From 905a6f96a1b18e490a75f810d733ced93c39b0e5 Mon Sep 17 00:00:00 2001 From: Hannes Frederic Sowa Date: Mon, 22 Jul 2013 23:45:53 +0200 Subject: ipv6: take rtnl_lock and mark mrt6 table as freed on namespace cleanup Otherwise we end up dereferencing the already freed net->ipv6.mrt pointer which leads to a panic (from Srivatsa S. Bhat): BUG: unable to handle kernel paging request at ffff882018552020 IP: [] ip6mr_sk_done+0x32/0xb0 [ipv6] PGD 290a067 PUD 207ffe0067 PMD 207ff1d067 PTE 8000002018552060 Oops: 0000 [#1] SMP DEBUG_PAGEALLOC Modules linked in: ebtable_nat ebtables nfs fscache nf_conntrack_ipv4 nf_defrag_ipv4 ipt_REJECT xt_CHECKSUM iptable_mangle iptable_filter ip_tables nfsd lockd nfs_acl exportfs auth_rpcgss autofs4 sunrpc 8021q garp bridge stp llc ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter +ip6_tables ipv6 vfat fat vhost_net macvtap macvlan vhost tun kvm_intel kvm uinput iTCO_wdt iTCO_vendor_support cdc_ether usbnet mii microcode i2c_i801 i2c_core lpc_ich mfd_core shpchp ioatdma dca mlx4_core be2net wmi acpi_cpufreq mperf ext4 jbd2 mbcache dm_mirror dm_region_hash dm_log dm_mod CPU: 0 PID: 7 Comm: kworker/u33:0 Not tainted 3.11.0-rc1-ea45e-a #4 Hardware name: IBM -[8737R2A]-/00Y2738, BIOS -[B2E120RUS-1.20]- 11/30/2012 Workqueue: netns cleanup_net task: ffff8810393641c0 ti: ffff881039366000 task.ti: ffff881039366000 RIP: 0010:[] [] ip6mr_sk_done+0x32/0xb0 [ipv6] RSP: 0018:ffff881039367bd8 EFLAGS: 00010286 RAX: ffff881039367fd8 RBX: ffff882018552000 RCX: dead000000200200 RDX: 0000000000000000 RSI: ffff881039367b68 RDI: ffff881039367b68 RBP: ffff881039367bf8 R08: ffff881039367b68 R09: 2222222222222222 R10: 2222222222222222 R11: 2222222222222222 R12: ffff882015a7a040 R13: ffff882014eb89c0 R14: ffff8820289e2800 R15: 0000000000000000 FS: 0000000000000000(0000) GS:ffff88103fc00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffff882018552020 CR3: 0000000001c0b000 CR4: 00000000000407f0 Stack: ffff881039367c18 ffff882014eb89c0 ffff882015e28c00 0000000000000000 ffff881039367c18 ffffffffa034d9d1 ffff8820289e2800 ffff882014eb89c0 ffff881039367c58 ffffffff815bdecb ffffffff815bddf2 ffff882014eb89c0 Call Trace: [] rawv6_close+0x21/0x40 [ipv6] [] inet_release+0xfb/0x220 [] ? inet_release+0x22/0x220 [] inet6_release+0x3f/0x50 [ipv6] [] sock_release+0x29/0xa0 [] sk_release_kernel+0x30/0x70 [] icmpv6_sk_exit+0x3b/0x80 [ipv6] [] ops_exit_list+0x39/0x60 [] cleanup_net+0xfb/0x1a0 [] process_one_work+0x1da/0x610 [] ? process_one_work+0x169/0x610 [] worker_thread+0x120/0x3a0 [] ? process_one_work+0x610/0x610 [] kthread+0xee/0x100 [] ? __init_kthread_worker+0x70/0x70 [] ret_from_fork+0x7c/0xb0 [] ? __init_kthread_worker+0x70/0x70 Code: 20 48 89 5d e8 4c 89 65 f0 4c 89 6d f8 66 66 66 66 90 4c 8b 67 30 49 89 fd e8 db 3c 1e e1 49 8b 9c 24 90 08 00 00 48 85 db 74 06 <4c> 39 6b 20 74 20 bb f3 ff ff ff e8 8e 3c 1e e1 89 d8 4c 8b 65 RIP [] ip6mr_sk_done+0x32/0xb0 [ipv6] RSP CR2: ffff882018552020 Reported-by: Srivatsa S. Bhat Tested-by: Srivatsa S. Bhat Signed-off-by: Hannes Frederic Sowa Signed-off-by: David S. Miller --- net/ipv6/ip6mr.