September 2007 - Wine-Devel

Re: winex11.drv[1/4]: fix two bugs in WTInfoA
by Alexandre Julliard Sept. 19, 2007

Sept. 19, 2007

"Mikolaj Zalewski" <mikolajz(a)google.com> writes: > Two bugs I've found working on previous implentations > From b233532f7adc140c4e36fb2288581fe0ff181935 Mon Sep 17 00:00:00 2001 > From: =?utf-8?q?Miko=C5=82aj_Zalewski?= <mikolaj(a)zalewski.pl> > Date: Mon, 17 Sep 2007 11:54:41 -0700 > Subject: [PATCH] winex11.drv: fix two bugs in WTInfoA Please write a better description of the changes, "fix bug" can apply to just about any commit (and in general, if you have two bugs to fix there should be two patches...) -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: A first implentation of services.exe
by Robert Shearman Sept. 18, 2007

Sept. 18, 2007

Mikolaj Zalewski wrote: > I don't want this patch to get committed but to maybe get some > comments. If the patch will be considered to be ok I can send it as > three smaller patches. It implements the basic infrastructure for > services.exe - a process that maintains the services database and > states. I think this can be useful in the long run. Absolutely. > This patch > currently doesn't implement anything useful - it connects to > services.exe in OpenSCManager and disconnect in CloseServiceHandle and > validating the database name is now done in services.exe but if this > approach is good I will add more functionality. > It looks good to me, but I'm a bit worried that you'll come across bugs/unimplemented features in widl when you start adding more calls. > Is there a way to check if a RPC connection over a named pipe was > closed by the client (probably we should close all it's handles in > such a case)? Yes, you can do this using context handles, but unfortunately we're lacking support for these on the RPC server side and in widl. > Also this program show a bug in our rpcrt4 - if we pass > NULL as a [in,unique,string] parameter rpcrt4 will crash because of a > strlen on it in NdrConformantStringBufferSize. To see it one can > remove the workaround in OpenSCManager and do a `make test`. That sounds like a bug in widl to me, and I'll take a look. > Macros > like RET_IF_ERROR should be removed from the final version or can they > stay? Definitely should be removed. > diff --git a/include/wine/svcctl_shared.h b/include/wine/svcctl_shared.h > new file mode 100644 > index 0000000..5687ddc > --- /dev/null > +++ b/include/wine/svcctl_shared.h > @@ -0,0 +1,35 @@ > +/* > + * SVCCTL (services.exe RPC interface) shared constants > + * > + * Copyright (C) 2007 Google (Mikolaj Zalewski) > + * > + * This library is free software; you can redistribute it and/or > + * modify it under the terms of the GNU Lesser General Public > + * License as published by the Free Software Foundation; either > + * version 2.1 of the License, or (at your option) any later version. > + * > + * This library is distributed in the hope that it will be useful, > + * but WITHOUT ANY WARRANTY; without even the implied warranty of > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU > + * Lesser General Public License for more details. > + * > + * You should have received a copy of the GNU Lesser General Public > + * License along with this library; if not, write to the Free Software > + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA > + */ > + > +#ifndef __WINE_SVCCTL_SHARED_H > +#define __WINE_SVCCTL_SHARED_H > + > +#include <stdarg.h> > +#include <windef.h> > +#include <winbase.h> > + > +#define SVCCTL_TRANSPORT {'n','c','a','c','n','_','n','p',0} > +#define SVCCTL_ENDPOINT {'\\','p','i','p','e','\\','s','v','c','c','t','l',0} > + > +/* Not the Windows event name - if needed the true one can be found in Inside Windows */ > +#define SVCCTL_STARTED_EVENT (const WCHAR[]){'S','v','c','c','t','l','S','t','a','r','t','e','d',0} > + > + > +#endif /* __WINE_RPCSS_SHARED_H */ > We shouldn't need another file for this. You can put this same stuff in svcctl.idl using cpp_quote. > +DWORD RPC_MainLoop() Make sure to fully specify the parameters for functions with no parameters by declaring them as "(void)". -- Rob Shearman

