Re: use-after-free in ip6_xmit

From: Eric Dumazet
Date: Mon Dec 07 2015 - 09:36:32 EST


On Mon, 2015-12-07 at 11:22 +0100, Dmitry Vyukov wrote:
> Hello,
>
> The following program triggers use-after-free in ip6_xmit:
>
> // autogenerated by syzkaller (http://github.com/google/syzkaller)
> #include <syscall.h>
> #include <string.h>
> #include <stdint.h>
> #include <pthread.h>
> #include <sys/types.h>
> #include <sys/socket.h>
> #include <linux/socket.h>
> #include <linux/in.h>
> #include <linux/in6.h>
> #include <unistd.h>
> #include <stdlib.h>
>
> void *thr0(void *arg)
> {
> *(uint32_t*)0x20000000 = 0x5;
> long r5 = syscall(SYS_setsockopt, arg, 0x29ul, 0x6ul,
> 0x20000000ul, 0x4ul, 0);
> return 0;
> }
>
> void *thr1(void *arg)
> {
> struct sockaddr_in6 sa = {};
> sa.sin6_family = AF_INET6;
> sa.sin6_port = getpid();
> sa.sin6_addr.s6_addr[15] = 1;
> syscall(SYS_bind, arg, &sa, sizeof(sa), 0, 0, 0);
> return 0;
> }
>
> void *thr2(void *arg)
> {
> struct sockaddr_in6 sa = {};
> sa.sin6_family = AF_INET6;
> sa.sin6_port = getpid();
> sa.sin6_addr.s6_addr[15] = 1;
> syscall(SYS_connect, arg, &sa, sizeof(sa), 0, 0, 0);
> return 0;
> }
>
> void *thr3(void *arg)
> {
> syscall(SYS_listen, arg, 0x3ul, 0, 0, 0, 0);
> return 0;
> }
>
> void *thr4(void *arg)
> {
> syscall(SYS_accept4, arg, 0, 0, 0x800ul, 0, 0);
> return 0;
> }
>
> int main()
> {
> srand(getpid());
> syscall(SYS_mmap, 0x20000000ul, 0x10000ul, 0x2ul, 0x32ul,
> 0xfffffffffffffffful, 0x0ul);
> int fd[2];
> fd[0] = syscall(SYS_socket, PF_INET6, SOCK_STREAM, IPPROTO_TCP);
> fd[1] = syscall(SYS_socket, PF_INET6, SOCK_STREAM, IPPROTO_TCP);
> pthread_t th;
> pthread_create(&th, 0, thr0, (void*)(long)fd[0]);
> pthread_create(&th, 0, thr1, (void*)(long)fd[0]);
> pthread_create(&th, 0, thr2, (void*)(long)fd[0]);
> pthread_create(&th, 0, thr3, (void*)(long)fd[0]);
> pthread_create(&th, 0, thr4, (void*)(long)fd[0]);
> pthread_create(&th, 0, thr0, (void*)(long)fd[0]);
> pthread_create(&th, 0, thr1, (void*)(long)fd[0]);
> pthread_create(&th, 0, thr2, (void*)(long)fd[0]);
> pthread_create(&th, 0, thr3, (void*)(long)fd[0]);
> pthread_create(&th, 0, thr4, (void*)(long)fd[0]);
> pthread_create(&th, 0, thr0, (void*)(long)fd[1]);
> pthread_create(&th, 0, thr1, (void*)(long)fd[1]);
> pthread_create(&th, 0, thr2, (void*)(long)fd[1]);
> pthread_create(&th, 0, thr3, (void*)(long)fd[1]);
> pthread_create(&th, 0, thr4, (void*)(long)fd[1]);
> pthread_create(&th, 0, thr0, (void*)(long)fd[1]);
> pthread_create(&th, 0, thr1, (void*)(long)fd[1]);
> pthread_create(&th, 0, thr2, (void*)(long)fd[1]);
> pthread_create(&th, 0, thr3, (void*)(long)fd[1]);
> pthread_create(&th, 0, thr4, (void*)(long)fd[1]);
>
