- Wine-devel - list.winehq.org

How to tell winemaker and winebuild what I want to tell it
by Bill Medland 27 Sep '02

27 Sep '02

Anyone able to tell me what I am doing wrong here before I spend days trying to figure it out? My winelib wrapper wants to call MultiByteToWideChar and DOSFS_GetFullName. As far as I am aware they are available from ntdll.dll (ntdll.dll.so) So I want to to get the build to "link" to ntdll.dll I thought it should simply require "-intdll" in the argument to winemaker, but during the link I still get complaints about not finding those two functions. What am I doing wrong? Bill p.s. Thanks to Francois for the fixes to Winemaker

2 5

patch to MZ_Exec() PE executables
by chrismorgan＠rcn.com 27 Sep '02

27 Sep '02

Look reasonable to all of the experts? ;-) dlls/winedos/module.c: Chris Morgan <cmorgan(a)alum.wpi.edu> Add PE support to MZ_Exec(). Index: module.c =================================================================== RCS file: /home/wine/wine/dlls/winedos/module.c,v retrieving revision 1.21 diff -u -r1.21 module.c --- module.c 30 Aug 2002 00:03:25 -0000 1.21 +++ module.c 27 Sep 2002 18:33:58 -0000 @@ -342,14 +342,54 @@ */ BOOL WINAPI MZ_Exec( CONTEXT86 *context, LPCSTR filename, BYTE func, LPVOID paramblk ) { + DWORD binType; + STARTUPINFOA st; + PROCESS_INFORMATION pe; + BOOL status; + /* this may only be called from existing DOS processes * (i.e. one DOS app spawning another) */ - /* FIXME: do we want to check binary type first, to check - * whether it's a NE/PE executable? */ HANDLE hFile = CreateFileA( filename, GENERIC_READ, FILE_SHARE_READ, NULL, OPEN_EXISTING, 0, 0); BOOL ret = FALSE; + + if(!GetBinaryTypeA(filename, &binType)) /* determine what kind of binary this is */ + { + return FALSE; /* binary is not an executable */ + } + if (hFile == INVALID_HANDLE_VALUE) return FALSE; + + /* handle PE executables */ + if(binType == SCS_32BIT_BINARY) + { + if(func == 0) /* load and execute */ + { + BYTE fullCmdLine[512]; + LPBYTE psp_start = (LPBYTE)((DWORD)DOSVM_psp << 4); + PDB16 *psp = (PDB16 *)psp_start; + ExecBlock *blk = (ExecBlock *)paramblk; + LPBYTE cmdline = PTR_REAL_TO_LIN(SELECTOROF(blk->cmdline),OFFSETOF(blk->cmdline)); + LPBYTE envblock = PTR_REAL_TO_LIN(psp->environment, 0); + + cmdline[strlen(cmdline) - 1] = 0; /* remove the \r at the end */ + cmdline++; /* the first character contains the length of the command line */ + + snprintf(fullCmdLine, sizeof(fullCmdLine), "%s %s", filename, cmdline); /* build the full command line */ + + ZeroMemory (&st, sizeof(STARTUPINFOA)); + st.cb = sizeof(STARTUPINFOA); + status = CreateProcessA (NULL, fullCmdLine, NULL, NULL, TRUE, + 0, envblock, NULL, &st, &pe); + return TRUE; + } else + { + FIXME("EXEC type of %d not implemented for PE executables\n", func); + return FALSE; + } + } /* if(binType == SCS_32BIT_BINARY) */ + + /* handle dos executables */ switch (func) { case 0: /* load and execute */ case 1: /* load but don't execute */

1 0

InitThreadInput16
by Dimitrie O. Paun 27 Sep '02

27 Sep '02

OK, This DLL stuff is a bit confusing. In dlls/user/user32.spec we have 409 pascal16 InitThreadInput(word word) InitThreadInput16 so the exported name is InitThreadInput, right? Then how come in ./dlls/x11drv/desktop.c we call it like so: win->hmemTaskQ = InitThreadInput16( 0, 0 ); Shouldn't this yield an error? -- Dimi.

2 5

WM_STYLECHANG{ING/ED}
by Dimitrie O. Paun 27 Sep '02

27 Sep '02

Hi folks, In my relentless quest for speed :), I discovered that we don't send a WM_STYLECHANGED message when we flip the WS_VISIBLE, or WS_VSCROLL, and maybe a bunch of others, I haven't checked. Shouldn't we send such a message, when we touch *any* WS_[EX_]* bits? -- Dimi.

2 3

-gstabs and '#warning "8 as alignment is not supported"'
by Greg Turner 27 Sep '02

