_present - is device available or removed
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Check if device has not been removed from system.
|
|
|
|
*/
|
2020-09-16 15:11:20 +08:00
|
|
|
static inline bool netif_device_present(const struct net_device *dev)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
return test_bit(__LINK_STATE_PRESENT, &dev->state);
|
|
|
|
}
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
void netif_device_detach(struct net_device *dev);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
void netif_device_attach(struct net_device *dev);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Network interface message level settings
|
|
|
|
*/
|
|
|
|
|
|
|
|
enum {
|
2020-01-27 06:11:04 +08:00
|
|
|
NETIF_MSG_DRV_BIT,
|
|
|
|
NETIF_MSG_PROBE_BIT,
|
|
|
|
NETIF_MSG_LINK_BIT,
|
|
|
|
NETIF_MSG_TIMER_BIT,
|
|
|
|
NETIF_MSG_IFDOWN_BIT,
|
|
|
|
NETIF_MSG_IFUP_BIT,
|
|
|
|
NETIF_MSG_RX_ERR_BIT,
|
|
|
|
NETIF_MSG_TX_ERR_BIT,
|
|
|
|
NETIF_MSG_TX_QUEUED_BIT,
|
|
|
|
NETIF_MSG_INTR_BIT,
|
|
|
|
NETIF_MSG_TX_DONE_BIT,
|
|
|
|
NETIF_MSG_RX_STATUS_BIT,
|
|
|
|
NETIF_MSG_PKTDATA_BIT,
|
|
|
|
NETIF_MSG_HW_BIT,
|
|
|
|
NETIF_MSG_WOL_BIT,
|
|
|
|
|
|
|
|
/* When you add a new bit above, update netif_msg_class_names array
|
|
|
|
* in net/ethtool/common.c
|
|
|
|
*/
|
|
|
|
NETIF_MSG_CLASS_COUNT,
|
2005-04-17 06:20:36 +08:00
|
|
|
};
|
2020-01-27 06:11:04 +08:00
|
|
|
/* Both ethtool_ops interface and internal driver implementation use u32 */
|
|
|
|
static_assert(NETIF_MSG_CLASS_COUNT <= 32);
|
|
|
|
|
|
|
|
#define __NETIF_MSG_BIT(bit) ((u32)1 << (bit))
|
|
|
|
#define __NETIF_MSG(name) __NETIF_MSG_BIT(NETIF_MSG_ ## name ## _BIT)
|
|
|
|
|
|
|
|
#define NETIF_MSG_DRV __NETIF_MSG(DRV)
|
|
|
|
#define NETIF_MSG_PROBE __NETIF_MSG(PROBE)
|
|
|
|
#define NETIF_MSG_LINK __NETIF_MSG(LINK)
|
|
|
|
#define NETIF_MSG_TIMER __NETIF_MSG(TIMER)
|
|
|
|
#define NETIF_MSG_IFDOWN __NETIF_MSG(IFDOWN)
|
|
|
|
#define NETIF_MSG_IFUP __NETIF_MSG(IFUP)
|
|
|
|
#define NETIF_MSG_RX_ERR __NETIF_MSG(RX_ERR)
|
|
|
|
#define NETIF_MSG_TX_ERR __NETIF_MSG(TX_ERR)
|
|
|
|
#define NETIF_MSG_TX_QUEUED __NETIF_MSG(TX_QUEUED)
|
|
|
|
#define NETIF_MSG_INTR __NETIF_MSG(INTR)
|
|
|
|
#define NETIF_MSG_TX_DONE __NETIF_MSG(TX_DONE)
|
|
|
|
#define NETIF_MSG_RX_STATUS __NETIF_MSG(RX_STATUS)
|
|
|
|
#define NETIF_MSG_PKTDATA __NETIF_MSG(PKTDATA)
|
|
|
|
#define NETIF_MSG_HW __NETIF_MSG(HW)
|
|
|
|
#define NETIF_MSG_WOL __NETIF_MSG(WOL)
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
#define netif_msg_drv(p) ((p)->msg_enable & NETIF_MSG_DRV)
|
|
|
|
#define netif_msg_probe(p) ((p)->msg_enable & NETIF_MSG_PROBE)
|
|
|
|
#define netif_msg_link(p) ((p)->msg_enable & NETIF_MSG_LINK)
|
|
|
|
#define netif_msg_timer(p) ((p)->msg_enable & NETIF_MSG_TIMER)
|
|
|
|
#define netif_msg_ifdown(p) ((p)->msg_enable & NETIF_MSG_IFDOWN)
|
|
|
|
#define netif_msg_ifup(p) ((p)->msg_enable & NETIF_MSG_IFUP)
|
|
|
|
#define netif_msg_rx_err(p) ((p)->msg_enable & NETIF_MSG_RX_ERR)
|
|
|
|
#define netif_msg_tx_err(p) ((p)->msg_enable & NETIF_MSG_TX_ERR)
|
|
|
|
#define netif_msg_tx_queued(p) ((p)->msg_enable & NETIF_MSG_TX_QUEUED)
|
|
|
|
#define netif_msg_intr(p) ((p)->msg_enable & NETIF_MSG_INTR)
|
|
|
|
#define netif_msg_tx_done(p) ((p)->msg_enable & NETIF_MSG_TX_DONE)
|
|
|
|
#define netif_msg_rx_status(p) ((p)->msg_enable & NETIF_MSG_RX_STATUS)
|
|
|
|
#define netif_msg_pktdata(p) ((p)->msg_enable & NETIF_MSG_PKTDATA)
|
|
|
|
#define netif_msg_hw(p) ((p)->msg_enable & NETIF_MSG_HW)
|
|
|
|
#define netif_msg_wol(p) ((p)->msg_enable & NETIF_MSG_WOL)
|
|
|
|
|
|
|
|
static inline u32 netif_msg_init(int debug_value, int default_msg_enable_bits)
|
|
|
|
{
|
|
|
|
/* use default */
|
|
|
|
if (debug_value < 0 || debug_value >= (sizeof(u32) * 8))
|
|
|
|
return default_msg_enable_bits;
|
|
|
|
if (debug_value == 0) /* no output */
|
|
|
|
return 0;
|
|
|
|
/* set low N bits */
|
2019-02-27 18:37:26 +08:00
|
|
|
return (1U << debug_value) - 1;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
2008-07-09 14:13:53 +08:00
|
|
|
static inline void __netif_tx_lock(struct netdev_queue *txq, int cpu)
|
2006-06-10 03:20:56 +08:00
|
|
|
{
|
2008-07-09 14:13:53 +08:00
|
|
|
spin_lock(&txq->_xmit_lock);
|
2021-12-01 01:01:55 +08:00
|
|
|
/* Pairs with READ_ONCE() in __dev_queue_xmit() */
|
|
|
|
WRITE_ONCE(txq->xmit_lock_owner, cpu);
|
2007-09-17 05:40:49 +08:00
|
|
|
}
|
|
|
|
|
2016-11-24 13:04:08 +08:00
|
|
|
static inline bool __netif_tx_acquire(struct netdev_queue *txq)
|
|
|
|
{
|
|
|
|
__acquire(&txq->_xmit_lock);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void __netif_tx_release(struct netdev_queue *txq)
|
|
|
|
{
|
|
|
|
__release(&txq->_xmit_lock);
|
|
|
|
}
|
|
|
|
|
2008-07-17 16:56:23 +08:00
|
|
|
static inline void __netif_tx_lock_bh(struct netdev_queue *txq)
|
|
|
|
{
|
|
|
|
spin_lock_bh(&txq->_xmit_lock);
|
2021-12-01 01:01:55 +08:00
|
|
|
/* Pairs with READ_ONCE() in __dev_queue_xmit() */
|
|
|
|
WRITE_ONCE(txq->xmit_lock_owner, smp_processor_id());
|
2008-07-17 16:56:23 +08:00
|
|
|
}
|
|
|
|
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool __netif_tx_trylock(struct netdev_queue *txq)
|
2008-08-01 07:58:50 +08:00
|
|
|
{
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
bool ok = spin_trylock(&txq->_xmit_lock);
|
2021-12-01 01:01:55 +08:00
|
|
|
|
|
|
|
if (likely(ok)) {
|
|
|
|
/* Pairs with READ_ONCE() in __dev_queue_xmit() */
|
|
|
|
WRITE_ONCE(txq->xmit_lock_owner, smp_processor_id());
|
|
|
|
}
|
2008-08-01 07:58:50 +08:00
|
|
|
return ok;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void __netif_tx_unlock(struct netdev_queue *txq)
|
|
|
|
{
|
2021-12-01 01:01:55 +08:00
|
|
|
/* Pairs with READ_ONCE() in __dev_queue_xmit() */
|
|
|
|
WRITE_ONCE(txq->xmit_lock_owner, -1);
|
2008-08-01 07:58:50 +08:00
|
|
|
spin_unlock(&txq->_xmit_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void __netif_tx_unlock_bh(struct netdev_queue *txq)
|
|
|
|
{
|
2021-12-01 01:01:55 +08:00
|
|
|
/* Pairs with READ_ONCE() in __dev_queue_xmit() */
|
|
|
|
WRITE_ONCE(txq->xmit_lock_owner, -1);
|
2008-08-01 07:58:50 +08:00
|
|
|
spin_unlock_bh(&txq->_xmit_lock);
|
|
|
|
}
|
|
|
|
|
2021-11-17 11:29:22 +08:00
|
|
|
/*
|
|
|
|
* txq->trans_start can be read locklessly from dev_watchdog()
|
|
|
|
*/
|
2009-05-26 13:58:01 +08:00
|
|
|
static inline void txq_trans_update(struct netdev_queue *txq)
|
|
|
|
{
|
|
|
|
if (txq->xmit_lock_owner != -1)
|
2021-11-17 11:29:22 +08:00
|
|
|
WRITE_ONCE(txq->trans_start, jiffies);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void txq_trans_cond_update(struct netdev_queue *txq)
|
|
|
|
{
|
|
|
|
unsigned long now = jiffies;
|
|
|
|
|
|
|
|
if (READ_ONCE(txq->trans_start) != now)
|
|
|
|
WRITE_ONCE(txq->trans_start, now);
|
2009-05-26 13:58:01 +08:00
|
|
|
}
|
|
|
|
|
2016-05-03 22:31:00 +08:00
|
|
|
/* legacy drivers only, netdev_start_xmit() sets txq->trans_start */
|
|
|
|
static inline void netif_trans_update(struct net_device *dev)
|
|
|
|
{
|
2016-05-03 22:33:14 +08:00
|
|
|
struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
|
|
|
|
|
2021-11-17 11:29:22 +08:00
|
|
|
txq_trans_cond_update(txq);
|
2016-05-03 22:31:00 +08:00
|
|
|
}
|
|
|
|
|
2008-07-23 05:09:06 +08:00
|
|
|
/**
|
|
|
|
* netif_tx_lock - grab network device transmit lock
|
|
|
|
* @dev: network device
|
|
|
|
*
|
|
|
|
* Get network device transmit lock
|
|
|
|
*/
|
2021-11-17 11:29:23 +08:00
|
|
|
void netif_tx_lock(struct net_device *dev);
|
2006-06-10 03:20:56 +08:00
|
|
|
|
|
|
|
static inline void netif_tx_lock_bh(struct net_device *dev)
|
|
|
|
{
|
2008-07-17 15:34:19 +08:00
|
|
|
local_bh_disable();
|
|
|
|
netif_tx_lock(dev);
|
2006-06-10 03:20:56 +08:00
|
|
|
}
|
|
|
|
|
2021-11-17 11:29:23 +08:00
|
|
|
void netif_tx_unlock(struct net_device *dev);
|
2006-06-10 03:20:56 +08:00
|
|
|
|
|
|
|
static inline void netif_tx_unlock_bh(struct net_device *dev)
|
|
|
|
{
|
2008-07-17 15:34:19 +08:00
|
|
|
netif_tx_unlock(dev);
|
|
|
|
local_bh_enable();
|
2006-06-10 03:20:56 +08:00
|
|
|
}
|
|
|
|
|
2008-07-09 14:13:53 +08:00
|
|
|
#define HARD_TX_LOCK(dev, txq, cpu) { \
|
2007-09-17 05:40:49 +08:00
|
|
|
if ((dev->features & NETIF_F_LLTX) == 0) { \
|
2008-07-09 14:13:53 +08:00
|
|
|
__netif_tx_lock(txq, cpu); \
|
2016-11-24 13:04:08 +08:00
|
|
|
} else { \
|
|
|
|
__netif_tx_acquire(txq); \
|
2007-09-17 05:40:49 +08:00
|
|
|
} \
|
|
|
|
}
|
|
|
|
|
2014-03-28 06:42:20 +08:00
|
|
|
#define HARD_TX_TRYLOCK(dev, txq) \
|
|
|
|
(((dev->features & NETIF_F_LLTX) == 0) ? \
|
|
|
|
__netif_tx_trylock(txq) : \
|
2016-11-24 13:04:08 +08:00
|
|
|
__netif_tx_acquire(txq))
|
2014-03-28 06:42:20 +08:00
|
|
|
|
2008-07-09 14:13:53 +08:00
|
|
|
#define HARD_TX_UNLOCK(dev, txq) { \
|
2007-09-17 05:40:49 +08:00
|
|
|
if ((dev->features & NETIF_F_LLTX) == 0) { \
|
2008-07-09 14:13:53 +08:00
|
|
|
__netif_tx_unlock(txq); \
|
2016-11-24 13:04:08 +08:00
|
|
|
} else { \
|
|
|
|
__netif_tx_release(txq); \
|
2007-09-17 05:40:49 +08:00
|
|
|
} \
|
|
|
|
}
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
static inline void netif_tx_disable(struct net_device *dev)
|
|
|
|
{
|
2008-07-17 16:56:23 +08:00
|
|
|
unsigned int i;
|
2008-08-01 07:58:50 +08:00
|
|
|
int cpu;
|
2008-07-17 16:56:23 +08:00
|
|
|
|
2008-08-01 07:58:50 +08:00
|
|
|
local_bh_disable();
|
|
|
|
cpu = smp_processor_id();
|
2021-02-06 09:37:32 +08:00
|
|
|
spin_lock(&dev->tx_global_lock);
|
2008-07-17 16:56:23 +08:00
|
|
|
for (i = 0; i < dev->num_tx_queues; i++) {
|
|
|
|
struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
|
2008-08-01 07:58:50 +08:00
|
|
|
|
|
|
|
__netif_tx_lock(txq, cpu);
|
2008-07-17 16:56:23 +08:00
|
|
|
netif_tx_stop_queue(txq);
|
2008-08-01 07:58:50 +08:00
|
|
|
__netif_tx_unlock(txq);
|
2008-07-17 16:56:23 +08:00
|
|
|
}
|
2021-02-06 09:37:32 +08:00
|
|
|
spin_unlock(&dev->tx_global_lock);
|
2008-08-01 07:58:50 +08:00
|
|
|
local_bh_enable();
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
2008-07-15 15:13:44 +08:00
|
|
|
static inline void netif_addr_lock(struct net_device *dev)
|
|
|
|
{
|
net: core: add nested_level variable in net_device
This patch is to add a new variable 'nested_level' into the net_device
structure.
