Re: [boost] portable precise garbage collector for C++

From: Achilleas Margaritis (axilmar_at_[hidden])
Date: 2007-09-19 14:37:00

O/H Larry Evans έγραψε:
> On 09/19/07 10:49, Achilleas Margaritis wrote:
> [snip]
>>>> I am downloading Fedora Core and I will install it under VMWare, so as 
>>>> that I can compile the code for Linux as well.
> I'm getting a segmentation fault:
> 
> gcc.link ../../bin.v2/sandbox/axilmar/gcc-4.1/debug/main_test
> Jamfile</home/evansl/prog_dev/boost-svn/ro/trunk/sandbox/axilmar>.capture-output 
> ../../bin.v2/sandbox/axilmar/gcc-4.1/debug/main_test.output
> ***running  ../../bin.v2/sandbox/axilmar/gcc-4.1/debug/main_test
> /bin/sh: line 2: 14060 Segmentation fault 
> ../../bin.v2/sandbox/axilmar/gcc-4.1/debug/main_test
> ...updated 4 targets...
> 
> I added some debug prints to the _page::new:
> 
> //allocate page from mspaces
> void *_page::operator new(size_t size) {
>      std::cout<<"_page::operator new at "<<__FILE__<<":"<<__LINE__<<"\n";
>      void*p=
>      #if 1
>        _CPPGC_MALLOC(gc::_mem_space, size);
>      #else
>        0;
>      #endif
>      std::cout<<"_page::operator new at p="
>      //<<p
>      <<"\n";
>      return p;
> }
> 
> When the '#if 1' is changed to '#if 0', I get:
> 
> ***running  ../../bin.v2/sandbox/axilmar/gcc-4.1/debug/main_test
> _page::operator new at ./cppgc/include/cppgc.hpp:648
> _page::operator new at p=
> running main_test.cpp
> p=0x8456008
> a_page=0
> ...updated 4 targets...
> 
> Compilation finished at Wed Sep 19 12:41:55
> 
> main_test only contains:
> 
> #include "cppgc.hpp"
> #include <iostream>
> int main(void)
> {
>      void*p=dlmalloc(20);
>      void*a_page=_page::operator new(sizeof(_page));
>      std::cout<<"running "<<__FILE__<<"\n";
>      std::cout<<"p="<<p<<"\n";
>      std::cout<<"a_page="<<a_page<<"\n";
>      return 0;
> }
> 

Should I remove dlmalloc temporarily? I need mspaces so as that when the 
code becomes threaded, each thread will have its own mspace.