July 2004 - Wine-Devel - list.winehq.org

Re: Reimplementation of CryptAcquireContextA
by Michael Jung July 19, 2004

July 19, 2004

Am Montag, 19. Juli 2004 02:47 schrieb Robert Shearman: > Michael Jung wrote: > >Hello, > > > >The original CryptAcquireContextA had some issues with memory management > >in failure conditions, resulting in heap corruption under certain > >cirumstances. I've reimplemented this function, checking behaviour against > >Windows XP Prof. There is also a unit test included in the patch, which > >tests CryptAcquireContext and successfully runs to completion on > >both Windows XP and Wine. The patch looks worse than it is, since diff > >believes to have found similarities between my new implementation and the > >original CryptAcquireContextA. > > Alexandre has already modified CryptAcquireContextA to fix the original > problem, but I'll comment on the patch anyway. ... > Two comments: > 1. This seems like overkill - a new function *and* a new structure to go > with it? > 2. Use CRYPT_Free instead of HeapFree in case someone in the future > changes the memory to routines to some non-Heap* (or remove the CRYPT > memory routines altogether) > > > ... > > Your patch does highlight one issue though - we should be freeing all of > the resource from one path, not multiple paths and the best way to do > this is how it is currently done - by using a goto. > Maybe we could simplify the function by putting chunks into new > functions, but certainly not a new function just to release local > resources. > > Rob Hi Rob, The original CryptAcquireContextA had a couple of memory management issues and Alexandre's patch only takes care of one. Furthermore, the original implementation does not comply to Windows behaviour regarding error reporting. The unit test given in the patch succesfully runs to completion on both Windows XP Professional and Wine, but only with the new CryptAcquireContext. The function and the structure are used to keep the CryptAcquireContext function body more readable. As you pointed out, this could also be achieved with goto's and a labeled cleanup section (Though not as compact, since you still would have to do a "SetLastError(..); goto error;", which requires an additional pair of curly braces ;) ). Personally I don't like goto's, but this is perhaps a position too academic. So if the wine community favours the goto approach, I will be happy to modify the code. The patch uses the HeapAlloc function only for memory blocks, which are completely memory managed inside CryptAcquireContext. This means that the implementation of CRYPT_Free is irrelevant here. As far as I understand it, HeapAlloc is the way to go in new code. Am I correct here? Since I'm new to wine, I don't have a firm understanding of the coding conventions. Could some more people comment on the given patch, please? This would help me to get a better picture. Thanks and greetings, Michael

