April 2002 - Wine-Devel - list.winehq.org

Followup: Urgent need for advice: POLLHUP and sockets

by Martin Wilck

I just might remark that with the two changes I discussed in the previous message (do not call set_select_events (..., -1), signal FD_CLOSE only once) I seem to finally get rid of the problems with the close()/shutdown() semantics that I talked about before. But only at the cost of very high wineserver CPU usage. Martin -- Martin Wilck Phone: +49 5251 8 15113 Fujitsu Siemens Computers Fax: +49 5251 8 20409 Heinz-Nixdorf-Ring 1 mailto:Martin.Wilck(a)Fujitsu-Siemens.com D-33106 Paderborn http://www.fujitsu-siemens.com/primergy

April 16, 2002

Re: Urgent need for advice: POLLHUP and sockets

by Martin Wilck

On Tue, 16 Apr 2002, Michael Cardenas wrote: > Could you use the set_select_events to say that you don't want any more > POLLHUP events? No you can't - that's how poll() works (at least on Linux). <man poll> int poll(struct pollfd *ufds, unsigned int nfds, int timeout); [...] The field revents is an output parameter, filled by the kernel with the events that actually occurred, either of the type requested, or of one of the types POLLERR or POLLHUP or POLLNVAL. (These three bits are meaningless in the events field, and will be set in the revents field whenever the corresponding condition is true.) </man poll> Martin -- Martin Wilck Phone: +49 5251 8 15113 Fujitsu Siemens Computers Fax: +49 5251 8 20409 Heinz-Nixdorf-Ring 1 mailto:Martin.Wilck(a)Fujitsu-Siemens.com D-33106 Paderborn http://www.fujitsu-siemens.com/primergy

April 16, 2002

Sorry for spamming wine bugs list

by Andriy Palamarchuk

I just changed a dependency of some task list bug and a message was sent to the wine bugs list for each entry in the task list. Probably we need better filter for notifications which sent to the bugs list. Andriy Palamarchuk __________________________________________________ Do You Yahoo!? Yahoo! Tax Center - online filing with TurboTax http://taxes.yahoo.com/

April 16, 2002

Re: Urgent need for advice: POLLHUP and sockets

by Michael Cardenas

Martin Wilck wrote: >My first reaction to this was to comment out the above call to >set_select_events() and continue listening to events from the socket. >However, due to the semantics of poll(), this will cause a POLLHUP event >on this socket for each and every iteration of the wineserver's main select >loop, putting the server basically into a busy-wait state. > Could you use the set_select_events to say that you don't want any more POLLHUP events? -- michael cardenas lead windows compatibility engineer www.lindows.com "Be the change you want to see in the world." -Mahatma Gandhi

April 16, 2002

Urgent need for advice: POLLHUP and sockets

by Martin Wilck

Hi all, please consider the following problem: On sockets, when POLLHUP is received, it is perfectly legitimate to continue reading data until a read() or recv() call returns 0 bytes. The current wine implementation calls set_select_events( &sock->obj, -1 ) when POLLHUP or POLLERR is received, disabling any future events from the socket. However, if an application uses asynchronous IO (or only asynchronous notification via WSAAsyncSelect() or WSAEventSelect()), this also inhibits reception of POLLIN events which can perfectly well occur if there are still unread data in the kernel buffers. Thus the app will never notice that there is still data to be read. My first reaction to this was to comment out the above call to set_select_events() and continue listening to events from the socket. However, due to the semantics of poll(), this will cause a POLLHUP event on this socket for each and every iteration of the wineserver's main select loop, putting the server basically into a busy-wait state. How can this be resolved? We seem to have the choice between a) busy loop in the server (bad) b) discarding data which could be legitimately read by the application (very bad). Any hints/comments/ideas? Please note that it is insufficient to queue all pending APCs on the socket when POLLHUP is received, because the app may very well queue some _after_ the close event arrives. A minor point is that sock_poll_event() also sets a FD_CLOSE event (if present) everytime it is called with POLLHUP set, effectively preventing any APC from being called, at least if the app waits for FD_CLOSE and asynchronous send/recv operations at the same time (likely). This can be worked around by setting FD_CLOSE only once and then blocking it, although I am not sure if that is correct behaviour (permanently signalling FD_CLOSE is obviously incorrect). I hope someone has a good idea how to resolve this, Martin -- Martin Wilck Phone: +49 5251 8 15113 Fujitsu Siemens Computers Fax: +49 5251 8 20409 Heinz-Nixdorf-Ring 1 mailto:Martin.Wilck(a)Fujitsu-Siemens.com D-33106 Paderborn http://www.fujitsu-siemens.com/primergy