This variable will be used as a parameter of spin_lock_nested() of
dev->addr_list_lock.
netif_addr_lock() can be called recursively so spin_lock_nested() is
used instead of spin_lock() and dev->lower_level is used as a parameter
of spin_lock_nested().
But, dev->lower_level value can be updated while it is being used.
So, lockdep would warn a possible deadlock scenario.
When a stacked interface is deleted, netif_{uc | mc}_sync() is
called recursively.
So, spin_lock_nested() is called recursively too.
At this moment, the dev->lower_level variable is used as a parameter of it.
dev->lower_level value is updated when interfaces are being unlinked/linked
immediately.
Thus, After unlinking, dev->lower_level shouldn't be a parameter of
spin_lock_nested().
A (macvlan)
|
B (vlan)
|
C (bridge)
|
D (macvlan)
|
E (vlan)
|
F (bridge)
A->lower_level : 6
B->lower_level : 5
C->lower_level : 4
D->lower_level : 3
E->lower_level : 2
F->lower_level : 1
When an interface 'A' is removed, it releases resources.
At this moment, netif_addr_lock() would be called.
Then, netdev_upper_dev_unlink() is called recursively.
Then dev->lower_level is updated.
There is no problem.
But, when the bridge module is removed, 'C' and 'F' interfaces
are removed at once.
If 'F' is removed first, a lower_level value is like below.
A->lower_level : 5
B->lower_level : 4
C->lower_level : 3
D->lower_level : 2
E->lower_level : 1
F->lower_level : 1
Then, 'C' is removed. at this moment, netif_addr_lock() is called
recursively.
The ordering is like this.
C(3)->D(2)->E(1)->F(1)
At this moment, the lower_level value of 'E' and 'F' are the same.
So, lockdep warns a possible deadlock scenario.
In order to avoid this problem, a new variable 'nested_level' is added.
This value is the same as dev->lower_level - 1.
But this value is updated in rtnl_unlock().
So, this variable can be used as a parameter of spin_lock_nested() safely
in the rtnl context.
Test commands:
ip link add br0 type bridge vlan_filtering 1
ip link add vlan1 link br0 type vlan id 10
ip link add macvlan2 link vlan1 type macvlan
ip link add br3 type bridge vlan_filtering 1
ip link set macvlan2 master br3
ip link add vlan4 link br3 type vlan id 10
ip link add macvlan5 link vlan4 type macvlan
ip link add br6 type bridge vlan_filtering 1
ip link set macvlan5 master br6
ip link add vlan7 link br6 type vlan id 10
ip link add macvlan8 link vlan7 type macvlan
ip link set br0 up
ip link set vlan1 up
ip link set macvlan2 up
ip link set br3 up
ip link set vlan4 up
ip link set macvlan5 up
ip link set br6 up
ip link set vlan7 up
ip link set macvlan8 up
modprobe -rv bridge
Splat looks like:
[ 36.057436][ T744] WARNING: possible recursive locking detected
[ 36.058848][ T744] 5.9.0-rc6+ #728 Not tainted
[ 36.059959][ T744] --------------------------------------------
[ 36.061391][ T744] ip/744 is trying to acquire lock:
[ 36.062590][ T744] ffff8c4767509280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_set_rx_mode+0x19/0x30
[ 36.064922][ T744]
[ 36.064922][ T744] but task is already holding lock:
[ 36.066626][ T744] ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.068851][ T744]
[ 36.068851][ T744] other info that might help us debug this:
[ 36.070731][ T744] Possible unsafe locking scenario:
[ 36.070731][ T744]
[ 36.072497][ T744] CPU0
[ 36.073238][ T744] ----
[ 36.074007][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.075290][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.076590][ T744]
[ 36.076590][ T744] *** DEADLOCK ***
[ 36.076590][ T744]
[ 36.078515][ T744] May be due to missing lock nesting notation
[ 36.078515][ T744]
[ 36.080491][ T744] 3 locks held by ip/744:
[ 36.081471][ T744] #0: ffffffff98571df0 (rtnl_mutex){+.+.}-{3:3}, at: rtnetlink_rcv_msg+0x236/0x490
[ 36.083614][ T744] #1: ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.085942][ T744] #2: ffff8c476c8da280 (&bridge_netdev_addr_lock_key/4){+...}-{2:2}, at: dev_uc_sync+0x39/0x80
[ 36.088400][ T744]
[ 36.088400][ T744] stack backtrace:
[ 36.089772][ T744] CPU: 6 PID: 744 Comm: ip Not tainted 5.9.0-rc6+ #728
[ 36.091364][ T744] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
[ 36.093630][ T744] Call Trace:
[ 36.094416][ T744] dump_stack+0x77/0x9b
[ 36.095385][ T744] __lock_acquire+0xbc3/0x1f40
[ 36.096522][ T744] lock_acquire+0xb4/0x3b0
[ 36.097540][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.098657][ T744] ? rtmsg_ifinfo+0x1f/0x30
[ 36.099711][ T744] ? __dev_notify_flags+0xa5/0xf0
[ 36.100874][ T744] ? rtnl_is_locked+0x11/0x20
[ 36.101967][ T744] ? __dev_set_promiscuity+0x7b/0x1a0
[ 36.103230][ T744] _raw_spin_lock_bh+0x38/0x70
[ 36.104348][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.105461][ T744] dev_set_rx_mode+0x19/0x30
[ 36.106532][ T744] dev_set_promiscuity+0x36/0x50
[ 36.107692][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.108929][ T744] dev_set_promiscuity+0x1e/0x50
[ 36.110093][ T744] br_port_set_promisc+0x1f/0x40 [bridge]
[ 36.111415][ T744] br_manage_promisc+0x8b/0xe0 [bridge]
[ 36.112728][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.113967][ T744] ? __hw_addr_sync_one+0x23/0x50
[ 36.115135][ T744] __dev_set_rx_mode+0x68/0x90
[ 36.116249][ T744] dev_uc_sync+0x70/0x80
[ 36.117244][ T744] dev_uc_add+0x50/0x60
[ 36.118223][ T744] macvlan_open+0x18e/0x1f0 [macvlan]
[ 36.119470][ T744] __dev_open+0xd6/0x170
[ 36.120470][ T744] __dev_change_flags+0x181/0x1d0
[ 36.121644][ T744] dev_change_flags+0x23/0x60
[ 36.122741][ T744] do_setlink+0x30a/0x11e0
[ 36.123778][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.124929][ T744] ? __nla_validate_parse.part.6+0x45/0x8e0
[ 36.126309][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.127457][ T744] __rtnl_newlink+0x546/0x8e0
[ 36.128560][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.129623][ T744] ? deactivate_slab.isra.85+0x6a1/0x850
[ 36.130946][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.132102][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.133176][ T744] ? is_bpf_text_address+0x5/0xe0
[ 36.134364][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.135445][ T744] ? rcu_read_lock_sched_held+0x32/0x60
[ 36.136771][ T744] ? kmem_cache_alloc_trace+0x2d8/0x380
[ 36.138070][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.139164][ T744] rtnl_newlink+0x47/0x70
[ ... ]
Fixes: 845e0ebb4408 ("net: change addr_list_lock back to static key")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-26 02:13:29 +08:00
|
|
|
unsigned char nest_level = 0;
|
2008-07-15 15:13:44 +08:00
|
|
|
|
net: core: add nested_level variable in net_device
This patch is to add a new variable 'nested_level' into the net_device
structure.
This variable will be used as a parameter of spin_lock_nested() of
dev->addr_list_lock.
netif_addr_lock() can be called recursively so spin_lock_nested() is
used instead of spin_lock() and dev->lower_level is used as a parameter
of spin_lock_nested().
But, dev->lower_level value can be updated while it is being used.
So, lockdep would warn a possible deadlock scenario.
When a stacked interface is deleted, netif_{uc | mc}_sync() is
called recursively.
So, spin_lock_nested() is called recursively too.
At this moment, the dev->lower_level variable is used as a parameter of it.
dev->lower_level value is updated when interfaces are being unlinked/linked
immediately.
Thus, After unlinking, dev->lower_level shouldn't be a parameter of
spin_lock_nested().
A (macvlan)
|
B (vlan)
|
C (bridge)
|
D (macvlan)
|
E (vlan)
|
F (bridge)
A->lower_level : 6
B->lower_level : 5
C->lower_level : 4
D->lower_level : 3
E->lower_level : 2
F->lower_level : 1
When an interface 'A' is removed, it releases resources.
At this moment, netif_addr_lock() would be called.
Then, netdev_upper_dev_unlink() is called recursively.
Then dev->lower_level is updated.
There is no problem.
But, when the bridge module is removed, 'C' and 'F' interfaces
are removed at once.
If 'F' is removed first, a lower_level value is like below.
A->lower_level : 5
B->lower_level : 4
C->lower_level : 3
D->lower_level : 2
E->lower_level : 1
F->lower_level : 1
Then, 'C' is removed. at this moment, netif_addr_lock() is called
recursively.
The ordering is like this.
C(3)->D(2)->E(1)->F(1)
At this moment, the lower_level value of 'E' and 'F' are the same.
So, lockdep warns a possible deadlock scenario.
In order to avoid this problem, a new variable 'nested_level' is added.
This value is the same as dev->lower_level - 1.
But this value is updated in rtnl_unlock().
So, this variable can be used as a parameter of spin_lock_nested() safely
in the rtnl context.
Test commands:
ip link add br0 type bridge vlan_filtering 1
ip link add vlan1 link br0 type vlan id 10
ip link add macvlan2 link vlan1 type macvlan
ip link add br3 type bridge vlan_filtering 1
ip link set macvlan2 master br3
ip link add vlan4 link br3 type vlan id 10
ip link add macvlan5 link vlan4 type macvlan
ip link add br6 type bridge vlan_filtering 1
ip link set macvlan5 master br6
ip link add vlan7 link br6 type vlan id 10
ip link add macvlan8 link vlan7 type macvlan
ip link set br0 up
ip link set vlan1 up
ip link set macvlan2 up
ip link set br3 up
ip link set vlan4 up
ip link set macvlan5 up
ip link set br6 up
ip link set vlan7 up
ip link set macvlan8 up
modprobe -rv bridge
Splat looks like:
[ 36.057436][ T744] WARNING: possible recursive locking detected
[ 36.058848][ T744] 5.9.0-rc6+ #728 Not tainted
[ 36.059959][ T744] --------------------------------------------
[ 36.061391][ T744] ip/744 is trying to acquire lock:
[ 36.062590][ T744] ffff8c4767509280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_set_rx_mode+0x19/0x30
[ 36.064922][ T744]
[ 36.064922][ T744] but task is already holding lock:
[ 36.066626][ T744] ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.068851][ T744]
[ 36.068851][ T744] other info that might help us debug this:
[ 36.070731][ T744] Possible unsafe locking scenario:
[ 36.070731][ T744]
[ 36.072497][ T744] CPU0
[ 36.073238][ T744] ----
[ 36.074007][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.075290][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.076590][ T744]
[ 36.076590][ T744] *** DEADLOCK ***
[ 36.076590][ T744]
[ 36.078515][ T744] May be due to missing lock nesting notation
[ 36.078515][ T744]
[ 36.080491][ T744] 3 locks held by ip/744:
[ 36.081471][ T744] #0: ffffffff98571df0 (rtnl_mutex){+.+.}-{3:3}, at: rtnetlink_rcv_msg+0x236/0x490
[ 36.083614][ T744] #1: ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.085942][ T744] #2: ffff8c476c8da280 (&bridge_netdev_addr_lock_key/4){+...}-{2:2}, at: dev_uc_sync+0x39/0x80
[ 36.088400][ T744]
[ 36.088400][ T744] stack backtrace:
[ 36.089772][ T744] CPU: 6 PID: 744 Comm: ip Not tainted 5.9.0-rc6+ #728
[ 36.091364][ T744] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
[ 36.093630][ T744] Call Trace:
[ 36.094416][ T744] dump_stack+0x77/0x9b
[ 36.095385][ T744] __lock_acquire+0xbc3/0x1f40
[ 36.096522][ T744] lock_acquire+0xb4/0x3b0
[ 36.097540][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.098657][ T744] ? rtmsg_ifinfo+0x1f/0x30
[ 36.099711][ T744] ? __dev_notify_flags+0xa5/0xf0
[ 36.100874][ T744] ? rtnl_is_locked+0x11/0x20
[ 36.101967][ T744] ? __dev_set_promiscuity+0x7b/0x1a0
[ 36.103230][ T744] _raw_spin_lock_bh+0x38/0x70
[ 36.104348][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.105461][ T744] dev_set_rx_mode+0x19/0x30
[ 36.106532][ T744] dev_set_promiscuity+0x36/0x50
[ 36.107692][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.108929][ T744] dev_set_promiscuity+0x1e/0x50
[ 36.110093][ T744] br_port_set_promisc+0x1f/0x40 [bridge]
[ 36.111415][ T744] br_manage_promisc+0x8b/0xe0 [bridge]
[ 36.112728][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.113967][ T744] ? __hw_addr_sync_one+0x23/0x50
[ 36.115135][ T744] __dev_set_rx_mode+0x68/0x90
[ 36.116249][ T744] dev_uc_sync+0x70/0x80
[ 36.117244][ T744] dev_uc_add+0x50/0x60
[ 36.118223][ T744] macvlan_open+0x18e/0x1f0 [macvlan]
[ 36.119470][ T744] __dev_open+0xd6/0x170
[ 36.120470][ T744] __dev_change_flags+0x181/0x1d0
[ 36.121644][ T744] dev_change_flags+0x23/0x60
[ 36.122741][ T744] do_setlink+0x30a/0x11e0
[ 36.123778][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.124929][ T744] ? __nla_validate_parse.part.6+0x45/0x8e0
[ 36.126309][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.127457][ T744] __rtnl_newlink+0x546/0x8e0
[ 36.128560][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.129623][ T744] ? deactivate_slab.isra.85+0x6a1/0x850
[ 36.130946][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.132102][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.133176][ T744] ? is_bpf_text_address+0x5/0xe0
[ 36.134364][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.135445][ T744] ? rcu_read_lock_sched_held+0x32/0x60
[ 36.136771][ T744] ? kmem_cache_alloc_trace+0x2d8/0x380
[ 36.138070][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.139164][ T744] rtnl_newlink+0x47/0x70
[ ... ]
Fixes: 845e0ebb4408 ("net: change addr_list_lock back to static key")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-26 02:13:29 +08:00
|
|
|
#ifdef CONFIG_LOCKDEP
|
|
|
|
nest_level = dev->nested_level;
|
|
|
|
#endif
|
|
|
|
spin_lock_nested(&dev->addr_list_lock, nest_level);
|
2020-06-09 05:53:01 +08:00
|
|
|
}
|
|
|
|
|
2008-07-15 15:13:44 +08:00
|
|
|
static inline void netif_addr_lock_bh(struct net_device *dev)
|
|
|
|
{
|
net: core: add nested_level variable in net_device
This patch is to add a new variable 'nested_level' into the net_device
structure.
This variable will be used as a parameter of spin_lock_nested() of
dev->addr_list_lock.
netif_addr_lock() can be called recursively so spin_lock_nested() is
used instead of spin_lock() and dev->lower_level is used as a parameter
of spin_lock_nested().
But, dev->lower_level value can be updated while it is being used.
So, lockdep would warn a possible deadlock scenario.
