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

Re: usp10: fix tests to allow atomic patches for new implementation
by James Hawkins Dec. 15, 2006

Dec. 15, 2006

On 12/15/06, Clinton Stimpson <cjstimpson(a)utwire.net> wrote: > Hi, > > Here's a patch to make unimplemented functions return E_NOTIMPL instead > of S_OK. Even dependent functions already return E_NOTIMPL. Also > fixing tests to check that the return value is S_OK before testing the > 'out' parameters. > This fix allows piecemeal patches to follow with implementation for each > function in a separate patch. > > Thanks, > Clinton Stimpson > > ChangeLog > Return E_NOTIMPL where appropriate. > And only check out parameters when appropriate. > There are a couple things wrong with this patch. First, you've made it so the tests won't be run for these functions (because they don't return S_OK as it stands). Second, there are probably applications that depend on these functions returning S_OK. You're really making it a lot harder than it should be. Implement each function one at a time. If tests start to succeed, just remove the todo_wine from them. So the implemented function and removed/added go in one patch. You just do that for each function you're implementing. -- James Hawkins

2 1

Re: ddraw: Fix warnings
by Stefan Dösinger Dec. 15, 2006

Dec. 15, 2006

Am Freitag 15 Dezember 2006 08:02 schrieb Dmitry Timoshkov: > Hello, > > Changelog: > ddraw: Fix warnings. > > --- > dlls/ddraw/ddraw_thunks.c | 4 ++-- > 1 files changed, 2 insertions(+), 2 deletions(-) > > diff --git a/dlls/ddraw/ddraw_thunks.c b/dlls/ddraw/ddraw_thunks.c > index ef8af6b..99518f0 100644 > --- a/dlls/ddraw/ddraw_thunks.c > +++ b/dlls/ddraw/ddraw_thunks.c > @@ -112,7 +112,7 @@ IDirectDraw3Impl_AddRef(LPDIRECTDRAW3 iface) > ICOM_THIS_FROM(IDirectDrawImpl, IDirectDraw3, iface); > ULONG ref = InterlockedIncrement(&This->ref4); > > - TRACE("(%p) : incrementing IDirectDraw4 refcount from %lu.\n", This, > ref -1); + TRACE("(%p) : incrementing IDirectDraw4 refcount from %u.\n", > This, ref -1); Sorry I forgot to check your original patch for this, but did you check if the IDirectDraw3 refcount is bound to the IDirectDraw4 refcount? I'd suspect that the IDirectDraw3 interface has its own reference count too. See dlls/ddraw/tests/refcount.c

2 2

Re: Removed as super maintainer
by Alexander Nicolaysen Sørnes Dec. 15, 2006

Dec. 15, 2006

Onsdag 13 desember 2006 19:14, skrev Chris Morgan: > If you didn't receive an email about how you haven't logged in in a > while then it isn't likely that the appdb cleanup system removed your > super maintainership. Could this possibly be related to the super > maintainer changes you put in recently? Like if you process a > submitted application does it remove your super maintainer status for > some other apps? > > We unfortunately don't log events on the appdb. If we did we could log > all kinds of stuff and go back and figure out when your entry was > removed. > > Chris I couldn't find anything suspicious in my changes, and everything was OK when I was logged in yesterday. I applied as super maintainer for an app and approved one in the queue, but that did not remove me. It seems like all the removals (not just mine) happened between today and yesterday. Regards, Alexander N. Sørnes > > On 12/13/06, Alexander Nicolaysen Sørnes <alex(a)thehandofagony.com> wrote: > > Hello, > > > > It seems like I have been removed from all the applications I was > > previously a super maintainer of. All my non-super maintainerships seem > > to be unaffected. Does anyone know the reason? Is it related to the > > recent server problems? > > > > > > Regards, > > > > Alexander N. Sørnes > > AppDB Administrator

5 5

Re: usp10: implement ScriptStringAnalyse, ScriptStringFree, ScriptStringXtoCP, ScriptStringCPtoX
by Clinton Stimpson Dec. 14, 2006

Dec. 14, 2006

Hi, Can I get some feedback on this patch? It appears to have been rejected. Thanks, Clinton Clinton Stimpson wrote: > > This patch implements ScriptStringAnalyse, ScriptStringFree, > ScriptStringXtoCP, ScriptStringCPtoX. > > Also, many todo_wine's are removed. > > > Thanks, > Clinton Stimpson > > ChangeLog > Implement ScriptStringAnalyse, ScriptStringFree, > ScriptStringXtoCP, ScriptStringCPtoX > > > ------------------------------------------------------------------------ > > Index: dlls/usp10/usp10.c > =================================================================== > RCS file: /home/wine/wine/dlls/usp10/usp10.c,v > retrieving revision 1.45 > diff -u -r1.45 usp10.c > --- dlls/usp10/usp10.c 12 Dec 2006 20:30:48 -0000 1.45 > +++ dlls/usp10/usp10.c 13 Dec 2006 03:28:30 -0000 > @@ -73,6 +73,46 @@ > HDC hdc; > } Scriptcache; > > +typedef struct { > + int numGlyphs; > + WORD* glyphs; > + WORD* pwLogClust; > + int* piAdvance; > + SCRIPT_VISATTR* psva; > + GOFFSET* pGoffset; > + ABC* abc; > +} StringGlyphs; > + > +typedef struct { > + BOOL invalid; > + HDC hdc; > + int cItems; > + int cMaxGlyphs; > + SCRIPT_ITEM* pItem; > + int numItems; > + StringGlyphs* glyphs; > + SIZE* sz; > +} StringAnalysis; > + > +static void ME_StringAnalysisFree(StringAnalysis* analysis) > +{ > + int i; > + for(i=0; i<analysis->numItems; i++) > + { > + HeapFree(GetProcessHeap(), 0, analysis->glyphs[i].glyphs); > + HeapFree(GetProcessHeap(), 0, analysis->glyphs[i].pwLogClust); > + HeapFree(GetProcessHeap(), 0, analysis->glyphs[i].piAdvance); > + HeapFree(GetProcessHeap(), 0, analysis->glyphs[i].psva); > + HeapFree(GetProcessHeap(), 0, analysis->glyphs[i].pGoffset); > + HeapFree(GetProcessHeap(), 0, analysis->glyphs[i].abc); > + } > + > + HeapFree(GetProcessHeap(), 0, analysis->glyphs); > + HeapFree(GetProcessHeap(), 0, analysis->pItem); > + > + HeapFree(GetProcessHeap(), 0, analysis); > +} > + > /*********************************************************************** > * DllMain > * > @@ -448,9 +488,16 @@ > const BYTE *pbInClass, > SCRIPT_STRING_ANALYSIS *pssa) > { > - FIXME("(%p,%p,%d,%d,%d,0x%x,%d,%p,%p,%p,%p,%p,%p): stub\n", > - hdc, pString, cString, cGlyphs, iCharset, dwFlags, > - iReqWidth, psControl, psState, piDx, pTabdef, pbInClass, pssa); > + HRESULT hr; > + StringAnalysis* analysis; > + int numItemizedItems; > + int i; > + SCRIPT_CACHE* sc = 0; > + > + TRACE("(%p,%p,%d,%d,%d,0x%x,%d,%p,%p,%p,%p,%p,%p)\n", > + hdc, pString, cString, cGlyphs, iCharset, dwFlags, > + iReqWidth, psControl, psState, piDx, pTabdef, pbInClass, pssa); > + > if (1 > cString || NULL == pString) { > return E_INVALIDARG; > } > @@ -458,7 +505,64 @@ > return E_PENDING; > } > > - return E_NOTIMPL; > + analysis = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, > + sizeof(StringAnalysis)); > + > + analysis->hdc = hdc; > + numItemizedItems = 255; > + analysis->pItem = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, > + numItemizedItems*sizeof(SCRIPT_ITEM)+1); > + > + hr = ScriptItemize(pString, cString, numItemizedItems, psControl, > + psState, analysis->pItem, &analysis->numItems); > + > + while(hr == E_OUTOFMEMORY) > + { > + numItemizedItems *= 2; > + HeapReAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, analysis->pItem, > + numItemizedItems*sizeof(SCRIPT_ITEM)+1); > + hr = ScriptItemize(pString, cString, numItemizedItems, psControl, > + psState, analysis->pItem, &analysis->numItems); > + } > + > + analysis->glyphs = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, > + sizeof(StringGlyphs)*analysis->numItems); > + sc = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(SCRIPT_CACHE)); > + > + for(i=0; i<analysis->numItems; i++) > + { > + int cChar = analysis->pItem[i+1].iCharPos - analysis->pItem[i].iCharPos; > + int numGlyphs = 1.5 * cChar + 16; > + WORD* glyphs = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(WORD)*numGlyphs); > + WORD* pwLogClust = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(WORD)*cChar); > + int* piAdvance = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(int)*numGlyphs); > + SCRIPT_VISATTR* psva = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(SCRIPT_VISATTR)*cChar); > + GOFFSET* pGoffset = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(GOFFSET)*numGlyphs); > + ABC* abc = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(ABC)); > + int numGlyphsReturned; > + > + /* FIXME: non unicode strings */ > + WCHAR* pStr = (WCHAR*)pString; > + hr = ScriptShape(hdc, sc, &pStr[analysis->pItem[i].iCharPos], > + cChar, numGlyphs, &analysis->pItem[i].a, > + glyphs, pwLogClust, psva, &numGlyphsReturned); > + hr = ScriptPlace(hdc, sc, glyphs, numGlyphsReturned, psva, &analysis->pItem[i].a, > + piAdvance, pGoffset, abc); > + > + analysis->glyphs[i].numGlyphs = numGlyphsReturned; > + analysis->glyphs[i].glyphs = glyphs; > + analysis->glyphs[i].pwLogClust = pwLogClust; > + analysis->glyphs[i].piAdvance = piAdvance; > + analysis->glyphs[i].psva = psva; > + analysis->glyphs[i].pGoffset = pGoffset; > + analysis->glyphs[i].abc = abc; > + } > + > + HeapFree(GetProcessHeap(), 0, sc); > + > + *pssa = analysis; > + > + return S_OK; > } > > /*********************************************************************** > @@ -489,9 +593,46 @@ > */ > HRESULT WINAPI ScriptStringCPtoX(SCRIPT_STRING_ANALYSIS ssa, int icp, BOOL fTrailing, int* pX) > { > - FIXME("(%p), %d, %d, (%p): stub\n", ssa, icp, fTrailing, pX); > - *pX = 0; /* Set a reasonable value */ > - return S_OK; > + int i, j; > + int runningX = 0; > + int runningCp = 0; > + StringAnalysis* analysis = ssa; > + TRACE("(%p), %d, %d, (%p)\n", ssa, icp, fTrailing, pX); > + > + if(!ssa || !pX) > + { > + return 1; > + } > + > + /* icp out of range */ > + if(icp < 0) > + { > + analysis->invalid = TRUE; > + return E_INVALIDARG; > + } > + > + for(i=0; i<analysis->numItems; i++) > + { > + for(j=0; j<analysis->glyphs[i].numGlyphs; j++) > + { > + if(runningCp == icp && fTrailing == FALSE) > + { > + *pX = runningX; > + return S_OK; > + } > + runningX += analysis->glyphs[i].piAdvance[j]; > + if(runningCp == icp && fTrailing == TRUE) > + { > + *pX = runningX; > + return S_OK; > + } > + runningCp++; > + } > + } > + > + /* icp out of range */ > + analysis->invalid = TRUE; > + return E_INVALIDARG; > } > > /*********************************************************************** > @@ -500,19 +641,79 @@ > */ > HRESULT WINAPI ScriptStringXtoCP(SCRIPT_STRING_ANALYSIS ssa, int iX, int* piCh, int* piTrailing) > { > - FIXME("(%p), %d, (%p), (%p): stub\n", ssa, iX, piCh, piTrailing); > - *piCh = 0; /* Set a reasonable value */ > - *piTrailing = 0; > + StringAnalysis* analysis = ssa; > + int i; > + int j; > + int runningX = 0; > + int runningCp = 0; > + int width; > + > + TRACE("(%p), %d, (%p), (%p)\n", ssa, iX, piCh, piTrailing); > + > + if(!ssa || !piCh || !piTrailing) > + { > + return 1; > + } > + > + /* out of range */ > + if(iX < 0) > + { > + *piCh = -1; > + *piTrailing = TRUE; > return S_OK; > + } > + > + for(i=0; i<analysis->numItems; i++) > + { > + for(j=0; j<analysis->glyphs[i].numGlyphs; j++) > + { > + width = analysis->glyphs[i].piAdvance[j]; > + if(iX < (runningX + width)) > + { > + *piCh = runningCp; > + if((iX - runningX) > width/2) > + *piTrailing = TRUE; > + else > + *piTrailing = FALSE; > + return S_OK; > + } > + runningX += width; > + runningCp++; > + } > + } > + > + /* out of range */ > + *piCh = analysis->pItem[analysis->numItems].iCharPos; > + *piTrailing = FALSE; > + > + return S_OK; > } > > /*********************************************************************** > * ScriptStringFree (USP10.@) > * > */ > -HRESULT WINAPI ScriptStringFree(SCRIPT_STRING_ANALYSIS *pssa) { > - FIXME("(%p): stub\n",pssa); > - return S_OK; > +HRESULT WINAPI ScriptStringFree(SCRIPT_STRING_ANALYSIS *pssa) > +{ > + StringAnalysis* analysis; > + BOOL invalid; > + TRACE("(%p)\n",pssa); > + > + if(!pssa) > + return E_INVALIDARG; > + > + analysis = *pssa; > + if(!analysis) > + return E_INVALIDARG; > + > + invalid = analysis->invalid; > + > + ME_StringAnalysisFree(analysis); > + > + if(invalid) > + return E_INVALIDARG; > + > + return S_OK; > } > > /*********************************************************************** > Index: dlls/usp10/tests/usp10.c > =================================================================== > RCS file: /home/wine/wine/dlls/usp10/tests/usp10.c,v > retrieving revision 1.30 > diff -u -r1.30 usp10.c > --- dlls/usp10/tests/usp10.c 12 Dec 2006 20:30:48 -0000 1.30 > +++ dlls/usp10/tests/usp10.c 13 Dec 2006 03:28:30 -0000 > @@ -701,21 +701,21 @@ > hr = ScriptStringAnalyse( hdc, teststr, String, Glyphs, Charset, Flags, > ReqWidth, &Control, &State, Dx, &Tabdef, > &InClass, &ssa); > - todo_wine ok(hr == S_OK, "ScriptStringAnalyse should return S_OK not %08x\n", hr); > + ok(hr == S_OK, "ScriptStringAnalyse should return S_OK not %08x\n", hr); > > /* test makes sure that a call with a valid pssa still works */ > hr = ScriptStringAnalyse( hdc, teststr, String, Glyphs, Charset, Flags, > ReqWidth, &Control, &State, Dx, &Tabdef, > &InClass, &ssa); > - todo_wine ok(hr == S_OK, "ScriptStringAnalyse should return S_OK not %08x\n", hr); > - todo_wine ok(ssa != NULL, "ScriptStringAnalyse pssa should not be NULL\n"); > + ok(hr == S_OK, "ScriptStringAnalyse should return S_OK not %08x\n", hr); > + ok(ssa != NULL, "ScriptStringAnalyse pssa should not be NULL\n"); > > if (hr == 0) > { > hr = ScriptStringOut(ssa, X, Y, Options, &rc, MinSel, MaxSel, Disabled); > todo_wine ok(hr == S_OK, "ScriptStringOut should return S_OK not %08x\n", hr); > hr = ScriptStringFree(&ssa); > - todo_wine ok(hr == S_OK, "ScriptStringFree should return S_OK not %08x\n", hr); > + ok(hr == S_OK, "ScriptStringFree should return S_OK not %08x\n", hr); > } > } > > @@ -774,8 +774,8 @@ > hr = ScriptStringAnalyse( hdc, String, String_len, Glyphs, Charset, Flags, > ReqWidth, &Control, &State, NULL, &Tabdef, > &InClass, &ssa); > - todo_wine ok(hr == S_OK, "ScriptStringAnalyse should return S_OK not %08x\n", hr); > - todo_wine ok(ssa != NULL, "ScriptStringAnalyse ssa should not be NULL\n"); > + ok(hr == S_OK, "ScriptStringAnalyse should return S_OK not %08x\n", hr); > + ok(ssa != NULL, "ScriptStringAnalyse ssa should not be NULL\n"); > if (hr == 0) > { > /* > @@ -792,25 +792,25 @@ > */ > fTrailing = FALSE; > hr = ScriptStringCPtoX(ssa, Cp, fTrailing, &X); > - todo_wine ok(hr == S_OK, "ScriptStringCPtoX should return S_OK not %08x\n", hr); > + ok(hr == S_OK, "ScriptStringCPtoX should return S_OK not %08x\n", hr); > hr = ScriptStringXtoCP(ssa, X, &Ch, &iTrailing); > - todo_wine ok(hr == S_OK, "ScriptStringXtoCP should return S_OK not %08x\n", hr); > - todo_wine ok(Cp == Ch, "ScriptStringXtoCP should return Ch = %d not %d for X = %d\n", Cp, Ch, X); > - todo_wine ok(iTrailing == FALSE, "ScriptStringXtoCP should return iTrailing = 0 not %d for X = %d\n", > + ok(hr == S_OK, "ScriptStringXtoCP should return S_OK not %08x\n", hr); > + ok(Cp == Ch, "ScriptStringXtoCP should return Ch = %d not %d for X = %d\n", Cp, Ch, X); > + ok(iTrailing == FALSE, "ScriptStringXtoCP should return iTrailing = 0 not %d for X = %d\n", > iTrailing, X); > fTrailing = TRUE; > hr = ScriptStringCPtoX(ssa, Cp, fTrailing, &X); > - todo_wine ok(hr == S_OK, "ScriptStringCPtoX should return S_OK not %08x\n", hr); > + ok(hr == S_OK, "ScriptStringCPtoX should return S_OK not %08x\n", hr); > hr = ScriptStringXtoCP(ssa, X, &Ch, &iTrailing); > - todo_wine ok(hr == S_OK, "ScriptStringXtoCP should return S_OK not %08x\n", hr); > + ok(hr == S_OK, "ScriptStringXtoCP should return S_OK not %08x\n", hr); > > /* > * Check that character position returned by ScriptStringXtoCP in Ch matches the > * one input to ScriptStringCPtoX. This means that the Cp to X position and back > * again works > */ > - todo_wine ok(Cp + 1 == Ch, "ScriptStringXtoCP should return Ch = %d not %d for X = %d\n", Cp + 1, Ch, X); > - todo_wine ok(iTrailing == FALSE, "ScriptStringXtoCP should return iTrailing = 0 not %d for X = %d\n", > + ok(Cp + 1 == Ch, "ScriptStringXtoCP should return Ch = %d not %d for X = %d\n", Cp + 1, Ch, X); > + ok(iTrailing == FALSE, "ScriptStringXtoCP should return iTrailing = 0 not %d for X = %d\n", > iTrailing, X); > } > > @@ -821,12 +821,12 @@ > fTrailing = TRUE; > Cp = 3; > hr = ScriptStringCPtoX(ssa, Cp, fTrailing, &X); > - todo_wine ok(hr == S_OK, "ScriptStringCPtoX should return S_OK not %08x\n", hr); > + ok(hr == S_OK, "ScriptStringCPtoX should return S_OK not %08x\n", hr); > X--; /* put X just inside the trailing edge */ > hr = ScriptStringXtoCP(ssa, X, &Ch, &iTrailing); > - todo_wine ok(hr == S_OK, "ScriptStringXtoCP should return S_OK not %08x\n", hr); > - todo_wine ok(Cp == Ch, "ScriptStringXtoCP should return Ch = %d not %d for X = %d\n", Cp, Ch, X); > - todo_wine ok(iTrailing == TRUE, "ScriptStringXtoCP should return iTrailing = 1 not %d for X = %d\n", > + ok(hr == S_OK, "ScriptStringXtoCP should return S_OK not %08x\n", hr); > + ok(Cp == Ch, "ScriptStringXtoCP should return Ch = %d not %d for X = %d\n", Cp, Ch, X); > + ok(iTrailing == TRUE, "ScriptStringXtoCP should return iTrailing = 1 not %d for X = %d\n", > iTrailing, X); > > /* > @@ -837,12 +837,12 @@ > fTrailing = TRUE; > Cp = 3; > hr = ScriptStringCPtoX(ssa, Cp, fTrailing, &X); > - todo_wine ok(hr == S_OK, "ScriptStringCPtoX should return S_OK not %08x\n", hr); > + ok(hr == S_OK, "ScriptStringCPtoX should return S_OK not %08x\n", hr); > X++; /* put X just outside the trailing edge */ > hr = ScriptStringXtoCP(ssa, X, &Ch, &iTrailing); > - todo_wine ok(hr == S_OK, "ScriptStringXtoCP should return S_OK not %08x\n", hr); > - todo_wine ok(Cp + 1 == Ch, "ScriptStringXtoCP should return Ch = %d not %d for X = %d\n", Cp + 1, Ch, X); > - todo_wine ok(iTrailing == FALSE, "ScriptStringXtoCP should return iTrailing = 0 not %d for X = %d\n", > + ok(hr == S_OK, "ScriptStringXtoCP should return S_OK not %08x\n", hr); > + ok(Cp + 1 == Ch, "ScriptStringXtoCP should return Ch = %d not %d for X = %d\n", Cp + 1, Ch, X); > + ok(iTrailing == FALSE, "ScriptStringXtoCP should return iTrailing = 0 not %d for X = %d\n", > iTrailing, X); > > /* > @@ -853,19 +853,19 @@ > fTrailing = FALSE; > Cp = 3; > hr = ScriptStringCPtoX(ssa, Cp, fTrailing, &X); > - todo_wine ok(hr == S_OK, "ScriptStringCPtoX should return S_OK not %08x\n", hr); > + ok(hr == S_OK, "ScriptStringCPtoX should return S_OK not %08x\n", hr); > X--; /* put X just outside the leading edge */ > hr = ScriptStringXtoCP(ssa, X, &Ch, &iTrailing); > - todo_wine ok(hr == S_OK, "ScriptStringXtoCP should return S_OK not %08x\n", hr); > - todo_wine ok(Cp - 1 == Ch, "ScriptStringXtoCP should return Ch = %d not %d for X = %d\n", Cp - 1, Ch, X); > - todo_wine ok(iTrailing == TRUE, "ScriptStringXtoCP should return iTrailing = 1 not %d for X = %d\n", > + ok(hr == S_OK, "ScriptStringXtoCP should return S_OK not %08x\n", hr); > + ok(Cp - 1 == Ch, "ScriptStringXtoCP should return Ch = %d not %d for X = %d\n", Cp - 1, Ch, X); > + ok(iTrailing == TRUE, "ScriptStringXtoCP should return iTrailing = 1 not %d for X = %d\n", > iTrailing, X); > > /* > * Cleanup the the SSA for the next round of tests > */ > hr = ScriptStringFree(&ssa); > - todo_wine ok(hr == S_OK, "ScriptStringFree should return S_OK not %08x\n", hr); > + ok(hr == S_OK, "ScriptStringFree should return S_OK not %08x\n", hr); > > /* > * Test to see that exceeding the number of chars returns E_INVALIDARG. First > @@ -874,7 +874,7 @@ > hr = ScriptStringAnalyse( hdc, String, String_len, Glyphs, Charset, Flags, > ReqWidth, &Control, &State, NULL, &Tabdef, > &InClass, &ssa); > - todo_wine ok(hr == S_OK, "ScriptStringAnalyse should return S_OK not %08x\n", hr); > + ok(hr == S_OK, "ScriptStringAnalyse should return S_OK not %08x\n", hr); > > /* > * When ScriptStringCPtoX is called with a character position Cp that exceeds the > @@ -884,13 +884,13 @@ > fTrailing = FALSE; > Cp = String_len + 1; > hr = ScriptStringCPtoX(ssa, Cp, fTrailing, &X); > - todo_wine ok(hr == E_INVALIDARG, "ScriptStringCPtoX should return E_INVALIDARG not %08x\n", hr); > + ok(hr == E_INVALIDARG, "ScriptStringCPtoX should return E_INVALIDARG not %08x\n", hr); > > hr = ScriptStringFree(&ssa); > /* > * ScriptStringCPtoX should free ssa, hence ScriptStringFree should fail > */ > - todo_wine ok(hr == E_INVALIDARG, "ScriptStringFree should return E_INVALIDARG not %08x\n", hr); > + ok(hr == E_INVALIDARG, "ScriptStringFree should return E_INVALIDARG not %08x\n", hr); > } > } > > > > ------------------------------------------------------------------------ > >