1 0

problem in dsound with DirectSoundCreate
by christoph kunz July 19, 2004

July 19, 2004

hi all, i´m trying to get the djing app "native instruments traktor dj studio" to work under recent wine versions. i´m running fc2 with kernel 2.6.6.1-435 on a dell laptop. when only using one stereo channel, the app really runs fine, but for djing you need monitor support, which means to use 4 channels of a quadrophonic sound card (my one is a audigy nx, but same for the intenal card). the means using the ds subsystem, since mm is not capable of using 4 channels.. i recently posted this question to wine-users, but i traced the error a bit more and think its more of interest to developers. if you guys can give me some hints one the sourcecode, may be i´m able to fix the problem.. the problem is: djstudio when started up the first time wants to know which device and which driver to use. therefor a dropdown lists pops up, wher the devices and the available drivers are listed (eg. mm, ds, asio, etc..) when choosing a driver i can then assign audio channels for monitoring and master. now, with wine and wineoss driver djstudio correctly lists the audio-devices available, and with mm two channels are provided for either the master or monitor audio channel. but when i choose the "ds driver" no channels are supported. i think the problem is located in the lowlevel wineoss driver, where DirectSoundCreate is something confusing with GUIDs. detail: wine-20040615, WINEDEBUG=trace+dsound,warn+dsound lists <--snip--> trace:dsound:DirectSoundEnumerateA calling lpDSEnumCallback(NULL,"Primary Sound Driver","wineoss.drv",0x556efb1c) trace:dsound:DirectSoundEnumerateA calling lpDSEnumCallback({bd6dd71a-3deb-11d1-b171-00c04fc20000},"SigmaTel STAC9750/51","wineoss.drv",0x556efb1c) trace:dsound:DirectSoundCaptureEnumerateA (0x60ce54,0x556efb1c) trace:dsound:GetDeviceID (0x55bd1868,0x556ef818) trace:dsound:DirectSoundCaptureEnumerateA calling lpDSEnumCallback(NULL,"Primary Sound Capture Driver","wineoss.drv",0x556efb1c) trace:dsound:DirectSoundCaptureEnumerateA calling lpDSEnumCallback({bd6dd71b-3deb-11d1-b171-00c04fc20000},"SigmaTel STAC9750/51","wineoss.drv",0x556efb1c) trace:dsound:DirectSoundCreate8 ({58d80060-2a50-5ba7-b171-00c04038a75b},0x5ba71010,(nil)) trace:dsound:setup_dsound_options appname = [TraktorDJStudio2Demo.exe\dsound] warn:dsound:setup_dsound_options ds_hw_accel = Emulation (default=Full) trace:dsound:GetDeviceID (0x556ef818,0x556edb84) trace:dsound:DirectSoundCreate8 expecting GUID {58d80060-2a50-5ba7-b171-00c04038a75b}. trace:dsound:DirectSoundCreate8 got GUID {bd6dd71a-3deb-11d1-b171-00c04fc20000} for wod 0. warn:dsound:DirectSoundCreate8 No device found matching given ID - trying with default one ! <--snap--> trace:dsound:IDirectSoundImpl_SetCooperativeLevel (0x55241ce8,00010024,2(DSSCL_PRIORITY)) fixme:dsound:IDirectSoundImpl_SetCooperativeLevel level=DSSCL_PRIORITY not fully supported <--snap II> as you can see the internal hw device "SigmaTel STAC9750/51" of my laptop is listed. then tow problems occur 1) DirectSoundCreate8 receives a wrong GUID (whatever that meens??) 2) SetCooperativeLevel level is not fully supported (whatever that meens??) in wine-20040716 the SetCooperativeLevel seems to be fixed and DirectSoundCreate8 switched to DirectSoundCreate but also recieves a wrong GUID, which i think thats the main reason traktor won´t list channels. i think, traktor is checking, if the device is really there now, i read something about a related problem (http://www.winehq.org/hypermail/wine-devel/2004/05/0022.html) the patch, which is in the actual sourcecode, ... This->pDirectSound = (IDirectSound*) pDirectSound; IDirectSound_AddRef((LPDIRECTSOUND) This->pDirectSound); } else { DirectSoundCreate8(NULL, (LPDIRECTSOUND8*) &This->pDirectSound, NULL); ... i think always trys to get the default device with the NULL paramter.. am i right, i dont really know?? please, can you help i some way, why the GUID is wrong (maybe my setup) or where to experiment with the sourcecode ?? thankx in advance&all the best christoph ps: the djstudio demo is available at www.nativeinstruments.com -- dipl.-ing. christoph kunz fraunhofer institut fuer arbeitswirtschaft und organisation competence center softwaretechnik nobelstrasse 12 d-70569 stuttgart germany fon: +49 (0)711 - 970 2314 fax: +49 (0)711 - 970 2300 mail: christoph.kunz(a)iao.fhg.de icq#: 49335597 web: http://www.iao.fhg.de pgp#: 12AF7964(a)wwwkey.pgp.net