27 Sep '02

OK, I did this: Index: configure =================================================================== RCS file: /home/wine/wine/configure,v retrieving revision 1.341 diff -u -r1.341 configure --- configure 25 Sep 2002 03:29:55 -0000 1.341 +++ configure 27 Sep 2002 15:20:40 -0000 @@ -1998,7 +1998,7 @@ CFLAGS=$ac_save_CFLAGS elif test $ac_cv_prog_cc_g = yes; then if test "$GCC" = yes; then - CFLAGS="-g -O2" + CFLAGS="-gstabs -O2" else CFLAGS="-g" fi and sure enough, debug symbols loaded for me. The only remaining annoyance: winedbg doesn't automagically find the sources like gdb. Presumably, I can fix this by issuing "dir" commands to winedbg. So I'm happy (although it'd be nice to not have to issue those "dir" commands). Thanks! Can anyone think of a reason not to submit this as a patch? I tried -gstabs in gcc 2.96 (the only other gcc readily available to me) and it was happy, so presumably its pretty portable, and it looks like the configure change above would only apply to gcc, which seems like precisely the right thing to me. Now one other question for y'all wine gurus out there: I was looking for the alignment issue I mentioned before, and it turns out not to be an "-malign=" problem like I was suspecting. Instead, I was seeing "8 as alignment is not supported" warnings from include/pshpack8.h: # if defined(__GNUC__) || defined(__SUNPRO_C) || defined(__SUNPRO_CC) || defined(_MSC_VER) # pragma pack(8) # warning "8 as alignment is not supported" # elif !defined(RC_INVOKED) # error "Adjusting the alignment is not supported with this compiler" # endif I don't understand, if the alignment isn't supported, why the #pragma pack(8) is there...? Anyhow, since it looks like everybody using gcc gets the same warning, it's probably not a gcc3.2 bug... but I was wondering if anyone could clarify what's going on, and whether this indicates a problem in crypt32.dll.so (the affected wine dll). Thanks for your patience and help, -gmt -- "Waiting periods are only a step. Registration is only a step. The prohibition of private firearms is the goal." -U.S. Attorney General Janet Reno, December 1993

1 0

Re: winemaker: wineapploader changes
by Alexandre Julliard 27 Sep '02

27 Sep '02

Francois Gouget <fgouget(a)codeweavers.com> writes: > However, looking for wine in bindir does not make sense: > * I don't install the Winelib applications I compile using Winelib. So > * I leave prefix to the default value: /usr/local > > * I always build my application using a Wine source tree. This means > * that there is strictly nothing of interest in /usr/local/bin. > > * configure.ac goes to great length to figure out where wrc, winebuild > * and *wine* are located. So at the end of configure.ac we know very > * well what the path to wine is. It's '$WINE'. There seems to be some confusion here. We have two very different problems: one is how to make things work on the build machine, and one is how to make things work when installed. configure can only detect where things are on the build machine, but in general that's not the right paths to put in the installed binaries. The installed stuff goes in $(prefix), and I think it makes sense to expect that Wine itself will be installed there too. Note that in Wine wineapploader is only used for make install; running from inside the tree is a different problem and is done by a different script. I think winemaker should do the same thing. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: winedbg is not finding debug symbols, anybody know what I'm doing wrong?
by Greg Turner 27 Sep '02

27 Sep '02

Eric POUECH wrote... > but I assume you double checked that more likely, aren't you using gcc 3.2 > by any chance ? if so, you need to force ELF symbol emission, not > Dwarf II which became default in 3.2 Bingo. Guilty as charged. Thank you SO MUCH for this, I should have figured gcc3.2 was the culprit, that thing is feisty! I'll take a look at whether I can find a generic solution to this and post a patch. Now that you mention it, I think gcc3.2 has some other interesting warnings, something about some alignment thingys not supported anymore (probably I just change "-m$X" to "-f$X" to fix), I'll see about those too, could well be important... -gmt -- "Waiting periods are only a step. Registration is only a step. The prohibition of private firearms is the goal." -U.S. Attorney General Janet Reno, December 1993

2 1

Re: Unimplemented function... Cause of exception?
by Andriy Palamarchuk 27 Sep '02

27 Sep '02

--- John Coppens <jcoppens(a)usa.net> wrote: > Hello. > > I'm trying to run WinPLACE with the last wine > package. The composition part > works, but the compiler doesn't. I get: > > fixme:int:DOSVM_Int10Handler Get Font Information - > Not Supported (BH: 0x0) > wine: Unhandled exception, starting debugger... The fixme message may or may not have the relation to the exception. What does debugger say about the exception? Run "bt" (backtrace) command in the debugger window and report what you found. Also you can get more information about the cause of the crash from relay trace. See: http://www.winehq.org/docs/wine-devel/dbg-param.shtml http://www.winehq.org/fom-meta/cache/266.html Andriy Boston, MA, USA As an employee of an oil company, let me assure you that nothing, and I mean NOTHING scares a big company more than a noisy, smelly, obnoxious, disaffected pressure group. Now go away and chain yourself to someone important's front door. TheRegister. Unknown __________________________________________________ Do you Yahoo!? New DSL Internet Access from SBC & Yahoo! http://sbc.yahoo.com

2 1

Where is the specfile documented nowadays?
by Bill Medland 27 Sep '02

27 Sep '02

I've just come back to WineLib after a long time and discovered that things have changed a bit. (As a result my code finds the builtin dll but complains that it loaded the so but the dll is not found; I guess it hasn't been named or something) I guess we need to update the winelib documentation and the "binary dlls" part. As I understand it the spec file format has changed significantly over the past year and half the contents go somewhere else. In particular all that seems to go in the spec file nowadays is the entrypoint and the actual functions. Can someone point me at the current documentation or is there none. Bill

4 8

dlls/xxx/xxx_main.c
by Dimitrie O. Paun 27 Sep '02

27 Sep '02

Can we please rename all these dlls/xxx/xxx_main.c to dlls/xxx/xxx.c It should be possible now, right? Should I send a patch in? -- Dimi.

2 1