NetXMS Forum

szll2010

Dear Victor,

There is a bug in the userdb_objects.cpp in trunk. The setpassword function does not consider the Unicode and MBCS. However, I changed the code to this:
void User::setPassword(const TCHAR *password, bool clearChangePasswdFlag)
{
#ifdef UNICODE
    char *mb = MBStringFromWideString(password);
    CalculateSHA1Hash((BYTE *)mb, strlen(mb), m_passwordHash);
    free(mb);
#else
    CalculateSHA1Hash((BYTE *)password, _tcslen(password), m_passwordHash);
#endif
    //CalculateSHA1Hash((BYTE *)password, _tcslen(password), m_passwordHash);
    m_graceLogins = MAX_GRACE_LOGINS;
    m_flags |= UF_MODIFIED;
    if (clearChangePasswdFlag)
        m_flags &= ~UF_CHANGE_PASSWORD;
}
Still has the problem with setting or changing the password.

Could you check?

best regards,

szll2010

szll2010

I've got it.

Thanks very much.

Best regards,

Szll2010

szll2010

Hi Victor,

The thread process code of server module need to delete the message object, but in main thread the message object will be deleted too. So could you consider a solution:
1 copy the the message pointer(msg) to another pointer
2 set the pointer (msg) to NULL
3 main thread delete msg(NULL)
4 server module is responsible to delete the msg (delete data->msg) in the pfHandler

Best regards,

szll2010

Hi Victor,

The DBAsyncSelect is no problem without using the StartCommandHandler to start new thread to process the command. It's very strange. The code is very simple, just a dbgprintf() and use your skeleton of module, crash with thread and ok with normal process.

BTW, the server code is the trunk's snapshot.

Best regards,

szll2010

Hi Victor

Can I use DBAsyncSelect function in module? I use it in my server module, it makes server crash.

Pls advise.

Thanks.

szll2010

Hi Victor,

Another question, how to use the db function in the module? I mean there are globals, like g_nDBSyntax and
g_hCoreDB . I want to add an audit log module the system using the module?

BTW, the new feature is not previewed in windows console, is the windows console deprecated?

szll2010

Hi Victor,

I do send this port to the community. For the reason of windows port, I erased some unix codes in the source and I trim the nxagentd.cpp to a simpler version. Two subagents were ported , WINNT and WINPERF.

How to send the files to you?

best regards,

szll2010

Hi Victor,

Sorry for mistake in the Config::print test using the setvbuf function to output. It should be OK.

Best regards,

szll2010

Hi Victor,

I tested the new config API again but it does not work well with east asian characters.

The program is Unicode version, and the config's coding is UTF-8 and ANSI, in the config file there is some east asian characters. In both scenarios, the east asian characters became ?? under Config::print , the others are right.

Thanks

szll2010

Hi Victor,

Thanks for your port. Now the file system function works well. I port the all the core agent and WINNT and winperf subagent to WINDOWS UNICODE except the hddinfo has a bug. Not really understand the StrStrip function.

I also test the new config relative function, it can read ANSI and UNICODE content in the config file, but using the config->print function, it has problem with MBCS, like east asian characters. Could you check?

szll2010

Hi Victor,

Almost port the nxagentd to UNICODE for windows based the trunk version. Whether it is possible to port the dir relative structure members to TCHAR and dir.c to support TCHAR?

szll2010

Hi Victor

Thanks. I noticed that some header files had been modified and I will try after the 1.1.0 is released.

And I want to know how to deal with the IP address changing problem. For example, some devices are in remote site and connected through carrier network. The node's IP address is not the same as last time connected. Does the agent register and server recognizes the agent then change the node's ip address, or there is other solution?

szll2010

Hi Victor,

I try to manage some WIN CE devices. Not sure that MBCS is no problem. Could you advise?

szll2010

Hi Victor,

How to construct the method in the server module to process the Agent's originated CMD(if custom define)? I read the source code and not found in the AgentConnection Class. Has it been put in the m_pMsgWaitQueue for processing by Session, for example CMD_GET_MY CONFIG.

szll2010

Hi Victor,

Thanks for your quick action.

