May 2002 - Wine-devel - list.winehq.org

Update 'IMPLEMENTING A NEW DLL' in DEVELOPORS-HINTS
by Ian D. Stewart 14 May '02

14 May '02

The attached patch updates Step 3 of 'IMPLEMENTING A NEW DLL' in DEVELOPERS-HINTS with the correct files (and sections of those files) to be modified. Regards, Ian

1 0

Unhandled exception in old DOS application
by M.H.VanLeeuwen 14 May '02

14 May '02

Hi, I hope this is the correct place to ask... Up until the 20020411 wine release my old DOS application worked well...for the last ~2 years. (Thanks all of you!) With both 20020411 and 20020509 releases the wine issues an Unhandled exception report and starts the debugger. bash-2.05$ wine cardload.exe -- -a340 -f"c:\\\winapps\\\winphone\\\cardload.hex" Could not stat /mnt/fd0 (No such file or directory), ignoring drive A: Warning: unprotecting the first 64KB of memory to allow real-mode calls. NULL pointer accesses will no longer be caught. wine: Unhandled exception, starting debugger... err:int:DOSVM_Loop MsgWaitForMultipleObjects returned unexpected value. This application just downloads firmware (cardload.hex) to an old ISA card. What can be done to fix this, where should I begin looking to fix it, who knows if there is a fix already, any ideas how to fix it? I can send the cardload.exe file if anyone is interested, its 48K. Martin ------------------------------------------------------------------------------------ wine --debugmsg +mmsys,+module,+int21 --managed cardload.exe gives: Major SNIP... trace:int21:ioctlGetDeviceInfo (1) trace:int21:DOS3Call returning: AX=4400 BX=0001 CX=7fea DX=0945 SI=36d7 DI=0448 DS=14b2 ES=102f EFL=000b0286 trace:int21:DOS3Call AX=4400 BX=0000 CX=7fea DX=0945 SI=36d7 DI=0448 DS=14b2 ES=102f EFL=000b0286 trace:int21:ioctlGetDeviceInfo (0) trace:int21:DOS3Call returning: AX=4400 BX=0000 CX=7fea DX=0945 SI=36d7 DI=0448 DS=14b2 ES=102f EFL=000b0286 trace:int21:DOSVM_Int21Handler GET INTERRUPT VECTOR 0x1c wine: Unhandled exception, starting debugger... trace:module:CreateProcessA app (null) cmdline "/usr/local/bin/winedbg 134675656 76" trace:module:get_file_name trying '/usr/local/bin/winedbg' -------------------------------------------------------------------------------------- debugger for cardload.exe in the gives me: WineDbg starting... on pid 806fcc8 No debug information in 32bit DLL 'F:\wine' (0x0804a000) Can't find file for ELF '/home/mhvl/wine' (0x00000000) No debug information in 32bit DLL 'NTDLL.DLL' (0x4005a000) No debug information in 32bit DLL 'KERNEL32.DLL' (0x407c4000) No debug information in 32bit DLL 'ADVAPI32.DLL' (0x40822000) No debug information in 32bit DLL 'GDI32.DLL' (0x40740000) No debug information in 32bit DLL 'USER32.DLL' (0x40629000) No debug information in 32bit DLL 'WINMM.DLL' (0x405b3000) No debug information in 32bit DLL 'X11DRV.DLL' (0x40883000) No debug information in 32bit DLL 'WINEOSS.DRV' (0x40b93000) No debug information in 32bit DLL 'MSACM32.DLL' (0x40bb7000) No debug information in 32bit DLL 'MSACM.DRV' (0x40bac000) No debug information in 32bit DLL 'MIDIMAP.DRV' (0x4036b000) No debug information in 32bit DLL 'WINEDOS.DLL' (0x40ce5000) Unhandled exception: page fault on write access to 0x3c8900b0 in 32-bit code (0x40cf12eb). In 32-bit mode. 0x40cf12eb (WINEDOS.DLL.LoadDosExe+0x113b): movb %al,0x0(%edx) Wine-dbg>bt Backtrace: =>0 0x40cf12eb (WINEDOS.DLL.LoadDosExe+0x113b) (ebp=40e26858) 1 0x40cf1435 (WINEDOS.DLL.LoadDosExe+0x1285) (ebp=40e26890) 2 0x40cec388 (WINEDOS.DLL.EntryPoint+0x2c28) (ebp=40e268b0) 3 0x40ced0d0 (WINEDOS.DLL.EntryPoint+0x3970) (ebp=40e268cc) 4 0x40ce969f (WINEDOS.DLL.GetTimer+0x11f) (ebp=40e268e4) 5 0x40ce85e2 (WINEDOS.DLL.ASPIHandler+0x1132) (ebp=40e26904) 6 0x40ce91df (WINEDOS.DLL.ASPIHandler+0x1d2f) (ebp=40e26934) 7 0x40068495 (NTDLL.DLL.__wine_exception_handler+0x45) (ebp=40e26958) 8 0x40067dc5 (NTDLL.DLL.wine_dbg_log+0x165) (ebp=40e2698c) 9 0x400680ef (NTDLL.DLL.RtlRaiseException+0x147) (ebp=40e26a20) 10 0x40071b0f (NTDLL.DLL.RtlConvertSidToUnicodeString+0x109f) (ebp=40e26b64) 11 0x40ce9359 (WINEDOS.DLL.ASPIHandler+0x1ea9) (ebp=40e26c2c) 12 0x40cf0527 (WINEDOS.DLL.LoadDosExe+0x377) (ebp=40e26f14) 13 0x400d8d5a (KERNEL32.DLL.RestoreThunkLock+0x59a) (ebp=40e26f38) 14 0x400d806c (KERNEL32.DLL.WaitForMultipleObjectsEx+0x44c) (ebp=40e26ff4) Wine-dbg>info stack Stack dump: 0x40e26840 (WINEDOS.DLL.LoadDosExe+0x136690): 40e26848 40e2684c 00007754 000040cf 0x40e26850 (WINEDOS.DLL.LoadDosExe+0x1366a0): 40cf7754 40cf6e0c 40e26890 40cf1435 0x40e26860 (WINEDOS.DLL.LoadDosExe+0x1366b0): 0000003c 00006830 000040e2 0000a60d 0x40e26870 (WINEDOS.DLL.LoadDosExe+0x1366c0): 00000000 3c106630 00006830 000040e2 0x40e26880 (WINEDOS.DLL.LoadDosExe+0x1366d0): 00006840 000040e2 40cf7754 0000003c 0x40e26890 (WINEDOS.DLL.LoadDosExe+0x1366e0): 40e268b0 40cec388 0000003c 00000001 0x40e268a0 (WINEDOS.DLL.LoadDosExe+0x1366f0): Wine-dbg>info reg Register dump: CS:0023 SS:002b DS:002b ES:002b FS:0257 GS:0000 EIP:40cf12eb ESP:40e26840 EBP:40e26858 EFLAGS:00010202( R- 00 I - - 1 ) EAX:0000003c EBX:40cf7754 ECX:00006830 EDX:3c8900b0 ESI:0000003c EDI:00000400 Wine-dbg>

