August 2001 - Wine-Devel - list.winehq.org

Re: documentation make fix
by Mike McCormack Aug. 21, 2001

Aug. 21, 2001

Yeah, that looks better, and seems to work too. thanks, Mike Original message from: Francois Gouget <fgouget(a)free.fr> > >On Tue, 21 Aug 2001, Mike McCormack wrote: > >> >> Not that anybody uses the documentation anyway... :-) >> >> ChangeLog: >> * recurse all directories when making docs. >> Don't fail if C_SRCS is empty. > > Maybe the attached patch would be a better way to do this: > - use the regular make trick for recursing > - test whether we have targets before entering the for loop > > Note that I did not really test it. > > > Well, I have done a little testing now. And c2man is not happy if you >don't create documentation/man3w beforehand... > So I added an mkdir too, same for doc-html. > > >-- >Francois Gouget fgouget(a)free.fr http://fgouget.free.fr/ > "Utilisateur" (nom commun) : > Mot utilis� par les informaticiens en lieu et place d'"idiot". ------------------------------------------ mailto:Mike_McCormack(a)start.com.au ph +82 16 430 0425 __________________________________________________________________ Get your free Australian email account at http://www.Looksmart.com.au

1 0

Re: reaktivate
by Malte Starostik Aug. 21, 2001

Aug. 21, 2001

Am Montag, 20. August 2001 08:59 schrieb Per Wigren: > I get this when I'm trying to load an Active-X component in konqueror: > > konqueror: /usr/lib/libreaktivate: could not load library > '/usr/lib/libreaktivate.so' as Winelib application: > /usr/X11R6/lib/libXv.so.1: undefined symbol: XGetErrorDatabaseText > > Any ideas what to do? Using cvs from today of both wine and reaktivate. Never seen that one before, but it's IMHO rather a wine problem. Has anyone seen this before? Or, did you by any chance update some but not all X libraries? -- Malte Starostik PGP: 1024D/D2F3C787 [C138 2121 FAF3 410A 1C2A 27CD 5431 7745 D2F3 C787]

1 0

Re: ReadFile and other issues
by Alexandre Julliard Aug. 21, 2001

Aug. 21, 2001

eric pouech <eric.pouech(a)wanadoo.fr> writes: > no, reading of input_records would be sone as it is today, with > the specific console request 'read_console_input' OK, we agree then. > the only thing we'd have to have is another request to get the fd from > a console input handle (get_handle_fd would no longer work), and we > still need a (background) mechanism to convert the unix xterm characters > into INPUT_RECORD (a get_console_fd should do). Anyway, those fd are > even rather static (they only change on Alloc|Free Console and at > process startup if we inherit them from parent), so we can savely keep > them around, process wide I'm not sure we should be converting xterm input into INPUT_RECORD. It may be better to have two modes, a simple console that returns an fd to ReadFile and no input records (and then ReadConsoleInput could generate them on the fly if necessary), and a complex console that supports all the features but that doesn't have an associated fd at all. IMO the current xterm-based console support is a ugly hack; we should do a proper Windows console using a normal Wine window with a message loop to retrieve input events. And usage of stdin/stdout should be limited to programs that don't need special console features, and should only support ReadFile/WriteFile. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: ReadFile and other issues
by Alexandre Julliard Aug. 21, 2001

Aug. 21, 2001

eric pouech <eric.pouech(a)wanadoo.fr> writes: > to implement properly this (mainly the ability to forward ReadFile calls > to ReadConsole for console handles), I'd like to do the following: > - extend the get_handle_fd to also return an id corresponding to the > type fd being used. As of today, handle / fd relationship mainly > assumes that we can map a Windows operation on a handle onto a similar > operation on a unix fd. The ReadFile/ReadConsoleInput doesn't fit in > this category. I'm not sure I understand you right; are you suggesting that get_handle_fd should return an fd from which you can read input records in the case of a console? It would seem preferable to have get_handle_fd simply fail with a special error code saying that the handle is a console, and then ReadFile can fall back to using ReadConsoleInput which doesn't require an fd at all. > - this type of feature would help in other areas too. For example, we > never check that Comm functions really get a Comm handle... Well, we check by the fact that we then do a serial ioctl on the fd, which will fail if it isn't serial. I'm not sure we need more checks. -- Alexandre Julliard julliard(a)winehq.com

2 1

Archive of overlapped serial I/O example code, experiences
by Dan Kegel Aug. 21, 2001

Aug. 21, 2001

For what it's worth, here is some example test code I wrote for win32 demoing overlapped serial I/O: http://www.alumni.caltech.edu/~dank/serio/serio.zip It might be of some use to people testing wine. I haven't tried it yet there. Also, http://www.alumni.caltech.edu/~dank/overlap.htm chronicles the problems many people have had with serial I/O under win95. The messages there get into interesting details of how people use overlapped serial I/O, and how serial I/O really behaves under various flavors of windows. Again, might be of some use, who knows. - Dan -- "I have seen the future, and it licks itself clean." -- Bucky Katt

1 0

ReadFile and other issues
by eric pouech Aug. 21, 2001

Aug. 21, 2001

while we're at talking about the ReadFile implementation with overlapped data, there's (an unrelated) point about ReadFile I'd like to discuss I've started looking at implementing proper console input in Wine (mainly the line mode editing). Apart from providing the functionnality, it would also allow to remove this unneeded code from the debugger the Win32 console input API is made of two layers 1/ the low level : each input event is stored into an INPUT_RECORD struct 2/ the high level which uses the low level events to make strings of them 1/ two functions are used: ReadConsoleInput(A|W) 2/ three functions are used: ReadFile and ReadConsole(A|W) the current implementation in Wine, makes the handle to the console input a file handle to the fd gotten from the console hence, the current implementation is broken in several ways - ReadFile is implemented as a raw read on the input fd, whereas it should be a higher level of function (forwarded from ReadFile to ReadConsole) - wait operations on handle are wrong: currently the handle is signaled as soon as a char is read (at the unix level) from the stdin fd, whereas it should be done at the level of the INPUT_RECORD queue to implement properly this (mainly the ability to forward ReadFile calls to ReadConsole for console handles), I'd like to do the following: - extend the get_handle_fd to also return an id corresponding to the type fd being used. As of today, handle / fd relationship mainly assumes that we can map a Windows operation on a handle onto a similar operation on a unix fd. The ReadFile/ReadConsoleInput doesn't fit in this category. - this type of feature would help in other areas too. For example, we never check that Comm functions really get a Comm handle... would such a mechanism be acceptable ? if so, should we do for every handle in the server, or just the ones we might convert into unix fd (file, consoles, pipes, sockets, "serials"...) A+ -- --------------- Eric Pouech (http://perso.wanadoo.fr/eric.pouech/) "The future will be better tomorrow", Vice President Dan Quayle

1 0

Re: icons uncoloured
by Gerard Patel Aug. 21, 2001

Aug. 21, 2001

At 01:39 PM 21/08/2001 GMT, you wrote: >I've just done a cvs update and recompiled and now the icons come up in >gray, if at all. > >Is this a known issue? I suspect it could be solved by this one I just sent to wine-patches Gerard Index: graphics/x11drv/bitmap.c =================================================================== RCS file: /home/wine/wine/graphics/x11drv/bitmap.c,v retrieving revision 1.32 diff -u -r1.32 bitmap.c --- graphics/x11drv/bitmap.c 2001/08/20 19:19:18 1.32 +++ graphics/x11drv/bitmap.c 2001/08/21 14:17:07 @@ -73,7 +73,7 @@ if (hbitmap == dc->hBitmap) return hbitmap; /* nothing to do */ if (!(bmp = GDI_GetObjPtr( hbitmap, BITMAP_MAGIC ))) return 0; - if (bmp->header.dwCount) + if (bmp->header.dwCount && (hbitmap != GetStockObject(DEFAULT_BITMAP))) { WARN( "Bitmap already selected in another DC\n" ); GDI_ReleaseObj( hbitmap ); ~ ~

2 1

icons uncoloured
by Bill Medland Aug. 21, 2001

Aug. 21, 2001

I've just done a cvs update and recompiled and now the icons come up in gray, if at all. Is this a known issue? Bill

1 0

Re: documentation make fix
by Francois Gouget Aug. 21, 2001

Aug. 21, 2001

On Tue, 21 Aug 2001, Mike McCormack wrote: > > Not that anybody uses the documentation anyway... :-) > > ChangeLog: > * recurse all directories when making docs. > Don't fail if C_SRCS is empty. Maybe the attached patch would be a better way to do this: - use the regular make trick for recursing - test whether we have targets before entering the for loop Note that I did not really test it. Well, I have done a little testing now. And c2man is not happy if you don't create documentation/man3w beforehand... So I added an mkdir too, same for doc-html. -- Francois Gouget fgouget(a)free.fr http://fgouget.free.fr/ "Utilisateur" (nom commun) : Mot utilisé par les informaticiens en lieu et place d'"idiot".