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'net/ipv6') diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c index 583e8d435f9a..03986d31fa41 100644 --- a/net/ipv6/ip6mr.c +++ b/net/ipv6/ip6mr.c @@ -259,10 +259,12 @@ static void __net_exit ip6mr_rules_exit(struct net *net) { struct mr6_table *mrt, *next; + rtnl_lock(); list_for_each_entry_safe(mrt, next, &net->ipv6.mr6_tables, list) { list_del(&mrt->list); ip6mr_free_table(mrt); } + rtnl_unlock(); fib_rules_unregister(net->ipv6.mr6_rules_ops); } #else @@ -289,7 +291,10 @@ static int __net_init ip6mr_rules_init(struct net *net) static void __net_exit ip6mr_rules_exit(struct net *net) { + rtnl_lock(); ip6mr_free_table(net->ipv6.mrt6); + net->ipv6.mrt6 = NULL; + rtnl_unlock(); } #endif -- cgit v1.2.1 From 2ac3ac8f86f2fe065d746d9a9abaca867adec577 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Michal=20Kube=C4=8Dek?= Date: Thu, 1 Aug 2013 10:04:14 +0200 Subject: ipv6: prevent fib6_run_gc() contention On a high-traffic router with many processors and many IPv6 dst entries, soft lockup in fib6_run_gc() can occur when number of entries reaches gc_thresh. This happens because fib6_run_gc() uses fib6_gc_lock to allow only one thread to run the garbage collector but ip6_dst_gc() doesn't update net->ipv6.ip6_rt_last_gc until fib6_run_gc() returns. On a system with many entries, this can take some time so that in the meantime, other threads pass the tests in ip6_dst_gc() (ip6_rt_last_gc is still not updated) and wait for the lock. They then have to run the garbage collector one after another which blocks them for quite long. Resolve this by replacing special value ~0UL of expire parameter to fib6_run_gc() by explicit "force" parameter to choose between spin_lock_bh() and spin_trylock_bh() and call fib6_run_gc() with force=false if gc_thresh is reached but not max_size. Signed-off-by: Michal Kubecek Signed-off-by: David S. Miller --- net/ipv6/ip6_fib.c | 19 ++++++++----------- net/ipv6/ndisc.c | 4 ++-- net/ipv6/route.c | 4 ++-- 3 files changed, 12 insertions(+), 15 deletions(-) (limited to 'net/ipv6') diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c index 5fc9c7a68d8d..d872553ca933 100644 --- a/net/ipv6/ip6_fib.c +++ b/net/ipv6/ip6_fib.c @@ -1632,19 +1632,16 @@ static int fib6_age(struct rt6_info *rt, void *arg) static DEFINE_SPINLOCK(fib6_gc_lock); -void fib6_run_gc(unsigned long expires, struct net *net) +void fib6_run_gc(unsigned long expires, struct net *net, bool force) { - if (expires != ~0UL) { + if (force) { spin_lock_bh(&fib6_gc_lock); - gc_args.timeout = expires ? (int)expires : - net->ipv6.sysctl.ip6_rt_gc_interval; - } else { - if (!spin_trylock_bh(&fib6_gc_lock)) { - mod_timer(&net->ipv6.ip6_fib_timer, jiffies + HZ); - return; - } - gc_args.timeout = net->ipv6.sysctl.ip6_rt_gc_interval; + } else if (!spin_trylock_bh(&fib6_gc_lock)) { + mod_timer(&net->ipv6.ip6_fib_timer, jiffies + HZ); + return; } + gc_args.timeout = expires ? (int)expires : + net->ipv6.sysctl.ip6_rt_gc_interval; gc_args.more = icmp6_dst_gc(); @@ -1661,7 +1658,7 @@ void fib6_run_gc(unsigned long expires, struct net *net) static void fib6_gc_timer_cb(unsigned long arg) { - fib6_run_gc(0, (struct net *)arg); + fib6_run_gc(0, (struct net *)arg, true); } static int __net_init fib6_net_init(struct net *net) diff --git a/net/ipv6/ndisc.c b/net/ipv6/ndisc.c index 24c03396e008..79aa9652ed86 100644 --- a/net/ipv6/ndisc.c +++ b/net/ipv6/ndisc.c @@ -1576,7 +1576,7 @@ static int ndisc_netdev_event(struct notifier_block *this, unsigned long event, switch (event) { case NETDEV_CHANGEADDR: neigh_changeaddr(&nd_tbl, dev); - fib6_run_gc(~0UL, net); + fib6_run_gc(0, net, false); idev = in6_dev_get(dev); if (!idev) break; @@ -1586,7 +1586,7 @@ static int ndisc_netdev_event(struct notifier_block *this, unsigned long event, break; case NETDEV_DOWN: neigh_ifdown(&nd_tbl, dev); - fib6_run_gc(~0UL, net); + fib6_run_gc(0, net, false); break; case NETDEV_NOTIFY_PEERS: ndisc_send_unsol_na(dev); diff --git a/net/ipv6/route.c b/net/ipv6/route.c index a8c891aa2464..824c424f9648 100644 --- a/net/ipv6/route.c +++ b/net/ipv6/route.c @@ -1326,7 +1326,7 @@ static int ip6_dst_gc(struct dst_ops *ops) goto out; net->ipv6.ip6_rt_gc_expire++; - fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net); + fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, entries > rt_max_size); net->ipv6.ip6_rt_last_gc = now; entries = dst_entries_get_slow(ops); if (entries < ops->gc_thresh) @@ -2827,7 +2827,7 @@ int ipv6_sysctl_rtcache_flush(struct ctl_table *ctl, int write, net = (struct net *)ctl->extra1; delay = net->ipv6.sysctl.flush_delay; proc_dointvec(ctl, write, buffer, lenp, ppos); - fib6_run_gc(delay <= 0 ? ~0UL : (unsigned long)delay, net); + fib6_run_gc(delay <= 0 ? 0 : (unsigned long)delay, net, delay > 0); return 0; } -- cgit v1.2.1 From 49a18d86f66d33a20144ecb5a34bba0d1856b260 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Michal=20Kube=C4=8Dek?= Date: Thu, 1 Aug 2013 10:04:24 +0200 Subject: ipv6: update ip6_rt_last_gc every time GC is run As pointed out by Eric Dumazet, net->ipv6.ip6_rt_last_gc should hold the last time garbage collector was run so that we should update it whenever fib6_run_gc() calls fib6_clean_all(), not only if we got there from ip6_dst_gc(). Signed-off-by: Michal Kubecek Signed-off-by: David S. Miller --- net/ipv6/ip6_fib.c | 6 +++++- net/ipv6/route.c | 4 +--- 2 files changed, 6 insertions(+), 4 deletions(-) (limited to 'net/ipv6') diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c index d872553ca933..bff3d821c7eb 100644 --- a/net/ipv6/ip6_fib.c +++ b/net/ipv6/ip6_fib.c @@ -1634,6 +1634,8 @@ static DEFINE_SPINLOCK(fib6_gc_lock); void fib6_run_gc(unsigned long expires, struct net *net, bool force) { + unsigned long now; + if (force) { spin_lock_bh(&fib6_gc_lock); } else if (!spin_trylock_bh(&fib6_gc_lock)) { @@ -1646,10 +1648,12 @@ void fib6_run_gc(unsigned long expires, struct net *net, bool force) gc_args.more = icmp6_dst_gc(); fib6_clean_all(net, fib6_age, 0, NULL); + now = jiffies; + net->ipv6.ip6_rt_last_gc = now; if (gc_args.more) mod_timer(&net->ipv6.ip6_fib_timer, - round_jiffies(jiffies + round_jiffies(now + net->ipv6.sysctl.ip6_rt_gc_interval)); else del_timer(&net->ipv6.ip6_fib_timer); diff --git a/net/ipv6/route.c b/net/ipv6/route.c index 824c424f9648..b70f8979003b 100644 --- a/net/ipv6/route.c +++ b/net/ipv6/route.c @@ -1311,7 +1311,6 @@ static void icmp6_clean_all(int (*func)(struct rt6_info *rt, void *arg), static int ip6_dst_gc(struct dst_ops *ops) { - unsigned long now = jiffies; struct net *net = container_of(ops, struct net, ipv6.ip6_dst_ops); int rt_min_interval = net->ipv6.sysctl.ip6_rt_gc_min_interval; int rt_max_size = net->ipv6.sysctl.ip6_rt_max_size; @@ -1321,13 +1320,12 @@ static int ip6_dst_gc(struct dst_ops *ops) int entries; entries = dst_entries_get_fast(ops); - if (time_after(rt_last_gc + rt_min_interval, now) && + if (time_after(rt_last_gc + rt_min_interval, jiffies) && entries <= rt_max_size) goto out; net->ipv6.ip6_rt_gc_expire++; fib6_run_gc(net->ipv6.ip6_rt_gc_expire, net, entries > rt_max_size); - net->ipv6.ip6_rt_last_gc = now; entries = dst_entries_get_slow(ops); if (entries < ops->gc_thresh) net->ipv6.ip6_rt_gc_expire = rt_gc_timeout>>1; -- cgit v1.2.1 From 3f8f52982ad020f0704548c46de66bf464d3b967 Mon Sep 17 00:00:00 2001 From: Jiri Pirko Date: Thu, 1 Aug 2013 10:41:27 +0200 Subject: ipv6: move peer_addr init into ipv6_add_addr() Signed-off-by: Jiri Pirko Signed-off-by: David S. Miller --- net/ipv6/addrconf.c | 20 +++++++++++--------- 1 file changed, 11 insertions(+), 9 deletions(-) (limited to 'net/ipv6') diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c index cfdcf7b2daf6..a0ce957fb671 100644 --- a/net/ipv6/addrconf.c +++ b/net/ipv6/addrconf.c @@ -813,7 +813,8 @@ static u32 inet6_addr_hash(const struct in6_addr *addr) /* On success it returns ifp with increased reference count */ static struct inet6_ifaddr * -ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, int pfxlen, +ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, + const struct in6_addr *peer_addr, int pfxlen, int scope, u32 flags) { struct inet6_ifaddr *ifa = NULL; @@ -863,6 +864,8 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, int pfxlen, } ifa->addr = *addr; + if (peer_addr) + ifa->peer_addr = *peer_addr; spin_lock_init(&ifa->lock); spin_lock_init(&ifa->state_lock); @@ -1123,8 +1126,8 @@ retry: ift = !max_addresses || ipv6_count_addresses(idev) < max_addresses ? - ipv6_add_addr(idev, &addr, tmp_plen, ipv6_addr_scope(&addr), - addr_flags) : NULL; + ipv6_add_addr(idev, &addr, NULL, tmp_plen, + ipv6_addr_scope(&addr), addr_flags) : NULL; if (IS_ERR_OR_NULL(ift)) { in6_ifa_put(ifp); in6_dev_put(idev); @@ -2179,7 +2182,8 @@ ok: */ if (!max_addresses || ipv6_count_addresses(in6_dev) < max_addresses) - ifp = ipv6_add_addr(in6_dev, &addr, pinfo->prefix_len, + ifp = ipv6_add_addr(in6_dev, &addr, NULL, + pinfo->prefix_len, addr_type&IPV6_ADDR_SCOPE_MASK, addr_flags); @@ -2455,15 +2459,13 @@ static int inet6_addr_add(struct net *net, int ifindex, const struct in6_addr *p prefered_lft = timeout; } - ifp = ipv6_add_addr(idev, pfx, plen, scope, ifa_flags); + ifp = ipv6_add_addr(idev, pfx, peer_pfx, plen, scope, ifa_flags); if (!IS_ERR(ifp)) { spin_lock_bh(&ifp->lock); ifp->valid_lft = valid_lft; ifp->prefered_lft = prefered_lft; ifp->tstamp = jiffies; - if (peer_pfx) - ifp->peer_addr = *peer_pfx; spin_unlock_bh(&ifp->lock); addrconf_prefix_route(&ifp->addr, ifp->prefix_len, dev, @@ -2557,7 +2559,7 @@ static void add_addr(struct inet6_dev *idev, const struct in6_addr *addr, { struct inet6_ifaddr *ifp; - ifp = ipv6_add_addr(idev, addr, plen, scope, IFA_F_PERMANENT); + ifp = ipv6_add_addr(idev, addr, NULL, plen, scope, IFA_F_PERMANENT); if (!IS_ERR(ifp)) { spin_lock_bh(&ifp->lock); ifp->flags &= ~IFA_F_TENTATIVE; @@ -2683,7 +2685,7 @@ static void addrconf_add_linklocal(struct inet6_dev *idev, const struct in6_addr #endif - ifp = ipv6_add_addr(idev, addr, 64, IFA_LINK, addr_flags); + ifp = ipv6_add_addr(idev, addr, NULL, 64, IFA_LINK, addr_flags); if (!IS_ERR(ifp)) { addrconf_prefix_route(&ifp->addr, ifp->prefix_len, idev->dev, 0, 0); addrconf_dad_start(ifp); -- cgit v1.2.1 From 8a226b2cfa776db6011fc84b71578513161cd3d3 Mon Sep 17 00:00:00 2001 From: Jiri Benc Date: Thu, 1 Aug 2013 10:41:28 +0200 Subject: ipv6: prevent race between address creation and removal There's a race in IPv6 automatic addess assignment. The address is created with zero lifetime when it's added to various address lists. Before it gets assigned the correct lifetime, there's a window where a new address may be configured. This causes the semi-initiated address to be deleted in addrconf_verify. This was discovered as a reference leak caused by concurrent run of __ipv6_ifa_notify for both RTM_NEWADDR and RTM_DELADDR with the same address. Fix this by setting the lifetime before the address is added to inet6_addr_lst. A few notes: 1. In addrconf_prefix_rcv, by setting update_lft to zero, the if (update_lft) { ... } condition is no longer executed for newly created addresses. This is okay, as the ifp fields are set in ipv6_add_addr now and ipv6_ifa_notify is called (and has been called) through addrconf_dad_start. 2. The removal of the whole block under ifp->lock in inet6_addr_add is okay, too, as tstamp is initialized to jiffies in ipv6_add_addr. Signed-off-by: Jiri Benc Signed-off-by: Jiri Pirko Signed-off-by: David S. Miller --- net/ipv6/addrconf.c | 31 +++++++++++++++---------------- 1 file changed, 15 insertions(+), 16 deletions(-) (limited to 'net/ipv6') diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c index a0ce957fb671..da4241c8c7da 100644 --- a/net/ipv6/addrconf.c +++ b/net/ipv6/addrconf.c @@ -815,7 +815,7 @@ static u32 inet6_addr_hash(const struct in6_addr *addr) static struct