Ice Compilation

[kraps33]

Hi,

I have a big problem with IceE since last week.
IceE run in native compil but not in cross compil...

The problem seems provide of upCast function...

LocalException.o(.sbss+0x0):IceE-1.2.0/src/IceE/LocalException.cpp:676: multiple definition of `in'
DefaultsAndOverrides.o(.sbss+0x0):IceE-1.2.0/src/IceE/DefaultsAndOverrides.cpp:18: first defined here
Network.o(.sbss+0x0):/inst_gcc-3.4.6///include/c++/3.4.6/bits/stl_vector.h:256: multiple definition of `in'
DefaultsAndOverrides.o(.sbss+0x0): first defined here
Acceptor.o(.sbss+0x0):../../src/TcpTransport/Acceptor.cpp:22: multiple definition of `in'
DefaultsAndOverrides.o(.sbss+0x0): first defined here
Connector.o(.sbss+0x0):../../src/TcpTransport/Connector.cpp:22: multiple definition of `in'
DefaultsAndOverrides.o(.sbss+0x0): first defined here
TcpEndpoint.o(.sbss+0x0):../../include/IceE/BasicStream.h:604: multiple definition of `in'
DefaultsAndOverrides.o(.sbss+0x0): first defined here
Transceiver.o(.sbss+0x0):../../src/TcpTransport/Transceiver.cpp:23: multiple definition of `in'
DefaultsAndOverrides.o(.sbss+0x0): first defined here
make[2]: *** [../../lib/libIceE.so.1.2.0] Error 1

Thanks for your help.

[matthew]

Please don't post a duplicate thread on an existing issue. If you want our attention then please add information to the existing thread.

[bernard]

Hi Sebastian,

I don't see any variable named 'in' in this Ice-E source code.

If you modified the Ice-E code to support your own system, it's going to be difficult to help you! Did you change headers/includes/variable names as part of your port?

Could you also specify the exact platform you're porting to, including compiler version?

Cheers,
Bernard

[kraps33]

Hi,

I have just add 3 headers to define errno.h variables or functions like gettimeofday, pthread_mutexattr_settype...
And a library to implement this.

I just add in 5 files macros likes #ifdef _MYOS #include <...> #endif (Config.h, Time.cpp, Network.h, Instance.cpp, Network.cpp)

My headers are protected (#ifndef __MYTYPES_H #define ...) and they countain only functions declarations, structures or #define VARIABLES.

I have found this in Acceptor.s and DefaultsAndOverrides.s (for "in"):

.file "Acceptor.cpp"
.globl in
.section ".sbss","aw",@nobits
.align 2
in:
.zero 4
.size in, 4
.type in, @object
.section ".text"
.align 2
.globl _ZN11IceInternal6upCastEPNS_8AcceptorE
.type _ZN11IceInternal6upCastEPNS_8AcceptorE, @function
_ZN11IceInternal6upCastEPNS_8AcceptorE:


.file "DefaultsAndOverrides.cpp"
.globl in
.section ".sbss","aw",@nobits
.align 2
in:
.zero 4
.size in, 4
.type in, @object
.section ".text"
.align 2
.globl _ZN11IceInternal6upCastEPNS_20DefaultsAndOverrides E
.type _ZN11IceInternal6upCastEPNS_20DefaultsAndOverrides E, @function
_ZN11IceInternal6upCastEPNS_20DefaultsAndOverrides E:

For system. I work under POSIX operating system (it is the target system, with newlib 1.8) on PowerPC processor.

Cheers,
Sebastien.

[bernard]

It looks like one of your headers define this 'in' global variable; or perhaps it's a macro that somehow defines 'in' in the upCast function definition.

You could pre-process one of these files (gcc -E) and try to locate this 'in' definition in the output.

Cheers,
Bernard

[kraps33]

struct in_addr in;

In a header file of target system...

It run better with an "extern" .

Thanks for your help.