December 2005 - Wine-Devel - list.winehq.org

Re: MSI patch getting some installer working thx deformat_environment did cut off string at beign of it.

by Uwe Bonnes

>>>>> "Magnus" == Magnus Olsen <magnus(a)itkonsult-olsen.com> writes: Magnus> in function deformat_environment did cut of one letter of Magnus> GetEnvironmentVariableW at beigner in second call. it try found Magnus> example LLUSERSPROFILE but it mean ALLUSERSPROFILE. fixed by me Magnus> and hpussin. in ReactOS Magnus Olsen greatlord(a)reactos.com Magnus> begin 666 format.c.patch Magnus> M26YD97(a)Z(&1L;',O;7-I+V9O<FUA="YC#0H]/3T]/3T]/3T]/3T]/3T]/3T] Magnus, please follow our patch convention: - unified diffs from below the wine directory - as plain text attachment (makes it easier to read for the list) If patches get not applied, resubmit. You'll get some reaction from Alexandre after some time. Bye -- Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt --------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

Dec. 20, 2005

Re: Steam Winsock regression

by Alexandre Julliard

James Liggett <jrliggett(a)cox.net> writes: > Yeah, that could be a feasible scenario, but the problem here is that > the router running my network failed, not the computers. The _router_ is > what needed to be reset, not the machines themselves. Like I said > before, I noticed some eerily similar behavior on windows, so it's > possible that it's not a wine bug. There may well be a problem in Wine that causes it to send wrong data, but if the router crashes that's obviously a bug in the router code, not in Wine. -- Alexandre Julliard julliard(a)winehq.org

Dec. 20, 2005

errors fixed with crossoffice

by Curro Amores

Hi i have a problem with wine and my access97 reports Letters are show separated and bigger. However using crossoffice i dont have this problem. How can i know where is the problem i have installed freetype, fontconfig last versions, copied all my ttf files... thanks

Dec. 20, 2005

Re: Alexandre Julliard : ntdll: More compatible exception information for protection faults.

by Alexandre Julliard

Paul Vriens <Paul.Vriens(a)xs4all.nl> writes: > this latest addition to the tests throws an exception on win98. > Screenshot attached. > > Probably Alexandre's area, as he wrote the test. Yes, but I don't have win98 here. Could you please try to find out which of the various exception tests causes that? -- Alexandre Julliard julliard(a)winehq.org

Dec. 20, 2005

Re: Steam Winsock regression

by James Liggett

I did some more testing and found that the fix to the problem was committed December 5: http://www.winehq.com/pipermail/wine-cvs/2005-December/019660.html Before that point, wine gives me these fixme's and my router dies: fixme:winsock:convert_af_w2u unhandled Windows address family 0 fixme:winsock:convert_proto_w2u unhandled Windows socket protocol 0 I hope they prove to be useful. James On Mon, 2005-12-19 at 08:39 +0100, Marcus Meissner wrote: > On Sun, Dec 18, 2005 at 11:26:01PM -0800, James Liggett wrote: > > On Mon, 2005-12-19 at 07:56 +0100, Marcus Meissner wrote: > > > On Sun, Dec 18, 2005 at 10:35:34PM -0800, James Liggett wrote: > > > > Hi, > > > > For a while now I've been noticing a very bad regression in winsock with > > > > steam. If I try to view information about a server using the "View game > > > > info command," my Internet connection fails completely. Not only that, > > > > but sometimes *all* of the machines on my network (both WinXP and Linux) > > > > lose their connections as well. I did the usual drudgery and found the > > > > offending patch: > > > > http://www.winehq.com/pipermail/wine-cvs/2005-November/019293.html > > > > > > Wow. Taking out *other* machines should not happen with a simple > > > userlevel patch. > > Yeah, tell me about it. :) Pretty crazy, huh? > > Yes. > > > > > Can someone take a look at this one ASAP? Thanks very much! > > > > > > Did you try WINE 0.9.3 and WINE CVS to see if it still happens there? > > > > > > Are there any error messages on the WINE console? > > This is still a problem as of Wine 0.9.3. I'm trying to compile current > > CVS right now, will get back to you on that. :) Thanks! > > Debugoutput would help. > > Also try to do a network sniff (tcpdump or tethereal) at the time of the > problem and see if there is anything strange. > > (Perhaps try a before - after version.) > > Ciao, Marcus

Dec. 20, 2005

Re: Steam Winsock regression

by James Liggett

On Mon, 2005-12-19 at 23:38 -0800, Scott Ritchie wrote: > Actually, I meant it might be a security issue for 2 reasons. The first > is that taking out other machines is obviously a problem. The second, > however, was that you reported it took out the internet of the machine > Wine was running on until a restart. A userlevel application shouldn't > be able to do that, which would mean there might be an internet-killing > exploit in the form of a winelib app. Yeah, that could be a feasible scenario, but the problem here is that the router running my network failed, not the computers. The _router_ is what needed to be reset, not the machines themselves. Like I said before, I noticed some eerily similar behavior on windows, so it's possible that it's not a wine bug. Now that I think about it a little more, I can remember getting some fixme's from winsock, something about unhandled protocol family 0 or something like that. Maybe it's more of a Steam bug then a Wine bug, but again I can't prove that for sure. In any case, such a bug, no matter whose it is, is a big problem. I'll put my tree back a couple of days and get the exact fixme's; maybe they'll help solve it? James