When a stacked interface is deleted, netif_{uc | mc}_sync() is
called recursively.
So, spin_lock_nested() is called recursively too.
At this moment, the dev->lower_level variable is used as a parameter of it.
dev->lower_level value is updated when interfaces are being unlinked/linked
immediately.
Thus, After unlinking, dev->lower_level shouldn't be a parameter of
spin_lock_nested().
A (macvlan)
|
B (vlan)
|
C (bridge)
|
D (macvlan)
|
E (vlan)
|
F (bridge)
A->lower_level : 6
B->lower_level : 5
C->lower_level : 4
D->lower_level : 3
E->lower_level : 2
F->lower_level : 1
When an interface 'A' is removed, it releases resources.
At this moment, netif_addr_lock() would be called.
Then, netdev_upper_dev_unlink() is called recursively.
Then dev->lower_level is updated.
There is no problem.
But, when the bridge module is removed, 'C' and 'F' interfaces
are removed at once.
If 'F' is removed first, a lower_level value is like below.
A->lower_level : 5
B->lower_level : 4
C->lower_level : 3
D->lower_level : 2
E->lower_level : 1
F->lower_level : 1
Then, 'C' is removed. at this moment, netif_addr_lock() is called
recursively.
The ordering is like this.
C(3)->D(2)->E(1)->F(1)
At this moment, the lower_level value of 'E' and 'F' are the same.
So, lockdep warns a possible deadlock scenario.
In order to avoid this problem, a new variable 'nested_level' is added.
This value is the same as dev->lower_level - 1.
But this value is updated in rtnl_unlock().
So, this variable can be used as a parameter of spin_lock_nested() safely
in the rtnl context.
Test commands:
ip link add br0 type bridge vlan_filtering 1
ip link add vlan1 link br0 type vlan id 10
ip link add macvlan2 link vlan1 type macvlan
ip link add br3 type bridge vlan_filtering 1
ip link set macvlan2 master br3
ip link add vlan4 link br3 type vlan id 10
ip link add macvlan5 link vlan4 type macvlan
ip link add br6 type bridge vlan_filtering 1
ip link set macvlan5 master br6
ip link add vlan7 link br6 type vlan id 10
ip link add macvlan8 link vlan7 type macvlan
ip link set br0 up
ip link set vlan1 up
ip link set macvlan2 up
ip link set br3 up
ip link set vlan4 up
ip link set macvlan5 up
ip link set br6 up
ip link set vlan7 up
ip link set macvlan8 up
modprobe -rv bridge
Splat looks like:
[ 36.057436][ T744] WARNING: possible recursive locking detected
[ 36.058848][ T744] 5.9.0-rc6+ #728 Not tainted
[ 36.059959][ T744] --------------------------------------------
[ 36.061391][ T744] ip/744 is trying to acquire lock:
[ 36.062590][ T744] ffff8c4767509280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_set_rx_mode+0x19/0x30
[ 36.064922][ T744]
[ 36.064922][ T744] but task is already holding lock:
[ 36.066626][ T744] ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.068851][ T744]
[ 36.068851][ T744] other info that might help us debug this:
[ 36.070731][ T744] Possible unsafe locking scenario:
[ 36.070731][ T744]
[ 36.072497][ T744] CPU0
[ 36.073238][ T744] ----
[ 36.074007][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.075290][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.076590][ T744]
[ 36.076590][ T744] *** DEADLOCK ***
[ 36.076590][ T744]
[ 36.078515][ T744] May be due to missing lock nesting notation
[ 36.078515][ T744]
[ 36.080491][ T744] 3 locks held by ip/744:
[ 36.081471][ T744] #0: ffffffff98571df0 (rtnl_mutex){+.+.}-{3:3}, at: rtnetlink_rcv_msg+0x236/0x490
[ 36.083614][ T744] #1: ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.085942][ T744] #2: ffff8c476c8da280 (&bridge_netdev_addr_lock_key/4){+...}-{2:2}, at: dev_uc_sync+0x39/0x80
[ 36.088400][ T744]
[ 36.088400][ T744] stack backtrace:
[ 36.089772][ T744] CPU: 6 PID: 744 Comm: ip Not tainted 5.9.0-rc6+ #728
[ 36.091364][ T744] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
[ 36.093630][ T744] Call Trace:
[ 36.094416][ T744] dump_stack+0x77/0x9b
[ 36.095385][ T744] __lock_acquire+0xbc3/0x1f40
[ 36.096522][ T744] lock_acquire+0xb4/0x3b0
[ 36.097540][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.098657][ T744] ? rtmsg_ifinfo+0x1f/0x30
[ 36.099711][ T744] ? __dev_notify_flags+0xa5/0xf0
[ 36.100874][ T744] ? rtnl_is_locked+0x11/0x20
[ 36.101967][ T744] ? __dev_set_promiscuity+0x7b/0x1a0
[ 36.103230][ T744] _raw_spin_lock_bh+0x38/0x70
[ 36.104348][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.105461][ T744] dev_set_rx_mode+0x19/0x30
[ 36.106532][ T744] dev_set_promiscuity+0x36/0x50
[ 36.107692][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.108929][ T744] dev_set_promiscuity+0x1e/0x50
[ 36.110093][ T744] br_port_set_promisc+0x1f/0x40 [bridge]
[ 36.111415][ T744] br_manage_promisc+0x8b/0xe0 [bridge]
[ 36.112728][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.113967][ T744] ? __hw_addr_sync_one+0x23/0x50
[ 36.115135][ T744] __dev_set_rx_mode+0x68/0x90
[ 36.116249][ T744] dev_uc_sync+0x70/0x80
[ 36.117244][ T744] dev_uc_add+0x50/0x60
[ 36.118223][ T744] macvlan_open+0x18e/0x1f0 [macvlan]
[ 36.119470][ T744] __dev_open+0xd6/0x170
[ 36.120470][ T744] __dev_change_flags+0x181/0x1d0
[ 36.121644][ T744] dev_change_flags+0x23/0x60
[ 36.122741][ T744] do_setlink+0x30a/0x11e0
[ 36.123778][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.124929][ T744] ? __nla_validate_parse.part.6+0x45/0x8e0
[ 36.126309][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.127457][ T744] __rtnl_newlink+0x546/0x8e0
[ 36.128560][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.129623][ T744] ? deactivate_slab.isra.85+0x6a1/0x850
[ 36.130946][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.132102][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.133176][ T744] ? is_bpf_text_address+0x5/0xe0
[ 36.134364][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.135445][ T744] ? rcu_read_lock_sched_held+0x32/0x60
[ 36.136771][ T744] ? kmem_cache_alloc_trace+0x2d8/0x380
[ 36.138070][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.139164][ T744] rtnl_newlink+0x47/0x70
[ ... ]
Fixes: 845e0ebb4408 ("net: change addr_list_lock back to static key")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-26 02:13:29 +08:00
|
|
|
unsigned char nest_level = 0;
|
|
|
|
|
|
|
|
#ifdef CONFIG_LOCKDEP
|
|
|
|
nest_level = dev->nested_level;
|
|
|
|
#endif
|
|
|
|
local_bh_disable();
|
|
|
|
spin_lock_nested(&dev->addr_list_lock, nest_level);
|
2008-07-15 15:13:44 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline void netif_addr_unlock(struct net_device *dev)
|
|
|
|
{
|
|
|
|
spin_unlock(&dev->addr_list_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void netif_addr_unlock_bh(struct net_device *dev)
|
|
|
|
{
|
|
|
|
spin_unlock_bh(&dev->addr_list_lock);
|
|
|
|
}
|
|
|
|
|
2009-05-05 10:48:28 +08:00
|
|
|
/*
|
2009-06-17 09:12:19 +08:00
|
|
|
* dev_addrs walker. Should be used only for read access. Call with
|
2009-05-05 10:48:28 +08:00
|
|
|
* rcu_read_lock held.
|
|
|
|
*/
|
|
|
|
#define for_each_dev_addr(dev, ha) \
|
2009-06-17 09:12:19 +08:00
|
|
|
list_for_each_entry_rcu(ha, &dev->dev_addrs.list, list)
|
2009-05-05 10:48:28 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/* These functions live elsewhere (drivers/net/net_init.c, but related) */
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
void ether_setup(struct net_device *dev);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
/* Support for loadable net-drivers */
|
2013-09-27 05:48:15 +08:00
|
|
|
struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
|
net: set name_assign_type in alloc_netdev()
Extend alloc_netdev{,_mq{,s}}() to take name_assign_type as argument, and convert
all users to pass NET_NAME_UNKNOWN.
Coccinelle patch:
@@
expression sizeof_priv, name, setup, txqs, rxqs, count;
@@
(
-alloc_netdev_mqs(sizeof_priv, name, setup, txqs, rxqs)
+alloc_netdev_mqs(sizeof_priv, name, NET_NAME_UNKNOWN, setup, txqs, rxqs)
|
-alloc_netdev_mq(sizeof_priv, name, setup, count)
+alloc_netdev_mq(sizeof_priv, name, NET_NAME_UNKNOWN, setup, count)
|
-alloc_netdev(sizeof_priv, name, setup)
+alloc_netdev(sizeof_priv, name, NET_NAME_UNKNOWN, setup)
)
v9: move comments here from the wrong commit
Signed-off-by: Tom Gundersen <teg@jklm.no>
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-07-14 22:37:24 +08:00
|
|
|
unsigned char name_assign_type,
|
2013-09-27 05:48:15 +08:00
|
|
|
void (*setup)(struct net_device *),
|
|
|
|
unsigned int txqs, unsigned int rxqs);
|
net: set name_assign_type in alloc_netdev()
Extend alloc_netdev{,_mq{,s}}() to take name_assign_type as argument, and convert
all users to pass NET_NAME_UNKNOWN.
Coccinelle patch:
@@
expression sizeof_priv, name, setup, txqs, rxqs, count;
@@
(
-alloc_netdev_mqs(sizeof_priv, name, setup, txqs, rxqs)
+alloc_netdev_mqs(sizeof_priv, name, NET_NAME_UNKNOWN, setup, txqs, rxqs)
|
-alloc_netdev_mq(sizeof_priv, name, setup, count)
+alloc_netdev_mq(sizeof_priv, name, NET_NAME_UNKNOWN, setup, count)
|
-alloc_netdev(sizeof_priv, name, setup)
+alloc_netdev(sizeof_priv, name, NET_NAME_UNKNOWN, setup)
)
v9: move comments here from the wrong commit
Signed-off-by: Tom Gundersen <teg@jklm.no>
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-07-14 22:37:24 +08:00
|
|
|
#define alloc_netdev(sizeof_priv, name, name_assign_type, setup) \
|
|
|
|
alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, 1, 1)
|
2011-01-10 03:36:31 +08:00
|
|
|
|
net: set name_assign_type in alloc_netdev()
Extend alloc_netdev{,_mq{,s}}() to take name_assign_type as argument, and convert
all users to pass NET_NAME_UNKNOWN.
Coccinelle patch:
@@
expression sizeof_priv, name, setup, txqs, rxqs, count;
@@
(
-alloc_netdev_mqs(sizeof_priv, name, setup, txqs, rxqs)
+alloc_netdev_mqs(sizeof_priv, name, NET_NAME_UNKNOWN, setup, txqs, rxqs)
|
-alloc_netdev_mq(sizeof_priv, name, setup, count)
+alloc_netdev_mq(sizeof_priv, name, NET_NAME_UNKNOWN, setup, count)
|
-alloc_netdev(sizeof_priv, name, setup)
+alloc_netdev(sizeof_priv, name, NET_NAME_UNKNOWN, setup)
)
v9: move comments here from the wrong commit
Signed-off-by: Tom Gundersen <teg@jklm.no>
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-07-14 22:37:24 +08:00
|
|
|
#define alloc_netdev_mq(sizeof_priv, name, name_assign_type, setup, count) \
|
|
|
|
alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, count, \
|
|
|
|
count)
|
2011-01-10 03:36:31 +08:00
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
int register_netdev(struct net_device *dev);
|
|
|
|
void unregister_netdev(struct net_device *dev);
|
2009-05-05 10:48:28 +08:00
|
|
|
|
2020-05-23 21:27:10 +08:00
|
|
|
int devm_register_netdev(struct device *dev, struct net_device *ndev);
|
|
|
|
|
2010-04-02 05:22:57 +08:00
|
|
|
/* General hardware address lists handling functions */
|
2013-09-27 05:48:15 +08:00
|
|
|
int __hw_addr_sync(struct netdev_hw_addr_list *to_list,
|
|
|
|
struct netdev_hw_addr_list *from_list, int addr_len);
|
|
|
|
void __hw_addr_unsync(struct netdev_hw_addr_list *to_list,
|
|
|
|
struct netdev_hw_addr_list *from_list, int addr_len);
|
2014-05-29 09:44:46 +08:00
|
|
|
int __hw_addr_sync_dev(struct netdev_hw_addr_list *list,
|
|
|
|
struct net_device *dev,
|
|
|
|
int (*sync)(struct net_device *, const unsigned char *),
|
|
|
|
int (*unsync)(struct net_device *,
|
|
|
|
const unsigned char *));
|
2018-11-09 04:27:54 +08:00
|
|
|
int __hw_addr_ref_sync_dev(struct netdev_hw_addr_list *list,
|
|
|
|
struct net_device *dev,
|
|
|
|
int (*sync)(struct net_device *,
|
|
|
|
const unsigned char *, int),
|
|
|
|
int (*unsync)(struct net_device *,
|
|
|
|
const unsigned char *, int));
|
|
|
|
void __hw_addr_ref_unsync_dev(struct netdev_hw_addr_list *list,
|
|
|
|
struct net_device *dev,
|
|
|
|
int (*unsync)(struct net_device *,
|
|
|
|
const unsigned char *, int));
|
2014-05-29 09:44:46 +08:00
|
|
|
void __hw_addr_unsync_dev(struct netdev_hw_addr_list *list,
|
|
|
|
struct net_device *dev,
|
|
|
|
int (*unsync)(struct net_device *,
|
|
|
|
const unsigned char *));
|
2013-09-27 05:48:15 +08:00
|
|
|
void __hw_addr_init(struct netdev_hw_addr_list *list);
|
2010-04-02 05:22:57 +08:00
|
|
|
|
2009-05-05 10:48:28 +08:00
|
|
|
/* Functions used for device addresses handling */
|
2021-11-19 22:21:51 +08:00
|
|
|
void dev_addr_mod(struct net_device *dev, unsigned int offset,
|
|
|
|
const void *addr, size_t len);
|
|
|
|
|
2021-09-03 02:10:37 +08:00
|
|
|
static inline void
|
2021-10-13 00:06:32 +08:00
|
|
|
__dev_addr_set(struct net_device *dev, const void *addr, size_t len)
|
2021-09-03 02:10:37 +08:00
|
|
|
{
|
2021-11-19 22:21:51 +08:00
|
|
|
dev_addr_mod(dev, 0, addr, len);
|
2021-09-03 02:10:37 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline void dev_addr_set(struct net_device *dev, const u8 *addr)
|
|
|
|
{
|
|
|
|
__dev_addr_set(dev, addr, dev->addr_len);
|
|
|
|
}
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
int dev_addr_add(struct net_device *dev, const unsigned char *addr,
|
|
|
|
unsigned char addr_type);
|
|
|
|
int dev_addr_del(struct net_device *dev, const unsigned char *addr,
|
|
|
|
unsigned char addr_type);
|
2010-04-02 05:22:09 +08:00
|
|
|
|
|
|
|
/* Functions used for unicast addresses handling */
|
2013-09-27 05:48:15 +08:00
|
|
|
int dev_uc_add(struct net_device *dev, const unsigned char *addr);
|
|
|
|
int dev_uc_add_excl(struct net_device *dev, const unsigned char *addr);
|
|
|
|
int dev_uc_del(struct net_device *dev, const unsigned char *addr);
|
|
|
|
int dev_uc_sync(struct net_device *to, struct net_device *from);
|
|
|
|
int dev_uc_sync_multiple(struct net_device *to, struct net_device *from);
|
|
|
|
void dev_uc_unsync(struct net_device *to, struct net_device *from);
|
|
|
|
void dev_uc_flush(struct net_device *dev);
|
|
|
|
void dev_uc_init(struct net_device *dev);
|
2009-05-05 10:48:28 +08:00
|
|
|
|
2014-05-29 09:44:46 +08:00
|
|
|
/**
|
|
|
|
* __dev_uc_sync - Synchonize device's unicast list
|
|
|
|
* @dev: device to sync
|
|
|
|
* @sync: function to call if address should be added
|
|
|
|
* @unsync: function to call if address should be removed
|
|
|
|
*
|
|
|
|
* Add newly added addresses to the interface, and release
|
|
|
|
* addresses that have been deleted.