April 16, 2002

Re: fixme:win32:PeekNamedPipe

by Michael Cardenas

Michael Riedel wrote: > Thanks for the hint, and yes it is a kernel32 function. ok, then there's some work to do. > fixme:win32:PeekNamedPipe (00000080, (nil), 00000000, (nil), > 0x406a2194, (nil)): stub > > this means that at least a valid handle was created so that there is > already some code for CreateNamedPipe or CreatePipe? This means there's a stub. So, there's a placeholder with no code in it. > Is there anybody knowing more before I am going to try looking into > the sources? I'm not more knowing about this function per se, I'm sure whoever else wrote the rest of the named pipe implementation will know. But I do know that you can do this to find the stub, and possibly the file that has the other named pipe functions: find (winesrcdir) -name *.c | xargs grep PeekNamedPipe -- michael cardenas lead windows compatibility engineer www.lindows.com "Be the change you want to see in the world." -Mahatma Gandhi

April 16, 2002

Re: Urgent need for advice: POLLHUP and sockets

by Michael Cardenas

> My first reaction to this was to comment out the above call to > set_select_events() and continue listening to events from the socket. > However, due to the semantics of poll(), this will cause a POLLHUP event > on this socket for each and every iteration of the wineserver's main select > loop, putting the server basically into a busy-wait state. well, can you just ignore the POLLHUP events in the server for that socket after the first one is received?

April 16, 2002

Re: fixme:win32:PeekNamedPipe

by Michael Riedel

Thanks for the hint, and yes it is a kernel32 function. BOOL PeekNamedPipe( HANDLE hNamedPipe, // handle to pipe LPVOID lpBuffer, // data buffer DWORD nBufferSize, // size of data buffer LPDWORD lpBytesRead, // number of bytes read LPDWORD lpTotalBytesAvail, // number of bytes available LPDWORD lpBytesLeftThisMessage // unread bytes ); Since I get fixme:win32:PeekNamedPipe (00000080, (nil), 00000000, (nil), 0x406a2194, (nil)): stub this means that at least a valid handle was created so that there is already some code for CreateNamedPipe or CreatePipe? Is there anybody knowing more before I am going to try looking into the sources? Thanks, Michael --

April 16, 2002

Re: fixme:win32:PeekNamedPipe

by Michael Cardenas

Try searching for PeekNamedPipe on msdn.microsoft.com and at the bottom it will tell you what dll the function is in. unfortunately, if it's in the kernel, you won't be able to use a native dll. On Tuesday 16 April 2002 03:40 am, Michael Riedel wrote: > First of all I want to thank the community for the efforts spent in > developing wine. It's simple a cool "tool" (sorry, it's more, I know). > > Now my problem: I installed wine-20020411 without having a native > windows installation. I installed successfully a simulator software > which obviously uses a tcl/tk-GUI to invoke other program moduls > (compiler, library manager, simulator core) re-directing their > stdin/stdout to the console window via pipes. It does not work for me > and I get the message (infinitivly repeated): > > fixme:win32:PeekNamedPipe (00000080, (nil), 00000000, (nil), 0x406a2194, > (nil)): stub > > Searchin the web I got no hit. > > Questions: > > (1) Can I avoid this by using a native windows-dll? If yes, which dll I > have to install (but I assume it is a kernel-dll function and so it is > not possible)? > > (2) I have no Windows programming background, but what do you think: > Should I dare to implement this function or was it left open because of > a tricky windows-API-Linux mapping problem? > > (3) Any other ideas? > > Thanks, > Michael

April 16, 2002

a problem using /tools/wineinstall

by davide madrisan

I had a problem using the wine installer script from wine-20020411. This is the brute change I've made to be able to end successfully the installation procedure. echo "Installing default Wine registry entries..." echo ## DEBUG : TMPREG == "/tmp/wineinstall.reg" ## DEBUG : REGAPI == "programs/regapi/regapi" # if ! $REGAPI setValue < $TMPREG > /dev/null <-- here is the problem # then { # rm -f $TMPREG # echo "Registry install failed." # conf_reset_question regapi_error # conf_question high regapi_error # exit 1 # } # else { # if we are converting from a .winerc file, running regapi once # will ONLY convert .winerc -> ~/.wine/config, it will not import the # registry data. so if we are converting we need to run regapi twice if [ "$CONVCONF" = 'yes' ] then if ! $REGAPI setValue < $TMPREG > /dev/null then rm -f $TMPREG echo "Registry install failed." conf_reset_question regapi_error conf_question high regapi_error exit 1 else echo echo "Registry entries successfully installed." fi else echo echo "Registry entries successfully installed." fi # } # fi I've choosen a wine-only installation, so no windows partition. If you need more informations, please let me know. Thanks. Saluti. -- Davide Madrisan GPG/PGP keys: <http://pgp.mit.edu/>

