将指针编组为结构异常的指针数组，“无法读取或写入受保护的内存”

Question

我正在将本机C ++库转换为C＃。原始函数分配结构的一些内存，对当前进程进行一些内存操作，然后将操作后的字节返回到结构内部的字节缓冲区中。

该结构定义如下：

typedef struct tdMEM_SCATTER_HEADER {
    DWORD magic;            
    WORD version;           
    WORD Future1;
    ULONG64 qwA;            
    DWORD cbMax;            
    DWORD cb;               
    PBYTE pb;             
    PVOID pvReserved1;      
    PVOID pvReserved2;      
    PVOID Future2[8];
} MEM_SCATTER_HEADER, *PMEM_SCATTER_HEADER, **PPMEM_SCATTER_HEADER;

原始功能定义如下：

DWORD Scatter(_Inout_ PPMEM_SCATTER_HEADER ppMEMs, _In_ DWORD count);

分配顺序为

pbBuffer = LocalAlloc(0, count * (sizeof(PMEM_SCATTER_HEADER) + sizeof(MEM_SCATTER_HEADER) + 0x1000));
ZeroMemory(pbBuffer, count * (sizeof(PMEM_SCATTER_HEADER) + sizeof(MEM_SCATTER_HEADER)));
for(i = 0; i < cMEMs; i++) {
     pMEM->pb = pbData + ((QWORD)i << 12); // Allocating return bytes buffer.
}

这是我尝试过的：

结构编组：

[StructLayout(LayoutKind.Sequential, CharSet = CharSet.Auto)]
public struct MEM_SCATTER_HEADER
{
     public uint magic;            
     public ushort version;        
     public ushort Future1;
     public ulong qwA;            
     public uint cbMax;            
     public uint cb;                
     [MarshalAs(UnmanagedType.ByValArray)]
     public byte[] pb;             
     public IntPtr pvReserved1;     
     public IntPtr pvReserved2;                
     [MarshalAs(UnmanagedType.ByValArray, SizeConst = 8)]
     public IntPtr[] Future2;
}

内存分配：

MEM_SCATTER_HEADER[] header = new MEM_SCATTER_HEADER[count];

for (int i = 0; i < count; i++)
{
     header[i].pb = new byte[0x1000];
     Array.Clear(header[i].pb, 0, header[i].pb.Length); // Fill array with 0
}

DLLImport：

[DllImport("testC.dll", CharSet = CharSet.Auto, SetLastError = true)]
public static extern uint Scatter(ref MEM_SCATTER_HEADER ppMems, uint count);

最后是函数调用：

uint count = Scatter(ref header, count);

我花了很多时间调试错误地封送数据，或者必须进行一些手动封送处理才能消除异常：“试图读取或写入受保护的内存。这通常表明其他内存已损坏。” “在函数调用时。

Answer 1

初始分配未将其分配的所有内存清零，并且看起来不正确。就像代码正在做的一样：

int bigchunk = sizeof(PMEM_SCATTER_HEADER) + sizeof(MEM_SCATTER_HEADER) + 0x1000;
int smallchunk = sizeof(PMEM_SCATTER_HEADER) + sizeof(MEM_SCATTER_HEADER);

pBuffer = LocalAlloc(0, count * bigchunk);
ZeroMemory(pBuffer, count * smallchunk);

这可能吗？