November 2003 - Wine-Devel - list.winehq.org

compiler warnings
by Ivan Leo Murray-Smith Nov. 27, 2003

Nov. 27, 2003

The current code compiles with a few warnings, in the attachment. BTW you can see how code degenerates rapidly when it's not free software any more, probably 10% of all the output when compiling winex 3.2 is compiler warnings.

1 0

Re: Prevent wildcards from being accepted in filenames (NT mode)
by Dmitry Timoshkov Nov. 27, 2003

Nov. 27, 2003

"Shachar Shemesh" <wine-patches(a)shemesh.biz> wrote: > + /* If wer'e in NT mode - don't allow wildcards in file name */ > + if( (strchrW(name, '?') || strchrW(name, '*')) && (GetVersion()&0x80000000)==0 ) (GetVersion() & 0x80000000)==0) construct is very inefficient. It forces compiler generate both a bit mask test and a comparison operation. While many modern processors have very effective bit test instructions, and we have to tell to the compiler that we want only a bit test. Simple !(GetVersion() & 0x80000000) is much better IMO. -- Dmitry.

7 8

More spare time = New Winetests build
by Jakob Eriksson Nov. 27, 2003

Nov. 27, 2003

winetests has still not made it to CVS so one month later I have another build for you: Tests built with MSVC 7 from CVS 20031127. (Testing shell cross built with mingw32.) http://vmlinux.org/jakov/Wine/ And as Brian Vincent put it: "All you have to do is download winetests.exe and just run it. All of the collected results get displayed in a browser window with the option of submitting them to the wine-tests-results mailing list." Don't be afraid it the screen turns blank for a second. It's normal. regards, Jakob .

1 0

winearts & KDE 3.2-beta1
by Kevin Koltzau Nov. 27, 2003

Nov. 27, 2003

I decided to give KDE 3.2-beta1 a try (just for reference I'm using the Gentoo ebuild), and found a problem with compiling winearts under KDE 3.1 "artsc-config --cflags" gives me -I/usr/kde/3.1/include/artsc but under KDE 3.2-beta1 I get -I/usr/kde/3.2/include/artsc -pthread -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include which causes a problem because this value is passed to makedep during compilation, which fails with "Unknown option '-pthread'"

2 1

Debug channel list
by Mike Hearn Nov. 27, 2003

Nov. 27, 2003

Here's a more readable version of what I just posted to wine-patches. There are quite a few, aren't there? Full marks to anybody who can actually describe what they all monitor :) accel adpcm advapi animate aspi atom avicap avifile bidi bitblt bitmap cabinet capi caret cdrom cfgmgr32 class clipboard clipping combo comboex comm commctrl commdlg computername console crtdll crypt curses cursor d3d d3d_shader d3d_surface datetime dc ddeml ddraw ddraw_fps ddraw_geom ddraw_tex debugstr devenum dialog dinput dll dma dmband dmcompos dmfile dmfiledat dmime dmloader dmscript dmstyle dmsynth dmusic dosfs dosmem dplay dplayx dpnhpast driver dsound dsound3d edit enhmetafile environ event eventlog exec file fixup font fps g711 gdi global glu graphics header heap hook hotkey icmp icon imagehlp imagelist imm int int21 int31 io ipaddress iphlpapi jack joystick key keyboard listbox listview loaddll local mapi mci mcianim mciavi mcicda mcimidi mciwave mdi menu menubuilder message metafile midi mmaux mmio mmsys mmtime module monthcal mpeg3 mpr msacm msdmo msg mshtml msi msimg32 msisys msrle32 msvcrt msvideo mswsock nativefont netapi32 netbios nls nonclient ntdll odbc ole oledlg olerelay opengl pager palette pidl powermgnt print process profile progress propsheet psapi psdrv qcap quartz ras rebar reg region relay resource richedit rundll32 sblaster scroll seh selector server setupapi shdocvw shell shlctrl snmpapi snoop sound static statusbar storage stress string syscolor system tab tape tapi task text thread thunk tid timer toolbar toolhelp tooltips trackbar treeview ttydrv twain typelib uninstaller updown urlmon uxtheme ver virtual vxd wave wc_font win win32 wineboot winecfg wineconsole wine_d3d winevdm wing winhelp wininet winmm winsock winspool wintab wintab32 wnet x11drv x11settings xdnd xrandr xrender xvidmode

1 0

Re: Fun projects?
by Ivan Leo Murray-Smith Nov. 27, 2003

Nov. 27, 2003

>The easiest way is simply to convert the Gimp into a >Windows program by compiling it with WineLib >That means that in order to use Photoshop >plugins in the Gimp you'd need a special build of the Gimp Special build? Wouldn't it be easier to use the native version of gimp for windows?