> usleep(20000);
> return 0;
> }
>
>
>
> ==================================================================
> BUG: KASAN: use-after-free in ip6_xmit+0x15e3/0x1f50 at addr ffff88003991c786
> Read of size 2 by task a.out/6864
> =============================================================================
> BUG kmalloc-64 (Not tainted): kasan: bad access detected
> -----------------------------------------------------------------------------
>
> Disabling lock debugging due to kernel taint
> INFO: Allocated in sock_kmalloc+0x93/0x100 age=40 cpu=1 pid=6850
> [< none >] ___slab_alloc+0x648/0x8c0 mm/slub.c:2438
> [< none >] __slab_alloc+0x4c/0x90 mm/slub.c:2467
> [< inline >] slab_alloc_node mm/slub.c:2530
> [< inline >] slab_alloc mm/slub.c:2572
> [< none >] __kmalloc+0x2d9/0x480 mm/slub.c:3532
> [< inline >] kmalloc include/linux/slab.h:463
> [< none >] sock_kmalloc+0x93/0x100 net/core/sock.c:1772
> [< none >] do_ipv6_setsockopt.isra.5+0xf4d/0x2d30
> net/ipv6/ipv6_sockglue.c:483
> [< none >] ipv6_setsockopt+0x4f/0x150 net/ipv6/ipv6_sockglue.c:885
> [< none >] sctp_setsockopt+0x194/0x4020 net/sctp/socket.c:3702
> [< none >] sock_common_setsockopt+0xb4/0x140 net/core/sock.c:2643
> [< inline >] SYSC_setsockopt net/socket.c:1757
> [< none >] SyS_setsockopt+0x161/0x290 net/socket.c:1736
> [< none >] entry_SYSCALL_64_fastpath+0x16/0x7a
> arch/x86/entry/entry_64.S:185
>
> INFO: Freed in sock_kfree_s+0x29/0x90 age=34 cpu=0 pid=6855
> [< none >] __slab_free+0x21e/0x3e0 mm/slub.c:2648
> [< inline >] slab_free mm/slub.c:2803
> [< none >] kfree+0x26f/0x3e0 mm/slub.c:3632
> [< inline >] __sock_kfree_s net/core/sock.c:1793
> [< none >] sock_kfree_s+0x29/0x90 net/core/sock.c:1799
> [< none >] do_ipv6_setsockopt.isra.5+0x100f/0x2d30
> net/ipv6/ipv6_sockglue.c:506
> [< none >] ipv6_setsockopt+0x4f/0x150 net/ipv6/ipv6_sockglue.c:885
> [< none >] sctp_setsockopt+0x194/0x4020 net/sctp/socket.c:3702
> [< none >] sock_common_setsockopt+0xb4/0x140 net/core/sock.c:2643
> [< inline >] SYSC_setsockopt net/socket.c:1757
> [< none >] SyS_setsockopt+0x161/0x290 net/socket.c:1736
> [< none >] entry_SYSCALL_64_fastpath+0x16/0x7a
> arch/x86/entry/entry_64.S:185
>
> INFO: Slab 0xffffea0000e64700 objects=21 used=19 fp=0xffff88003991cc00
> flags=0x1fffc0000004080
> INFO: Object 0xffff88003991c780 @offset=1920 fp=0x (null)
> CPU: 2 PID: 6864 Comm: a.out Tainted: G B 4.4.0-rc3+ #150
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
> 0000000000000002 ffff88005f3f6e18 ffffffff82c6f2a8 0000000041b58ab3
> ffffffff8788bf8d ffffffff82c6f1f6 ffff880062045a00 ffffffff878ac723
> ffff88003e807700 0000000000000008 ffff88003991c780 ffff88005f3f6e18
>
> Call Trace:
