ipv4: fix data races in fib_alias_hw_flags_set
Subsystems (4)
PATCH 1/2 ipv4: fix data races in fib_alias_hw_flags_set
From: Eric Dumazet <[email protected]>
fib_alias_hw_flags_set() can be used by concurrent threads, and is only RCU protected.
We need to annotate accesses to following fields of struct fib_alias:
offload, trap, offload_failed
Because of READ_ONCE()WRITE_ONCE() limitations, make these field u8.
BUG: KCSAN: data-race in fib_alias_hw_flags_set / fib_alias_hw_flags_set
read to 0xffff888134224a6a of 1 bytes by task 2013 on cpu 1:
fib_alias_hw_flags_set+0x28a/0x470 net/ipv4/fib_trie.c:1050 nsim_fib4_rt_hw_flags_set drivers/net/netdevsim/fib.c:350 [inline] nsim_fib4_rt_add drivers/net/netdevsim/fib.c:367 [inline] nsim_fib4_rt_insert drivers/net/netdevsim/fib.c:429 [inline] nsim_fib4_event drivers/net/netdevsim/fib.c:461 [inline] nsim_fib_event drivers/net/netdevsim/fib.c:881 [inline] nsim_fib_event_work+0x1852/0x2cf0 drivers/net/netdevsim/fib.c:1477 process_one_work+0x3f6/0x960 kernel/workqueue.c:2307 process_scheduled_works kernel/workqueue.c:2370 [inline] worker_thread+0x7df/0xa70 kernel/workqueue.c:2456 kthread+0x1bf/0x1e0 kernel/kthread.c:377 ret_from_fork+0x1f/0x30
write to 0xffff888134224a6a of 1 bytes by task 4872 on cpu 0:
fib_alias_hw_flags_set+0x2d5/0x470 net/ipv4/fib_trie.c:1054 nsim_fib4_rt_hw_flags_set drivers/net/netdevsim/fib.c:350 [inline] nsim_fib4_rt_add drivers/net/netdevsim/fib.c:367 [inline] nsim_fib4_rt_insert drivers/net/netdevsim/fib.c:429 [inline] nsim_fib4_event drivers/net/netdevsim/fib.c:461 [inline] nsim_fib_event drivers/net/netdevsim/fib.c:881 [inline] nsim_fib_event_work+0x1852/0x2cf0 drivers/net/netdevsim/fib.c:1477 process_one_work+0x3f6/0x960 kernel/workqueue.c:2307 process_scheduled_works kernel/workqueue.c:2370 [inline] worker_thread+0x7df/0xa70 kernel/workqueue.c:2456 kthread+0x1bf/0x1e0 kernel/kthread.c:377 ret_from_fork+0x1f/0x30
value changed: 0x00 -> 0x02
Reported by Kernel Concurrency Sanitizer on:
CPU: 0 PID: 4872 Comm: kworker/0:0 Not tainted 5.17.0-rc3-syzkaller-00188-g1d41d2e82623-dirty #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Workqueue: events nsim_fib_event_work
--- net/ipv4/fib_lookup.h | 7 +++---- net/ipv4/fib_semantics.c | 6 +++--- net/ipv4/fib_trie.c | 22 +++++++++++++--------- net/ipv4/route.c | 4 ++-- 4 files changed, 21 insertions(+), 18 deletions(-)
net/ipv4/fib_lookup.h +3 -4
@@ -16,10 +16,9 @@ struct fib_alias {u8 fa_slen;u32 tb_id;s16 fa_default; - u8 offload:1, - trap:1, - offload_failed:1, - unused:5; +u8 offload; +u8 trap; +u8 offload_failed;struct rcu_head rcu; };
net/ipv4/fib_semantics.c +3 -3