2 1

Re: Winealsa : add support for Alsa 1.0
by Dimitrie O. Paun Nov. 27, 2003

Nov. 27, 2003

On November 27, 2003 06:21 am, Sylvain Petreolle wrote: > Is this second attempt better ? > (not tested with alsa 0.9) Why don't you define the snd_pcm_hw_params_* macros for alsa 0.9. For example, this bit: - snd_pcm_format_t format = snd_pcm_hw_params_get_format(hw_params); - snd_pcm_access_t access = snd_pcm_hw_params_get_access(hw_params); + snd_pcm_format_t format; + snd_pcm_access_t access; + snd_pcm_uframes_t buffer_size, bufsize_min, bufsize_max; + int dir=0, channels, channels_min, channels_max; + long int period_size, periods_min=0, periods_max=0; + +#if (SND_LIB_MAJOR == 0) + format = snd_pcm_hw_params_get_format(hw_params); + access = snd_pcm_hw_params_get_access(hw_params); +#else + snd_pcm_hw_params_get_format(hw_params,&format); + snd_pcm_hw_params_get_access(hw_params,&access); +#endif Could have been coded like so: +#if (SND_LIB_MAJOR == 0) +# define snd_pcm_hw_params_get_format(hw_params,format) do{ *format = snd_pcm_hw_params_get_format(hw_params);} while(0) +# define snd_pcm_hw_params_get_access(hw_params,access) do{ *access = snd_pcm_hw_params_get_access(hw_params);} while(0) ... more snd_pcm_hw_params_* definitions +#endif ... - snd_pcm_format_t format = snd_pcm_hw_params_get_format(hw_params); - snd_pcm_access_t access = snd_pcm_hw_params_get_access(hw_params); + snd_pcm_format_t format; + snd_pcm_access_t access; + snd_pcm_uframes_t buffer_size, bufsize_min, bufsize_max; + int dir=0, channels, channels_min, channels_max; + long int period_size, periods_min=0, periods_max=0; + + snd_pcm_hw_params_get_format(hw_params,&format); + snd_pcm_hw_params_get_access(hw_params,&access); Eliminates a lot of #ifdefs, cleaner code, etc. -- Dimi.

2 1

Wine lecture slides
by Shachar Shemesh Nov. 27, 2003

Nov. 27, 2003

Hi all, The updated slides for the lecture I'll be giving on Sunday are at http://www.shemesh.biz/wine/slides.pdf. I tried to incorporate all of the feedback from you guys. I would, still, appretiate any further feedback you may wish to give. I also have specific questions, if anyone knows the answers: - What is the current status of "--with-nptl" on RH9? Is it positively necessary to include that when running ./configure? I know it's not necessary when running wineinstall. - Can someone please verify my knowledge of Corel's wine history Hmm, that list turned out shorter Corel-wine: As far as I could discern from the good ol' web, Corel started it's Wine initiative in 1999 as an alternative to actually porting WordPerfect and friends to Linux. They created a branch called "Corel-wine", which was also open source (wine was X11 at the time, so they did have the choice). At some point, corel-wine died. Why did it die? When did that happen? Shachar -- Shachar Shemesh Open Source integration & consulting Home page & resume - http://www.shemesh.biz/

3 2

MSVC build log of wine conformance tests
by Jakob Eriksson Nov. 27, 2003

Nov. 27, 2003

MSVC build log of wine conformance tests: http://vmlinux.org/jakov/Wine/build_log_MSVC_CVS_2003-11-27.txt All in all, 14 tests compile and 7 do not. AFAIK I use an up to date VC++ 7. (I reinstalled Windows though, so I _might_ have missed something.) regards, Jakob

1 0

Re: Winealsa : add support for Alsa 1.0
by Alexandre Julliard Nov. 27, 2003

Nov. 27, 2003

Sylvain Petreolle <spetreolle(a)yahoo.fr> writes: > this patch is an attempt to provide Alsa 1.0 support into winealsa. > > the file audio_09.c is the current audio.c for alsa 0.9 with small > modifications to > ensure compatibility with the new alsa API. I don't think we want yet another duplicate of that code. Please try to support both in the same file, there aren't that many differences. -- Alexandre Julliard julliard(a)winehq.com