1 0

Want to contact Mike McCormack
by Stefan Ensslen Sept. 18, 2007

Sept. 18, 2007

Dear Sirs, I wanted to contact Mike McCormack from the WinE developer team, but my email to mike(a)codeweavers.com bounced. If you could send me his new email address, I'd be very happy. The question I wanted to ask Mr. McCormack is the following: Dear Mike, I have found your code for generating proxy/stub-code at www.winehq.org. I wanted to ask where you have found a documentation of the functions used in this code (for example NdrProxyInitialize, NdrProxySendReceive, and so on). The MSDN doesn't provide any _valuable_ documentation for this function, so you surely have another source of documentation. Thanks in advance, Stefan Ensslen -- GMX FreeMail: 1 GB Postfach, 5 E-Mail-Adressen, 10 Free SMS. Alle Infos und kostenlose Anmeldung: http://www.gmx.net/de/go/freemail

2 1

AppDB down - too many SQL connections?
by Alexander Nicolaysen Sørnes Sept. 18, 2007

Sept. 18, 2007

Hello, I am unable to access the database now because of too many MySQL connections. Does anyone have a log file to indicate whether this is some sort of attack, or whether we just have too many legitimate users? Alexander N. Sørnes

2 1

make crosstest is missing -lcabinet
by Pedro Araújo Sept. 18, 2007

Sept. 18, 2007

Hi all, I've been trying to cross-compile the Wine tests, but I keep getting this error: $ ./configure && make crosstest [SNIP] i586-mingw32msvc-gcc advpack.cross.o files.cross.o install.cross.o testlist.cross.o -o advpack_crosstest.exe -lcabinet -luser32 -ladvapi32 -lkernel32 /usr/lib/gcc/i586-mingw32msvc/3.4.5/../../../../i586-mingw32msvc/bin/ld: cannot find -lcabinet collect2: ld returned 1 exit status make[2]: *** [advpack_crosstest.exe] Error 1 make[2]: Leaving directory `/.../dlls/advpack/tests' make[1]: *** [advpack/tests/__crosstest__] Error 2 make[1]: Leaving directory `/.../dlls' make: *** [dlls/__crosstest__] Error 2 Am I missing anything? Thanks in advance. -- - Pedro.

2 2

Problem with TCP Option 0x00000008
by Christian Mahnke Sept. 18, 2007

Sept. 18, 2007

Hello, i have a problem with a program that uses a network license manager. I've installed native DCOM using the winetricks script. When i start the program i got the message "fixme:winsock:WS_setsockopt Unknown IPPROTO_TCP optname 0x00000008" after the loading of native rpcltccm.dll. To work around this i've compiled a modified wine 0.9.44 with an empty case statement for 0x00000008 in socket.c. It doesn't help much, the program just fails a bit later. The TCP Option 0x00000008 is defined as TCP_ATMARK in ws2_ipdef.h in the Windows SDK. Does anyone know how this should be mapped ? Cheers, Christian

4 6

compile wine from sources - strange messages
by Andrey Gavrilin Sept. 18, 2007

Sept. 18, 2007

Hello. I wish to compile wine from sources. Has received such message. The distribution: Ubuntu 7.04. What headers files does not suffice? PS I used the program the translator from Russian to English. configure: WARNING: jack/jack.h: present but cannot be compiled configure: WARNING: jack/jack.h: check for missing prerequisite headers? configure: WARNING: jack/jack.h: see the Autoconf documentation configure: WARNING: jack/jack.h: section "Present But Cannot Be Compiled" configure: WARNING: jack/jack.h: proceeding with the preprocessor's result configure: WARNING: jack/jack.h: in the future, the compiler will take precedence configure: WARNING: ## ------------------------------------ ## configure: WARNING: ## Report this to wine-devel(a)winehq.org ## configure: WARNING: ## ------------------------------------ ## configure: WARNING: pthread.h: present but cannot be compiled configure: WARNING: pthread.h: check for missing prerequisite headers? configure: WARNING: pthread.h: see the Autoconf documentation configure: WARNING: pthread.h: section "Present But Cannot Be Compiled" configure: WARNING: pthread.h: proceeding with the preprocessor's result configure: WARNING: pthread.h: in the future, the compiler will take precedence