Is there any chance to port Agent to Unicode?

szll2010

Hi Victor,

Is there any sample code for the server module development like the agent's skeleton for reference.

szll2010

I patched the message.c file for the two functions, but still need you check the release of pStr:
void *CSCPMessage::Set(DWORD dwVarId, BYTE bType, const void *pValue, DWORD dwSize)
{
DWORD dwIndex, dwLength, dwLenWC;
CSCP_DF *pVar;
#if !defined(UNICODE_UCS2) || !defined(UNICODE)
UCS2CHAR *pBuffer;
#endif

// Create CSCP_DF structure
switch(bType)
{
case CSCP_DT_INTEGER:
pVar = (CSCP_DF *)malloc(12);
pVar->df_int32 = *((const DWORD *)pValue);
break;
case CSCP_DT_INT16:
pVar = (CSCP_DF *)malloc(8);
pVar->df_int16 = *((const WORD *)pValue);
break;
case CSCP_DT_INT64:
pVar = (CSCP_DF *)malloc(16);
pVar->df_int64 = *((const QWORD *)pValue);
break;
case CSCP_DT_FLOAT:
pVar = (CSCP_DF *)malloc(16);
pVar->df_real = *((const double *)pValue);
break;
case CSCP_DT_STRING:
dwLength = (DWORD)_tcslen((const TCHAR *)pValue);
//pVar = (CSCP_DF *)malloc(12 + dwLength * 2);
//pVar->df_string.dwLen = dwLength * 2;
#ifdef UNICODE
#ifdef UNICODE_UCS2
         pVar = (CSCP_DF *)malloc(12 + dwLength * 2);
pVar->df_string.dwLen = dwLength * 2;
memcpy(pVar->df_string.szValue, pValue, pVar->df_string.dwLen);
#else /* assume UNICODE_UCS4 */
         pVar = (CSCP_DF *)malloc(12 + dwLength * 2);
pVar->df_string.dwLen = dwLength * 2;
pBuffer = (UCS2CHAR *)malloc(dwLength * 2 + 2);
ucs4_to_ucs2((WCHAR *)pValue, dwLength, pBuffer, dwLength + 1);
memcpy(pVar->df_string.szValue, pBuffer, pVar->df_string.dwLen);
free(pBuffer);
#endif
#else /* not UNICODE */
         dwLenWC = mb_to_ucs2((const char *)pValue, dwLength, NULL, 0);
pBuffer = (UCS2CHAR *)malloc(dwLenWC * 2 + 2);
mb_to_ucs2((const char *)pValue, dwLength, pBuffer, dwLenWC + 1);
         pVar = (CSCP_DF *)malloc(12 + dwLenWC * 2);
         pVar->df_string.dwLen = dwLenWC* 2;
memcpy(pVar->df_string.szValue, pBuffer, pVar->df_string.dwLen);
free(pBuffer);
#endif
break;
case CSCP_DT_BINARY:
pVar = (CSCP_DF *)malloc(12 + dwSize);
pVar->df_string.dwLen = dwSize;
if ((pVar->df_string.dwLen > 0) && (pValue != NULL))
memcpy(pVar->df_string.szValue, pValue, pVar->df_string.dwLen);
break;
default:
return NULL; // Invalid data type, unable to handle
}
pVar->dwVarId = dwVarId;
pVar->bType = bType;

// Check if variable exists
dwIndex = FindVariable(pVar->dwVarId);
if (dwIndex == INVALID_INDEX) // Add new variable to list
{
m_ppVarList = (CSCP_DF **)realloc(m_ppVarList, sizeof(CSCP_DF *) * (m_dwNumVar + 1));
m_ppVarList[m_dwNumVar] = pVar;
m_dwNumVar++;
}
else // Replace existing variable
{
free(m_ppVarList[dwIndex]);
m_ppVarList[dwIndex] = pVar;
}

return (bType == CSCP_DT_INT16) ? ((void *)((BYTE *)pVar + 6)) : ((void *)((BYTE *)pVar + 8));
}