1 0

Re: dsound test: Remove unneeded Caps/3D test
by Robert Reif July 19, 2004

July 19, 2004

Please don't apply this patch. The point of the test is to test the QueryInterface reference counting which you just removed. Please just fix the flag bug and not remove the point of the test. Francois Gouget wrote: > This fixes another dsound Windows crash: we first create a buffer > without DSBCAPS_CTRL3D, then we query the IDirectSound3DBuffer on it > which obviously fails so that we try to do an AddRef on a NULL pointer > (buffer3d). > > But this 3D buffer is not really needed for this test since we already > have the secondary buffer reference. Also MSDN recommends only using > CDSBCAPS_TRL3D when necessary because DirectSound can optimize things > when it's not specified. It may not yet be the case in Wine but I'd > prefer to keep the dsound.c test as simple as possible. So I removed > all references to this 3D buffer. I also went through the caps we > request on our buffers and remove all the unneeded ones. > > > Changelog: > > * dlls/dsound/tests/dsound.c > > Francois Gouget <fgouget(a)codeweavers.com> > Don't try to create a IDirectSound3DBuffer, partly to keep this > test simple and partly because it's going to fail since we did not > request DSBCAPS_CTRL3D. Fixes a crash on Windows. > Don't request unneeded caps. > >------------------------------------------------------------------------ > >Index: dlls/dsound/tests/dsound.c >=================================================================== >RCS file: /var/cvs/wine/dlls/dsound/tests/dsound.c,v >retrieving revision 1.24 >diff -u -r1.24 dsound.c >--- dlls/dsound/tests/dsound.c 13 Jul 2004 23:35:09 -0000 1.24 >+++ dlls/dsound/tests/dsound.c 16 Jul 2004 17:07:59 -0000 >@@ -336,18 +342,13 @@ > init_format(&wfx,WAVE_FORMAT_PCM,11025,8,1); > ZeroMemory(&bufdesc, sizeof(bufdesc)); > bufdesc.dwSize=sizeof(bufdesc); >- bufdesc.dwFlags=DSBCAPS_GETCURRENTPOSITION2; >- bufdesc.dwFlags|=(DSBCAPS_CTRLVOLUME|DSBCAPS_CTRLPAN); >+ bufdesc.dwFlags=DSBCAPS_GETCURRENTPOSITION2 | DSBCAPS_CTRLVOLUME; > bufdesc.dwBufferBytes=wfx.nAvgBytesPerSec*BUFFER_LEN/1000; > bufdesc.lpwfxFormat=&wfx; > rc=IDirectSound_CreateSoundBuffer(dso,&bufdesc,&secondary,NULL); > ok(rc==DS_OK && secondary!=NULL,"CreateSoundBuffer failed to create a secondary buffer 0x%lx\n", rc); > if (rc==DS_OK && secondary!=NULL) { >- LPDIRECTSOUND3DBUFFER buffer3d; >- rc=IDirectSound_QueryInterface(secondary, &IID_IDirectSound3DBuffer, (void **)&buffer3d); >- ok(rc==DS_OK && buffer3d!=NULL,"QueryInterface failed: %s\n",DXGetErrorString9(rc)); > /* add some more refs */ >- IDirectSound3DBuffer_AddRef(buffer3d); > IDirectSoundBuffer_AddRef(secondary); > } > /* release with buffer */ >@@ -459,18 +460,13 @@ > init_format(&wfx,WAVE_FORMAT_PCM,11025,8,1); > ZeroMemory(&bufdesc, sizeof(bufdesc)); > bufdesc.dwSize=sizeof(bufdesc); >- bufdesc.dwFlags=DSBCAPS_GETCURRENTPOSITION2; >- bufdesc.dwFlags|=(DSBCAPS_CTRLVOLUME|DSBCAPS_CTRLPAN); >+ bufdesc.dwFlags=DSBCAPS_GETCURRENTPOSITION2 | DSBCAPS_CTRLVOLUME; > bufdesc.dwBufferBytes=wfx.nAvgBytesPerSec*BUFFER_LEN/1000; > bufdesc.lpwfxFormat=&wfx; > rc=IDirectSound8_CreateSoundBuffer(dso,&bufdesc,&secondary,NULL); > ok(rc==DS_OK && secondary!=NULL,"CreateSoundBuffer failed to create a secondary buffer 0x%lx\n", rc); > if (rc==DS_OK && secondary!=NULL) { >- LPDIRECTSOUND3DBUFFER buffer3d; >- rc=IDirectSound8_QueryInterface(secondary, &IID_IDirectSound3DBuffer, (void **)&buffer3d); >- ok(rc==DS_OK && buffer3d!=NULL,"QueryInterface failed: %s\n",DXGetErrorString9(rc)); > /* add some more refs */ >- IDirectSound3DBuffer_AddRef(buffer3d); > IDirectSoundBuffer8_AddRef(secondary); > } > /* release with buffer */ >@@ -614,7 +617,6 @@ > ZeroMemory(&bufdesc, sizeof(bufdesc)); > bufdesc.dwSize=sizeof(bufdesc); > bufdesc.dwFlags=DSBCAPS_PRIMARYBUFFER; >- bufdesc.dwFlags|=(DSBCAPS_CTRLVOLUME|DSBCAPS_CTRLPAN); > rc=IDirectSound_CreateSoundBuffer(dso,&bufdesc,&primary,NULL); > ok(rc==DS_OK && primary!=NULL,"CreateSoundBuffer failed to create a primary buffer 0x%lx\n", rc); > >@@ -625,7 +627,6 @@ > ZeroMemory(&bufdesc, sizeof(bufdesc)); > bufdesc.dwSize=sizeof(bufdesc); > bufdesc.dwFlags=DSBCAPS_GETCURRENTPOSITION2; >- bufdesc.dwFlags|=(DSBCAPS_CTRLFREQUENCY|DSBCAPS_CTRLVOLUME|DSBCAPS_CTRLPAN); > bufdesc.dwBufferBytes=wfx.nAvgBytesPerSec*BUFFER_LEN/1000; > bufdesc.lpwfxFormat=&wfx; > trace(" Testing a secondary buffer at %ldx%dx%d\n", > >