2 1

Re: kernel32: Don't crash accessing an invalid handle in GlobalSize
by Robert Shearman Sept. 18, 2007

Sept. 18, 2007

James Hawkins wrote: > + if (IsBadReadPtr(pintern, 1)) > + { > + SetLastError(ERROR_INVALID_HANDLE); > + return 0; > + } > + > if(pintern->Magic==MAGIC_GLOBAL_USED) > { > if (!pintern->Pointer) /* handle case of GlobalAlloc( ??,0) */ It would be better to wrap the Magic access with an exception handler rather than use IsBadReadPtr. -- Rob Shearman

1 0

Re: fix for bug 9665
by James Hawkins Sept. 17, 2007

Sept. 17, 2007

On 9/17/07, Martin <martin.marsan(a)gmail.com> wrote: > > > 2007-09-17 Martin Sand <deadlock(a)start.no> > > * dlls/ddraw/direct3d.c: > ddraw: fixes a bug which prevents Kings Quest 8 from starting(bug 9665) > A good changelog entry details what you changed, not what apps are fixed by a change. -- James Hawkins

2 1

Re: fix for bug 9665
by Stefan Dösinger Sept. 17, 2007

Sept. 17, 2007

Am Montag, 17. September 2007 20:45:04 schrieb Martin: I doubt this fix is correct, as explained in bug 9665 a few minutes ago. It should fail the d3d test too

1 0

[AppDB] Bug: Let objectManager delete bug links
by Alexander Nicolaysen Sørnes Sept. 17, 2007

Sept. 17, 2007

Bug: Let objectManager delete bug links Regards, Alexander N. Sørnes

1 0

Re: user32: better fix how to handle GetWindowLongPtr[AW](..., GWLP_WNDPROC) for builtin winprocs [resend]
by Alexandre Julliard Sept. 17, 2007

Sept. 17, 2007

"Mikolaj Zalewski" <mikolajz(a)google.com> writes: > Is there something wrong with this patch? > From 50eacd2465df53e5cfc15458ff5ab80f0cc07e7e Mon Sep 17 00:00:00 2001 > From: Mikolaj Zalewski <mikolaj(a)zalewski.pl> > Date: Mon, 10 Sep 2007 14:03:49 -0700 > Subject: [PATCH] user32: better fix how to handle GetWindowLongPtr[AW](..., GWLP_WNDPROC) for builtin winprocs There are several other places where winprocs are returned, like SetWindowLong or GetClassLong. You probably have to handle them too, or at least write test cases for them. -- Alexandre Julliard julliard(a)winehq.org

1 0

An enumeration of network resources
by Konstantin Kondratyuk Sept. 17, 2007

Sept. 17, 2007

Hello! I try to organize an enumeration of network resources by means of functions WNetOpenEnum and WNetEnumResource. Function WNetOpenEnumW contains a line: " ret = providerTable->table[index].openEnum( dwScope, dwType, dwUsage, lpNet, &handle); " which addresses to function in dll. The information about loaded dll is absent in the wine registry. Whether it means, what there are no preparations of DLL's for work with a network (similar ntlanman.dll in Windows)? -- Best regards, Konstantin Kondratyuk.

