September 2004 - Wine-devel

Re: SystemFunction036 debugging problem
by James Hawkins 28 Sep '04

28 Sep '04

> Is this program linking to SystemFunction036 statically or dynamically? I'm not really sure. Is there another way to tell besides doing a trace+all? > If it is linking dynamically (i.e. with GetProcAddresa), one option to try > is to completly remove SystemFunction036 so it is not exported from > advapi32.dll. It might then be that the apps that call it will see that it > is not present and do something else. Actually I just added the stub to advapi32, because the program will crash when it tries to call an umimplemented function. The main problem is that the program is expecting some result from SystemFunction036 (Filip thinks it might be a random generator thing), but it's next to impossible (or so it seems) to implement it correctly without documentation of what it does. I wanted to be able to see which function calls this to see whether I can get in there and keep it from calling. On Tue, 28 Sep 2004 07:01:33 +0800, Jonathan Wilson <jonwil(a)tpgi.com.au> wrote: > Is this program linking to SystemFunction036 statically or dynamically? > > If it is linking dynamically (i.e. with GetProcAddresa), one option to try > is to completly remove SystemFunction036 so it is not exported from > advapi32.dll. It might then be that the apps that call it will see that it > is not present and do something else. > > Another option is to try to contact the author of the app in question > (MyPhotoCalendars or whatever it is) and ask them about the function. > > -- James Hawkins

2 1

SystemFunction036 debugging problem
by James Hawkins 28 Sep '04

28 Sep '04

It turns out that SystemFunction036 is actually required to do something for MyPhotoCalendars.exe to install correctly because the installer exits with a program-level exception (ie not wine) What I would like to do is run MyPhot...exe with winedbg and then set a breakpoint for SystemFunction036 so that execution will stop when that function is called so I can see who is calling it or something along those lines, but I set the break point and when SystemFunction036 is called, the program doesnt stop at all. winedbg MyPhotoCalendars.exe WineDbg starting on pid 0xa In 32 bit mode. 1249 static inline void WINAPI DbgBreakPoint(void) { __asm__ __volatile__("int3"); } Wine-dbg>break SystemFunction036 fixme:dbghelp:elf_load_debug_info Unsupported Dwarf2 information for ld-linux.so.2 Breakpoint 1 at 0x4089a670 SystemFunction036 [/home/truiken/wine/dlls/advapi32/crypt.c:1914] in advapi32 Wine-dbg>step 1019 ExitProcess( entry( peb ) ); Wine-dbg>step Single stepping until exit from function, which has no line number information. ......... fixme:crypt:SystemFunction036 ( _A)l(a)H¦X@`ùm@%!D, 256): stub fixme:win:WIN_CreateWindowEx Parent is HWND_MESSAGE fixme:crypt:SystemFunction036 (H¦X@, 256): stub fixme:actctx:FindActCtxSectionGuid stub! Invalid address (0x00404aa2) for breakpoint 0, disabling it Invalid address (0x4089a670) for breakpoint 1, disabling it Process of pid=0x0000000a has terminated Wine-dbg>bt You must be attached to a process to run this command. Any thoughts? -- James Hawkins

2 1

missing advapi32 prototype: SystemFunction036
by James Hawkins 28 Sep '04

28 Sep '04

While trying to install MyPhotoCalendars, an exception is raised because of a call to the unimplemented function advapi32.dll.SystemFunction036. >From reading the following post, http://www.winehq.com/hypermail/wine-devel/2003/09/0265.html I can see that it won't be very easy to even stub this function because it's a hidden api. What methods could I use to track down any information about this function or its prototype? -- James Hawkins