2 2

Re: Reimplementation of CryptAcquireContextA
by Robert Shearman July 19, 2004

July 19, 2004

Michael Jung wrote: >Hello, > >The original CryptAcquireContextA had some issues with memory management >in failure conditions, resulting in heap corruption under certain >cirumstances. I've reimplemented this function, checking behaviour against >Windows XP Prof. There is also a unit test included in the patch, which >tests CryptAcquireContext and successfully runs to completion on >both Windows XP and Wine. The patch looks worse than it is, since diff >believes to have found similarities between my new implementation and the >original CryptAcquireContextA. > > Alexandre has already modified CryptAcquireContextA to fix the original problem, but I'll comment on the patch anyway. >Index: dlls/advapi32/crypt.c >=================================================================== >RCS file: /home/wine/wine/dlls/advapi32/crypt.c,v >retrieving revision 1.39 >diff -u -r1.39 crypt.c >--- dlls/advapi32/crypt.c 16 Jul 2004 19:19:00 -0000 1.39 >+++ dlls/advapi32/crypt.c 18 Jul 2004 22:55:19 -0000 >@@ -239,6 +239,45 @@ > #undef CRYPT_GetProvFunc > #undef CRYPT_GetProvFuncOpt > >+struct acquire_context_resources { >+ HKEY hKey; >+ PCHAR pszProvKeyName; >+ PCHAR pszTemp; >+ PCHAR pszImagePath; >+ PBYTE pSignature; >+}; >+ >+/****************************************************************************** >+ * acquire_context_release_resources [Internal] >+ * >+ * Helper function for CryptAcquireContext. Releases the resources used by >+ * CryptAcquireContext, sets last error to error_code and returns TRUE if >+ * error_code == ERROR_SUCCESS, else FALSE. >+ * >+ * PARAMS >+ * resources [I] Pointer to an acquire_context_resources struct, which holds >+ * pointers to resources used by CryptAcquireContext >+ * error_code [I] Error code, which the "last error" is set to. >+ * >+ * RETURNS >+ * TRUE, iff error_code == ERROR_SUCCESS >+ * FALSE, otherwise >+ */ >+inline BOOL acquire_context_release_resources(struct acquire_context_resources *resources, DWORD error_code) >+{ >+ if (resources->hKey != (HKEY)INVALID_HANDLE_VALUE) >+ RegCloseKey(resources->hKey); >+ if (resources->pszProvKeyName != NULL) >+ HeapFree(GetProcessHeap(), 0, resources->pszProvKeyName); >+ if (resources->pszTemp != NULL) >+ HeapFree(GetProcessHeap(), 0, resources->pszTemp); >+ if (resources->pszImagePath != NULL) >+ HeapFree(GetProcessHeap(), 0, resources->pszImagePath); >+ if (resources->pSignature != NULL) >+ HeapFree(GetProcessHeap(), 0, resources->pSignature); >+ SetLastError(error_code); >+ return error_code == ERROR_SUCCESS; >+} > > Two comments: 1. This seems like overkill - a new function *and* a new structure to go with it? 2. Use CRYPT_Free instead of HeapFree in case someone in the future changes the memory to routines to some non-Heap* (or remove the CRYPT memory routines altogether) ... Your patch does highlight one issue though - we should be freeing all of the resource from one path, not multiple paths and the best way to do this is how it is currently done - by using a goto. Maybe we could simplify the function by putting chunks into new functions, but certainly not a new function just to release local resources. Rob

2 1

Re: dsound tests propset.c windows compile patch
by Robert Reif July 19, 2004

July 19, 2004

Robert Reif wrote: > Fix so it compiles with VC7 on XP. > > > Here is the proper fix. Don't include dsconf IIDs in dxguid.lib. Index: dlls/dsound/tests/propset.c =================================================================== RCS file: /home/wine/wine/dlls/dsound/tests/propset.c,v retrieving revision 1.5 diff -u -r1.5 propset.c --- dlls/dsound/tests/propset.c 5 Sep 2003 23:08:38 -0000 1.5 +++ dlls/dsound/tests/propset.c 19 Jul 2004 04:14:11 -0000 @@ -29,9 +29,12 @@ #include "windef.h" #include "wingdi.h" #include "dsound.h" +#include "initguid.h" #include "dsconf.h" -#include "initguid.h" +#ifndef DSBCAPS_CTRLDEFAULT +#define DSBCAPS_CTRLDEFAULT DSBCAPS_CTRLFREQUENCY|DSBCAPS_CTRLPAN|DSBCAPS_CTRLVOLUME +#endif DEFINE_GUID(DSPROPSETID_VoiceManager,0x62A69BAE,0xDF9D,0x11D1,0x99,0xA6,0x00,0xC0,0x4F,0xC9,0x9D,0x46); DEFINE_GUID(DSPROPSETID_EAX20_ListenerProperties,0x306a6a8,0xb224,0x11d2,0x99,0xe5,0x0,0x0,0xe8,0xd8,0xc7,0x22); Index: dlls/dxguid/dxguid.c =================================================================== RCS file: /home/wine/wine/dlls/dxguid/dxguid.c,v retrieving revision 1.2 diff -u -r1.2 dxguid.c --- dlls/dxguid/dxguid.c 3 Jun 2004 00:03:23 -0000 1.2 +++ dlls/dxguid/dxguid.c 19 Jul 2004 04:14:11 -0000 @@ -37,7 +37,6 @@ #include "mmsystem.h" #include "dsound.h" #include "dsdriver.h" -#include "dsconf.h" #include "dmusici.h" #include "dmusics.h" #include "dplay.h"