2 1

Re: msi(1/3): Don't crash if rec is NULL
by James Hawkins Sept. 17, 2007

Sept. 17, 2007

On 9/14/07, Juan Lang <juan.lang(a)gmail.com> wrote: > This series of patches is in response to Jeremy's troubles with > iTunes. They may or may not fix the installer for him, but they > prevent potential crashes. > --Juan > MSI_FormatRecordW is an internal function and we should be crashing if rec is NULL. -- James Hawkins

3 5

old dos/windows program crashing
by Joris Huizer Sept. 17, 2007

Sept. 17, 2007

Hello, I have a program, that I got like ages ago. It effectively is an atari emulator for dos, by the name "pacifist", but I originally got it installed under windows ME. It has an old atari-basic program "installed" in it, (a simple game) written by a friend in it, which I wanted to play. I decided to try to get it to run again under Linux. The website it came from originally doesn't seem to have the program available for download anymore - so I just copied the files from the windows ME disc. I obviously tried WINE then, but it didn't work, however, dosbox works with it (though slow, obviously) so I know I'm not missing required files (though, perhaps, registry settings - not sure how likely that is, though) Anyway. This isn't such a big deal, though I'd like to look at what's going on in WINE, and possibly fix it - though I need help here. This program is, among things, a "PMODE W/DOS Extender" - so that means it probably does some obscure real-mode/protected-mode stuff. Console output is as follows, Warning: unprotecting memory to allow real-mode calls. NULL pointer accesses will no longer be caught. err:int:__wine_emulate_instruction mov cr0,eax at 0x00000544 err:int:DOSVM_Int33Handler int33: unknown/not implemented parameters: int33: AX 0015, BX 0003, CX 11f7, DX 1165, SI 0055, DI 03ac, DS 11df, ES 11e7 fixme:int:DOSVM_Int10Handler Select vertical resolution - not supported fixme:int:DOSVM_Int10Handler Load ROM 8x8 Double Dot Patterns - Not Supported fixme:int31:DOSVM_Int31Handler physical address mapping (0x00000000) - unimplemented wine: Unhandled page fault on read access to 0xffffffff at address 0x101f:0x0003ed71 (thread 002d), starting debugger... fixme:dbghelp:addr_to_linear Failed to linearize address 08f0:00000002 (mode 0) fixme:dbghelp:addr_to_linear Failed to linearize address 08f0:00000000 (mode 0) Also I attached the visual display the "winevdm" shows as the program crashes. (It seems it crashes somewhere in generated code in kernel32.dll - in DPMI_PendingEventCheck() - written by the winebuild tool) I know some things about real-mode code, but not too much. Could you give me pointers on what to look at next? BTW, please CC me, I'm not on the mailing list currently. Thanks, Joris ____________________________________________________________________________________ Check out the hottest 2008 models today at Yahoo! Autos. http://autos.yahoo.com/new_cars.html

3 4

AppDB: cannot submit a screenshot even when logged in
by Vit Hrachovy Sept. 16, 2007

Sept. 16, 2007

Hello everybody, I may missed something, but I cannot submit a screenshot to the application I maintain in AppDB and I'm logged in. There's always a nasty 'Log in to submit screenshots' instead of Submit button. Is it planned downtime? Regards Vit P.S. It's http://appdb.winehq.org/screenshots.php?iAppId=4025

2 2

DirectDraw opengl renderer
by Alexander Nicolaysen Sørnes Sept. 16, 2007

Sept. 16, 2007

Hello everyone, I know this has been discussed before (and not that long ago), but I think I have some wonderful new information to add to the debate of whether or not to add an option for this in winecfg. As I understand it, there is still a long time before the opengl renderer will become the default one. With Wine 0.9.45, some bugs were fixed allowing the opengl renderer to be used for C&C Tiberian Sun, and I must say the speed improvements are dramatic! Without any FPS tools I would definitely say performance has increased by several hundred per cent. Now the game is so fast I may have to turn down the game speed setting. :) If you move the mouse it still slows down, though, as it does with the gdi renderer. Anyway, I think this should apply to several other games as well (and I will test more). But I was wondering is this is enough to add an option in winecfg? Alexander N. Sørnes