Dec. 20, 2005

Re: cvs.winehq.org slow?

by Ivan Leo Puoti

Aric Cyr wrote: > I was having similar problems over the weekend, except my "cvs up" would timeout > and fail. Pinging cvs.winehq.com resulted in a few packet drops then it would > starting pinging properly. After that cvs works fine. Network issues at > Codeweavers? Well, you can take load off the codeweavers server by using the mirror in Germany, instructions on how to do so are here http://www.winehq.org/site/cvs Ivan.

Dec. 20, 2005

[wined3d] Code questions

by Aric Cyr

I just have a couple wined3d code questions/comments that I'd like to clear up and possible submit patches if my understanding is correct. If I am misunderstanding the code please correct me. 1) basetexture.c:BindTexture() Just _before_ we call glBindTexture we call glTexParameteri and either set GL_TEXTURE_MAX_LEVEL or GL_TEXTURE_MIN_FILTER. The comment in the code is: "Always need to reset the number of mipmap levels when rebinding as it is a property of the active texture unit, and another texture may have set it to a different value" However, this doesn't make any sense to me and I believe it to be incorrect. glTexParameter affects the currently bound texture object. If the current and next texture to be bound have different mipmaps levels or parameters it should not be necessary to change these settings as all the state is encapsulated correctly in the OpenGL texture object, which should have been initialized correctly in Device::CreateTexture or BaseTexture::ApplyStateChanges. The reason the current code is bad is two-fold: 1) unneeded state changes on a performance critical path, as texture binding is very frequent and 2) changing the state of a _previous_ texture object, not the current one. The next time we go to use that texture object we will have to reapply all of it's state again, another waste of time. I spent a lot of the weekend debugging wined3d (trying to get anisotropic working in World of Warcraft) and just commenting out the block of code greatly cut down on d3d trace messages and makes the code easier to understand (and faster). 2) device.c::SetupTextureStates() Similar to my previous remark, there seems to be an incorrect state change here as well. glTexEnvf(TEXTURE_FILTER_CONTROL, LOAD_BIAS), glBlendColor() and then glTexEnvfv(TEXTURE_ENV, TEXTURE_ENV_COLOR) are called. All of these are called for each texture state update. However these are all global render states, not per-texture contrary to the comment in the code. Since these states are already being set in Device::SetRenderState() I'm pretty sure they don't need to be set here as well. Well that's all for now. Having started to really dig into wined3d, it seems that some effort in cleaning it up (coding style is way off) and stablizing it before it gets much more unwieldy is due. I'm also willing to help with such a task if there is a consensus that it should be done. Regards, Aric

Dec. 20, 2005

Re: Steam Winsock regression

by Scott Ritchie

On Mon, 2005-12-19 at 14:59 -0800, James Liggett wrote: > On Mon, 2005-12-19 at 10:00 +0100, Andreas Mohr wrote: > > I don't think Scott meant it that way ;-) > > Who cares about Wine, here's a change that makes half the machines in the > > network lose connection randomly, that's much bigger things to worry about > > potentially than simply non-working Wine socket functionality... > Of course that's the case here, maybe I didn't say it like I should have > and it came off not quite right. Sorry, it was late at night and I was > tired ;-) In fact, I was thinking the same thing. Hell, I almost > uninstalled wine because of it thinking I should stay away from > something that unstable and potentially dangerous. But maybe it's not > all Wine's fault. Last week my brother and I got together with a friend > of his to play a couple late-night matches of CS Source. I was playing > on Windows XP since Source games don't work so well on Wine yet. For > some reason I can't explain, my network's router kept failing > spontaneously while we were playing, in a similar manner in wine with > the patch. It's possible that Marcus's winsock patch might have exposed > a bug in Source/Steam network code, but I can't prove that yet. > > > > Thus one should probably attempt to investigate a bit more. > Absolutely. What can I do to help out here? > > James Actually, I meant it might be a security issue for 2 reasons. The first is that taking out other machines is obviously a problem. The second, however, was that you reported it took out the internet of the machine Wine was running on until a restart. A userlevel application shouldn't be able to do that, which would mean there might be an internet-killing exploit in the form of a winelib app. Thanks, Scott Ritchie

Dec. 20, 2005

Re: cvs.winehq.org slow?

by Aric Cyr

Dan Kegel <dank <at> kegel.com> writes: > > I'm having trouble doing a cvs update from cvs.winehq.org > lately. The connection is made, but nothing happens, > even after a minute or so. > The European site seems healthier. > What's up? > > -- > Wine for Windows ISVs: http://kegel.com/wine/isv > > I was having similar problems over the weekend, except my "cvs up" would timeout and fail. Pinging cvs.winehq.com resulted in a few packet drops then it would starting pinging properly. After that cvs works fine. Network issues at Codeweavers?

Dec. 20, 2005