wine-devel

wine-devel@winehq.org

2 participants
84469 discussions

queue management problem
by eric pouech 19 May '01

19 May '01

after last night move of queue management to the server, one of my test app stopped working. What happens is that it's waiting on a WM_TIMER message, but which is no longer triggered. After some investigation, it turned out that message retrieval did change a bit: - in previous code, if the QS_PAINT bit was set, but no window to redraw was found, then the QS_TIMER bit was tested - current code, returns from the server a WM_PAINT message in any case, and if no window can be found, then we retry the whole loop, and we don't check the timer case for my, the wPaintCount was always > 1, meaning that the QS_PAINT bit was never reset then, I investigated why the wPaintCount was > 1. It seems that when a visible window, needing some redraw, is hidden before the update takes place, then we don't update the paint count (which should be decreased) the attached patch does take care of it (basically, if we find a hidden window with an update region, then we destroy the region, and clear the internal paint flag), but I don't think it's the best place to do it. any better idea ? A+ -- --------------- Eric Pouech (http://perso.wanadoo.fr/eric.pouech/) "The future will be better tomorrow", Vice President Dan Quayle Index: windows/win.c =================================================================== RCS file: /usr/share/cvs/cvsroot/wine/wine/windows/win.c,v retrieving revision 1.130 diff -u -r1.130 win.c --- windows/win.c 2001/05/18 22:51:57 1.130 +++ windows/win.c 2001/05/19 11:39:54 @@ -367,8 +367,15 @@ { if (!(pWnd->dwStyle & WS_VISIBLE)) { - TRACE("skipping window %04x\n", - pWnd->hwndSelf ); + TRACE("skipping window %04x\n", pWnd->hwndSelf); + if ((pWnd->flags & WIN_INTERNAL_PAINT) || pWnd->hrgnUpdate) + { + if (pWnd->hrgnUpdate) DeleteObject(pWnd->hrgnUpdate); + pWnd->hrgnUpdate = 0; + pWnd->flags &= ~WIN_INTERNAL_PAINT; + QUEUE_DecPaintCount(hQueue); + } + } else if ((pWnd->hmemTaskQ == hQueue) && (pWnd->hrgnUpdate || (pWnd->flags & WIN_INTERNAL_PAINT)))

1 0

Re: Patch for a bug in server/select.c
by Morten Welinder 19 May '01

19 May '01

> +#ifdef __NetBSD__ > + if (pollfd[i].revents & !(pollfd[i].revents & POLLNVAL)) > +#else > if (pollfd[i].revents) > +#endif I would hate to see this kind of __NetBSD__ pollution of Wine. (I believe something like it slipped through recently -- not good.) It used to be that documentation/porting contained lots of reasons why mentioning any OS name in places like this is a bad idea. I suggest a suitable configure test or a porting wrapper for poll. Morten

2 1

Re: PATCH: PeekMessage - reducing load
by Francois Gouget 18 May '01

18 May '01

On Sat, 19 May 2001, Ove Kaaven wrote: > > On Fri, 18 May 2001, Marcus Meissner wrote: > > > winword.exe apparently uses a while (PeekMessage()) loop. > > > > Both wineserver and 1 wine thread take up 90% CPU even if idle. > > > > This adds a timeout of 10 milliseconds so we are still interactive, > > but do not take up as much CPU. (We basically wait 1 timeslice if 100HZ > > scheduling). > > I question this... an app could be doing something important in such a > loop, i.e. while (!PeekMessage()) do_a_piece_of_important_stuff(), and > this would severely slow down things like that. In particular, I expect a > lot of games to use such loops. If you need CPU time reduction, at least > make it configurable. Me too. There are (badly written) applications that suffer greatly from this. Furthermore I did some tests on Win 95 and Word 97 only does that with some files, not all. Also, Word 2000 no longer behaves that badly. Now, if it were a per application setting it could make sense. -- Francois Gouget fgouget(a)free.fr http://fgouget.free.fr/ Any sufficiently advanced Operating System is indistinguishable from Linux

