- Wine-Devel - list.winehq.org

Re: VirtualProtect() fixup
by Alexandre Julliard June 8, 2001

June 8, 2001

520053692817-0001(a)t-online.de (Andreas Mohr) writes: > What isn't the case ? Both fixups ? Or only one of these ? The return value: NT returns a BOOL, 0 or 1. I haven't checked the old protection behavior; but anyway this doesn't belong in VirtualProtect, if we have to ignore the execute permission it must be done consistently throughout the code. > Well, to be exact, no application. > I just found this out by accident. That's what I suspected ;-) > IMHO the probability that some applications exist that are too lazy > to supply their own old_prot variable and that check the return value instead > is rather high, so I think that patch is necessary. We can't start breaking things based only on your suspicion that someone somewhere will have used a feature that isn't documented at all (plus there is a documented way of doing exactly the same thing). If you have an app that needs it sure, otherwise we'll just keep returning a BOOL, as the function is documented to do. > Maybe we should add a winver check, though. No thanks. -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: wcmd fix
by Francois Gouget June 8, 2001

June 8, 2001

On Fri, 8 Jun 2001, Andreas Mohr wrote: > Hi all, > > file_size.XXXPart should be file_size.s.XXXPart. > > Thanks to Lasse Melsaeter for spotting this. That's the 'Dummy struct name problem in wcmd' I sent this morning. But actually whether there should an 's.' depends on the compiler (for Winelib applications) so you should test for NONAMELESSSTRUCT as I did in my patch. -- Francois Gouget fgouget(a)free.fr http://fgouget.free.fr/ If it stinks, it's chemistry. If it moves, it's biology. If it does not work, It's computer science.

1 0

Re: VirtualProtect() fixup
by Alexandre Julliard June 8, 2001

June 8, 2001

Andreas Mohr <a.mohr(a)mailto.de> writes: > VirtualProtect had two big problems: > > - undocumented return value > - should return protection flags only instead of whole flag setting This isn't the case under NT. Which application depends on that? -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: Fwd: Re: WINE fix for HOMM3 [bluefire@zauberwald.net]
by Uwe Bonnes June 7, 2001

June 7, 2001

>>>>> "BlueFire" == BlueFire <bluefire(a)zauberwald.net> writes: BlueFire> Now with the missing diff..... sorry... On Thu, 07 Jun 2001 BlueFire> 11:37:14 BlueFire wrote: >> This is messed up (totally unreadable) formatting. Anyway, how about >> doing a diff -u of your change, and then submitting that patch (in >> more readable form) either to Debian's BTS, or directly to >> wine-patches(a)winehq.com? BlueFire> Sorry, I'm not familiar with either diff or the debian BlueFire> bts. That's why I tried mailing the stuff to you (and to BlueFire> wine-patches). I did a BlueFire> diff /tmp/dos_fs.c dos_fs.c > dos_fs.c.diff Hallo, find some help at http://www.winehq.com/dev.shtml BlueFire> diff -u is preferred due to it's better readability. Appending the plain file makes it easier for others to read your patch. But don't " cut and paste", as that will ofter reformat and make the patch a pain to apply. Bye -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

1 0

Question about the new window design
by Gerard Patel June 7, 2001

June 7, 2001

Since the patch of 4 June, all windows are created as X windows. This makes one of my test apps to crash; it's creating a child window in the WM_NCCREATE event handling of a window. When the child window is created, X bombs because the X window parent does not exist yet - Wine is trying to call XCreateWindow with parent = 0. What is the best way of fixing this ? Gerard

1 0

Re: stack limit
by Alexandre Julliard June 6, 2001

June 6, 2001

Stephen Crowley <stephenc(a)placemark.com> writes: > Works like a charm! Thanks. I'm curious though.. I set the stack size > to 64 megs and the application immediately uses up that much ram. Is > that the default behaviour on windows? What stops it from behaving > like a normal stack and just allocating what's needed from the heap? The ram isn't actually used, it's only reserved from the address space with an anonymous mapping. Only the pages that are really accessed will use physical memory/swap, so it doesn't use any more ram than a normal stack. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: stack limit
by Alexandre Julliard June 6, 2001