> [< inline >] __dump_stack lib/dump_stack.c:15
> [<ffffffff82c6f2a8>] dump_stack+0xb2/0xfa lib/dump_stack.c:50
> [<ffffffff81788e3c>] print_trailer+0x12c/0x210 mm/slub.c:652
> [<ffffffff81793c5f>] object_err+0x2f/0x40 mm/slub.c:659
> [< inline >] print_address_description mm/kasan/report.c:138
> [<ffffffff817979c9>] kasan_report_error+0x5d9/0x860 mm/kasan/report.c:251
> [< inline >] kasan_report mm/kasan/report.c:274
> [<ffffffff81797d14>] __asan_report_load2_noabort+0x54/0x70
> mm/kasan/report.c:293
> [<ffffffff85bc06e3>] ip6_xmit+0x15e3/0x1f50 net/ipv6/ip6_output.c:176
> [<ffffffff8635a0b9>] sctp_v6_xmit+0x319/0x530 net/sctp/ipv6.c:223
> [<ffffffff8634595d>] sctp_packet_transmit+0x113d/0x2680 net/sctp/output.c:595
> [<ffffffff8630e897>] sctp_outq_flush+0x717/0x2840 net/sctp/outqueue.c:1090
> [<ffffffff86315d67>] sctp_outq_uncork+0x57/0x80 net/sctp/outqueue.c:692
> [< inline >] sctp_cmd_interpreter net/sctp/sm_sideeffect.c:1750
> [< inline >] sctp_side_effects net/sctp/sm_sideeffect.c:1153
> [<ffffffff862d9e83>] sctp_do_sm+0x38a3/0x5be0 net/sctp/sm_sideeffect.c:1125
> [<ffffffff863423c9>] sctp_primitive_SHUTDOWN+0xa9/0xd0 net/sctp/primitive.c:104
> [<ffffffff8632e94d>] sctp_close+0x6fd/0x9b0 net/sctp/socket.c:1519
> [<ffffffff85a4b2a1>] inet_release+0x111/0x270 net/ipv4/af_inet.c:413
> [<ffffffff85bac8c5>] inet6_release+0x55/0x90 net/ipv6/af_inet6.c:406
> [<ffffffff85501586>] sock_release+0x96/0x260 net/socket.c:571
> [<ffffffff85501766>] sock_close+0x16/0x20 net/socket.c:1022
> [<ffffffff817f0014>] __fput+0x244/0x860 fs/file_table.c:208
> [<ffffffff817f06c5>] ____fput+0x15/0x20 fs/file_table.c:244
> [<ffffffff8133a5e0>] task_work_run+0x130/0x240 kernel/task_work.c:115
> [< inline >] exit_task_work include/linux/task_work.h:21
> [<ffffffff812d4a05>] do_exit+0x885/0x3050 kernel/exit.c:750
> [<ffffffff812d732c>] do_group_exit+0xec/0x390 kernel/exit.c:880
> [<ffffffff81302177>] get_signal+0x677/0x1bf0 kernel/signal.c:2307
> [<ffffffff8118645e>] do_signal+0x7e/0x2170 arch/x86/kernel/signal.c:709
> [<ffffffff81003a1e>] exit_to_usermode_loop+0xfe/0x1e0
> arch/x86/entry/common.c:247
> [< inline >] prepare_exit_to_usermode arch/x86/entry/common.c:282
> [<ffffffff8100733b>] syscall_return_slowpath+0x16b/0x240
> arch/x86/entry/common.c:344
> [<ffffffff868dafe2>] int_ret_from_sys_call+0x25/0x9f
> arch/x86/entry/entry_64.S:281
> ==================================================================
>
> On commit 31ade3b83e1821da5fbb2f11b5b3d4ab2ec39db8.
>
> The allocation/free stacks look similar to the ones I reported in
> "memory leak in do_ipv6_setsockopt". However, the fix for that leak
> suggests that it could lead only to leaks, not to use-after-frees.
> --

