current.con.getInfo(): segmentation fault (core dumped) python ./server.py
RedSquirrel
Martin KulasOrganization: PersonalProject: control softwareMember ✭
in Bug Reports
Hi ZeroC users and developers!
Here is the problematic code:
Is the server and client code fine?
Starting the server works:
The client connects to the server and fails:
The server has crashed with a segmentation fault:
Here comes my system configuration:
What am I doing wrong?
Thank you in advance!
Here is the problematic code:
$ cat tux.ice
module chaos
{
interface Tux
{
void fly();
};
};
$ cat client.py
import Ice
import sys
import chaos
ic= Ice.initialize(sys.argv)
base= ic.stringToProxy("theTux:default -p 12345")
tux= chaos.TuxPrx.checkedCast(base)
assert tux is not None
tux.fly()
print "OK"
$ cat server.py
import sys
import Ice
import chaos
class TuxI(chaos.Tux):
def fly(self, current):
print "Flying ..."
print "dir current.con:", dir(current.con)
clientInfo= current.con.getInfo()
print "Done: Flying"
ic= Ice.initialize(sys.argv)
adapter= ic.createObjectAdapterWithEndpoints("theAdapter", "default -p 12345")
obj= TuxI()
adapter.add(obj, ic.stringToIdentity("theTux"))
adapter.activate()
print "Waiting for shutdown"
ic.waitForShutdown()
Is the server and client code fine?
Starting the server works:
$ python ./server.py
Waiting for shutdown
The client connects to the server and fails:
$ python ./client.py
Traceback (most recent call last):
File "./client.py", line 12, in <module>
tux.fly()
File "/home/mkulas/tmp/ice_remote_addr/tux_ice.py", line 56, in fly
return _M_chaos.Tux._op_fly.invoke(self, ((), _ctx))
Ice.ConnectionLostException: Ice.ConnectionLostException:
recv() returned zero
The server has crashed with a segmentation fault:
$ python ./server.py
Waiting for shutdown
Flying ...
dir current.con: ['__class__', '__delattr__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__gt__', '__hash__', '__init__', '__le__', '__lt__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', 'begin_flushBatchRequests', 'close', 'createProxy', 'end_flushBatchRequests', 'flushBatchRequests', 'getAdapter', 'getEndpoint', 'getInfo', 'setAdapter', 'timeout', 'toString', 'type']
zsh: segmentation fault (core dumped) python ./server.py
$ gdb `which python` core
GNU gdb (Ubuntu 7.8-1ubuntu4) 7.8.0.20141001-cvs
Copyright (C) 2014 Free Software Foundation, Inc.
[... SNIPPED ...]
warning: core file may not match specified executable file.
[New LWP 22932]
[New LWP 22928]
[New LWP 22929]
[New LWP 22931]
[New LWP 22933]
[New LWP 22930]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Core was generated by `python ./server.py'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0 0x00007fccaed94838 in __dynamic_cast ()
from /usr/lib/x86_64-linux-gnu/libstdc++.so.6
(gdb) bt
#0 0x00007fccaed94838 in __dynamic_cast ()
from /usr/lib/x86_64-linux-gnu/libstdc++.so.6
#1 0x00007fccafbb64fb in dynamicCast<Ice::ConnectionInfo> (r=...)