June 6, 2001

Stephen Crowley <stephenc(a)placemark.com> writes: > On Mon, Jun 04, 2001 at 12:02:09PM -0700, Alexandre Julliard wrote: > > There is no way to do this at the moment; but it shouldn't be too hard > > to add support for a 'stack' statement in the application spec file. > > Thanks. Where would I look to change this in the source? I'm using some linear > regression DLLs under linux and we have to deal with some fairly large > matrices. Try this: Index: scheduler/process.c =================================================================== RCS file: /opt/cvs-commit/wine/scheduler/process.c,v retrieving revision 1.154 diff -u -r1.154 process.c --- scheduler/process.c 2001/06/06 20:24:13 1.154 +++ scheduler/process.c 2001/06/06 22:46:20 @@ -319,12 +319,6 @@ LPTHREAD_START_ROUTINE entry; WINE_MODREF *wm; - /* build command line */ - if (!ENV_BuildCommandLine( main_exe_argv )) goto error; - - /* create 32-bit module for main exe */ - if (!(current_process.module = BUILTIN32_LoadExeModule( current_process.module ))) goto error; - /* use original argv[0] as name for the main module */ if (!main_exe_name[0]) { @@ -498,7 +492,6 @@ { if (PE_HEADER(current_process.module)->FileHeader.Characteristics & IMAGE_FILE_DLL) ExitProcess( ERROR_BAD_EXE_FORMAT ); - stack_size = PE_HEADER(current_process.module)->OptionalHeader.SizeOfStackReserve; goto found; } @@ -512,6 +505,13 @@ _EnterWin16Lock(); found: + /* build command line */ + if (!ENV_BuildCommandLine( main_exe_argv )) goto error; + + /* create 32-bit module for main exe */ + if (!(current_process.module = BUILTIN32_LoadExeModule( current_process.module ))) goto error; + stack_size = PE_HEADER(current_process.module)->OptionalHeader.SizeOfStackReserve; + /* allocate main thread stack */ if (!THREAD_InitStack( NtCurrentTeb(), stack_size )) goto error; Index: tools/winebuild/README =================================================================== RCS file: /opt/cvs-commit/wine/tools/winebuild/README,v retrieving revision 1.4 diff -u -r1.4 README --- tools/winebuild/README 2000/12/26 01:22:34 1.4 +++ tools/winebuild/README 2001/06/06 22:46:21 @@ -6,6 +6,7 @@ [file WINFILENAME] [mode dll|cuiexe|guiexe|cuiexe_unicode|guiexe_unicode] [heap SIZE] +[stack SIZE] [init FUNCTION] [import [IMP_FLAGS] DLL] [rsrc RESFILE] @@ -39,6 +40,9 @@ "heap" is the size of the module local heap (only valid for Win16 modules); default is no local heap. + +"stack" is the stack size for Win32 exe modules, in kilobytes; default +size is 1024 (1Mb stack). "file" gives the name of the Windows file that is replaced by the builtin. <name>.DLL is assumed if none is given. (This is important Index: tools/winebuild/build.h =================================================================== RCS file: /opt/cvs-commit/wine/tools/winebuild/build.h,v retrieving revision 1.17 diff -u -r1.17 build.h --- tools/winebuild/build.h 2001/05/22 19:55:51 1.17 +++ tools/winebuild/build.h 2001/06/06 22:46:21 @@ -161,6 +161,7 @@ extern int DLLHeapSize; extern int UsePIC; extern int debugging; +extern int stack_size; extern int nb_debug_channels; extern int nb_lib_paths; Index: tools/winebuild/main.c =================================================================== RCS file: /opt/cvs-commit/wine/tools/winebuild/main.c,v retrieving revision 1.17 diff -u -r1.17 main.c --- tools/winebuild/main.c 2001/05/22 19:55:51 1.17 +++ tools/winebuild/main.c 2001/06/06 22:46:21 @@ -27,6 +27,7 @@ int Limit = 0; int DLLHeapSize = 0; int UsePIC = 0; +int stack_size = 0; int nb_entry_points = 0; int nb_names = 0; int nb_debug_channels = 0; Index: tools/winebuild/parser.c =================================================================== RCS file: /opt/cvs-commit/wine/tools/winebuild/parser.c,v retrieving revision 1.18 diff -u -r1.18 parser.c --- tools/winebuild/parser.c 2001/05/22 19:55:51 1.18 +++ tools/winebuild/parser.c 2001/06/06 22:46:21 @@ -559,6 +559,12 @@ if (!IsNumberString(token)) fatal_error( "Expected number after heap\n" ); DLLHeapSize = atoi(token); } + else if (strcmp(token, "stack") == 0) + { + token = GetToken(0); + if (!IsNumberString(token)) fatal_error( "Expected number after stack\n" ); + stack_size = atoi(token); + } else if (strcmp(token, "init") == 0) { if (SpecType == SPEC_WIN16) Index: tools/winebuild/spec32.c =================================================================== RCS file: /opt/cvs-commit/wine/tools/winebuild/spec32.c,v retrieving revision 1.28 diff -u -r1.28 spec32.c --- tools/winebuild/spec32.c 2001/05/22 19:55:51 1.28 +++ tools/winebuild/spec32.c 2001/06/06 22:46:21 @@ -656,7 +656,10 @@ fprintf( outfile, " %ld,\n", page_size ); /* SizeOfHeaders */ fprintf( outfile, " 0,\n" ); /* CheckSum */ fprintf( outfile, " 0x%04x,\n", subsystem ); /* Subsystem */ - fprintf( outfile, " 0, 0, 0, 0, 0, 0,\n" ); + fprintf( outfile, " 0,\n" ); /* DllCharacteristics */ + fprintf( outfile, " %d, 0,\n", stack_size*1024 ); /* SizeOfStackReserve/Commit */ + fprintf( outfile, " %d, 0,\n", DLLHeapSize*1024 );/* SizeOfHeapReserve/Commit */ + fprintf( outfile, " 0,\n" ); /* LoaderFlags */ fprintf( outfile, " %d,\n", IMAGE_NUMBEROF_DIRECTORY_ENTRIES ); /* NumberOfRvaAndSizes */ fprintf( outfile, " {\n" ); fprintf( outfile, " { %s, %d },\n", /* IMAGE_DIRECTORY_ENTRY_EXPORT */ -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: Question about the new window design
by Alexandre Julliard June 6, 2001