I thought I already fixed these, based on your report ?

commit 6bd4f355df2eae80b8a5c7b097371cd1e05f20d5
Author: Eric Dumazet <edumazet@xxxxxxxxxx>
Date: Wed Dec 2 21:53:57 2015 -0800

ipv6: kill sk_dst_lock

While testing the np->opt RCU conversion, I found that UDP/IPv6 was
using a mixture of xchg() and sk_dst_lock to protect concurrent changes
to sk->sk_dst_cache, leading to possible corruptions and crashes.

ip6_sk_dst_lookup_flow() uses sk_dst_check() anyway, so the simplest
way to fix the mess is to remove sk_dst_lock completely, as we did for
IPv4.

__ip6_dst_store() and ip6_dst_store() share same implementation.

sk_setup_caps() being called with socket lock being held or not,
we have to use sk_dst_set() instead of __sk_dst_set()

Note that I had to move the "np->dst_cookie = rt6_get_cookie(rt);"
in ip6_dst_store() before the sk_setup_caps(sk, dst) call.

This is because ip6_dst_store() can be called from process context,
without any lock held.

As soon as the dst is installed in sk->sk_dst_cache, dst can be freed
from another cpu doing a concurrent ip6_dst_store()

Doing the dst dereference before doing the install is needed to make
sure no use after free would trigger.

Signed-off-by: Eric Dumazet <edumazet@xxxxxxxxxx>
Reported-by: Dmitry Vyukov <dvyukov@xxxxxxxxxx>
Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx>

commit 45f6fad84cc305103b28d73482b344d7f5b76f39
Author: Eric Dumazet <edumazet@xxxxxxxxxx>
Date: Sun Nov 29 19:37:57 2015 -0800

ipv6: add complete rcu protection around np->opt

This patch addresses multiple problems :

UDP/RAW sendmsg() need to get a stable struct ipv6_txoptions
while socket is not locked : Other threads can change np->opt
concurrently. Dmitry posted a syzkaller
(http://github.com/google/syzkaller) program desmonstrating
use-after-free.

Starting with TCP/DCCP lockless listeners, tcp_v6_syn_recv_sock()
and dccp_v6_request_recv_sock() also need to use RCU protection
to dereference np->opt once (before calling ipv6_dup_options())

This patch adds full RCU protection to np->opt

Reported-by: Dmitry Vyukov <dvyukov@xxxxxxxxxx>
Signed-off-by: Eric Dumazet <edumazet@xxxxxxxxxx>
Acked-by: Hannes Frederic Sowa <hannes@xxxxxxxxxxxxxxxxxxx>
Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx>


commit 602dd62dfbda3e63a2d6a3cbde953ebe82bf5087
Author: Eric Dumazet <edumazet@xxxxxxxxxx>
Date: Tue Dec 1 07:20:07 2015 -0800

ipv6: sctp: implement sctp_v6_destroy_sock()

Dmitry Vyukov reported a memory leak using IPV6 SCTP sockets.

We need to call inet6_destroy_sock() to properly release
inet6 specific fields.

Reported-by: Dmitry Vyukov <dvyukov@xxxxxxxxxx>
Signed-off-by: Eric Dumazet <edumazet@xxxxxxxxxx>
Acked-by: Daniel Borkmann <daniel@xxxxxxxxxxxxx>
Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx>

commit c836a8ba93869d6a0290a6ae0047fbef09066871
Author: Eric Dumazet <edumazet@xxxxxxxxxx>
Date: Wed Dec 2 21:48:14 2015 -0800

ipv6: sctp: add rcu protection around np->opt

This patch completes the work I did in commit 45f6fad84cc3
("ipv6: add complete rcu protection around np->opt"), as I missed
sctp part.

This simply makes sure np->opt is used with proper RCU locking
and accessors.

Signed-off-by: Eric Dumazet <edumazet@xxxxxxxxxx>
Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx>

Can you double check based on linux-4.4-rc4 instead of linux-4.4-rc3 ?

Thanks


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/