1 0

Re: dsound tests propset.c windows compile patch
by Robert Reif July 19, 2004

July 19, 2004

Please disregard this patch. It breaks wine compiles. I'll dig into a better fix that works for both wine and windows. Robert Reif wrote: > Fix so it compiles with VC7 on XP. > > >

1 0

Re: dsound test: Remove unneeded Caps/3D test
by Robert Reif July 19, 2004

July 19, 2004

Francois Gouget wrote: > This fixes another dsound Windows crash: we first create a buffer > without DSBCAPS_CTRL3D, then we query the IDirectSound3DBuffer on it > which obviously fails so that we try to do an AddRef on a NULL pointer > (buffer3d). > Fixes the right bugs and keeps intent of test. Adds more reference count checks. Tested on XP with DX9. Index: dlls/dsound/tests/dsound.c =================================================================== RCS file: /home/wine/wine/dlls/dsound/tests/dsound.c,v retrieving revision 1.27 diff -u -r1.27 dsound.c --- dlls/dsound/tests/dsound.c 17 Jul 2004 00:04:17 -0000 1.27 +++ dlls/dsound/tests/dsound.c 19 Jul 2004 00:34:23 -0000 @@ -342,14 +342,21 @@ init_format(&wfx,WAVE_FORMAT_PCM,11025,8,1); ZeroMemory(&bufdesc, sizeof(bufdesc)); bufdesc.dwSize=sizeof(bufdesc); - bufdesc.dwFlags=DSBCAPS_GETCURRENTPOSITION2 | DSBCAPS_CTRLVOLUME; + bufdesc.dwFlags=DSBCAPS_GETCURRENTPOSITION2 | DSBCAPS_CTRL3D; bufdesc.dwBufferBytes=wfx.nAvgBytesPerSec*BUFFER_LEN/1000; bufdesc.lpwfxFormat=&wfx; rc=IDirectSound_CreateSoundBuffer(dso,&bufdesc,&secondary,NULL); ok(rc==DS_OK && secondary!=NULL,"CreateSoundBuffer failed to create a secondary buffer 0x%lx\n", rc); if (rc==DS_OK && secondary!=NULL) { - /* add some more refs */ - IDirectSoundBuffer_AddRef(secondary); + LPDIRECTSOUND3DBUFFER buffer3d; + rc=IDirectSound_QueryInterface(secondary, &IID_IDirectSound3DBuffer, (void **)&buffer3d); + ok(rc==DS_OK && buffer3d!=NULL,"QueryInterface failed: %s\n",DXGetErrorString9(rc)); + if (rc==DS_OK && buffer3d!=NULL) { + ref=IDirectSound3DBuffer_AddRef(buffer3d); + ok(ref==2,"IDirectSound3DBuffer_AddRef has %d references, should have 2\n",ref); + } + ref=IDirectSoundBuffer_AddRef(secondary); + ok(ref==2,"IDirectSoundBuffer_AddRef has %d references, should have 2\n",ref); } /* release with buffer */ ref=IDirectSound_Release(dso); @@ -460,14 +467,27 @@ init_format(&wfx,WAVE_FORMAT_PCM,11025,8,1); ZeroMemory(&bufdesc, sizeof(bufdesc)); bufdesc.dwSize=sizeof(bufdesc); - bufdesc.dwFlags=DSBCAPS_GETCURRENTPOSITION2 | DSBCAPS_CTRLVOLUME; + bufdesc.dwFlags=DSBCAPS_GETCURRENTPOSITION2 | DSBCAPS_CTRL3D; bufdesc.dwBufferBytes=wfx.nAvgBytesPerSec*BUFFER_LEN/1000; bufdesc.lpwfxFormat=&wfx; rc=IDirectSound8_CreateSoundBuffer(dso,&bufdesc,&secondary,NULL); ok(rc==DS_OK && secondary!=NULL,"CreateSoundBuffer failed to create a secondary buffer 0x%lx\n", rc); if (rc==DS_OK && secondary!=NULL) { - /* add some more refs */ - IDirectSoundBuffer8_AddRef(secondary); + LPDIRECTSOUND3DBUFFER buffer3d; + LPDIRECTSOUNDBUFFER8 buffer8; + rc=IDirectSound_QueryInterface(secondary, &IID_IDirectSound3DBuffer, (void **)&buffer3d); + ok(rc==DS_OK && buffer3d!=NULL,"QueryInterface failed: %s\n",DXGetErrorString9(rc)); + if (rc==DS_OK && buffer3d!=NULL) { + ref=IDirectSound3DBuffer_AddRef(buffer3d); + ok(ref==2,"IDirectSound3DBuffer_AddRef has %d references, should have 2\n",ref); + } + rc=IDirectSound_QueryInterface(secondary, &IID_IDirectSoundBuffer8, (void **)&buffer8); + if (rc==DS_OK && buffer8!=NULL) { + ref=IDirectSoundBuffer8_AddRef(buffer8); + ok(ref==3,"IDirectSoundBuffer8_AddRef has %d references, should have 3\n",ref); + } + ref=IDirectSoundBuffer_AddRef(secondary); + ok(ref==4,"IDirectSoundBuffer_AddRef has %d references, should have 4\n",ref); } /* release with buffer */ ref=IDirectSound8_Release(dso);