April 16, 2002

fixme:win32:PeekNamedPipe

by Michael Riedel

First of all I want to thank the community for the efforts spent in developing wine. It's simple a cool "tool" (sorry, it's more, I know). Now my problem: I installed wine-20020411 without having a native windows installation. I installed successfully a simulator software which obviously uses a tcl/tk-GUI to invoke other program moduls (compiler, library manager, simulator core) re-directing their stdin/stdout to the console window via pipes. It does not work for me and I get the message (infinitivly repeated): fixme:win32:PeekNamedPipe (00000080, (nil), 00000000, (nil), 0x406a2194, (nil)): stub Searchin the web I got no hit. Questions: (1) Can I avoid this by using a native windows-dll? If yes, which dll I have to install (but I assume it is a kernel-dll function and so it is not possible)? (2) I have no Windows programming background, but what do you think: Should I dare to implement this function or was it left open because of a tricky windows-API-Linux mapping problem? (3) Any other ideas? Thanks, Michael --

April 16, 2002

Re: building documentation failed

by Stefan Leichter

Michael Stefaniuc wrote: > > On Sat, Apr 13, 2002 at 01:36:37PM +0200, Stefan Leichter wrote: > > when i run make inside the ~wine/documentation directory i get an error when > > the pdf is created. The fix is easy for me, just an additional parameter (-s) > > for the command db2pdf. > > > > But before i send the attached patch to wine-patches i like to know if anyone > > can run make successful in the documentation directory. In this case we may > > have to add a check in the configure script. > > > > If someone is able to build the documentation successful please post the > > output of "db2pdf -v". > Yes, it works without the "-s" for me. > camus:~/work/wine/documentation$ db2pdf -v > DocBook-utils version 0.6.9 (jw version 1.1) > Hello, does the documentation build with '-s' too? Tschüß Stefan

April 16, 2002

Re: wine/ misc/options.c scheduler/process.c

by Dimitrie O. Paun

On April 15, 2002 10:14 am, Michael Cardenas wrote: > When was this patch applied? I don't see it on the wine-patches list. Maybe you should look in the wine-cvs list? :-) -- Dimi.

April 15, 2002

*update*: fixed tools/wineconf and tools/wineinstall

by Speeddymon

ok guys, I know some of you are going to flame me for my post to wine-patches, but I can explain...I did rtfm (the patch submitting archive) for the format of submitting a patch, but I have no internet at home my college's computers are all windows-only, so please, forgive me for sending the whole files and not just 2 patch files... __________________________________________________ Do You Yahoo!? Yahoo! Tax Center - online filing with TurboTax http://taxes.yahoo.com/

April 15, 2002

Re: PATCH: accept-deferred.diff (5 in series)

by Dimitrie O. Paun

On April 15, 2002 09:38 am, Martin Wilck wrote: > On Fri, 12 Apr 2002, Dimitrie O. Paun wrote: > > > + if ( lpCallerData || lpCalleeData || lpSQOS || lpGQOS ) > > > + WARN ("unsupported parameters!"); > > > > ^^^^ > > Shouldn't this be a FIXME instead? > > I don't think so - there is little to fix on the part of wine. > AFAICS the Linux network stack simply doesn't support passing of > additional parameters along with a connection request. But this is irrelevant. Wine is supposed to implement the Win32 API, whether we know how to do it or not (on some platforms) is another matter. But we should be vocal (i.e. FIXME :)) when we can not implement all the features, this way people will implement them in the future, even if this requires modifications to the kernel. There are only few cases where WARN is the right answer. Most messages are either ERR, FIXME or TRACE. -- Dimi.