1 0

Re: Fix in Variant's coerce function
by Alexandre Julliard 18 May '01

18 May '01

Francois Gouget <fgouget(a)free.fr> writes: > So I'm in favor of uncommenting the calls in Coerce, and if these > cases are really being used then either we do a full-fledged > implementation or we write pseudo-stubs for these functions. > > Do we have a know app that crashes with this patch? It won't ever get to crashing, since it doesn't even link! Try a make checklink and you'll see. Unfortunately because of the way we link libraries you can actually create a .so that contains undefined symbols; it will only fail at load-time. This is why we have make checklink to spot this kind of problems at compile-time. -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: Fix in Variant's coerce function
by Alexandre Julliard 18 May '01

18 May '01

Francois Gouget <fgouget(a)free.fr> writes: > It seemed wrong, as did the 32 suffixes, and leaving all the > xxxFromDisp functions commented out. It may be a good idea to actually implement the functions before uncommenting the calls ;-) -- Alexandre Julliard julliard(a)winehq.com

2 1

Re: Question on exposing non-windows functions from dlls...
by Alexandre Julliard 18 May '01

18 May '01

Mike Bond <mbond(a)cox.rr.com> writes: > If you can point me to some way of implementing _exec[lv][ep] using the > standard win32 API I would be more than willing to. Unfortunately that > was one arguable shortcoming of win32 process management in that it did not > have any interface that allowed loading and running an image in-process. You cannot do this with the Win32 API, that's true. And since the native msvcrt is implemented on top of the Win32 API, this means that msvcrt exec functions cannot possibly behave exactly like the Unix ones. What you must do is find out the expected behavior of the msvcrt exec, and implement the same behavior; you should then be able to do this using the Win32 API, since this is how the native one does it. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: Networking with Mandrake 8.0 and SuSE 7.1
by Steve Fox 18 May '01

18 May '01

Unfortunately I can't really try the 2.4 kernel because Mandrake didn't get the PCMCIA token ring patch in there. I failed miserably to patch it in the last time I tried :( I plan to examine all the patches that Mandrake applies to their 2.2.19 kernel and back out some of them. Steve Fox IBM Linux Technology Center http://www.ibm.com/linux/ltc http://k-lug.com

2 1

Re: Networking with Mandrake 8.0 and SuSE 7.1
by gerard patel 18 May '01

18 May '01

At 10:14 AM 17/05/2001 -0500, you wrote: >I'm running Lotus Notes 5.04a under WINE with one of the January releases >and its been working great for months. However, any release since February >does not seem to do reliable networking on my Mandrake 8.0 system. With the other replies to this post, I'd say that it would be interesting to try to configure your Mandrake system with kernel 2.4 (I suspect that you choosed 2.2 kernel). Gerard

1 0

Re: Networking with Mandrake 8.0 and SuSE 7.1
by Steve Fox 18 May '01

18 May '01

In my case I'm using Mandrake's 2.2.19 kernel, but apparently the straight unpatched 2.2.19 works ok. I guess I'll have to go look through the patches Mandrake applied to the kernel. Thanks for the response. Steve Fox IBM Linux Technology Center http://www.ibm.com/linux/ltc http://k-lug.com

1 0

Re: Networking with Mandrake 8.0 and SuSE 7.1
by r.maurizzi＠gvs.it 18 May '01

18 May '01

> I'm running Lotus Notes 5.04a under WINE with one of the January releases > and its been working great for months. However, any release since February > does not seem to do reliable networking on my Mandrake 8.0 system. (...) > On a related note, another cow-orker who is using Mandrake Cooker > (development distro) and his own compiled kernel does not have this > problem. I am on a token ring network, which I don't think should matter. I had the exact problem with SuSE Linux: starting from Kernel SuSE-2.2.18, wine doesn't work anymore, but if I use a self compiled "stock" kernel, it works again. It's a pity I know little of both Wine and Kernel tcp/ip, and I don't have time to check kernels and diffs... :-( I don't know about Mandrake, but SuSE distributes a tar.gz with all the patches they apply to their kernels. If Mandrake does something similar, I guess a few "apply a part of the patches - compile - test" cycle should help spot the problem(s). If anyone on the list has ideas on tests I could do to gather more info, I'll be glad to help (I still have SuSE stock kernels installed, so I can test differences in behaviours between different versions rather quickly) Ciao, Rob!

2 1

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

wine-devel