1 0

Re: fix for Lawson's CLOCAL problem
by Mike McCormack Aug. 21, 2001

Aug. 21, 2001

Hi Alexandre, >Of course, I have no problem with that. The create_file request takes >attributes, so create_serial can do the same (it would be better if >create_serial were done inside create_file, but that's another >story...) agreed. >> Another question: do you think it would be possible to implement a >> "write memory" message from the wineserver to the client? .. >But that would require multiple copies of the data (and many context >switches for large reads); that wouldn't be very efficient. I think >the current approach is much better. What is the problem with doing >CancelIo? IMO it would be messy to implement CancelIo with the current architecture. If it were all in the server (including actual read and write calls) it might be simpler... but probably not efficient, like you say. i think most of the data transferred from a serial port is pretty small though... but if the scheme was expanded to handle overlapped socket operations that wouldn't be true. Additionally, i though it might be useful to have a way to write small chunks of data to the client's memory. For example, when implementing WaitForNamedPipe and ConnectNamedPipe, i need to write a result code into an overlapped structure... it's only 4 bytes, but i need to create a new type of APC to deal with it. If i had a method to write directly into client memory, it would be simpler. Mike ------------------------------------------ mailto:Mike_McCormack(a)start.com.au ph +82 16 430 0425 __________________________________________________________________ Get your free Australian email account at http://www.Looksmart.com.au

1 0

Re: fix for Lawson's CLOCAL problem
by Alexandre Julliard Aug. 21, 2001

Aug. 21, 2001

Mike McCormack <mike_mccormack(a)start.com.au> writes: > Hmmm. i'm not exactly sure about this. i don't think the server ever > blocked on serial port reads or opens... i think the reason Lawson had > problems was that the client thread was blocking in the overlapped > read code. (FILE_AsyncReadService in files/file.c) I don't know if this was Lawson problem, but open() can potentially block on a serial port, and we cannot have that in the server; so your change to set O_NONBLOCK is good, but it shouldn't depend on FILE_FLAG_OVERLAPPED. > If we use non-blocking fds unconditionally, a non overlapped ReadFile > may result in a busy loop in the client thread. Sure, you need to reset the O_NONBLOCK flag to do blocking reads. > Additionally, overlapped read (and write) requests should fail unless > the handle is opened with FILE_FLAG_OVERLAPPED. There are programs > that depend on that behaviour (as reported by Eric Kohl, 3 Feb 2001 on > this list). This is another reason to propagate CreateFile's > attributes parameter into the wineserver. Of course, I have no problem with that. The create_file request takes attributes, so create_serial can do the same (it would be better if create_serial were done inside create_file, but that's another story...) > Another question: do you think it would be possible to implement a > "write memory" message from the wineserver to the client? This message > would be carried over wait_fd and would be received by the client > asynchronously in wait_reply(). The message would contain a length, > address and data to be written. If this is possible, i could move > async i/o back into the server, keep it efficient and perhaps > implement CancelIo properly... But that would require multiple copies of the data (and many context switches for large reads); that wouldn't be very efficient. I think the current approach is much better. What is the problem with doing CancelIo? -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: winsock stuff
by Daniel Walker Aug. 21, 2001

Aug. 21, 2001

This seems to work.. Daniel Walker --- server/sock.c 2001/05/14 20:09:39 1.18 +++ server/sock.c 2001/08/21 01:31:04 @@ -113,9 +113,14 @@ { struct sock *sock = (struct sock *)obj; unsigned int emask; + char rBuf; assert( sock->obj.ops == &sock_ops ); if (debug_level) fprintf(stderr, "socket %d select event: %x\n", sock->obj.fd, event); + + if (event == POLLIN && !recvfrom(sock->obj.fd, &rBuf, sizeof(rBuf), MSG_PEEK, NULL, NULL)) + event = POLLHUP; + if (sock->state & WS_FD_CONNECT) { /* connecting */

1 0

Re: fix for Lawson's CLOCAL problem
by Mike McCormack Aug. 21, 2001

Aug. 21, 2001

Hi Alexandre, Hmmm. i'm not exactly sure about this. i don't think the server ever blocked on serial port reads or opens... i think the reason Lawson had problems was that the client thread was blocking in the overlapped read code. (FILE_AsyncReadService in files/file.c) If we use non-blocking fds unconditionally, a non overlapped ReadFile may result in a busy loop in the client thread. Additionally, overlapped read (and write) requests should fail unless the handle is opened with FILE_FLAG_OVERLAPPED. There are programs that depend on that behaviour (as reported by Eric Kohl, 3 Feb 2001 on this list). This is another reason to propagate CreateFile's attributes parameter into the wineserver. Another question: do you think it would be possible to implement a "write memory" message from the wineserver to the client? This message would be carried over wait_fd and would be received by the client asynchronously in wait_reply(). The message would contain a length, address and data to be written. If this is possible, i could move async i/o back into the server, keep it efficient and perhaps implement CancelIo properly... Mike > The open on the server side must always be non-blocking, we cannot > have the server block on anything. If there are cases where we need to > block on open, it will have to be done on the client side. ------------------------------------------ mailto:Mike_McCormack(a)start.com.au ph +82 16 430 0425 __________________________________________________________________ Get your free Australian email account at http://www.Looksmart.com.au

1 0

Re: winsock stuff
by Alexandre Julliard Aug. 21, 2001

Aug. 21, 2001

Daniel Walker <diwalker(a)earthlink.net> writes: > How is it unreliable? Why is MSG_PEEK more reliable? I'm not opposed to > using MSG_PEEK , just curios.. FIONREAD is unreliable because you cannot distinguish between an EOF condition and an empty receive buffer. recvmsg(MSG_PEEK) is supposed to return 0 on EOF, and EWOULDBLOCK on empty buffer, so you can determine the real status. > Another issue is why doesn't poll() give > us a POLLHUP? That would be ideal, get poll() to work the way we need it > too. As I understand it poll() returns POLLHUP only when both directions of the socket have been closed; but if you shutdown() only the read or write side, you don't get POLLHUP so that you can continue using poll() for the other direction. This has apparently been changed between kernels 2.2 and 2.4; in 2.2 POLLHUP is returned as soon as the read side of the socket has been closed. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: winsock stuff
by Alexandre Julliard Aug. 21, 2001

Aug. 21, 2001

Daniel Walker <diwalker(a)earthlink.net> writes: > To improve on this patch we could only do the FIONREAD if the sock is > in a non-blocking state. That's the only time that doing the FIONREAD > would be an advantage. Also we wouldn't need to worry about screwing up > blocking sockets. Even further, we could only do the FIONREAD if socket > is non-blocking and looking for an FD_CLOSE event. Using FIONREAD is simply not an option, it cannot be made reliable. A possibility to detect EOF could be a recvmsg with MSG_PEEK flag, though you should probably verify first that it behaves as expected. -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: winsock stuff
by Alexandre Julliard Aug. 20, 2001