1 0

Re: [PATCH] FIXME: Explain how to un-tar, compile, and install Wine from a tarball.
by Tony Lambregts 14 May '02

14 May '02

william stinson wrote: >Hi > >this is my first patch for wine. It is for the documentation > FIXME: Explain how to un-tar, compile, and install Wine from a tarball. > >I compiled the documentation html and it looks OK (no warnings) > >Best regards >William Stinson (wstinson(a)infonie.fr) > >--- wine-20020509/documentation/getting.sgml Mon May 13 23:00:16 2002 >+++ wine-local/documentation/getting.sgml Mon May 13 23:00:34 2002 >@@ -201,7 +201,22 @@ > one. > </para> > <para> >- FIXME: Explain how to un-tar, compile, and install Wine from a >tarball. >+ To install Wine from a tarball you need to first un-tar the archive. >This >+ will create a subdirectory containing the Wine sources. Using the cd >+ command change the current directory to the top level wine directory >+ and finish the installation using the wineinstall command. >+ A typical installation might look like this: >+ <screen> > cut out the next two lines >+$ su - >+<emphasis><Type in root password></emphasis> > **** >+# cd /home/user >+# tar -xvzf Wine-20020310.tar.gz >+# cd wine-20020310 >+# ./tools/wineinstall >+ </screen> >+ More detailed step by step instructions can be found in the README file >+ in the top level wine directory. Note that to install wine you must be >+ logged in as root. > </para> > > <para></para> > > > > This is wrong as of the last two months. wine is not installed as root, in fact ./tools/wininstall will not run as root . I would suggest this as an alternative. --- wine-20020509/documentation/getting.sgml Mon May 13 23:00:16 2002 +++ wine-local/documentation/getting.sgml Mon May 13 23:00:34 2002 @@ -201,7 +201,22 @@ one. </para> <para> - FIXME: Explain how to un-tar, compile, and install Wine from a tarball. + To install Wine from a tarball you need to first un-tar the archive. + this will create a subdirectory containing the Wine sources. Using the + cd command change the current directory to the top level wine directory + and finish the installation using the wineinstall command. + </para> + <para> + A typical installation might look like this: + </para> + <screen> +# cd /home/user +# tar -xvzf Wine-20020310.tar.gz +# cd wine-20020310 +# ./tools/wineinstall + </screen> + <para> + More detailed step by step instructions can be found in the README file + in the top level wine directory. Note: During Installation you will be + asked for the root password. </para> <para></para> Thankyou for bringing this up. Tony Lambregts