1 0

Yard sale windows app fun: Lotus SmartSuite '96. 1-2-3 works, WordPro doesn't.
by Dan Kegel July 18, 2004

July 18, 2004

I went to a yard sale today, and that's always dangerous. Today I ended up buying a copy of Lotus SmartSuite '96. I then installed Wine-20040615.tar.gz and tried installing my new disc with $ wine d:\\install.exe It got all confused, started iterating through every file underneath my home directory, and printed out err:shell:SHGetFileInfoW pidl is null! for each one! On a hunch, I made my home directory look empty, $ mkdir .blort $ mv * .blort $ wine d:\\install.exe That seemed much better; Wine only printed out the SHGetFileInfoW error once, presumably for the lotustmp.000 directory it created in my home directory. The next problem was when a dialog box "Select a Program Group..." popped up. At about the same time, an error box with title "Setup Message" and body "Setup has encountered a problem. Please report File='COMMON\SYMTAB11.C',Line='618' to Microsoft Product Support Services." This looks a bit lot like http://support.microsoft.com/default.aspx?scid=kb;EN-US;113252 which happens when GetWindowsDirectory() returns a string with lowercase letters, but clicking OK let me past with no ill affects, and I was able to install. At least when all I installed was Word Pro. When I also tried installing 1-2-3 and ScreenCam, I got the error "Lotus Setup" : "Install encountered an error trying to start Regedit.exe." That popped up twice, after which install thought for a long time, then warned fixme:ole:CoCreateInstance no classfactory created for CLSID {00021401-0000-0000-c000-000000000046}, hres is 0x80040154 and seemed to get stuck. Still, things seemed to be more or less well installed, so I tried running two of the apps. $ cd c/lotus/123/programs $ wine 123w.exe worked quite nicely. (Well, it did say the date was 6/17/104, but nobody's perfect!) Next I tried Word Pro: $ cd ~/.wine/c/lotus/wordpro $ wine wordpro.exe That failed immediately, so quickly that WINEDEBUG=+all wine wordpro.exe was practical to run. Here's part of its output. It looks as if the very first instruction accesses a null pointer in a strange way. 0009:Starting process L"C:\\lotus\\wordpro\\wordpro.exe" (entryproc=0x7a6aee) 0009:trace:seh:EXC_RtlRaiseException code=c0000005 flags=0 addr=0x7a6aee 0009:trace:seh:EXC_RtlRaiseException info[0]=00000001 0009:trace:seh:EXC_RtlRaiseException info[1]=00000000 0009:trace:seh:EXC_RtlRaiseException eax=00000000 ebx=403f1668 ecx=400c0be8 edx=400c0c37 esi=400c1000 edi=007a6aee 0009:trace:seh:EXC_RtlRaiseException ebp=4070fff4 esp=4070ff30 cs=0023 ds=002b es=002b fs=003b gs=0033 flags=00010246 0009:trace:seh:EXC_CallHandler calling handler at 0x403e1708 code=c0000005 flags=0 0009:Call ntdll.NtQueryVirtualMemory(ffffffff,00000000,00000000,4070fac0,0000001c,4070fa84) ret=4037aeb8 0009:Ret ntdll.NtQueryVirtualMemory() retval=00000000 ret=4037aeb8 0009:Call ntdll.RtlImageDirectoryEntryToData(00000000,00000001,00000002,4070fabc) ret=4033fbbe 0009:trace:seh:EXC_RtlRaiseException code=c0000005 flags=0 addr=0x40091368 0009:trace:seh:EXC_RtlRaiseException info[0]=00000000 0009:trace:seh:EXC_RtlRaiseException