Aug. 20, 2001

Daniel Walker <diwalker(a)earthlink.net> writes: > I noticed that WSAEnumNetworkEvents() was returning all events that had > occurred and not just the events that had been selected with > WSAEventsSelect(). Win98 will trigger the event object for all events > but WSAEnumNetworkEvents() (in win98) would return zero if the event > hadn't been selected. I was trying to gain compatibility with win98. I'd suggest doing the masking in WSAEnumNetworkEvents then. > An FD_READ event with a zero buffer would mean that the other side > closed the connection so it would be equivalent to a POLLHUP from poll() > .. I didn't find an occasion when poll() actually returned POLLHUP when > the socket closed, on Linux anyway. The application that I was testing > got confused when recv() returned zero. Even though recv() can legally > return zero, I think non-blocking applications are going to be looking > for FD_CLOSE events as opposed to recv() returning zero. But FIONREAD can return 0 bytes even when the socket isn't closed (for instance if another thread got the data first). You can only determine that it was really closed when you do the actual read. So I don't really see how we can avoid having recv() return zero; well maybe we can make it return EWOULDBLOCK and tell the server the connection is closed, but I'm not sure it's any better. -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: winsock stuff
by Alexandre Julliard Aug. 20, 2001

Aug. 20, 2001

Daniel Walker <diwalker(a)earthlink.net> writes: > -changed get_socket_event() so it only returns events that > the application is looking for. > -Changed sock_poll_event() to interpret a POLLIN event with > zero bytes waiting to be read as a POLLHUP. This doesn't look really right to me. Could you please explain a bit more what it is supposed to fix? -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: Refcounting in SelectObject
by Alexandre Julliard Aug. 20, 2001

Aug. 20, 2001

Uwe Bonnes <bon(a)elektron.ikp.physik.tu-darmstadt.de> writes: > Appended code tries to clean up better by: > - Remembering an unsuccessfull DeleteObject by adding 0x8000 to the Refcount > - Call DeleteObject in SelectObject when the RefCount of the released handle > is 0x8000 > - Decrementing the RefCount when a hBitmap is still selected into a DC when > DeleteDC is called for that DC > - Call DeleteObject in DeleteDC when RefCount is 0x8000 > > For some applications I tested, I always saw the delayed DeleteObject done > after the refused DeleteObject in the first place. Is this approach better? Yes, looks good. Though we can probably safely use the dwCount field in GDIOBJHDR since it is not used for anything useful right now, instead of adding new fields. I've changed that and applied the patch. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: fix for Lawson's CLOCAL problem
by Alexandre Julliard Aug. 20, 2001

Aug. 20, 2001

Mike McCormack <mike_mccormack(a)start.com.au> writes: > Open serial ports with O_NONBLOCK if FILE_FLAG_OVERLAPPED is > specified. The open on the server side must always be non-blocking, we cannot have the server block on anything. If there are cases where we need to block on open, it will have to be done on the client side. -- Alexandre Julliard julliard(a)winehq.com

1 0

-clocal and the wineserver
by lawson_whitney＠juno.com Aug. 19, 2001

Aug. 19, 2001

Good day! Any time clocal falls out of the modem's termios (say I use cu -l on it) the wineserver hangs trying to open it, or at least the app hangs, and I haven't found a way to kill it without killing the wineserver. Both 16 and 32 bit versions of juno are affected. I haven't really investigated it very thouroughly, but I wasted a lot of time restoring backup versions of the app and wine. Should I look into mickey-mousing the open to avoid the problem, or - is it really a wineserver problem which I am probably not even marginally competent to muck with? I'd be glad to make traces or winedbg output, but it seems the problem should be easy to duplicate. Hmmm, setting clocal back on from another terminal unhangs the app. That is neater than killing the wineserver, I guess. terminal.exe from wfw 3.11 behaves similarly, with an extra twist: it hangs on the Win16Mutex until somebody sets clocal (well, I guess carrier detect would work, too). [whit(a)giftie user0001]$ wine terminal.exe err:ntdll:RtlpWaitForCriticalSection section 0x400fad6c "syslevel.c: Win16Mutex" wait timed out, retrying (60 sec) fs=008f fixme:pthread_kill_other_threads_np Console: Making console complex (creating an xterm)... fixme:pthread_kill_other_threads_np fixme:pthread_kill_other_threads_np fixme:pthread_kill_other_threads_np fixme:pthread_kill_other_threads_np Warning: locale not supported by Xlib, locale set to C Warning: X locale modifiers not supported, using default xterm: unable to open font "vga", trying "fixed".... Failed to open input method passed master doesn't match slave name passed master doesn't match slave name [whit(a)giftie user0001]$ Regards, Lawson ---oof--- ________________________________________________________________ GET INTERNET ACCESS FROM JUNO! Juno offers FREE or PREMIUM Internet access for less! Join Juno today! For your FREE software, visit: http://dl.www.juno.com/get/tagj.

1 0

Re: PATCH: olepictures black & white
by Alexandre Julliard Aug. 19, 2001

Aug. 19, 2001

Marcus Meissner <marcus(a)jet.franken.de> writes: > Somehow the loaded JPEGs are now black & white in the IPicture implementation. > > Exchanging CreateCompatibleDC(0) by CreateDCA("DISPLAY",0,0,0); makes > them colored again. > > But I am not sure whether the problem itself was a bug on my side, or > on the WINE side... Maybe a bit of both... I think the patch below should fix the GDI part; but is there any reason you don't use GetDC(0) instead of CreateCompatibleDC? It would avoid creating a new DC every time. Index: objects/dib.c =================================================================== RCS file: /opt/cvs-commit/wine/objects/dib.c,v retrieving revision 1.60 diff -u -r1.60 dib.c --- objects/dib.c 2001/08/17 00:07:42 1.60 +++ objects/dib.c 2001/08/19 19:27:42 @@ -875,7 +875,9 @@ /* Now create the bitmap */ - if (fColor) handle = CreateCompatibleBitmap( hdc, width, height ); + if (fColor) + handle = CreateBitmap( width, height, GetDeviceCaps( hdc, PLANES ), + GetDeviceCaps( hdc, BITSPIXEL ), NULL ); else handle = CreateBitmap( width, height, 1, 1, NULL ); if (!handle) return 0; -- Alexandre Julliard julliard(a)winehq.com

1 0

PATCH: oleauto header problem
by Nikolas Zimmermann Aug. 19, 2001

Aug. 19, 2001

Hi there, Reaktivate doesn't compile anymore. We need this patch. Bye Bye Niko -- Nikolas Zimmermann wildfox(a)kde.org

1 0

Re: WM_GETTEXTLENGTH change in winproc.c
by Gerard Patel Aug. 19, 2001

Aug. 19, 2001

At 02:35 PM 18/08/2001 -0700, you wrote: >For standard controls like buttons it should never be a problem since >there is both an ASCII and Unicode win proc, so in a pure ASCII app we >should never do the conversion. I think the problem is that the >dialog code does a CreateWindowW, so the button is Unicode even though >the app does everything in ASCII. I guess we'll have to maintain both >A and W win procs inside each window, not only inside the class. How about creating Ascii windows when the app calls DialogBox..A variants ? Gerard

1 0

Re: WM_GETTEXTLENGTH change in winproc.c
by Alexandre Julliard Aug. 18, 2001

Aug. 18, 2001

Gerard Patel <gerard.patel(a)nerim.net> writes: > How about creating Ascii windows when the app calls DialogBox..A variants ? That could be a workaround, but since dialog resources are Unicode it's better to avoid converting back and forth. Besides, Windows maintains two different win procs for windows of standard classes so we should do it this way too. I'll try to come up with something. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: WM_GETTEXTLENGTH change in winproc.c
by Gerard Patel Aug. 18, 2001