4 9

Re: urlmon: IInternetSecurityManager
by Robert Shearman 28 Sep '04

28 Sep '04

Huw D M Davies wrote: > Huw Davies <huw(a)codeweavers.com> > Stub implementation of IInternetSecurityManager > Add a classfactory > Fix calling convention of CoInternetCreateSecurityManager > > >+ >+static ULONG WINAPI CF_AddRef(LPCLASSFACTORY iface) { >+ IClassFactoryImpl *This = (IClassFactoryImpl *)iface; >+ return ++(This->ref); >+} >+ > Please try to use Interlocked* functions for ref-counting as it is currently a janitorial task to clean up this type of code and I don't think they are desparate for more work. >+static HRESULT WINAPI CF_CreateInstance(LPCLASSFACTORY iface, LPUNKNOWN pOuter, >+ REFIID riid, LPVOID *ppobj) { >+ IClassFactoryImpl *This = (IClassFactoryImpl *)iface; >+ HRESULT hres; >+ LPUNKNOWN punk; >+ >+ TRACE("(%p)->(%p,%s,%p)\n",This,pOuter,debugstr_guid(riid),ppobj); >+ >+ hres = This->pfnCreateInstance(pOuter, (LPVOID *) &punk); >+ if (FAILED(hres)) { >+ *ppobj = NULL; >+ return hres; >+ } >+ hres = IUnknown_QueryInterface(punk, riid, ppobj); >+ if (FAILED(hres)) { >+ *ppobj = NULL; >+ return hres; > You leak a reference on failure here. >+ } >+ IUnknown_Release(punk); >+ return hres; >+} > >+/***************************************************************************** >+ * IInternetSecurityManager interface >+ */ >+cpp_quote("#define SID_SInternetSecurityManager IID_IInternetSecurityManager") >+ >+[ >+ object, >+ uuid(79eac9ee-baf9-11ce-8c82-00aa004ba90b), >+ pointer_default(unique) >+] >+interface IInternetSecurityManager : IUnknown >+{ >+ HRESULT SetSecuritySite( >+ [in, unique] IInternetSecurityMgrSite *pSite); >+ >+ HRESULT GetSecuritySite( >+ [out] IInternetSecurityMgrSite **ppSite); >+ >+ HRESULT MapUrlToZone( >+ [in] LPCWSTR pwszUrl, >+ [out] DWORD *pdwZone, >+ [in] DWORD dwFlags); >+ >+cpp_quote("#define MAX_SIZE_SECURITY_ID 512") > It is probably better to put this outside the interface declaration as I'm not sure if midl and widl are consistent about not outputting this twice. >+HRESULT SecManagerImpl_Construct(IUnknown *pUnkOuter, LPVOID *ppobj) >+{ >+ SecManagerImpl *This; > >+ >+ TRACE("(%p,%p)\n",pUnkOuter,ppobj); >+ This = HeapAlloc(GetProcessHeap(), 0, sizeof(*This)); > > The convention with COM interfaces is to use the CoTaskMemAlloc / CoTaskMemFree memory management functions instead of the Heap ones (even though they are the same). This is just a slight nit-pick. >+ >+ /* Initialize the virtual function table. */ >+ This->lpvtbl1 = &VT_SecManagerImpl; >+ This->ref = 1; >+ >+ *ppobj = This; >+ return S_OK; >+} > Rob

2 1

Re: PATCH: imagelist: change ERR to TRACE
by Andreas Mohr 28 Sep '04

28 Sep '04

Hi, On Tue, Sep 28, 2004 at 06:44:16AM +0200, Michael Stefaniuc wrote: > Hello, > > License: LGPL, X11 > Changelog: > Michael Stefaniuc <mstefani(a)redhat.com> > - native ImageList_Remove dosn't spit out an error message when > deleting an index out of range so don't do that either Probably true, but native ImageList_Remove doesn't have its whole Windows environment completely buggy either. ;-) IOW: this was perhaps meant for Wine code bug tracking purposes (AFAIK Wine uses imagelists a lot internally, e.g. shell code or so). While an ERR is too verbose, TRACE is ill-suited for that (a wrong index certainly is an anomaly); thus I'd say it should be a WARN. Greetings, Andreas Mohr