|
2016-03-24 02:47:23 +08:00
|
|
|
*/
|
2014-05-29 09:44:46 +08:00
|
|
|
static inline int __dev_uc_sync(struct net_device *dev,
|
|
|
|
int (*sync)(struct net_device *,
|
|
|
|
const unsigned char *),
|
|
|
|
int (*unsync)(struct net_device *,
|
|
|
|
const unsigned char *))
|
|
|
|
{
|
|
|
|
return __hw_addr_sync_dev(&dev->uc, dev, sync, unsync);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2014-09-04 22:44:36 +08:00
|
|
|
* __dev_uc_unsync - Remove synchronized addresses from device
|
2014-05-29 09:44:46 +08:00
|
|
|
* @dev: device to sync
|
|
|
|
* @unsync: function to call if address should be removed
|
|
|
|
*
|
|
|
|
* Remove all addresses that were added to the device by dev_uc_sync().
|
2016-03-24 02:47:23 +08:00
|
|
|
*/
|
2014-05-29 09:44:46 +08:00
|
|
|
static inline void __dev_uc_unsync(struct net_device *dev,
|
|
|
|
int (*unsync)(struct net_device *,
|
|
|
|
const unsigned char *))
|
|
|
|
{
|
|
|
|
__hw_addr_unsync_dev(&dev->uc, dev, unsync);
|
|
|
|
}
|
|
|
|
|
2010-04-02 05:22:57 +08:00
|
|
|
/* Functions used for multicast addresses handling */
|
2013-09-27 05:48:15 +08:00
|
|
|
int dev_mc_add(struct net_device *dev, const unsigned char *addr);
|
|
|
|
int dev_mc_add_global(struct net_device *dev, const unsigned char *addr);
|
|
|
|
int dev_mc_add_excl(struct net_device *dev, const unsigned char *addr);
|
|
|
|
int dev_mc_del(struct net_device *dev, const unsigned char *addr);
|
|
|
|
int dev_mc_del_global(struct net_device *dev, const unsigned char *addr);
|
|
|
|
int dev_mc_sync(struct net_device *to, struct net_device *from);
|
|
|
|
int dev_mc_sync_multiple(struct net_device *to, struct net_device *from);
|
|
|
|
void dev_mc_unsync(struct net_device *to, struct net_device *from);
|
|
|
|
void dev_mc_flush(struct net_device *dev);
|
|
|
|
void dev_mc_init(struct net_device *dev);
|
2009-05-05 10:48:28 +08:00
|
|
|
|
2014-05-29 09:44:46 +08:00
|
|
|
/**
|
|
|
|
* __dev_mc_sync - Synchonize device's multicast list
|
|
|
|
* @dev: device to sync
|
|
|
|
* @sync: function to call if address should be added
|
|
|
|
* @unsync: function to call if address should be removed
|
|
|
|
*
|
|
|
|
* Add newly added addresses to the interface, and release
|
|
|
|
* addresses that have been deleted.
|
2016-03-24 02:47:23 +08:00
|
|
|
*/
|
2014-05-29 09:44:46 +08:00
|
|
|
static inline int __dev_mc_sync(struct net_device *dev,
|
|
|
|
int (*sync)(struct net_device *,
|
|
|
|
const unsigned char *),
|
|
|
|
int (*unsync)(struct net_device *,
|
|
|
|
const unsigned char *))
|
|
|
|
{
|
|
|
|
return __hw_addr_sync_dev(&dev->mc, dev, sync, unsync);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2014-09-04 22:44:36 +08:00
|
|
|
* __dev_mc_unsync - Remove synchronized addresses from device
|
2014-05-29 09:44:46 +08:00
|
|
|
* @dev: device to sync
|
|
|
|
* @unsync: function to call if address should be removed
|
|
|
|
*
|
|
|
|
* Remove all addresses that were added to the device by dev_mc_sync().
|
2016-03-24 02:47:23 +08:00
|
|
|
*/
|
2014-05-29 09:44:46 +08:00
|
|
|
static inline void __dev_mc_unsync(struct net_device *dev,
|
|
|
|
int (*unsync)(struct net_device *,
|
|
|
|
const unsigned char *))
|
|
|
|
{
|
|
|
|
__hw_addr_unsync_dev(&dev->mc, dev, unsync);
|
|
|
|
}
|
|
|
|
|
2007-06-27 16:28:10 +08:00
|
|
|
/* Functions used for secondary unicast and multicast support */
|
2013-09-27 05:48:15 +08:00
|
|
|
void dev_set_rx_mode(struct net_device *dev);
|
|
|
|
int dev_set_promiscuity(struct net_device *dev, int inc);
|
|
|
|
int dev_set_allmulti(struct net_device *dev, int inc);
|
|
|
|
void netdev_state_change(struct net_device *dev);
|
2020-12-15 05:19:28 +08:00
|
|
|
void __netdev_notify_peers(struct net_device *dev);
|
2013-09-27 05:48:15 +08:00
|
|
|
void netdev_notify_peers(struct net_device *dev);
|
|
|
|
void netdev_features_change(struct net_device *dev);
|
2005-04-17 06:20:36 +08:00
|
|
|
/* Load a device via the kmod */
|
2013-09-27 05:48:15 +08:00
|
|
|
void dev_load(struct net *net, const char *name);
|
|
|
|
struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
|
|
|
|
struct rtnl_link_stats64 *storage);
|
|
|
|
void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
|
|
|
|
const struct net_device_stats *netdev_stats);
|
2020-10-12 16:01:27 +08:00
|
|
|
void dev_fetch_sw_netstats(struct rtnl_link_stats64 *s,
|
|
|
|
const struct pcpu_sw_netstats __percpu *netstats);
|
2020-11-08 04:49:07 +08:00
|
|
|
void dev_get_tstats64(struct net_device *dev, struct rtnl_link_stats64 *s);
|
2008-11-20 13:40:23 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
extern int netdev_max_backlog;
|
2016-12-30 04:37:21 +08:00
|
|
|
extern int dev_rx_weight;
|
|
|
|
extern int dev_tx_weight;
|
net: use listified RX for handling GRO_NORMAL skbs
When GRO decides not to coalesce a packet, in napi_frags_finish(), instead
of passing it to the stack immediately, place it on a list in the napi
struct. Then, at flush time (napi_complete_done(), napi_poll(), or
napi_busy_loop()), call netif_receive_skb_list_internal() on the list.
We'd like to do that in napi_gro_flush(), but it's not called if
!napi->gro_bitmask, so we have to do it in the callers instead. (There are
a handful of drivers that call napi_gro_flush() themselves, but it's not
clear why, or whether this will affect them.)
Because a full 64 packets is an inefficiently large batch, also consume the
list whenever it exceeds gro_normal_batch, a new net/core sysctl that
defaults to 8.
Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-08-06 21:53:55 +08:00
|
|
|
extern int gro_normal_batch;
|
2013-01-04 06:48:49 +08:00
|
|
|
|
net: core: add nested_level variable in net_device
This patch is to add a new variable 'nested_level' into the net_device
structure.
This variable will be used as a parameter of spin_lock_nested() of
dev->addr_list_lock.
netif_addr_lock() can be called recursively so spin_lock_nested() is
used instead of spin_lock() and dev->lower_level is used as a parameter
of spin_lock_nested().
But, dev->lower_level value can be updated while it is being used.
So, lockdep would warn a possible deadlock scenario.
When a stacked interface is deleted, netif_{uc | mc}_sync() is
called recursively.
So, spin_lock_nested() is called recursively too.
At this moment, the dev->lower_level variable is used as a parameter of it.
dev->lower_level value is updated when interfaces are being unlinked/linked
immediately.
Thus, After unlinking, dev->lower_level shouldn't be a parameter of
spin_lock_nested().
A (macvlan)
|
B (vlan)
|
C (bridge)
|
D (macvlan)
|
E (vlan)
|
F (bridge)
A->lower_level : 6
B->lower_level : 5
C->lower_level : 4
D->lower_level : 3
E->lower_level : 2
F->lower_level : 1
When an interface 'A' is removed, it releases resources.
At this moment, netif_addr_lock() would be called.
Then, netdev_upper_dev_unlink() is called recursively.
Then dev->lower_level is updated.
There is no problem.
But, when the bridge module is removed, 'C' and 'F' interfaces
are removed at once.
If 'F' is removed first, a lower_level value is like below.
A->lower_level : 5
B->lower_level : 4
C->lower_level : 3
D->lower_level : 2
E->lower_level : 1
F->lower_level : 1
Then, 'C' is removed. at this moment, netif_addr_lock() is called
recursively.
The ordering is like this.
C(3)->D(2)->E(1)->F(1)
At this moment, the lower_level value of 'E' and 'F' are the same.
So, lockdep warns a possible deadlock scenario.
In order to avoid this problem, a new variable 'nested_level' is added.
This value is the same as dev->lower_level - 1.
But this value is updated in rtnl_unlock().
So, this variable can be used as a parameter of spin_lock_nested() safely
in the rtnl context.
Test commands:
ip link add br0 type bridge vlan_filtering 1
ip link add vlan1 link br0 type vlan id 10
ip link add macvlan2 link vlan1 type macvlan
ip link add br3 type bridge vlan_filtering 1
ip link set macvlan2 master br3
ip link add vlan4 link br3 type vlan id 10
ip link add macvlan5 link vlan4 type macvlan
ip link add br6 type bridge vlan_filtering 1
ip link set macvlan5 master br6
ip link add vlan7 link br6 type vlan id 10
ip link add macvlan8 link vlan7 type macvlan
ip link set br0 up
ip link set vlan1 up
ip link set macvlan2 up
ip link set br3 up
ip link set vlan4 up
ip link set macvlan5 up
ip link set br6 up
ip link set vlan7 up
ip link set macvlan8 up
modprobe -rv bridge
Splat looks like:
[ 36.057436][ T744] WARNING: possible recursive locking detected
[ 36.058848][ T744] 5.9.0-rc6+ #728 Not tainted
[ 36.059959][ T744] --------------------------------------------
[ 36.061391][ T744] ip/744 is trying to acquire lock:
[ 36.062590][ T744] ffff8c4767509280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_set_rx_mode+0x19/0x30
[ 36.064922][ T744]
[ 36.064922][ T744] but task is already holding lock:
[ 36.066626][ T744] ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.068851][ T744]
[ 36.068851][ T744] other info that might help us debug this:
[ 36.070731][ T744] Possible unsafe locking scenario:
[ 36.070731][ T744]
[ 36.072497][ T744] CPU0
[ 36.073238][ T744] ----
[ 36.074007][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.075290][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.076590][ T744]
[ 36.076590][ T744] *** DEADLOCK ***
[ 36.076590][ T744]
[ 36.078515][ T744] May be due to missing lock nesting notation
[ 36.078515][ T744]
[ 36.080491][ T744] 3 locks held by ip/744:
[ 36.081471][ T744] #0: ffffffff98571df0 (rtnl_mutex){+.+.}-{3:3}, at: rtnetlink_rcv_msg+0x236/0x490
[ 36.083614][ T744] #1: ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.085942][ T744] #2: ffff8c476c8da280 (&bridge_netdev_addr_lock_key/4){+...}-{2:2}, at: dev_uc_sync+0x39/0x80
[ 36.088400][ T744]
[ 36.088400][ T744] stack backtrace:
[ 36.089772][ T744] CPU: 6 PID: 744 Comm: ip Not tainted 5.9.0-rc6+ #728
[ 36.091364][ T744] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
[ 36.093630][ T744] Call Trace:
[ 36.094416][ T744] dump_stack+0x77/0x9b
[ 36.095385][ T744] __lock_acquire+0xbc3/0x1f40
[ 36.096522][ T744] lock_acquire+0xb4/0x3b0
[ 36.097540][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.098657][ T744] ? rtmsg_ifinfo+0x1f/0x30
[ 36.099711][ T744] ? __dev_notify_flags+0xa5/0xf0
[ 36.100874][ T744] ? rtnl_is_locked+0x11/0x20
[ 36.101967][ T744] ? __dev_set_promiscuity+0x7b/0x1a0
[ 36.103230][ T744] _raw_spin_lock_bh+0x38/0x70
[ 36.104348][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.105461][ T744] dev_set_rx_mode+0x19/0x30
[ 36.106532][ T744] dev_set_promiscuity+0x36/0x50
[ 36.107692][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.108929][ T744] dev_set_promiscuity+0x1e/0x50
[ 36.110093][ T744] br_port_set_promisc+0x1f/0x40 [bridge]
[ 36.111415][ T744] br_manage_promisc+0x8b/0xe0 [bridge]
[ 36.112728][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.113967][ T744] ? __hw_addr_sync_one+0x23/0x50
[ 36.115135][ T744] __dev_set_rx_mode+0x68/0x90
[ 36.116249][ T744] dev_uc_sync+0x70/0x80
[ 36.117244][ T744] dev_uc_add+0x50/0x60
[ 36.118223][ T744] macvlan_open+0x18e/0x1f0 [macvlan]
[ 36.119470][ T744] __dev_open+0xd6/0x170
[ 36.120470][ T744] __dev_change_flags+0x181/0x1d0
[ 36.121644][ T744] dev_change_flags+0x23/0x60
[ 36.122741][ T744] do_setlink+0x30a/0x11e0
[ 36.123778][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.124929][ T744] ? __nla_validate_parse.part.6+0x45/0x8e0
[ 36.126309][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.127457][ T744] __rtnl_newlink+0x546/0x8e0
[ 36.128560][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.129623][ T744] ? deactivate_slab.isra.85+0x6a1/0x850
[ 36.130946][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.132102][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.133176][ T744] ? is_bpf_text_address+0x5/0xe0
[ 36.134364][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.135445][ T744] ? rcu_read_lock_sched_held+0x32/0x60
[ 36.136771][ T744] ? kmem_cache_alloc_trace+0x2d8/0x380
[ 36.138070][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.139164][ T744] rtnl_newlink+0x47/0x70
[ ... ]
Fixes: 845e0ebb4408 ("net: change addr_list_lock back to static key")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-26 02:13:29 +08:00
|
|
|
enum {
|
|
|
|
NESTED_SYNC_IMM_BIT,
|
|
|
|
NESTED_SYNC_TODO_BIT,
|
|
|
|
};
|
|
|
|
|
|
|
|
#define __NESTED_SYNC_BIT(bit) ((u32)1 << (bit))
|
|
|
|
#define __NESTED_SYNC(name) __NESTED_SYNC_BIT(NESTED_SYNC_ ## name ## _BIT)
|
|
|
|
|
|
|
|
#define NESTED_SYNC_IMM __NESTED_SYNC(IMM)
|
|
|
|
#define NESTED_SYNC_TODO __NESTED_SYNC(TODO)
|
|
|
|
|
2020-09-26 02:13:12 +08:00
|
|
|
struct netdev_nested_priv {
|
net: core: add nested_level variable in net_device
This patch is to add a new variable 'nested_level' into the net_device
structure.