3 2

IoCompletion port
by Christian Heimes Dec. 14, 2006

Dec. 14, 2006

Hello! I have a program that needs IoCompletion port to run. The program crashes at startup because WINE head has only stubs for the functions. I found a patch created by Robert Shearman which adds IoCompletion to WINE. The patch is very old (2004?) and doesn't apply against the current development version of wine. I merged the patch manually into the source tree but my knowdledge of WINE is limited (zero *g*). My patched source tree doesn't compile: make[2]: Betrete Verzeichnis '/home/heimes/dev/misc/wine-io/dlls/winecrt0' gcc -c -I. -I. -I../../include -I../../include -D__WINESRC__ -D_REENTRANT -fPIC -Wall -pipe -fno-strict-aliasing -Wdeclaration-after-statement -Wwrite-strings -Wpointer-arith -fno-stack-protector -o drv_entry.o drv_entry.c In file included from drv_entry.c:23: ../../include/winternl.h:2377: error: expected declaration specifiers or ‘...’ before ‘LPOVERLAPPED’ ../../include/winternl.h:2378: error: expected declaration specifiers or ‘...’ before ‘LPOVERLAPPED’ make[2]: *** [drv_entry.o] Fehler 1 I'm attaching my patch to this posting in the hope somebody with more wisdom stands up. The patch has two ports (iofiles and new_iocompletion) because git diff didn't include the new files although I did a git add. bug report: http://bugs.winehq.org/show_bug.cgi?id=6155 iocompletion: original patch by Robert Shearman iofiles + new_iocompletion: patches against GIT head Thank you very much! Christian diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/dlls/kernel/iocompletion.c wine/dlls/kernel/iocompletion.c --- wine.orig/dlls/kernel/iocompletion.c 1969-12-31 19:00:00.000000000 -0500 +++ wine/dlls/kernel/iocompletion.c 2005-05-04 14:08:38.000000000 -0400 @@ -0,0 +1,173 @@ +/* + * I/O Completion Ports + * + * Copyright (C) 2003 Robert Shearman + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include <stdarg.h> + +#include "windef.h" +#include "winbase.h" +#include "winnt.h" +#include "winreg.h" +#include "winternl.h" +#include "ntstatus.h" +#include "wine/debug.h" + +WINE_DEFAULT_DEBUG_CHANNEL(iocompletion); + +HANDLE WINAPI CreateIoCompletionPort( + HANDLE FileHandle, + HANDLE ExistingCompletionPort, + ULONG_PTR CompletionKey, + DWORD NumberOfConcurrentThreads) +{ + HANDLE CompletionPort; + NTSTATUS Status; + + TRACE("(%p, %p, %lx, %ld)\n", + FileHandle, + ExistingCompletionPort, + CompletionKey, + NumberOfConcurrentThreads); + + if (((FileHandle == INVALID_HANDLE_VALUE) && (ExistingCompletionPort != NULL)) || + (ExistingCompletionPort == INVALID_HANDLE_VALUE)) + { + SetLastError(ERROR_INVALID_PARAMETER); + return INVALID_HANDLE_VALUE; + } + if (ExistingCompletionPort == NULL) + { + Status = NtCreateIoCompletion( + &CompletionPort, + GENERIC_ALL, 0, + NumberOfConcurrentThreads); + + if (Status != STATUS_SUCCESS) + { + SetLastError(RtlNtStatusToDosError(Status)); + return INVALID_HANDLE_VALUE; + } + } + else + CompletionPort = ExistingCompletionPort; + + if (FileHandle != INVALID_HANDLE_VALUE) + { + IO_STATUS_BLOCK IoStatusBlock; + FILE_COMPLETION_INFORMATION CompletionInfo; + CompletionInfo.CompletionKey = CompletionKey; + CompletionInfo.CompletionPort = CompletionPort; + + Status = NtSetInformationFile( + FileHandle, + &IoStatusBlock, + (PVOID)&CompletionInfo, + sizeof(FILE_COMPLETION_INFORMATION), + FileCompletionInformation); + + if (Status != STATUS_SUCCESS) + { + SetLastError(RtlNtStatusToDosError(Status)); + return INVALID_HANDLE_VALUE; + } + } + return CompletionPort; +} + +BOOL WINAPI GetQueuedCompletionStatus( + HANDLE CompletionPort, + LPDWORD lpNumberOfBytesTransferred, + PULONG_PTR lpCompletionKey, + LPOVERLAPPED * lplpOverlapped, + DWORD dwMilliseconds) +{ + IO_STATUS_BLOCK CompletionBlock; + NTSTATUS Status; + + TRACE("(CompletionPort %p, %p, %p, %p, %ld)\n", + CompletionPort, + lpNumberOfBytesTransferred, + lpCompletionKey, + lplpOverlapped, + dwMilliseconds); + + if (dwMilliseconds == INFINITE) + Status = NtRemoveIoCompletion( + CompletionPort, + lpCompletionKey, + lplpOverlapped, + &CompletionBlock, + NULL); + else + { + LARGE_INTEGER WaitTime; + /* multiplying two LONGLONGs with at least one LONGLONG having its + * higher long part not zero makes the multiplying a bit harder, + * therefore we do an easy multiply and negate afterwards rather than + * making it a hard multiply by doing "* -10000" + */ + WaitTime.QuadPart = (LONGLONG)dwMilliseconds * (LONGLONG)10000; + WaitTime.QuadPart = -WaitTime.QuadPart; + Status = NtRemoveIoCompletion( + CompletionPort, + lpCompletionKey, + lplpOverlapped, + &CompletionBlock, + &WaitTime); + } + if (Status == STATUS_SUCCESS) + { + *lpNumberOfBytesTransferred = CompletionBlock.Information; + return TRUE; + } + else + { + SetLastError(RtlNtStatusToDosError(Status)); + return FALSE; + } +} + +BOOL WINAPI PostQueuedCompletionStatus( + HANDLE CompletionPort, + DWORD dwNumberOfBytesTransferred, + ULONG_PTR dwCompletionKey, + LPOVERLAPPED lpOverlapped) +{ + NTSTATUS Status; + + TRACE("(CompletionPort %p, %ld, %lx, %p)\n", + CompletionPort, + dwNumberOfBytesTransferred, + dwCompletionKey, + lpOverlapped); + + Status = NtSetIoCompletion( + CompletionPort, + dwCompletionKey, + lpOverlapped, 0, + dwNumberOfBytesTransferred); + + if (Status == STATUS_SUCCESS) + return TRUE; + else + { + SetLastError(RtlNtStatusToDosError(Status)); + return FALSE; + } +} diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/dlls/kernel/kernel32.spec wine/dlls/kernel/kernel32.spec --- wine.orig/dlls/kernel/kernel32.spec 2005-04-27 07:08:47.000000000 -0400 +++ wine/dlls/kernel/kernel32.spec 2005-05-04 13:58:00.000000000 -0400 @@ -657,7 +657,7 @@ @ stdcall PeekConsoleInputA(ptr ptr long ptr) @ stdcall PeekConsoleInputW(ptr ptr long ptr) @ stdcall PeekNamedPipe(long ptr long ptr ptr ptr) -@ stub PostQueuedCompletionStatus +@ stdcall PostQueuedCompletionStatus(ptr long long ptr) @ stdcall PrepareTape(ptr long long) @ stub PrivMoveFileIdentityW @ stdcall Process32First (ptr ptr) diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/dlls/kernel/Makefile.in wine/dlls/kernel/Makefile.in --- wine.orig/dlls/kernel/Makefile.in 2005-04-20 11:43:36.000000000 -0400 +++ wine/dlls/kernel/Makefile.in 2005-05-04 14:19:47.000000000 -0400 @@ -33,6 +33,7 @@ file.c \ file16.c \ format_msg.c \ + iocompletion.c \ global16.c \ heap.c \ instr.c \ diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/dlls/kernel/sync.c wine/dlls/kernel/sync.c --- wine.orig/dlls/kernel/sync.c 2005-04-24 13:36:34.000000000 -0400 +++ wine/dlls/kernel/sync.c 2005-05-04 19:33:16.000000000 -0400 @@ -1743,31 +1743,6 @@ /****************************************************************************** - * CreateIoCompletionPort (KERNEL32.@) - */ -HANDLE WINAPI CreateIoCompletionPort(HANDLE hFileHandle, HANDLE hExistingCompletionPort, - ULONG_PTR CompletionKey, DWORD dwNumberOfConcurrentThreads) -{ - FIXME("(%p, %p, %08lx, %08lx): stub.\n", - hFileHandle, hExistingCompletionPort, CompletionKey, dwNumberOfConcurrentThreads); - return NULL; -} - - -/****************************************************************************** - * GetQueuedCompletionStatus (KERNEL32.@) - */ -BOOL WINAPI GetQueuedCompletionStatus( HANDLE CompletionPort, LPDWORD lpNumberOfBytesTransferred, - PULONG_PTR pCompletionKey, LPOVERLAPPED *lpOverlapped, - DWORD dwMilliseconds ) -{ - FIXME("(%p,%p,%p,%p,%ld), stub!\n", - CompletionPort,lpNumberOfBytesTransferred,pCompletionKey,lpOverlapped,dwMilliseconds); - SetLastError(ERROR_CALL_NOT_IMPLEMENTED); - return FALSE; -} - -/****************************************************************************** * CreateJobObjectW (KERNEL32.@) */ HANDLE WINAPI CreateJobObjectW( LPSECURITY_ATTRIBUTES attr, LPCWSTR name ) diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/dlls/ntdll/iocompletion.c wine/dlls/ntdll/iocompletion.c --- wine.orig/dlls/ntdll/iocompletion.c 1969-12-31 19:00:00.000000000 -0500 +++ wine/dlls/ntdll/iocompletion.c 2005-05-04 20:05:27.000000000 -0400 @@ -0,0 +1,216 @@ +/* + * I/O Completion Ports + * + * Copyright (C) 2003 Robert Shearman + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include "config.h" +#include <stdarg.h> + +#define NONAMELESSUNION +#include "windef.h" +#include "winbase.h" +#include "winnt.h" +#include "winreg.h" +#include "winternl.h" + +#include "ntdll_misc.h" +#include "wine/server.h" +#include "wine/debug.h" + +WINE_DEFAULT_DEBUG_CHANNEL(ntdll); + +extern int NTDLL_wait_reply( void *cookie ); + +/************************************************************************** + * NtCreateIoCompletion (NTDLL.@) + * + * Params: + * CompletionPort [O]: the handle created + * DesiredAccess [I}: the access desired (e.g. GENERIC_ALL) + * Reserved [I}: unknown + * NumberOfConcurrentThreads [I]: the desired number of concurrent + * threads + * Returns: + * Status + * Notes: + * It is effectively a FIFO queue for data and + * a LIFO queue for threads to "minimize context switches". + * The aim is to keep a small number of threads constantly + * active. + * See: + * MSDN for CreateIoCompletionPort spec and + * the article "Inside I/O Completion Ports" + * (http://www.sysinternals.com/ntw2k/info/comport.shtml) + */ +NTSTATUS WINAPI NtCreateIoCompletion ( + OUT PHANDLE CompletionPort, + IN ACCESS_MASK DesiredAccess, + IN ULONG_PTR Reserved, + IN ULONG NumberOfConcurrentThreads + ) + { + NTSTATUS ret; + + TRACE("(%p, %lx, %lx, %ld)\n", + CompletionPort, + DesiredAccess, + Reserved, + NumberOfConcurrentThreads); + + if (Reserved != 0) + { + FIXME("Reserved != 0 not supported\n"); + return STATUS_INVALID_PARAMETER; + } + + if (DesiredAccess && GENERIC_ALL) + DesiredAccess |= GENERIC_READ | GENERIC_WRITE; + + SERVER_START_REQ( create_io_completion ) + { + req->access = DesiredAccess; + req->concurrent_threads = NumberOfConcurrentThreads; + ret = wine_server_call( req ); + *CompletionPort = reply->handle; + } + SERVER_END_REQ; + + TRACE("returning %lx\n", ret); + return ret; +} + +/************************************************************************** + * NtSetIoCompletion (NTDLL.@) + * + * Params: + * CompletionPort [I]: port to send data to + * CompletionKey [I}: user key to identify this set of data + * lpOverlapped [I}: OVERLAPPED structure to send to port + * NumberOfBytesTransferred [I}: unknown - seems to be set to zero always + * NumberOfBytesToTransfer [I]: Bytes to transfer in this packet of data + * Returns: + * Status + * See: + * MSDN for PostQueuedCompletionStatus spec and + * the article "Inside I/O Completion Ports" + * (http://www.sysinternals.com/ntw2k/info/comport.shtml) + */ +NTSTATUS WINAPI NtSetIoCompletion( + IN HANDLE CompletionPort, + IN ULONG_PTR CompletionKey, + IN LPOVERLAPPED lpOverlapped, + IN ULONG NumberOfBytesTransferred, /* normally set to 0 */ + IN ULONG NumberOfBytesToTransfer /* will become number of bytes transferred in the io operation */ + ) +{ + NTSTATUS ret; + + TRACE("(%p, %lx, %p, %ld, %ld)\n", + CompletionPort, + CompletionKey, + lpOverlapped, + NumberOfBytesTransferred, + NumberOfBytesToTransfer); + + if (NumberOfBytesTransferred != 0) + { + FIXME("NumberOfBytesTransferred != 0 not supported\n"); + return STATUS_INVALID_PARAMETER; + } + + SERVER_START_REQ( set_io_completion ) + { + req->handle = CompletionPort; + req->completion_key = (void *)CompletionKey; + req->overlapped = lpOverlapped; + req->bytes_transferred = NumberOfBytesToTransfer; + ret = wine_server_call( req ); + } + SERVER_END_REQ; + + TRACE("returning %lx\n", ret); + return ret; +} + +/************************************************************************** + * NtRemoveIoCompletion (NTDLL.@) + * + * See: MSDN for GetQueuedCompletionStatus spec and + * the article "Inside I/O Completion Ports" + * (http://www.sysinternals.com/ntw2k/info/comport.shtml) + */ +NTSTATUS WINAPI NtRemoveIoCompletion ( + IN HANDLE CompletionPort, + OUT PULONG_PTR CompletionKey, + OUT LPOVERLAPPED * lplpOverlapped, + OUT PIO_STATUS_BLOCK CompletionStatus, + IN PLARGE_INTEGER WaitTime + ) +{ + NTSTATUS ret; + int cookie; + + TRACE("(%p, %p, %p, %p, %p)\n", + CompletionPort, + CompletionKey, + lplpOverlapped, + CompletionStatus, + WaitTime); + + for (;;) + { + SERVER_START_REQ( remove_io_completion ) + { + req->handle = CompletionPort; + req->cookie = &cookie; + NTDLL_get_server_timeout( &req->timeout, WaitTime ); + ret = wine_server_call( req ); + if (ret == STATUS_SUCCESS) + { + *CompletionKey = (ULONG_PTR)reply->completion_key; + *lplpOverlapped = reply->overlapped; + CompletionStatus->u.Status = STATUS_SUCCESS; + CompletionStatus->Information = reply->bytes_transferred; + } + } + SERVER_END_REQ; + if (ret == STATUS_PENDING) + ret = NTDLL_wait_reply( &cookie ); + if (ret == STATUS_ABANDONED) + { + SERVER_START_REQ( remove_io_completion_assigned ) + { + req->handle = CompletionPort; + ret = wine_server_call( req ); + if (ret == STATUS_SUCCESS) + { + *CompletionKey = (ULONG_PTR)reply->completion_key; + *lplpOverlapped = reply->overlapped; + CompletionStatus->u.Status = STATUS_SUCCESS; + CompletionStatus->Information = reply->bytes_transferred; + } + } + SERVER_END_REQ; + } + break; + } + + TRACE("returning %lx\n", ret); + return ret; +} + diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/dlls/ntdll/Makefile.in wine/dlls/ntdll/Makefile.in --- wine.orig/dlls/ntdll/Makefile.in 2004-12-07 09:47:13.000000000 -0500 +++ wine/dlls/ntdll/Makefile.in 2005-05-04 14:25:17.000000000 -0400 @@ -18,6 +18,7 @@ file.c \ handletable.c \ heap.c \ + iocompletion.c \ large_int.c \ loader.c \ loadorder.c \ diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/dlls/ntdll/ntdll.spec wine/dlls/ntdll/ntdll.spec --- wine.orig/dlls/ntdll/ntdll.spec 2005-04-22 17:17:16.000000000 -0400 +++ wine/dlls/ntdll/ntdll.spec 2005-05-04 14:24:47.000000000 -0400 @@ -86,7 +86,7 @@ @ stdcall NtCreateEvent(long long long long long) @ stub NtCreateEventPair @ stdcall NtCreateFile(ptr long ptr ptr long long long ptr long long ptr) -@ stub NtCreateIoCompletion +@ stdcall NtCreateIoCompletion(ptr long long long) @ stdcall NtCreateKey(ptr long ptr long ptr long long) @ stdcall NtCreateMailslotFile(long long long long long long long long) @ stdcall NtCreateMutant(ptr long ptr long) @@ -205,7 +205,7 @@ @ stdcall NtReleaseMutant(long ptr) @ stub NtReleaseProcessMutant @ stdcall NtReleaseSemaphore(long long ptr) -@ stub NtRemoveIoCompletion +@ stdcall NtRemoveIoCompletion(ptr ptr ptr ptr ptr) @ stdcall NtReplaceKey(ptr long ptr) @ stub NtReplyPort @ stdcall NtReplyWaitReceivePort(ptr ptr ptr ptr) @@ -234,7 +234,7 @@ @ stdcall NtSetInformationThread(long long ptr long) @ stdcall NtSetInformationToken(long long ptr long) @ stdcall NtSetIntervalProfile(long long) -@ stub NtSetIoCompletion +@ stdcall NtSetIoCompletion(ptr long ptr long long) @ stub NtSetLdtEntries @ stub NtSetLowEventPair @ stub NtSetLowWaitHighEventPair diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/dlls/ntdll/sync.c wine/dlls/ntdll/sync.c --- wine.orig/dlls/ntdll/sync.c 2005-04-24 13:35:53.000000000 -0400 +++ wine/dlls/ntdll/sync.c 2005-05-04 14:27:33.000000000 -0400 @@ -611,7 +611,7 @@ * * Wait for a reply on the waiting pipe of the current thread. */ -static int wait_reply( void *cookie ) +int NTDLL_wait_reply( void *cookie ) { int signaled; struct wake_up_reply reply; @@ -624,7 +624,7 @@ if (!reply.cookie) break; /* thread got killed */ if (reply.cookie == cookie) return reply.signaled; /* we stole another reply, wait for the real one */ - signaled = wait_reply( cookie ); + signaled = NTDLL_wait_reply( cookie ); /* and now put the wrong one back in the pipe */ for (;;) { @@ -719,7 +719,7 @@ ret = wine_server_call( req ); } SERVER_END_REQ; - if (ret == STATUS_PENDING) ret = wait_reply( &cookie ); + if (ret == STATUS_PENDING) ret = NTDLL_wait_reply( &cookie ); if (ret != STATUS_USER_APC) break; call_apcs( (flags & SELECT_ALERTABLE) != 0 ); if (flags & SELECT_ALERTABLE) break; diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/include/winbase.h wine/include/winbase.h --- wine.orig/include/winbase.h 2005-04-25 12:23:32.000000000 -0400 +++ wine/include/winbase.h 2005-05-04 13:58:00.000000000 -0400 @@ -1602,6 +1602,7 @@ HANDLE WINAPI OpenWaitableTimerW(DWORD,BOOL,LPCWSTR); #define OpenWaitableTimer WINELIB_NAME_AW(OpenWaitableTimer) BOOL WINAPI PeekNamedPipe(HANDLE,PVOID,DWORD,PDWORD,PDWORD,PDWORD); +BOOL WINAPI PostQueuedCompletionStatus(HANDLE,DWORD,ULONG_PTR,LPOVERLAPPED); DWORD WINAPI PrepareTape(HANDLE,DWORD,BOOL); BOOL WINAPI PrivilegedServiceAuditAlarmA(LPCSTR,LPCSTR,HANDLE,PPRIVILEGE_SET,BOOL); BOOL WINAPI PrivilegedServiceAuditAlarmW(LPCWSTR,LPCWSTR,HANDLE,PPRIVILEGE_SET,BOOL); diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/include/winternl.h wine/include/winternl.h --- wine.orig/include/winternl.h 2005-04-27 04:14:18.000000000 -0400 +++ wine/include/winternl.h 2005-05-04 19:24:46.000000000 -0400 @@ -1467,7 +1467,6 @@ NTSTATUS WINAPI NtClose(HANDLE); NTSTATUS WINAPI NtCreateEvent(PHANDLE,ACCESS_MASK,const OBJECT_ATTRIBUTES *,BOOLEAN,BOOLEAN); NTSTATUS WINAPI NtCreateFile(PHANDLE,ACCESS_MASK,POBJECT_ATTRIBUTES,PIO_STATUS_BLOCK,PLARGE_INTEGER,ULONG,ULONG,ULONG,ULONG,PVOID,ULONG); -NTSTATUS WINAPI NtCreateIoCompletion(PHANDLE,ACCESS_MASK,POBJECT_ATTRIBUTES,ULONG); NTSTATUS WINAPI NtCreateKey(PHKEY,ACCESS_MASK,const OBJECT_ATTRIBUTES*,ULONG,const UNICODE_STRING*,ULONG,PULONG); NTSTATUS WINAPI NtCreateMailslotFile(PHANDLE,ACCESS_MASK,POBJECT_ATTRIBUTES,PIO_STATUS_BLOCK,ULONG,ULONG,ULONG,PLARGE_INTEGER); NTSTATUS WINAPI NtCreateMutant(HANDLE*,ACCESS_MASK,const OBJECT_ATTRIBUTES*,BOOLEAN); @@ -1543,7 +1542,6 @@ NTSTATUS WINAPI NtReadVirtualMemory(HANDLE,const void*,void*,SIZE_T,SIZE_T*); NTSTATUS WINAPI NtReleaseMutant(HANDLE,PLONG); NTSTATUS WINAPI NtReleaseSemaphore(HANDLE,ULONG,PULONG); -NTSTATUS WINAPI NtRemoveIoCompletion(HANDLE,PULONG,PULONG,PIO_STATUS_BLOCK,PLARGE_INTEGER); NTSTATUS WINAPI NtReplaceKey(POBJECT_ATTRIBUTES,HKEY,POBJECT_ATTRIBUTES); NTSTATUS WINAPI NtResetEvent(HANDLE,PULONG); NTSTATUS WINAPI NtRestoreKey(HKEY,HANDLE,ULONG); @@ -1554,12 +1552,10 @@ NTSTATUS WINAPI NtSetDefaultUILanguage(LANGID); NTSTATUS WINAPI NtSetEaFile(HANDLE,PIO_STATUS_BLOCK,PVOID,ULONG); NTSTATUS WINAPI NtSetEvent(HANDLE,PULONG); -NTSTATUS WINAPI NtSetInformationFile(HANDLE,PIO_STATUS_BLOCK,PVOID,ULONG,FILE_INFORMATION_CLASS); NTSTATUS WINAPI NtSetInformationKey(HKEY,const int,PVOID,ULONG); NTSTATUS WINAPI NtSetInformationObject(HANDLE, OBJECT_INFORMATION_CLASS, PVOID, ULONG); NTSTATUS WINAPI NtSetInformationThread(HANDLE,THREADINFOCLASS,LPCVOID,ULONG); NTSTATUS WINAPI NtSetInformationToken(HANDLE,TOKEN_INFORMATION_CLASS,PVOID,ULONG); -NTSTATUS WINAPI NtSetIoCompletion(HANDLE,ULONG,ULONG,NTSTATUS,ULONG); NTSTATUS WINAPI NtSetSecurityObject(HANDLE,SECURITY_INFORMATION,PSECURITY_DESCRIPTOR); NTSTATUS WINAPI NtSetSystemTime(const LARGE_INTEGER*,LARGE_INTEGER*); NTSTATUS WINAPI NtSetTimer(HANDLE, const LARGE_INTEGER*, PTIMERAPCROUTINE, PVOID, BOOLEAN, ULONG, BOOLEAN*); @@ -1968,6 +1964,23 @@ NTSTATUS WINAPI LdrUnloadDll(HMODULE); NTSTATUS WINAPI LdrUnlockLoaderLock(ULONG,ULONG); +/************************************************************************* + * I/O completion functions and structures. + * + * These are not part of standard Winternl.h + */ +typedef struct _FILE_COMPLETION_INFORMATION { + HANDLE CompletionPort; + ULONG_PTR CompletionKey; +} FILE_COMPLETION_INFORMATION; +typedef FILE_COMPLETION_INFORMATION *PFILE_COMPLETION_INFORMATION; + +NTSTATUS WINAPI NtCreateIoCompletion(PHANDLE,ACCESS_MASK,ULONG_PTR,ULONG); +NTSTATUS WINAPI NtSetIoCompletion(HANDLE,ULONG_PTR,LPOVERLAPPED,ULONG,ULONG); +NTSTATUS WINAPI NtRemoveIoCompletion(HANDLE,PULONG_PTR,LPOVERLAPPED*,PIO_STATUS_BLOCK,PLARGE_INTEGER); +NTSTATUS WINAPI NtSetInformationFile(HANDLE,PIO_STATUS_BLOCK,PVOID,ULONG,FILE_INFORMATION_CLASS); + + /* list manipulation macros */ #define InitializeListHead(le) (void)((le)->Flink = (le)->Blink = (le)) #define InsertHeadList(le,e) do { PLIST_ENTRY f = (le)->Flink; (e)->Flink = f; (e)->Blink = (le); f->Blink = (e); (le)->Flink = (e); } while (0) diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/server/iocompletion.c wine/server/iocompletion.c --- wine.orig/server/iocompletion.c 1969-12-31 19:00:00.000000000 -0500 +++ wine/server/iocompletion.c 2005-05-04 14:13:16.000000000 -0400 @@ -0,0 +1,282 @@ +/* + * I/O Completion Ports + * + * Copyright (C) 2003 Robert Shearman + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include "config.h" +#include "wine/port.h" + +#include <stdio.h> + +#include "windef.h" + +#include "handle.h" +#include "thread.h" +#include "request.h" +#include "wine/list.h" + +static void io_completion_dump( struct object *obj, int verbose ); +static void io_completion_destroy( struct object *obj ); +static int io_completion_signaled( struct object * obj, struct thread * thread ); +static int io_completion_satisfied( struct object * obj, struct thread * thread ); + +extern void select_on( int count, void *cookie, const obj_handle_t *handles, + int flags, const abs_time_t *timeout ); + +struct io_completion_data +{ + struct list entry; + unsigned int bytes_transferred; + void * completion_key; + void * overlapped; +}; + +struct io_completion_assigned_data +{ + struct list entry; + struct thread * thread; + struct io_completion_data * data; +}; + +struct io_completion_port +{ + struct object obj; + unsigned int concurrent_threads; + unsigned int max_concurrent_threads; /* FIXME: should we honour this? */ + struct list data; /* fifo queue for io_completion_data */ + + struct list assigned_data; /* io_completion_assigned_data */ + + /* Used to determine whether we have initiated the select() + * through GetQueuedCompletionStatus or whether the client + * has done WaitForSingleObject */ + int satisfied; +}; + +static const struct object_ops io_completion_ops = +{ + sizeof(struct io_completion_port), /* size */ + io_completion_dump, /* dump */ + add_queue, /* add_queue */ + remove_queue, /* remove_queue */ + io_completion_signaled, /* signaled */ + io_completion_satisfied, /* satisfied */ + no_get_fd, /* get_fd */ + io_completion_destroy /* destroy */ +}; + +static void io_completion_dump( struct object *obj, int verbose ) +{ + struct io_completion_port *port = (struct io_completion_port *)obj; + assert( obj->ops == &io_completion_ops ); + fprintf( stderr, "I/O completion port max_threads=%d\n", + port->max_concurrent_threads); +} + +static void io_completion_destroy( struct object *obj ) +{ + struct list * current; + struct io_completion_port *port = (struct io_completion_port *)obj; + assert( obj->ops == &io_completion_ops ); + + /* free data queue */ + for (current = list_head(&port->data); + !list_empty(&port->data); current = list_head(&port->data)) + { + list_remove(current); + free( LIST_ENTRY(current, struct io_completion_data, entry) ); + } + + /* free assigned data queue */ + for (current = list_head(&port->assigned_data); + !list_empty(&port->assigned_data); current = list_head(&port->assigned_data)) + { + list_remove(current); + free( LIST_ENTRY(current, struct io_completion_assigned_data, entry)->data ); + free( LIST_ENTRY(current, struct io_completion_assigned_data, entry) ); + } +} + +static int io_completion_signaled( struct object * obj, struct thread * thread ) +{ + struct io_completion_port * port = (struct io_completion_port *)obj; + assert( obj->ops == &io_completion_ops ); + return !list_empty(&port->data); +} + +static int io_completion_satisfied( struct object * obj, struct thread * thread ) +{ + struct io_completion_port * port = (struct io_completion_port *)obj; + assert( obj->ops == &io_completion_ops ); + return port->satisfied; +} + +static struct object * create_io_completion(unsigned int concurrent_threads) +{ + struct io_completion_port * port; + if (!(port = alloc_object( &io_completion_ops ))) + { + return NULL; + } + + list_init(&port->data); + list_init(&port->assigned_data); + port->concurrent_threads = 0; + port->max_concurrent_threads = concurrent_threads; + port->satisfied = 1; /* abandon any waits on the port immediately */ + + return &port->obj; +} + +static void assign_data(struct io_completion_port * port, struct thread * thread) +{ + struct io_completion_data * data = + LIST_ENTRY(list_head(&port->data), struct io_completion_data, entry); + struct io_completion_assigned_data * assigned_data; + if ((assigned_data = mem_alloc(sizeof(*data))) != NULL) + { + list_init(&assigned_data->entry); + assigned_data->data = data; + assigned_data->thread = thread; + + list_add_head( &port->assigned_data, &assigned_data->entry ); + + list_remove( &data->entry ); + } +} + +DECL_HANDLER(create_io_completion) +{ + struct object * obj; + + reply->handle = 0; + if ((obj = create_io_completion(req->concurrent_threads)) != NULL) + { + reply->handle = alloc_handle(current->process, obj, req->access | SYNCHRONIZE, FALSE /*inherit flag*/); + release_object( obj ); + } +} + +DECL_HANDLER(remove_io_completion) +{ + struct io_completion_port * port = (struct io_completion_port *)get_handle_obj( + current->process, + req->handle, + GENERIC_READ, + &io_completion_ops); + + if (!port) + { + reply->bytes_transferred = 0; + reply->completion_key = NULL; + reply->overlapped = NULL; + return; + } + + if (!list_empty(&port->data)) /* there is waiting data */ + { + struct io_completion_data * data = + LIST_ENTRY(list_head(&port->data), struct io_completion_data, entry); + reply->bytes_transferred = data->bytes_transferred; + reply->completion_key = data->completion_key; + reply->overlapped = data->overlapped; + /* remove the data from the completion port */ + list_remove( &data->entry ); + free( data ); + } + else /* there is no waiting data */ + { + port->satisfied = 0; /* don't abandon wait on the port */ + select_on(1, req->cookie, &req->handle, SELECT_TIMEOUT, &req->timeout); + port->satisfied = 1; /* abandon any waits on the port immediately */ + reply->bytes_transferred = 0; + reply->completion_key = NULL; + reply->overlapped = NULL; + } + release_object( &port->obj ); +} + +DECL_HANDLER(set_io_completion) +{ + struct io_completion_data * data; + struct io_completion_port * port = (struct io_completion_port *)get_handle_obj( + current->process, + req->handle, + GENERIC_WRITE, + &io_completion_ops); + + if (!port) + return; + + if ((data = mem_alloc(sizeof(*data))) != NULL) + { + list_init(&data->entry); + data->bytes_transferred = req->bytes_transferred; + data->completion_key = req->completion_key; + data->overlapped = req->overlapped; + + list_add_tail(&port->data, &data->entry); + + if (port->obj.tail != NULL) /* there is a waiting thread */ + { + struct wait_queue_entry * waiting = port->obj.tail; + assign_data( port, waiting->thread ); + wake_thread( waiting->thread ); + } + } + release_object( &port->obj ); +} + +DECL_HANDLER(remove_io_completion_assigned) +{ + struct io_completion_assigned_data * assigned_data; + struct list * cursor; + struct io_completion_port * port = (struct io_completion_port *)get_handle_obj( + current->process, + req->handle, + GENERIC_WRITE, + &io_completion_ops); + + if (!port) + { + reply->bytes_transferred = 0; + reply->completion_key = NULL; + reply->overlapped = NULL; + return; + } + + LIST_FOR_EACH(cursor, &port->assigned_data) + { + assigned_data = LIST_ENTRY(cursor, struct io_completion_assigned_data, entry); + if (assigned_data->thread == current) + { + reply->bytes_transferred = assigned_data->data->bytes_transferred; + reply->completion_key = assigned_data->data->completion_key; + reply->overlapped = assigned_data->data->overlapped; + list_remove( &assigned_data->entry ); + free( assigned_data->data ); + free( assigned_data ); + return; + } + } + + set_error(STATUS_INVALID_PARAMETER); + reply->bytes_transferred = 0; + reply->completion_key = NULL; + reply->overlapped = NULL; +} diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/server/Makefile.in wine/server/Makefile.in --- wine.orig/server/Makefile.in 2005-03-30 14:02:15.000000000 -0500 +++ wine/server/Makefile.in 2005-05-04 14:38:33.000000000 -0400 @@ -21,6 +21,7 @@ file.c \ handle.c \ hook.c \ + iocompletion.c \ mailslot.c \ main.c \ mapping.c \ diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/server/request.h wine/server/request.h --- wine.orig/server/request.h 2005-04-28 08:04:14.000000000 -0400 +++ wine/server/request.h 2005-05-04 14:40:31.000000000 -0400 @@ -280,6 +280,10 @@ DECL_HANDLER(start_hook_chain); DECL_HANDLER(finish_hook_chain); DECL_HANDLER(get_next_hook); +DECL_HANDLER(create_io_completion); +DECL_HANDLER(set_io_completion); +DECL_HANDLER(remove_io_completion); +DECL_HANDLER(remove_io_completion_assigned); DECL_HANDLER(create_class); DECL_HANDLER(destroy_class); DECL_HANDLER(set_class_info); @@ -476,6 +480,10 @@ (req_handler)req_start_hook_chain, (req_handler)req_finish_hook_chain, (req_handler)req_get_next_hook, + (req_handler)req_create_io_completion, + (req_handler)req_set_io_completion, + (req_handler)req_remove_io_completion, + (req_handler)req_remove_io_completion_assigned, (req_handler)req_create_class, (req_handler)req_destroy_class, (req_handler)req_set_class_info, diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/server/thread.c wine/server/thread.c --- wine.orig/server/thread.c 2005-04-24 13:35:52.000000000 -0400 +++ wine/server/thread.c 2005-05-04 14:41:21.000000000 -0400 @@ -533,7 +533,7 @@ } /* select on a list of handles */ -static void select_on( int count, void *cookie, const obj_handle_t *handles, +void select_on( int count, void *cookie, const obj_handle_t *handles, int flags, const abs_time_t *timeout, obj_handle_t signal_obj ) { int ret, i; diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/server/trace.c wine/server/trace.c --- wine.orig/server/trace.c 2005-04-28 08:04:14.000000000 -0400 +++ wine/server/trace.c 2005-05-04 14:32:57.000000000 -0400 @@ -2602,6 +2602,33 @@ fprintf( stderr, " client_ptr=%p", req->client_ptr ); } +static void dump_remove_io_completion_request( const struct remove_io_completion_request *req ) +{ + fprintf( stderr, " handle=%p,", req->handle ); + fprintf( stderr, " cookie=%p,", req->cookie ); + fprintf( stderr, " timeout=" ); + dump_abs_time( &req->timeout ); +} + +static void dump_remove_io_completion_reply( const struct remove_io_completion_reply *req ) +{ + fprintf( stderr, " bytes_transferred=%08x,", req->bytes_transferred ); + fprintf( stderr, " completion_key=%p,", req->completion_key ); + fprintf( stderr, " overlapped=%p", req->overlapped ); +} + +static void dump_remove_io_completion_assigned_request( const struct remove_io_completion_assigned_request *req ) +{ + fprintf( stderr, " handle=%p", req->handle ); +} + +static void dump_remove_io_completion_assigned_reply( const struct remove_io_completion_assigned_reply *req ) +{ + fprintf( stderr, " bytes_transferred=%08x,", req->bytes_transferred ); + fprintf( stderr, " completion_key=%p,", req->completion_key ); + fprintf( stderr, " overlapped=%p", req->overlapped ); +} + static void dump_set_class_info_request( const struct set_class_info_request *req ) { fprintf( stderr, " window=%p,", req->window ); @@ -2949,6 +2976,10 @@ (dump_func)dump_start_hook_chain_request, (dump_func)dump_finish_hook_chain_request, (dump_func)dump_get_next_hook_request, + (dump_func)dump_create_io_completion_request, + (dump_func)dump_set_io_completion_request, + (dump_func)dump_remove_io_completion_request, + (dump_func)dump_remove_io_completion_assigned_request, (dump_func)dump_create_class_request, (dump_func)dump_destroy_class_request, (dump_func)dump_set_class_info_request, @@ -3142,6 +3173,10 @@ (dump_func)dump_start_hook_chain_reply, (dump_func)0, (dump_func)dump_get_next_hook_reply, + (dump_func)dump_create_io_completion_reply, + (dump_func)0, + (dump_func)dump_remove_io_completion_reply, + (dump_func)dump_remove_io_completion_assigned_reply, (dump_func)0, (dump_func)dump_destroy_class_reply, (dump_func)dump_set_class_info_reply, @@ -3335,6 +3370,10 @@ "start_hook_chain", "finish_hook_chain", "get_next_hook", + "create_io_completion", + "set_io_completion", + "remove_io_completion", + "remove_io_completion_assigned", "create_class", "destroy_class", "set_class_info", diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/dlls/kernel32/iocompletion.c wine/dlls/kernel32/iocompletion.c --- wine.orig/dlls/kernel32/iocompletion.c 1969-12-31 19:00:00.000000000 -0500 +++ wine/dlls/kernel32/iocompletion.c 2005-05-04 14:08:38.000000000 -0400 @@ -0,0 +1,173 @@ +/* + * I/O Completion Ports + * + * Copyright (C) 2003 Robert Shearman + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include <stdarg.h> + +#include "windef.h" +#include "winbase.h" +#include "winnt.h" +#include "winreg.h" +#include "winternl.h" +#include "ntstatus.h" +#include "wine/debug.h" + +WINE_DEFAULT_DEBUG_CHANNEL(iocompletion); + +HANDLE WINAPI CreateIoCompletionPort( + HANDLE FileHandle, + HANDLE ExistingCompletionPort, + ULONG_PTR CompletionKey, + DWORD NumberOfConcurrentThreads) +{ + HANDLE CompletionPort; + NTSTATUS Status; + + TRACE("(%p, %p, %lx, %ld)\n", + FileHandle, + ExistingCompletionPort, + CompletionKey, + NumberOfConcurrentThreads); + + if (((FileHandle == INVALID_HANDLE_VALUE) && (ExistingCompletionPort != NULL)) || + (ExistingCompletionPort == INVALID_HANDLE_VALUE)) + { + SetLastError(ERROR_INVALID_PARAMETER); + return INVALID_HANDLE_VALUE; + } + if (ExistingCompletionPort == NULL) + { + Status = NtCreateIoCompletion( + &CompletionPort, + GENERIC_ALL, 0, + NumberOfConcurrentThreads); + + if (Status != STATUS_SUCCESS) + { + SetLastError(RtlNtStatusToDosError(Status)); + return INVALID_HANDLE_VALUE; + } + } + else + CompletionPort = ExistingCompletionPort; + + if (FileHandle != INVALID_HANDLE_VALUE) + { + IO_STATUS_BLOCK IoStatusBlock; + FILE_COMPLETION_INFORMATION CompletionInfo; + CompletionInfo.CompletionKey = CompletionKey; + CompletionInfo.CompletionPort = CompletionPort; + + Status = NtSetInformationFile( + FileHandle, + &IoStatusBlock, + (PVOID)&CompletionInfo, + sizeof(FILE_COMPLETION_INFORMATION), + FileCompletionInformation); + + if (Status != STATUS_SUCCESS) + { + SetLastError(RtlNtStatusToDosError(Status)); + return INVALID_HANDLE_VALUE; + } + } + return CompletionPort; +} + +BOOL WINAPI GetQueuedCompletionStatus( + HANDLE CompletionPort, + LPDWORD lpNumberOfBytesTransferred, + PULONG_PTR lpCompletionKey, + LPOVERLAPPED * lplpOverlapped, + DWORD dwMilliseconds) +{ + IO_STATUS_BLOCK CompletionBlock; + NTSTATUS Status; + + TRACE("(CompletionPort %p, %p, %p, %p, %ld)\n", + CompletionPort, + lpNumberOfBytesTransferred, + lpCompletionKey, + lplpOverlapped, + dwMilliseconds); + + if (dwMilliseconds == INFINITE) + Status = NtRemoveIoCompletion( + CompletionPort, + lpCompletionKey, + lplpOverlapped, + &CompletionBlock, + NULL); + else + { + LARGE_INTEGER WaitTime; + /* multiplying two LONGLONGs with at least one LONGLONG having its + * higher long part not zero makes the multiplying a bit harder, + * therefore we do an easy multiply and negate afterwards rather than + * making it a hard multiply by doing "* -10000" + */ + WaitTime.QuadPart = (LONGLONG)dwMilliseconds * (LONGLONG)10000; + WaitTime.QuadPart = -WaitTime.QuadPart; + Status = NtRemoveIoCompletion( + CompletionPort, + lpCompletionKey, + lplpOverlapped, + &CompletionBlock, + &WaitTime); + } + if (Status == STATUS_SUCCESS) + { + *lpNumberOfBytesTransferred = CompletionBlock.Information; + return TRUE; + } + else + { + SetLastError(RtlNtStatusToDosError(Status)); + return FALSE; + } +} + +BOOL WINAPI PostQueuedCompletionStatus( + HANDLE CompletionPort, + DWORD dwNumberOfBytesTransferred, + ULONG_PTR dwCompletionKey, + LPOVERLAPPED lpOverlapped) +{ + NTSTATUS Status; + + TRACE("(CompletionPort %p, %ld, %lx, %p)\n", + CompletionPort, + dwNumberOfBytesTransferred, + dwCompletionKey, + lpOverlapped); + + Status = NtSetIoCompletion( + CompletionPort, + dwCompletionKey, + lpOverlapped, 0, + dwNumberOfBytesTransferred); + + if (Status == STATUS_SUCCESS) + return TRUE; + else + { + SetLastError(RtlNtStatusToDosError(Status)); + return FALSE; + } +} diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/dlls/ntdll/iocompletion.c wine/dlls/ntdll/iocompletion.c --- wine.orig/dlls/ntdll/iocompletion.c 1969-12-31 19:00:00.000000000 -0500 +++ wine/dlls/ntdll/iocompletion.c 2005-05-04 20:05:27.000000000 -0400 @@ -0,0 +1,216 @@ +/* + * I/O Completion Ports + * + * Copyright (C) 2003 Robert Shearman + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include "config.h" +#include <stdarg.h> + +#define NONAMELESSUNION +#include "windef.h" +#include "winbase.h" +#include "winnt.h" +#include "winreg.h" +#include "winternl.h" + +#include "ntdll_misc.h" +#include "wine/server.h" +#include "wine/debug.h" + +WINE_DEFAULT_DEBUG_CHANNEL(ntdll); + +extern int NTDLL_wait_reply( void *cookie ); + +/************************************************************************** + * NtCreateIoCompletion (NTDLL.@) + * + * Params: + * CompletionPort [O]: the handle created + * DesiredAccess [I}: the access desired (e.g. GENERIC_ALL) + * Reserved [I}: unknown + * NumberOfConcurrentThreads [I]: the desired number of concurrent + * threads + * Returns: + * Status + * Notes: + * It is effectively a FIFO queue for data and + * a LIFO queue for threads to "minimize context switches". + * The aim is to keep a small number of threads constantly + * active. + * See: + * MSDN for CreateIoCompletionPort spec and + * the article "Inside I/O Completion Ports" + * (http://www.sysinternals.com/ntw2k/info/comport.shtml) + */ +NTSTATUS WINAPI NtCreateIoCompletion ( + OUT PHANDLE CompletionPort, + IN ACCESS_MASK DesiredAccess, + IN ULONG_PTR Reserved, + IN ULONG NumberOfConcurrentThreads + ) + { + NTSTATUS ret; + + TRACE("(%p, %lx, %lx, %ld)\n", + CompletionPort, + DesiredAccess, + Reserved, + NumberOfConcurrentThreads); + + if (Reserved != 0) + { + FIXME("Reserved != 0 not supported\n"); + return STATUS_INVALID_PARAMETER; + } + + if (DesiredAccess && GENERIC_ALL) + DesiredAccess |= GENERIC_READ | GENERIC_WRITE; + + SERVER_START_REQ( create_io_completion ) + { + req->access = DesiredAccess; + req->concurrent_threads = NumberOfConcurrentThreads; + ret = wine_server_call( req ); + *CompletionPort = reply->handle; + } + SERVER_END_REQ; + + TRACE("returning %lx\n", ret); + return ret; +} + +/************************************************************************** + * NtSetIoCompletion (NTDLL.@) + * + * Params: + * CompletionPort [I]: port to send data to + * CompletionKey [I}: user key to identify this set of data + * lpOverlapped [I}: OVERLAPPED structure to send to port + * NumberOfBytesTransferred [I}: unknown - seems to be set to zero always + * NumberOfBytesToTransfer [I]: Bytes to transfer in this packet of data + * Returns: + * Status + * See: + * MSDN for PostQueuedCompletionStatus spec and + * the article "Inside I/O Completion Ports" + * (http://www.sysinternals.com/ntw2k/info/comport.shtml) + */ +NTSTATUS WINAPI NtSetIoCompletion( + IN HANDLE CompletionPort, + IN ULONG_PTR CompletionKey, + IN LPOVERLAPPED lpOverlapped, + IN ULONG NumberOfBytesTransferred, /* normally set to 0 */ + IN ULONG NumberOfBytesToTransfer /* will become number of bytes transferred in the io operation */ + ) +{ + NTSTATUS ret; + + TRACE("(%p, %lx, %p, %ld, %ld)\n", + CompletionPort, + CompletionKey, + lpOverlapped, + NumberOfBytesTransferred, + NumberOfBytesToTransfer); + + if (NumberOfBytesTransferred != 0) + { + FIXME("NumberOfBytesTransferred != 0 not supported\n"); + return STATUS_INVALID_PARAMETER; + } + + SERVER_START_REQ( set_io_completion ) + { + req->handle = CompletionPort; + req->completion_key = (void *)CompletionKey; + req->overlapped = lpOverlapped; + req->bytes_transferred = NumberOfBytesToTransfer; + ret = wine_server_call( req ); + } + SERVER_END_REQ; + + TRACE("returning %lx\n", ret); + return ret; +} + +/************************************************************************** + * NtRemoveIoCompletion (NTDLL.