tcp: gso: avoid refcount_t warning from tcp_gso_segment()

When a GSO skb of truesize O is segmented into 2 new skbs of truesize N1
and N2, we want to transfer socket ownership to the new fresh skbs.

In order to avoid expensive atomic operations on a cache line subject to
cache bouncing, we replace the sequence :

refcount_add(N1, &sk->sk_wmem_alloc);
refcount_add(N2, &sk->sk_wmem_alloc); // repeated by number of segments

refcount_sub(O, &sk->sk_wmem_alloc);

by a single

refcount_add(sum_of(N) - O, &sk->sk_wmem_alloc);

Problem is :

In some pathological cases, sum(N) - O might be a negative number, and
syzkaller bot was apparently able to trigger this trace [1]

atomic_t was ok with this construct, but we need to take care of the
negative delta with refcount_t

[1]
refcount_t: saturated; leaking memory.
------------[ cut here ]------------
WARNING: CPU: 0 PID: 8404 at lib/refcount.c:77 refcount_add_not_zero+0x198/0x200 lib/refcount.c:77
Kernel panic - not syncing: panic_on_warn set ...

CPU: 0 PID: 8404 Comm: syz-executor2 Not tainted 4.14.0-rc5-mm1+ #20
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:16 [inline]
 dump_stack+0x194/0x257 lib/dump_stack.c:52
 panic+0x1e4/0x41c kernel/panic.c:183
 __warn+0x1c4/0x1e0 kernel/panic.c:546
 report_bug+0x211/0x2d0 lib/bug.c:183
 fixup_bug+0x40/0x90 arch/x86/kernel/traps.c:177
 do_trap_no_signal arch/x86/kernel/traps.c:211 [inline]
 do_trap+0x260/0x390 arch/x86/kernel/traps.c:260
 do_error_trap+0x120/0x390 arch/x86/kernel/traps.c:297
 do_invalid_op+0x1b/0x20 arch/x86/kernel/traps.c:310
 invalid_op+0x18/0x20 arch/x86/entry/entry_64.S:905
RIP: 0010:refcount_add_not_zero+0x198/0x200 lib/refcount.c:77
RSP: 0018:ffff8801c606e3a0 EFLAGS: 00010282
RAX: 0000000000000026 RBX: 0000000000001401 RCX: 0000000000000000
RDX: 0000000000000026 RSI: ffffc900036fc000 RDI: ffffed0038c0dc68
RBP: ffff8801c606e430 R08: 0000000000000001 R09: 0000000000000000
R10: ffff8801d97f5eba R11: 0000000000000000 R12: ffff8801d5acf73c
R13: 1ffff10038c0dc75 R14: 00000000ffffffff R15: 00000000fffff72f
 refcount_add+0x1b/0x60 lib/refcount.c:101
 tcp_gso_segment+0x10d0/0x16b0 net/ipv4/tcp_offload.c:155
 tcp4_gso_segment+0xd4/0x310 net/ipv4/tcp_offload.c:51
 inet_gso_segment+0x60c/0x11c0 net/ipv4/af_inet.c:1271
 skb_mac_gso_segment+0x33f/0x660 net/core/dev.c:2749
 __skb_gso_segment+0x35f/0x7f0 net/core/dev.c:2821
 skb_gso_segment include/linux/netdevice.h:3971 [inline]
 validate_xmit_skb+0x4ba/0xb20 net/core/dev.c:3074
 __dev_queue_xmit+0xe49/0x2070 net/core/dev.c:3497
 dev_queue_xmit+0x17/0x20 net/core/dev.c:3538
 neigh_hh_output include/net/neighbour.h:471 [inline]
 neigh_output include/net/neighbour.h:479 [inline]
 ip_finish_output2+0xece/0x1460 net/ipv4/ip_output.c:229
 ip_finish_output+0x85e/0xd10 net/ipv4/ip_output.c:317
 NF_HOOK_COND include/linux/netfilter.h:238 [inline]
 ip_output+0x1cc/0x860 net/ipv4/ip_output.c:405
 dst_output include/net/dst.h:459 [inline]
 ip_local_out+0x95/0x160 net/ipv4/ip_output.c:124
 ip_queue_xmit+0x8c6/0x18e0 net/ipv4/ip_output.c:504
 tcp_transmit_skb+0x1ab7/0x3840 net/ipv4/tcp_output.c:1137
 tcp_write_xmit+0x663/0x4de0 net/ipv4/tcp_output.c:2341
 __tcp_push_pending_frames+0xa0/0x250 net/ipv4/tcp_output.c:2513
 tcp_push_pending_frames include/net/tcp.h:1722 [inline]
 tcp_data_snd_check net/ipv4/tcp_input.c:5050 [inline]
 tcp_rcv_established+0x8c7/0x18a0 net/ipv4/tcp_input.c:5497
 tcp_v4_do_rcv+0x2ab/0x7d0 net/ipv4/tcp_ipv4.c:1460
 sk_backlog_rcv include/net/sock.h:909 [inline]
 __release_sock+0x124/0x360 net/core/sock.c:2264
 release_sock+0xa4/0x2a0 net/core/sock.c:2776
 tcp_sendmsg+0x3a/0x50 net/ipv4/tcp.c:1462
 inet_sendmsg+0x11f/0x5e0 net/ipv4/af_inet.c:763
 sock_sendmsg_nosec net/socket.c:632 [inline]
 sock_sendmsg+0xca/0x110 net/socket.c:642
 ___sys_sendmsg+0x31c/0x890 net/socket.c:2048
 __sys_sendmmsg+0x1e6/0x5f0 net/socket.c:2138

Fixes: 14afee4b6092 ("net: convert sock.sk_wmem_alloc from atomic_t to refcount_t")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/ipv4/tcp_offload.c b/net/ipv4/tcp_offload.c
index 11f69bbf93072b7b4dbc3a0485c9f9e0b9ba30b3..b6a2aa1dcf56cbaa029b0ec6c5e512e33c90fff9 100644 (file)
--- a/net/ipv4/tcp_offload.c
+++ b/net/ipv4/tcp_offload.c
@@ -149,11 +149,19 @@ struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
         * is freed by GSO engine
         */
        if (copy_destructor) {
+               int delta;
+
                swap(gso_skb->sk, skb->sk);
                swap(gso_skb->destructor, skb->destructor);
                sum_truesize += skb->truesize;
-               refcount_add(sum_truesize - gso_skb->truesize,
-                          &skb->sk->sk_wmem_alloc);
+               delta = sum_truesize - gso_skb->truesize;
+               /* In some pathological cases, delta can be negative.
+                * We need to either use refcount_add() or refcount_sub_and_test()
+                */
+               if (likely(delta >= 0))
+                       refcount_add(delta, &skb->sk->sk_wmem_alloc);
+               else
+                       WARN_ON_ONCE(refcount_sub_and_test(-delta, &skb->sk->sk_wmem_alloc));
        }
 
        delta = htonl(oldlen + (skb_tail_pointer(skb) -