linux/include/net/busy_poll.h

165 lines
3.9 KiB
C
Raw Normal View History

/* SPDX-License-Identifier: GPL-2.0-only */
/*
* net busy poll support
* Copyright(c) 2013 Intel Corporation.
*
* Author: Eliezer Tamir
*
* Contact Information:
* e1000-devel Mailing List <e1000-devel@lists.sourceforge.net>
*/
#ifndef _LINUX_NET_BUSY_POLL_H
#define _LINUX_NET_BUSY_POLL_H
#include <linux/netdevice.h>
#include <linux/sched/clock.h>
#include <linux/sched/signal.h>
#include <net/ip.h>
/* 0 - Reserved to indicate value not set
* 1..NR_CPUS - Reserved for sender_cpu
* NR_CPUS+1..~0 - Region available for NAPI IDs
*/
#define MIN_NAPI_ID ((unsigned int)(NR_CPUS + 1))
#define BUSY_POLL_BUDGET 8
#ifdef CONFIG_NET_RX_BUSY_POLL
struct napi_struct;
extern unsigned int sysctl_net_busy_read __read_mostly;
extern unsigned int sysctl_net_busy_poll __read_mostly;
static inline bool net_busy_loop_on(void)
{
return sysctl_net_busy_poll;
}
static inline bool sk_can_busy_loop(const struct sock *sk)
{
net: annotate data race around sk_ll_usec sk_ll_usec is read locklessly from sk_can_busy_loop() while another thread can change its value in sock_setsockopt() This is correct but needs annotations. BUG: KCSAN: data-race in __skb_try_recv_datagram / sock_setsockopt write to 0xffff88814eb5f904 of 4 bytes by task 14011 on cpu 0: sock_setsockopt+0x1287/0x2090 net/core/sock.c:1175 __sys_setsockopt+0x14f/0x200 net/socket.c:2100 __do_sys_setsockopt net/socket.c:2115 [inline] __se_sys_setsockopt net/socket.c:2112 [inline] __x64_sys_setsockopt+0x62/0x70 net/socket.c:2112 do_syscall_64+0x4a/0x90 arch/x86/entry/common.c:47 entry_SYSCALL_64_after_hwframe+0x44/0xae read to 0xffff88814eb5f904 of 4 bytes by task 14001 on cpu 1: sk_can_busy_loop include/net/busy_poll.h:41 [inline] __skb_try_recv_datagram+0x14f/0x320 net/core/datagram.c:273 unix_dgram_recvmsg+0x14c/0x870 net/unix/af_unix.c:2101 unix_seqpacket_recvmsg+0x5a/0x70 net/unix/af_unix.c:2067 ____sys_recvmsg+0x15d/0x310 include/linux/uio.h:244 ___sys_recvmsg net/socket.c:2598 [inline] do_recvmmsg+0x35c/0x9f0 net/socket.c:2692 __sys_recvmmsg net/socket.c:2771 [inline] __do_sys_recvmmsg net/socket.c:2794 [inline] __se_sys_recvmmsg net/socket.c:2787 [inline] __x64_sys_recvmmsg+0xcf/0x150 net/socket.c:2787 do_syscall_64+0x4a/0x90 arch/x86/entry/common.c:47 entry_SYSCALL_64_after_hwframe+0x44/0xae value changed: 0x00000000 -> 0x00000101 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 14001 Comm: syz-executor.3 Not tainted 5.13.0-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2021-06-29 17:12:45 +03:00
return READ_ONCE(sk->sk_ll_usec) && !signal_pending(current);
}
bool sk_busy_loop_end(void *p, unsigned long start_time);
void napi_busy_loop(unsigned int napi_id,
bool (*loop_end)(void *, unsigned long),
void *loop_end_arg, bool prefer_busy_poll, u16 budget);
