From 4648dc97af9d496218a05353b0e442b3dfa6aaab Mon Sep 17 00:00:00 2001 From: Neal Cardwell Date: Mon, 5 Mar 2012 19:35:04 +0000 Subject: tcp: fix tcp_shift_skb_data() to not shift SACKed data below snd_una This commit fixes tcp_shift_skb_data() so that it does not shift SACKed data below snd_una. This fixes an issue whose symptoms exactly match reports showing tp->sacked_out going negative since 3.3.0-rc4 (see "WARNING: at net/ipv4/tcp_input.c:3418" thread on netdev). Since 2008 (832d11c5cd076abc0aa1eaf7be96c81d1a59ce41) tcp_shift_skb_data() had been shifting SACKed ranges that were below snd_una. It checked that the *end* of the skb it was about to shift from was above snd_una, but did not check that the end of the actual shifted range was above snd_una; this commit adds that check. Shifting SACKed ranges below snd_una is problematic because for such ranges tcp_sacktag_one() short-circuits: it does not declare anything as SACKed and does not increase sacked_out. Before the fixes in commits cc9a672ee522d4805495b98680f4a3db5d0a0af9 and daef52bab1fd26e24e8e9578f8fb33ba1d0cb412, shifting SACKed ranges below snd_una happened to work because tcp_shifted_skb() was always (incorrectly) passing in to tcp_sacktag_one() an skb whose end_seq tcp_shift_skb_data() had already guaranteed was beyond snd_una. Hence tcp_sacktag_one() never short-circuited and always increased tp->sacked_out in this case. After those two fixes, my testing has verified that shifting SACKed ranges below snd_una could cause tp->sacked_out to go negative with the following sequence of events: (1) tcp_shift_skb_data() sees an skb whose end_seq is beyond snd_una, then shifts a prefix of that skb that is below snd_una (2) tcp_shifted_skb() increments the packet count of the already-SACKed prev sk_buff (3) tcp_sacktag_one() sees the end of the new SACKed range is below snd_una, so it short-circuits and doesn't increase tp->sacked_out (5) tcp_clean_rtx_queue() sees the SACKed skb has been ACKed, decrements tp->sacked_out by this "inflated" pcount that was missing a matching increase in tp->sacked_out, and hence tp->sacked_out underflows to a u32 like 0xFFFFFFFF, which casted to s32 is negative. (6) this leads to the warnings seen in the recent "WARNING: at net/ipv4/tcp_input.c:3418" thread on the netdev list; e.g.: tcp_input.c:3418 WARN_ON((int)tp->sacked_out < 0); More generally, I think this bug can be tickled in some cases where two or more ACKs from the receiver are lost and then a DSACK arrives that is immediately above an existing SACKed skb in the write queue. This fix changes tcp_shift_skb_data() to abort this sequence at step (1) in the scenario above by noticing that the bytes are below snd_una and not shifting them. Signed-off-by: Neal Cardwell Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index d9b83d198c3d..b5e315f13641 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -1585,6 +1585,10 @@ static struct sk_buff *tcp_shift_skb_data(struct sock *sk, struct sk_buff *skb, } } + /* tcp_sacktag_one() won't SACK-tag ranges below snd_una */ + if (!after(TCP_SKB_CB(skb)->seq + len, tp->snd_una)) + goto fallback; + if (!skb_shift(prev, skb, len)) goto fallback; if (!tcp_shifted_skb(sk, skb, state, pcount, len, mss, dup_sack)) -- cgit v1.2.3 From 5faa5df1fa2024bd750089ff21dcc4191798263d Mon Sep 17 00:00:00 2001 From: