Microsoft Outlook email - Memory leak bug in inetcomm!CreateNewsHeader?

Outlook Banter (http://www.outlookbanter.com/)

- Outlook Express (http://www.outlookbanter.com/outlook-express/)

- - Memory leak bug in inetcomm!CreateNewsHeader? (http://www.outlookbanter.com/outlook-express/89843-memory-leak-bug-inetcomm-createnewsheader.html)

oen	May 3rd 09 05:56 AM

Memory leak bug in inetcomm!CreateNewsHeader?

OE uses inetcomm.dll(6.0.2900.5512) to create the news header, when I tried to interpret the disassemble codes as C codes, I found that INETCOMM!CreateNewsHeader calls INETCOMM!EscapePound to MemAlloc a LPWSTR (if needed), and uses the returned LPWSTR in eax, but at last, does not call ole32!CRetailMalloc_Free to free the LPWSTR! Is this a Memory leak bug?

HRESULT _stdcall CreateNewsHeader(IMimeMessageTree *pMTree, DWORD ReplyFlags, IVirtualStream *pVStream) {
DWORD countByte; //used as esi
WCHAR *pMessageIDEscapePound; //used as eax
WCHAR *pFormtString = 0; //ebp-404h
HRESULT hr; //ebp-408h
WCHAR *pAddress = 0; //ebp-40Ch
WCHAR *pMessageID = 0; //ebp-410h
WCHAR *pMessageIDDropAngles = 0; //ebp-414h
WCHAR *pNewsHeader = 0; //ebp-418h
WCHAR *pNewsHeaderHtml; //ebp-420h

pMTree-GetAddressFormatW(IAT_FROM, AFT_DISPLAY_BOTH, &pAddress);
MimeOleGetBodyPropW(pMTree, HBODY_ROOT, PID_HDR_MESSAGEID, 0, &pMessageID);
countByte = 2 * LoadStringWrapW(hModuleInetresDll, 1167, &pFormtString, 512);
if (pAddress) {
countByte += 2 * lstrlenW(pAddress);
else
pAddress = g_wszEmpty;
if (pMessageID) {
if (pMessageIDEscapePound = EscapePound(pMessageID)) {
DropAngles(pMessageIDEscapePound, &pMessageIDDropAngles);
else
DropAngles(pMessageID, &pMessageIDDropAngles);
countByte += 2 * lstrlenW(pMessageIDDropAngles);
} else
pMessageIDDropAngles = g_wszEmpty;
countByte += 2;
if (S_OK != MemAlloc(&pNewsHeader, countByte)) {
hr = MAKE_SCODE(SEVERITY_ERROR, FACILITY_WIN32, ERROR_OUTOFMEMORY);
goto epilog;
}
if (S_OK != (hr = pVStream-Write(L"DIV", 2 * lstrlenW(L"DIV"), NULL)))
goto epilog;
wnsprintfW(pNewsHeader, countByte, pFormtString, pAddress, pMessageIDDropAngles);
if (S_OK != (hr = EscapeStringToHTML(pNewsHeader, &pNewsHeaderHtml)))
goto epilog;
pVStream-Write(pNewsHeaderHtml, 2 * lstrlenW(pNewsHeaderHtml), NULL); //why not check hr?
RetailMalloc_Free(pNewsHeaderHtml);
hr = pVStream-Write(L"/DIV", 2 * lstrlenW(L"/DIV"), NULL);
epilog:
if (pAddress != g_wszEmpty)
RetailMalloc_Free(pAddress);
RetailMalloc_Free(pMessageID);
RetailMalloc_Free(pNewsHeader);
//no RetailMalloc_Free(pMessageIDEscapePound)!
return hr;
}

PA Bear [MS MVP]

May 3rd 09 06:30 AM

Memory leak bug in inetcomm!CreateNewsHeader?

cf. http://groups.google.com/group/micro...a7703d6886547a

Windows Mail (formerly Outlook Express)Documentation
http://msdn.microsoft.com/en-us/library/ms709546.aspx
--
~PA Bear

"oen" wrote in message ...
OE uses inetcomm.dll(6.0.2900.5512) to create the news header, when I tried to interpret the disassemble codes as C codes, I found that INETCOMM!CreateNewsHeader calls INETCOMM!EscapePound to MemAlloc a LPWSTR (if needed), and uses the returned LPWSTR in eax, but at last, does not call ole32!CRetailMalloc_Free to free the LPWSTR! Is this a Memory leak bug?

HRESULT _stdcall CreateNewsHeader(IMimeMessageTree *pMTree, DWORD ReplyFlags, IVirtualStream *pVStream) {
DWORD countByte; //used as esi
WCHAR *pMessageIDEscapePound; //used as eax
WCHAR *pFormtString = 0; //ebp-404h
HRESULT hr; //ebp-408h
WCHAR *pAddress = 0; //ebp-40Ch
WCHAR *pMessageID = 0; //ebp-410h
WCHAR *pMessageIDDropAngles = 0; //ebp-414h
WCHAR *pNewsHeader = 0; //ebp-418h
WCHAR *pNewsHeaderHtml; //ebp-420h

