trafficserver-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "xuchao@Gmail" <xuc...@gmail.com>
Subject Re: Crash - looks like 1765 -
Date Fri, 06 Jul 2018 16:10:52 GMT
I'm try to fix the TVC double close bug with https://github.com/apache/trafficserver/pull/1765
.

But the PR#1765 does not fix it and makes new crash.

The 2nd solution to fix the issue https://github.com/apache/trafficserver/pull/2907
.

That's all about the bug.

- oknet xu

发自我的 iPhone

> 在 2018年6月29日,00:23,gksalil@gmail.com <gksalil@gmail.com> 写道:
> 
> 
> 
>> On 2018/05/22 08:56:56, Chao Xu <oknet@apache.org> wrote: 
>> Hi salil,
>> 
>> please try to apply
>> https://github.com/apache/trafficserver/commit/49c903b0c13a2fd2c085325f6942326c4f9c926a
>> .
>> 
>> - Oknet
>> 
>> 2018-05-14 22:03 GMT+08:00 salil GK <gksalil@gmail.com>:
>> 
>>> Hello
>>> 
>>>   I am using ATS 6.2.2 and found a crash happening in
>>> TransformVConnection::do_io_close:467. This looks like bug 1765. But
>>> while looking at that bug - it looks like that defect is not resolved
>>> completely.
>>> 
>>> Stack trace from my machine is as follows
>>> 
>>>>>> 
>>> 
>>> #1 0x00007f42727282a8 in _Unwind_Backtrace () from
>>> /lib/x86_64/libgcc_s.so.1
>>> #2 0x00007f427245fa88 in backtrace () from /lib/x86_64/libc.so.6
>>> #3 0x00007f4274c9bd7f in ink_stack_trace_dump () at ink_stack_trace.cc:61
>>> #4 0x00007f4274c9e2f3 in signal_crash_handler (signo=11) at signals.cc:180
>>> #5 0x00005575fdf9f68a in crash_logger_invoke (signo=11,
>>> info=0x7f427174e630, ctx=0x7f427174e500) at Crash.cc:169
>>> #6 <signal handler called>
>>> #7 0x00000000000000b8 in ?? ()
>>> #8 0x00005575fdff8b5e in TransformVConnection::do_io_close
>>> (this=0x5575ff57eb20, error=20600) at Transform.cc:467
>>> #9 0x00005575fe08d691 in HttpTunnel::chain_abort_all
>>> (this=0x7f426a9d54e8, p=0x7f426a9d56e8) at HttpTunnel.cc:1444
>>> #10 0x00005575fe03761a in HttpSM::tunnel_handler_post_ua
>>> (this=0x7f426a9d41b0, event=104, p=0x7f426a9d56e8) at HttpSM.cc:3505
>>> #11 0x00005575fe08cdeb in HttpTunnel::producer_handler
>>> (this=0x7f426a9d54e8, event=104, p=0x7f426a9d56e8) at
>>> HttpTunnel.cc:1240
>>> #12 0x00005575fe08dc3b in HttpTunnel::main_handler
>>> (this=0x7f426a9d54e8, event=104, data=0x5575ff201e18) at
>>> HttpTunnel.cc:1623
>>> #13 0x00005575fdfa282f in Continuation::handleEvent
>>> (this=0x7f426a9d54e8, event=104, data=0x5575ff201e18) at
>>> ../iocore/eventsystem/I_Continuation.h:153
>>> #14 0x00005575fe1f9a08 in read_signal_and_update (event=104,
>>> vc=0x5575ff201d00) at UnixNetVConnection.cc:148
>>> #15 0x00005575fe1f9da1 in read_signal_done (event=104,
>>> nh=0x7f4271858cd0, vc=0x5575ff201d00) at UnixNetVConnection.cc:209
>>> #16 0x00005575fe1fa61f in read_from_net (nh=0x7f4271858cd0,
>>> vc=0x5575ff201d00, thread=0x7f4271855010) at UnixNetVConnection.cc:359
>>> #17 0x00005575fe1fc78a in UnixNetVConnection::net_read_io
>>> (this=0x5575ff201d00, nh=0x7f4271858cd0, lthread=0x7f4271855010) at
>>> UnixNetVConnection.cc:933
>>> #18 0x00005575fe1f0bc1 in NetHandler::mainNetEvent
>>> (this=0x7f4271858cd0, event=5, e=0x5575ff06ec40) at UnixNet.cc:513
>>> #19 0x00005575fdfa282f in Continuation::handleEvent
>>> (this=0x7f4271858cd0, event=5, data=0x5575ff06ec40) at
>>> ../iocore/eventsystem/I_Continuation.h:153
>>> #20 0x00005575fe21fe1f in EThread::process_event (this=0x7f4271855010,
>>> e=0x5575ff06ec40, calling_code=5) at UnixEThread.cc:148
>>> #21 0x00005575fe22037c in EThread::execute (this=0x7f4271855010) at
>>> UnixEThread.cc:275
>>> #22 0x00005575fe21f312 in spawn_thread_internal (a=0x5575ff03ee10) at
>>> Thread.cc:86
>>> #23 0x00007f4272ecb633 in start_thread (arg=0x7f4271753700) at
>>> pthread_create.c:463
>>> #24 0x00007f427245198f in clone () from /lib/x86_64/libc.so.6
>>> 
>>> <<<<<<
>>> 
>>> What can I do to handle this ...  Can I apply the solution mentioned
>>> in that bug in my 6.2.2 source.
>>> 
>>> Thanks in advance
>>> ~S
>>> 
>> 
>> 
>> 
>> -- 
>> - Oknet Xu
>> 
> 
> Do you have pointers on what circumstance, this issue could happen ? 

Mime
  • Unnamed multipart/alternative (inline, 7-Bit, 0 bytes)
View raw message