Compilation of mico head

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

_______________________________________________
Mico-devel mailing list
Mico-devel <at> lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mico-devel

Re: Compilation of mico head

Hi Dmitry,

could you be so kind and give a try to 
http://mico.org/snapshots/mico-snapshot-2012-07-18.tar.bz2 ? I've just 
tested this on Solaris 11 + GCC 4.7.1 both in 32bit and 64bit and 
everything went fine, i.e. no compilation failure.

Thanks,
Karel

On 07/18/12 11:37 AM, Dmitry Kachan wrote:
> Hi,
>
> I just tried to install mico head on my machine (OpenSUSE 11.4; kernel
> 2.6.37.6 ; gcc version 4.7.0)
>
> I've downloaded distributive*/mico-head-2011-01-18.tar.bz2/*
> (18-Jan-2011 18:18 7.5M) from http://mico.org/snapshots/
>
> I faced the following mistakes during "make" command execution:
>
> make[1]: Entering directory `/<path-to-mico>/mico-head-2011-01-18/orb'
> c++ -I../include -O2 -Wall -Wwrite-strings -fno-strict-aliasing
> -D_REENTRANT -DPIC -fPIC -c orb.cc -o orb.pic.o
> orb.cc: In member function 'CORBA::ORBInvokeRec*
> CORBA::ORB::get_invoke(CORBA::ORB::MsgId)':
> orb.cc:2397:29: error: redeclaration of 'std::map<unsigned int,
> CORBA::ORBInvokeRec*, std::less<unsigned int> >::iterator i'
> orb.cc:2392:38: error: 'std::map<long unsigned int,
> MICOMT::Locked<std::map<unsigned int, CORBA::ORBInvokeRec*,
> std::less<unsigned int> > >*>::iterator i' previously declared here
> orb.cc: In member function 'void
> CORBA::ORB::del_invoke(CORBA::ORB::MsgId)':
> orb.cc:2448:29: error: redeclaration of 'std::map<unsigned int,
> CORBA::ORBInvokeRec*, std::less<unsigned int> >::iterator i'
> orb.cc:2443:38: error: 'std::map<long unsigned int,
> MICOMT::Locked<std::map<unsigned int, CORBA::ORBInvokeRec*,
> std::less<unsigned int> > >*>::iterator i' previously declared here
> make[1]: *** [orb.pic.o] Error 1
> make[1]: Leaving directory `/<path-to-mico>/mico-head-2011-01-18/orb'
> make: *** [system] Error 1
>
> Here are the parts of code where the mistakes occurs (file orb.cc):
>
> 2392 for (ThreadIDInvokeMap::iterator i = _invokes.begin();
> 2393 i != _invokes.end();
> 2394 i++) {
> 2395 Locked<InvokeMap>* map = (*i).second;
> 2396 AutoLock l2(*map);
> 2397 InvokeMap::iterator i = map->find(id);
> 2398 if (i != map->end()) {
> 2399 rec = (*i).second;
> 2400 return rec;
> 2401 }
> 2402 }
>
>
> and
>
> 2443 for (ThreadIDInvokeMap::iterator i = _invokes.begin();
> 2444 i != _invokes.end();
> 2445 i++) {
> 2446 Locked<InvokeMap>* map = (*i).second;
> 2447 AutoLock l2(*map);
> 2448 InvokeMap::iterator i = map->find(id);
> 2449 if (i != map->end()) {
> 2450 delete (*i).second;
> 2451 map->erase(i);
> 2452 return;
> 2453 }
> 2454 }
>
> Does anybody know how it should works? two same-name iterators?
>
> To avoid these mistakes I've changed the name of iterators (second
> declaration and further in both cases), so, I have:
>
> 2392 for (ThreadIDInvokeMap::iterator i = _invokes.begin();
> 2393 i != _invokes.end();
> 2394 i++) {
> 2395 Locked<InvokeMap>* map = (*i).second;
> 2396 AutoLock l2(*map);
> 2397 InvokeMap::iterator k = map->find(id);
> 2398 if (k != map->end()) {
> 2399 rec = (*k).second;
> 2400 return rec;
> 2401 }
> 2402 }
>
>
> and
>
> 2443 for (ThreadIDInvokeMap::iterator i = _invokes.begin();
> 2444 i != _invokes.end();
> 2445 i++) {
> 2446 Locked<InvokeMap>* map = (*i).second;
> 2447 AutoLock l2(*map);
> 2448 InvokeMap::iterator k = map->find(id);
> 2449 if (k != map->end()) {
> 2450 delete (*k).second;
> 2451 map->erase(k);
> 2452 return;
> 2453 }
> 2454 }
>
>
> These changes allowed me to go further with compilation of mico head.
>
>
>
> Than I was forced to set the -fpermissive flag to compile a few files
> because of mistakes like thise one:
>
> c++ -I../include -O2 -Wall -Wwrite-strings -fno-strict-aliasing
> -D_REENTRANT -D_GNU_SOURCE -DPIC -fPIC -c poa_impl.cc -o
> poa_impl.pic.o
> poa_impl.cc: In constructor 'MICOPOA::POA_impl::POA_impl(const
> char*, PortableServer::POAManager_ptr, const PolicyList&,
> MICOPOA::POA_impl*, CORBA::ORB_ptr)':
> poa_impl.cc:1955:80: error: invalid user-defined conversion from
> 'const ObjVar<CORBA::Policy>' to 'CORBA::Object*' [-fpermissive]
> In file included from ../include/CORBA.h:153:0,
> from poa_impl.cc:38:
> ../include/mico/template.h:96:3: note: candidate is:
> ObjVar<T>::operator T*() [with T = CORBA::Policy] <near match>
> ../include/mico/template.h:96:3: note: no known conversion for
> implicit 'this' parameter from 'const ObjVar<CORBA::Policy>*' to
> 'ObjVar<CORBA::Policy>*'
> poa_impl.cc:1955:80: error: passing 'const ObjVar<CORBA::Policy>' as
> 'this' argument of 'ObjVar<T>::operator T*() [with T =
> CORBA::Policy]' discards qualifiers [-fpermissive]
> make[1]: *** [poa_impl.pic.o] Error 1
> make[1]: Leaving directory `/<path-to-mico>/mico-head-2011-01-18/orb'
> make: *** [system] Error 1
>
>
> is it okay? from my perspective this flag could make a unexpected
> changes in workflow.
>
>
> Anyway, after all these manipulations I've compile and install mico head.
>
>
> with new mico head my project compiles perfectly, however during the
> execution the project shows me:
>
> my_projrct: pthread_mutex_lock.c:62: __pthread_mutex_lock: Assertion
> `mutex->__data.__owner == 0' failed.
> Aborted
>
>
>
> (this project worked perfectly with usual mico 2.3.13 and gcc v 4.6;
> however I was forced to use gcc v 4.7, and the project started show me
> segmentation fault on the mico functions during project execution.)
>
> Does anybody knows how can I fix it?
>
> Best regards
>
> -Dmitry Kachan.
>
>
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
>
>
>
> _______________________________________________
> Mico-devel mailing list
> Mico-devel <at> lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/mico-devel

--

-- 
Karel Gardas                  kgardas <at> objectsecurity.com
ObjectSecurity Ltd.           http://www.objectsecurity.com

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/