June 6, 2001

Gerard Patel <gerard.patel(a)nerim.net> writes: > This makes one of my test apps to crash; it's creating a child window > in the WM_NCCREATE event handling of a window. When the > child window is created, X bombs because the X window parent > does not exist yet - Wine is trying to call XCreateWindow with > parent = 0. > > What is the best way of fixing this ? The X window should be created before the WM_NCCREATE. I'm planning to fix it, but it needs to correctly handle style changes happening in WM_NCCREATE (otherwise we get the managed tooltips bug), and this needs a bit more work. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: uname(1) on UnixWare (Solaris)
by Alexandre Julliard June 6, 2001

June 6, 2001

Bang Jun-Young <bjy(a)mogua.org> writes: > I've nearly completed new DLL configuration code in configure.in. > Before making it available, I need to know what uname(1) command > prints out on UnixWare (and Solaris?). Why do you need that? Using uname(1) in a configure script is almost certainly the wrong thing to do. -- Alexandre Julliard julliard(a)winehq.com

4 9

Re: [PATCH] Flag unicode structure for user create window
by Uwe Bonnes June 6, 2001

June 6, 2001

>>>>> "Gerard" == Gerard Patel <gerard.patel(a)nerim.net> writes: Gerard> In the CreateWindow user driver function, it is not enough to Gerard> test the window procedure to know if the cs structure is unicode Gerard> or not : the window can be subclassed, in the cbt hook, for Gerard> example. Gerard> ChangeLog: Hallo, yesterday I tried to find why the findfile95 dialog only displayed the first character ( "o" versus "open"). Your patch fixed it :-) Bye -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

1 0