1/1 
|
compiling C++
by ag09 on Aug 24, 2011 |
ag09
Posts: 38 Joined: Aug 22, 2010 Last seen: Jan 13, 2012 |
||
|
Hi all,
in the generated toolchain we the compiler or32-elf-g++ hat been generated I am trying to compile a simple HelloWord implemented in class data structure. The application hat been compiled well but I got the following error while executing. are there any constrictions to compile C++ code? ./meta-test: cannot execute binary file
//helloworld.cpp
==================
#include
#include
using namespace std;
class test {
public:
test();
~test();
void print();
};
test::test(){}
test::~test(){}
void test::print(){
cout
|
|||
|
RE: compiling C++
by olof on Aug 24, 2011 |
olof
Posts: 218 Joined: Feb 10, 2010 Last seen: Dec 17, 2018 |
||
|
Are you running it under or1ksim or on a real FPGA board?
|
|||
|
RE: compiling C++
by ag09 on Aug 24, 2011 |
ag09
Posts: 38 Joined: Aug 22, 2010 Last seen: Jan 13, 2012 |
||
|
Are you running it under or1ksim or on a real FPGA board?
On Or1ksim |
|||
|
RE: compiling C++
by jeremybennett on Aug 24, 2011 |
jeremybennett
Posts: 815 Joined: May 29, 2008 Last seen: Jun 13, 2019 |
||
|
Hi ag09, Can you post the commands you used to compile the program and the Or1ksim command line you used to run it, together with the full error message output. That doesn't look like an Or1ksim error message. BTW, C++ for OR1K currently only supports statically linked libraries. Many C++ programs assume shared object libraries and dynamic loading are available. HTH Jeremy
-- |
|||
|
RE: compiling C++
by ag09 on Aug 24, 2011 |
ag09
Posts: 38 Joined: Aug 22, 2010 Last seen: Jan 13, 2012 |
||
|
Hi ag09, Can you post the commands you used to compile the program and the Or1ksim command line you used to run it, together with the full error message output. That doesn't look like an Or1ksim error message. Sorry, I checked it again yet and it works fine. Apologize!! BTW, C++ for OR1K currently only supports statically linked libraries. Many C++ programs assume shared object libraries and dynamic loading are available. Which efforts can be needed to allow shared object libraries and dynamic loading to be linking on or1k? Thanks for your support. |
|||
|
RE: compiling C++
by jeremybennett on Aug 25, 2011 |
jeremybennett
Posts: 815 Joined: May 29, 2008 Last seen: Jun 13, 2019 |
||
|
Which efforts can be needed to allow shared object libraries and dynamic loading to be linking on or1k? Hi ag09, Here are the major things:
I have a full project plan worked out to do this. The time taken is completely dominated by testing. One of the past problems with the OpenRISC tool chain is that it has been implemented, but not properly tested. There has also been a substantial amount of code that does not follow the GNU design and coding conventions, making for tools that break badly as you try to migrate them to new versions. It is also code that could never be committed to the mainline GNU tool chain distribution. Over the past year or so, Joern Rennecke and I worked on GCC for C/C++ with static libraries, so it is now reasonably reliable and complies with the GNU design conventions, although the static pthreads library gets some things wrong. That work has revealed how very broken the binutils implementation is, although we have fixed some of the most egregious errors. The uClibc implementation, which is used for all applications running on Linux is to all intents and purposes untested. The only tests which have been run are those which are consequent on running the GNU regression tests (which allowed us to fix some bugs). uClibc has a regression test suite, but until recently it was only capable of running with dynamic libraries and natively. We've fixed the test suite so it can handle static libraries. However we need to write a DejaGnu harness that will allow the tests to be run against a remote embedded target. So a key OpenRISC objective for the future is that we only allow properly tested tools into the code base. It is not terribly useful to have the latest feature, if it brings with it several thousand regression test failures. Hope this explains what is needed. Jeremy |
|||
1/1