forked from Minki/linux
03cfda4fa6
KMSAN is still not happy [1]. I missed that passive connections do not inherit their sk_rx_queue_mapping values from the request socket, but instead tcp_child_process() is calling sk_mark_napi_id(child, skb) We have many sk_mark_napi_id() callers, so I am providing a new helper, forcing the setting sk_rx_queue_mapping and sk_napi_id. Note that we had no KMSAN report for sk_napi_id because passive connections got a copy of this field from the listener. sk_rx_queue_mapping in the other hand is inside the sk_dontcopy_begin/sk_dontcopy_end so sk_clone_lock() leaves this field uninitialized. We might remove dead code populating req->sk_rx_queue_mapping in the future. [1] BUG: KMSAN: uninit-value in __sk_rx_queue_set include/net/sock.h:1924 [inline] BUG: KMSAN: uninit-value in sk_rx_queue_update include/net/sock.h:1938 [inline] BUG: KMSAN: uninit-value in sk_mark_napi_id include/net/busy_poll.h:136 [inline] BUG: KMSAN: uninit-value in tcp_child_process+0xb42/0x1050 net/ipv4/tcp_minisocks.c:833 __sk_rx_queue_set include/net/sock.h:1924 [inline] sk_rx_queue_update include/net/sock.h:1938 [inline] sk_mark_napi_id include/net/busy_poll.h:136 [inline] tcp_child_process+0xb42/0x1050 net/ipv4/tcp_minisocks.c:833 tcp_v4_rcv+0x3d83/0x4ed0 net/ipv4/tcp_ipv4.c:2066 ip_protocol_deliver_rcu+0x760/0x10b0 net/ipv4/ip_input.c:204 ip_local_deliver_finish net/ipv4/ip_input.c:231 [inline] NF_HOOK include/linux/netfilter.h:307 [inline] ip_local_deliver+0x584/0x8c0 net/ipv4/ip_input.c:252 dst_input include/net/dst.h:460 [inline] ip_sublist_rcv_finish net/ipv4/ip_input.c:551 [inline] ip_list_rcv_finish net/ipv4/ip_input.c:601 [inline] ip_sublist_rcv+0x11fd/0x1520 net/ipv4/ip_input.c:609 ip_list_rcv+0x95f/0x9a0 net/ipv4/ip_input.c:644 __netif_receive_skb_list_ptype net/core/dev.c:5505 [inline] __netif_receive_skb_list_core+0xe34/0x1240 net/core/dev.c:5553 __netif_receive_skb_list+0x7fc/0x960 net/core/dev.c:5605 netif_receive_skb_list_internal+0x868/0xde0 net/core/dev.c:5696 gro_normal_list net/core/dev.c:5850 [inline] napi_complete_done+0x579/0xdd0 net/core/dev.c:6587 virtqueue_napi_complete drivers/net/virtio_net.c:339 [inline] virtnet_poll+0x17b6/0x2350 drivers/net/virtio_net.c:1557 __napi_poll+0x14e/0xbc0 net/core/dev.c:7020 napi_poll net/core/dev.c:7087 [inline] net_rx_action+0x824/0x1880 net/core/dev.c:7174 __do_softirq+0x1fe/0x7eb kernel/softirq.c:558 run_ksoftirqd+0x33/0x50 kernel/softirq.c:920 smpboot_thread_fn+0x616/0xbf0 kernel/smpboot.c:164 kthread+0x721/0x850 kernel/kthread.c:327 ret_from_fork+0x1f/0x30 Uninit was created at: __alloc_pages+0xbc7/0x10a0 mm/page_alloc.c:5409 alloc_pages+0x8a5/0xb80 alloc_slab_page mm/slub.c:1810 [inline] allocate_slab+0x287/0x1c20 mm/slub.c:1947 new_slab mm/slub.c:2010 [inline] ___slab_alloc+0xbdf/0x1e90 mm/slub.c:3039 __slab_alloc mm/slub.c:3126 [inline] slab_alloc_node mm/slub.c:3217 [inline] slab_alloc mm/slub.c:3259 [inline] kmem_cache_alloc+0xbb3/0x11c0 mm/slub.c:3264 sk_prot_alloc+0xeb/0x570 net/core/sock.c:1914 sk_clone_lock+0xd6/0x1940 net/core/sock.c:2118 inet_csk_clone_lock+0x8d/0x6a0 net/ipv4/inet_connection_sock.c:956 tcp_create_openreq_child+0xb1/0x1ef0 net/ipv4/tcp_minisocks.c:453 tcp_v4_syn_recv_sock+0x268/0x2710 net/ipv4/tcp_ipv4.c:1563 tcp_check_req+0x207c/0x2a30 net/ipv4/tcp_minisocks.c:765 tcp_v4_rcv+0x36f5/0x4ed0 net/ipv4/tcp_ipv4.c:2047 ip_protocol_deliver_rcu+0x760/0x10b0 net/ipv4/ip_input.c:204 ip_local_deliver_finish net/ipv4/ip_input.c:231 [inline] NF_HOOK include/linux/netfilter.h:307 [inline] ip_local_deliver+0x584/0x8c0 net/ipv4/ip_input.c:252 dst_input include/net/dst.h:460 [inline] ip_sublist_rcv_finish net/ipv4/ip_input.c:551 [inline] ip_list_rcv_finish net/ipv4/ip_input.c:601 [inline] ip_sublist_rcv+0x11fd/0x1520 net/ipv4/ip_input.c:609 ip_list_rcv+0x95f/0x9a0 net/ipv4/ip_input.c:644 __netif_receive_skb_list_ptype net/core/dev.c:5505 [inline] __netif_receive_skb_list_core+0xe34/0x1240 net/core/dev.c:5553 __netif_receive_skb_list+0x7fc/0x960 net/core/dev.c:5605 netif_receive_skb_list_internal+0x868/0xde0 net/core/dev.c:5696 gro_normal_list net/core/dev.c:5850 [inline] napi_complete_done+0x579/0xdd0 net/core/dev.c:6587 virtqueue_napi_complete drivers/net/virtio_net.c:339 [inline] virtnet_poll+0x17b6/0x2350 drivers/net/virtio_net.c:1557 __napi_poll+0x14e/0xbc0 net/core/dev.c:7020 napi_poll net/core/dev.c:7087 [inline] net_rx_action+0x824/0x1880 net/core/dev.c:7174 __do_softirq+0x1fe/0x7eb kernel/softirq.c:558 Fixes:342159ee39
("net: avoid dirtying sk->sk_rx_queue_mapping") Fixes:a37a0ee4d2
("net: avoid uninit-value from tcp_conn_request") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Tested-by: Alexander Potapenko <glider@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
178 lines
4.3 KiB
C
178 lines
4.3 KiB
C
/* 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)
|
|
{
|
|
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)
|
|
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_update(sk, skb);
|
|
}
|
|
|
|
/* Variant of sk_mark_napi_id() for passive flow setup,
|
|
* as sk->sk_napi_id and sk->sk_rx_queue_mapping content
|
|
* needs to be set.
|
|
*/
|
|
static inline void sk_mark_napi_id_set(struct sock *sk,
|
|
const struct sk_buff *skb)
|
|
{
|
|
#ifdef CONFIG_NET_RX_BUSY_POLL
|
|
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
|
|
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 */
|