pMTree-GetAddressFormatW(IAT_FROM, AFT_DISPLAY_BOTH, &pAddress);
MimeOleGetBodyPropW(pMTree, HBODY_ROOT, PID_HDR_MESSAGEID, 0, &pMessageID);
countByte = 2 * LoadStringWrapW(hModuleInetresDll, 1167, &pFormtString, 512);
if (pAddress) {
countByte += 2 * lstrlenW(pAddress);
else
pAddress = g_wszEmpty;
if (pMessageID) {
if (pMessageIDEscapePound = EscapePound(pMessageID)) {
DropAngles(pMessageIDEscapePound, &pMessageIDDropAngles);
else
DropAngles(pMessageID, &pMessageIDDropAngles);
countByte += 2 * lstrlenW(pMessageIDDropAngles);
} else
pMessageIDDropAngles = g_wszEmpty;
countByte += 2;
if (S_OK != MemAlloc(&pNewsHeader, countByte)) {
hr = MAKE_SCODE(SEVERITY_ERROR, FACILITY_WIN32, ERROR_OUTOFMEMORY);
goto epilog;
}
if (S_OK != (hr = pVStream-Write(L"DIV", 2 * lstrlenW(L"DIV"), NULL)))
goto epilog;
wnsprintfW(pNewsHeader, countByte, pFormtString, pAddress, pMessageIDDropAngles);
if (S_OK != (hr = EscapeStringToHTML(pNewsHeader, &pNewsHeaderHtml)))
goto epilog;
pVStream-Write(pNewsHeaderHtml, 2 * lstrlenW(pNewsHeaderHtml), NULL); //why not check hr?
RetailMalloc_Free(pNewsHeaderHtml);
hr = pVStream-Write(L"/DIV", 2 * lstrlenW(L"/DIV"), NULL);
epilog:
if (pAddress != g_wszEmpty)
RetailMalloc_Free(pAddress);
RetailMalloc_Free(pMessageID);
RetailMalloc_Free(pNewsHeader);
//no RetailMalloc_Free(pMessageIDEscapePound)!
return hr;
}

PA Bear [MS MVP]

May 3rd 09 06:30 AM

Memory leak bug in inetcomm!CreateNewsHeader?

Steve Cochran

May 13th 09 02:24 PM

Memory leak bug in inetcomm!CreateNewsHeader?

No telling. That function is completely undocumented and most at MS don't
understand the OE code anyway, so you won't get much help here.

cheers,

steve

"oen" wrote in message ...
OE uses inetcomm.dll(6.0.2900.5512) to create the news header, when I tried
to interpret the disassemble codes as C codes, I found that
INETCOMM!CreateNewsHeader calls INETCOMM!EscapePound to MemAlloc a LPWSTR
(if needed), and uses the returned LPWSTR in eax, but at last, does not call
ole32!CRetailMalloc_Free to free the LPWSTR! Is this a Memory leak bug?

HRESULT _stdcall CreateNewsHeader(IMimeMessageTree *pMTree, DWORD
ReplyFlags, IVirtualStream *pVStream) {
DWORD countByte; //used as esi
WCHAR *pMessageIDEscapePound; //used as eax
WCHAR *pFormtString = 0; //ebp-404h
HRESULT hr; //ebp-408h
WCHAR *pAddress = 0; //ebp-40Ch
WCHAR *pMessageID = 0; //ebp-410h
WCHAR *pMessageIDDropAngles = 0; //ebp-414h
WCHAR *pNewsHeader = 0; //ebp-418h
WCHAR *pNewsHeaderHtml; //ebp-420h

pMTree-GetAddressFormatW(IAT_FROM, AFT_DISPLAY_BOTH, &pAddress);
MimeOleGetBodyPropW(pMTree, HBODY_ROOT, PID_HDR_MESSAGEID, 0,
&pMessageID);
countByte = 2 * LoadStringWrapW(hModuleInetresDll, 1167, &pFormtString,
512);
if (pAddress) {
countByte += 2 * lstrlenW(pAddress);
else
pAddress = g_wszEmpty;
if (pMessageID) {
if (pMessageIDEscapePound = EscapePound(pMessageID)) {
DropAngles(pMessageIDEscapePound, &pMessageIDDropAngles);
else
DropAngles(pMessageID, &pMessageIDDropAngles);
countByte += 2 * lstrlenW(pMessageIDDropAngles);
} else
pMessageIDDropAngles = g_wszEmpty;
countByte += 2;
if (S_OK != MemAlloc(&pNewsHeader, countByte)) {
hr = MAKE_SCODE(SEVERITY_ERROR, FACILITY_WIN32, ERROR_OUTOFMEMORY);
goto epilog;
}
if (S_OK != (hr = pVStream-Write(L"DIV", 2 * lstrlenW(L"DIV"),
NULL)))
goto epilog;
wnsprintfW(pNewsHeader, countByte, pFormtString, pAddress,
pMessageIDDropAngles);
if (S_OK != (hr = EscapeStringToHTML(pNewsHeader, &pNewsHeaderHtml)))
goto epilog;
pVStream-Write(pNewsHeaderHtml, 2 * lstrlenW(pNewsHeaderHtml), NULL);
//why not check hr?
RetailMalloc_Free(pNewsHeaderHtml);
hr = pVStream-Write(L"/DIV", 2 * lstrlenW(L"/DIV"), NULL);
epilog:
if (pAddress != g_wszEmpty)
RetailMalloc_Free(pAddress);
RetailMalloc_Free(pMessageID);
RetailMalloc_Free(pNewsHeader);
//no RetailMalloc_Free(pMessageIDEscapePound)!
return hr;
}

Steve Cochran

May 13th 09 02:24 PM

Memory leak bug in inetcomm!CreateNewsHeader?

All times are GMT +1. The time now is 11:14 AM.