2 1

Re: MakeSureDirectoryPathExists improvement
by Alexandre Julliard 27 Sep '04

27 Sep '04

Wolfgang Schwotzer <wolfgang.schwotzer(a)gmx.net> writes: > @@ -91,13 +91,28 @@ > */ > BOOL WINAPI MakeSureDirectoryPathExists(LPCSTR DirPath) > { > - if (CreateDirectoryA(DirPath, NULL)) return TRUE; > - if (GetLastError() == ERROR_ALREADY_EXISTS) > + char path[MAX_PATH]; > + const char *p = DirPath; > + int n; > + > + if (strchr(p, ':') != NULL) > + p = strchr(p, ':'); > + while ((p = strchr(p, '\\')) != NULL) > { > - SetLastError(ERROR_SUCCESS); > - return TRUE; > + n = p - DirPath + 1; > + strncpy(path, DirPath, n); > + path[n] = '\0'; > + if( !CreateDirectoryA(path, NULL) && > + (GetLastError() != ERROR_ACCESS_DENIED) && > + (GetLastError() != ERROR_ALREADY_EXISTS)) > + return FALSE; > + p++; > } > - return FALSE; > + if (GetLastError() == ERROR_ACCESS_DENIED || > + (GetLastError() == ERROR_ALREADY_EXISTS)) > + SetLastError(ERROR_SUCCESS); > + > + return TRUE; I don't think you want to simply ignore ERROR_ACCESS_DENIED. You should fix your algorithm to not try to create the drive root and then ignore only ERROR_ALREADY_EXISTS. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: winecfg: Add Makefile rules to rebuild resource file when translations change
by Alexandre Julliard 27 Sep '04

27 Sep '04

Mike Hearn <mike(a)navi.cx> writes: > Add Makefile rules to rebuild resource file when translations change 'make depend' is a convenient way to do that, maybe you should give it a try ;-) -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: stub implementation SystemFunction036
by Filip Navara 27 Sep '04

27 Sep '04

James Hawkins wrote: >+ >+VOID WINAPI SystemFunction036(PCHAR Buffer, INT BufferSize) >+{ >+ FIXME("(%s, %i): stub\n", Buffer, BufferSize); >+} >+ > > If you really want to stub this function then please don't print the buffer contents. It's an output buffer and contains garbage (maybe the garbage means something, but still it's garbage). - Filip

2 1

Re: SystemFunction036 debugging problem
by James Hawkins 27 Sep '04

27 Sep '04

> Is the target (the thing calling SystemFunction036) available to download > anywhere? Yea. here's the link: http://www.download.com/redir?pid=10287361&merid=107235&mfgid=107235&lop=li… On Tue, 28 Sep 2004 07:29:14 +0800, Jonathan Wilson <jonwil(a)tpgi.com.au> wrote: > Is the target (the thing calling SystemFunction036) available to download > anywhere? > > -- James Hawkins

1 0

Problems installing office 2k.
by James Pellow 27 Sep '04

27 Sep '04

Hi List, I just installed wine from cvs as of this morning and am still having problems installing office. I get the following error when starting the installer with the msi debugging channel on: fixme:msi:MsiEnableLogA 00000516 "c:\\windows\\temp\\Office 2000 Professional Setup (0014)_MsiExec.txt" 00000000 trace:msi:MsiEnableLogW 00000516 L"c:\\windows\\temp\\Office 2000 Professional Setup (0014)_MsiExec.txt" 00000000 trace:msi:MsiInstallProductA "/I" "" fixme:msi:MsiInstallProductW L"/I" L"" trace:msi:MsiVerifyPackageW L"/I" trace:msi:MsiOpenDatabaseW L"/I" (null) 0x4055fd78 trace:msi:MSI_OpenDatabaseW L"/I" (null) fixme:msi:MSI_OpenDatabaseW open failed r = 80004005! trace:msi:MsiCloseHandle 405aa5aa Installation of /I () failed. Mike had suggested I not use the builtin msi.dll. Since no native msi system was installed, the office 2k installer to tried to install one resulting in the following error: fixme:thread:NtSetInformationThread Set ThreadImpersonationToken handle to (nil) fixme:thread:NtSetInformationThread Set ThreadImpersonationToken handle to 0x50 err:ntdll:NtQueryInformationToken Unhandled Token Information class 8! fixme:thread:NtSetInformationThread Set ThreadImpersonationToken handle to 0x44 fixme:powermgnt:SetThreadExecutionState (0x80000001): stub, harmless. fixme:powermgnt:SetThreadExecutionState (0x80000000): stub, harmless. fixme:thread:NtSetInformationThread Set ThreadImpersonationToken handle to (nil) Installation of /i (REBOOT=REALLYSUPRESS) failed. It is suprising to me that nobody else has reported this office 2k msi problem. I guess everyone is using OpenOffice these days :) I really don't use wine much, but I have been following its progress over the years and compile and test it once in a while. I am happy to do a little more debugging if someone wanted to give me some pointers. Thanks. Cheers, James -- ***************************** James A. Pellow, President Alent Design Solutions www.alentdesignsolutions.com (509) 526-0682 *****************************

1 0