This variable will be used as a parameter of spin_lock_nested() of
dev->addr_list_lock.
netif_addr_lock() can be called recursively so spin_lock_nested() is
used instead of spin_lock() and dev->lower_level is used as a parameter
of spin_lock_nested().
But, dev->lower_level value can be updated while it is being used.
So, lockdep would warn a possible deadlock scenario.
When a stacked interface is deleted, netif_{uc | mc}_sync() is
called recursively.
So, spin_lock_nested() is called recursively too.
At this moment, the dev->lower_level variable is used as a parameter of it.
dev->lower_level value is updated when interfaces are being unlinked/linked
immediately.
Thus, After unlinking, dev->lower_level shouldn't be a parameter of
spin_lock_nested().
A (macvlan)
|
B (vlan)
|
C (bridge)
|
D (macvlan)
|
E (vlan)
|
F (bridge)
A->lower_level : 6
B->lower_level : 5
C->lower_level : 4
D->lower_level : 3
E->lower_level : 2
F->lower_level : 1
When an interface 'A' is removed, it releases resources.
At this moment, netif_addr_lock() would be called.
Then, netdev_upper_dev_unlink() is called recursively.
Then dev->lower_level is updated.
There is no problem.
But, when the bridge module is removed, 'C' and 'F' interfaces
are removed at once.
If 'F' is removed first, a lower_level value is like below.
A->lower_level : 5
B->lower_level : 4
C->lower_level : 3
D->lower_level : 2
E->lower_level : 1
F->lower_level : 1
Then, 'C' is removed. at this moment, netif_addr_lock() is called
recursively.
The ordering is like this.
C(3)->D(2)->E(1)->F(1)
At this moment, the lower_level value of 'E' and 'F' are the same.
So, lockdep warns a possible deadlock scenario.
In order to avoid this problem, a new variable 'nested_level' is added.
This value is the same as dev->lower_level - 1.
But this value is updated in rtnl_unlock().
So, this variable can be used as a parameter of spin_lock_nested() safely
in the rtnl context.
Test commands:
ip link add br0 type bridge vlan_filtering 1
ip link add vlan1 link br0 type vlan id 10
ip link add macvlan2 link vlan1 type macvlan
ip link add br3 type bridge vlan_filtering 1
ip link set macvlan2 master br3
ip link add vlan4 link br3 type vlan id 10
ip link add macvlan5 link vlan4 type macvlan
ip link add br6 type bridge vlan_filtering 1
ip link set macvlan5 master br6
ip link add vlan7 link br6 type vlan id 10
ip link add macvlan8 link vlan7 type macvlan
ip link set br0 up
ip link set vlan1 up
ip link set macvlan2 up
ip link set br3 up
ip link set vlan4 up
ip link set macvlan5 up
ip link set br6 up
ip link set vlan7 up
ip link set macvlan8 up
modprobe -rv bridge
Splat looks like:
[ 36.057436][ T744] WARNING: possible recursive locking detected
[ 36.058848][ T744] 5.9.0-rc6+ #728 Not tainted
[ 36.059959][ T744] --------------------------------------------
[ 36.061391][ T744] ip/744 is trying to acquire lock:
[ 36.062590][ T744] ffff8c4767509280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_set_rx_mode+0x19/0x30
[ 36.064922][ T744]
[ 36.064922][ T744] but task is already holding lock:
[ 36.066626][ T744] ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.068851][ T744]
[ 36.068851][ T744] other info that might help us debug this:
[ 36.070731][ T744] Possible unsafe locking scenario:
[ 36.070731][ T744]
[ 36.072497][ T744] CPU0
[ 36.073238][ T744] ----
[ 36.074007][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.075290][ T744] lock(&vlan_netdev_addr_lock_key);
[ 36.076590][ T744]
[ 36.076590][ T744] *** DEADLOCK ***
[ 36.076590][ T744]
[ 36.078515][ T744] May be due to missing lock nesting notation
[ 36.078515][ T744]
[ 36.080491][ T744] 3 locks held by ip/744:
[ 36.081471][ T744] #0: ffffffff98571df0 (rtnl_mutex){+.+.}-{3:3}, at: rtnetlink_rcv_msg+0x236/0x490
[ 36.083614][ T744] #1: ffff8c4767769280 (&vlan_netdev_addr_lock_key){+...}-{2:2}, at: dev_uc_add+0x1e/0x60
[ 36.085942][ T744] #2: ffff8c476c8da280 (&bridge_netdev_addr_lock_key/4){+...}-{2:2}, at: dev_uc_sync+0x39/0x80
[ 36.088400][ T744]
[ 36.088400][ T744] stack backtrace:
[ 36.089772][ T744] CPU: 6 PID: 744 Comm: ip Not tainted 5.9.0-rc6+ #728
[ 36.091364][ T744] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
[ 36.093630][ T744] Call Trace:
[ 36.094416][ T744] dump_stack+0x77/0x9b
[ 36.095385][ T744] __lock_acquire+0xbc3/0x1f40
[ 36.096522][ T744] lock_acquire+0xb4/0x3b0
[ 36.097540][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.098657][ T744] ? rtmsg_ifinfo+0x1f/0x30
[ 36.099711][ T744] ? __dev_notify_flags+0xa5/0xf0
[ 36.100874][ T744] ? rtnl_is_locked+0x11/0x20
[ 36.101967][ T744] ? __dev_set_promiscuity+0x7b/0x1a0
[ 36.103230][ T744] _raw_spin_lock_bh+0x38/0x70
[ 36.104348][ T744] ? dev_set_rx_mode+0x19/0x30
[ 36.105461][ T744] dev_set_rx_mode+0x19/0x30
[ 36.106532][ T744] dev_set_promiscuity+0x36/0x50
[ 36.107692][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.108929][ T744] dev_set_promiscuity+0x1e/0x50
[ 36.110093][ T744] br_port_set_promisc+0x1f/0x40 [bridge]
[ 36.111415][ T744] br_manage_promisc+0x8b/0xe0 [bridge]
[ 36.112728][ T744] __dev_set_promiscuity+0x123/0x1a0
[ 36.113967][ T744] ? __hw_addr_sync_one+0x23/0x50
[ 36.115135][ T744] __dev_set_rx_mode+0x68/0x90
[ 36.116249][ T744] dev_uc_sync+0x70/0x80
[ 36.117244][ T744] dev_uc_add+0x50/0x60
[ 36.118223][ T744] macvlan_open+0x18e/0x1f0 [macvlan]
[ 36.119470][ T744] __dev_open+0xd6/0x170
[ 36.120470][ T744] __dev_change_flags+0x181/0x1d0
[ 36.121644][ T744] dev_change_flags+0x23/0x60
[ 36.122741][ T744] do_setlink+0x30a/0x11e0
[ 36.123778][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.124929][ T744] ? __nla_validate_parse.part.6+0x45/0x8e0
[ 36.126309][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.127457][ T744] __rtnl_newlink+0x546/0x8e0
[ 36.128560][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.129623][ T744] ? deactivate_slab.isra.85+0x6a1/0x850
[ 36.130946][ T744] ? __lock_acquire+0x92c/0x1f40
[ 36.132102][ T744] ? lock_acquire+0xb4/0x3b0
[ 36.133176][ T744] ? is_bpf_text_address+0x5/0xe0
[ 36.134364][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.135445][ T744] ? rcu_read_lock_sched_held+0x32/0x60
[ 36.136771][ T744] ? kmem_cache_alloc_trace+0x2d8/0x380
[ 36.138070][ T744] ? rtnl_newlink+0x2e/0x70
[ 36.139164][ T744] rtnl_newlink+0x47/0x70
[ ... ]
Fixes: 845e0ebb4408 ("net: change addr_list_lock back to static key")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-09-26 02:13:29 +08:00
|
|
|
unsigned char flags;
|
2020-09-26 02:13:12 +08:00
|
|
|
void *data;
|
|
|
|
};
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
bool netdev_has_upper_dev(struct net_device *dev, struct net_device *upper_dev);
|
2014-05-17 05:20:38 +08:00
|
|
|
struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
|
|
|
|
struct list_head **iter);
|
2013-08-29 05:25:08 +08:00
|
|
|
|
2014-05-17 05:20:38 +08:00
|
|
|
/* iterate through upper list, must be called under RCU read lock */
|
|
|
|
#define netdev_for_each_upper_dev_rcu(dev, updev, iter) \
|
|
|
|
for (iter = &(dev)->adj_list.upper, \
|
|
|
|
updev = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
|
|
|
|
updev; \
|
|
|
|
updev = netdev_upper_get_next_dev_rcu(dev, &(iter)))
|
|
|
|
|
2016-10-18 10:15:44 +08:00
|
|
|
int netdev_walk_all_upper_dev_rcu(struct net_device *dev,
|
|
|
|
int (*fn)(struct net_device *upper_dev,
|
2020-09-26 02:13:12 +08:00
|
|
|
struct netdev_nested_priv *priv),
|
|
|
|
struct netdev_nested_priv *priv);
|
2016-10-18 10:15:44 +08:00
|
|
|
|
|
|
|
bool netdev_has_upper_dev_all_rcu(struct net_device *dev,
|
|
|
|
struct net_device *upper_dev);
|
|
|
|
|
2017-09-01 16:52:31 +08:00
|
|
|
bool netdev_has_any_upper_dev(struct net_device *dev);
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
void *netdev_lower_get_next_private(struct net_device *dev,
|
|
|
|
struct list_head **iter);
|
|
|
|
void *netdev_lower_get_next_private_rcu(struct net_device *dev,
|
|
|
|
struct list_head **iter);
|
2013-09-25 15:20:12 +08:00
|
|
|
|
|
|
|
#define netdev_for_each_lower_private(dev, priv, iter) \
|
|
|
|
for (iter = (dev)->adj_list.lower.next, \
|
|
|
|
priv = netdev_lower_get_next_private(dev, &(iter)); \
|
|
|
|
priv; \
|
|
|
|
priv = netdev_lower_get_next_private(dev, &(iter)))
|
|
|
|
|
|
|
|
#define netdev_for_each_lower_private_rcu(dev, priv, iter) \
|
|
|
|
for (iter = &(dev)->adj_list.lower, \
|
|
|
|
priv = netdev_lower_get_next_private_rcu(dev, &(iter)); \
|
|
|
|
priv; \
|
|
|
|
priv = netdev_lower_get_next_private_rcu(dev, &(iter)))
|
|
|
|
|
2014-05-17 05:04:53 +08:00
|
|
|
void *netdev_lower_get_next(struct net_device *dev,
|
|
|
|
struct list_head **iter);
|
2016-07-04 14:23:12 +08:00
|
|
|
|
2014-05-17 05:04:53 +08:00
|
|
|
#define netdev_for_each_lower_dev(dev, ldev, iter) \
|
2016-02-18 01:00:31 +08:00
|
|
|
for (iter = (dev)->adj_list.lower.next, \
|
2014-05-17 05:04:53 +08:00
|
|
|
ldev = netdev_lower_get_next(dev, &(iter)); \
|
|
|
|
ldev; \
|
|
|
|
ldev = netdev_lower_get_next(dev, &(iter)))
|
|
|
|
|
2020-02-15 18:50:21 +08:00
|
|
|
struct net_device *netdev_next_lower_dev_rcu(struct net_device *dev,
|
2016-07-04 14:23:12 +08:00
|
|
|
struct list_head **iter);
|
2016-10-18 10:15:44 +08:00
|
|
|
int netdev_walk_all_lower_dev(struct net_device *dev,
|
|
|
|
int (*fn)(struct net_device *lower_dev,
|
2020-09-26 02:13:12 +08:00
|
|
|
struct netdev_nested_priv *priv),
|
|
|
|
struct netdev_nested_priv *priv);
|
2016-10-18 10:15:44 +08:00
|
|
|
int netdev_walk_all_lower_dev_rcu(struct net_device *dev,
|
|
|
|
int (*fn)(struct net_device *lower_dev,
|
2020-09-26 02:13:12 +08:00
|
|
|
struct netdev_nested_priv *priv),
|
|
|
|
struct netdev_nested_priv *priv);
|
2016-10-18 10:15:44 +08:00
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
void *netdev_adjacent_get_private(struct list_head *adj_list);
|
2013-12-13 10:19:55 +08:00
|
|
|
void *netdev_lower_get_first_private_rcu(struct net_device *dev);
|
2013-09-27 05:48:15 +08:00
|
|
|
struct net_device *netdev_master_upper_dev_get(struct net_device *dev);
|
|
|
|
struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev);
|
2017-10-05 08:48:47 +08:00
|
|
|
int netdev_upper_dev_link(struct net_device *dev, struct net_device *upper_dev,
|
|
|
|
struct netlink_ext_ack *extack);
|
2013-09-27 05:48:15 +08:00
|
|
|
int netdev_master_upper_dev_link(struct net_device *dev,
|
2015-12-03 19:12:10 +08:00
|
|
|
struct net_device *upper_dev,
|
2017-10-05 08:48:47 +08:00
|
|
|
void *upper_priv, void *upper_info,
|
|
|
|
struct netlink_ext_ack *extack);
|
2013-09-27 05:48:15 +08:00
|
|
|
void netdev_upper_dev_unlink(struct net_device *dev,
|
|
|
|
struct net_device *upper_dev);
|
2019-10-22 02:47:56 +08:00
|
|
|
int netdev_adjacent_change_prepare(struct net_device *old_dev,
|
|
|
|
struct net_device *new_dev,
|
|
|
|
struct net_device *dev,
|
|
|
|
struct netlink_ext_ack *extack);
|
|
|
|
void netdev_adjacent_change_commit(struct net_device *old_dev,
|
|
|
|
struct net_device *new_dev,
|
|
|
|
struct net_device *dev);
|
|
|
|
void netdev_adjacent_change_abort(struct net_device *old_dev,
|
|
|
|
struct net_device *new_dev,
|
|
|
|
struct net_device *dev);
|
2014-01-15 04:58:51 +08:00
|
|
|
void netdev_adjacent_rename_links(struct net_device *dev, char *oldname);
|
2013-09-27 05:48:15 +08:00
|
|
|
void *netdev_lower_dev_get_private(struct net_device *dev,
|
|
|
|
struct net_device *lower_dev);
|
2015-12-03 19:12:15 +08:00
|
|
|
void netdev_lower_state_changed(struct net_device *lower_dev,
|
|
|
|
void *lower_state_info);
|
2014-11-16 22:23:05 +08:00
|
|
|
|
|
|
|
/* RSS keys are 40 or 52 bytes long */
|
|
|
|
#define NETDEV_RSS_KEY_LEN 52
|
2016-02-02 11:51:16 +08:00
|
|
|
extern u8 netdev_rss_key[NETDEV_RSS_KEY_LEN] __read_mostly;
|
2014-11-16 22:23:05 +08:00
|
|
|
void netdev_rss_key_fill(void *buffer, size_t len);
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
int skb_checksum_help(struct sk_buff *skb);
|
2017-05-18 21:44:38 +08:00
|
|
|
int skb_crc32c_csum_help(struct sk_buff *skb);
|
2017-05-18 21:44:41 +08:00
|
|
|
int skb_csum_hwoffload_help(struct sk_buff *skb,
|
|
|
|
const netdev_features_t features);
|
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
struct sk_buff *__skb_gso_segment(struct sk_buff *skb,
|
|
|
|
netdev_features_t features, bool tx_path);
|
2022-03-07 20:11:41 +08:00
|
|
|
struct sk_buff *skb_eth_gso_segment(struct sk_buff *skb,
|
|
|
|
netdev_features_t features, __be16 type);
|
2013-09-27 05:48:15 +08:00
|
|
|
struct sk_buff *skb_mac_gso_segment(struct sk_buff *skb,
|
|
|
|
netdev_features_t features);
|