Aug. 18, 2001

At 01:19 PM 18/08/2001 -0700, you wrote: > >The idea is that 6 Unicode chars can yield more than 6 ASCII chars >with DBCS text. The MSDN doc states that WM_GETTEXTLENGTH can return >more than the actual size when mixing ASCII and Unicode. > >What does the app do exactly? Maybe there are cases where the >conversion can be avoided. WM_GETTEXTLENGTH, WM_GETTEXT, then DrawText with the length returned by WM_GETTEXTLENGTH. If the app had used the length returned by WM_GETTEXT, it would have worked all right of course - maybe that's the reason why the problem does not appear more often ? In the following trace Cancel is the text of a button in a dialog box. Gerard 080682e8:Call window proc 0x5f404ae6 (hwnd=000026dc,msg=WM_DRAWITEM,wp=00000002,lp=40586474) 080682e8:Call user32.GetPropA(000026dc,5f49c94c "AfxOldWndProc423") ret=5f404b09 080682e8:Ret user32.GetPropA() retval=40ac0259 ret=5f404b09 080682e8:Call user32.CallWindowProcA(40ac0259,000026dc,0000002b,00000002,40586474) ret=5f404b4b 080682e8:Call window proc 0x406a7980 (hwnd=000026dc,msg=WM_DRAWITEM,wp=00000002,lp=40586474) 080682e8:Call window proc 0x43d715 (hwnd=000026dc,msg=WM_DRAWITEM,wp=00000002,lp=40586474) 080682e8:Call user32.GetWindowLongA(000026dc,00000008) ret=0043d727 080682e8:Ret user32.GetWindowLongA() retval=40586914 ret=0043d727 080682e8:Call user32.GetWindowTextLengthA(00002e0c) ret=0043bb82 trace:message:SPY_EnterMessage (00002e0c) "Cancel" message [000e] WM_GETTEXTLENGTH sent from self wp=00000000 lp=00000000 080682e8:Call window proc 0x43d28b (hwnd=00002e0c,msg=WM_GETTEXTLENGTH,wp=00000000,lp=00000000) 080682e8:Call user32.CallWindowProcA(40ac00f1,00002e0c,0000000e,00000000,00000000) ret=0043d2b8 080682e8:Call window proc 0x40679e8c (hwnd=00002e0c,msg=WM_GETTEXTLENGTH,wp=00000000,lp=00000000) trace:message:SPY_EnterMessage (00002e0c) DefWindowProc32: WM_GETTEXTLENGTH [000e] wp=00000000 lp=00000000 trace:message:SPY_ExitMessage (00002e0c) DefWindowProc32: WM_GETTEXTLENGTH [000e] returned 00000006 080682e8:Ret window proc 0x40679e8c (hwnd=00002e0c,msg=WM_GETTEXTLENGTH,wp=00000000,lp=00000000) retval=00000006 080682e8:Ret user32.CallWindowProcA() retval=0000000c ret=0043d2b8 080682e8:Ret window proc 0x43d28b (hwnd=00002e0c,msg=WM_GETTEXTLENGTH,wp=00000000,lp=00000000) retval=0000000c trace:message:SPY_ExitMessage (00002e0c) "Cancel" message [000e] WM_GETTEXTLENGTH returned 0000000c 080682e8:Ret user32.GetWindowTextLengthA() retval=0000000c ret=0043bb82 080682e8:Call ntdll.RtlEnterCriticalSection(5f4d0088) ret=5f4020bd 080682e8:Ret ntdll.RtlEnterCriticalSection() retval=00000000 ret=5f4020bd 080682e8:Call ntdll.RtlLeaveCriticalSection(5f4d0088) ret=5f402103 080682e8:Ret ntdll.RtlLeaveCriticalSection() retval=00000000 ret=5f402103 080682e8:Call user32.GetWindowTextA(00002e0c,413904e0,0000000d) ret=0043bba9 trace:message:SPY_EnterMessage (00002e0c) "Cancel" message [000d] WM_GETTEXT sent from self wp=0000000d lp=413904e0 080682e8:Call window proc 0x43d28b (hwnd=00002e0c,msg=WM_GETTEXT,wp=0000000d,lp=413904e0) 080682e8:Call user32.CallWindowProcA(40ac00f1,00002e0c,0000000d,0000000d,413904e0) ret=0043d2b8 080682e8:Call window proc 0x40679e8c (hwnd=00002e0c,msg=WM_GETTEXT,wp=0000000d,lp=403a7328) trace:message:SPY_EnterMessage (00002e0c) DefWindowProc32: WM_GETTEXT [000d] wp=0000000d lp=403a7328 trace:message:SPY_ExitMessage (00002e0c) DefWindowProc32: WM_GETTEXT [000d] returned 00000006 080682e8:Ret window proc 0x40679e8c (hwnd=00002e0c,msg=WM_GETTEXT,wp=0000000d,lp=403a7328) retval=00000006 080682e8:Ret user32.CallWindowProcA() retval=00000006 ret=0043d2b8 080682e8:Ret window proc 0x43d28b (hwnd=00002e0c,msg=WM_GETTEXT,wp=0000000d,lp=413904e0) retval=00000006 trace:message:SPY_ExitMessage (00002e0c) "Cancel" message [000d] WM_GETTEXT returned 00000006 080682e8:Ret user32.GetWindowTextA() retval=00000006 ret=0043bba9 <snip unrelated painting instructions> 080682e8:Call gdi32.GetTextExtentPoint32A(00000088,413b00d0,00000006,40585f68) ret=0043c022 080682e8:Call x11drv.GetTextExtentPoint(4037c808,4039d238,00000006,40585f68) ret=40747c04 trace:text:X11DRV_GetTextExtentPoint L"Cancel"... 6 080682e8:Ret x11drv.GetTextExtentPoint() retval=00000001 ret=40747c04 080682e8:Ret gdi32.GetTextExtentPoint32A() retval=00000001 ret=0043c022 <snip a few text painting preparation> 080682e8:Call user32.DrawTextA(00000088,413904e0 "Cancel",0000000c,40585f28,00000100) ret=0043c13a trace:text:DrawTextExW L"Cancel", 12 , [(9,4),(51,20)]

1 0

Re: WM_GETTEXTLENGTH change in winproc.c
by Alexandre Julliard Aug. 18, 2001

Aug. 18, 2001

Gerard Patel <gerard.patel(a)nerim.net> writes: > WM_GETTEXTLENGTH, WM_GETTEXT, then DrawText with the > length returned by WM_GETTEXTLENGTH. > If the app had used the length returned by WM_GETTEXT, it would > have worked all right of course - maybe that's the reason why > the problem does not appear more often ? For standard controls like buttons it should never be a problem since there is both an ASCII and Unicode win proc, so in a pure ASCII app we should never do the conversion. I think the problem is that the dialog code does a CreateWindowW, so the button is Unicode even though the app does everything in ASCII. I guess we'll have to maintain both A and W win procs inside each window, not only inside the class. -- Alexandre Julliard julliard(a)winehq.com

1 0

WM_GETTEXTLENGTH change in winproc.c
by Gerard Patel Aug. 18, 2001

Aug. 18, 2001

I don't understand it : case WM_GETTEXTLENGTH: 719 case CB_GETLBTEXTLEN: 720 case LB_GETTEXTLEN: 721 /* there may be one DBCS char for each Unicode char */ 722 return result * 2; This has a bad effect on one of my test apps - returns a length of 12 for a text of 6 characters, displays funny symbols after the text. The Unicode window proc returns 6, why the unicode/ascii conversion should make it 12 ? The Api returns number of chars, not bytes. Gerard

