Date:   Sun, 1 Mar 2020 18:49:20 -0800
From:   Eric Dumazet <eric.dumazet@...il.com>
To:     syzbot <syzbot+dd803bc0e8adf0003261@...kaller.appspotmail.com>,
        andriin@...com, ast@...nel.org, bpf@...r.kernel.org,
        daniel@...earbox.net, davem@...emloft.net, edumazet@...gle.com,
        hawk@...nel.org, jasowang@...hat.com, jhs@...atatu.com,
        jiri@...nulli.us, john.fastabend@...il.com, kafai@...com,
        kuba@...nel.org, linux-kernel@...r.kernel.org,
        netdev@...r.kernel.org, songliubraving@...com,
        syzkaller-bugs@...glegroups.com, xiyou.wangcong@...il.com,
        yhs@...com
Subject: Re: WARNING: refcount bug in __sk_destruct



On 3/1/20 6:38 PM, Eric Dumazet wrote:
> 
> 
> On 3/1/20 6:22 PM, Eric Dumazet wrote:
>>
>>
>> On 3/1/20 5:33 PM, syzbot wrote:
>>> Hello,
>>>
>>> syzbot found the following crash on:
>>>
>>> HEAD commit:    fd786fb1 net: convert suitable drivers to use phy_do_ioctl..
>>> git tree:       net-next
>>> console output: https://syzkaller.appspot.com/x/log.txt?x=14e9726ee00000
>>> kernel config:  https://syzkaller.appspot.com/x/.config?x=7f93900a7904130d
>>> dashboard link: https://syzkaller.appspot.com/bug?extid=dd803bc0e8adf0003261
>>> compiler:       gcc (GCC) 9.0.0 20181231 (experimental)
>>> syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=12af9369e00000
>>> C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=10583d76e00000
>>>
>>> The bug was bisected to:
>>>
>>> commit 14215108a1fd7e002c0a1f9faf8fbaf41fdda50d
>>> Author: Cong Wang <xiyou.wangcong@...il.com>
>>> Date:   Thu Feb 21 05:37:42 2019 +0000
>>>
>>>     net_sched: initialize net pointer inside tcf_exts_init()
>>>
>>> bisection log:  https://syzkaller.appspot.com/x/bisect.txt?x=175b66bee00000
>>> final crash:    https://syzkaller.appspot.com/x/report.txt?x=14db66bee00000
>>> console output: https://syzkaller.appspot.com/x/log.txt?x=10db66bee00000
>>>
>>> IMPORTANT: if you fix the bug, please add the following tag to the commit:
>>> Reported-by: syzbot+dd803bc0e8adf0003261@...kaller.appspotmail.com
>>> Fixes: 14215108a1fd ("net_sched: initialize net pointer inside tcf_exts_init()")
>>>
>>> RBP: 0000000000000000 R08: 0000000000000002 R09: 00000000bb1414ac
>>> R10: 0000000000000000 R11: 0000000000000246 R12: ffffffffffffffff
>>> R13: 0000000000000009 R14: 0000000000000000 R15: 0000000000000000
>>> ------------[ cut here ]------------
>>> refcount_t: underflow; use-after-free.
>>> WARNING: CPU: 1 PID: 9577 at lib/refcount.c:28 refcount_warn_saturate+0x1dc/0x1f0 lib/refcount.c:28
>>> Kernel panic - not syncing: panic_on_warn set ...
>>> CPU: 1 PID: 9577 Comm: syz-executor327 Not tainted 5.5.0-rc6-syzkaller #0
>>> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
>>> Call Trace:
>>>  __dump_stack lib/dump_stack.c:77 [inline]
>>>  dump_stack+0x197/0x210 lib/dump_stack.c:118
>>>  panic+0x2e3/0x75c kernel/panic.c:221
>>>  __warn.cold+0x2f/0x3e kernel/panic.c:582
>>>  report_bug+0x289/0x300 lib/bug.c:195
>>>  fixup_bug arch/x86/kernel/traps.c:174 [inline]
>>>  fixup_bug arch/x86/kernel/traps.c:169 [inline]
>>>  do_error_trap+0x11b/0x200 arch/x86/kernel/traps.c:267
>>>  do_invalid_op+0x37/0x50 arch/x86/kernel/traps.c:286
>>>  invalid_op+0x23/0x30 arch/x86/entry/entry_64.S:1027
>>> RIP: 0010:refcount_warn_saturate+0x1dc/0x1f0 lib/refcount.c:28
>>> Code: e9 d8 fe ff ff 48 89 df e8 c1 f8 16 fe e9 85 fe ff ff e8 a7 77 d8 fd 48 c7 c7 e0 44 71 88 c6 05 9e 86 db 06 01 e8 93 27 a9 fd <0f> 0b e9 ac fe ff ff 0f 1f 00 66 2e 0f 1f 84 00 00 00 00 00 55 48
>>> RSP: 0018:ffffc9000c067b00 EFLAGS: 00010282
>>> RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000
>>> RDX: 0000000000000000 RSI: ffffffff815e5dd6 RDI: fffff5200180cf52
>>> RBP: ffffc9000c067b10 R08: ffff8880a8630340 R09: ffffed1015d26621
>>> R10: ffffed1015d26620 R11: ffff8880ae933107 R12: 0000000000000003
>>> R13: ffff888090256000 R14: ffff8880a7e5c040 R15: ffff8880a7e5c044
>>>  refcount_sub_and_test include/linux/refcount.h:261 [inline]