TCHAR *CSCPMessage::GetVariableStr(DWORD dwVarId, TCHAR *pszBuffer, DWORD dwBufSize)
{
void *pValue;
TCHAR *pStr = NULL;
DWORD dwLen, dwLenMB;

if ((pszBuffer != NULL) && (dwBufSize == 0))
return NULL; // non-sense combination

pValue = Get(dwVarId, CSCP_DT_STRING);
if (pValue != NULL)
{
if (pszBuffer == NULL)
{
#if defined(UNICODE) && defined(UNICODE_UCS4)
pStr = (TCHAR *)malloc(*((DWORD *)pValue) * 2 + 4);
         dwLen = (pszBuffer == NULL) ? (*((DWORD *)pValue) / 2) : min(*((DWORD *)pValue) / 2, dwBufSize - 1);
      ucs2_to_ucs4((UCS2CHAR *)((BYTE *)pValue + 4), dwLen, pStr, dwLen + 1);
      pStr[dwLen] = 0;
#elif defined(UNICODE) && defined(UNICODE_UCS2)
pStr = (TCHAR *)malloc(*((DWORD *)pValue) + 2);
         dwLen = (pszBuffer == NULL) ? (*((DWORD *)pValue) / 2) : min(*((DWORD *)pValue) / 2, dwBufSize - 1);
memcpy(pStr, (BYTE *)pValue + 4, dwLen * 2);
      pStr[dwLen] = 0;
#else

         dwLenMB = ucs2_to_mb((UCS2CHAR *)((BYTE *)pValue + 4), *((DWORD *)pValue) / 2, NULL, 0);
         pStr = (TCHAR *)malloc(dwLenMB + 1);
         dwLen = (pszBuffer == NULL) ? dwLenMB : min(dwLenMB, dwBufSize - 1);
      ucs2_to_mb((UCS2CHAR *)((BYTE *)pValue + 4), *((DWORD *)pValue) / 2, pStr, dwLen + 1);
      pStr[dwLen] = 0;
#endif


      }
else
{
pStr = pszBuffer;
#if defined(UNICODE) && defined(UNICODE_UCS4)
      dwLen = (pszBuffer == NULL) ? (*((DWORD *)pValue) / 2) : min(*((DWORD *)pValue) / 2, dwBufSize - 1);
         ucs2_to_ucs4((UCS2CHAR *)((BYTE *)pValue + 4), dwLen, pStr, dwLen + 1);
      pStr[dwLen] = 0;
#elif defined(UNICODE) && defined(UNICODE_UCS2)
      dwLen = (pszBuffer == NULL) ? (*((DWORD *)pValue) / 2) : min(*((DWORD *)pValue) / 2, dwBufSize - 1);
memcpy(pStr, (BYTE *)pValue + 4, dwLen * 2);
      pStr[dwLen] = 0;
#else

         dwLenMB = ucs2_to_mb((UCS2CHAR *)((BYTE *)pValue + 4), *((DWORD *)pValue) / 2, NULL, 0);
         dwLen = (pszBuffer == NULL) ? dwLenMB : min(dwLenMB, dwBufSize - 1);
      ucs2_to_mb((UCS2CHAR *)((BYTE *)pValue + 4), *((DWORD *)pValue) / 2, pStr, dwLen + 1);
      pStr[dwLen] = 0;
#endif

}
}
else
{
if (pszBuffer != NULL)
{
pStr = pszBuffer;
pStr[0] = 0;
}
}
return pStr;
}

szll2010

I found problems with displaying east asian characters in NETxms console, the following is the environment:

Server on Windows (build with MBCS)
Console on Windows (Build with Unicode)

In this scenario, east asian character can be input, but after transfering to server, then the these characters was wrong in server's database. I read the source code, found that:

In message.c file, GetVariableStr and SetVariable function's allocating memory is assume the MBCS is Single byte for one character, so it allocates double or half buffer for the converting, but the fact is not.
In GetVariable Str, allocating memory for the pStr, but not found where to release the memory, chance to leak ?

Could you help to check the source file.

NetXMS Forum

szll2010

Posts