1 0

Re: WM_GETTEXTLENGTH change in winproc.c
by Alexandre Julliard Aug. 18, 2001

Aug. 18, 2001

Gerard Patel <gerard.patel(a)nerim.net> writes: > I don't understand it : > > case WM_GETTEXTLENGTH: > 719 case CB_GETLBTEXTLEN: > 720 case LB_GETTEXTLEN: > 721 /* there may be one DBCS char for each Unicode char */ > 722 return result * 2; > > This has a bad effect on one of my test apps - returns a length of 12 > for a text of 6 characters, displays funny symbols after the text. > > The Unicode window proc returns 6, why the unicode/ascii conversion > should make it 12 ? The Api returns number of chars, not bytes. The idea is that 6 Unicode chars can yield more than 6 ASCII chars with DBCS text. The MSDN doc states that WM_GETTEXTLENGTH can return more than the actual size when mixing ASCII and Unicode. What does the app do exactly? Maybe there are cases where the conversion can be avoided. -- Alexandre Julliard julliard(a)winehq.com

1 0

not-yet-connected sockets?
by Daniel Walker Aug. 18, 2001

Aug. 18, 2001

The problem is that poll returns POLLHUP when a socket isn't connected, then wine ignores the socket and it's events .. I realize this has been known for a some time(since linux 2.4), but why hasn't it been fix? Doesn't this effect a lot of non-blocking socket applications? Daniel Walker

3 13

Re: Toolbar background error
by Gerard Patel Aug. 18, 2001

Aug. 18, 2001

At 06:54 PM 17/08/2001 -0700, you wrote: <snip> >But their solution is still not correct IMO. How about something like >this? All clear for me on this one, tested without problem for all of my test apps - at least the one surviving the Gdi lock slaughter :-) Gerard

1 0

Re:CryptoAPI
by Eric Pouech Aug. 18, 2001

Aug. 18, 2001

from a pure technical point of view, the crypto API is in a pure WOSA style : - a set of high level API (in advapi) - an extendable set of low level service providers (here for example the crypto algos - RSA....) - the high level maps the fonctional request of the user onto on of the available SP. so, my advice here would be to implement this type of scheme. This would allow to: - use native DLLs if they exist (and go beyond what's supported by Linux) - still allow to use Linux lib (just build a wrapper DLL to that lib) and you should be all set from a copyright point of view: - Wine use a X11 like licence, so make sure all the external libs you use are not GPL (LGPL is fine). If they are GPL, you won't be able to ship the result with Wine. - Use of native DLL is granted as long as you have a Windows licence on the target machine (native DLLs shall not be shipped with Wine, Wine should be configured to use the local resources) from an exportation point of view: - some countries (US as an example) puts lots of contraints on the type of crypto algo to be used outside of the US. So those shouldn't be shipped with Wine's, but be provided as an external resource) > Hi, > > I was going to start implementing the Cypto API (in advapi32.dll) over the > next few weeks. As of yet these functions are no more than stubs. However, > I am finding increased use of this API as more applications becomed > networkable and relying more heavily on internet connections. > > I was hoping to implement these functions by using the OpenSSL library > (if/when available). This library seems to have similar capabilities. > > One of my concerns is about legal issues. What special concerns should I > note about cryptographic software? Are there any special copyright issues I > should be aware of? > > Another question is about design. M$ seems to split the cryptographic > software into different dlls (rsabase.dll, rsasig.dll, dssbase.dll, etc.). > Should I do this as well so that, applications can use the native dlls if > necessary or should I simply implement it entirely though advapi32.dll which > and avoid the need to create several new (and very small) dlls under wine? > Because most of the code is already in another library (OpenSSL) creating > extra dlls seems abit unnecessary. > > - Travis > > _________________________________________________________ > Do You Yahoo!? > Get your free @yahoo.com address at http://mail.yahoo.com > > > > --------------- Eric Pouech (http://perso.wanadoo.fr/eric.pouech/) The future will be better tomorrow, Vice President Dan Quayle __________________________________________________________ Faites un vu et puis Voila ! www.voila.fr Pendant tout l'été, gagnez une Ford Ka sur Voila Gratter ! http://cybergrattage.voila.fr/voila

2 1

Re: Toolbar background error
by Alexandre Julliard Aug. 18, 2001

Aug. 18, 2001

Marcus Meissner <marcus(a)jet.franken.de> writes: > I suggest reading the Corel CVS for this part of the code. > > They rewrote this part like _5_ times. But their solution is still not correct IMO. How about something like this? Index: windows/defwnd.c =================================================================== RCS file: /opt/cvs-commit/wine/windows/defwnd.c,v retrieving revision 1.59 diff -u -r1.59 defwnd.c --- windows/defwnd.c 2001/08/10 22:51:43 1.59 +++ windows/defwnd.c 2001/08/18 00:21:43 @@ -461,10 +461,13 @@ HBRUSH hbr = GetClassLongW( wndPtr->hwndSelf, GCL_HBRBACKGROUND ); if (!hbr) return 0; - /* GetClientRect used to be GetClipBox, but it is not what - * Windows does, and it breaks badly with CS_PARENTDC */ - GetClientRect( wndPtr->hwndSelf, &rect ); - DPtoLP( hdc, (LPPOINT)&rect, 2 ); + if (GetClassLongW( wndPtr->hwndSelf, GCL_STYLE ) & CS_PARENTDC) + { + /* can't use GetClipBox with a parent DC or we fill the whole parent */ + GetClientRect( wndPtr->hwndSelf, &rect ); + DPtoLP( hdc, (LPPOINT)&rect, 2 ); + } + else GetClipBox( hdc, &rect ); FillRect( hdc, &rect, hbr ); return 1; } -- Alexandre Julliard julliard(a)winehq.com

1 0

dsound ??
by Bobby Bingham Aug. 17, 2001

Aug. 17, 2001

Hello all, I completely rebuilt my computer a while back, and updated wine, so I'm not sure which is causing it, but not dsound doesn't seem to work where it did before. Starcraft doesn't play any music, sound, etc, but it's installer still plays it's sound (without dsound). Before this I hadn't updated WINE for a while, so I'm not sure when it broke. Bobby Bingham

1 0

Re: Toolbar background error
by Gerard Patel Aug. 17, 2001

Aug. 17, 2001

At 11:08 AM 17/08/2001 -0700, you wrote: >That's not what Windows does. If you pass a window DC to WM_ERASEBKGND >it only erases an area equivalent to the client area (shifted to be >relative to the whole window), so GetClientRect is correct IMO. The >non-client area should be painted upon WM_NCPAINT, not WM_ERASEBKGND. How do you test this under Windows ? In my test app, I have a white brush as class brush. So my window is white. If I add after the BeginPaint : GetClientRect(hwnd, &r); r.top += 200; FillRect(ps.hdc, &r, (HBRUSH) COLOR_WINDOW) the lower part of my window is painted with the beautiful window color (brown) Now if I add immediately after : hdc = GetWindowDC(hwnd); SendMessage(hwnd, WM_ERASEBKGND, (WPARAM) hdc, 0); ReleaseDC(hwnd, hdc); I get a window that is completely white; maybe I am missing something but I should see a small brown strip at the lower part of the window, corresponding to the caption + menu height, no ? Gerard

1 0

Re: Toolbar background error
by Gerard Patel Aug. 17, 2001

Aug. 17, 2001

At 02:33 PM 17/08/2001 GMT, you wrote: <snip> >So how does an application modify the location of the client rectangle >after it has been created? By changing rgrc[0] when responding to WM_NCCALCSIZE. It seems that it is only possible to erase properly the 'custom' nonclient area by using GetWindowDC... So : --- defwnd.c.orig Sat Aug 11 10:23:18 2001 +++ defwnd.c Fri Aug 17 19:47:17 2001 @@ -462,8 +462,12 @@ if (!hbr) return 0; /* GetClientRect used to be GetClipBox, but it is not what - * Windows does, and it breaks badly with CS_PARENTDC */ - GetClientRect( wndPtr->hwndSelf, &rect ); + * Windows does, and it breaks badly with CS_PARENTDC + * GetClientRect breaks too when we are called with a window + * DC (!!) - what next ? */ + rect.left = rect.top = 0; + rect.right = wndPtr->rectWindow.right - wndPtr->rectWindow.left; + rect.bottom = wndPtr->rectWindow.bottom - wndPtr->rectWindow.top; DPtoLP( hdc, (LPPOINT)&rect, 2 ); FillRect( hdc, &rect, hbr ); return 1; What do you think ? Gerard