at ../../../cpp/include/Ice/Handle.h:175
#2 IcePy::createConnectionInfo (connectionInfo=...) at ConnectionInfo.cpp:423
#3 0x00007fccafbb47e6 in connectionGetInfo (self=<optimized out>)
at Connection.cpp:436
#4 0x00000000004cb00d in PyEval_EvalFrameEx ()
#5 0x00000000004e6970 in ?? ()
#6 0x0000000000505128 in ?? ()
#7 0x00000000004b3ffe in PyObject_Call ()
#8 0x00007fccafbd5b41 in IcePy::TypedUpcall::dispatch (this=0x7fcca4001790,
servant=0x7fccae438b50, inBytes=..., current=...) at Operation.cpp:3337
#9 0x00007fccafbd0fb2 in IcePy::TypedServantWrapper::ice_invoke_async (
this=0x2acfec0, cb=..., inParams=..., current=...) at Operation.cpp:4106
#10 0x00007fccaf7b9e17 in Ice::BlobjectArrayAsync::__dispatch (this=0x2acfec0,
in=..., current=...) at Object.cpp:436
#11 0x00007fccaf759cf4 in IceInternal::Incoming::invoke (
[email protected]=0x7fccacbba490, servantManager=...,
[email protected]=0x7fccacbbaaa8) at Incoming.cpp:677
#12 0x00007fccaf732803 in Ice::ConnectionI::invokeAll (
[email protected]=0x7fcca4000ca0, stream=..., [email protected]=1,
[email protected]=2, [email protected]=0 '\000',
servantManager=..., adapter=...) at ConnectionI.cpp:3124
#13 0x00007fccaf732b98 in Ice::ConnectionI::dispatch (
[email protected]=0x7fcca4000ca0, startCB=...,
sentCBs=std::vector of length 0, capacity 0, compress=<optimized out>,
requestId=2, invokeNum=1, servantManager=..., adapter=..., outAsync=...,
stream=...) at ConnectionI.cpp:1629
#14 0x00007fccaf7346e9 in Ice::ConnectionI::message (this=0x7fcca4000ca0,
current=...) at ConnectionI.cpp:1576
#15 0x00007fccaf86eabc in IceInternal::ThreadPool::run (this=0x2ad0140,
thread=...) at ThreadPool.cpp:643
[... SNIPPED ...]
Here comes my system configuration:
$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 14.10
Release: 14.10
Codename: utopic
$ slice2py --version
3.5.1
$ python --version
Python 2.7.8
What am I doing wrong?
Thank you in advance!
Comments
I cannot reproduce the crash with the provided test case.
Using same OS, Ice and python versions
How did you build and install Ice-3.5.1? seems to me like for some reason the type info for TCPConnectionInfo ins't available to the IcePy python extension and this result in the dynamic_cast crash
My programs were working fine with Ice 3.4.x on Ubuntu 13.10. Then, I have
updated to Ubuntu 14.10 which caused to update the Ice libraries.
Thus, I am using the Ice binaries coming from Ubuntu.
Which libraries do you recommend for Ubuntu 14.10? Do you compile
Ice yourself?
After reinstalling the Ice packages from Ubuntu, the failure was
still there. Then I have built Ice from scratch and then the failure
is gone. Thank you!
Looking at their source builds I see that their C++ Ice packages are build with -std=c++0x but the IcePy extension is not I wonder if this is causing the rtti crash.
If you want the Ubuntu packages to be fixed you will need to report the bug to Ubuntu package maintainers.
[X] Done [0]
Thank you again for your effort, xdm!
[0] https://bugs.launchpad.net/ubuntu/+source/zeroc-ice/+bug/1410265
I had similar errors on Debian Jessie with python 3.4.2 and zeroc-ice 3.5.1.
I also tried Debian testing and Ubuntu testing with the same result. Both those systems ran python 3.5.1 and and zeroc-ice 3.5.1.
The bug is indeed caused by different use of the "-std=c++0x" compiler flag during the build of the C++ Ice packages and the IcePy extension. But this is not a bug in the Debian/Ubuntu
build script. The bug is just triggered by the use of the build flag "CPP11=yes" in the Debian/Ubuntu build script.
The "CPP11=yes" build flag causes the compiler flag "-std=c++0x" to be included like it should be. But this compiler flag is reset during the build of the IcePy extension by the line
"CPPFLAGS =" in the second half of zeroc-ice-3.5.1/py/config/Make.rules.
The following patch solved the bug as far as the python extension is concerned:
--- zeroc-ice-3.5.1/py/config/Make.rules.original 2013-10-04 17:48:15.000000000 +0200
+++ zeroc-ice-3.5.1/py/config/Make.rules 2016-02-24 16:08:58.531547408 +0100
endif
CPPFLAGS =
+ifeq ($(CXX),g++)
+ ifeq ($(CPP11), yes)
+ CPPFLAGS = -std=c++0x
+ endif
+endif
+
ICECPPFLAGS = -I$(slicedir)
SLICE2PYFLAGS = $(ICECPPFLAGS)
LDFLAGS = $(LDPLATFORMFLAGS) $(CXXFLAGS) -L$(libdir)
We don't use zeroc-ice with ruby or PHP so we didn't test the ruby and PHP extentions. But looking at the other Make.rules* files I expect similar errors with ruby and PHP.
Best regards,
Kees Theunissen