July 2004 - Wine-Devel - list.winehq.org

taskmgs hangs on edit debug channels
by Andrei Barbu July 2, 2004

July 2, 2004

The taskmanager dies when trying to edit the debug channels of any programs with the newest CVS (checked out 2 days ago) It just sits there with this: 002e:Ret x11drv.MsgWaitForMultipleObjectsEx() retval=00000102 ret=409cd43c 002e:Call kernel32._ConfirmSysLevel(40a08ac0) ret=409a3f15 002e:Ret kernel32._ConfirmSysLevel() retval=00000000 ret=409a3f15 002e:Call kernel32.GlobalLock16(0000116f) ret=40999b70 002e:Ret kernel32.GlobalLock16() retval=404326d0 ret=40999b70 002e:Call kernel32.ReleaseThunkLock(41625f7c) ret=409cd30f 002e:Ret kernel32.ReleaseThunkLock() retval=00000000 ret=409cd30f 002e:trace:msg:GetMessageW (116f) mask=0000007f, bits=00000000, changed=00000000, waiting 002e:Call kernel32.ReleaseThunkLock(41625fe4) ret=409cd684 002e:Ret kernel32.ReleaseThunkLock() retval=00000000 ret=409cd684 002e:Call x11drv.MsgWaitForMultipleObjectsEx(00000001,404326d4,ffffffff,00000000,00000000) ret=409cd6c6 002e:Call ntdll.RtlEnterCriticalSection(40d76680) ret=40d54902 002e:Ret ntdll.RtlEnterCriticalSection() retval=00000000 ret=40d54902 002e:Call ntdll.RtlLeaveCriticalSection(40d76680) ret=40d54932 002e:Ret ntdll.RtlLeaveCriticalSection() retval=00000000 ret=40d54932 002e:Call ntdll.RtlEnterCriticalSection(40d76680) ret=40d54902 002e:Ret ntdll.RtlEnterCriticalSection() retval=00000000 ret=40d54902 002e:Call ntdll.RtlLeaveCriticalSection(40d76680) ret=40d54932 002e:Ret ntdll.RtlLeaveCriticalSection() retval=00000000 ret=40d54932 002e:Call kernel32.WaitForMultipleObjectsEx(00000002,41625dac,00000000,ffffffff,00000000) ret=40d383db 002e:Call ntdll.NtWaitForMultipleObjects(00000002,41625ba0,00000000,00000000,00000000) ret=40591c85 If I move the mouse it spits out the correct mouse messages which indicate that wine itself is running just fine. The taskmanager window doesn't get redrawn (the black backround around it does). The backtrace from that point is: =>1 0x400636a9 (0x41b2be48) 2 0x402335c2 NTDLL_wait_for_multiple_objects+0x122(count=0x1, handles=0x41b2bf38, flags=0x4, timeout=0x0) [sync.c:587] in ntdll (0x41b2beec) 3 0x40233636 NtWaitForMultipleObjects+0x66(count=0x1, handles=0x41b2bf38, wait_all=0x0, alertable=0x0, timeout=0x0) [sync.c:605] in ntdll (0x41b2bf14) 4 0x40591c85 WaitForMultipleObjectsEx+0xb5(count=0x1, handles=0x41b2c06c, wait_all=0x0, timeout=0xffffffff, alertable=0x0) [sync.c:171] in kernel32 (0x41b2c044) 5 0x40591b1a WaitForSingleObject+0x3a(handle=0x6c, timeout=0xffffffff) [sync.c:99] in kernel32 (0x41b2c064) 6 0x40656867 ProcessPageRefreshThread+0x87(lpParameter=0x0) [procpage.c:511] in taskmgr (0x41b2c1a4) 7 0x40598fba THREAD_Start+0x11a(ptr=0x40450220) [thread.c:107] in kernel32 (0x41b2c27c) 8 0x40234fe9 start_thread+0x199(info=0x40450240) [thread.c:198] in ntdll (0x41b2cacc) 9 0x4005df70 (0x41b2cbcc) 10 0x4017e40a (0x00000000) Any ideas? I'm not quite versed enough in wine to know what's happening too well. I realise it's a thread waiting for another that apparently never finishes.. or I think it is, and I'm kind of lost from there. Thanks, Andrei