2 2

Re: RFH: *ReAlloc* mess
by Juan Lang Nov. 27, 2003

Nov. 27, 2003

NetApiBufferReallocate: - does not allocate memory when NULL is the original pointer passed in - frees memory when the new size is 0 HTH, --Juan __________________________________ Do you Yahoo!? Free Pop-Up Blocker - Get it now http://companion.yahoo.com/

1 0

Descrepancy between Windows and GLIBC version of [v]snprintf
by Daniel Gehriger Nov. 26, 2003

Nov. 26, 2003

Hi, I found a problem with the _snprintf() / _vsnprintf() functions. Wine seems to use the forward these function calls to their native implementation (eg. in GLIBC). The problem is that they behave differently under Windows and GLIBC: Both function take a parameter to specify the maximum number of characters to write. Under Windows, if the formated string is equal in length to or exceeds the specified number, the terminating NULL will NOT be written to the buffer. Under Linux, the terminating NULL is ALWAYS written. The test program below produces the following output (only when linked dynamically, of course) : Under Windows: 0000000001**** Under Wine: 000000000 When compiled on Linux: 000000000 --------------------------------------------------------- #include <stdio.h> #include <memory.h> int main() { char buffer[16]; memset(buffer, 0, 16); memset(buffer, '*', 15); int len = _snprintf(buffer, 10, "%010d", 1); printf("%d : %s\n", len, buffer); return 0; } --------------------------------------------------------- Regards, Daniel

3 3

PRIVATE flag sends dllwrap for a loop.
by Steven Edwards Nov. 26, 2003

Nov. 26, 2003

Hello, This is a problem I have had for a while building WINE dlls for use on Windows and ReactOS. When linking the dlls dllwrap always complains that it cant find the symbol private that is added to the .def file. When I remove this symbol and rebuild my import libs and dlls I dont have any problems so lets disable it for now. Changelog: Disable PRIVATE name on for exports in def files. Index: spec32.c =================================================================== RCS file: /home/wine/wine/tools/winebuild/spec32.c,v retrieving revision 1.72 diff -u -r1.72 spec32.c --- spec32.c 3 Nov 2003 22:19:44 -0000 1.72 +++ spec32.c 20 Nov 2003 18:21:56 -0000 @@ -900,7 +900,9 @@ fprintf( outfile, " @%d", odp->ordinal ); if (!odp->name) fprintf( outfile, " NONAME" ); if (is_data) fprintf( outfile, " DATA" ); +#if 0 if (odp->flags & FLAG_PRIVATE) fprintf( outfile, " PRIVATE" ); +#endif fprintf( outfile, "\n" ); } } __________________________________ Do you Yahoo!? Free Pop-Up Blocker - Get it now http://companion.yahoo.com/

2 3

Re: netapi32: implement Netbios()
by Juan Lang Nov. 26, 2003

Nov. 26, 2003

Hi Uwe et al.. Uwe said: > Saying a little more may help. > > What's needed to test you patch? What programs show > improvement? Good questions. I have little test programs that I've been using, but NetBIOS is hard to test without knowing something about your network configuration, so they're not much use for netapi32/tests. I don't know of any programs that work now that didn't, except what I'm working on: I'm trying to implement the Network Neighborhood. I've been tweaking WINE's SMB code to use Netbios() rather than implementing the netbios-over-tcpip stuff directly. That work isn't done yet, watch this space for more :) There was an article a while back linked here about some HP management program (for their NAS appliance) that didn't discover their box under WINE because of lack of NetBIOS support. I suspect this may now work, but I don't know. > Any configuration items needed in a .ini file or the > registry? Needed? No, should work as-is. You may configure some things in the registry if you need to. Many are configured in the same way as Win98's reg, but not all. I didn't know a good place to document these; comments in winedefault.reg??? > And thanks for your work :-) You're very welcome :) --Juan __________________________________ Do you Yahoo!? Free Pop-Up Blocker - Get it now http://companion.yahoo.com/