@@ -525,9 +525,9 @@ void rtmsg_fib(int event, __be32 key, struct fib_alias *fa, fri.dst_len = dst_len; fri.tos = fa->fa_tos; fri.type = fa->fa_type; - fri.offload = fa->offload; - fri.trap = fa->trap; - fri.offload_failed = fa->offload_failed; + fri.offload =READ_ONCE (fa->offload); + fri.trap =READ_ONCE (fa->trap); + fri.offload_failed =READ_ONCE (fa->offload_failed); err =fib_dump_info (skb, info->portid, seq, event, &fri, nlm_flags);if (err <0 ) {/* -EMSGSIZE implies BUG in fib_nlmsg_size() */
net/ipv4/fib_trie.c +13 -9
@@ -1047,19 +1047,23 @@ void fib_alias_hw_flags_set(struct net *net, const struct fib_rt_info *fri)if (!fa_match) goto out; -if (fa_match->offload == fri->offload && fa_match->trap == fri->trap && - fa_match->offload_failed == fri->offload_failed) +/* These are paired with the WRITE_ONCE() happening in this function. +* The reason is that we are only protected by RCU at this point. +*/ +if (READ_ONCE (fa_match->offload) == fri->offload && +READ_ONCE (fa_match->trap) == fri->trap && +READ_ONCE (fa_match->offload_failed) == fri->offload_failed) goto out; - fa_match->offload = fri->offload; - fa_match->trap = fri->trap; +WRITE_ONCE (fa_match->offload, fri->offload); +WRITE_ONCE (fa_match->trap, fri->trap);/* 2 means send notifications only if offload_failed was changed. */ if (net->ipv4.sysctl_fib_notify_on_flag_change ==2 && - fa_match->offload_failed == fri->offload_failed) +READ_ONCE (fa_match->offload_failed) == fri->offload_failed) goto out; - fa_match->offload_failed = fri->offload_failed; +WRITE_ONCE (fa_match->offload_failed, fri->offload_failed);if (!net->ipv4.sysctl_fib_notify_on_flag_change) goto out;
@@ -2297,9 +2301,9 @@ static int fn_trie_dump_leaf(struct key_vector *l, struct fib_table *tb, fri.dst_len =KEYLENGTH - fa->fa_slen; fri.tos = fa->fa_tos; fri.type = fa->fa_type; - fri.offload = fa->offload; - fri.trap = fa->trap; - fri.offload_failed = fa->offload_failed; + fri.offload =READ_ONCE (fa->offload); + fri.trap =READ_ONCE (fa->trap); + fri.offload_failed =READ_ONCE (fa->offload_failed); err = fib_dump_info(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
net/ipv4/route.c +2 -2
@@ -3395,8 +3395,8 @@ static int inet_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh, fa->fa_tos == fri.tos && fa->fa_info == res.fi && fa->fa_type == fri.type) { - fri.offload = fa->offload; - fri.trap = fa->trap; + fri.offload =READ_ONCE (fa->offload); + fri.trap =READ_ONCE (fa->trap);break ; } }
Signature · 1 line
2.35.1.265.g69c8d7142f-goog
PATCH 2/2 ipv6: fix data-race in fib6_info_hw_flags_set / fib6_purge_rt
From: Eric Dumazet <[email protected]>
Because fib6_info_hw_flags_set() is called without any synchronization, all accesses to gi6->offload, fi->trap and fi->offload_failed need some basic protection like READ_ONCE()/WRITE_ONCE().