2 1

Re: Toolbar background error
by Alexandre Julliard Aug. 17, 2001

Aug. 17, 2001

Gerard Patel <gerard.patel(a)nerim.net> writes: > Now if I add immediately after : > > hdc = GetWindowDC(hwnd); > SendMessage(hwnd, WM_ERASEBKGND, (WPARAM) hdc, 0); > ReleaseDC(hwnd, hdc); > > I get a window that is completely white; maybe I am missing something but > I should see a small brown strip at the lower part of the window, corresponding > to the caption + menu height, no ? Yes; actually I realize now that my test app uses the CS_PARENTDC style, which is why I see that. The behavior that you observe is what happens without CS_PARENTDC; so it's actually more complex that I thought, we'll need to explicitly test the CS_PARENTDC flag. Sigh... -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: Problems building with extern inline functions in winebase.h
by Alexandre Julliard Aug. 17, 2001

Aug. 17, 2001

Mike Bond <mbond(a)cox.rr.com> writes: > A couple days ago a number of inline functions were put into winebase.h. > Since then I have been unable to build, using gcc 2.96-81 (RedHat). Apparently > this version of gcc thinks that because there is an extern prefixing the > inline, even though the function does in fact have a body, that the > function really is external. Removing the extern, just having inline as > one would do in C++, does not appear to work correctly either, and causes > the function to appear as a exported symbol in all the object files that > header gets included with, thus resulting in multiple definitions. This patch should fix the build failure, though it doesn't explain why gcc doesn't inline such a simple function. Index: dlls/kernel/kernel32.spec =================================================================== RCS file: /opt/cvs-commit/wine/dlls/kernel/kernel32.spec,v retrieving revision 1.39 diff -u -r1.39 kernel32.spec --- dlls/kernel/kernel32.spec 2001/08/06 17:50:42 1.39 +++ dlls/kernel/kernel32.spec 2001/08/16 16:30:03 @@ -922,7 +922,9 @@ @ stub GetProcessPriorityBoost @ stdcall GetThreadPriorityBoost(long ptr) GetThreadPriorityBoost @ stdcall InterlockedCompareExchange (ptr long long) InterlockedCompareExchange +@ stdcall InterlockedCompareExchangePointer(ptr ptr ptr) InterlockedCompareExchangePointer @ stdcall InterlockedExchangeAdd (ptr long ) InterlockedExchangeAdd +@ stdcall InterlockedExchangePointer(ptr ptr) InterlockedExchangePointer @ stdcall IsProcessorFeaturePresent(long) IsProcessorFeaturePresent @ stdcall OpenWaitableTimerA(long long str) OpenWaitableTimerA @ stdcall OpenWaitableTimerW(long long wstr) OpenWaitableTimerW Index: scheduler/critsection.c =================================================================== RCS file: /opt/cvs-commit/wine/scheduler/critsection.c,v retrieving revision 1.30 diff -u -r1.30 critsection.c --- scheduler/critsection.c 2001/08/09 21:21:13 1.30 +++ scheduler/critsection.c 2001/08/16 16:30:11 @@ -86,7 +86,7 @@ /*********************************************************************** * InterlockedCompareExchange (KERNEL32.@) */ -/* PVOID WINAPI InterlockedCompareExchange( PVOID *dest, PVOID xchg, PVOID compare ); */ +/* LONG WINAPI InterlockedCompareExchange( PLONG dest, LONG xchg, LONG compare ); */ __ASM_GLOBAL_FUNC(InterlockedCompareExchange, "movl 12(%esp),%eax\n\t" "movl 8(%esp),%ecx\n\t" @@ -95,10 +95,31 @@ "ret $12"); /*********************************************************************** + * InterlockedCompareExchangePointer (KERNEL32.@) + */ +/* PVOID WINAPI InterlockedCompareExchangePointer( PVOID *dest, PVOID xchg, PVOID compare ); */ +__ASM_GLOBAL_FUNC(InterlockedCompareExchangePointer, + "movl 12(%esp),%eax\n\t" + "movl 8(%esp),%ecx\n\t" + "movl 4(%esp),%edx\n\t" + "lock; cmpxchgl %ecx,(%edx)\n\t" + "ret $12"); + +/*********************************************************************** * InterlockedExchange (KERNEL32.@) */ /* LONG WINAPI InterlockedExchange( PLONG dest, LONG val ); */ __ASM_GLOBAL_FUNC(InterlockedExchange, + "movl 8(%esp),%eax\n\t" + "movl 4(%esp),%edx\n\t" + "lock; xchgl %eax,(%edx)\n\t" + "ret $8"); + +/*********************************************************************** + * InterlockedExchangePointer (KERNEL32.@) + */ +/* PVOID WINAPI InterlockedExchangePointer( PVOID *dest, PVOID val ); */ +__ASM_GLOBAL_FUNC(InterlockedExchangePointer, "movl 8(%esp),%eax\n\t" "movl 4(%esp),%edx\n\t" "lock; xchgl %eax,(%edx)\n\t" -- Alexandre Julliard julliard(a)winehq.com

4 5

Re: Toolbar background error
by Alexandre Julliard Aug. 17, 2001

Aug. 17, 2001

Gerard Patel <gerard.patel(a)nerim.net> writes: > By changing rgrc[0] when responding to WM_NCCALCSIZE. > It seems that it is only possible to erase properly the 'custom' nonclient > area by using GetWindowDC... So : > > --- defwnd.c.orig Sat Aug 11 10:23:18 2001 > +++ defwnd.c Fri Aug 17 19:47:17 2001 > @@ -462,8 +462,12 @@ > if (!hbr) return 0; > > /* GetClientRect used to be GetClipBox, but it is not what > - * Windows does, and it breaks badly with CS_PARENTDC */ > - GetClientRect( wndPtr->hwndSelf, &rect ); > + * Windows does, and it breaks badly with CS_PARENTDC > + * GetClientRect breaks too when we are called with a window > + * DC (!!) - what next ? */ > + rect.left = rect.top = 0; > + rect.right = wndPtr->rectWindow.right - wndPtr->rectWindow.left; > + rect.bottom = wndPtr->rectWindow.bottom - wndPtr->rectWindow.top; > DPtoLP( hdc, (LPPOINT)&rect, 2 ); > FillRect( hdc, &rect, hbr ); > return 1; > > What do you think ? That's not what Windows does. If you pass a window DC to WM_ERASEBKGND it only erases an area equivalent to the client area (shifted to be relative to the whole window), so GetClientRect is correct IMO. The non-client area should be painted upon WM_NCPAINT, not WM_ERASEBKGND. -- Alexandre Julliard julliard(a)winehq.com

1 0

err:module:map_image Could not map section .data, file probably truncated
by mdew Aug. 17, 2001

Aug. 17, 2001

a number of apps that wouldve worked normally ..are now failing to load due to this error.. err:module:map_image Could not map section .data, file probably truncated /usr/local/bin/wine: can't exec './xxx.exe': error=21 -- id: 9DF3262C key fingerprint: B556 04DA CFD9 78DD 9215 A970 8196 FB6D 9DF3 262C

3 3

Re: Does Wine's winhlp work?
by Gerard Patel Aug. 17, 2001

Aug. 17, 2001

At 02:43 PM 17/08/2001 +0200, you wrote: >I was about to ask the same question. My impression was that: >i) 'winhelp' is completely non-functional <snip> I think so - I have never found one help file that it can read. Gerard

