[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CANn89i+KL0=p2mchoZCOsZ1YoF9xhoUoubkub6YyLOY2wpSJtg@mail.gmail.com>
Date: Thu, 31 Oct 2024 15:08:20 +0100
From: Eric Dumazet <edumazet@...gle.com>
To: Wang Liang <wangliang74@...wei.com>
Cc: davem@...emloft.net, kuba@...nel.org, pabeni@...hat.com, horms@...nel.org,
dsahern@...nel.org, yuehaibing@...wei.com, zhangchangzhong@...wei.com,
netdev@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH net] net: fix data-races around sk->sk_forward_alloc
On Thu, Oct 31, 2024 at 1:06 PM Wang Liang <wangliang74@...wei.com> wrote:
>
> Syzkaller reported this warning:
Was this a public report ?
> [ 65.568203][ C0] ------------[ cut here ]------------
> [ 65.569339][ C0] WARNING: CPU: 0 PID: 16 at net/ipv4/af_inet.c:156 inet_sock_destruct+0x1c5/0x1e0
> [ 65.575017][ C0] Modules linked in:
> [ 65.575699][ C0] CPU: 0 UID: 0 PID: 16 Comm: ksoftirqd/0 Not tainted 6.12.0-rc5 #26
> [ 65.577086][ C0] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014
> [ 65.577094][ C0] RIP: 0010:inet_sock_destruct+0x1c5/0x1e0
> [ 65.577100][ C0] Code: 24 12 4c 89 e2 5b 48 c7 c7 98 ec bb 82 41 5c e9 d1 18 17 ff 4c 89 e6 5b 48 c7 c7 d0 ec bb 82 41 5c e9 bf 18 17 ff 0f 0b eb 83 <0f> 0b eb 97 0f 0b eb 87 0f 0b e9 68 ff ff ff 66 66 2e 0f 1f 84 00
> [ 65.577107][ C0] RSP: 0018:ffffc9000008bd90 EFLAGS: 00010206
> [ 65.577113][ C0] RAX: 0000000000000300 RBX: ffff88810b172a90 RCX: 0000000000000007
> [ 65.577117][ C0] RDX: 0000000000000002 RSI: 0000000000000300 RDI: ffff88810b172a00
> [ 65.577120][ C0] RBP: ffff88810b172a00 R08: ffff888104273c00 R09: 0000000000100007
> [ 65.577123][ C0] R10: 0000000000020000 R11: 0000000000000006 R12: ffff88810b172a00
> [ 65.577125][ C0] R13: 0000000000000004 R14: 0000000000000000 R15: ffff888237c31f78
> [ 65.577131][ C0] FS: 0000000000000000(0000) GS:ffff888237c00000(0000) knlGS:0000000000000000
> [ 65.592485][ C0] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [ 65.592489][ C0] CR2: 00007ffc63fecac8 CR3: 000000000342e000 CR4: 00000000000006f0
> [ 65.592491][ C0] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> [ 65.592492][ C0] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> [ 65.592495][ C0] Call Trace:
> [ 65.596277][ C0] <TASK>
> [ 65.598171][ C0] ? __warn+0x88/0x130
> [ 65.598874][ C0] ? inet_sock_destruct+0x1c5/0x1e0
> [ 65.598879][ C0] ? report_bug+0x18e/0x1a0
> [ 65.598883][ C0] ? handle_bug+0x53/0x90
> [ 65.598886][ C0] ? exc_invalid_op+0x18/0x70
> [ 65.598888][ C0] ? asm_exc_invalid_op+0x1a/0x20
> [ 65.598893][ C0] ? inet_sock_destruct+0x1c5/0x1e0
> [ 65.598897][ C0] __sk_destruct+0x2a/0x200
> [ 65.604664][ C0] rcu_do_batch+0x1aa/0x530
> [ 65.605450][ C0] ? rcu_do_batch+0x13b/0x530
> [ 65.605456][ C0] rcu_core+0x159/0x2f0
> [ 65.605466][ C0] handle_softirqs+0xd3/0x2b0
> [ 65.607689][ C0] ? __pfx_smpboot_thread_fn+0x10/0x10
> [ 65.607695][ C0] run_ksoftirqd+0x25/0x30
> [ 65.607699][ C0] smpboot_thread_fn+0xdd/0x1d0
> [ 65.610152][ C0] kthread+0xd3/0x100
> [ 65.610158][ C0] ? __pfx_kthread+0x10/0x10
> [ 65.610160][ C0] ret_from_fork+0x34/0x50
> [ 65.610170][ C0] ? __pfx_kthread+0x10/0x10
> [ 65.610172][ C0] ret_from_fork_asm+0x1a/0x30
> [ 65.610181][ C0] </TASK>
> [ 65.610182][ C0] ---[ end trace 0000000000000000 ]---
>
> Its possible that two threads call tcp_v6_do_rcv()/sk_forward_alloc_add()
> concurrently when sk->sk_state == TCP_LISTEN with sk->sk_lock unlocked,
> which triggers a data-race around sk->sk_forward_alloc:
> tcp_v6_rcv
> tcp_v6_do_rcv
> skb_clone_and_charge_r
> sk_rmem_schedule
> __sk_mem_schedule
> sk_forward_alloc_add()
> skb_set_owner_r
> sk_mem_charge
> sk_forward_alloc_add()
> __kfree_skb
> skb_release_all
> skb_release_head_state
> sock_rfree
> sk_mem_uncharge
> sk_forward_alloc_add()
> sk_mem_reclaim
> // set local var reclaimable
> __sk_mem_reclaim
> sk_forward_alloc_add()
>
> In this syzkaller testcase, two threads call tcp_v6_do_rcv() with
> skb->truesize=768, the sk_forward_alloc changes like this:
> (cpu 1) | (cpu 2) | sk_forward_alloc
> ... | ... | 0
> __sk_mem_schedule() | | +4096 = 4096
> | __sk_mem_schedule() | +4096 = 8192
> sk_mem_charge() | | -768 = 7424
> | sk_mem_charge() | -768 = 6656
> ... | ... |
> sk_mem_uncharge() | | +768 = 7424
> reclaimable=7424 | |
> | sk_mem_uncharge() | +768 = 8192
> | reclaimable=8192 |
> __sk_mem_reclaim() | | -4096 = 4096
> | __sk_mem_reclaim() | -8192 = -4096 != 0
>
> Add lock around tcp_v6_do_rcv() in tcp_v6_rcv() will have some the
> performance impacts, only add lock when opt_skb clone occurs. In some
> scenes, tcp_v6_do_rcv() is embraced by sk->sk_lock, add
> TCP_SKB_CB(skb)->sk_lock_capability to avoid re-locking.