info[1]=00000000 0009:trace:seh:EXC_RtlRaiseException eax=00000000 ebx=400beec4 ecx=00000001 edx=00000000 esi=00000000 edi=00000002 0009:trace:seh:EXC_RtlRaiseException ebp=4070f9e4 esp=4070f928 cs=0023 ds=002b es=002b fs=003b gs=0033 flags=00010246 0009:trace:seh:EXC_CallHandler calling handler at 0x40089948 code=c0000005 flags=0 0009:trace:seh:EXC_RtlUnwind code=c0000005 flags=2 0009:trace:seh:EXC_CallHandler calling handler at 0x40088fc0 code=c0000005 flags=2 0009:trace:seh:EXC_CallHandler handler returned 1 0009:Ret ntdll.RtlImageDirectoryEntryToData() retval=00000000 ret=4033fbbe 0009:Call ntdll.NtCreateEvent(4070fac0,001f0003,4070fac4,00000001,00000000) ret=4033f836 0009:Ret ntdll.NtCreateEvent() retval=00000000 ret=4033f836 wine: Unhandled exception (thread 0009), starting debugger... ... Unhandled exception: page fault on write access to 0x00000000 in 32-bit code (0x007a6aee). In 32 bit mode. Register dump: CS:0023 SS:002b DS:002b ES:002b FS:003b GS:0033 EIP:007a6aee ESP:4070ff30 EBP:4070fff4 EFLAGS:00010246( - 00 -RIZP1) EAX:00000000 EBX:403f1668 ECX:400c0be8 EDX:400c0c37 ESI:400c1000 EDI:007a6aee Stack dump: 0x00000000: 00000000 00000000 00000000 00000000 0x00000010: 00000000 00000000 00000000 00000000 0x00000020: 00000000 00000000 00000000 00000000 0x00000030: 00000000 00000000 00000000 00000000 0x00000040: 00000000 00000000 00000000 00000000 0x00000050: 00000000 00000000 00000000 00000000 0x00000060: Backtrace: =>1 0x007a6aee (0x4070fff4) 2 0x4002e83d (0x00000000) 0x007a6aee: addb %al,0x0(%eax) Modules: 0x40070000-400c2000 Deferred ntdll 0x40320000-403f3000 Deferred kernel32 0x00400000-00888000 --none-- wordpro 0x402f0000-403f3000 Deferred kernel32<elf> 0x40019000-40025000 Deferred libnss_files.so.2 0x401b5000-401d7000 Deferred libm.so.6 0x400c2000-401b5000 Deferred libwine_unicode.so.1 0x40053000-400c2000 Deferred ntdll<elf> 0x40000000-40016000 Deferred ld-linux.so.2 0x4004f000-40052000 Deferred libdl.so.2 0x420155f0-841485f0 Deferred libc.so.6 0x40042000-4004f000 Deferred libpthread.so.0 0x4002a000-40042000 Sym libwine.so.1 0x420155f0-841485f0 Deferred libc.so.6 Module Address Debug info Name (14 modules) PE 0x00400000-00888000 --none-- wordpro ELF 0x40000000-40016000 Deferred ld-linux.so.2 ELF 0x40019000-40025000 Deferred libnss_files.so.2 ELF 0x4002a000-40042000 Sym libwine.so.1 ELF 0x40042000-4004f000 Deferred libpthread.so.0 ELF 0x4004f000-40052000 Deferred libdl.so.2 ELF 0x40053000-400c2000 Deferred ntdll<elf> \-PE 0x40070000-400c2000 Deferred ntdll ELF 0x400c2000-401b5000 Deferred libwine_unicode.so.1 ELF 0x401b5000-401d7000 Deferred libm.so.6 ELF 0x402f0000-403f3000 Deferred kernel32<elf> \-PE 0x40320000-403f3000 Deferred kernel32 ELF 0x420155f0-841485f0 Deferred libc.so.6 ELF 0x420155f0-841485f0 Deferred libc.so.6 Threads: process tid prio (all id:s are in hex) 0000000b (D) C:\lotus\wordpro\wordpro.exe 00000009 0 <== 00000010 00000011 0 WineDbg terminated on pid 0xb So I guess I got a couple hours worth of entertainment out of my $5, and I can now run Lotus 1-2-3 if I feel like it. Cool. - Dan -- My technical stuff: http://kegel.com My politics: see http://www.misleader.org for examples of why I'm for regime change