2013-02-06 00:36:38 +08:00
|
|
|
|
2015-02-03 22:48:29 +08:00
|
|
|
struct netdev_bonding_info {
|
|
|
|
ifslave slave;
|
|
|
|
ifbond master;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct netdev_notifier_bonding_info {
|
|
|
|
struct netdev_notifier_info info; /* must be first */
|
|
|
|
struct netdev_bonding_info bonding_info;
|
|
|
|
};
|
|
|
|
|
|
|
|
void netdev_bonding_info_change(struct net_device *dev,
|
|
|
|
struct netdev_bonding_info *bonding_info);
|
|
|
|
|
2019-12-27 22:55:33 +08:00
|
|
|
#if IS_ENABLED(CONFIG_ETHTOOL_NETLINK)
|
|
|
|
void ethtool_notify(struct net_device *dev, unsigned int cmd, const void *data);
|
|
|
|
#else
|
|
|
|
static inline void ethtool_notify(struct net_device *dev, unsigned int cmd,
|
|
|
|
const void *data)
|
|
|
|
{
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2013-02-06 00:36:38 +08:00
|
|
|
static inline
|
|
|
|
struct sk_buff *skb_gso_segment(struct sk_buff *skb, netdev_features_t features)
|
|
|
|
{
|
|
|
|
return __skb_gso_segment(skb, features, true);
|
|
|
|
}
|
2014-03-28 05:26:18 +08:00
|
|
|
__be16 skb_network_protocol(struct sk_buff *skb, int *depth);
|
2013-03-07 17:28:01 +08:00
|
|
|
|
|
|
|
static inline bool can_checksum_protocol(netdev_features_t features,
|
|
|
|
__be16 protocol)
|
|
|
|
{
|
2015-12-15 03:19:44 +08:00
|
|
|
if (protocol == htons(ETH_P_FCOE))
|
|
|
|
return !!(features & NETIF_F_FCOE_CRC);
|
|
|
|
|
|
|
|
/* Assume this is an IP checksum (not SCTP CRC) */
|
|
|
|
|
|
|
|
if (features & NETIF_F_HW_CSUM) {
|
|
|
|
/* Can checksum everything */
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
switch (protocol) {
|
|
|
|
case htons(ETH_P_IP):
|
|
|
|
return !!(features & NETIF_F_IP_CSUM);
|
|
|
|
case htons(ETH_P_IPV6):
|
|
|
|
return !!(features & NETIF_F_IPV6_CSUM);
|
|
|
|
default:
|
|
|
|
return false;
|
|
|
|
}
|
2013-03-07 17:28:01 +08:00
|
|
|
}
|
2013-02-06 00:36:38 +08:00
|
|
|
|
2005-11-11 05:01:24 +08:00
|
|
|
#ifdef CONFIG_BUG
|
2018-11-13 06:47:18 +08:00
|
|
|
void netdev_rx_csum_fault(struct net_device *dev, struct sk_buff *skb);
|
2005-11-11 05:01:24 +08:00
|
|
|
#else
|
2018-11-13 06:47:18 +08:00
|
|
|
static inline void netdev_rx_csum_fault(struct net_device *dev,
|
|
|
|
struct sk_buff *skb)
|
2005-11-11 05:01:24 +08:00
|
|
|
{
|
|
|
|
}
|
|
|
|
#endif
|
2005-04-17 06:20:36 +08:00
|
|
|
/* rx skb timestamps */
|
2013-09-27 05:48:15 +08:00
|
|
|
void net_enable_timestamp(void);
|
|
|
|
void net_disable_timestamp(void);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2022-05-07 04:01:40 +08:00
|
|
|
static inline ktime_t netdev_get_tstamp(struct net_device *dev,
|
|
|
|
const struct skb_shared_hwtstamps *hwtstamps,
|
|
|
|
bool cycles)
|
|
|
|
{
|
|
|
|
const struct net_device_ops *ops = dev->netdev_ops;
|
|
|
|
|
|
|
|
if (ops->ndo_get_tstamp)
|
|
|
|
return ops->ndo_get_tstamp(dev, hwtstamps, cycles);
|
|
|
|
|
|
|
|
return hwtstamps->hwtstamp;
|
|
|
|
}
|
|
|
|
|
2014-08-23 07:21:53 +08:00
|
|
|
static inline netdev_tx_t __netdev_start_xmit(const struct net_device_ops *ops,
|
2014-08-30 12:55:22 +08:00
|
|
|
struct sk_buff *skb, struct net_device *dev,
|
|
|
|
bool more)
|
2014-08-23 07:21:53 +08:00
|
|
|
{
|
2019-04-01 22:42:14 +08:00
|
|
|
__this_cpu_write(softnet_data.xmit.more, more);
|
2014-08-26 06:51:53 +08:00
|
|
|
return ops->ndo_start_xmit(skb, dev);
|
2014-08-23 07:21:53 +08:00
|
|
|
}
|
|
|
|
|
2019-04-01 22:42:13 +08:00
|
|
|
static inline bool netdev_xmit_more(void)
|
|
|
|
{
|
|
|
|
return __this_cpu_read(softnet_data.xmit.more);
|
|
|
|
}
|
|
|
|
|
2014-08-30 12:07:24 +08:00
|
|
|
static inline netdev_tx_t netdev_start_xmit(struct sk_buff *skb, struct net_device *dev,
|
2014-08-30 12:55:22 +08:00
|
|
|
struct netdev_queue *txq, bool more)
|
2014-08-23 07:21:53 +08:00
|
|
|
{
|
|
|
|
const struct net_device_ops *ops = dev->netdev_ops;
|
2018-11-24 17:01:41 +08:00
|
|
|
netdev_tx_t rc;
|
2014-08-23 07:21:53 +08:00
|
|
|
|
2014-08-30 12:55:22 +08:00
|
|
|
rc = __netdev_start_xmit(ops, skb, dev, more);
|
2014-08-30 12:07:24 +08:00
|
|
|
if (rc == NETDEV_TX_OK)
|
|
|
|
txq_trans_update(txq);
|
|
|
|
|
|
|
|
return rc;
|
2014-08-23 07:21:53 +08:00
|
|
|
}
|
|
|
|
|
2017-08-19 04:46:20 +08:00
|
|
|
int netdev_class_create_file_ns(const struct class_attribute *class_attr,
|
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next
Pull networking updates from David Miller:
1) The addition of nftables. No longer will we need protocol aware
firewall filtering modules, it can all live in userspace.
At the core of nftables is a, for lack of a better term, virtual
machine that executes byte codes to inspect packet or metadata
(arriving interface index, etc.) and make verdict decisions.
Besides support for loading packet contents and comparing them, the
interpreter supports lookups in various datastructures as
fundamental operations. For example sets are supports, and
therefore one could create a set of whitelist IP address entries
which have ACCEPT verdicts attached to them, and use the appropriate
byte codes to do such lookups.
Since the interpreted code is composed in userspace, userspace can
do things like optimize things before giving it to the kernel.
Another major improvement is the capability of atomically updating
portions of the ruleset. In the existing netfilter implementation,
one has to update the entire rule set in order to make a change and
this is very expensive.
Userspace tools exist to create nftables rules using existing
netfilter rule sets, but both kernel implementations will need to
co-exist for quite some time as we transition from the old to the
new stuff.
Kudos to Patrick McHardy, Pablo Neira Ayuso, and others who have
worked so hard on this.
2) Daniel Borkmann and Hannes Frederic Sowa made several improvements
to our pseudo-random number generator, mostly used for things like
UDP port randomization and netfitler, amongst other things.
In particular the taus88 generater is updated to taus113, and test
cases are added.
3) Support 64-bit rates in HTB and TBF schedulers, from Eric Dumazet
and Yang Yingliang.
4) Add support for new 577xx tigon3 chips to tg3 driver, from Nithin
Sujir.
5) Fix two fatal flaws in TCP dynamic right sizing, from Eric Dumazet,
Neal Cardwell, and Yuchung Cheng.
6) Allow IP_TOS and IP_TTL to be specified in sendmsg() ancillary
control message data, much like other socket option attributes.
From Francesco Fusco.
7) Allow applications to specify a cap on the rate computed
automatically by the kernel for pacing flows, via a new
SO_MAX_PACING_RATE socket option. From Eric Dumazet.
8) Make the initial autotuned send buffer sizing in TCP more closely
reflect actual needs, from Eric Dumazet.
9) Currently early socket demux only happens for TCP sockets, but we
can do it for connected UDP sockets too. Implementation from Shawn
Bohrer.
10) Refactor inet socket demux with the goal of improving hash demux
performance for listening sockets. With the main goals being able
to use RCU lookups on even request sockets, and eliminating the
listening lock contention. From Eric Dumazet.
11) The bonding layer has many demuxes in it's fast path, and an RCU
conversion was started back in 3.11, several changes here extend the
RCU usage to even more locations. From Ding Tianhong and Wang
Yufen, based upon suggestions by Nikolay Aleksandrov and Veaceslav
Falico.
12) Allow stackability of segmentation offloads to, in particular, allow
segmentation offloading over tunnels. From Eric Dumazet.
13) Significantly improve the handling of secret keys we input into the
various hash functions in the inet hashtables, TCP fast open, as
well as syncookies. From Hannes Frederic Sowa. The key fundamental
operation is "net_get_random_once()" which uses static keys.
Hannes even extended this to ipv4/ipv6 fragmentation handling and
our generic flow dissector.
14) The generic driver layer takes care now to set the driver data to
NULL on device removal, so it's no longer necessary for drivers to
explicitly set it to NULL any more. Many drivers have been cleaned
up in this way, from Jingoo Han.
15) Add a BPF based packet scheduler classifier, from Daniel Borkmann.
16) Improve CRC32 interfaces and generic SKB checksum iterators so that
SCTP's checksumming can more cleanly be handled. Also from Daniel
Borkmann.
17) Add a new PMTU discovery mode, IP_PMTUDISC_INTERFACE, which forces
using the interface MTU value. This helps avoid PMTU attacks,
particularly on DNS servers. From Hannes Frederic Sowa.
18) Use generic XPS for transmit queue steering rather than internal
(re-)implementation in virtio-net. From Jason Wang.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1622 commits)
random32: add test cases for taus113 implementation
random32: upgrade taus88 generator to taus113 from errata paper
random32: move rnd_state to linux/random.h
random32: add prandom_reseed_late() and call when nonblocking pool becomes initialized
random32: add periodic reseeding
random32: fix off-by-one in seeding requirement
PHY: Add RTL8201CP phy_driver to realtek
xtsonic: add missing platform_set_drvdata() in xtsonic_probe()
macmace: add missing platform_set_drvdata() in mace_probe()
ethernet/arc/arc_emac: add missing platform_set_drvdata() in arc_emac_probe()
ipv6: protect for_each_sk_fl_rcu in mem_check with rcu_read_lock_bh
vlan: Implement vlan_dev_get_egress_qos_mask as an inline.
ixgbe: add warning when max_vfs is out of range.
igb: Update link modes display in ethtool
netfilter: push reasm skb through instead of original frag skbs
ip6_output: fragment outgoing reassembled skb properly
MAINTAINERS: mv643xx_eth: take over maintainership from Lennart
net_sched: tbf: support of 64bit rates
ixgbe: deleting dfwd stations out of order can cause null ptr deref
ixgbe: fix build err, num_rx_queues is only available with CONFIG_RPS
...
2013-11-13 16:40:34 +08:00
|
|
|
const void *ns);
|
2017-08-19 04:46:20 +08:00
|
|
|
void netdev_class_remove_file_ns(const struct class_attribute *class_attr,
|
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next
Pull networking updates from David Miller:
1) The addition of nftables. No longer will we need protocol aware
firewall filtering modules, it can all live in userspace.
At the core of nftables is a, for lack of a better term, virtual
machine that executes byte codes to inspect packet or metadata
(arriving interface index, etc.) and make verdict decisions.
Besides support for loading packet contents and comparing them, the
interpreter supports lookups in various datastructures as
fundamental operations. For example sets are supports, and
therefore one could create a set of whitelist IP address entries
which have ACCEPT verdicts attached to them, and use the appropriate
byte codes to do such lookups.
Since the interpreted code is composed in userspace, userspace can
do things like optimize things before giving it to the kernel.
Another major improvement is the capability of atomically updating
portions of the ruleset. In the existing netfilter implementation,
one has to update the entire rule set in order to make a change and
this is very expensive.
Userspace tools exist to create nftables rules using existing
netfilter rule sets, but both kernel implementations will need to
co-exist for quite some time as we transition from the old to the
new stuff.
Kudos to Patrick McHardy, Pablo Neira Ayuso, and others who have
worked so hard on this.
2) Daniel Borkmann and Hannes Frederic Sowa made several improvements
to our pseudo-random number generator, mostly used for things like
UDP port randomization and netfitler, amongst other things.
In particular the taus88 generater is updated to taus113, and test
cases are added.
3) Support 64-bit rates in HTB and TBF schedulers, from Eric Dumazet
and Yang Yingliang.
4) Add support for new 577xx tigon3 chips to tg3 driver, from Nithin
Sujir.
5) Fix two fatal flaws in TCP dynamic right sizing, from Eric Dumazet,
Neal Cardwell, and Yuchung Cheng.
6) Allow IP_TOS and IP_TTL to be specified in sendmsg() ancillary
control message data, much like other socket option attributes.
From Francesco Fusco.
7) Allow applications to specify a cap on the rate computed
automatically by the kernel for pacing flows, via a new
SO_MAX_PACING_RATE socket option. From Eric Dumazet.
8) Make the initial autotuned send buffer sizing in TCP more closely
reflect actual needs, from Eric Dumazet.
9) Currently early socket demux only happens for TCP sockets, but we
can do it for connected UDP sockets too. Implementation from Shawn
Bohrer.