#else /* CONFIG_NET_RX_BUSY_POLL */
static inline unsigned long net_busy_loop_on(void)
{
return 0;
}
static inline bool sk_can_busy_loop(struct sock *sk)
{
return false;
}
#endif /* CONFIG_NET_RX_BUSY_POLL */
static inline unsigned long busy_loop_current_time(void)
{
#ifdef CONFIG_NET_RX_BUSY_POLL
return (unsigned long)(local_clock() >> 10);
#else
return 0;
#endif
}
/* in poll/select we use the global sysctl_net_ll_poll value */
static inline bool busy_loop_timeout(unsigned long start_time)
{
#ifdef CONFIG_NET_RX_BUSY_POLL
unsigned long bp_usec = READ_ONCE(sysctl_net_busy_poll);
if (bp_usec) {
unsigned long end_time = start_time + bp_usec;
unsigned long now = busy_loop_current_time();
return time_after(now, end_time);
}
#endif
return true;
}
static inline bool sk_busy_loop_timeout(struct sock *sk,
unsigned long start_time)
{
#ifdef CONFIG_NET_RX_BUSY_POLL
unsigned long bp_usec = READ_ONCE(sk->sk_ll_usec);
if (bp_usec) {
unsigned long end_time = start_time + bp_usec;
unsigned long now = busy_loop_current_time();
return time_after(now, end_time);
}
#endif
return true;
}
static inline void sk_busy_loop(struct sock *sk, int nonblock)
{
#ifdef CONFIG_NET_RX_BUSY_POLL
unsigned int napi_id = READ_ONCE(sk->sk_napi_id);
if (napi_id >= MIN_NAPI_ID)
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
2020-11-30 21:51:56 +03:00
napi_busy_loop(napi_id, nonblock ? NULL : sk_busy_loop_end, sk,
READ_ONCE(sk->sk_prefer_busy_poll),
READ_ONCE(sk->sk_busy_poll_budget) ?: BUSY_POLL_BUDGET);
#endif
}
/* used in the NIC receive handler to mark the skb */
static inline void skb_mark_napi_id(struct sk_buff *skb,
struct napi_struct *napi)
{
#ifdef CONFIG_NET_RX_BUSY_POLL
/* If the skb was already marked with a valid NAPI ID, avoid overwriting
* it.
*/
if (skb->napi_id < MIN_NAPI_ID)
skb->napi_id = napi->napi_id;
#endif
}
/* used in the protocol hanlder to propagate the napi_id to the socket */
static inline void sk_mark_napi_id(struct sock *sk, const struct sk_buff *skb)
{
#ifdef CONFIG_NET_RX_BUSY_POLL
if (unlikely(READ_ONCE(sk->sk_napi_id) != skb->napi_id))
WRITE_ONCE(sk->sk_napi_id, skb->napi_id);
#endif
sk_rx_queue_set(sk, skb);
}
static inline void __sk_mark_napi_id_once(struct sock *sk, unsigned int napi_id)
{
#ifdef CONFIG_NET_RX_BUSY_POLL
net: annotate lockless accesses to sk->sk_napi_id We already annotated most accesses to sk->sk_napi_id We missed sk_mark_napi_id() and sk_mark_napi_id_once() which might be called without socket lock held in UDP stack. KCSAN reported : BUG: KCSAN: data-race in udpv6_queue_rcv_one_skb / udpv6_queue_rcv_one_skb write to 0xffff888121c6d108 of 4 bytes by interrupt on cpu 0: sk_mark_napi_id include/net/busy_poll.h:125 [inline] __udpv6_queue_rcv_skb net/ipv6/udp.c:571 [inline] udpv6_queue_rcv_one_skb+0x70c/0xb40 