BUG: KCSAN: data-race in fib6_info_hw_flags_set / fib6_purge_rt
read to 0xffff8881087d5886 of 1 bytes by task 13953 on cpu 0:
fib6_drop_pcpu_from net/ipv6/ip6_fib.c:1007 [inline] fib6_purge_rt+0x4f/0x580 net/ipv6/ip6_fib.c:1033 fib6_del_route net/ipv6/ip6_fib.c:1983 [inline] fib6_del+0x696/0x890 net/ipv6/ip6_fib.c:2028 __ip6_del_rt net/ipv6/route.c:3876 [inline] ip6_del_rt+0x83/0x140 net/ipv6/route.c:3891 __ipv6_dev_ac_dec+0x2b5/0x370 net/ipv6/anycast.c:374 ipv6_dev_ac_dec net/ipv6/anycast.c:387 [inline] __ipv6_sock_ac_close+0x141/0x200 net/ipv6/anycast.c:207 ipv6_sock_ac_close+0x79/0x90 net/ipv6/anycast.c:220 inet6_release+0x32/0x50 net/ipv6/af_inet6.c:476 __sock_release net/socket.c:650 [inline] sock_close+0x6c/0x150 net/socket.c:1318 __fput+0x295/0x520 fs/file_table.c:280 ____fput+0x11/0x20 fs/file_table.c:313 task_work_run+0x8e/0x110 kernel/task_work.c:164 tracehook_notify_resume include/linux/tracehook.h:189 [inline] exit_to_user_mode_loop kernel/entry/common.c:175 [inline] exit_to_user_mode_prepare+0x160/0x190 kernel/entry/common.c:207 __syscall_exit_to_user_mode_work kernel/entry/common.c:289 [inline] syscall_exit_to_user_mode+0x20/0x40 kernel/entry/common.c:300 do_syscall_64+0x50/0xd0 arch/x86/entry/common.c:86 entry_SYSCALL_64_after_hwframe+0x44/0xae
write to 0xffff8881087d5886 of 1 bytes by task 1912 on cpu 1:
fib6_info_hw_flags_set+0x155/0x3b0 net/ipv6/route.c:6230 nsim_fib6_rt_hw_flags_set drivers/net/netdevsim/fib.c:668 [inline] nsim_fib6_rt_add drivers/net/netdevsim/fib.c:691 [inline] nsim_fib6_rt_insert drivers/net/netdevsim/fib.c:756 [inline] nsim_fib6_event drivers/net/netdevsim/fib.c:853 [inline] nsim_fib_event drivers/net/netdevsim/fib.c:886 [inline] nsim_fib_event_work+0x284f/0x2cf0 drivers/net/netdevsim/fib.c:1477 process_one_work+0x3f6/0x960 kernel/workqueue.c:2307 worker_thread+0x616/0xa70 kernel/workqueue.c:2454 kthread+0x2c7/0x2e0 kernel/kthread.c:327 ret_from_fork+0x1f/0x30
value changed: 0x22 -> 0x2a
Reported by Kernel Concurrency Sanitizer on:
CPU: 1 PID: 1912 Comm: kworker/1:3 Not tainted 5.16.0-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Workqueue: events nsim_fib_event_work
--- drivers/net/netdevsim/fib.c | 4 ++-- include/net/ip6_fib.h | 10 ++++++---- net/ipv6/route.c | 19 ++++++++++--------- 3 files changed, 18 insertions(+), 15 deletions(-)
drivers/net/netdevsim/fib.c +2 -2
@@ -623,14 +623,14 @@ static int nsim_fib6_rt_append(struct nsim_fib_data *data,if (err) goto err_fib6_rt_nh_del; - fib6_event->rt_arr[i]->trap = true; +WRITE_ONCE (fib6_event->rt_arr[i]->trap, true); }return 0 ; err_fib6_rt_nh_del:for (i--; i >=0 ; i--) { - fib6_event->rt_arr[i]->trap = false; +WRITE_ONCE (fib6_event->rt_arr[i]->trap, false);nsim_fib6_rt_nh_del (fib6_rt, fib6_event->rt_arr[i]); }return err;
include/net/ip6_fib.h +6 -4
@@ -190,14 +190,16 @@ struct fib6_info {u32 fib6_metric;u8 fib6_protocol;u8 fib6_type; + +u8 offload; +u8 trap; +u8 offload_failed; + u8 should_flush:1, dst_nocount:1, dst_nopolicy:1, fib6_destroying:1, - offload:1, - trap:1, - offload_failed:1, - unused:1; + unused:4;struct rcu_head rcu;struct nexthop *nh;
net/ipv6/route.c +10 -9