10) Refactor inet socket demux with the goal of improving hash demux
performance for listening sockets. With the main goals being able
to use RCU lookups on even request sockets, and eliminating the
listening lock contention. From Eric Dumazet.
11) The bonding layer has many demuxes in it's fast path, and an RCU
conversion was started back in 3.11, several changes here extend the
RCU usage to even more locations. From Ding Tianhong and Wang
Yufen, based upon suggestions by Nikolay Aleksandrov and Veaceslav
Falico.
12) Allow stackability of segmentation offloads to, in particular, allow
segmentation offloading over tunnels. From Eric Dumazet.
13) Significantly improve the handling of secret keys we input into the
various hash functions in the inet hashtables, TCP fast open, as
well as syncookies. From Hannes Frederic Sowa. The key fundamental
operation is "net_get_random_once()" which uses static keys.
Hannes even extended this to ipv4/ipv6 fragmentation handling and
our generic flow dissector.
14) The generic driver layer takes care now to set the driver data to
NULL on device removal, so it's no longer necessary for drivers to
explicitly set it to NULL any more. Many drivers have been cleaned
up in this way, from Jingoo Han.
15) Add a BPF based packet scheduler classifier, from Daniel Borkmann.
16) Improve CRC32 interfaces and generic SKB checksum iterators so that
SCTP's checksumming can more cleanly be handled. Also from Daniel
Borkmann.
17) Add a new PMTU discovery mode, IP_PMTUDISC_INTERFACE, which forces
using the interface MTU value. This helps avoid PMTU attacks,
particularly on DNS servers. From Hannes Frederic Sowa.
18) Use generic XPS for transmit queue steering rather than internal
(re-)implementation in virtio-net. From Jason Wang.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1622 commits)
random32: add test cases for taus113 implementation
random32: upgrade taus88 generator to taus113 from errata paper
random32: move rnd_state to linux/random.h
random32: add prandom_reseed_late() and call when nonblocking pool becomes initialized
random32: add periodic reseeding
random32: fix off-by-one in seeding requirement
PHY: Add RTL8201CP phy_driver to realtek
xtsonic: add missing platform_set_drvdata() in xtsonic_probe()
macmace: add missing platform_set_drvdata() in mace_probe()
ethernet/arc/arc_emac: add missing platform_set_drvdata() in arc_emac_probe()
ipv6: protect for_each_sk_fl_rcu in mem_check with rcu_read_lock_bh
vlan: Implement vlan_dev_get_egress_qos_mask as an inline.
ixgbe: add warning when max_vfs is out of range.
igb: Update link modes display in ethtool
netfilter: push reasm skb through instead of original frag skbs
ip6_output: fragment outgoing reassembled skb properly
MAINTAINERS: mv643xx_eth: take over maintainership from Lennart
net_sched: tbf: support of 64bit rates
ixgbe: deleting dfwd stations out of order can cause null ptr deref
ixgbe: fix build err, num_rx_queues is only available with CONFIG_RPS
...
2013-11-13 16:40:34 +08:00
|
|
|
const void *ns);
|
2013-09-12 10:29:04 +08:00
|
|
|
|
2017-08-19 04:46:22 +08:00
|
|
|
extern const struct kobj_ns_type_operations net_ns_type_operations;
|
2010-08-05 23:45:15 +08:00
|
|
|
|
2013-09-27 05:48:15 +08:00
|
|
|
const char *netdev_drivername(const struct net_device *dev);
|
2008-07-22 04:31:48 +08:00
|
|
|
|
2014-05-20 14:29:25 +08:00
|
|
|
static inline netdev_features_t netdev_intersect_features(netdev_features_t f1,
|
|
|
|
netdev_features_t f2)
|
|
|
|
{
|
2015-12-15 03:19:44 +08:00
|
|
|
if ((f1 ^ f2) & NETIF_F_HW_CSUM) {
|
|
|
|
if (f1 & NETIF_F_HW_CSUM)
|
2016-01-12 02:19:10 +08:00
|
|
|
f1 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
|
2015-12-15 03:19:44 +08:00
|
|
|
else
|
2016-01-12 02:19:10 +08:00
|
|
|
f2 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
|
2015-12-15 03:19:44 +08:00
|
|
|
}
|
2014-05-20 14:29:25 +08:00
|
|
|
|
2015-12-15 03:19:44 +08:00
|
|
|
return f1 & f2;
|
2014-05-20 14:29:25 +08:00
|
|
|
}
|
|
|
|
|
2011-11-15 23:29:55 +08:00
|
|
|
static inline netdev_features_t netdev_get_wanted_features(
|
|
|
|
struct net_device *dev)
|
2011-02-16 00:59:17 +08:00
|
|
|
{
|
|
|
|
return (dev->features & ~dev->hw_features) | dev->wanted_features;
|
|
|
|
}
|
2011-11-15 23:29:55 +08:00
|
|
|
netdev_features_t netdev_increment_features(netdev_features_t all,
|
|
|
|
netdev_features_t one, netdev_features_t mask);
|
2013-05-16 15:34:53 +08:00
|
|
|
|
|
|
|
/* Allow TSO being used on stacked device :
|
|
|
|
* Performing the GSO segmentation before last device
|
|
|
|
* is a performance improvement.
|
|
|
|
*/
|
|
|
|
static inline netdev_features_t netdev_add_tso_features(netdev_features_t features,
|
|
|
|
netdev_features_t mask)
|
|
|
|
{
|
|
|
|
return netdev_increment_features(features, NETIF_F_ALL_TSO, mask);
|
|
|
|
}
|
|
|
|
|
2011-04-03 13:48:47 +08:00
|
|
|
int __netdev_update_features(struct net_device *dev);
|
2011-02-16 00:59:17 +08:00
|
|
|
void netdev_update_features(struct net_device *dev);
|
2011-05-07 11:22:17 +08:00
|
|
|
void netdev_change_features(struct net_device *dev);
|
2007-08-11 06:47:58 +08:00
|
|
|
|
2009-12-04 07:59:22 +08:00
|
|
|
void netif_stacked_transfer_operstate(const struct net_device *rootdev,
|
|
|
|
struct net_device *dev);
|
|
|
|
|
2015-03-27 13:31:13 +08:00
|
|
|
netdev_features_t passthru_features_check(struct sk_buff *skb,
|
|
|
|
struct net_device *dev,
|
|
|
|
netdev_features_t features);
|
2014-05-05 21:00:44 +08:00
|
|
|
netdev_features_t netif_skb_features(struct sk_buff *skb);
|
2010-10-29 20:14:55 +08:00
|
|
|
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool net_gso_ok(netdev_features_t features, int gso_type)
|
2006-06-28 04:22:38 +08:00
|
|
|
{
|
2016-04-26 02:13:17 +08:00
|
|
|
netdev_features_t feature = (netdev_features_t)gso_type << NETIF_F_GSO_SHIFT;
|
2011-11-16 22:05:33 +08:00
|
|
|
|
|
|
|
/* check flags correspondence */
|
|
|
|
BUILD_BUG_ON(SKB_GSO_TCPV4 != (NETIF_F_TSO >> NETIF_F_GSO_SHIFT));
|
|
|
|
BUILD_BUG_ON(SKB_GSO_DODGY != (NETIF_F_GSO_ROBUST >> NETIF_F_GSO_SHIFT));
|
|
|
|
BUILD_BUG_ON(SKB_GSO_TCP_ECN != (NETIF_F_TSO_ECN >> NETIF_F_GSO_SHIFT));
|
2016-04-11 09:44:51 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_TCP_FIXEDID != (NETIF_F_TSO_MANGLEID >> NETIF_F_GSO_SHIFT));
|
2011-11-16 22:05:33 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_TCPV6 != (NETIF_F_TSO6 >> NETIF_F_GSO_SHIFT));
|
|
|
|
BUILD_BUG_ON(SKB_GSO_FCOE != (NETIF_F_FSO >> NETIF_F_GSO_SHIFT));
|
2014-06-15 14:23:52 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_GRE != (NETIF_F_GSO_GRE >> NETIF_F_GSO_SHIFT));
|
|
|
|
BUILD_BUG_ON(SKB_GSO_GRE_CSUM != (NETIF_F_GSO_GRE_CSUM >> NETIF_F_GSO_SHIFT));
|
2016-05-19 00:06:10 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_IPXIP4 != (NETIF_F_GSO_IPXIP4 >> NETIF_F_GSO_SHIFT));
|
|
|
|
BUILD_BUG_ON(SKB_GSO_IPXIP6 != (NETIF_F_GSO_IPXIP6 >> NETIF_F_GSO_SHIFT));
|
2014-06-15 14:23:52 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL != (NETIF_F_GSO_UDP_TUNNEL >> NETIF_F_GSO_SHIFT));
|
|
|
|
BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM != (NETIF_F_GSO_UDP_TUNNEL_CSUM >> NETIF_F_GSO_SHIFT));
|
2016-04-11 09:45:03 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_PARTIAL != (NETIF_F_GSO_PARTIAL >> NETIF_F_GSO_SHIFT));
|
2014-11-05 01:06:54 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM != (NETIF_F_GSO_TUNNEL_REMCSUM >> NETIF_F_GSO_SHIFT));
|
2016-06-03 02:05:43 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_SCTP != (NETIF_F_GSO_SCTP >> NETIF_F_GSO_SHIFT));
|
2017-04-14 16:05:36 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_ESP != (NETIF_F_GSO_ESP >> NETIF_F_GSO_SHIFT));
|
net: accept UFO datagrams from tuntap and packet
Tuntap and similar devices can inject GSO packets. Accept type
VIRTIO_NET_HDR_GSO_UDP, even though not generating UFO natively.
Processes are expected to use feature negotiation such as TUNSETOFFLOAD
to detect supported offload types and refrain from injecting other
packets. This process breaks down with live migration: guest kernels
do not renegotiate flags, so destination hosts need to expose all
features that the source host does.
Partially revert the UFO removal from 182e0b6b5846~1..d9d30adf5677.
This patch introduces nearly(*) no new code to simplify verification.
It brings back verbatim tuntap UFO negotiation, VIRTIO_NET_HDR_GSO_UDP
insertion and software UFO segmentation.
It does not reinstate protocol stack support, hardware offload
(NETIF_F_UFO), SKB_GSO_UDP tunneling in SKB_GSO_SOFTWARE or reception
of VIRTIO_NET_HDR_GSO_UDP packets in tuntap.
To support SKB_GSO_UDP reappearing in the stack, also reinstate
logic in act_csum and openvswitch. Achieve equivalence with v4.13 HEAD
by squashing in commit 939912216fa8 ("net: skb_needs_check() removes
CHECKSUM_UNNECESSARY check for tx.") and reverting commit 8d63bee643f1
("net: avoid skb_warn_bad_offload false positives on UFO").
(*) To avoid having to bring back skb_shinfo(skb)->ip6_frag_id,
ipv6_proxy_select_ident is changed to return a __be32 and this is
assigned directly to the frag_hdr. Also, SKB_GSO_UDP is inserted
at the end of the enum to minimize code churn.
Tested
Booted a v4.13 guest kernel with QEMU. On a host kernel before this
patch `ethtool -k eth0` shows UFO disabled. After the patch, it is
enabled, same as on a v4.13 host kernel.
A UFO packet sent from the guest appears on the tap device:
host:
nc -l -p -u 8000 &
tcpdump -n -i tap0
guest:
dd if=/dev/zero of=payload.txt bs=1 count=2000
nc -u 192.16.1.1 8000 < payload.txt
Direct tap to tap transmission of VIRTIO_NET_HDR_GSO_UDP succeeds,
packets arriving fragmented:
./with_tap_pair.sh ./tap_send_ufo tap0 tap1
(from https://github.com/wdebruij/kerneltools/tree/master/tests)
Changes
v1 -> v2
- simplified set_offload change (review comment)
- documented test procedure
Link: http://lkml.kernel.org/r/<CAF=yD-LuUeDuL9YWPJD9ykOZ0QCjNeznPDr6whqZ9NGMNF12Mw@mail.gmail.com>
Fixes: fb652fdfe837 ("macvlan/macvtap: Remove NETIF_F_UFO advertisement.")
Reported-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: Willem de Bruijn <willemb@google.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-11-21 23:22:25 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_UDP != (NETIF_F_GSO_UDP >> NETIF_F_GSO_SHIFT));
|
2018-04-27 01:42:21 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_UDP_L4 != (NETIF_F_GSO_UDP_L4 >> NETIF_F_GSO_SHIFT));
|
2020-01-25 18:26:42 +08:00
|
|
|
BUILD_BUG_ON(SKB_GSO_FRAGLIST != (NETIF_F_GSO_FRAGLIST >> NETIF_F_GSO_SHIFT));
|
2011-11-16 22:05:33 +08:00
|
|
|
|
2006-06-30 03:25:53 +08:00
|
|
|
return (features & feature) == feature;
|
2006-06-28 04:22:38 +08:00
|
|
|
}
|
|
|
|
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
static inline bool skb_gso_ok(struct sk_buff *skb, netdev_features_t features)
|
2006-07-01 04:36:35 +08:00
|
|
|
{
|
2009-06-04 12:20:51 +08:00
|
|
|
return net_gso_ok(features, skb_shinfo(skb)->gso_type) &&
|
2010-08-23 15:13:46 +08:00
|
|
|
(!skb_has_frag_list(skb) || (features & NETIF_F_FRAGLIST));
|
2006-07-01 04:36:35 +08:00
|
|
|
}
|
|
|
|
|
2015-04-17 21:45:04 +08:00
|
|
|
static inline bool netif_needs_gso(struct sk_buff *skb,
|
net: Use bool in netdevice.h helpers.
Specifically use it in napi_disable_pending(), napi_schedule_prep(),
napi_reschedule(), netif_tx_queue_stopped(), netif_queue_stopped(),
netif_xmit_stopped(), netif_xmit_frozen_or_stopped(), netif_running(),
__netif_subqueue_stopped(), netif_subqueue_stopped(),
netif_is_multiquue(), netif_carrier_ok(), netif_dormant(),
netif_oper_up(), netif_device_present(), __netif_tx_trylock(),
net_gso_ok(), skb_gso_ok(), netif_needs_gso(), and
netif_is_bond_slave().