@) + * + * See: MSDN for GetQueuedCompletionStatus spec and + * the article "Inside I/O Completion Ports" + * (http://www.sysinternals.com/ntw2k/info/comport.shtml) + */ +NTSTATUS WINAPI NtRemoveIoCompletion ( + IN HANDLE CompletionPort, + OUT PULONG_PTR CompletionKey, + OUT LPOVERLAPPED * lplpOverlapped, + OUT PIO_STATUS_BLOCK CompletionStatus, + IN PLARGE_INTEGER WaitTime + ) +{ + NTSTATUS ret; + int cookie; + + TRACE("(%p, %p, %p, %p, %p)\n", + CompletionPort, + CompletionKey, + lplpOverlapped, + CompletionStatus, + WaitTime); + + for (;;) + { + SERVER_START_REQ( remove_io_completion ) + { + req->handle = CompletionPort; + req->cookie = &cookie; + NTDLL_get_server_timeout( &req->timeout, WaitTime ); + ret = wine_server_call( req ); + if (ret == STATUS_SUCCESS) + { + *CompletionKey = (ULONG_PTR)reply->completion_key; + *lplpOverlapped = reply->overlapped; + CompletionStatus->u.Status = STATUS_SUCCESS; + CompletionStatus->Information = reply->bytes_transferred; + } + } + SERVER_END_REQ; + if (ret == STATUS_PENDING) + ret = NTDLL_wait_reply( &cookie ); + if (ret == STATUS_ABANDONED) + { + SERVER_START_REQ( remove_io_completion_assigned ) + { + req->handle = CompletionPort; + ret = wine_server_call( req ); + if (ret == STATUS_SUCCESS) + { + *CompletionKey = (ULONG_PTR)reply->completion_key; + *lplpOverlapped = reply->overlapped; + CompletionStatus->u.Status = STATUS_SUCCESS; + CompletionStatus->Information = reply->bytes_transferred; + } + } + SERVER_END_REQ; + } + break; + } + + TRACE("returning %lx\n", ret); + return ret; +} + diff -Naur -u --exclude-from=autogen.list --exclude-from=excludes.list wine.orig/server/iocompletion.c wine/server/iocompletion.c --- wine.orig/server/iocompletion.c 1969-12-31 19:00:00.000000000 -0500 +++ wine/server/iocompletion.c 2005-05-04 14:13:16.000000000 -0400 @@ -0,0 +1,282 @@ +/* + * I/O Completion Ports + * + * Copyright (C) 2003 Robert Shearman + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include "config.h" +#include "wine/port.h" + +#include <stdio.h> + +#include "windef.h" + +#include "handle.h" +#include "thread.h" +#include "request.h" +#include "wine/list.h" + +static void io_completion_dump( struct object *obj, int verbose ); +static void io_completion_destroy( struct object *obj ); +static int io_completion_signaled( struct object * obj, struct thread * thread ); +static int io_completion_satisfied( struct object * obj, struct thread * thread ); + +extern void select_on( int count, void *cookie, const obj_handle_t *handles, + int flags, const abs_time_t *timeout ); + +struct io_completion_data +{ + struct list entry; + unsigned int bytes_transferred; + void * completion_key; + void * overlapped; +}; + +struct io_completion_assigned_data +{ + struct list entry; + struct thread * thread; + struct io_completion_data * data; +}; + +struct io_completion_port +{ + struct object obj; + unsigned int concurrent_threads; + unsigned int max_concurrent_threads; /* FIXME: should we honour this? */ + struct list data; /* fifo queue for io_completion_data */ + + struct list assigned_data; /* io_completion_assigned_data */ + + /* Used to determine whether we have initiated the select() + * through GetQueuedCompletionStatus or whether the client + * has done WaitForSingleObject */ + int satisfied; +}; + +static const struct object_ops io_completion_ops = +{ + sizeof(struct io_completion_port), /* size */ + io_completion_dump, /* dump */ + add_queue, /* add_queue */ + remove_queue, /* remove_queue */ + io_completion_signaled, /* signaled */ + io_completion_satisfied, /* satisfied */ + no_get_fd, /* get_fd */ + io_completion_destroy /* destroy */ +}; + +static void io_completion_dump( struct object *obj, int verbose ) +{ + struct io_completion_port *port = (struct io_completion_port *)obj; + assert( obj->ops == &io_completion_ops ); + fprintf( stderr, "I/O completion port max_threads=%d\n", + port->max_concurrent_threads); +} + +static void io_completion_destroy( struct object *obj ) +{ + struct list * current; + struct io_completion_port *port = (struct io_completion_port *)obj; + assert( obj->ops == &io_completion_ops ); + + /* free data queue */ + for (current = list_head(&port->data); + !list_empty(&port->data); current = list_head(&port->data)) + { + list_remove(current); + free( LIST_ENTRY(current, struct io_completion_data, entry) ); + } + + /* free assigned data queue */ + for (current = list_head(&port->assigned_data); + !list_empty(&port->assigned_data); current = list_head(&port->assigned_data)) + { + list_remove(current); + free( LIST_ENTRY(current, struct io_completion_assigned_data, entry)->data ); + free( LIST_ENTRY(current, struct io_completion_assigned_data, entry) ); + } +} + +static int io_completion_signaled( struct object * obj, struct thread * thread ) +{ + struct io_completion_port * port = (struct io_completion_port *)obj; + assert( obj->ops == &io_completion_ops ); + return !list_empty(&port->data); +} + +static int io_completion_satisfied( struct object * obj, struct thread * thread ) +{ + struct io_completion_port * port = (struct io_completion_port *)obj; + assert( obj->ops == &io_completion_ops ); + return port->satisfied; +} + +static struct object * create_io_completion(unsigned int concurrent_threads) +{ + struct io_completion_port * port; + if (!(port = alloc_object( &io_completion_ops ))) + { + return NULL; + } + + list_init(&port->data); + list_init(&port->assigned_data); + port->concurrent_threads = 0; + port->max_concurrent_threads = concurrent_threads; + port->satisfied = 1; /* abandon any waits on the port immediately */ + + return &port->obj; +} + +static void assign_data(struct io_completion_port * port, struct thread * thread) +{ + struct io_completion_data * data = + LIST_ENTRY(list_head(&port->data), struct io_completion_data, entry); + struct io_completion_assigned_data * assigned_data; + if ((assigned_data = mem_alloc(sizeof(*data))) != NULL) + { + list_init(&assigned_data->entry); + assigned_data->data = data; + assigned_data->thread = thread; + + list_add_head( &port->assigned_data, &assigned_data->entry ); + + list_remove( &data->entry ); + } +} + +DECL_HANDLER(create_io_completion) +{ + struct object * obj; + + reply->handle = 0; + if ((obj = create_io_completion(req->concurrent_threads)) != NULL) + { + reply->handle = alloc_handle(current->process, obj, req->access | SYNCHRONIZE, FALSE /*inherit flag*/); + release_object( obj ); + } +} + +DECL_HANDLER(remove_io_completion) +{ + struct io_completion_port * port = (struct io_completion_port *)get_handle_obj( + current->process, + req->handle, + GENERIC_READ, + &io_completion_ops); + + if (!port) + { + reply->bytes_transferred = 0; + reply->completion_key = NULL; + reply->overlapped = NULL; + return; + } + + if (!list_empty(&port->data)) /* there is waiting data */ + { + struct io_completion_data * data = + LIST_ENTRY(list_head(&port->data), struct io_completion_data, entry); + reply->bytes_transferred = data->bytes_transferred; + reply->completion_key = data->completion_key; + reply->overlapped = data->overlapped; + /* remove the data from the completion port */ + list_remove( &data->entry ); + free( data ); + } + else /* there is no waiting data */ + { + port->satisfied = 0; /* don't abandon wait on the port */ + select_on(1, req->cookie, &req->handle, SELECT_TIMEOUT, &req->timeout); + port->satisfied = 1; /* abandon any waits on the port immediately */ + reply->bytes_transferred = 0; + reply->completion_key = NULL; + reply->overlapped = NULL; + } + release_object( &port->obj ); +} + +DECL_HANDLER(set_io_completion) +{ + struct io_completion_data * data; + struct io_completion_port * port = (struct io_completion_port *)get_handle_obj( + current->process, + req->handle, + GENERIC_WRITE, + &io_completion_ops); + + if (!port) + return; + + if ((data = mem_alloc(sizeof(*data))) != NULL) + { + list_init(&data->entry); + data->bytes_transferred = req->bytes_transferred; + data->completion_key = req->completion_key; + data->overlapped = req->overlapped; + + list_add_tail(&port->data, &data->entry); + + if (port->obj.tail != NULL) /* there is a waiting thread */ + { + struct wait_queue_entry * waiting = port->obj.tail; + assign_data( port, waiting->thread ); + wake_thread( waiting->thread ); + } + } + release_object( &port->obj ); +} + +DECL_HANDLER(remove_io_completion_assigned) +{ + struct io_completion_assigned_data * assigned_data; + struct list * cursor; + struct io_completion_port * port = (struct io_completion_port *)get_handle_obj( + current->process, + req->handle, + GENERIC_WRITE, + &io_completion_ops); + + if (!port) + { + reply->bytes_transferred = 0; + reply->completion_key = NULL; + reply->overlapped = NULL; + return; + } + + LIST_FOR_EACH(cursor, &port->assigned_data) + { + assigned_data = LIST_ENTRY(cursor, struct io_completion_assigned_data, entry); + if (assigned_data->thread == current) + { + reply->bytes_transferred = assigned_data->data->bytes_transferred; + reply->completion_key = assigned_data->data->completion_key; + reply->overlapped = assigned_data->data->overlapped; + list_remove( &assigned_data->entry ); + free( assigned_data->data ); + free( assigned_data ); + return; + } + } + + set_error(STATUS_INVALID_PARAMETER); + reply->bytes_transferred = 0; + reply->completion_key = NULL; + reply->overlapped = NULL; +} diff --git a/dlls/kernel32/Makefile.in b/dlls/kernel32/Makefile.in index c725509..be5da9e 100644 --- a/dlls/kernel32/Makefile.in +++ b/dlls/kernel32/Makefile.in @@ -35,6 +35,7 @@ C_SRCS = \ file.c \ file16.c \ format_msg.c \ + iocompletion.c \ global16.c \ heap.c \ instr.c \ diff --git a/dlls/kernel32/kernel32.spec b/dlls/kernel32/kernel32.spec index 7502bb8..76ee2dc 100644 --- a/dlls/kernel32/kernel32.spec +++ b/dlls/kernel32/kernel32.spec @@ -828,6 +828,7 @@ # @ stub NumaVirtualQueryNode @ stdcall PeekConsoleInputA(ptr ptr long ptr) @ stdcall PeekConsoleInputW(ptr ptr long ptr) @ stdcall PeekNamedPipe(long ptr long ptr ptr ptr) +# PATCH: @ stdcall PostQueuedCompletionStatus(ptr long long ptr) @ stdcall PostQueuedCompletionStatus(long long ptr ptr) @ stdcall PrepareTape(ptr long long) @ stub PrivCopyFileExW diff --git a/dlls/kernel32/sync.c b/dlls/kernel32/sync.c index eaa0f70..c517744 100644 --- a/dlls/kernel32/sync.c +++ b/dlls/kernel32/sync.c @@ -1766,41 +1766,6 @@ BOOL WINAPI SetMailslotInfo( HANDLE hMai return TRUE; } - -/****************************************************************************** - * CreateIoCompletionPort (KERNEL32.@) - */ -HANDLE WINAPI CreateIoCompletionPort(HANDLE hFileHandle, HANDLE hExistingCompletionPort, - ULONG_PTR CompletionKey, DWORD dwNumberOfConcurrentThreads) -{ - FIXME("(%p, %p, %08lx, %08x): stub.\n", - hFileHandle, hExistingCompletionPort, CompletionKey, dwNumberOfConcurrentThreads); - SetLastError(ERROR_CALL_NOT_IMPLEMENTED); - return NULL; -} - - -/****************************************************************************** - * GetQueuedCompletionStatus (KERNEL32.@) - */ -BOOL WINAPI GetQueuedCompletionStatus( HANDLE CompletionPort, LPDWORD lpNumberOfBytesTransferred, - PULONG_PTR pCompletionKey, LPOVERLAPPED *lpOverlapped, - DWORD dwMilliseconds ) -{ - FIXME("(%p,%p,%p,%p,%d), stub!\n", - CompletionPort,lpNumberOfBytesTransferred,pCompletionKey,lpOverlapped,dwMilliseconds); - SetLastError(ERROR_CALL_NOT_IMPLEMENTED); - return FALSE; -} - -BOOL WINAPI PostQueuedCompletionStatus( HANDLE CompletionPort, DWORD dwNumberOfBytes, - ULONG_PTR dwCompletionKey, LPOVERLAPPED lpOverlapped) -{ - FIXME("%p %d %08lx %p\n", CompletionPort, dwNumberOfBytes, dwCompletionKey, lpOverlapped ); - SetLastError(ERROR_CALL_NOT_IMPLEMENTED); - return FALSE; -} - /****************************************************************************** * CreateJobObjectW (KERNEL32.@) */ diff --git a/dlls/ntdll/Makefile.in b/dlls/ntdll/Makefile.in index 37b18f1..7496930 100644 --- a/dlls/ntdll/Makefile.in +++ b/dlls/ntdll/Makefile.in @@ -21,6 +21,7 @@ C_SRCS = \ file.c \ handletable.c \ heap.c \ + iocompletion.c \ large_int.c \ loader.c \ loadorder.c \ diff --git a/dlls/ntdll/ntdll.spec b/dlls/ntdll/ntdll.spec index 803ed53..baf5ff8 100644 --- a/dlls/ntdll/ntdll.spec +++ b/dlls/ntdll/ntdll.spec @@ -124,6 +124,7 @@ # @ stub NtCreateDebugObject @ stdcall NtCreateEvent(long long long long long) @ stub NtCreateEventPair @ stdcall NtCreateFile(ptr long ptr ptr long long long ptr long long ptr) +# PATCH: @ stdcall NtCreateIoCompletion(ptr long long long) @ stdcall NtCreateIoCompletion(ptr long ptr long) # @ stub NtCreateJobObject # @ stub NtCreateJobSet @@ -328,6 +329,7 @@ # @ stub NtSetInformationJobObject @ stdcall NtSetInformationThread(long long ptr long) @ stdcall NtSetInformationToken(long long ptr long) @ stdcall NtSetIntervalProfile(long long) +# PATCH: @ stdcall NtSetIoCompletion(ptr long ptr long long) @ stdcall NtSetIoCompletion(ptr long ptr long long) @ stub NtSetLdtEntries @ stub NtSetLowEventPair diff --git a/dlls/ntdll/sync.c b/dlls/ntdll/sync.c index 852bfde..bc0470b 100644 --- a/dlls/ntdll/sync.c +++ b/dlls/ntdll/sync.c @@ -618,7 +618,7 @@ NTSTATUS WINAPI NtSetTimerResolution(IN * * Wait for a reply on the waiting pipe of the current thread. */ -static int wait_reply( void *cookie ) +int NTDLL_wait_reply( void *cookie ) { int signaled; struct wake_up_reply reply; @@ -631,7 +631,7 @@ static int wait_reply( void *cookie ) if (!reply.cookie) break; /* thread got killed */ if (reply.cookie == cookie) return reply.signaled; /* we stole another reply, wait for the real one */ - signaled = wait_reply( cookie ); + signaled = NTDLL_wait_reply( cookie ); /* and now put the wrong one back in the pipe */ for (;;) { @@ -726,7 +726,7 @@ NTSTATUS NTDLL_wait_for_multiple_objects ret = wine_server_call( req ); } SERVER_END_REQ; - if (ret == STATUS_PENDING) ret = wait_reply( &cookie ); + if (ret == STATUS_PENDING) ret = NTDLL_wait_reply( &cookie ); if (ret != STATUS_USER_APC) break; call_apcs( (flags & SELECT_ALERTABLE) != 0 ); if (flags & SELECT_ALERTABLE) break; diff --git a/include/winternl.h b/include/winternl.h index 19b2ac9..d27d4f2 100644 --- a/include/winternl.h +++ b/include/winternl.h @@ -1763,7 +1763,6 @@ NTSTATUS WINAPI NtCreateDirectoryObject NTSTATUS WINAPI NtCreateEvent(PHANDLE,ACCESS_MASK,const OBJECT_ATTRIBUTES *,BOOLEAN,BOOLEAN); NTSTATUS WINAPI NtCreateEventPair(PHANDLE,ACCESS_MASK,POBJECT_ATTRIBUTES); NTSTATUS WINAPI NtCreateFile(PHANDLE,ACCESS_MASK,POBJECT_ATTRIBUTES,PIO_STATUS_BLOCK,PLARGE_INTEGER,ULONG,ULONG,ULONG,ULONG,PVOID,ULONG); -NTSTATUS WINAPI NtCreateIoCompletion(PHANDLE,ACCESS_MASK,POBJECT_ATTRIBUTES,ULONG); NTSTATUS WINAPI NtCreateKey(PHANDLE,ACCESS_MASK,const OBJECT_ATTRIBUTES*,ULONG,const UNICODE_STRING*,ULONG,PULONG); NTSTATUS WINAPI NtCreateMailslotFile(PHANDLE,ACCESS_MASK,POBJECT_ATTRIBUTES,PIO_STATUS_BLOCK,ULONG,ULONG,ULONG,PLARGE_INTEGER); NTSTATUS WINAPI NtCreateMutant(HANDLE*,ACCESS_MASK,const OBJECT_ATTRIBUTES*,BOOLEAN); @@ -1882,7 +1881,6 @@ NTSTATUS WINAPI NtReadVirtualMemory(HAN NTSTATUS WINAPI NtRegisterThreadTerminatePort(HANDLE); NTSTATUS WINAPI NtReleaseMutant(HANDLE,PLONG); NTSTATUS WINAPI NtReleaseSemaphore(HANDLE,ULONG,PULONG); -NTSTATUS WINAPI NtRemoveIoCompletion(HANDLE,PULONG_PTR,PIO_STATUS_BLOCK,PULONG,PLARGE_INTEGER); NTSTATUS WINAPI NtReplaceKey(POBJECT_ATTRIBUTES,HANDLE,POBJECT_ATTRIBUTES); NTSTATUS WINAPI NtReplyPort(HANDLE,PLPC_MESSAGE); NTSTATUS WINAPI NtReplyWaitReceivePort(HANDLE,PULONG,PLPC_MESSAGE,PLPC_MESSAGE); @@ -1904,14 +1902,12 @@ NTSTATUS WINAPI NtSetEvent(HANDLE,PULON NTSTATUS WINAPI NtSetHighEventPair(HANDLE); NTSTATUS WINAPI NtSetHighWaitLowEventPair(HANDLE); NTSTATUS WINAPI NtSetHighWaitLowThread(VOID); -NTSTATUS WINAPI NtSetInformationFile(HANDLE,PIO_STATUS_BLOCK,PVOID,ULONG,FILE_INFORMATION_CLASS); NTSTATUS WINAPI NtSetInformationKey(HANDLE,const int,PVOID,ULONG); NTSTATUS WINAPI NtSetInformationObject(HANDLE, OBJECT_INFORMATION_CLASS, PVOID, ULONG); NTSTATUS WINAPI NtSetInformationProcess(HANDLE,PROCESS_INFORMATION_CLASS,PVOID,ULONG); NTSTATUS WINAPI NtSetInformationThread(HANDLE,THREADINFOCLASS,LPCVOID,ULONG); NTSTATUS WINAPI NtSetInformationToken(HANDLE,TOKEN_INFORMATION_CLASS,PVOID,ULONG); NTSTATUS WINAPI NtSetIntervalProfile(ULONG,KPROFILE_SOURCE); -NTSTATUS WINAPI NtSetIoCompletion(HANDLE,ULONG_PTR,PIO_STATUS_BLOCK,ULONG,ULONG); NTSTATUS WINAPI NtSetLdtEntries(ULONG,LDT_ENTRY,ULONG,LDT_ENTRY); NTSTATUS WINAPI NtSetLowEventPair(HANDLE); NTSTATUS WINAPI NtSetLowWaitHighEventPair(HANDLE); @@ -2367,11 +2363,21 @@ NTSTATUS WINAPI LdrQueryProcessModuleInf NTSTATUS WINAPI LdrUnloadDll(HMODULE); NTSTATUS WINAPI LdrUnlockLoaderLock(ULONG,ULONG); +/************************************************************************* + * I/O completion functions and structures. + * + * These are not part of standard Winternl.h + */ typedef struct _FILE_COMPLETION_INFORMATION { HANDLE CompletionPort; ULONG_PTR CompletionKey; } FILE_COMPLETION_INFORMATION; typedef FILE_COMPLETION_INFORMATION *PFILE_COMPLETION_INFORMATION; +NTSTATUS WINAPI NtCreateIoCompletion(PHANDLE,ACCESS_MASK,ULONG_PTR,ULONG); +NTSTATUS WINAPI NtSetIoCompletion(HANDLE,ULONG_PTR,LPOVERLAPPED,ULONG,ULONG); +NTSTATUS WINAPI NtRemoveIoCompletion(HANDLE,PULONG_PTR,LPOVERLAPPED*,PIO_STATUS_BLOCK,PLARGE_INTEGER); +NTSTATUS WINAPI NtSetInformationFile(HANDLE,PIO_STATUS_BLOCK,PVOID,ULONG,FILE_INFORMATION_CLASS); + /* list manipulation macros */ #define InitializeListHead(le) (void)((le)->Flink = (le)->Blink = (le)) diff --git a/server/Makefile.in b/server/Makefile.in index f62bd1d..2c0daa4 100644 --- a/server/Makefile.in +++ b/server/Makefile.in @@ -24,6 +24,7 @@ C_SRCS = \ file.c \ handle.c \ hook.c \ + iocompletion.c \ mailslot.c \ main.c \ mapping.c \ diff --git a/server/request.h b/server/request.h index 8cfeccd..c6b6a8b 100644 --- a/server/request.h +++ b/server/request.h @@ -302,6 +302,10 @@ DECL_HANDLER(remove_hook); DECL_HANDLER(start_hook_chain); DECL_HANDLER(finish_hook_chain); DECL_HANDLER(get_hook_info); +DECL_HANDLER(create_io_completion); +DECL_HANDLER(set_io_completion); +DECL_HANDLER(remove_io_completion); +DECL_HANDLER(remove_io_completion_assigned); DECL_HANDLER(create_class); DECL_HANDLER(destroy_class); DECL_HANDLER(set_class_info); @@ -521,6 +525,10 @@ static const req_handler req_handlers[RE (req_handler)req_start_hook_chain, (req_handler)req_finish_hook_chain, (req_handler)req_get_hook_info, + (req_handler)req_create_io_completion, + (req_handler)req_set_io_completion, + (req_handler)req_remove_io_completion, + (req_handler)req_remove_io_completion_assigned, (req_handler)req_create_class, (req_handler)req_destroy_class, (req_handler)req_set_class_info, diff --git a/server/thread.c b/server/thread.c index 05cf3cd..8581b0b 100644 --- a/server/thread.c +++ b/server/thread.c @@ -548,7 +548,7 @@ static int signal_object( obj_handle_t h } /* select on a list of handles */ -static void select_on( int count, void *cookie, const obj_handle_t *handles, +void select_on( int count, void *cookie, const obj_handle_t *handles, int flags, const abs_time_t *timeout, obj_handle_t signal_obj ) { int ret, i; diff --git a/server/trace.c b/server/trace.c index b128ad5..1dd2f1b 100644 --- a/server/trace.c +++ b/server/trace.c @@ -2975,6 +2975,33 @@ static void dump_destroy_class_reply( co fprintf( stderr, " client_ptr=%p", req->client_ptr ); } +static void dump_remove_io_completion_request( const struct remove_io_completion_request *req ) +{ + fprintf( stderr, " handle=%p,", req->handle ); + fprintf( stderr, " cookie=%p,", req->cookie ); + fprintf( stderr, " timeout=" ); + dump_abs_time( &req->timeout ); +} + +static void dump_remove_io_completion_reply( const struct remove_io_completion_reply *req ) +{ + fprintf( stderr, " bytes_transferred=%08x,", req->bytes_transferred ); + fprintf( stderr, " completion_key=%p,", req->completion_key ); + fprintf( stderr, " overlapped=%p", req->overlapped ); +} + +static void dump_remove_io_completion_assigned_request( const struct remove_io_completion_assigned_request *req ) +{ + fprintf( stderr, " handle=%p", req->handle ); +} + +static void dump_remove_io_completion_assigned_reply( const struct remove_io_completion_assigned_reply *req ) +{ + fprintf( stderr, " bytes_transferred=%08x,", req->bytes_transferred ); + fprintf( stderr, " completion_key=%p,", req->completion_key ); + fprintf( stderr, " overlapped=%p", req->overlapped ); +} + static void dump_set_class_info_request( const struct set_class_info_request *req ) { fprintf( stderr, " window=%p,", req->window ); @@ -3456,6 +3483,10 @@ static const dump_func req_dumpers[REQ_N (dump_func)dump_start_hook_chain_request, (dump_func)dump_finish_hook_chain_request, (dump_func)dump_get_hook_info_request, + (dump_func)dump_create_io_completion_request, + (dump_func)dump_set_io_completion_request, + (dump_func)dump_remove_io_completion_request, + (dump_func)dump_remove_io_completion_assigned_request, (dump_func)dump_create_class_request, (dump_func)dump_destroy_class_request, (dump_func)dump_set_class_info_request, @@ -3672,6 +3703,10 @@ static const dump_func reply_dumpers[REQ (dump_func)dump_start_hook_chain_reply, (dump_func)0, (dump_func)dump_get_hook_info_reply, + (dump_func)dump_create_io_completion_reply, + (dump_func)0, + (dump_func)dump_remove_io_completion_reply, + (dump_func)dump_remove_io_completion_assigned_reply, (dump_func)0, (dump_func)dump_destroy_class_reply, (dump_func)dump_set_class_info_reply, @@ -3888,6 +3923,10 @@ static const char * const req_names[REQ_ "start_hook_chain", "finish_hook_chain", "get_hook_info", + "create_io_completion", + "set_io_completion", + "remove_io_completion", + "remove_io_completion_assigned", "create_class", "destroy_class", "set_class_info",