net/ipv6/udp.c:672 udpv6_queue_rcv_skb+0xb5/0x400 net/ipv6/udp.c:689 udp6_unicast_rcv_skb.isra.0+0xd7/0x180 net/ipv6/udp.c:832 __udp6_lib_rcv+0x69c/0x1770 net/ipv6/udp.c:913 udpv6_rcv+0x2b/0x40 net/ipv6/udp.c:1015 ip6_protocol_deliver_rcu+0x22a/0xbe0 net/ipv6/ip6_input.c:409 ip6_input_finish+0x30/0x50 net/ipv6/ip6_input.c:450 NF_HOOK include/linux/netfilter.h:305 [inline] NF_HOOK include/linux/netfilter.h:299 [inline] ip6_input+0x177/0x190 net/ipv6/ip6_input.c:459 dst_input include/net/dst.h:442 [inline] ip6_rcv_finish+0x110/0x140 net/ipv6/ip6_input.c:76 NF_HOOK include/linux/netfilter.h:305 [inline] NF_HOOK include/linux/netfilter.h:299 [inline] ipv6_rcv+0x1a1/0x1b0 net/ipv6/ip6_input.c:284 __netif_receive_skb_one_core+0xa7/0xe0 net/core/dev.c:5010 __netif_receive_skb+0x37/0xf0 net/core/dev.c:5124 process_backlog+0x1d3/0x420 net/core/dev.c:5955 napi_poll net/core/dev.c:6392 [inline] net_rx_action+0x3ae/0xa90 net/core/dev.c:6460 write to 0xffff888121c6d108 of 4 bytes by interrupt on cpu 1: sk_mark_napi_id include/net/busy_poll.h:125 [inline] __udpv6_queue_rcv_skb net/ipv6/udp.c:571 [inline] udpv6_queue_rcv_one_skb+0x70c/0xb40 net/ipv6/udp.c:672 udpv6_queue_rcv_skb+0xb5/0x400 net/ipv6/udp.c:689 udp6_unicast_rcv_skb.isra.0+0xd7/0x180 net/ipv6/udp.c:832 __udp6_lib_rcv+0x69c/0x1770 net/ipv6/udp.c:913 udpv6_rcv+0x2b/0x40 net/ipv6/udp.c:1015 ip6_protocol_deliver_rcu+0x22a/0xbe0 net/ipv6/ip6_input.c:409 ip6_input_finish+0x30/0x50 net/ipv6/ip6_input.c:450 NF_HOOK include/linux/netfilter.h:305 [inline] NF_HOOK include/linux/netfilter.h:299 [inline] ip6_input+0x177/0x190 net/ipv6/ip6_input.c:459 dst_input include/net/dst.h:442 [inline] ip6_rcv_finish+0x110/0x140 net/ipv6/ip6_input.c:76 NF_HOOK include/linux/netfilter.h:305 [inline] NF_HOOK include/linux/netfilter.h:299 [inline] ipv6_rcv+0x1a1/0x1b0 net/ipv6/ip6_input.c:284 __netif_receive_skb_one_core+0xa7/0xe0 net/core/dev.c:5010 __netif_receive_skb+0x37/0xf0 net/core/dev.c:5124 process_backlog+0x1d3/0x420 net/core/dev.c:5955 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 10890 Comm: syz-executor.0 Not tainted 5.4.0-rc3+ #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Fixes: e68b6e50fa35 ("udp: enable busy polling for all sockets") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-10-29 20:54:44 +03:00
if (!READ_ONCE(sk->sk_napi_id))
WRITE_ONCE(sk->sk_napi_id, napi_id);
#endif
}
/* variant used for unconnected sockets */
static inline void sk_mark_napi_id_once(struct sock *sk,
const struct sk_buff *skb)
{
#ifdef CONFIG_NET_RX_BUSY_POLL
__sk_mark_napi_id_once(sk, skb->napi_id);
#endif
}
static inline void sk_mark_napi_id_once_xdp(struct sock *sk,
const struct xdp_buff *xdp)
{
#ifdef CONFIG_NET_RX_BUSY_POLL
__sk_mark_napi_id_once(sk, xdp->rxq->napi_id);
#endif
}
#endif /* _LINUX_NET_BUSY_POLL_H */