@@ -5753,11 +5753,11 @@ static int rt6_fill_node(struct net *net, struct sk_buff *skb, }if (!dst) { -if (rt->offload) +if (READ_ONCE (rt->offload)) rtm->rtm_flags |=RTM_F_OFFLOAD ; -if (rt->trap) +if (READ_ONCE (rt->trap)) rtm->rtm_flags |=RTM_F_TRAP ; -if (rt->offload_failed) +if (READ_ONCE (rt->offload_failed)) rtm->rtm_flags |=RTM_F_OFFLOAD_FAILED ; }
@@ -6215,19 +6215,20 @@ void fib6_info_hw_flags_set(struct net *net, struct fib6_info *f6i,struct sk_buff *skb;int err; -if (f6i->offload == offload && f6i->trap == trap && - f6i->offload_failed == offload_failed) +if (READ_ONCE (f6i->offload) == offload && +READ_ONCE (f6i->trap) == trap && +READ_ONCE (f6i->offload_failed) == offload_failed)return ; - f6i->offload = offload; - f6i->trap = trap; +WRITE_ONCE (f6i->offload, offload); +WRITE_ONCE (f6i->trap, trap);/* 2 means send notifications only if offload_failed was changed. */ if (net->ipv6.sysctl.fib_notify_on_flag_change ==2 && - f6i->offload_failed == offload_failed) +READ_ONCE (f6i->offload_failed) == offload_failed)return ; - f6i->offload_failed = offload_failed; +WRITE_ONCE (f6i->offload_failed, offload_failed);if (!rcu_access_pointer (f6i->fib6_node))/* The route was removed from the tree, do not send
Signature · 1 line
2.35.1.265.g69c8d7142f-goog
Quoted 11 lines · Eric Dumazet
From: Eric Dumazet <[email protected]>
fib_alias_hw_flags_set()can be used by concurrent threads, and is only RCU protected.We need to annotate accesses to following fields of struct
fib_alias:offload, trap,
offload_failedBecause of READ_ONCE()WRITE_ONCE() limitations, make these field u8.
[...]
Quoted 4 lines
Fixes 90b93f1b31f8 ("ipv4: Add "offload" and "trap" indications to routes")Signed-off-by Eric Dumazet <[email protected]>Cc Ido Schimmel <[email protected]>Reported-by syzbot <[email protected]>
Thanks, this is better than taking RTNL
Quoted 12 lines
--- net/ipv4/fib_lookup.h | 7 +++---- net/ipv4/fib_semantics.c | 6 +++--- net/ipv4/fib_trie.c | 22 +++++++++++++--------- net/ipv4/route.c | 4 ++-- 4 files changed, 21 insertions(+), 18 deletions(-)net/ipv4/fib_lookup.h +0
-0
net/ipv4/fib_lookup.h struct fib_alias { · line 21
Quoted 10 lines
net/ipv4/fib_lookup.h +3
-4@@ -16,10 +16,9 @@ struct fib_alias {u8 fa_slen;u32 tb_id;s16 fa_default; - u8 offload:1, - trap:1, - offload_failed:1, - unused:5; +u8 offload; +u8 trap; +u8 offload_failed;
There is a 5 bytes hole here, so this shouldn't increase the struct and it should still fit in one cache line
Quoted 2 lines
net/ipv4/fib_lookup.h +0
-0struct rcu_head rcu; };
This series was applied to netdev/net.git (master) by Jakub Kicinski <[email protected]>:
Hello:
This series was applied to netdev/net.git (master) by Jakub Kicinski <[email protected]>:
Quoted 8 lines · On Wed, 16 Feb 2022 09:32:16 -0800 you wrote:
From: Eric Dumazet <[email protected]>
fib_alias_hw_flags_set()can be used by concurrent threads, and is only RCU protected.We need to annotate accesses to following fields of struct
fib_alias:[...]
Here is the summary with links:
- [net,1/2] ipv4: fix data races in fib_alias_hw_flags_set
https://git.kernel.org/netdev/net/c/9fcf986cc4bc - [net,2/2] ipv6: fix data-race in fib6_info_hw_flags_set / fib6_purge_rt https://git.kernel.org/netdev/net/c/d95d6320ba7a
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html