1 0

Re: oleaut32: use setlocale() around sprintfW to force use of period as decimal separator (RESEND)
by Alexandre Julliard Dec. 14, 2006

Dec. 14, 2006

Alex Villacís Lasso <a_villacis(a)palosanto.com> writes: > @@ -4159,7 +4161,9 @@ > { > WCHAR buff[256]; > > + setlocale(LC_ALL, "C"); > sprintfW( buff, szFloatFormatW, fltIn ); > + setlocale(LC_ALL, ""); > return VarDecFromStr(buff, LOCALE_EN_US, 0, pDecOut); You can't do that, setlocale() affects the whole process. In any case we shouldn't need to go through a string to convert a float to a decimal. -- Alexandre Julliard julliard(a)winehq.org

2 4

Re: mshtml #5: Set default print template in exec_print.
by Frank Richter Dec. 14, 2006

Dec. 14, 2006

On 14.12.2006 00:21, Jacek Caban wrote: > --- a/dlls/mshtml/resource.h > +++ b/dlls/mshtml/resource.h > @@ -28,6 +28,9 @@ > > #define IDS_MESSAGE_BOX_TITLE 2213 > > +#define IDS_PRINT_HEADER_TEMPLATE 8403 > +#define IDS_PRINT_FOOTER_TEMPLATE 8404 > + Could it be you forgot to submit the .rc file changes for these new strings? -f.r.