Steffen Klassert Date: Tue, 6 Mar 2012 21:20:26 +0000 Subject: inetpeer: Invalidate the inetpeer tree along with the routing cache We initialize the routing metrics with the values cached on the inetpeer in rt_init_metrics(). So if we have the metrics cached on the inetpeer, we ignore the user configured fib_metrics. To fix this issue, we replace the old tree with a fresh initialized inet_peer_base. The old tree is removed later with a delayed work queue. Signed-off-by: Steffen Klassert Signed-off-by: David S. Miller --- net/ipv4/inetpeer.c | 80 ++++++++++++++++++++++++++++++++++++++++++++++++++++- net/ipv4/route.c | 1 + 2 files changed, 80 insertions(+), 1 deletion(-) (limited to 'net/ipv4') diff --git a/net/ipv4/inetpeer.c b/net/ipv4/inetpeer.c index bf4a9c4808e1..deea2e96b7f2 100644 --- a/net/ipv4/inetpeer.c +++ b/net/ipv4/inetpeer.c @@ -17,6 +17,7 @@ #include #include #include +#include #include #include #include @@ -66,6 +67,11 @@ static struct kmem_cache *peer_cachep __read_mostly; +static LIST_HEAD(gc_list); +static const int gc_delay = 60 * HZ; +static struct delayed_work gc_work; +static DEFINE_SPINLOCK(gc_lock); + #define node_height(x) x->avl_height #define peer_avl_empty ((struct inet_peer *)&peer_fake_node) @@ -102,6 +108,50 @@ int inet_peer_threshold __read_mostly = 65536 + 128; /* start to throw entries m int inet_peer_minttl __read_mostly = 120 * HZ; /* TTL under high load: 120 sec */ int inet_peer_maxttl __read_mostly = 10 * 60 * HZ; /* usual time to live: 10 min */ +static void inetpeer_gc_worker(struct work_struct *work) +{ + struct inet_peer *p, *n; + LIST_HEAD(list); + + spin_lock_bh(&gc_lock); + list_replace_init(&gc_list, &list); + spin_unlock_bh(&gc_lock); + + if (list_empty(&list)) + return; + + list_for_each_entry_safe(p, n, &list, gc_list) { + + if(need_resched()) + cond_resched(); + + if (p->avl_left != peer_avl_empty) { + list_add_tail(&p->avl_left->gc_list, &list); + p->avl_left = peer_avl_empty; + } + + if (p->avl_right != peer_avl_empty) { + list_add_tail(&p->avl_right->gc_list, &list); + p->avl_right = peer_avl_empty; + } + + n = list_entry(p->gc_list.next, struct inet_peer, gc_list); + + if (!atomic_read(&p->refcnt)) { + list_del(&p->gc_list); + kmem_cache_free(peer_cachep, p); + } + } + + if (list_empty(&list)) + return; + + spin_lock_bh(&gc_lock); + list_splice(&list, &gc_list); + spin_unlock_bh(&gc_lock); + + schedule_delayed_work(&gc_work, gc_delay); +} /* Called from ip_output.c:ip_init */ void __init inet_initpeers(void) @@ -126,6 +176,7 @@ void __init inet_initpeers(void) 0, SLAB_HWCACHE_ALIGN | SLAB_PANIC, NULL); + INIT_DELAYED_WORK_DEFERRABLE(&gc_work, inetpeer_gc_worker); } static int addr_compare(const struct inetpeer_addr *a, @@ -449,7 +500,7 @@ relookup: p->pmtu_orig = 0; p->redirect_genid = 0; memset(&p->redirect_learned, 0, sizeof(p->redirect_learned)); - + INIT_LIST_HEAD(&p->gc_list); /* Link the node. */ link_to_pool(p, base); @@ -509,3 +560,30 @@ bool inet_peer_xrlim_allow(struct inet_peer *peer, int timeout) return rc; } EXPORT_SYMBOL(inet_peer_xrlim_allow); + +void inetpeer_invalidate_tree(int family) +{ + struct inet_peer *old, *new, *prev; + struct inet_peer_base *base = family_to_base(family); + + write_seqlock_bh(&base->lock); + + old = base->root; + if (old == peer_avl_empty_rcu) + goto out; + + new = peer_avl_empty_rcu; + + prev = cmpxchg(&base->root, old, new); + if (prev == old) { + base->total = 0; + spin_lock(&gc_lock); + list_add_tail(&prev->gc_list, &gc_list); + spin_unlock(&gc_lock); + schedule_delayed_work(&gc_work, gc_delay); + } + +out: + write_sequnlock_bh(&base->lock); +} +EXPORT_SYMBOL(inetpeer_invalidate_tree); diff --git a/net/ipv4/route.c b/net/ipv4/route.c index bcacf54e5418..23ce0c1287ab 100644 --- a/net/ipv4/route.c +++ b/net/ipv4/route.c @@ -938,6 +938,7 @@ static void rt_cache_invalidate(struct net *net) get_random_bytes(&shuffle, sizeof(shuffle)); atomic_add(shuffle + 1U, &net->ipv4.rt_genid); redirect_genid++; + inetpeer_invalidate_tree(AF_INET); } /* -- cgit v1.2.3 From ac3f48de09d8f4b73397047e413fadff7f65cfa7 Mon Sep 17 00:00:00 2001 From: Steffen Klassert Date: Tue, 6 Mar 2012 21:21:10 +0000 Subject: route: Remove redirect_genid As we invalidate the inetpeer tree along with the routing cache now, we don't need a genid to reset the redirect handling when the routing cache is flushed. Signed-off-by: Steffen Klassert Signed-off-by: David S. Miller --- net/ipv4/inetpeer.c | 1 - net/ipv4/route.c | 11 ++--------- 2 files changed, 2 insertions(+), 10 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/inetpeer.c b/net/ipv4/inetpeer.c index deea2e96b7f2..d4d61b694fab 100644 --- a/net/ipv4/inetpeer.c +++ b/net/ipv4/inetpeer.c @@ -498,7 +498,6 @@ relookup: p->rate_last = 0; p->pmtu_expires = 0; p->pmtu_orig = 0; - p->redirect_genid = 0; memset(&p->redirect_learned, 0, sizeof(p->redirect_learned)); INIT_LIST_HEAD(&p->gc_list); diff --git a/net/ipv4/route.c b/net/ipv4/route.c index 23ce0c1287ab..019774796174 100644 --- a/net/ipv4/route.c +++ b/net/ipv4/route.c @@ -132,7 +132,6 @@ static int ip_rt_mtu_expires __read_mostly = 10 * 60 * HZ; static int ip_rt_min_pmtu __read_mostly = 512 + 20 + 20; static int ip_rt_min_advmss __read_mostly = 256; static int rt_chain_length_max __read_mostly = 20; -static int redirect_genid; static struct delayed_work expires_work; static unsigned long expires_ljiffies; @@ -937,7 +936,6 @@ static void rt_cache_invalidate(struct net *net) get_random_bytes(&shuffle, sizeof(shuffle)); atomic_add(shuffle + 1U, &net->ipv4.rt_genid); - redirect_genid++; inetpeer_invalidate_tree(AF_INET); } @@ -1486,10 +1484,8 @@ void ip_rt_redirect(__be32 old_gw, __be32 daddr, __be32 new_gw, peer = rt->peer; if (peer) { - if (peer->redirect_learned.a4 != new_gw || - peer->redirect_genid != redirect_genid) { + if (peer->redirect_learned.a4 != new_gw) { peer->redirect_learned.a4 = new_gw; - peer->redirect_genid = redirect_genid; atomic_inc(&__rt_peer_genid); } check_peer_redir(&rt->dst, peer); @@ -1794,8 +1790,6 @@ static void ipv4_validate_peer(struct rtable *rt) if (peer) { check_peer_pmtu(&rt->dst, peer); - if (peer->redirect_genid != redirect_genid) - peer->redirect_learned.a4 = 0; if (peer->redirect_learned.a4 && peer->redirect_learned.a4 != rt->rt_gateway) check_peer_redir(&rt->dst, peer); @@ -1959,8 +1953,7 @@ static void rt_init_metrics(struct rtable *rt, const struct flowi4 *fl4, dst_init_metrics(&rt->dst, peer->metrics, false); check_peer_pmtu(&rt->dst, peer); - if (peer->redirect_genid != redirect_genid) - peer->redirect_learned.a4 = 0; + if (peer->redirect_learned.a4 && peer->redirect_learned.a4 != rt->rt_gateway) { rt->rt_gateway = peer->redirect_learned.a4; -- cgit v1.2.3