December 2006 - Wine-devel - list.winehq.org

Re: Bug 50
by Pedro Araújo Chaves Jr. 19 Dec '06

19 Dec '06

On 12/19/06, Alexandre Julliard <julliard(a)winehq.org> wrote: > Passing the DC is not an option, that's a private gdi32 structure that > the driver doesn't know about. You shouldn't need to change the > function signature in any case, it already takes an lpDx array, all > you have to do is put better spacing values in there. Well, I had dropped that idea altogether as soon as I started working early in the morning, before reading your e-mail... besides, my next move was to play around with the spacing array, of course. Anyway, for a number of reasons, the idea of changing the function signature didn't sound well from the start to me neither. Yet I wanted to give it a shot, since it once fixed the issue. I'll keep you informed. Regards, Pedro.

1 0

Bug 50
by Pedro Araújo Chaves Jr. 19 Dec '06

19 Dec '06

First, sorry for the blank mail I accidentally sent earlier... my bad. Ok, so I have been assigned bug 50 [1] here at my job, and I'm trying to apply the patch [2] Keith Dunwoody proposed in June 2005, which fixed the issue but was rejected for technical reasons. Right now I can't think of another way around that bug except a patch that big, but I think we could break it into two smaller ones: the first one would change what must be changed for the second one ― the actual correction ― to work. Well, that first one involves changing the signatures of the ExtTextOut functions in winex11.drv, in a similar fashion to what Keith had proposed; but, instead of adding 'INT *breakTotalExtra, INT *breakCount' to their argument lists, I suggest that we pass them the whole DC instead of just its physDev ― that way we'd have access to the relevant attributes while still being able to get its physDev (assigning it to a variable right at the beginning of the functions, in order to keep their current functionality in this first step). I'm only afraid that may bring a little overhead, so I'd really like to see some opinions, or some better ideas... [1] http://bugs.winehq.org/show_bug.cgi?id=50 [2] http://bugs.winehq.org/attachment.cgi?id=912&action=view I'll be testing that here, anyway. Cheers, Pedro. -- "People think computers will keep them from making mistakes. They're wrong. With computers you make mistakes faster." (Adam Osborne)

2 1

Running winetest on Vista
by Paul Vriens 19 Dec '06

19 Dec '06

Hi, I've started to run our winetest executable on Vista. The current output is 7.8 MB or so, so we will never be able to sent it out just yet (1.5MB limit). I will start reducing the number of test failures on Vista (if I can of course). One thing that struck me and most likely it's due to Vista security is that when I run winetest.exe it tells me that it couldn't retrieve the list for the setupapi tests. While investigating that (several hours up to now) I found (not 100% sure though) that if an executable (like our setupapi_test.exe) has the word setup in it and is not properly signed, Vista starts complaining !!! Just copying our setupapi_test.exe to paul.exe makes it possible to run 'paul.exe --list'. Any ideas about the latter? Cheers, Paul.

4 5

Re: [1] WineD3D: Add the state dirtification infrastructure
by Stefan Dösinger 18 Dec '06

18 Dec '06

Am Montag 18 Dezember 2006 19:25 schrieb Stefan Dösinger: Another issue I stumbled uppon: This patch moves the + 1 for the array declaration to the right place. > Am Montag 18 Dezember 2006 14:06 schrieb Stefan Dösinger: > Another update, it fixes an spotted by Henri issue where the state was used > instead of the representative and uses shifting and bitmasks to calculate > the index and shift for the bitmap. > > > This is another updated version, it uses bitmaps for the isStateDirty > > member. > > > > Am Sonntag 17 Dezember 2006 14:45 schrieb Stefan Dösinger: > > > Am Freitag 15 Dezember 2006 18:31 schrieb Stefan Dösinger: > > > > This patch adds the dirty list infrastructure. A dirty marker > > > > consists of a DWORD containing the state it marks dirty, and the list > > > > structure to build the list. > > > > > > > > The wined3ddevice has 3 new members: > > > > dirtyStateList: The list of dirty states. This will be moved to a > > > > per-context stucture once we add multithreading support > > > > freeListElems: A list containing unused dirty markers to avoid > > > > unnecessary HeapAlloc/HeapFree calls. I have seen D3D games spending > > > > 15% cpu time in the Heap code, we don't want to increase that. > > > > dirtyStates: An array that tells if a state is dirty. It is used to > > > > avoid double dirtification. I do not use the > > > > stateblock->changed.XXXX[] array because this is used for different > > > > purposes in the stateblock code. Ivan, do we still need it? > > > > > > Attached is an updated patch that uses an array for dirtification > > > instead of a list. Actually there are 2 arrays, one to quickly find out > > > if a specific state is dirty, and one to quickly find all dirty states. > > > > > > Checking if a state is dirty is done very often, so we need the > > > isStateDirty. A linear search on the dirtyArray would take way too much > > > time. On the other hand, my experiance so far shows that usually not > > > more than 10-20 states are dirty at the beginning of a drawprim call, > > > so iterating over all states would be a waste of time too. That's why > > > the dirtyArray is there. > > > > > > In the end there will be around 1000 states, so the size of those > > > arrays will be 8-10 kilobytes. Even oldest games load textures in the > > > area of megabytes, with recent games it's about gigabytes, so I don't > > > think those 10 kb's are an issue :-) . I could easilly change the > > > DWORDs to shorts, but all d3d states are defined as DWORDs, so I > > > decided to use DWORDS. > > > > > > If I used a bitmap I could squeeze the same information in about 128 > > > bytes + extra size for the code dealing with the bitmap + extra runtime > > > needs for bit shifting and checking all DWORDs in the bitmap. D3D games > > > are usually cpu limited, so I prefer the more memory intensive version. > > > > > > The other patches should apply cleanly over this patch, so I do not > > > resend them.