1 0

shortcuts doesn't work under wcmd
by Insyde 13 May '02

13 May '02

Hi, I have a system that's working ok except for one more thing (Property sheet is under tests since Guy sends a patch): It execute an external .pif file. I've noticed that it doesn't work, old stuff, but it still works under win2k. It's used to call a bat. I've compiled wcmd to substitutes cmd as it is a link to a .bat file. I made as a test 3 .bat (1.bat, 2.bat and 3.bat) 1 calls 2 2 calls 3 3 prints "DONE!", pause if i execute $wcmd /c 1 it works just the same as win/dos I have a 123.pif that is a link to 1.bat. I execute: $wcmd /c 123.pif wcmd prints: invalid executable. What could it be? Wasn't .pif support implemented? If not, will it? Thanks, Ricardo.

2 1

RFC: CommDlg changes for bug#344
by Ann and Jason Edmeades 13 May '02

13 May '02

Hello, The attached patch resolves the 4 issues raised in bug #344 plus lots more. However, before submitting to wine-patches I would like comments it. One particular area of concern is that the behaviour of commdlg should be different depending on what release of windows is currently being run. Also attached is the program I used to test it with (commenting out various parts as applicable) What is the correct way of handling this? Should I do this? The msdn talks about win2000 behaviour and win98 behaviour. I have no experience with winME to know how this would work so I assumed 98 behaviour. Same with WinXP, so I assumed Win2K behaviour. The code currently does this: win95 isWin98plus = false isWin2Kplus = false win98 isWin98plus = true isWin2Kplus = false winME isWin98plus = true isWin2Kplus = false win2k isWin98plus = true isWin2Kplus = true winxp isWin98plus = true isWin2Kplus = true Fixed: 1. InitialDir handling and InitialFile handling now agrees with MSDN description, and testing on win2000 appears to agree 2. Filename field not cleared when directory selected/clicked on 3. ofNoValidate field no longer causes traps, and appears to behave as per w2k Comments please? No comments within a few days I'll submit to wine-patches and see how many people I can break! Thanks Jason

1 0

Bug tracking organization
by Andreas Mohr 13 May '02

13 May '02

Hi all, I think our bug tracking is still a bit too chaotic. Thus I intend to get some info on what bug tracking should look like. So far, we've got "Wine 0.9.0 TaskList" and "Wine 1.x WishList" as the two main metabugs. IMHO Wine 0.9.0 TaskList is the main bug for all development work up to Wine 1.0. And Wine 1.x WishList is everything that will definitely not be included in Wine 1.0 (and possibly even more). Now how about the layout of Wine 0.9.0 TaskList ? I'd suggest the following layout: 0.9.0 Presentation metabug Documentation Website Stable packaging (.deb, .rpm, .tar.gz) programs metabug (what WineLib programs to supply ?) GUI metabug window management metabug messaging metabug common controls metabug shell namespace fonts/printing metabug multimedia metabug dsound winmm directx (direct3d etc.) opengl video DLL separation metabug RPC/wineserver metabug OLE metabug (ok, not 100% correct, but main hole is DCOM RPC stuff) lowlevel metabug file system metabug registry profile lzexpand memory management metabug loader metabug PE NE DOS metabug comm metabug winsock wininet, icmp, url, urlmon, snmpapi, mapi32, netapi32 serial TAPI "developer stuff" metabug WineLib winedbg regression testing/test suite metabug msvcrt, msvcrt20, crtdll Misc. metabug (for other stuff) twain, winaspi "others" metabug (in order to deposit bug reports of "alien" wine packages here) Is this structure ok ? How should it be reorganized, what should be changed ? If we get to a final result, then I'll make sure the 0.9.0 bug gets to look like that. OK, on to the next topic: How should we organize the bug work ? We need tons of people to take UNCONFIRMED bugs and make sure that they reach enough quality in order to be reassigned as NEW (I expect ~ 30% of all bugs to get dropped right away). The second line of bug tracking would be experienced developers that are mainly responsible for certain areas (like we already started to discuss). There might even be a third line of "emergency" people for the really tough cases... BTW, I'm currently subscribed to wine-bugs. I guess I'll stay subscribed for a while, and once the initial bug tracking structure is solid and doesn't need special attention, I'll be able to concentrate on specific areas, and then I'll probably unsubscribe. Any comments ? -- Andreas Mohr Stauferstr. 6, D-71272 Renningen, Germany Tel. +49 7159 800604 http://home.arcor.de/andi.mohr/