>>>  refcount_dec_and_test include/linux/refcount.h:281 [inline]
>>>  put_net include/net/net_namespace.h:259 [inline]
>>>  __sk_destruct+0x6d8/0x7f0 net/core/sock.c:1723
>>>  sk_destruct+0xd5/0x110 net/core/sock.c:1739
>>>  __sk_free+0xfb/0x3f0 net/core/sock.c:1750
>>>  sk_free+0x83/0xb0 net/core/sock.c:1761
>>>  sock_put include/net/sock.h:1719 [inline]
>>>  __tun_detach+0xbe0/0x1150 drivers/net/tun.c:728
>>>  tun_detach drivers/net/tun.c:740 [inline]
>>>  tun_chr_close+0xe0/0x180 drivers/net/tun.c:3455
>>>  __fput+0x2ff/0x890 fs/file_table.c:280
>>>  ____fput+0x16/0x20 fs/file_table.c:313
>>>  task_work_run+0x145/0x1c0 kernel/task_work.c:113
>>>  exit_task_work include/linux/task_work.h:22 [inline]
>>>  do_exit+0xba9/0x2f50 kernel/exit.c:801
>>>  do_group_exit+0x135/0x360 kernel/exit.c:899
>>>  __do_sys_exit_group kernel/exit.c:910 [inline]
>>>  __se_sys_exit_group kernel/exit.c:908 [inline]
>>>  __x64_sys_exit_group+0x44/0x50 kernel/exit.c:908
>>>  do_syscall_64+0xfa/0x790 arch/x86/entry/common.c:294
>>>  entry_SYSCALL_64_after_hwframe+0x49/0xbe
>>> RIP: 0033:0x441a48
>>> Code: Bad RIP value.
>>> RSP: 002b:00007fffe55809a8 EFLAGS: 00000246 ORIG_RAX: 00000000000000e7
>>> RAX: ffffffffffffffda RBX: 0000000000000001 RCX: 0000000000441a48
>>> RDX: 0000000000000001 RSI: 000000000000003c RDI: 0000000000000001
>>> RBP: 00000000004c8430 R08: 00000000000000e7 R09: ffffffffffffffd0
>>> R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000001
>>> R13: 00000000006dba80 R14: 0000000000000000 R15: 0000000000000000
>>> Kernel Offset: disabled
>>> Rebooting in 86400 seconds..
>>>
>>>
>>> ---
>>> This bug is generated by a bot. It may contain errors.
>>> See https://goo.gl/tpsmEJ for more information about syzbot.
>>> syzbot engineers can be reached at syzkaller@...glegroups.com.
>>>
>>> syzbot will keep track of this bug report. See:
>>> https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
>>> For information about bisection process see: https://goo.gl/tpsmEJ#bisection
>>> syzbot can test patches for this bug, for details see:
>>> https://goo.gl/tpsmEJ#testing-patches
>>>
>>
>>
>> Hmmm... maybe this patch would fix it ?
>>
>> diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h
>> index a972244ab1931a2fd52a41efa4f7a6cd78d8746a..6e258a6c1328e38a757d9711f4116a6672995b8f 100644
>> --- a/include/net/pkt_cls.h
>> +++ b/include/net/pkt_cls.h
>> @@ -206,11 +206,12 @@ static inline int tcf_exts_init(struct tcf_exts *exts, struct net *net,
>>  #ifdef CONFIG_NET_CLS_ACT
>>         exts->type = 0;
>>         exts->nr_actions = 0;
>> -       exts->net = net;
>> +       exts->net = NULL;
>>         exts->actions = kcalloc(TCA_ACT_MAX_PRIO, sizeof(struct tc_action *),
>>                                 GFP_KERNEL);
>>         if (!exts->actions)
>>                 return -ENOMEM;
>> +       exts->net = net;
>>  #endif
>>         exts->action = action;
>>         exts->police = police;
>>
> 
> Hmm no.
> 
> Instead, make sure cls_u32 gets a reference so that calls
> to u32_destroy_key(new, false); do not end up decrementing netns refcount.
> 
> 
> diff --git a/net/sched/cls_u32.c b/net/sched/cls_u32.c
> index e15ff335953deef36e55901d8f4ce34e15ed676f..587d71def49157680cf54f84558c4f4e5dad37f1 100644
> --- a/net/sched/cls_u32.c
> +++ b/net/sched/cls_u32.c
> @@ -898,6 +898,7 @@ static int u32_change(struct net *net, struct sk_buff *in_skb,
>                 new = u32_init_knode(net, tp, n);
>                 if (!new)
>                         return -ENOMEM;
> +               tcf_exts_get_net(&new->exts);
>  
>                 err = u32_set_parms(net, tp, base, new, tb,
>                                     tca[TCA_RATE], ovr, extack);
> @@ -918,7 +919,6 @@ static int u32_change(struct net *net, struct sk_buff *in_skb,
>  
>                 u32_replace_knode(tp, tp_c, new);
>                 tcf_unbind_filter(tp, &n->res);
> -               tcf_exts_get_net(&n->exts);
>                 tcf_queue_work(&n->rwork, u32_delete_key_work);
>                 return 0;
>         }
> 
> 