1 0

re: mscoree: Add a stub implementation for CorExitProcess and GetRequestedRuntimeInfo.
by Dan Kegel 18 Dec '06

18 Dec '06

Hans wrote: > These stubs make the installer happy that is common to many of the .NET apps I tried. Which apps? Now that your patches seem to have been committed, I've started trying to find and install apps that require .net, but for some reason I'm having trouble finding many. The first one I tried is Life Balance (http://appdb.winehq.org/appview.php?iAppId=2928) but even if Mono-1.2 is installed, Life Balance seems to still try to download and install the MS c# runtime. +relay shows this before the download starts: 0009:Call advapi32.RegOpenKeyExA(80000002,0042f148 "SOFTWARE\\Microsoft\\NET Framework Setup\\NDP",00000000,00020019,0033bb90) ret=0040fa26 0009:Ret advapi32.RegOpenKeyExA() retval=00000002 ret=0040fa26 And bingo, http://msdn2.microsoft.com/en-us/library/aa480243.aspx documents this as part of the check to see if .net is installed. Has somebody figured out a canned .reg file to use to fake out this install check? - Dan

3 4

Re: ws2_32: Implement the SIO_ADDRESS_LIST_QUERY ioctl.
by Dmitry Timoshkov 18 Dec '06

18 Dec '06

"Hans Leidekker" <hans(a)it.vu.nl> wrote: > case WS_SIO_ADDRESS_LIST_QUERY: > - FIXME("-> SIO_ADDRESS_LIST_QUERY request: stub\n"); > - break; > + { > + int fd; > + DWORD size; > + > + TRACE("-> SIO_ADDRESS_LIST_QUERY request\n"); > + > + if (!lpcbBytesReturned || (fd = get_sock_fd(s, 0, NULL)) == -1) It appears that fd is not used later at all. -- Dmitry.

1 0

Bug 50
by Pedro Araújo Chaves Jr. 18 Dec '06

18 Dec '06

-- "People think computers will keep them from making mistakes. They're wrong. With computers you make mistakes faster." (Adam Osborne)

1 0

Re: DDraw: Implement depth filling through blit. (re-resubmit) try3
by Stefan Dösinger 18 Dec '06

18 Dec '06

Am Sonntag 17 Dezember 2006 23:24 schrieb greg87(a)online.de: > Original Author: > Elie Morisse > lachienne(a)wanadoo.fr > > Updated by: > Gregor Münch > greg87(a)online.de > > This is modified version of this patch: > http://www.winehq.org/pipermail/wine-patches/2006-September/030815.html > > With the help from Stefan Dösinger I altered 2 lines. Full credit > belongs to Elie Morisse > > fixes: http://bugs.winehq.org/show_bug.cgi?id=4306 A small issue: The indentions in IWineD3DSurfaceImpl_BltZ look a bit strange. --- a/dlls/ddraw/ddraw.c +++ b/dlls/ddraw/ddraw.c @@ -1787,7 +1787,7 @@ IDirectDrawImpl_CreateNewSurface(IDirect { Usage |= WINED3DUSAGE_OVERLAY; } - if(This->depthstencil) + if(This->depthstencil || (pDDSD->ddsCaps.dwCaps & DDSCAPS_ZBUFFER)) { Perhaps you can get rid of This->depthstencil by setting the DDSCAPS_ZBUFFER flag instead on the requested surface desc in the auto depth stencil creation callback. Mind you, this auto depth stencil creation isn't used nowadays any more, it was needed earlier to prevent wined3d from crashing. You can remove the function body and replace it by an ERR entirely I think, it shouldn't be called if everything works correctly in wined3d.

1 0

A wine success story
by Dan Kegel 18 Dec '06

18 Dec '06

I admit, I cruise new blog posts just to see what people are writing about Wine. Today I saw one ( http://www.ubuntux.org/shockwave-player-ubuntu-linux ) that says "I have discovered a way to get Shockwave working under Linux. I am using Ubuntu 6.06 (Dapper Drake), Wine 0.99.22, Firefox 1.5.0.7, and Shockwave 10.1.4.020. 1. Install Wine. ... Wine will automatically associate .exe files with itself. More information on Wine is available at http://www.winehq.com . 2. Download the Windows version of Firefox and install. Wine will ensure the installer functions correctly. 3. Install the Windows version of Shockwave for Mozilla. I downloaded and installed it manually (slim install). After these steps, I had a working Shockwave player from within Firefox. " Now that's the kind of user story I like to see! - Dan

7 8

Re: WineD3D: paletted texture emulation using fragment shaders
by Frank Richter 17 Dec '06

17 Dec '06

On 16.12.2006 22:24, Roderick Colenbrander wrote: > + "MUL index.x, index.x, constants.x;\n" /* Scale the index by 255/256 */ > + "ADD index.x, index.x, constants.y;\n" /* Add a bias of '0.5' in order to sample in the middle */ FWIW, this can be conflated to "MAD index.x, index.x, constants.x, constants.y;". -f.r.

1 0