inet6_ifaddr * ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, const struct in6_addr *peer_addr, int pfxlen, - int scope, u32 flags) + int scope, u32 flags, u32 valid_lft, u32 prefered_lft) { struct inet6_ifaddr *ifa = NULL; struct rt6_info *rt; @@ -875,6 +875,8 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, ifa->scope = scope; ifa->prefix_len = pfxlen; ifa->flags = flags | IFA_F_TENTATIVE; + ifa->valid_lft = valid_lft; + ifa->prefered_lft = prefered_lft; ifa->cstamp = ifa->tstamp = jiffies; ifa->tokenized = false; @@ -1127,7 +1129,8 @@ retry: ift = !max_addresses || ipv6_count_addresses(idev) < max_addresses ? ipv6_add_addr(idev, &addr, NULL, tmp_plen, - ipv6_addr_scope(&addr), addr_flags) : NULL; + ipv6_addr_scope(&addr), addr_flags, + tmp_valid_lft, tmp_prefered_lft) : NULL; if (IS_ERR_OR_NULL(ift)) { in6_ifa_put(ifp); in6_dev_put(idev); @@ -1139,8 +1142,6 @@ retry: spin_lock_bh(&ift->lock); ift->ifpub = ifp; - ift->valid_lft = tmp_valid_lft; - ift->prefered_lft = tmp_prefered_lft; ift->cstamp = now; ift->tstamp = tmp_tstamp; spin_unlock_bh(&ift->lock); @@ -2185,14 +2186,16 @@ ok: ifp = ipv6_add_addr(in6_dev, &addr, NULL, pinfo->prefix_len, addr_type&IPV6_ADDR_SCOPE_MASK, - addr_flags); + addr_flags, valid_lft, + prefered_lft); if (IS_ERR_OR_NULL(ifp)) { in6_dev_put(in6_dev); return; } - update_lft = create = 1; + update_lft = 0; + create = 1; ifp->cstamp = jiffies; ifp->tokenized = tokenized; addrconf_dad_start(ifp); @@ -2213,7 +2216,7 @@ ok: stored_lft = ifp->valid_lft - (now - ifp->tstamp) / HZ; else stored_lft = 0; - if (!update_lft && stored_lft) { + if (!update_lft && !create && stored_lft) { if (valid_lft > MIN_VALID_LIFETIME || valid_lft > stored_lft) update_lft = 1; @@ -2459,15 +2462,10 @@ static int inet6_addr_add(struct net *net, int ifindex, const struct in6_addr *p prefered_lft = timeout; } - ifp = ipv6_add_addr(idev, pfx, peer_pfx, plen, scope, ifa_flags); + ifp = ipv6_add_addr(idev, pfx, peer_pfx, plen, scope, ifa_flags, + valid_lft, prefered_lft); if (!IS_ERR(ifp)) { - spin_lock_bh(&ifp->lock); - ifp->valid_lft = valid_lft; - ifp->prefered_lft = prefered_lft; - ifp->tstamp = jiffies; - spin_unlock_bh(&ifp->lock); - addrconf_prefix_route(&ifp->addr, ifp->prefix_len, dev, expires, flags); /* @@ -2559,7 +2557,8 @@ static void add_addr(struct inet6_dev *idev, const struct in6_addr *addr, { struct inet6_ifaddr *ifp; - ifp = ipv6_add_addr(idev, addr, NULL, plen, scope, IFA_F_PERMANENT); + ifp = ipv6_add_addr(idev, addr, NULL, plen, + scope, IFA_F_PERMANENT, 0, 0); if (!IS_ERR(ifp)) { spin_lock_bh(&ifp->lock); ifp->flags &= ~IFA_F_TENTATIVE; @@ -2685,7 +2684,7 @@ static void addrconf_add_linklocal(struct inet6_dev *idev, const struct in6_addr #endif - ifp = ipv6_add_addr(idev, addr, NULL, 64, IFA_LINK, addr_flags); + ifp = ipv6_add_addr(idev, addr, NULL, 64, IFA_LINK, addr_flags, 0, 0); if (!IS_ERR(ifp)) { addrconf_prefix_route(&ifp->addr, ifp->prefix_len, idev->dev, 0, 0); addrconf_dad_start(ifp); -- cgit v1.2.1