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-03-08 10:02:35 +08:00
|
|
|
netdev_features_t features)
|
2006-06-22 17:40:14 +08:00
|
|
|
{
|
2011-01-09 14:23:32 +08:00
|
|
|
return skb_is_gso(skb) && (!skb_gso_ok(skb, features) ||
|
2012-03-17 07:08:11 +08:00
|
|
|
unlikely((skb->ip_summed != CHECKSUM_PARTIAL) &&
|
|
|
|
(skb->ip_summed != CHECKSUM_UNNECESSARY)));
|
2006-06-22 17:40:14 +08:00
|
|
|
}
|
|
|
|
|
2022-05-06 10:51:32 +08:00
|
|
|
void netif_set_tso_max_size(struct net_device *dev, unsigned int size);
|
|
|
|
void netif_set_tso_max_segs(struct net_device *dev, unsigned int segs);
|
2022-05-06 10:51:31 +08:00
|
|
|
void netif_inherit_tso_max(struct net_device *to,
|
|
|
|
const struct net_device *from);
|
|
|
|
|
2013-12-27 05:10:22 +08:00
|
|
|
static inline void skb_gso_error_unwind(struct sk_buff *skb, __be16 protocol,
|
|
|
|
int pulled_hlen, u16 mac_offset,
|
|
|
|
int mac_len)
|
|
|
|
{
|
|
|
|
skb->protocol = protocol;
|
|
|
|
skb->encapsulation = 1;
|
|
|
|
skb_push(skb, pulled_hlen);
|
|
|
|
skb_reset_transport_header(skb);
|
|
|
|
skb->mac_header = mac_offset;
|
|
|
|
skb->network_header = skb->mac_header + mac_len;
|
|
|
|
skb->mac_len = mac_len;
|
|
|
|
}
|
|
|
|
|
2016-03-12 01:07:32 +08:00
|
|
|
static inline bool netif_is_macsec(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_MACSEC;
|
|
|
|
}
|
|
|
|
|
2015-12-04 22:01:31 +08:00
|
|
|
static inline bool netif_is_macvlan(const struct net_device *dev)
|
2013-11-07 01:54:46 +08:00
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_MACVLAN;
|
|
|
|
}
|
|
|
|
|
2015-12-04 22:01:31 +08:00
|
|
|
static inline bool netif_is_macvlan_port(const struct net_device *dev)
|
2014-12-07 07:53:04 +08:00
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_MACVLAN_PORT;
|
|
|
|
}
|
|
|
|
|
2015-12-04 22:01:31 +08:00
|
|
|
static inline bool netif_is_bond_master(const struct net_device *dev)
|
2013-03-12 10:49:01 +08:00
|
|
|
{
|
|
|
|
return dev->flags & IFF_MASTER && dev->priv_flags & IFF_BONDING;
|
|
|
|
}
|
|
|
|
|
2015-12-04 22:01:31 +08:00
|
|
|
static inline bool netif_is_bond_slave(const struct net_device *dev)
|
2011-02-12 14:48:36 +08:00
|
|
|
{
|
|
|
|
return dev->flags & IFF_SLAVE && dev->priv_flags & IFF_BONDING;
|
|
|
|
}
|
|
|
|
|
2012-02-11 23:39:30 +08:00
|
|
|
static inline bool netif_supports_nofcs(struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_SUPP_NOFCS;
|
|
|
|
}
|
|
|
|
|
ipvlan, l3mdev: fix broken l3s mode wrt local routes
While implementing ipvlan l3 and l3s mode for kubernetes CNI plugin,
I ran into the issue that while l3 mode is working fine, l3s mode
does not have any connectivity to kube-apiserver and hence all pods
end up in Error state as well. The ipvlan master device sits on
top of a bond device and hostns traffic to kube-apiserver (also running
in hostns) is DNATed from 10.152.183.1:443 to 139.178.29.207:37573
where the latter is the address of the bond0. While in l3 mode, a
curl to https://10.152.183.1:443 or to https://139.178.29.207:37573
works fine from hostns, neither of them do in case of l3s. In the
latter only a curl to https://127.0.0.1:37573 appeared to work where
for local addresses of bond0 I saw kernel suddenly starting to emit
ARP requests to query HW address of bond0 which remained unanswered
and neighbor entries in INCOMPLETE state. These ARP requests only
happen while in l3s.
Debugging this further, I found the issue is that l3s mode is piggy-
backing on l3 master device, and in this case local routes are using
l3mdev_master_dev_rcu(dev) instead of net->loopback_dev as per commit
f5a0aab84b74 ("net: ipv4: dst for local input routes should use l3mdev
if relevant") and 5f02ce24c269 ("net: l3mdev: Allow the l3mdev to be
a loopback"). I found that reverting them back into using the
net->loopback_dev fixed ipvlan l3s connectivity and got everything
working for the CNI.
Now judging from 4fbae7d83c98 ("ipvlan: Introduce l3s mode") and the
l3mdev paper in [0] the only sole reason why ipvlan l3s is relying
on l3 master device is to get the l3mdev_ip_rcv() receive hook for
setting the dst entry of the input route without adding its own
ipvlan specific hacks into the receive path, however, any l3 domain
semantics beyond just that are breaking l3s operation. Note that
ipvlan also has the ability to dynamically switch its internal
operation from l3 to l3s for all ports via ipvlan_set_port_mode()
at runtime. In any case, l3 vs l3s soley distinguishes itself by
'de-confusing' netfilter through switching skb->dev to ipvlan slave
device late in NF_INET_LOCAL_IN before handing the skb to L4.
Minimal fix taken here is to add a IFF_L3MDEV_RX_HANDLER flag which,
if set from ipvlan setup, gets us only the wanted l3mdev_l3_rcv() hook
without any additional l3mdev semantics on top. This should also have
minimal impact since dev->priv_flags is already hot in cache. With
this set, l3s mode is working fine and I also get things like
masquerading pod traffic on the ipvlan master properly working.
[0] https://netdevconf.org/1.2/papers/ahern-what-is-l3mdev-paper.pdf
Fixes: f5a0aab84b74 ("net: ipv4: dst for local input routes should use l3mdev if relevant")
Fixes: 5f02ce24c269 ("net: l3mdev: Allow the l3mdev to be a loopback")
Fixes: 4fbae7d83c98 ("ipvlan: Introduce l3s mode")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Mahesh Bandewar <maheshb@google.com>
Cc: David Ahern <dsa@cumulusnetworks.com>
Cc: Florian Westphal <fw@strlen.de>
Cc: Martynas Pumputis <m@lambda.lt>
Acked-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-01-30 19:49:48 +08:00
|
|
|
static inline bool netif_has_l3_rx_handler(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_L3MDEV_RX_HANDLER;
|
|
|
|
}
|
|
|
|
|
2015-09-30 11:07:10 +08:00
|
|
|
static inline bool netif_is_l3_master(const struct net_device *dev)
|
2015-08-14 04:59:00 +08:00
|
|
|
{
|
2015-09-30 11:07:10 +08:00
|
|
|
return dev->priv_flags & IFF_L3MDEV_MASTER;
|
2015-08-14 04:59:00 +08:00
|
|
|
}
|
|
|
|
|
2015-10-05 23:51:24 +08:00
|
|
|
static inline bool netif_is_l3_slave(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_L3MDEV_SLAVE;
|
|
|
|
}
|
|
|
|
|
2015-08-27 15:31:19 +08:00
|
|
|
static inline bool netif_is_bridge_master(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_EBRIDGE;
|
|
|
|
}
|
|
|
|
|
2015-11-17 04:43:45 +08:00
|
|
|
static inline bool netif_is_bridge_port(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_BRIDGE_PORT;
|
|
|
|
}
|
|
|
|
|
2015-08-27 15:31:20 +08:00
|
|
|
static inline bool netif_is_ovs_master(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_OPENVSWITCH;
|
|
|
|
}
|
|
|
|
|
2017-04-18 22:55:36 +08:00
|
|
|
static inline bool netif_is_ovs_port(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_OVS_DATAPATH;
|
|
|
|
}
|
|
|
|
|
ipv4: route: Ignore output interface in FIB lookup for PMTU route
Currently, processes sending traffic to a local bridge with an
encapsulation device as a port don't get ICMP errors if they exceed
the PMTU of the encapsulated link.
David Ahern suggested this as a hack, but it actually looks like
the correct solution: when we update the PMTU for a given destination
by means of updating or creating a route exception, the encapsulation
might trigger this because of PMTU discovery happening either on the
encapsulation device itself, or its lower layer. This happens on
bridged encapsulations only.
The output interface shouldn't matter, because we already have a
valid destination. Drop the output interface restriction from the
associated route lookup.
For UDP tunnels, we will now have a route exception created for the
encapsulation itself, with a MTU value reflecting its headroom, which
allows a bridge forwarding IP packets originated locally to deliver
errors back to the sending socket.
The behaviour is now consistent with IPv6 and verified with selftests
pmtu_ipv{4,6}_br_{geneve,vxlan}{4,6}_exception introduced later in
this series.
v2:
- reset output interface only for bridge ports (David Ahern)
- add and use netif_is_any_bridge_port() helper (David Ahern)
Suggested-by: David Ahern <dsahern@gmail.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-08-04 13:53:42 +08:00
|
|
|
static inline bool netif_is_any_bridge_port(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return netif_is_bridge_port(dev) || netif_is_ovs_port(dev);
|
|
|
|
}
|
|
|
|
|
2015-12-04 22:01:31 +08:00
|
|
|
static inline bool netif_is_team_master(const struct net_device *dev)
|
2015-12-03 19:12:06 +08:00
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_TEAM;
|
|
|
|
}
|
|
|
|
|
2015-12-04 22:01:31 +08:00
|
|
|
static inline bool netif_is_team_port(const struct net_device *dev)
|
2015-12-03 19:12:07 +08:00
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_TEAM_PORT;
|
|
|
|
}
|
|
|
|
|
2015-12-04 22:01:31 +08:00
|
|
|
static inline bool netif_is_lag_master(const struct net_device *dev)
|
2015-12-03 19:12:08 +08:00
|
|
|
{
|
|
|
|
return netif_is_bond_master(dev) || netif_is_team_master(dev);
|
|
|
|
}
|
|
|
|
|
2015-12-04 22:01:31 +08:00
|
|
|
static inline bool netif_is_lag_port(const struct net_device *dev)
|
2015-12-03 19:12:09 +08:00
|
|
|
{
|
|
|
|
return netif_is_bond_slave(dev) || netif_is_team_port(dev);
|
|
|
|
}
|
|
|
|
|
2016-02-09 08:05:03 +08:00
|
|
|
static inline bool netif_is_rxfh_configured(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_RXFH_CONFIGURED;
|
|
|
|
}
|
|
|
|
|
2018-05-25 00:55:13 +08:00
|
|
|
static inline bool netif_is_failover(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_FAILOVER;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool netif_is_failover_slave(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
return dev->priv_flags & IFF_FAILOVER_SLAVE;
|
|
|
|
}
|
|
|
|
|
2014-10-06 09:38:35 +08:00
|
|
|
/* This device needs to keep skb dst for qdisc enqueue or ndo_start_xmit() */
|
|
|
|
static inline void netif_keep_dst(struct net_device *dev)
|
|
|
|
{
|
|
|
|
dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM);
|
|
|
|
}
|
|
|
|
|
2016-07-15 00:00:10 +08:00
|
|
|
/* return true if dev can't cope with mtu frames that need vlan tag insertion */
|
|
|
|
static inline bool netif_reduces_vlan_mtu(struct net_device *dev)
|
|
|
|
{
|
|
|
|
/* TODO: reserve and use an additional IFF bit, if we get more users */
|
2021-10-10 12:03:29 +08:00
|
|
|
return netif_is_macsec(dev);
|
2016-07-15 00:00:10 +08:00
|
|
|
}
|
|
|
|
|
2008-11-08 14:54:20 +08:00
|
|
|
extern struct pernet_operations __net_initdata loopback_net_ops;
|
2009-04-20 12:49:28 +08:00
|
|
|
|
2010-02-09 19:49:47 +08:00
|
|
|
/* Logging, debugging and troubleshooting/diagnostic helpers. */
|
|
|
|
|
|
|
|
/* netdev_printk helpers, similar to dev_printk */
|
|
|
|
|
|
|
|
static inline const char *netdev_name(const struct net_device *dev)
|
|
|
|
{
|
2014-07-18 01:46:09 +08:00
|
|
|
if (!dev->name[0] || strchr(dev->name, '%'))
|
|
|
|
return "(unnamed net_device)";
|
2010-02-09 19:49:47 +08:00
|
|
|
return dev->name;
|
|
|
|
}
|
|
|
|
|
2014-07-18 01:46:10 +08:00
|
|
|
static inline const char *netdev_reg_state(const struct net_device *dev)
|
|
|
|
{
|
|
|
|
switch (dev->reg_state) {
|
|
|
|
case NETREG_UNINITIALIZED: return " (uninitialized)";
|
|
|
|
case NETREG_REGISTERED: return "";
|
|
|
|
case NETREG_UNREGISTERING: return " (unregistering)";
|
|
|
|
case NETREG_UNREGISTERED: return " (unregistered)";
|
|
|
|
case NETREG_RELEASED: return " (released)";
|
|
|
|
case NETREG_DUMMY: return " (dummy)";
|
|
|
|
}
|
|
|
|
|
|
|
|
WARN_ONCE(1, "%s: unknown reg_state %d\n", dev->name, dev->reg_state);
|
|
|
|
return " (unknown)";
|
|
|
|
}
|
|
|
|
|
2011-03-02 05:33:13 +08:00
|
|
|
#define MODULE_ALIAS_NETDEV(device) \
|
|
|
|
MODULE_ALIAS("netdev-" device)
|
|
|
|
|
2010-02-09 19:49:47 +08:00
|
|
|
/*
|
|
|
|
* netdev_WARN() acts like dev_printk(), but with the key difference
|
|
|
|
* of using a WARN/WARN_ON to get the message out, including the
|
|
|
|
* file/line information and a backtrace.
|
|
|
|
*/
|
|
|
|
#define netdev_WARN(dev, format, args...) \
|
2018-01-07 18:08:36 +08:00
|
|
|
WARN(1, "netdevice: %s%s: " format, netdev_name(dev), \
|
2014-07-18 01:46:10 +08:00
|
|
|
netdev_reg_state(dev), ##args)
|
2010-02-09 19:49:47 +08:00
|
|
|
|
2018-01-07 18:08:35 +08:00
|
|
|
#define netdev_WARN_ONCE(dev, format, args...) \
|
2018-01-07 18:08:36 +08:00
|
|
|
WARN_ONCE(1, "netdevice: %s%s: " format, netdev_name(dev), \
|
2017-09-17 18:43:58 +08:00
|
|
|
netdev_reg_state(dev), ##args)
|
|
|
|
|
2013-02-19 03:20:33 +08:00
|
|
|
/*
|
|
|
|
* The list of packet types we will receive (as opposed to discard)
|
|
|
|
* and the routines to invoke.
|
|
|
|
*
|
|
|
|
* Why 16. Because with 16 the only overlap we get on a hash of the
|
|
|
|
* low nibble of the protocol value is RARP/SNAP/X.25.
|
|
|
|
*
|
|
|
|
* 0800 IP
|
|
|
|
* 0001 802.3
|
|
|
|
* 0002 AX.25
|
|
|
|
* 0004 802.2
|
|
|
|
* 8035 RARP
|
|
|
|
* 0005 SNAP
|
|
|
|
* 0805 X.25
|
|
|
|
* 0806 ARP
|
|
|
|
* 8137 IPX
|
|
|
|
* 0009 Localtalk
|
|
|
|
* 86DD IPv6
|
|
|
|
*/
|
|
|
|
#define PTYPE_HASH_SIZE (16)
|
|
|
|
#define PTYPE_HASH_MASK (PTYPE_HASH_SIZE - 1)
|
|
|
|
|
2021-03-22 19:31:48 +08:00
|
|
|
extern struct list_head ptype_all __read_mostly;
|
|
|
|
extern struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
|
|
|
|
|
2019-07-02 05:38:49 +08:00
|
|
|
extern struct net_device *blackhole_netdev;
|
|
|
|
|
2009-05-28 06:48:07 +08:00
|
|
|
#endif /* _LINUX_NETDEVICE_H */
|