5 5

Attn: component owners list
by Andriy Palamarchuk 13 May '02

13 May '02

Please, take a minute to review the list of component owners and give me a feedback privately or throught the mailing list. We need to define component owners to assign bugs to. These people do not necessarily have to fix the bugs forwarded to them, but will have more detailed knowlege about the status of the component, overview progress in bugs fixing. The component owners will receive bugs after they will be processed by the first line of bugs handling, confirmed, probably already researched, with detailed information. Goal of the whole process is to reduce load on you more experienced developers. You won't have to tell "update your Wine distribution" in 100'th time. A few contributors are already doing this for you ;-) So far only a few people took ownership over separate components: * Guy L. Albetelli - common controls * Dustin Navea - wine tools * Andriy Palamarchuk - wine applications List of candidates: * Alexandre Julliard - wineserver, kernel, IPC * Andreas Mohr - documentation, X11 driver * Dimitrie O. Paun - GUI * Laurent Pinchart - copy protection * Hidenori TAKESHIMA - quartz, DirectShow (is the scope too narrow?), DirectX (?) * Dmitry Timoshkov - NLS, Unicode, keyboard I/O * Francois Gouget - COM, OLE * Martin Wick - sockets * Huw Davies - TrueType support, printing Available components: * DOS support * multimedia * console * files * gdi * loader * net Let me know what you would like to work on, if you agree or not agree to own the component, change scope, know good candidates. I don't want to send bug reports to "black hole", assigning them to a person which is not going to work on them. Look forward for your responses. Andriy Palamarchuk __________________________________________________ Do You Yahoo!? Yahoo! Shopping - Mother's Day is May 12th! http://shopping.yahoo.com

12 19

Re: Fix for commdlg on Mingw
by Marcus Meissner 13 May '02

13 May '02

On Sun, May 12, 2002 at 10:45:08PM -0400, Steven Edwards wrote: > I didn't think wcstol should be moved out of ntddk and in to port.h so I > did the patch this way. If this is another thing that we shouldn't work > around I will add it to the bitch list for the mingw developers that I > am sending them in a week or two. If they don't fix things shortly after > that then I submit my end-all-be-all mingw buggy fix patch to you. > > There are only one or two more places where I will need to do this. > > Changelog: Fix for commdlg. Mingw stdlib.h defines wcstol so we check > for it and use it else we use the normal wine one. Unfortunately wcstol exists in Linux. However it is implemented differently. wide characters are 4 Byte here, not just 2. So your patch will most likely break the Linux side of things. Ciao, Marcus

1 0

Pre-regression regression error :)
by Paul Millar 12 May '02

12 May '02

Hi, Looks like the bug picked up in WRT [1] is genuine (I didn't want to post anything until I was reasonably sure). Juergen shlwapi tests, patch http://cvs.winehq.com/patch.py?id=1020973688889206600449146 seem to have uncovered a bug in registry expansion, but strangely only when wine is emulating an NT-like versions of windows. If you do: export WINEOPTIONS="-winver win2k" make testclean test you find two errors in tests/shreg.c: the tests at line 125 and at line 145. But, if the tests are done with wine in a Win95-like mode: export WINEOPTIONS="-winver win95" make testclean test the tests pass. The errors are: > tests/shreg.c:125: Test failed: (21,20) > tests/shreg.c:145: Test failed: (21,20) which suggests an out-by-one error (the value "21" should be "20"). As far as I've tracked the problem, it seems to be with RegQueryValueExA returning the wrong size. Unfortunately I'm rather pushed for time right now, so if anyone fancies getting to the bottom of this, feel free. Cheers, ---- Paul Millar [1] - http://www.astro.gla.ac.uk/users/paulm/WRT/wrt.php

1 0

Re: Second try for GetCalendarInfo
by Andreas Mohr 12 May '02

12 May '02

On Sat, May 11, 2002 at 11:35:47PM -0400, Vincent Béron wrote: > There's still a couple of things unimplemented, as I don't have access > to a compiler able to produce working code on Windows for me to get the > missing data. (VC98 bails out on _GetCalendarInfoA, and it's the only > one I have easy access to). Err, "man GetProcAddress" ? :-) -- Andreas Mohr Stauferstr. 6, D-71272 Renningen, Germany Tel. +49 7159 800604 http://home.arcor.de/andi.mohr/

3 2