April 15, 2002

Re: wine/ misc/options.c scheduler/process.c

by Alexandre Julliard

Uwe Bonnes <bon(a)elektron.ikp.physik.tu-darmstadt.de> writes: > The wine executable now doesn't understand OPTIONS any more. It is > nescessary to give options in the WINEOPTIONS environment variable. Is this > intended? No, stupid typo on my part. Should be fixed now. -- Alexandre Julliard julliard(a)winehq.com

April 15, 2002

Re: Patch series for async socket IO & winsock2

by Michael Cardenas

AOL7, when used over a modem, uses async sockets with overlapped IO. I will be working on this again in a few weeks, and I'll keep you posted. On Friday 12 April 2002 11:10 am, Martin Wilck wrote: > Hi, > > I will soon start submitting this week's work on async sockets. > > As I said in yesterday's "good news/bad news" posting, this is work in > progress and far from being complete. However, I strongly need feedback > and perhaps some testing by adventurous people. > > Anybody out there with an app that uses overlapped socket IO ??

April 15, 2002

Re: wine/ misc/options.c scheduler/process.c

by Uwe Bonnes

>>>>> "Alexandre" == Alexandre Julliard <julliard(a)wine.codeweavers.com> writes: Alexandre> Log message: Fixed inheriting command-line options from Alexandre> parent process. The wine executable now doesn't understand OPTIONS any more. It is nescessary to give options in the WINEOPTIONS environment variable. Is this intended? Bye -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

April 15, 2002

Re: wine/ misc/options.c scheduler/process.c

by Michael Cardenas

When was this patch applied? I don't see it on the wine-patches list. On Monday 15 April 2002 10:31 am, Uwe Bonnes wrote: > >>>>> "Alexandre" == Alexandre Julliard <julliard(a)wine.codeweavers.com> > >>>>> writes: > > Alexandre> Log message: Fixed inheriting command-line options from > Alexandre> parent process. > > > The wine executable now doesn't understand OPTIONS any more. It is > nescessary to give options in the WINEOPTIONS environment variable. Is this > intended? > > Bye

April 15, 2002

Re: Patch: sock-shutdown-file.diff (11 in series)

by Martin Wilck

On Fri, 12 Apr 2002, Dimitrie O. Paun wrote: > On April 12, 2002 11:25 am, Martin Wilck wrote: > > + if ( flags & FD_FLAG_RECV_SHUTDOWN ) { > > + /* Fixme: WSAESHUTDOWN is unavailable */ > ^^^^^^ > Don't we want these things as FIXME("WSAESHUTDOWN is unavailable") > so we see when we need to fix things? I need to figure out what Windows returns here. I think WSAESHUTDOWN cannot be returned because it is a socket error unknown to the ReadFile() function. We might be able to fix this by handling the error in the server. Anyway, thanks for looking at my code so thoroughly! Martin -- Martin Wilck Phone: +49 5251 8 15113 Fujitsu Siemens Computers Fax: +49 5251 8 20409 Heinz-Nixdorf-Ring 1 mailto:Martin.Wilck(a)Fujitsu-Siemens.com D-33106 Paderborn http://www.fujitsu-siemens.com/primergy

April 15, 2002

Re: PATCH: accept-deferred.diff (5 in series)

by Martin Wilck

On Fri, 12 Apr 2002, Dimitrie O. Paun wrote: > > + if ( lpCallerData || lpCalleeData || lpSQOS || lpGQOS ) > > + WARN ("unsupported parameters!"); > ^^^^ > Shouldn't this be a FIXME instead? I don't think so - there is little to fix on the part of wine. AFAICS the Linux network stack simply doesn't support passing of additional parameters along with a connection request. I am not so sure about the QOS parameters, but that appears to be a secondary issue. Martin -- Martin Wilck Phone: +49 5251 8 15113 Fujitsu Siemens Computers Fax: +49 5251 8 20409 Heinz-Nixdorf-Ring 1 mailto:Martin.Wilck(a)Fujitsu-Siemens.com D-33106 Paderborn http://www.fujitsu-siemens.com/primergy