Another possibility would be to have two different pointers to reduce
possibility of a refcount mismatch in some complex error paths.

tcf_exts_put_net() would have no effect if no prior
call to tcf_exts_get_net() has been done.

diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h
index a972244ab1931a2fd52a41efa4f7a6cd78d8746a..163caee65cd4ff07ce32699beab287ceffae0916 100644
--- a/include/net/pkt_cls.h
+++ b/include/net/pkt_cls.h
@@ -192,6 +192,7 @@ struct tcf_exts {
        int nr_actions;
        struct tc_action **actions;
        struct net *net;
+       struct net *ref_net;
 #endif
        /* Map to export classifier specific extension TLV types to the
         * generic extensions API. Unsupported extensions must be set to 0.
@@ -207,6 +208,7 @@ static inline int tcf_exts_init(struct tcf_exts *exts, struct net *net,
        exts->type = 0;
        exts->nr_actions = 0;
        exts->net = net;
+       exts->ref_net = NULL;
        exts->actions = kcalloc(TCA_ACT_MAX_PRIO, sizeof(struct tc_action *),
                                GFP_KERNEL);
        if (!exts->actions)
@@ -224,8 +226,9 @@ static inline int tcf_exts_init(struct tcf_exts *exts, struct net *net,
 static inline bool tcf_exts_get_net(struct tcf_exts *exts)
 {
 #ifdef CONFIG_NET_CLS_ACT
-       exts->net = maybe_get_net(exts->net);
-       return exts->net != NULL;
+       WARN_ON_ONCE(exts->ref_net);
+       exts->ref_net = maybe_get_net(exts->net);
+       return exts->ref_net != NULL;
 #else
        return true;
 #endif
@@ -234,8 +237,8 @@ static inline bool tcf_exts_get_net(struct tcf_exts *exts)
 static inline void tcf_exts_put_net(struct tcf_exts *exts)
 {
 #ifdef CONFIG_NET_CLS_ACT
-       if (exts->net)
-               put_net(exts->net);
+       if (exts->ref_net)
+               put_net(exts->ref_net);
 #endif
 }
 

Powered by blists - more mailing lists