[使用DLL函数崩溃后C ++分配内存

Question

我最近发布了有关使用DLL函数的问题，并且自从我试图缩小问题的范围以来。我重新创建了代码部分，仅包括DLL部分，这似乎是问题所在。该库是一个64位的库，我正在使用Mingw64-g ++进行编译。下面的代码是加载dll，加载库函数，然后使用它们。使用库函数可以正常工作，但是在我有机会做任何事情之前，为随后的调用分配内存失败（崩溃）。这是我当前拥有的代码。

#include <iostream>
#include <fstream>
#include <windows.h>
using namespace std;

typedef int WINAPI CompressFunc(uint32_t codec, char *src_buf, int64_t src_len, char *dst_buf, int64_t level,
            void *opts, int64_t offs);

typedef int64_t WINAPI DecompressFunc(
            char *src_buf, int64_t src_len, char *dst_buf, int64_t dst_size, int fuzz, int crc, int verbose,
            uint8_t *dst_base, size_t e, void *cb, void *cb_ctx, void *scratch, size_t scratch_size, int threadPhase
        );


int main()
{
    string dll_path = "oo2core_5_win64.dll";
    HINSTANCE dll_object = LoadLibraryA(dll_path.c_str());
    cout<<GetLastError()<<endl;
    string compress_function_name = "OodleLZ_Compress";
    string decompress_function_name = "OodleLZ_Decompress";

    CompressFunc* CompressFunc_Func;
    DecompressFunc* DecompressFunc_Func;

    CompressFunc_Func = (CompressFunc*)GetProcAddress(dll_object, compress_function_name.c_str());
    DecompressFunc_Func = (DecompressFunc*)GetProcAddress(dll_object, decompress_function_name.c_str());
    cout<<bool(CompressFunc_Func)<<" "<<bool(DecompressFunc_Func)<<endl;

    ifstream fin("0.decompressed", ios::binary);
    cout<<fin.good()<<endl; // Prints 1
    fin.seekg(0, ios::end);
    uint64_t decompressed_size = fin.tellg(); 
    cout<<decompressed_size<<endl; //Prints 558, correct
    fin.seekg(0);
    cout<<fin.tellg()<<endl; // Prints 0
    cout.flush();
    char* decompressed_data = new char [decompressed_size]; 
    fin.read(decompressed_data, decompressed_size);
    cout<<"Read Data"<<endl;
    cout.flush();

    char* compressed_data;
    compressed_data = new char [decompressed_size + 0x10000]; //Upper Bound
    int64_t compressed_size = CompressFunc_Func(7, decompressed_data, decompressed_size, compressed_data, 7, 0, 0);
    cout<<"Compressed Successfully"<<endl;
    cout<<"Compressed Size: "<<compressed_size<<endl; // Prints 225
    cout.flush();

    for (int i = 0; i < 4; i++)
    {
        cout<<hex<<(uint16_t)compressed_data[i]<<" ";
    }
    cout<<dec<<endl;

    delete [] decompressed_data;

    cout<<"Still Deco Size: "<<decompressed_size<<endl; //Prints 558, correct

    char* x = new char [decompressed_size]; // Freezes here then stops

    //DecompressFunc_Func(compressed_data, compressed_size, decompressed_data, decompressed_size, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);



}

输出：

0
1 1
1
558
0
Read Data

当我在底部评论内存分配时，输出：

0
1 1
1
558
0
Read Data
Compressed Successfully
Compressed Size: 225
ff8c 5 40 ffdc
Still Deco Size: 558

仅在此后才存在内存分配的情况下，运行GDB调试器才会在Compress函数上产生分段错误。

Answer 1

我拥有的文档与我正在使用的DLL的实际版本之间不匹配。我在IDA Pro中打开DLL，发现我缺少“压缩函数”调用的一个额外参数。

添加参数后，我发现还有2个附加参数，因此该函数最终需要使用int64_t ununsed, void* scratch, int64_t scratch_size。