2 1

Delayed breakpoints in winedbg
by Adrian Rees July 2, 2004

July 2, 2004

Hi all, Apologies in advance for what is probably a dumb question... I'm trying to debug a windows application by breaking on a call to a DLL function - in fact it's the NtCreateKey function within the ntdll library. I must be missing something fundamental because the debugger indicates that it has set a delayed breakpoint, but the breakpoint is never reached, although turning on debugging messages clearly indicates that the function is being called. Here's a typical attempt: $ WINDEBUG=reg winedbg MSMONEY.EXE ... (lots of startup calls to reg) ... In 32 bit mode. Wine-dbg>break NtCreateKey fixme:dbghelp:elf_new_wine_thunks Duplicate in midimap.drv<elf>: _edata<5b828a10 -00000000> completed.1<5b828a10-00000000> ... (snip) ... No symbols found for NtCreateKey Unable to add breakpoint, will check again when a new DLL is loaded Wine-dbg>info break No breakpoints No watchpoints Delayed breakpoints: 0: NtCreateKey Wine-dbg>cont trace:reg:NtCreateKey(0x28,L"Software\\Microsoft\\Windows\\CurrentVersion\\AppPaths\\MSMONEY.EXE",L"",1,1,0xb9fd04) trace:reg:NtCreateKey <- 0x50 trace:reg:RegQueryValueExA(0x50,"Path",(nil),0xb9fcf8,0xb9f318,0xb9fd00=260) ... (etc till program ends) ... I've spent a couple of hours searching for more information on how to use this capability of the debugger, but with no luck - can anyone tell me what I'm doing wrong? Thanks, Adrian

2 3

Re: collation tables
by Dmitry Timoshkov July 2, 2004

July 2, 2004

"Jonathan Wilson" <jonwil(a)tpgi.com.au> wrote: > Given that simply calling whatever the locale functions are on windows and > using their results is considered too legally risky (because the results > are almost certainly covered under a MS copyright), the best solution is to > simply take the best possible implementation we have now then build a > testcase. Build a testcase that calls the locale functions and checks that > the output matches. From there, you can fix WINE untill it passes the tests. > > This is how it is done for most other things in WINE, so why not this? Look at the dlls/kernel/tests/locale.c,test_CompareStringA/test_LCMapStringA > Another way is to look at what ReactOS has, do they have collation tables? > Are they good? Can WINE use them? (I would assume they would need to be > relicenced but other than that...) I don't think ReactOS has anything except .nls and case mapping files generated with the tool I wrote. Even those files aren't used AFAIK. -- Dmitry.

1 0

Re: winedbg crashes
by Jon Griffiths July 2, 2004

July 2, 2004

Hi Eric, <cc'd to wine-devel so anyone with a similar problem can google for the fix> > no, it's normal you get this. In fact, we should simply abort the > loading. > Does this work better ? I can't test it unfortunately, but it looks like the right thing to do. I asked on bugs(a)binutils and was told to upgrade to binutils 2.15 to fix the bad stab generation. Sure enough, this fixes the issue (which means I can't duplicate this behaviour any more). It also fixed a bunch of other random crashes in my app; There were obviously some needed fixes for huge (40 MB in this case) .so generation. So perhaps printing a message to upgrade binutils to >=2.15 when the stabs are found to be corrupted would be appropriate? I'm not sure how prevalent this version is in the wild (my mdk 10 community uses 2.13). I'm investigating now to see if debugging works as expected (so far, so good) Thanks, Jon ===== "Don't wait for the seas to part, or messiahs to come; Don't you sit around and waste this chance..." - Live jon_p_griffiths(a)yahoo.com __________________________________ Do you Yahoo!? Yahoo! Mail Address AutoComplete - You start. We finish. http://promotions.yahoo.com/new_mail

1 0

IE6 installation error. would you please give me a suggession
by Shiva Ram Krishna July 2, 2004

July 2, 2004

Dear Sir/Madam, I downloaded the latest cvs of wine and compiled it. All went well but it seems that the regsvr32.exe.so file wasn't created properly. If check the dependency of expand.exe.so, winver.exe.so, winepath.exe.so, I can see all the proper files that it needs but if I do it on regsvr32.exe.so, it says "regsvr32.exe.so: Not a dynamic executable". I have that his problem for the last 3-4 months. I used a text editor and it seems that the header is the regsvr32.exe.so looks like an msdos .exe file (ex, first 2 characters have "MZ") but expand,winver, and winepath start with "^^". The reason I have notified it is when I run ie 5/6, It stops at the point saying "/usr/local/bin/wine: cannot open builtin library for 'C:\windows\system\regsvr32.exe': /usr/local/lib/wine/regsvr32.exe.so: invalid ELF header". would you please suggest me what I have to do ________________________________________________________________________ Yahoo! India Careers: Over 50,000 jobs online Go to: http://yahoo.naukri.com/

1 0

Re: [Mono-winforms-list] Fw: [Mono-list] System.Windows.Forms plans.
by Peter Dennis Bartok July 2, 2004

July 2, 2004

>You are asking for stable interfaces and those can only come once there >is a Wine 1.0. Its not here and dont get me wrong I think the Wine TODO >list is way to long and things should be close to frozen by now but >thats not the case. All you are doing is pointing out that its not >stable which is well known. I wish it was stable now but its not and >until the TODO is done its not going to be. Paul's message was very specific, asking about a versioning interface (which I would have loved to have, too, instead of trying to guess in code what version of Wine I might be dealing with). If very reasonable API requests are being refused while it is still under heavy development why would those be granted once 1.0 is reached? And, thus, how is working down the TODO going to help his request? With SWF we luckily have a choice whether or not use use Wine, we tried with and it didn't quite work out, so now we're gonna try without. Paul isn't as lucky, his code is dependent on the Win32 API. I would have thought that Alexandre is exited that his code is being put to new, cool uses even he never imagined... With that I should probably stop participating in this thread, since it's really not contributing to SWF development :-) Regards, Peter