1 0

Re: netapi32: implement Netbios()
by Uwe Bonnes Nov. 26, 2003

Nov. 26, 2003

>>>>> "Juan" == Juan Lang <juan_lang(a)yahoo.com> writes: Juan> Apologies for the size of the patch, I didn't see any useful way Juan> to shrink it. Juan> ChangeLog: implement rather a lot of Netbios() Juan> I could say a bit more than that, but it wouldn't mean much to Juan> most people. Where there was nearly none, now there is some. Saying a little more may help. What's needed to test you patch? What programs show improvement? Any configuration items needed in a .ini file or the registry? Just some thoughts. And thanks for your work :-) 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

debugger problems
by Chris Morgan Nov. 26, 2003

Nov. 26, 2003

I'm trying to debug a crash in some changes to the jack audio driver and can't seem to have the wine debugger show me the actual source code of the crash. I start up the debugger and get this: oaded debug information from ELF 'wine' ((nil)) No debug information in 32bit DLL 'C:\Program Files\winamp\Winamp.exe' (0x400000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\NTDLL.DLL' (0x401d0000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\KERNEL32.DLL' (0x404b0000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\ADVAPI32.DLL' (0x408c0000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\GDI32.DLL' (0x40840000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\USER32.DLL' (0x406f0000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\RPCRT4.DLL' (0x40a00000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\OLE32.DLL' (0x40980000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\SHLWAPI.DLL' (0x40a50000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\COMCTL32.DLL' (0x40aa0000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\SHELL32.DLL' (0x40900000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\WINMM.DLL' (0x40b50000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\WINSPOOL.DRV' (0x40c40000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\COMDLG32.DLL' (0x40bd0000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\OLEAUT32.DLL' (0x40c80000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\X11DRV.DLL' (0x40e10000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\WINEJACK.DRV' (0x410e0000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\MSACM32.DLL' (0x41130000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\MSACM.DRV' (0x41120000) No debug information in 32bit DLL 'C:\WINDOWS\SYSTEM\MIDIMAP.DRV' (0x41270000) In 32-bit mode. Wine-dbg> I run and get a crash: Backtrace: =>0 0x41101541 (WINEJACK.DRV..data+0x14541) (ebp=42b87748) 1 0x410e6cbf (WINEJACK.DRV..text+0x5cbf in WINEJACK.DRV) (ebp=42b877cc) 2 0x410e7e10 (WINEJACK.DRV..text+0x6e10 in WINEJACK.DRV) (ebp=42b87804) 3 0x410e9111 (WINEJACK.DRV.wodMessage+0x131 in WINEJACK.DRV) (ebp=42b8783c) 4 0x40b70999 (WINMM.DLL.joySetThreshold+0xd89 in WINMM.DLL) (ebp=42b878ac) 5 0x40b711a4 (WINMM.DLL.joySetThreshold+0x1594 in WINMM.DLL) (ebp=42b878dc) 6 0x40b813e6 (WINMM.DLL.midiStreamStop+0x266 in WINMM.DLL) (ebp=42b87948) 7 0x40b818f4 (WINMM.DLL.waveOutOpen+0x54 in WINMM.DLL) (ebp=42b87974) 8 0x41121362 (MSACM.DRV..text+0x362 in MSACM.DRV) (ebp=42b879a8) 9 0x4112158b (MSACM.DRV..text+0x58b in MSACM.DRV) (ebp=42b87a14) 10 0x411225e7 (MSACM.DRV.wodMessage+0xe7 in MSACM.DRV) (ebp=42b87a4c) 11 0x40b70999 (WINMM.DLL.joySetThreshold+0xd89 in WINMM.DLL) (ebp=42b87abc) 12 0x40b711a4 (WINMM.DLL.joySetThreshold+0x1594 in WINMM.DLL) (ebp=42b87aec) 13 0x40b813e6 (WINMM.DLL.midiStreamStop+0x266 in WINMM.DLL) (ebp=42b87b58) 14 0x40b818f4 (WINMM.DLL.waveOutOpen+0x54 in WINMM.DLL) (ebp=42b87b84) 15 0x42a69ca6 (DSOUND.DLL.DirectSoundCreate8+0x5d6 in DSOUND.DLL) (ebp=42b87c44) 16 0x425f90eb (OUT_DS.DLL..text+0x10eb in OUT_DS.DLL) (ebp=00000000) 0x41101541 (WINEJACK.DRV..data+0x14541): movb %al,0x0(%edx) Wine-dbg>list Enter path to file '': What am I doing wrong here? Chris

3 3

Re: Intresting page
by Ivan Leo Murray-Smith Nov. 26, 2003

Nov. 26, 2003

> Hmm. I'd be wary of trusting a page with such obviously manipulated > screenshots (check out the different background colours/fonts of some of > the labels), and "Write in HERE or die LMAO!" as the link to the > guestbook :-) I think they manipulated the screenshots to avoid someone hosting them and saying "I took this screenshot on my pc". Also, if you look on google you can find links to pages that talk about neptune, nashville and so on.