2 2

Re: [1/2] tests: Add skip_wine, a standard way to skip tests that would cause Wine to crash.
by Alexandre Julliard Dec. 14, 2006

Dec. 14, 2006

Francois Gouget <fgouget(a)free.fr> writes: > Currently there is no standard way of adding a test that works on > Windows but is known to cause Wine to crash. For instance this is > the case for bug 6827: > > Bug 6827: The SimSig installer crashes > http://bugs.winehq.org/show_bug.cgi?id=6827 > > This essentially means that such tests just don't get written, or get > put inside '#if 0' or 'if (0)' sections so that they are never executed, > not even on Windows. This means we won't get good Windows coverage until > Wine has been 'fixed'. It's not really better with skip_wine; since the test will never be run on Wine it's useless for regression testing, and we won't even notice when the bug gets fixed. If the test doesn't crash on Windows, the right way is to fix Wine to not crash either, that's usually not too hard (making the test succeed may be harder, but that's why we have todo_wine). -- Alexandre Julliard julliard(a)winehq.org

2 2

Re: msxml.dll: nextNode and reset functions
by Huw Davies Dec. 14, 2006

Dec. 14, 2006

On Thu, Dec 14, 2006 at 11:41:25AM +0100, Colin Pitrat wrote: > Change Log: > Add an iterator to _xmlnodemap struct > Implement nextNode function that give nodes one by one > Implement resetNode that reset the iterator > > It's my first patch, so please tell me if something is wrong. I saw on > http://www.winehq.com/site/sending_patches that I should write a test > case, but I don't know how to do this (do I need windows ? because I > don't have it). Yes, you really should write a test for this. For this dll you could even use a (legal) native version of msxml3.dll run under Wine. Some specific points about your patch: * You seem to have your editor set to display tabs as width 4, please don't do this. Don't use tabs at all and use 4 spaces per indent level. * You don't initialize the iterator member of xmlnodemap, this should be done in create_nodemap. Huw. -- Huw Davies huw(a)codeweavers.com

1 0

Re: Add ddrawex dll
by Alexandre Julliard Dec. 14, 2006

Dec. 14, 2006

Ulrich Czekalla <ulrich.czekalla(a)utoronto.ca> writes: > diff --git a/dlls/dxguid/dxguid.c b/dlls/dxguid/dxguid.c > index c2c102f..7ea8d17 100644 > --- a/dlls/dxguid/dxguid.c > +++ b/dlls/dxguid/dxguid.c > @@ -44,3 +44,4 @@ > #include "ddrawi.h" > #include "dxfile.h" > #include "d3drm.h" > +#include "ddrawex.h" There's no ddrawex.h in your patch (note that there isn't one either in recent SDKs, so you may want to keep these definitions private). -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: mshtml #1: Added test of CompareUrl.
by Paul Vriens Dec. 13, 2006

Dec. 13, 2006

On Tue, 2006-12-12 at 21:52 +0100, Jacek Caban wrote: > + hres = IInternetProtocolInfo_CompareUrl(protocol_info, NULL, > NULL, 0xdeadbeef); > + ok(hres == E_NOTIMPL, "CompareUrl failed: %08x\n", hres); > + Will these tests also return E_NOTIMPL on Windows? The tests should be against Windows and worked around if it's not implemented in Wine. Cheers, Paul.

2 1

ntdll/signal_i386.c: __wine_enter_vm86 issue
by Christoph Bumiller Dec. 13, 2006

Dec. 13, 2006

I tried to run some DOS games with wine recently and since most of them crashed with a page fault I tried to find out why and so, examining the last function calls before faulting revealed that when __wine_enter_vm86 in signal_i386.c returns from res = vm86_enter(...) and encounters a VM86TYPE(res) == VM86_UNKNOWN it calls the raise_segv_exception, which in turn calls NtSetContextThread. This one calls set_cpu_context which tries to reload the context of the VM86 thread an, of course, crashes since it can't load the segment registers with invalid (non GDT) values. Backtrace: =>1 0x7efe2a57 in ntdll (+0x52a57) (0x7d55e348) 2 0x7efd9a8d NtSetContextThread+0x77() in ntdll (0x7d55e478) 3 0x7efd412e in ntdll (+0x4412e) (0x7d55e498) 4 0x7efd5032 __wine_enter_vm86+0x1d6() in ntdll (0x7d55e608) 5 0x7eeb9b15 K32WOWCallback16Ex+0x417() in kernel32 (0x7d55e668) fixme:dbghelp_dwarf:dwarf2_parse_variable Unsupported constant max_clusters in function fixme:dbghelp_dwarf:dwarf2_parse_variable Unsupported constant max_sectors_per_cluster in function fixme:dbghelp_dwarf:dwarf2_parse_variable Unsupported constant max_bytes_per_sector in function 6 0x7d6f50bb DOSVM_Enter+0xab(context=0x7d55e760) [/home/chr/WINE/CVS_BUILD/dlls/winedos/dosvm.c:586] in winedos (0x7d55e748) 7 0x7d71254b MZ_DOSVM+0xbb(lpExtra=0x0) [/home/chr/WINE/CVS_BUILD/dlls/winedos/module.c:638] in winedos (0x7d55ea48) 8 0x7eeacdf6 in kernel32 (+0x6cdf6) (0x7d55eb28) 9 0x7efd9939 in ntdll (+0x49939) (0x7d55f448) 10 0xb7de1512 start_thread+0x84() in libpthread.so.0 (0x7d55f4d8) 11 0xb7d7afde __clone+0x5e() in libc.so.6 (0x00000000) 0x7efe2a57: pop %es My question, is this a bug, and, if so, what should be done to correct it? Is it sufficient to just make raise_segv_exception return if context->Eflags indicate VM86 instead of calling NtSetContextThread ? In my example (Realms of Arkania: Blade of Destiny) this lets the app continue and I get the initial text box to select difficulty level. But ... the process doesn't handle any input - events are queued but never handled p.e. by DOSVM_Wait or similar ... Unfortunately I don't know much about this stuff ... what procedure is responsible to handle these asynchronous events and when ?

2 3

Re: Removed as super maintainer
by darckness Dec. 13, 2006

Dec. 13, 2006

I've been removed from super maintainer and maintainer of World of Warcraft and I know I've been on within the last month even if I haven't changed anything.

1 0

Re: kernel32: Add a couple of weird PE images
by Michael Jung Dec. 13, 2006

Dec. 13, 2006

On Wednesday 13 December 2006 18:00, Dmitry Timoshkov wrote: > +static BOOL is_power_of_two(DWORD alignment) > +{ > + if (!alignment) return FALSE; > + > + while (!(alignment & 1)) > + { > + alignment >>= 1; > + } > + return alignment == 1; > +} Although speed should not be an issue here, a faster way to test for a power of two is: alignment && !(alignment & (alignment-1)) Bye, -- Michael Jung mjung(a)iss.tu-darmstadt.de

1 0

Removed as super maintainer
by Alexander Nicolaysen Sørnes Dec. 13, 2006

Dec. 13, 2006

Hello, It seems like I have been removed from all the applications I was previously a super maintainer of. All my non-super maintainerships seem to be unaffected. Does anyone know the reason? Is it related to the recent server problems? Regards, Alexander N. Sørnes AppDB Administrator

5 5

Re: kernel32: Allow the prefered language to be different from the prefered locale on Mac OS X. [take3]
by Ken Thomases Dec. 13, 2006

Dec. 13, 2006

Wait, wait... On Dec 12, 2006, at 6:55 AM, Pierre d'Herbemont wrote: > This time with Alexandre suggestion to use directly parse_locale_name. > > We still leave the possibility not to use the System preferences > language if LC_MESSAGES env variable LC_MESSAGES was set to a value > different than the default ("C"). > > Pierre. > --- > dlls/kernel32/locale.c | 37 ++++++++++++++++++++++++++++++++++++- > 1 files changed, 36 insertions(+), 1 deletions(-) > diff --git a/dlls/kernel32/locale.c b/dlls/kernel32/locale.c > index 8792623..c70e08c 100644 > --- a/dlls/kernel32/locale.c > +++ b/dlls/kernel32/locale.c > @@ -2832,11 +2839,39 @@ void LOCALE_Init(void) > unix_cp = CP_UTF8; /* default to utf-8 even if we don't get a > valid locale */ > setenv( "LANG", user_locale, 0 ); > TRACE( "setting locale to '%s'\n", user_locale ); > -#endif > + > + /* We still want to set the retrieve the prefered language as > choosen in > + System Preferences.app, because it can differ from > CFLocaleCopyCurrent(). > + */ > + all_locales = CFLocaleCopyAvailableLocaleIdentifiers(); > + preferred_locales = CFBundleCopyLocalizationsForPreferences > ( all_locales, NULL ); I thought I explained that exactly the above combination of calls produces unusable results. For example, for me it produces "en_ZW" -- English (Zimbabwe). In System Preferences, I have set "English" (no country) as my preferred language and U.S. for other locale settings. In other words, it is picking a random country. > + if (preferred_locales) > + { > + if (CFArrayGetCount( preferred_locales )) > + user_language_string_ref = CFArrayGetValueAtIndex > ( preferred_locales, 0 ); > + CFRelease( preferred_locales ); > + } > + CFRelease( all_locales ); > +#endif /* __APPLE__ */ > + > setlocale( LC_ALL, "" ); > > unix_cp = setup_unix_locales(); > if (!lcid_LC_MESSAGES) lcid_LC_MESSAGES = lcid_LC_CTYPE; > +#ifdef __APPLE__ > + /* Override lcid_LC_MESSAGES with user_language if LC_MESSAGES > is set to default */ > + if (lcid_LC_MESSAGES == lcid_LC_CTYPE && > user_language_string_ref) > + { > + struct locale_name locale_name; > + WCHAR buffer[128]; > + CFStringGetCString( user_language_string_ref, user_locale, > sizeof(user_locale), kCFStringEncodingUTF8 ); > + strcpynAtoW( buffer, user_locale, sizeof(buffer)/sizeof > (WCHAR) ); > + parse_locale_name( buffer, &locale_name ); > + lcid_LC_MESSAGES = locale_name.lcid; > + TRACE( "setting lcid_LC_MESSAGES to '%s'\n", user_locale ); > + } > +#endif > + > NtSetDefaultUILanguage( LANGIDFROMLCID(lcid_LC_MESSAGES) ); > NtSetDefaultLocale( TRUE, lcid_LC_MESSAGES ); > NtSetDefaultLocale( FALSE, lcid_LC_CTYPE ); > >

2 2

Re: kernel32: suppress an unneeded warning
by Juan Lang Dec. 13, 2006

Dec. 13, 2006

Please don't apply, it disallows integers preceded by whitespace, which I don't think is correct. Thanks, --Juan http://www.winehq.org/pipermail/wine-patches/2006-December/033747.html ____________________________________________________________________________________ Do you Yahoo!? Everyone is raving about the all-new Yahoo! Mail beta. http://new.mail.yahoo.com

1 0

Re: Wine HQ server.
by Scott Ritchie Dec. 13, 2006

Dec. 13, 2006

On Tue, 2006-12-12 at 11:09 -0700, Tony Lambregts wrote: > On 12/12/06, Scott Ritchie <scott(a)open-vote.org> wrote: > > On Mon, 2006-12-11 at 19:27 -0700, Brian Vincent wrote: > > > > I don't know; that strikes me as overkill. We do have rack space > > > > for more servers, so we certainly could drop more in; if someone > > > > wants to buy 1U servers, we can put those in. > > > > > > Looking at the traffic reports, it seems like the WineHQ server is > > > serving A LOT more than it ever has. We're at 300GB a month compared > > > to 200GB a year ago at this time. And I seem to recall a year before > > > that we were a lot lower. The only thing that looks simple to offload > > > is the Changelog, which accounts for 16GB of traffic a month. > > > > > Once again, I'd like to offer up the server we already have hosted by > > budgetdedicated.com. Currently it just hosts the Ubuntu APT repository, > > but it's quite capable of doing a LOT more. > > > > In November, it served up 1.5 TB of data, at a rate maxing out around > > 3mb per second on release days. > > > > I've already asked, and they have no problem with us offloading some of > > our meager site onto it :) > > > > Thanks, > > Scott Ritchie > > > > The main site uses php a lot, is that available? Yeah, just about anything is. We have our own (virtual) server, currently running Ubuntu Dapper. So far it's got an uptime of like 2 months, only having to reboot because it was moved. Currently only I have the passwords. > > Having our main site (www.winehq.org) (mirrored or hosted) at a > separate site poses a problem of what to do for the mail archives IE: > http://www.winehq.org/pipermail/wine-announce/ (not that I find them > as useful as before) I think that we could address this issue if > changed the hosting of the archive to be http://forums.winehq.org (or > something similar) like we do with git and cvs. > That could be done, though I'm tempted to think we should have the current winehq server be the one with all the stuff other than the main site (where most of the traffic is), and put the main site at budgetdedicated. > I would like to see mirrored solution myself where we have two (or > more) sites that respond to www.winehq.org and this seems to be an > opportunity to do it. > > -- That could work - setting up a static mirror of winehq shouldn't be too hard, should it? Thanks, Scott Ritchie

1 0

re: Delphi Toolbar Icons not displayed correctly
by Dan Kegel Dec. 12, 2006

Dec. 12, 2006

viny wrote: >Here http://home.scarlet.be/linux/compteco/test.zip , a little test >program (with delphi 5 sources) that demonstrate a wrong display of the >toolbar. With native comctl32, it's OK, but not with the builtin one. >I would like to run this program >(http://home.scarlet.be/linux/compteco/index.html) without native DLL. I filed http://bugs.winehq.org/show_bug.cgi?id=6879 for you. Can you add the missing details there, e.g. which version of wine? - Dan -- Wine for Windows ISVs: http://kegel.com/wine/isv

1 0

Wine HQ server.
by Tony Lambregts Dec. 12, 2006

Dec. 12, 2006

What would it take (how much would it cost us) to move the www.winehq.org site to another server? We have had this problem for a while now (over a week) that our main site has been sporaticly unavailable. I appreciate all the work Jeremy Newman has put into trying to find the problem but I think that having all of websites (excepting the wiki) on one box is a bad idea in the long run. I think that we should dedicate a box to www.winehq.org since having a uninterupted presence on the web is critical to Wine's image. Right now we host quite a number of different services on the one box. Ideally we could have a redundant/mirrored solution but that is not esential at this point. This is my list in order of importance (wiki is not on this list because it is already on a separate box) Git - Essential service (easy to miror) www.winehq.org - critical to our image (easy to mirror ???) mysql - critical to appdb and bugzilla (cannot mirror effectively) bugs.winehq.org - important to fixing bugs ;^) (hard to mirror) appdb.winehq.org - userland mostly. ( yes it's important but not critical) (hard to miror) mailing lists archives - we already have redundancy here. cvs - needed for web site development and people not converted to git (easy to mirror) Buying an used box for $500-1000(us) and dedicating it to www.winehq.org is one option I see (and am willing to help pay for). I am sure there are other options but I wanted to at least open the dicussion. -- Tony Lambregts

4 4

Re: mshtml #1: Added alert implementation. (resend)
by Alexandre Julliard Dec. 12, 2006

Dec. 12, 2006

Jacek Caban <jacek(a)codeweavers.com> writes: > --- > dlls/mshtml/htmlwindow.c | 14 ++++++++++++-- > 1 files changed, 12 insertions(+), 2 deletions(-) It doesn't compile: gcc -c -I. -I. -I../../include -I../../include -D__WINESRC__ -DCOM_NO_WINDOWS_H -D_REENTRANT -fPIC -Wall -pipe -fno-strict-aliasing -Wdeclaration-after-statement -Wwrite-strings -Wpointer-arith -g -O2 -o htmlwindow.o htmlwindow.c htmlwindow.c: In function ‘HTMLWindow2_alert’: htmlwindow.c:200: error: ‘IDS_MESSAGE_BOX_TITLE’ undeclared (first use in this function) htmlwindow.c:200: error: (Each undeclared identifier is reported only once htmlwindow.c:200: error: for each function it appears in.) make[2]: *** [htmlwindow.o] Error 1 -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: kernel32: Allow the prefered language to be different from the prefered locale on Mac OS X by correctly setting LC_MESSAGES.
by Ken Thomases Dec. 12, 2006

Dec. 12, 2006

On Dec 11, 2006, at 1:46 PM, Pierre d'Herbemont wrote: > This patch allows Mac OS X Users that set in System Preferences.app > a language (say English) and that don't use the default number and > text formatting currently associated with this language (for > instance if they use French number and text formatting), to have > Wine displaying text in their choosen language. And thus it allows > Wine to behave like 'regular' Mac OS X Applications. > > The only problem is that we have to translate the language id to a > locale id. > > Pierre. > --- > dlls/kernel32/locale.c | 54 +++++++++++++++++++++++++++++++++++++ > ++++++++++- > 1 files changed, 53 insertions(+), 1 deletions(-) > diff --git a/dlls/kernel32/locale.c b/dlls/kernel32/locale.c > index 8792623..345780f 100644 > --- a/dlls/kernel32/locale.c > +++ b/dlls/kernel32/locale.c > @@ -2832,7 +2841,50 @@ void LOCALE_Init(void) > unix_cp = CP_UTF8; /* default to utf-8 even if we don't get a > valid locale */ > setenv( "LANG", user_locale, 0 ); > TRACE( "setting locale to '%s'\n", user_locale ); > -#endif > + > + /* We still want to set the LC_MESSAGES env to what is the > prefered language in > + System Preferences.app, because it can differ from > CFLocaleCopyCurrent(). > + > + However this prefered language is stored using the general > language denotation > + that may not include the country code (en,fr,en-GB,...) > which is not accepted > + by setlocale (setlocale would expect en_US,fr_FR,en_GB,...). > + > + So we retrieve possible locales from /usr/share/locale and > intersect them > + with the prefered languages using > CFBundleCopyLocalizationsForPreferences. */ > + dir = opendir("/usr/share/locale"); > + available_locales = CFArrayCreateMutable(kCFAllocatorDefault, > 0, NULL); > + while ((file = readdir(dir))) > + { > + /* This filters the 'right' locales (xx_xx.UTF-8) */ > + if (strstr(file->d_name, ".UTF-8")) > + CFArrayAppendValue(available_locales, (void*) > CFStringCreateWithCString(kCFAllocatorDefault, > + file- > >d_name, kCFStringEncodingUTF8)); The above should use CFStringCreateWithFileSystemRepresentation instead of CFStringCreateWithCString. Also, you need to CFRelease the created string after adding it to the array. (The array manages its references, but that doesn't relieve you of the responsibility to manage your own.) > + } > + closedir(dir); I think the above code to scan /usr/share/locale is unnecessary. Doesn't CFLocaleCopyAvailableLocaleIdentifiers provide the equivalent? > + > + CFPreferencesAppSynchronize(kCFPreferencesAnyApplication); > + prefered_languages = CFPreferencesCopyValue( CFSTR > ("AppleLanguages"), kCFPreferencesAnyApplication, > + > kCFPreferencesCurrentUser, kCFPreferencesAnyHost); > + if(prefered_languages) > + { > + CFArrayRef intersected_locales = > CFBundleCopyLocalizationsForPreferences(available_locales, > prefered_languages); Do you need prefered_languages? The docs say that if you pass NULL for the second parameter of CFBundleCopyLocalizationsForPreferences, it uses the user's preferences. That would avoid you having to look them up, above. Also, "preferred" is spelled with three R's. > + CFStringRef user_language = NULL; > + > + if(CFArrayGetCount(intersected_locales)) > + user_language = CFArrayGetValueAtIndex > (intersected_locales, 0); > + if(user_language) > + { > + CFStringGetCString( user_language, user_locale, sizeof > (user_locale), kCFStringEncodingUTF8 ); > + CFRelease( user_language ); > + TRACE( "setting LC_MESSAGES to '%s'\n", user_locale ); > + setenv( "LC_MESSAGES", user_locale, 0 ); > + } > + CFRelease(prefered_languages); > + CFRelease(intersected_locales); > + } > + CFRelease(available_locales); > +#endif /* __APPLE__ */ > + > setlocale( LC_ALL, "" ); > > unix_cp = setup_unix_locales(); > >

3 4

Delphi Toolbar Icons not displayed correctly
by viny Dec. 12, 2006

Dec. 12, 2006

Here http://home.scarlet.be/linux/compteco/test.zip , a little test program (with delphi 5 sources) that demonstrate a wrong display of the toolbar. With native comctl32, it's OK, but not with the builtin one. I would like to run this program (http://home.scarlet.be/linux/compteco/index.html) without native DLL. Thanks.

