Re: [PATCH 2/2] tracing: Fix wild-memory-access in register_synth_event()

[shangxiaojing <shangxiaojing@huawei.com>]

On 2022/11/13 5:24, Tom Zanussi wrote:
> Hi Shang,
> 
> On Tue, 2022-11-08 at 16:31 +0800, Shang XiaoJing wrote:
>> In register_synth_event(), if set_synth_event_print_fmt() failed, then
>> both trace_remove_event_call() and unregister_trace_event() will be
>> called. If call->event.funcs is not NULL, then the trace_event_call will
>> call __unregister_trace_event() twice. As the result, the second
>> __unregister_trace_event() will causes the wild-memory-access.
>>
>> register_synth_event
>>      set_synth_event_print_fmt failed
>>      trace_remove_event_call
>>          event_remove
>>              if call->event.funcs then
>>              __unregister_trace_event (first call)
>>      unregister_trace_event
>>          __unregister_trace_event (second call)
>>
>> Fix the bug by avoiding to call the second __unregister_trace_event() by
>> checking if the first one is called.
>>
>> general protection fault, probably for non-canonical address
>>          0xfbd59c0000000024: 0000 [#1] SMP KASAN PTI
>> KASAN: maybe wild-memory-access in range
>> [0xdead000000000120-0xdead000000000127]
>> CPU: 0 PID: 3807 Comm: modprobe Not tainted
>> 6.1.0-rc1-00186-g76f33a7eedb4 #299
>> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
>> rel-1.15.0-0-g2dd4b9b3f840-prebuilt.qemu.org 04/01/2014
>> RIP: 0010:unregister_trace_event+0x6e/0x280
>> Code: 00 fc ff df 4c 89 ea 48 c1 ea 03 80 3c 02 00 0f 85 0e 02 00 00 48
>> b8 00 00 00 00 00 fc ff df 4c 8b 63 08 4c 89 e2 48 c1 ea 03 <80> 3c 02
>> 00 0f 85 e2 01 00 00 49 89 2c 24 48 85 ed 74 28 e8 7a 9b
>> RSP: 0018:ffff88810413f370 EFLAGS: 00010a06
>> RAX: dffffc0000000000 RBX: ffff888105d050b0 RCX: 0000000000000000
>> RDX: 1bd5a00000000024 RSI: ffff888119e276e0 RDI: ffffffff835a8b20
>> RBP: dead000000000100 R08: 0000000000000000 R09: fffffbfff0913481
>> R10: ffffffff8489a407 R11: fffffbfff0913480 R12: dead000000000122
>> R13: ffff888105d050b8 R14: 0000000000000000 R15: ffff888105d05028
>> FS:  00007f7823e8d540(0000) GS:ffff888119e00000(0000)
>> knlGS:0000000000000000
>> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>> CR2: 00007f7823e7ebec CR3: 000000010a058002 CR4: 0000000000330ef0
>> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
>> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
>> Call Trace:
>>   <TASK>
>>   __create_synth_event+0x1e37/0x1eb0
>>   create_or_delete_synth_event+0x110/0x250
>>   synth_event_run_command+0x2f/0x110
>>   test_gen_synth_cmd+0x170/0x2eb [synth_event_gen_test]
>>   synth_event_gen_test_init+0x76/0x9bc [synth_event_gen_test]
>>   do_one_initcall+0xdb/0x480
>>   do_init_module+0x1cf/0x680
>>   load_module+0x6a50/0x70a0
>>   __do_sys_finit_module+0x12f/0x1c0
>>   do_syscall_64+0x3f/0x90
>>   entry_SYSCALL_64_after_hwframe+0x63/0xcd
>>
>> Fixes: 4b147936fa50 ("tracing: Add support for 'synthetic' events")
>> Signed-off-by: Shang XiaoJing <shangxiaojing@huawei.com>
>> Cc: stable@vger.kernel.org
>> ---
>>   kernel/trace/trace_events_synth.c | 2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/kernel/trace/trace_events_synth.c b/kernel/trace/trace_events_synth.c
>> index e310052dc83c..a51280a153e3 100644
>> --- a/kernel/trace/trace_events_synth.c
>> +++ b/kernel/trace/trace_events_synth.c
>> @@ -830,6 +830,8 @@ static int register_synth_event(struct synth_event *event)
>>          ret = set_synth_event_print_fmt(call);
>>          if (ret < 0) {
>>                  trace_remove_event_call(call);
>> +               if (call->event.funcs)
>> +                       return ret;
>>                  goto err;
>>          }
>>    out:
> 
> Good catch, thanks for finding this bug!
> 
> It looks like call->event.funcs will always be true here since it's set
> to &synth_event_funcs above.
> 
> So it seems like you could just call trace_remove_event_call() and fall
> through for this (ret < 0) case?  If so, it might be good to put a
> comment there noting that trace_remove_event_call() will call
> unregister_trace_event(), so it's ok to just return.
> 

Ok, will fix in v2.

Thanks,
-- 
Shang XiaoJing