1 0

Re: [Mono-winforms-list] Fw: [Mono-list] System.Windows.Forms plans.
by Steven Edwards July 2, 2004

July 2, 2004

Hello Peter, I am not trying to be rude but... --- Peter Dennis Bartok <peter(a)novonyx.com> wrote: > >Rather than spending a whole lot of time rewritting things wouldnt > it > >be better to help the Wine project get stable and move to the 1.0 > goal? > >There is a roadmap and TODO on Winehq of things that are needed for > >1.0. > No, it wouldn't. Even if Wine already was 1.0 we still would not have > the > portability we're trying to get (ie. run on Solaris Sparc, Mac OS, > etc) and Well Winelib does run on PPC and the port to Sparc is mostly done. 99% of Wine even compiles on Alpha here for me. See http://www.winehq.com/site/status_porting > debugging would still be almost impossible (Wine does funky stuff > with > register for thread storage, special stuff would be required to setup > Mono-created threads before they'd be able to call Wine functions, > etc). > Also, there doesn't seem to be much interest from the Wine community > to do > anything that helps using Wine with SWF/Mono. We've had to resort to > some > quite complicated mechanisms to even be able to use Wine as a shared, > runtime attached library, after Alexandre rejected a patch that only > touched > six lines of existing Wine code and added a new library to Wine, to > allow > straightforward use of Wine as a library. Having to take instead the > complicated route it now means that many people are having problems > getting > SWF to run at all, with due to strict path dependencies. I was there when Miguel came online to ask why the patch was not merged and I also seem to recall Alexandre proposed a fix the would work and seemed to make Miguel happy. See: http://www.winehq.org/hypermail/wine-devel/2004/03/0150.html > Another problem was that we had to spend way too much time tracking > what > changed from one Wine version to the next. For example from April to > May, > the wine_get_unix_file_name() function arguments were changed (I'm > not even > asking why it was changed instead of introducing a new function with > different arguments and leaving the old one for those people who > might use > it). Since I can't require always the latest Wine version, I have to > come up > with code figuring out what Wine version might be running, and write > code to > make version dependent Wine calls. I'd rather spend that time > improving SWF. Once again how can you fuss about unstable interfaces on a project that does not have a stable release? You could work with Winehq to create those stable interfaces. Besides how can you blame the Wine project for tracking changes when the first patch Mono submitted was this? http://primates.ximian.com/~duncan/mono-wine/sources/mono-wine.patch A whole bunch of #ifdefs and other kludges that violates Winehq coding standards. > Notice how people complain on this mailing that they're having > problems > getting even basic stuff going? The SWF code works, it's the > interaction > with Wine that's mostly causing the problems. Again, as I said, it > will > become much easier for the regular user once the Wine dependency is > removed > and the controls are all in managed code. Alright well I am not claiming to know very much about the project. Like I said I only lurk here because I care about how its going to run on ReactOS. I guess my final though on the matter is that if Vertigo Software could make a mananged Quake2, how hard can it be for you to make a managed Wine? Thanks Steven __________________________________ Do you Yahoo!? Yahoo! Mail - 50x more storage than other providers! http://promotions.yahoo.com/new_mail

1 0

collation tables
by Vitaly Lipatov July 1, 2004

July 1, 2004

In test dlls/kernel/tests/locale.c I see comment /* this requires collation table patch to make it MS compatible */ I get trouble with one program it needed collation table to be identical to MS table. Does anyone knows what is the better way to fix this tables, in which place? Can I fix libs/unicode/cpmap.pl for it? Any suggestions? -- Lav Vitaly Lipatov Saint-Petersburg GNU! ALT Linux Team! LaTeX! LyX!

2 1

Re: fix bug in GetCPInfo
by Uwe Bonnes July 1, 2004

July 1, 2004

>>>>> "Vitaly" == Vitaly Lipatov <lav(a)altlinux.ru> writes: Vitaly> Bug: When call GetCPInfo with CP_ACP for example, in returned Vitaly> structure field CodePage filled with CP_ACP constant, not real Vitaly> codepage of current ACP. The attached patch fix this bug. I Vitaly> have compared behavior of this functions with Win2003. This implies, that you have written a test. Please provide the test for our test suite. Bye -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

2 1

psapi.dll tests
by Ferenc Wagner July 1, 2004

July 1, 2004

Hi, please somebody enlighten me! How does this test work once psapi_test.exe is linked with -lpsapi? How can a program start at all without one of its hard dependencies? START_TEST(module) { dll = LoadLibrary("psapi.dll"); if (!dll) { trace("LoadLibraryA(psapi.dll) failed: skipping tests with target module\n"); return; } The above seems to be a common idiom which I don't understand as opposed to using GetProcAddress and invoking functions through pointers. What am I missing here? -- Feri.

2 2