1 0

Intresting page
by Ivan Leo Murray-Smith Nov. 26, 2003

Nov. 26, 2003

Lists (And screenshots) of lots of stuff m$ never released, like windows neptune, Whistler Advanced Server, Cairo, Nashville (Windows 96), windows titanium, windows 99, and more. http://ad-iso.mine.nu/~air101/

2 1

PE stack trick not working on lowmem machines
by Marcus Meissner Nov. 26, 2003

Nov. 26, 2003

Hi, On SuSE 8.2, i586 glibc, loader/wine-* crash due to the 256 MB static array in main(). The test machine has just 64MB RAM and 200MB swap. If I start decreasing the array it works, up to 201MB. $ /lib/ld-linux.so.2 ./wine-kthread ./wine-kthread: error while loading shared libraries: ./wine-kthread: cannot map zero-fill pages: Cannot allocate memory $ So ld.so really wants to get 20xMB memory but fails. -> Not good on low memory machines. (Not sure if we should fix it.) Btw, why don't we use -Wa,--noexecstack ? Ciao, Marcus

3 3

Re: Reject wildcards in directory names
by Rolf Kalbermatter Nov. 26, 2003

Nov. 26, 2003

Alexandre Julliard <julliard(a)winehq.org> wrote: > > As for the invalid characters, [Create|Remove]File[A|W] only check > > for '?', '*', which I just copied for [Create|Remove]Directory[A|W]. > > I take it you will accept a patch that also extends the checks in > > [Create|Remove]File*? > > Yes, and of course it should use a common routine to do the checks, so > that it's consistent everywhere. Also the Create and Remove cases > should be handled differently IMO, a Remove should succeed if the > specified file exists, even if it contains illegal characters. I would actually agree with Alexandre here, but W2K at least behaves differently. My tests show that RemoveDirectory just as CreateDirectory rejects all paths which contain one of the wildcards ": * ? \" < > |" with error 123 completely independent where that character occurs (not just in the path element which is created). And as far as I can say DeleteFile just as much behaves also this way. So should we attempt to be smarter than Windows in this case? Rolf Kalbermatter

5 12

Re: winehq name
by Alexandre Julliard Nov. 26, 2003

Nov. 26, 2003

"Dimitrie O. Paun" <dpaun(a)rogers.com> writes: > I know we've discussed this before, but I feel things > have changed and it's worth another try :) Namely, > have the winehq.org name be the default one. It seems > virtually everybody (except you) prefer the .org name > over the .com name, it's way more community friendly, > and it seems to be the default one on WineHQ for a > while, without any negative side effects. It's not that I prefer the .com name, frankly I couldn't care less, it's just that I think it's silly to change a perfectly good name that everybody was used to, simply to achieve some questionable ideal of "purity" of the domain name; and even more silly to have to change all references to try to reduce the confusion that would never have happened if we hadn't changed it in the first place. But I don't really care enough to keep fighting about it, so I put your patch in... -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: wine/programs/winedbg debug.l
by Dimitrie O. Paun Nov. 26, 2003

Nov. 26, 2003

On November 25, 2003 11:10 pm, Alexandre Julliard wrote: > Modified files: > programs/winedbg: debug.l > > Log message: > Fixed one more HeapReAlloc call. Good catch! I didn't lie when I said all calls were reviewed, it's just that I didn't search .l files... -- Dimi.