3 8

HttpOpenRequest: Who has to clean up lpszObjectName?
by Uwe Bonnes July 18, 2004

July 18, 2004

Hallo, appended test fails with builtin wininet, but succeeds with native wininet. The failing call, triggered but the ending \r, seems to be caused by the www-server side handling. www.winehq.org seems to cope with the ending \r, so the test has to be carried out on the xilinx server. Ise webupdate fails caused by this error. What function has to clean up the ending \r? Thanks -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ---------- Index: wine/dlls/wininet/tests/http.c =================================================================== RCS file: /home/wine/wine/dlls/wininet/tests/http.c,v retrieving revision 1.13 diff -u -r1.13 http.c --- wine/dlls/wininet/tests/http.c 23 Jan 2004 22:44:26 -0000 1.13 +++ wine/dlls/wininet/tests/http.c 18 Jul 2004 20:36:28 -0000 @@ -275,10 +275,30 @@ ok((strcmp(TEST_URL_PATH,path) == 0),"path cracked wrong\n"); } +void InternetOpenRequest_test() +{ + DWORD rc; + HINTERNET hi, hic = 0, hor = 0; + const char *types[2] = { "*/*", NULL }; + hi = InternetOpenA("",0x0,0x0,0x0,0x0); + ok((hi != 0x0),"InternetOpen Failed\n"); + trace("InternetOpenA -->\n"); + hic=InternetConnectA(hi,"www.xilinx.com",0x0, 0x0, 0x0, 0x3,0x0, 0xdeadbeef); + ok((hic != 0x0),"InternetConnect Failed\n"); + trace("InternetConnect Succeeded\n"); + hor = HttpOpenRequestA(hic, "GET", + "/direct/swhelp/series6/6.2i_latestservicepack.dat\r", + "HTTP/1.0", + 0x0,types,04000000,00000000); + rc= HttpSendRequestA(hor,0x0,0x0,0x0,0x0); + ok((rc != 0x0),"HttpSendRequestA Failed\n"); +} + START_TEST(http) { winapi_test(0x10000000); winapi_test(0x00000000); InternetCrackUrl_test(); InternetOpenUrlA_test(); + InternetOpenRequest_test(); }

1 0

Re: msiexec.exe replacement
by Robert Shearman July 18, 2004

July 18, 2004

Vincent Béron wrote: >Note: Much functionnality is still absent, but it's already able to >begin the installation of iTunes.msi and netfx.msi. >I plan on updating this in the coming days. > > Nice work. >Changelog: >First draft for a msiexec.exe replacement. > >Vincent > > ... >+ printf("Dll %s does not implement function %s\n", DllName, ProcName); > > Minor nitpick: this (and similar messages) should be fprintf(stderr,...) >+static void DllRegisterServer(LPCSTR DllName) >+{ >+ HRESULT hr; >+ DLLREGISTERSERVER pfDllRegisterServer = NULL; >+ HMODULE DllHandle = NULL; >+ >+ pfDllRegisterServer = LoadProc(DllName, "DllRegisterServer", &DllHandle); >+ >+ hr = pfDllRegisterServer(); >+ if(FAILED(hr)) >+ { >+ printf("Failed to register dll %s\n", DllName); >+ ExitProcess(1); >+ } >+ printf("Successfully registered dll %s\n", DllName); >+ if(DllHandle) >+ FreeLibrary(DllHandle); >+} >+ >+static void DllUnregisterServer(LPCSTR DllName) >+{ >+ HRESULT hr; >+ DLLUNREGISTERSERVER pfDllUnregisterServer = NULL; >+ HMODULE DllHandle = NULL; >+ >+ pfDllUnregisterServer = LoadProc(DllName, "DllUnregisterServer", &DllHandle); >+ >+ hr = pfDllUnregisterServer(); >+ if(FAILED(hr)) >+ { >+ printf("Failed to unregister dll %s\n", DllName); >+ ExitProcess(1); >+ } >+ printf("Successfully unregistered dll %s\n", DllName); >+ if(DllHandle) >+ FreeLibrary(DllHandle); >+} >+ > > ... >+ >+ ExitProcess(0); >+ return 0; >+} > > The ExitProcess call should not be needed. Rob

1 0