1 0

Re: Toolbar background error
by Gerard Patel Aug. 17, 2001

Aug. 17, 2001

At 05:10 PM 16/08/2001 GMT, you wrote: >It's the client rectangle in the toolbar (I believe). I've handed the >issue back to Guy. I think that's a general issue, not specific to toolbars. We are seeing it for toolbars because it's a trick that is used mainly for toolbars. What happens I think is that some apps are not satisfied with the standard toolbar look. They want to have a bigger border around the buttons. That is not provided by the standard control, at least when these apps were written (maybe it is now, I am not following these matters closely). What these apps are doing is playing with WM_NCCALCSIZE to restrict the client rectangle to something smaller than what remains when the system clips the true nonclient area. In the case of one of my test apps, this is done by subclassing (the Wine code is never called). Since GetWindowRect returns only the width and the height of the client area, and is setting the origin to 0 - 0, the custom client area appears higher that it is to WM_ERASEBKGND and the lower part of the window is not painted (toolbar in the case of the app I see) Now, I don't know yet how should be done the erasing of this custom nonclient area :-/ Gerard

2 1

Re: Does Wine's winhlp work?
by Waldek Hebisch Aug. 17, 2001

Aug. 17, 2001

David D. Hagood wrote: >Is Wine's WinHelp program supposed to work? By that I mean, if I invoke >that program with a Windows help file, is it supposed to be able to >display the contents of the help file as Window's winhelp.exe would? I was about to ask the same question. My impression was that: i) 'winhelp' is completely non-functional ii) part reading help files only handles very old help files I implemented code to read somewhat newer help files, to degree that 'hlp2sgml' program is able to get some output, but since data structures used by graphical part are largly NOT filled, 'winhelp' itself has more reasons not to work (I have not sent patch to 'hlpfile.c' in case 'winhelp' actually can read some files). -- Waldek Hebisch hebisch(a)math.uni.wroc.pl or hebisch(a)hera.math.uni.wroc.pl

1 0

Does Wine's winhlp work?
by David D. Hagood Aug. 17, 2001

Aug. 17, 2001

Is Wine's WinHelp program supposed to work? By that I mean, if I invoke that program with a Windows help file, is it supposed to be able to display the contents of the help file as Window's winhelp.exe would? I ask because 1) I have a program (DeLorme's AAA MapNGo) that uses WinHelp format documents for some of its normal operation, and 2) I have a WinHelp format file I'd dearly like to be able to access directly from Wine. When I try to launch Wine's Winhelp, it brings up a window, but the window is empty. If I go to File->Open and try to open a WinHelp file, I get the same result. Is there some setup I'm missing?

3 2

Refcounting in SelectObject
by Uwe Bonnes Aug. 17, 2001

Aug. 17, 2001

Hallo, we discussed this January 2001 with no final outcome. Some application (gcprevue version 9) does 080673a0:Call gdi32.SelectObject(000011d4,000011d8) ret=5f485bb1 080673a0:Ret gdi32.SelectObject() retval=00000044 ret=5f485bb1 080673a0:Call gdi32.DeleteObject(000011d8) ret=5f487813 080673a0:Ret gdi32.DeleteObject() retval=00000001 ret=5f487813 080673a0:Call gdi32.BitBlt(000011d4,00000000,00000000,0000008c,000001a7,000009d0,00000000,00000000,00cc0020) ret=0048d1e6 and gets an XErrorEvent and fails to Debugbreak. Starting theapplication with winedbg still gets this error, but DebugBreak isn't executed and the application proceeds. The Windows Api explicitly tells for DeleteObject: Return Value: ... If the specified handle is not valid or is currently selected into a device context, the return value is FALSE. Windows Internals tells about some elements in GDIOBJHDR in debugbuild of windows and has pseudo code for incrementing some refrence counters in SelectObject. Appended patch tries to implement this scheme. The original discussion dissolved into nothing after Andi saw the need for such refcounting, but also mentioned possible resource leakage. Appended code tries to clean up better by: - Remembering an unsuccessfull DeleteObject by adding 0x8000 to the Refcount - Call DeleteObject in SelectObject when the RefCount of the released handle is 0x8000 - Decrementing the RefCount when a hBitmap is still selected into a DC when DeleteDC is called for that DC - Call DeleteObject in DeleteDC when RefCount is 0x8000 For some applications I tested, I always saw the delayed DeleteObject done after the refused DeleteObject in the first place. Is this approach better? Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Free Software: If you contribute nothing, expect nothing -- Index: wine/include/gdi.h =================================================================== RCS file: /home/wine/wine/include/gdi.h,v retrieving revision 1.50 diff -u -r1.50 gdi.h --- wine/include/gdi.h 2001/08/15 23:33:20 1.50 +++ wine/include/gdi.h 2001/08/16 23:23:12 @@ -45,6 +45,9 @@ HANDLE16 hNext; WORD wMagic; DWORD dwCount; + WORD wMetaList; + WORD wSelCount;/* only in debug builds of windows */ + WORD wObjTask;/* only in debug builds of windows */ } GDIOBJHDR; Index: wine/objects/gdiobj.c =================================================================== RCS file: /home/wine/wine/objects/gdiobj.c,v retrieving revision 1.54 diff -u -r1.54 gdiobj.c --- wine/objects/gdiobj.c 2001/08/16 19:13:52 1.54 +++ wine/objects/gdiobj.c 2001/08/16 23:23:13 @@ -374,6 +374,7 @@ obj->hNext = 0; obj->wMagic = magic|OBJECT_NOSYSTEM; obj->dwCount = ++count; + obj->wSelCount = 0; TRACE_SEC( *handle, "enter" ); return obj; @@ -545,6 +546,15 @@ GDI_ReleaseObj( obj ); return TRUE; } + + if(header->wSelCount) + { + TRACE("delayed for %04x because object in use, count %d\n", + obj,header->wSelCount); + header->wSelCount+=0x8000; /* mark for delete */ + GDI_ReleaseObj( obj ); + return FALSE; + } TRACE("%04x\n", obj ); @@ -814,12 +824,32 @@ HGDIOBJ WINAPI SelectObject( HDC hdc, HGDIOBJ handle ) { HGDIOBJ ret = 0; + GDIOBJHDR *header; + DC * dc = DC_GetDCUpdate( hdc ); if (!dc) return 0; TRACE("hdc=%04x %04x\n", hdc, handle ); if (dc->funcs->pSelectObject) ret = dc->funcs->pSelectObject( dc, handle ); GDI_ReleaseObj( hdc ); + if (ret) + { + if (!(header = GDI_GetObjPtr( handle, MAGIC_DONTCARE ))) return 0; + header->wSelCount++; + GDI_ReleaseObj(handle); + if (!(header = GDI_GetObjPtr( ret, MAGIC_DONTCARE ))) return 0; + if(header->wSelCount) + header->wSelCount--; + if(header->wSelCount == 0x8000) /* handle delayed DeleteObject*/ + { + header->wSelCount = 0; + GDI_ReleaseObj(ret); + TRACE("executing delayed DeleteObject for %04x\n",ret); + DeleteObject(ret); + } + else + GDI_ReleaseObj(ret); + } return ret; } Index: wine/objects/dc.c =================================================================== RCS file: /home/wine/wine/objects/dc.c,v retrieving revision 1.62 diff -u -r1.62 dc.c --- wine/objects/dc.c 2001/08/16 19:01:23 1.62 +++ wine/objects/dc.c 2001/08/16 23:23:14 @@ -766,6 +766,25 @@ if (dc->hGCClipRgn) DeleteObject( dc->hGCClipRgn ); if (dc->pAbortProc) THUNK_Free( (FARPROC)dc->pAbortProc ); if (dc->hookThunk) THUNK_Free( (FARPROC)dc->hookThunk ); + if (dc->hBitmap) + { + GDIOBJHDR * header; + if ((header = GDI_GetObjPtr( dc->hBitmap, MAGIC_DONTCARE ))) + { + if(header->wSelCount ) + header->wSelCount--; + if(header->wSelCount == 0x8000) + { + header->wSelCount = 0; + GDI_ReleaseObj(dc->hBitmap); + TRACE("executing delayed DeleteObject for %04x\n",dc->hBitmap); + DeleteObject(dc->hBitmap); + } + else + GDI_ReleaseObj(dc->hBitmap); + + } + } PATH_DestroyGdiPath(&dc->path); GDI_FreeObject( hdc, dc );