April 15, 2002

Re: Regression tests, UNICODE vs ASCII

by Lina Kemmel

You can write something like the following: const WCHAR zeroWidthNonJoiner = 0x200C; const WCHAR zeroWidthJoiner = 0x200D; const WCHAR leftToRightMark = 0x200E; const WCHAR rightToLeftMark = 0x200F; const WCHAR leftToRightEmbedding = 0x202A; const WCHAR rightToLeftEmbedding = 0x202B; const WCHAR popDirectionalFormatting = 0x202C; const WCHAR leftToRightOverride = 0x202D; const WCHAR rightToLeftOverride = 0x202E; ( or enum { zeroWidthNonJoiner = 0x200C, zeroWidthJoiner, leftToRightMark, rightToLeftMark, leftToRightEmbedding = 0x202A, rightToLeftEmbedding, popDirectionalFormatting, leftToRightOverride, rightToLeftOverride }; ) const WCHAR test_string[] = {zeroWidthNonJoiner, zeroWidthJoiner, leftToRightMark, ... , rightToLeftOverride}; Lina Francois Gouget <fgouget(a)free.fr> Sent by: wine-devel-admin(a)winehq.com 14/04/02 10:02 م Please respond to Francois Gouget To: Shachar Shemesh <wine-devel(a)sun.consumer.org.il> cc: Andriy Palamarchuk <apa3a(a)yahoo.com>, Geoffrey Hausheer <i8e7fkwmsl1(a)phracturedblue.com>, <wine-devel(a)winehq.com> Subject: Re: Regression tests, UNICODE vs ASCII On Sun, 14 Apr 2002, Shachar Shemesh wrote: > One thing you may try are the explicit BiDi specifiers (Right to left, > left to right etc.). These are Unicode characters designed to explicitly > specify the direction of otherwise amigious characters. > > Ansii does not have, to the best of my knowledge, an equivalent encoded > characters. But what would these Unicode characters look like? I.e. what would I write in my code: const WCHAR test_string[]={'???','???','???'}; -- Francois Gouget fgouget(a)free.fr http://fgouget.free.fr/ Broadcast message : fin du monde dans cinq minutes, repentez vous !

April 15, 2002

Re: 16-bit apps starting and stopping notification

by Joshua Thielen

Andreas Mohr wrote: >On Sat, Apr 13, 2002 at 09:53:57AM -0400, Joshua Thielen wrote: > >>>Can somebody give me a hint how is better to approach >>>implementation of >>>NFY_EXITTASK, NFY_STARTTASK events generation? >>> >>Does anyone know what ToolHelpHook does (KERNEL.341)? Could this be used by windows to hook a toolhelp notification function into >>kernel.dll? >> >It installs notification handlers before important system events >(dll load, task exit, ...). >It returns the address of the previously installed notification handler, >thus allowing for chaining handlers. >It's basically a Win 3.1 replacement for RegisterPtrace(). >(same notification messages being used etc.) >The notification msgs are NFY_xxx in toolhelp.h, BTW. > >It's all described in "Undocumented Windows". > OK., it looks like toolhelp.dll should install a callback function into the kernel using ToolHelpHook. Whenever a kernel event occurs, this callback function would call each function in the NotifyRegister function list corresponding to the kernel event. But this presents a problem: How would you install a 32-bit wine callback function using ToolHelpHook16? I guess we could have separate ToolHelpHook32 and ToolHelpHook16 functions. I don't think it would be possible to chain the 32-bit and 16-bit functions together like they are in windows/hook.c. The kernel code would have to call the 16-bit and 32-bit hooknotify functions separately. Any better ideas? As I'm no wine pro, I hope I'm not spouting off meaningless nonsense here. Josh

April 15, 2002

Re: Directory reorganization

by Alexandre Julliard

Joshua Thielen <thielen(a)netprince.net> writes: > I have an idea for reorganizing some of the directories in wine. I > hope I'm not stating an obvious idea here, but I would like to group > everything into the following first level directories: > > dlls > documentation > include > programs > server > tools That's more or less the plan yes. We are slowly getting there. I have not moved everything yet because it's not yet clear where everything will go (most notably for the ntdll/kernel proper separation) and I don't want to move things twice given CVS weaknesses in that area. -- Alexandre Julliard julliard(a)winehq.com

April 14, 2002

Re: building documentation failed

by Michael Stefaniuc

Hello! On Sat, Apr 13, 2002 at 01:36:37PM +0200, Stefan Leichter wrote: > when i run make inside the ~wine/documentation directory i get an error when > the pdf is created. The fix is easy for me, just an additional parameter (-s) > for the command db2pdf. > > But before i send the attached patch to wine-patches i like to know if anyone > can run make successful in the documentation directory. In this case we may > have to add a check in the configure script. > > If someone is able to build the documentation successful please post the > output of "db2pdf -v". Yes, it works without the "-s" for me. camus:~/work/wine/documentation$ db2pdf -v DocBook-utils version 0.6.9 (jw version 1.1) bye michael -- Michael Stefaniuc Tel.: +49-711-96437-199 System Administration Fax.: +49-711-96437-111 Red Hat GmbH Email: mstefani(a)redhat.com Hauptstaetterstr. 58 http://www.redhat.de/ D-70178 Stuttgart

April 14, 2002