3 3

Re: RICHED20: additional tests plus RFC
by Alexander Nicolaysen Sørnes Sept. 16, 2007

Sept. 16, 2007

On Sunday 16. September 2007 04:34:17 a_villacis(a)palosanto.com wrote: > I am tracking down a richedit bug in FACETS (bug #1713). > > Apparently, the OCX control used by this app expects a particular > SendMessage() > with the message EM_SETCHARFORMAT to return 1. Currently builtin returns 0 > unconditionally, making this app bail out with "Invalid property value". By > changing the return value to 1 unconditionally, the app gets "fixed", but > this > is not a proper solution, since other apps might break with this change. > > This patch adds additional testing of EM_SETCHARFORMAT. All todo_wine's > succeed > in an installation of WinXP SP2. Please confirm this is so for all of you. > > There are two code blocks commented out. The first one makes the app crash > on WinXP, and is listed for completeness. The other one was supposed to > test what > happens when the riched20 control receives a blanked-out CHARFORMAT2 > structure. > However this triggers an assertion in Wine at style.c line 63. WinXP > doesn't bat an eye, and possibly ignores these structure values. Unless > somebody objects, I will shortly send a patch that replaces the assertion > with a NULL return in style.c, plus a NULL check in editor.c . > > There are more riched20 tests that fail in WinXP SP2, unrelated to > EM_SETCHARFORMAT, but I would like confirmation that (1) somebody else sees > failing tests in native Windows, not just me, and (2) this patch does not > introduce any new failing tests in native Windows. > > Changelog: > * Add additional tests for EM_SETCHARFORMAT message of riched20 control Hey, great to see some work on EM_SETCHARFORMAT. If you're interested, Wine has a builtin WordPad (just type 'wine wordpad') which exposes some problems. Try to select a font in the font list when no text is selected, for instance. Alexander N. Sørnes

1 0

Re: [PATCH 2/7] gdi: Implement BiDi classification of characters
by Steven Edwards Sept. 16, 2007

Sept. 16, 2007

Hi Maarten, Could you configure your git scripts to output different filenames when generating patches? Somehow they get managed for me in my gmail inbox. It makes it quite a pain for me to review. Thanks -- Steven Edwards "There is one thing stronger than all the armies in the world, and that is an idea whose time has come." - Victor Hugo

6 10

Re: patches to make wine compile on osx
by Roderick Colenbrander Sept. 16, 2007

Sept. 16, 2007

Hi Daniel, A few comments regarding the patches. First of all you should use a single patch file for each patch (this makes tracking patches easier). Now over to the patches. Regarding wineosxtypeerrors.patch: - 'GLenum' should be 'unsigned int' instead of int. I don't understand why the casts are needed as it should just get the type info from gl.h. I think that using '%u' for unsigned data could fix most of your problems. Regarding wineosxglexterrors.patch: - Same comment regarding unsigned int data. (make this a separate patch) - GL_TEXTURE_RECTANGLE_NV should be handled in include/wine/wgl.h, strange that I didn't receive compile failures for linux about it .. perhaps it is already in the mesa headers. (make this a separate patch) Anyway I would define GL_TEXTURE_RECTANGLE_NV like this: #ifndef GL_NV_texture_rectangle #define GL_TEXTURE_RECTANGLE_NV 0x84F5 #define GL_TEXTURE_BINDING_RECTANGLE_NV 0x84F6 #define GL_PROXY_TEXTURE_RECTANGLE_NV 0x84F7 #define GL_MAX_RECTANGLE_TEXTURE_SIZE_NV 0x84F8 #endif The extension is fully compatible with GL_ARB_texture_rectangle which osx should have and GL_EXT_texture_rectangle is likely the same. (the constants between the ARB and NV one are also the same, I expect the same for EXT and ARB) Regards, Roderick Colenbrander

1 0

Dfw5 draws graphs black (regression)
by Willi Mann Sept. 15, 2007

Sept. 15, 2007

Hi! Derive 5 no longer draws graphs in color. It should look like that: http://wserver.wm1.at/~willi/wine/dfw.png With current versions, it draw the blue graph in black. I could track it down to this commit (between 0.9.25 and 0.9.26): http://wserver.wm1.at/~willi/wine/wine_dfw.diff.txt (In fact, these are two commits, but the intermediate version does not compile) Please let me know if I can do anything to help fixing this bug. Willi Please CC: me on replies.