>
> Fixes: e994b2f0fb92 ("tcp: do not lock listener to process SYN packets")
> Signed-off-by: Wang Liang <wangliang74@...wei.com>
> ---
> include/net/tcp.h | 3 ++-
> net/ipv6/tcp_ipv6.c | 21 ++++++++++++++++-----
> 2 files changed, 18 insertions(+), 6 deletions(-)
>
> diff --git a/include/net/tcp.h b/include/net/tcp.h
> index d1948d357dad..110a23dda1eb 100644
> --- a/include/net/tcp.h
> +++ b/include/net/tcp.h
> @@ -961,7 +961,8 @@ struct tcp_skb_cb {
> __u8 txstamp_ack:1, /* Record TX timestamp for ack? */
> eor:1, /* Is skb MSG_EOR marked? */
> has_rxtstamp:1, /* SKB has a RX timestamp */
> - unused:5;
> + sk_lock_capability:1, /* Avoid re-lock flag */
> + unused:4;
> __u32 ack_seq; /* Sequence number ACK'd */
> union {
> struct {
Oh the horror, this is completely wrong and unsafe anyway.
TCP listen path MUST be lockless, and stay lockless.
Ask yourself : Why would a listener even hold a pktoptions in the first place ?
Normally, each request socket can hold an ireq->pktopts (see in
tcp_v6_init_req())
The skb_clone_and_charge_r() happen later in tcp_v6_syn_recv_sock()
The correct fix is to _not_ call skb_clone_and_charge_r() for a
listener socket, of course, this never made _any_ sense.
The following patch should fix both TCP and DCCP, and as a bonus make
TCP SYN processing faster
for listeners requesting these IPV6_PKTOPTIONS things.
diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c
index da5dba120bc9a55c5fd9d6feda791b0ffc887423..d6649246188d72b3df6c74750779b7aa5910dcb7
100644
--- a/net/dccp/ipv6.c
+++ b/net/dccp/ipv6.c
@@ -618,7 +618,7 @@ static int dccp_v6_do_rcv(struct sock *sk, struct
sk_buff *skb)
by tcp. Feel free to propose better solution.
--ANK (980728)
*/
- if (np->rxopt.all)
+ if (np->rxopt.all && sk->sk_state != DCCP_LISTEN)
opt_skb = skb_clone_and_charge_r(skb, sk);
if (sk->sk_state == DCCP_OPEN) { /* Fast path */
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index d71ab4e1efe1c6598cf3d3e4334adf0881064ce9..e643dbaec9ccc92eb2d9103baf185c957ad1dd2e
100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -1605,25 +1605,12 @@ int tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb)
* is currently called with bh processing disabled.
*/
- /* Do Stevens' IPV6_PKTOPTIONS.
-
- Yes, guys, it is the only place in our code, where we
- may make it not affecting IPv4.
- The rest of code is protocol independent,
- and I do not like idea to uglify IPv4.
-
- Actually, all the idea behind IPV6_PKTOPTIONS
- looks not very well thought. For now we latch
- options, received in the last packet, enqueued
- by tcp. Feel free to propose better solution.
- --ANK (980728)
- */
- if (np->rxopt.all)
- opt_skb = skb_clone_and_charge_r(skb, sk);
if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */
struct dst_entry *dst;
+ if (np->rxopt.all)
+ opt_skb = skb_clone_and_charge_r(skb, sk);
dst = rcu_dereference_protected(sk->sk_rx_dst,
lockdep_sock_is_held(sk));
@@ -1656,13 +1643,13 @@ int tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb)
if (reason)
goto reset;
}
- if (opt_skb)
- __kfree_skb(opt_skb);
return 0;
}
} else
sock_rps_save_rxhash(sk, skb);
+ if (np->rxopt.all)
+ opt_skb = skb_clone_and_charge_r(skb, sk);
reason = tcp_rcv_state_process(sk, skb);
if (reason)
goto reset;
Powered by blists - more mailing lists