1 0

RE: wine/ windows/win.c windows/message.c programs ...
by Robert Shearman Nov. 26, 2003

Nov. 26, 2003

> -----Original Message----- > From: wine-cvs-admin(a)winehq.com [mailto:wine-cvs-admin(a)winehq.com]On > Behalf Of Alexandre Julliard > Sent: 26 November 2003 04:09 > To: wine-cvs(a)winehq.com > Subject: wine/ windows/win.c windows/message.c programs ... > > Log message: > Fixed declarations of BroadcastSystemMessage and > TranslateAccelerator > in winuser.h. Added GetWindowModuleFileName. > +LONG WINAPI BroadcastSystemMessageA(DWORD,LPDWORD,UINT,WPARAM,LPARAM); > +LONG WINAPI BroadcastSystemMessageW(DWORD,LPDWORD,UINT,WPARAM,LPARAM); > +#define BroadcastSystemMessage WINELIB_NAME_A(BroadcastSystemMessage) Should this not be WINELIB_NAME_AW? Rob

1 0

Strange problem with wine
by Gerhard W. Gruber Nov. 25, 2003

Nov. 25, 2003

I'm using a build that is about one month old, but I don't think that this is the problem. I will update in the next few days and then I will see if this goes away. OK. I have written this small tool which you can download at http://sourceforge.net/projects/launchmen. The Linux port uses a normal fork() and then executes the specifed application when the entry is selected from the menu. It works for every application but only with wine/Agent I have to select the entry twice to make it run. When I select the entry wich should run wine with Agent nothing happens. When I look into the process list the first time I can see this: 1506 ? S 0:00 /usr/local/bin/wine /home/user/wine/D/home/agent/agent.exe 1507 ? S 0:00 \_ wineserver 1508 ? S 0:00 \_ wineserver Which is strange because I have two wineservers running. When I launch exactly the same commandline from the shell everything is normal. I see one Agent, one wineserver and Agent comes up fine. In the case of my application I can see the above list and Agent doesn't start. Only when I launch Agent a second time from my app then it starts fine. When I exit Agent then and look again I can still see the old Agent hanging, but as long as this one is there my app can launch Agent normally. When I kill this Agent and wineserver the same happens again and I have to start it twice. Any ideas what this could be? Since I only have this problem with wine I thought I ask if somebody could say something about this. -- Gerhard Gruber Maintainer of SoftICE for Linux - http://sourceforge.net/projects/pice Fast application launcher - http://sourceforge.net/projects/launchmenu

1 0

Re: unicows update and PE memory format
by Rolf Kalbermatter Nov. 25, 2003

Nov. 25, 2003

Message: 4 Subject: Re: PE stack trick not working on lowmem machines From: Mike Hearn <mike(a)theoretic.com> To: Alexandre Julliard <julliard(a)winehq.org> Cc: Marcus Meissner <meissner(a)suse.de>, wine-devel(a)winehq.com Date: Mon, 24 Nov 2003 21:00:33 +0000 On Mon, 2003-11-24 at 20:38, Alexandre Julliard wrote: > > Btw, why don't we use -Wa,--noexecstack ? > > It doesn't help with prelink AFAICT. Is it really too late to get some kind of hook into prelink to stop it interfering with Wine? It seems daft that we need such a big hack when surely a flag in the elf headers would work just as well with co-operation from the rtld. --__--__-- >Shachar Shemesh <wine-devel(a)shemesh.biz> writes: >Now here's the puzzle - GetProcAddress is never called. Neither are >these functions imported from the DLL in the PE header. In fact, >unicows.dll does not even appear in the PE header (which makes sense - >on NT you don't even want to install it). This means that unicows.lib >manages to call GetProcAddress, without actually calling GetProcAddress. > >Some tracing leads me to suspect (I have not done enough research to >determine yet) that it accesses the PE header of kernel32 directly. It >appears that unicows.dll enountered the exact same problem we have vis a >vis GetProcAddress, and have happily hacked their way around it using >direct memory access. What I see them do is call "GetModuleHandle" on >"kernel32.dll", and then use the resulting handle as a struct/array >pointer, and access offsets into it. This is not exactly very nice but it is also not to surprising. The PE header is actually documented on MSDN including the import and export tables. GetModuleHandle actually also just returns the HMODULE which at least for Win32 is the same as the HINSTANCE returned by LoadLibrary. The difference is if I'm not mistaken, that GetModuleHandle will usually not try to load the module if it isn't already loaded, but for kernel32 you can safely assume that a process has it already loaded implicitedly or even explicitedly. An HMODULE on Win32 is nothing else than the base address of a loaded module image, and with some offset calculations using Macros from winnt.h you can get at anything in that image including the export table and from there you can then search GetProcAddress. One possible problem is that the unicows GetProcAddress implementation might not search the function in the export table by name but rather by ordinal and our ordinal or ordinal base is different from what Unicows expects. Rolf Kalbermatter

2 1