1 0

Re: comctl32: fix imagelist regression from "comctl32: Remove Nx1 assumptions in ImageList_Read"
by Peter Oberndorfer Dec. 11, 2006

Dec. 11, 2006

On Sunday 10 December 2006 16:57, Peter Oberndorfer wrote: > The patch comctl32: Remove Nx1 assumptions in ImageList_Read (e74b0ee9102470d1b866e94c54527b34806cf876) > would produce black images on the toolbar of IDA Pro Demo after clicking one of the listviews on the right > > > Changelog: > comctl32: fix imagelist regression from "comctl32: Remove Nx1 assumptions in ImageList_Read" > remove unused HBITMAP hbmColor > It still leaks memory, when it fails to read the bitmap or the mask. I'll try to write a testcase to see what happens if one of them is missing. Greetings Peter

1 0

Re: fonts:add "framework" for tahoma font
by Dmitry Timoshkov Dec. 11, 2006

Dec. 11, 2006

Louis. Lenders <xerox_xerox2000(a)yahoo.co.uk> wrote: > Hi, as there hasn't been any sign of life from someone who's doing > this now or in near future, and as no-one seems to know how exactly > this font could be added in a legal way, i decided to take the gauntlet > and give it a try. You seem to not be aware of Greenville, another (much more complete and compatible) Tahoma replacement. -- Dmitry.

3 3

Re: mscoree: Take advantage of an installed Mono for Windows to run .NET applications.
by Alexandre Julliard Dec. 11, 2006

Dec. 11, 2006

Hans Leidekker <hans(a)it.vu.nl> writes: > + memset(&si, 0, sizeof(si)); > + si.cb = sizeof(si); > + if (!CreateProcessW(NULL, cmd_line, NULL, NULL, FALSE, 0, NULL, NULL, &si, &pi)) > + { > + HeapFree(GetProcessHeap(), 0, cmd_line); > + return -1; > + } > + > + if (WAIT_OBJECT_0 != WaitForSingleObject(pi.hProcess, 10000)) > + WARN("Timed out waiting for command to start\n"); This waits for the process to end. If you really want to wait for it to start you have to use something like WaitForInputIdle. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: DDraw: Protect IDirectDrawImpl against race conditions
by Alexandre Julliard Dec. 11, 2006

Dec. 11, 2006

Stefan Dösinger <stefandoesinger(a)gmx.at> writes: > @@ -356,6 +373,8 @@ IDirectDrawImpl_SetupFullscreenWindow(ID > * Helper function that restores a windows' properties when taking it out > * of fullscreen mode > * > + * Assumes that the critical section is properly held > + * This seems dangerous, changing the window size will potentially send messages to other threads, so doing it inside the lock can cause deadlocks. -- Alexandre Julliard julliard(a)winehq.org

2 3

Re: OpenAL32.dll thunk and OpenAL winmm driver
by Stefan Dösinger Dec. 11, 2006

Dec. 11, 2006

Am Sonntag 26 November 2006 00:25 schrieb Nick Burns: > dlls/openal32/openal.c: main thunk file > dlls/openal32/Makefile.in: simple thunk makefile > dlls/openal32/openal32.spec: openal+alut spec file > dlls/winmm/wineopenal/audio.c: main winmm driver file > dlls/winmm/wineopenal/Makefile.in: simple winmm makefile > dlls/winmm/wineopenal/openal.c: simple interface > dlls/winmm/wineopenal/openal.h: simple header for simple interface > dlls/winmm/wineopenal/wineopenal.drv.spec: winmm spec file I think you should send the thunk and winmm driver in 2 seperate patches

11 18

Hoiga salaray software
by Mattias Eriksson Dec. 11, 2006

Dec. 11, 2006

Hi, I'm a user of Hogia salary software under wine. It works fine with wine 0.9.11 but something broke in 0.9.12. It has gotten better with the latest versions but it still has a bug that makes it not work. I have been in contact with the Hogia developers and they are willing to provide a demo version of the software for you to look at but they need some information about who they should contact? So if you just mail me some contact information and I'll forward them to the Hogia developers. //Mattias

3 2

Yet another benchmark: Sandra
by Dan Kegel Dec. 11, 2006

Dec. 11, 2006

While looking around for windows benchmarks, I ran into http://www.mysimon.com/4520-6603_8-5020816-1.html which mentions a benchmark called Sandra. Sandra Lite is a free download available from http://www.sisoftware.net/?dir=dload&location=sware_dl_all There are two versions available for download: the one for win2k and up (done in 2006), and the one for win98 (done in 2004). The win2k one installs, but fails silently on startup. Wine says: ~/.wine/drive_c/Program Files/SiSoftware/SiSoftware Sandra Lite XIb$ ~/wine-git/wine sandra.exe fixme:actctx:FindActCtxSectionStringW 00000000 (null) 2 L"msvcr80.dll" 0x337acc fixme:actctx:CreateActCtxW 0x33fa30 00000088 ... fixme:wintrust:WinVerifyTrust 0xffffffff {00aac56b-cd44-11d0-8cc2-00c04fc295ee} 0x33edc0 fixme:crypt:CryptVerifyMessageSignature stub: 0x33ede0, 0, 0x6e81d8, 4296, (nil), 0x33edc4, 0x33edc0 fixme:actctx:ReleaseActCtx 0xf00baa ... and exits. The win98 one wants mdac 2.5 installed before installing, and fails to run because winscard.dll is not present. :-( So I guess Sanda doesn't like wine much at the moment. - Dan -- Wine for Windows ISVs: http://kegel.com/wine/isv

1 0

old benchmark #2: business winstone?
by Dan Kegel Dec. 11, 2006

Dec. 11, 2006

There's an update to business winstone online at ftp://ftp.pcmag.com/benchmarks/bwinstone, but according to http://web.archive.org/web/20041024235010/www.etestinglabs.com/benchmarks/b… the full benchmark was only available on cd-rom. Anyone have a copy of that around still? Thanks, Dan

1 0

winbench '99?
by Dan Kegel Dec. 11, 2006

Dec. 11, 2006

I noticed that, way back six years ago, there was interest in getting winbench '99 running on wine: http://www.kernel-traffic.com/wine/wn20000214_30.html#3 Winbench '99 is still available as a free download from ftp://ftp.pcmag.com/benchmarks/winbench (check out the cool Y2K bug in the directory listing there). I tried it just now with current wine, and it installs ok, but when you start it, it aborts with the error Create system info object failed and wine shows ~/.wine/drive_c/ZDBENCH/UI32$ ~/wine-git/wine ZDBUI32.exe err:ole:CoGetClassObject class {18483923-9673-11cf-ad24-149efa000000} not registered err:ole:CoGetClassObject no class object {18483923-9673-11cf-ad24-149efa000000} could be created for context 0x1 Anyone know what's up with that, or whether this worked with Wine at one time? - Dan p.s. I'd have opened up a report in Bugzilla, but bugs.google.com doesn't seem to be listening on port 80 at the moment.

1 0

Re: user32/winex11: Move CursorClip functionality from user to winex11.
by Pierre d'Herbemont Dec. 10, 2006

Dec. 10, 2006

Vitaliy, Would it be possible to keep that code in user? You could probably compute virtual_screen_rect from GetMonitorInfo() function? I am trying to get as much driver independant code from winex11.drv to user32, so that winequartz.drv won't have to duplicate too much winex11.drv code. Pierre.

1 0

How to deal with render targets in wined3d
by Stefan Dösinger Dec. 10, 2006

Dec. 10, 2006

Hi, The render target support in wined3d is a bit jerky. Basically we're dealing with 2 sorts of render targets: Offscreen textures and onscreen swapchains. Onscreen swapchains have their own drawable and glx context right now, for offscreen textures have that too, except if fbos are used. The seperate context makes a problem. Due to that the render target change has to be performed before any other opengl states are set for drawprim, and it requires keeping track of multiple sets of gl states. The question is, do we need a seperate context? Primary swapchain: essentially, no Back buffer offscreen rendering: not really PBuffer offscreen rendering: Yes? No? FBO offscreen rendering: No Secondary swapchain: Yes? No? Technically a swapchain and offscreen render target would match a glx drawable. However, can we switch the drawable, but keep the context? Can we switch the drawable that the context is used on? What is better performance-wise? Switching a context's drawable, or switching context and drawable and applying all states that changed since the last use of that context? The main thing I'm concerned about are driver bugs. I know that the dri drivers keep track of per-drawable settings per context, what might other drivers do? Another question is what do do about pbuffers. Pbuffers are pretty nasty. Are there any cards out there which have a working pbuffer support, but no fbo support? If we slightly modify the back buffer rendering we can render on auxiliary buffers instead of the pbuffer(If there are any cards out there supporting that). My new state management can handle multiple contexts per device, it has to do that for multithreading. However, if we can do the render target stuff with only one context things will get MUCH easier. Stefan

1 0

Re: WGL: pixel formats
by Tomas Carnecky Dec. 10, 2006

Dec. 10, 2006

Roderick Colenbrander wrote: > Hi, > > This patch makes the WGL code independant of the number of supported pixelformats. At the moment only a single format is supported but this will be extended with a few other 'compatible' formats to provide more opengl functionality like multisampling support. > Please don't apply this patch: DescribePixelFormat(hDC, 0, 0, NULL) should succeed and return the number of available pixel formats - WoW expects this behavior. But someone should probably write a test for that. tom

1 0

winehq.org flakiness?
by Dan Kegel Dec. 10, 2006

Dec. 10, 2006

What is up with winehq.org? Seems like it's down half the time lately. Now, for instance.

2 1

file/directory changes not working with openSUSE 10.2/ kernel 2.6.18?
by Marcus Meissner Dec. 10, 2006

Dec. 10, 2006

Hi, While running make check on the (64bit) kernel 2.6.18 supplied with openSUSE 10.2, I get failures in the kernel/tests/change and ntdll/tests/change tests: ../../../tools/runtest -q -P wine -M kernel32.dll -T ../../.. -p kernel32_test.exe.so change.c && touch change.ok change.c:272: Test failed: failed to create directory change.c:284: Test failed: failed to create subdir change.c:287: Test failed: should be ready change.c:290: Test failed: should be ready change.c:299: Test failed: failed to remove subdir change.c:302: Test failed: should be ready change.c:305: Test failed: should be ready change.c:317: Test failed: failed to remove dir change.c:415: Test failed: ov.Internal wrong (res 259) change.c:416: Test failed: ov.InternalHigh wrong change.c:420: Test failed: action wrong change.c:421: Test failed: len wrong change.c:422: Test failed: name wrong change.c:457: Test failed: ov.Internal wrong change.c:458: Test failed: ov.InternalHigh wrong change.c:469: Test failed: action wrong change.c:470: Test failed: len wrong change.c:471: Test failed: name wrong change.c:483: Test failed: ov.Internal wrong change.c:484: Test failed: ov.InternalHigh wrong change.c:494: Test failed: should be ready change.c:496: Test failed: ov.Internal wrong change.c:497: Test failed: ov.InternalHigh wrong change.c:501: Test failed: action wrong change.c:502: Test failed: len wrong change.c:503: Test failed: name wrong change.c:521: Test failed: action wrong change.c:522: Test failed: len wrong change.c:523: Test failed: name wrong change.c:525: Test failed: ov.Internal wrong change.c:526: Test failed: ov.InternalHigh wrong change.c:585: Test failed: should timeout change.c:593: Test failed: ov.Internal wrong change.c:612: Test failed: should be ready change.c:614: Test failed: ov.Internal wrong change.c:672: Test failed: should timeout change.c:682: Test failed: event should be ready change.c:684: Test failed: ov.Internal wrong change.c:685: Test failed: ov.InternalHigh wrong change.c:688: Test failed: offset wrong change.c:689: Test failed: action wrong change.c:690: Test failed: len wrong change.c:691: Test failed: name wrong make: *** [change.ok] Fehler 43 I tried setting a breakpoint on inotify_poll_event and inotify_get_poll_events but they do not trigger even once. My suspicion is that the new kernel don't wake the queue correctly. :/ Ciao, Marcus

2 1

Wine performance (opengl)
by nlataill Dec. 10, 2006

Dec. 10, 2006

Hi, I am using wine to play World of warcraft (this game is using opengl, and is threaded - 2/3 threads IIRC). I did recently find a reproductable performance regression on versions 0.9.24+. Under the same conditions (always difficult to get in an MMORPG game, but this is reproductible), version 0.9.23 gives me 20/25% more FPS than versions 0.9.24->0.9.26 (have not tried .27 yet). - Is this performance regression known/expected on opengl/threaded applications (new codebase) ? - Is someone using WOW here (and has found a similar preformance regression) ? - It is worth for me to start looking at it (more measures, throw oprofile at it, bissect the change...) Thanks, Norbert

2 2

Re: d3d: Fix EnumAdapterModes and add an adapter mode format test
by H. Verbeet Dec. 9, 2006

Dec. 9, 2006

On 09/12/06, Chris Robinson <chris.kcat(a)gmail.com> wrote: >+ * Copyright (C) 2006 Henri Verbeet Except for perhaps the creation of the device, there's not a lot of code I wrote in there. You should at least include your own name in there.

1 0

wine-rpm on mandriva2007_x86-64
by Daniele Bagaglini Dec. 9, 2006

Dec. 9, 2006

For installing the latest *wine-rpms* and working with it, on mandriva-2007.0_x86-64, we need to install libxxf86vm1-1.0.1-2mdv2007.0.i586.rpm downloadable on a mandriva2007_x86_32 repository, like this for example: ftp://mirror.pacific.net.au/MandrivaLinux/official/2007.0/i586/media/main/release Regards -- Daniele Bagaglini

1 0

Re: d3d: Fix EnumAdapterModes and add an adapter mode format test
by Stefan Dösinger Dec. 9, 2006

Dec. 9, 2006

Am Samstag 09 Dezember 2006 19:28 schrieb Chris Robinson: > This patch fixes D3D adapter mode enumeration, which clears some issues > some games had when starting (eg. Morrowind being unable to set a mode). > I've added an adapter test (modified from the texture test) to D3D 9 which > tests all the D3DFORMAT types. > > Apologies for the macro (a)buse in the test, but it would've been more > combersome to have to manually make strings for each of the modes without > it. Do not apply this patch yet. The WINED3DFMT_X1R5G5B5 test fails on my brothers windows xp system. Win2k system also seem to have sporadic test failures.

1 0

re: First case of a real virus infection on wine
by Dan Kegel Dec. 9, 2006

Dec. 9, 2006

Stefan wrote: >What we noticed pretty fast was that at least 3 windows boxes were infested >with a bunch of malware, and trying to infect each other over network shares. >The virus alert messages popping up made playing impossible. The obvious >solution: Disable the virus scanners. I think that qualifies you for the Darwin award! Can you try Christoph's ClamAV integration, http://www.christoph-probst.com/soc2006/wine/ and see if it would have prevented the infection? - Dan

3 2

First case of a real virus infection on wine
by Stefan Dösinger Dec. 9, 2006

Dec. 9, 2006

Hi, Seems like I just got the victim of the first infection of a windows virus on wine :-( . I got infected with some virus called W32.Parite.B. Unlike most of the biests out there it is a real virus that attaches itself to existing .exe files, not a stand-alone worm. I caught it yesterday on a small private gaming session. I didn't have Warcraft 3 installed, so instead of installing and messing with patches I would have had to download I decided to copy an existing installation over. That seemed to work fine at first :-/ What we noticed pretty fast was that at least 3 windows boxes were infested with a bunch of malware, and trying to infect each other over network shares. The virus alert messages popping up made playing impossible. The obvious solution: Disable the virus scanners. So with the protection disabled those windows boxes were able to play, everything seemed to work fine. I noticed that something was wrong when my Battlefield 1942 crashed, which worked a few hours before when tested for regression in my new patches. ClamAV showed up a W32.Parite.B infection in bf1942.exe. That virus wasn't only in bf1942.exe, I found it in the war3 installation I copied over too. Looks like it came from there. I found it in all .exe files on my fake C:\ drive, except of our fake .exe's in C:\windows\system32. Looks like it didn't like those. Next I had a look at my real windows installation mounted in /media/windows, and found it infected too. Well, that was easy to clean up with a mkfs because I didn't have anything valueable in there. Well, the thing that is rather bad is that it infected my downloaded game demos and other files on my home drive. Luckily I didn't have my external hard drive with the rest of my stuff attached when I ran wine. So I've now deleted my wine installation, windows installation and all .exe files on my disk. I'm scanning my whole linux drives to be sure, but I didn't run wine as root and I'm confident that the Linux file security prevented the worst problems. To summarize, I got into the trouble mainly because I ignored the basic security guidelines. I ran executables from a really not trustworthy source, knowing that my friends' windows boxes are in a bad shape quite often. My real windows installation got infected because I had it world writeable for no real reason. On the bright side, running as a least priviledged user prevented the worst problems. If anyone wants to play around with that virus, I kept my infected GenuineCheck.exe :-) Stefan