1 0

Re: DC lock in CreateDIBitmap
by Alexandre Julliard Aug. 16, 2001

Aug. 16, 2001

Uwe Bonnes <bon(a)elektron.ikp.physik.tu-darmstadt.de> writes: > CreateDIBitmap is called with the HDC lock hold, and when the (fColor) clause > is true, CreateDCA() checks that the lock is not hold. A laymans approach to > release hdc before CreateDCA() and reopen it afterwards resulted in this > report asr startup, with no visual output: I don't think we should be using CreateDC at all in CreateDIBitmap. Could you try this? Index: objects/dib.c =================================================================== RCS file: /opt/cvs-commit/wine/objects/dib.c,v retrieving revision 1.59 diff -u -r1.59 dib.c --- objects/dib.c 2001/08/15 23:33:20 1.59 +++ objects/dib.c 2001/08/16 21:08:12 @@ -875,12 +875,7 @@ /* Now create the bitmap */ - if (fColor) - { - HDC tmpdc = CreateDCA( "DISPLAY", NULL, NULL, NULL ); - handle = CreateCompatibleBitmap( tmpdc, width, height ); - DeleteDC( tmpdc ); - } + if (fColor) handle = CreateCompatibleBitmap( hdc, width, height ); else handle = CreateBitmap( width, height, 1, 1, NULL ); if (!handle) return 0; -- Alexandre Julliard julliard(a)winehq.com

2 1

Problems building with extern inline functions in winebase.h
by Mike Bond Aug. 16, 2001

Aug. 16, 2001

A couple days ago a number of inline functions were put into winebase.h. Since then I have been unable to build, using gcc 2.96-81 (RedHat). Apparently this version of gcc thinks that because there is an extern prefixing the inline, even though the function does in fact have a body, that the function really is external. Removing the extern, just having inline as one would do in C++, does not appear to work correctly either, and causes the function to appear as a exported symbol in all the object files that header gets included with, thus resulting in multiple definitions. I'm not sure what the correct solution would be, other than using the older function call method for this version of gcc. Perhaps I'm just missing some variant on the use of inline? -- TTFN MikeB

5 5

Re: patch.py?id=997918367324549717828631
by Alexandre Julliard Aug. 16, 2001

Aug. 16, 2001

lawson_whitney(a)juno.com writes: > I can't remember that I've ever seen the fixme that was here before, > but I'll check again when I send this. No, the CheckNotSysLevel calls were not done before. I added a few of them in GDI to find potential locking bugs; seems pretty successful so far ;-) I think this should fix your problem: Index: objects/dc.c =================================================================== RCS file: /opt/cvs-commit/wine/objects/dc.c,v retrieving revision 1.61 diff -u -r1.61 dc.c --- objects/dc.c 2001/08/15 23:33:20 1.61 +++ objects/dc.c 2001/08/16 17:34:45 @@ -524,7 +524,9 @@ success=FALSE; } GDI_ReleaseObj( hdcs ); + GDI_ReleaseObj( hdc ); DeleteDC( hdcs ); + if (!(dc = DC_GetDCPtr( hdc ))) return FALSE; } GDI_ReleaseObj( hdc ); return success; -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: [resend] Don't crash when CUPS is installed
by mm＠caldera.de Aug. 16, 2001

Aug. 16, 2001

In article <997995321.1690.4.camel(a)bugg> you wrote: > Hi all, > Having CUPS installed and a printer in CUPS configured would crash an > application I'm looking at. The reason was that the printer would be > added with a NULL name. > - if (!AddPrinterA(NULL,2,(LPBYTE)&pinfo2a)) { > + if (!AddPrinterA(printers[i],2,(LPBYTE)&pinfo2a)) { This will just make the printer not added AT ALL, since there is the following code in AddPrinterW: if(pName != NULL) { ERR("pName = %s - unsupported\n", debugstr_w(pName)); SetLastError(ERROR_INVALID_PARAMETER); return 0; } Please do not apply. Ciao, Marcus

1 0

DC lock in CreateDIBitmap
by Uwe Bonnes Aug. 16, 2001

Aug. 16, 2001

Hallo, with the new GDI_CheckNotLock() stuff, some application (gcprevue 9) fails with > err:win32:_CheckNotSysLevel Holding lock 0x4079b9e8 level 3 Backtrace is: Backtrace: =>0 0x400e21d1 (_CheckNotSysLevel+0x51(lock=0x4079b9e8) [syslevel.c:152] in libntdll.so) (ebp=405c6090) 1 0x4078261c (GDI_CheckNotLock+0x1c [gdiobj.c:510] in libgdi32.so) (ebp=405c60a0) 2 0x4077823d (CreateDCA+0x1d(driver=0x407909d3, device=0x0, output=0x0, initData=0x0, s=0x0, s=0x0, s=0x407909d3) [dc.c:558] in libgdi32.so) (ebp=405c61e8) 3 0x4077bbc6 (CreateDIBitmap+0x166(hdc=0xdec, header=0x41eadb50, init=0x0, bits=0x0, data=0x41eadb50, coloruse=0x0) [dib.c:881] in libgdi32.so) (ebp=405c6218) 4 0x409f9479 (X11DRV_DIB_CreateDIBSection+0x359(dc=0x404225f4, bmi=0x41eadb50, usage=0x0, bits=0x411000d8, section=0x0, offset=0x0, ovr_pitch=0x0) [dib.c:3914] in libx11drv.so) (ebp=405c62a0) 5 0x4077bd75 (DIB_CreateDIBSection+0x65(hdc=0xdec, bmi=0x41eadb50, usage=0x0, bits=0x411000d8, section=0x0, offset=0x0, ovr_pitch=0x0) [dib.c:944] in libgdi32.so) (ebp=405c62d4) 6 0x4077bdd1 (CreateDIBSection+0x31(hdc=0xdec, bmi=0x41eadb50, usage=0x0, bits=0x411000d8, section=0x0, offset=0x0) [dib.c:961] in libgdi32.so) (ebp=405c6304) 7 0x00401fba (gcprevue.exe..text+0xfba in D:\cae\gcp9.0.0.8\gcprevue.exe) (ebp=ffffffff) CreateDIBitmap is called with the HDC lock hold, and when the (fColor) clause is true, CreateDCA() checks that the lock is not hold. A laymans approach to release hdc before CreateDCA() and reopen it afterwards resulted in this report asr startup, with no visual output: fixme:win32:PE_CreateModule Security directory ignored err:win32:_LeaveSysLevel (0x4079ba08, level 3): Invalid state: count 0 mutex (nil). err:ntdll:RtlpWaitForCriticalSection section 0x4010fb88 "rtl.c: peb_lock" wait timed out, retrying (60 sec) fs=030f err:ntdll:RtlpWaitForCriticalSection section 0x4010fb88 "rtl.c: peb_lock" wait timed out, retrying (60 sec) fs=031f err:ntdll:RtlpWaitForCriticalSection section 0x4079ba08 "gdiobj.c: GDI_level" wait timed out, retrying (60 sec) fs=008f How to cure this problem? Bye Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Free Software: If you contribute nothing, expect nothing --

1 0

Stupid Winedbg question
by Bobby Bingham Aug. 16, 2001

Aug. 16, 2001

Ok, I feel like an idiot for having to ask this, but it's my first time realing using the debugger. When wine starts the debugger on an exception, it opens it up in another xterm. Is there an easy way to copy it's output? The usual select and middle-click isn't working because it doesn't let me select the text to begin with. Thanks in advance for your patience. Bobby Bingham

3 2