2 2

Re: add shell32.755
by Maarten Lankhorst Sept. 15, 2007

Sept. 15, 2007

Louis. Lenders schreef: > Hi, i'm sending in 2 patches from Anastasius Focht for review, and > hopefully for inclusion in git. Otherwise they might get "lost" and > forgotten. > > They're helpful for testing apps with native comdlg32.dll. So although > they do not really fix any real life app (not that i know of I'm missing a few tests to see if this behavior is actually what it should be like in windows. You should always add them to implementations of new functions. Cheers, Maarten

1 0

Re: fnt2fon: janitorial IgnoredReturnValues (1/5)
by Alexandre Julliard Sept. 14, 2007

Sept. 14, 2007

Lionel_Debroux <lionel_debroux(a)yahoo.fr> writes: > tools/fnt2fon is flagged in the janitorial page "IgnoredReturnValues". > This page is outdated, but its content is still valid for fnt2fon, which > ignores a number of return values. I don't think that's worth fixing, the warnings are essentially useless, it just makes the code uglier. There may be a few legitimate errors in there, like checking the fopen() result, and those could be fixed; but please don't add checking to every call that gcc complains about. -- Alexandre Julliard julliard(a)winehq.org

2 2

Problem with TCP Option 0x00000008
by Christian Mahnke Sept. 14, 2007

Sept. 14, 2007

Hello, i have a problem with a program that uses a network license manager. I've installed native DCOM using the winetricks script. When i start the program i got the message "fixme:winsock:WS_setsockopt Unknown IPPROTO_TCP optname 0x00000008" after the loading of native rpcltccm.dll. To work around this i've compiled a modified wine 0.9.44 with an empty case statement for 0x00000008 in socket.c. It doesn't help much, the program just fails a bit later. The TCP Option 0x00000008 is defined as TCP_ATMARK in ws2_ipdef.h in the Windows SDK. Does anyone know how this should be mapped ? Cheers, Christian

1 0

Re: winex11.drv/wintab32[1/5]: store categories field-by-field (resend)
by Alexandre Julliard Sept. 14, 2007

Sept. 14, 2007

"Mikolaj Zalewski" <mikolajz(a)google.com> writes: > /*********************************************************************** > * X11DRV_WTInfoA (X11DRV.@) > + * > + * NOTES > + * If cchOnlyCountString is not NULL and the data is a string then no > + * data is stored in the output buffer but only the string length is > + * returned in *cchOnlyCountString (and is >= 1 as contains at least a NUL > + * character). It have no effect on non-string data. > + * > + * Apart from the cchOnlyCountString parameter this function differs from the > + * WTInfo[AW] as for nIndex == 0 it returns the number of items in the > + * category. It's wintab32.dll that concatenates the data > + * > + * It is valid to call WTInfoA with lpOutput == NULL, as per standard. > + * lpOutput == NULL signifies the user only wishes to find the size of > + * the data. > */ > -UINT X11DRV_WTInfoA(UINT wCategory, UINT nIndex, LPVOID lpOutput) > +UINT X11DRV_WTInfoA(UINT wCategory, UINT nIndex, LPVOID lpOutput, DWORD *cchOnlyCountString) You shouldn't change the prototype. Driver functions should follow the signature of the corresponding APIs as far as possible. -- Alexandre Julliard julliard(a)winehq.org

2 2