1 0

Re: appdb/include version.php
by Tony Lambregts Dec. 9, 2006

Dec. 9, 2006

WineHQ wrote: > ChangeSet ID: 30433 > CVSROOT: /opt/cvs-commit > Module name: appdb > Changes by: wineowner(a)winehq.org 2006/12/08 22:28:20 > > Modified files: > include : version.php > > Log message: > Alexander Nicolaysen Sørnes <alex(a)thehandofagony.com> > Colourize the maintainer's rating and version fields with the appropriate > rating colours, both in the application and version view. > This should make it easier for users to identify working versions, and it also > prevents the maintainer's rating from being 'overshadowed' by the test > results. > > Patch: http://cvs.winehq.org/patch.py?id=30433 > > Old revision New revision Changes Path > 1.86 1.87 +12 -4 appdb/include/version.php > > Index: appdb/include/version.php > diff -u -p appdb/include/version.php:1.86 appdb/include/version.php:1.87 > --- appdb/include/version.php:1.86 9 Dec 2006 4:28:20 -0000 > +++ appdb/include/version.php 9 Dec 2006 4:28:20 -0000 > @@ -662,9 +662,14 @@ class Version { > echo " </td></tr>\n"; > } > > + if($this->sTestedRating != "/" && $this->sTestedRating) > + $sMaintainerColor = $this->sTestedRating; > + else > + $sMaintainerColor = "color0"; > + > // rating Area > - echo "<tr class=\"color1\" valign=\"top\"><td>Maintainer Rating</td><td>".$this->sTestedRating."</td></tr>\n"; > - echo "<tr class=\"color0\" valign=\"top\"><td>Maintainers Version</td><td>".$this->sTestedRelease."</td></tr>\n"; > + echo "<tr class=\"$sMaintainerColor\" valign=\"top\"><td>Maintainer’s Rating</td><td>".$this->sTestedRating."</td></tr>\n"; > + echo "<tr class=\"$sMaintainerColor\" valign=\"top\"><td>Maintainer’s Version</td><td>".$this->sTestedRelease."</td></tr>\n"; > > // image > $img = Screenshot::get_random_screenshot_img($oApp->iAppId, $this->iVersionId, false); > @@ -923,12 +928,15 @@ class Version { > // set row color > $bgcolor = ($c % 2 == 0) ? "color0" : "color1"; We need to initialize aRatingColor like this... $sRatingColor = "class=\$bgcolor\""; > > + if($oVersion->sTestedRating && $oVersion->sTestedRating != "/") > + $sRatingColor = "class=\"$oVersion->sTestedRating\""; > + or it needs an else like so... else $sRatingColor = "class=\$bgcolor\""; Otherwise we get this problem. http://appdb.winehq.org/appview.php?iAppId=8 > //display row > echo "<tr class=$bgcolor>\n"; > echo " <td><a href=\"".BASE."appview.php?iVersionId=".$iVersionId."\">".$oVersion->sName."</a></td>\n"; > echo " <td>".util_trim_description($oVersion->sDescription)."</td>\n"; > - echo " <td align=center>".$oVersion->sTestedRating."</td>\n"; > - echo " <td align=center>".$oVersion->sTestedRelease."</td>\n"; > + echo " <td $sRatingColor align=center>".$oVersion->sTestedRating."</td>\n"; > + echo " <td $sRatingColor align=center>".$oVersion->sTestedRelease."</td>\n"; > echo " <td align=center>".Comment::get_comment_count_for_versionid($oVersion->iVersionId)."</td>\n"; > echo "</tr>\n\n"; > > > >

1 0

XRenderFindDisplay crash / user32 winex11 initialisation order problem
by Peter Oberndorfer Dec. 8, 2006

Dec. 8, 2006

Hi, today i wanted to create a test for CreateCompatibleDC so i created a exe with this content [1] just to find out it crashes wine inside XRenderFindDisplay [2] The reason is that the CreateCompatibleDC call in gdi executes LoadLibrary winex11.drv which in turn also loads user32.dll, because winex11.drv links to it. Then in DllMain user32 registers some builtin classes which causes it to call into winex11. But at this point DllMain of winex11 had no chance to execute and XRenderQueryExtension gets passed a NULL pointer which makes it unhappy. Unfortunately i have no idea how to properly solve this problem :-( * Use UserClientDllInitialize for initializing user32 stuff as the comment above DllMain mentions it? * I tried linking gdi32 to user32(windows seems to do that) and inserted a dummy IsCharAlphaA('c'); to make sure it is *really* linked :-) And it does not crash anymore, but i don't know if i break any assumptions with that. Also i think it is preferable to defer loading the driver until it is really needed. If you have a better idea of how to fix this please tell me. Greetings Peter 1: simplified version, compile with winegcc foo.c -lgdi32 #include <windows.h> int main(int argc, char ** argv){ CreateCompatibleDC(NULL); } 2: see attachment

1 0

Re: advapi32: Fix comparisons of unsigned expressions with zero
by Mike McCormack Dec. 8, 2006

Dec. 8, 2006

Andrew Talbot wrote: > - if( n < 0 ) > + if( (LONG)n < 0 ) Why not change the type of n to LONG rather than using a cast? Mike

2 1

mshtml as html parser
by Kirill K. Smirnov Dec. 8, 2006

Dec. 8, 2006

Hi, all! Just a little question: can wine's current implementation of mshtml.dll be used as html parser? I need to convert html code into a tree without displaying it. I tried to look through the code, but I'm not familiar with OLE. BTW some E_NOIMPL confused me. Thanks in advance

3 2

Re: mshtml #1: Added get_parentWindow implementation.
by Detlef Riekenberg Dec. 8, 2006

Dec. 8, 2006

Hi. I have no Idea about COM, but while i was reading your Patch to learn, I do not understand this pair: > + > +#define HTMLWINDOW2_THIS(iface) DEFINE_THIS(HTMLWindow, HTMLWindow2, > iface) > + > +#undef HTMLWINDOW_THIS > + Did you mean +#undef HTMLWINDOW2_THIS > here? -- By by ... Detlef

2 1

audio glitch patch by mike hearn
by kfs1＠online.no Dec. 8, 2006

Dec. 8, 2006

I'm trying to implement this patch: http://wiki.winehq.org/Implement_SetThreadPriority by Mike Hearn which will enable wine to work without all the nasty audio glitches when using alsa. However hunk 1 fails on configure.ac as the patch hasn't been updated to latest wine. Can anyone help me get this baby compiled as i want to play my games without the nasty audio glitches and don't care if i have to be root/sudo/whatever. If there are any details (does wineserver/wine* has to be setuid root? do i need any special settings in winecfg, will it work with sudo or do i have to be root,etc etc) I would like to get informed of them also as it will probably take me days to figure it out on my own <:) some guy rediffed(whatever that means) it here: http://bbs.archlinux.org/viewtopic.php?t=27529&highlight= but it doesn't seem to work. Thanks for all the help. I'm asking here as Mike is too busy to fix it, but he says it should be an easy fix. -- Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

8 16

Re: [doc]: Fix system32, winecfg and Printers
by Kai Blin Dec. 8, 2006

Dec. 8, 2006

On Friday 08 December 2006 01:03, Detlef Riekenberg wrote: > - <para> > - The next few settings primarily affect games and are somewhat > + The first few settingsare primarily affect games and are You missed a space there. > somewhat self-explanatory. You can prevent the mouse from leaving the > window of a DirectX program (i.e. a game.) and the default is Cheers, Kai -- Kai Blin, <kai Dot blin At gmail Dot com> WorldForge developer http://www.worldforge.org/ Wine developer http://wiki.winehq.org/KaiBlin/ -- Will code for cotton.

1 0

Re: appdb/include application.php
by Tony Lambregts Dec. 8, 2006

Dec. 8, 2006

WineHQ wrote: > ChangeSet ID: 30409 > CVSROOT: /opt/cvs-commit > Module name: appdb > Changes by: wineowner(a)winehq.org 2006/12/07 21:38:27 > > Modified files: > include : application.php > > Log message: > Alexander Nicolaysen Sørnes <alex(a)thehandofagony.com> > Use $this instead of $oApp since this method is referring to its class > > Patch: http://cvs.winehq.org/patch.py?id=30409 > > Old revision New revision Changes Path > 1.76 1.77 +2 -2 appdb/include/application.php > > Index: appdb/include/application.php > diff -u -p appdb/include/application.php:1.76 appdb/include/application.php:1.77 > --- appdb/include/application.php:1.76 8 Dec 2006 3:38:27 -0000 > +++ appdb/include/application.php 8 Dec 2006 3:38:27 -0000 > @@ -345,12 +345,12 @@ class Application { > { > case "add": > $sSubject = "Submitted application accepted"; > - $sMsg = "The application you submitted (".$oApp->sName." ".$this->sName.") has been accepted by ".$_SESSION['current']->sRealname."."; > + $sMsg = "The application you submitted (".$this->sName." ".$this->sName.") has been accepted by ".$_SESSION['current']->sRealname."."; This is not right. this will print the application name twice. it should be + $sMsg = "The application you submitted (".$this->sName.") has been accepted by .$_SESSION['current']->sRealname."."; > $sMsg .= "Administrators Responce:\n"; > break; > case "reject": > $sSubject = "Submitted application rejected"; > - $sMsg = "The application you submitted (".$oApp->sName." ".$this->sName.") has been rejected by ".$_SESSION['current']->sRealname."."; > + $sMsg = "The application you submitted (".$this->sName." ".$this->sName.") has been rejected by ".$_SESSION['current']->sRealname."."; same thing here > $sMsg .= "Clicking on the link in this email will allow you to modify and resubmit the application. "; > $sMsg .= "A link to your queue of applications and versions will also show up on the left hand side of the Appdb site once you have logged in. "; > $sMsg .= APPDB_ROOT."appsubmit.php?sub=view&apptype=application&appId=".$this->iAppId."\n"; > > >

1 0

Re: mshtml as html parser
by Kirill K. Smirnov Dec. 7, 2006

Dec. 7, 2006

>> Just a little question: can wine's current implementation of mshtml.dll be >> used as html parser? I need to convert html code into a tree without >> displaying it. >You might want to use tidy and an XML pareser : >http://www-128.ibm.com/developerworks/library/x-tiptidy.shtml This way looks too complicated in the following case: I switched off from winhelp in favor of chmhelp - winhelp works somehow, but chmhelp not. chm content is an simple HTML file, which can be parsered by hands (like xCHM utility does), but if there is a library like mshtml (or msxml) why not use it? The only problem is I am afraid wine's mshtml has E_NOTIMPL in piece of codes looks interesting to me. (I do not understand the architcture of mshtml.dll to be sure). To stop wandering I decided to write a question here. So the question: can the wine's mshtml (as it is now) be used? I looked for smth like CreateHTMLDocument2(...) but failed. -- Kirill K. Smirnov

1 0

Change to fix Mac OS X
by Michael Casadevall Dec. 7, 2006

Dec. 7, 2006

I've been tracking Wine for sometime in git, and attempting to build it on Mac OS X just to have repetitive failure. I finally had some time to look at the cause, and figured out that it was failing due to a prototype mismatch in dlls/dbghelp/dbghelp_private.h and dlls/ dbghelp/elf_loader.c. The problem is that the function elf_load_module uses a DWORD in elf_loader.c and an unsigned load in the dbghelp_private.h. I changed the header to a DWORD which fixed the compile right up and allowed it to successfully build Wine, but after talking to vitamin, this change might cause issues with Wine on 64-bit systems. Since I only own 32-bit, and know little about differences when programming 64-bit systems, I thought I should ask before I submit the patch to wine-patches. The full prototype and arguments are: dbghelp_private.h: extern struct module* elf_load_module(struct process* pcs, const char* name, unsigned long) elf_loader.c: struct module* elf_load_module(struct process* pcs, const char* name, DWORD addr) For good measure, I've also attached the patch to this email. Regard, Michael

3 4

Question about allocating memory for a Windows PE file
by Stephen Torri Dec. 7, 2006

Dec. 7, 2006

I am trying to create a custom Windows PE loader (binary & DLL) for the purpose of security research. I am having a difficult time finding how to allocate memory for a binary at the desired address in memory (especially if its non-relocatable). I would like to see why I cannot get memory allocated at the exact address request in the binary headers. Is there a program or system call that allows me to see a list of memory address ranges allocated to the running processes on a system? Stephen

3 2

Re: [Wine] Re: Difference Between WINE and an Emulator
by Edward Savage Dec. 7, 2006

Dec. 7, 2006

>An emulator is a specific type of program that mimics hardware. Windows > is not a piece of hardware, ergo Wine Is Not an Emulator. It's merely a > clone of the Windows API. This is probably the best answer you will get that is clear and simple. Out of interest the term I use to explain to others the concept is that WINE is a compatibility layer between Linux and Windows. Does that seem an accurate use of the word? It tends to set the right impression in the minds of the people I've explained it too.

3 2

Re: winex11.drv: Update cached cursor position after calling hook.
by Alexandre Julliard Dec. 7, 2006

Dec. 7, 2006

Vitaliy Margolen <wine-patches(a)kievinfo.com> writes: > @@ -276,10 +274,12 @@ void X11DRV_send_mouse_input( HWND hwnd, > if ((injected_flags & LLMHF_INJECTED) && > ((flags & MOUSEEVENTF_ABSOLUTE) || x || y)) /* we have to actually move the cursor */ > { > - TRACE( "warping to (%d,%d)\n", pt.x, pt.y ); > + X11DRV_SetCursorPos( pt.x, pt.y ); > + } > + else > + { > wine_tsx11_lock(); > - XWarpPointer( thread_display(), root_window, root_window, 0, 0, 0, 0, > - pt.x - virtual_screen_rect.left, pt.y - virtual_screen_rect.top ); > + cursor_pos = pt; > wine_tsx11_unlock(); > } You need to update the position for button events too. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: ole32: Support reading and writing custom clipformats in the data cache.
by Alexandre Julliard Dec. 7, 2006

Dec. 7, 2006

Robert Shearman <rob(a)codeweavers.com> writes: > --- > dlls/ole32/datacache.c | 106 > +++++++++++++++++++++++++++++++++++++++++------- > 1 files changed, 91 insertions(+), 15 deletions(-) This breaks make test: ../../../tools/runtest -q -P wine -M ole32.dll -T ../../.. -p ole32_test.exe.so ole2.c && touch ole2.ok ole2.c:1257: Test failed: IViewObject_Draw failed with error 0x80040007 ole2.c:1266: Test failed: IViewObject_Draw failed with error 0x80040007 ole2.c:1272: Test failed: IViewObject_Draw failed with error 0x80040007 -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: mshtml #3: Use HTMLWindow::alert in nsPromptService::Alert.
by Alexandre Julliard Dec. 7, 2006

Dec. 7, 2006

Jacek Caban <jacek(a)codeweavers.com> writes: > @@ -39,6 +39,13 @@ WINE_DEFAULT_DEBUG_CHANNEL(mshtml); > > #define HTMLWINDOW2_THIS(iface) DEFINE_THIS(HTMLWindow, HTMLWindow2, iface) > > +typedef struct { > + struct list entry; > + HTMLWindow *window; > +} window_list_t; > + > +static struct list window_list = LIST_INIT(window_list); > + The right way to build a list of windows is to put the list entry directly in the HTMLWindow structure, there's no need for a separate structure. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [PATCH] [WineCfg]: a few fixes for the French translation
by Jonathan Ernst Dec. 7, 2006

Dec. 7, 2006

Le mercredi 06 décembre 2006 à 21:35 +0100, Eric Pouech a écrit : > - remove typos (ortaugraffe) > - set correct size for controls > - used correct space usage (for French) > [...] > - LTEXT "&Nouveau suclassage pour:",IDC_STATIC,16,58,100,8 > + LTEXT "&Nouveau suclassage pour :",IDC_STATIC,16,58,100,8 Could you please fix "suclassage" as well ;-) Thanks. Jonathan

1 0

Re: [PATCH 4/4] ole32: Support reading and writing custom clipformats in the data cache.
by Mike McCormack Dec. 7, 2006

Dec. 7, 2006

Robert Shearman wrote: > + { > + char *format_name = HeapAlloc(GetProcessHeap(), 0, length); > + if (!format_name) > + return E_OUTOFMEMORY; > + GetClipboardFormatNameA(clipformat, format_name, length); > + hr = IStream_Write(stream, format_name, length, NULL); > + } You might want to free format_name at the end of that block. Mike

2 1

Re: [Wine] Cyber Cafe Pro on Linux
by Edward Savage Dec. 6, 2006

Dec. 6, 2006

I'm going to get eaten alive for this though depending on your requirements it might be a better idea to stick with Windows and really lock down your installation. I've been directly involved in two Linux migrations for Internet Cafes (and advised on several others) - one successful (net cafe) and the other unsuccessful (gaming cafe) - and in both cases it was more trouble than it was worth. The problem isn't the software it is the users; When they go to use the software they notice, for example, that they aren't using IE and when they come up to some page that doesn't display perfectly problems occur. There are many more examples that just make the mind hurt. So unless this Cycber Cafe Pro provides _everything_ you really need to be asking yourself if you really want to open this can of worms. On 12/5/06, vulani mabunda <vulani(a)mac.com> wrote: > > Hi, > > I'm running an internet cafe using Cyber Cafe Pro software. This software > is design soley for Windows platforms. We're constantly experiencing > problems with Windows with regards to viruses and stability issues. > Recently, we lost all our data and had to re-install. > > May you please help us in flawlessly porting this application so that it > can run on Linux both on the server and client sides. If this happens, it > could mean an end to our nightmare with Windows! > > We have already tried Open Kiosk (which natively runs on Linux) prior to > making my request. Open Kiosk is free but is not advanced enough to meet our > requirements. For example, it can't allow us to control and charge for > excessive bandwidth. We've done extensive research on what is available and > so far Cyber Cafe Pro seems to meet all our requirements with the exception > of running only on Windows. Hodoman works on Linux but isn't as advanced as > Cyber Cafe Pro is. > > This is the reason why we are interested in trying to get Cyber Cafe Pro > to work on Linux. Cyber Cafe Pro people aren't interested in assisting us > with regards to porting their application to Windows. > > > Regards > > Vulani Mabunda > CEO > Internet CAfe Jack > South Africa > > _______________________________________________ > wine-users mailing list > wine-users(a)winehq.org > http://www.winehq.org/mailman/listinfo/wine-users >

3 3

Putting my hand up for WWN
by Edward Savage Dec. 6, 2006

Dec. 6, 2006

I'm prepared to write the WWN releases in Brian Vincent seeming absence if it pleases the powers that be. If this is alright I'd like some one to give an overview of what topics should be covered and then for some one to review the first one or two created. I've read the last hundred or so WWN issues over the years so I have a good idea of whats required though clarification would be nice. I already got wine-devel/wine-users - are there any more that need to be monitored. Finally I would need to know how to submit the newsletter and the proper syntax. Edward

2 2

Re: [winetest] Make sure we can send larger files
by Dimi Paun Dec. 5, 2006

Dec. 5, 2006

On Thu, November 30, 2006 1:01 pm, Paul Vriens wrote: > I'm hitting the 1MB limit already with my win98 tests. Most other tests > are about to reach that threshold as well. I'd make it 2MB, what's 1MB these days... :) -- Dimi Paun <dimi(a)lattica.com> Lattica, Inc.

3 2

Putting my hand up for WWN
by Edward Savage Dec. 5, 2006

Dec. 5, 2006

I'm prepared to write the WWN releases in Brian Vincent seeming absence if it pleases the powers that be. If this is alright I'd like some one to give an overview of what topics should be covered and then for some one to review the first one or two created. I've read the last hundred or so WWN issues over the years so I have a good idea of whats required though clarification would be nice. I already got wine-devel/wine-users - are there any more that need to be monitored. Finally I would need to know how to submit the newsletter and the proper syntax. Edward

3 2

Re: [PATCH][DbgHelp] Recognize DT_GNU_HASH
by Thomas Fitzsimmons Dec. 5, 2006

Dec. 5, 2006

Thomas Fitzsimmons wrote: > This patch fixes #6830. > > Tom > > 2006-12-05 Thomas Fitzsimmons <fitzsim(a)redhat.com> > > * dlls/dbghelp/elf_module.c (is_dt_flag_valid): Recognize > DT_GNU_HASH. > > > ------------------------------------------------------------------------ > > diff --git a/dlls/dbghelp/elf_module.c b/dlls/dbghelp/elf_module.c > index 85494ca..a7fa2f8 100644 > --- a/dlls/dbghelp/elf_module.c > +++ b/dlls/dbghelp/elf_module.c > @@ -1041,6 +1041,9 @@ #endif > #ifndef DT_EXTRANUM > #define DT_EXTRANUM 0 > #endif > +#ifndef DT_GNU_HASH > +#define DT_GNU_HASH 0x6ffffef5 > +#endif > return (d_tag >= 0 && d_tag < DT_NUM + DT_PROCNUM + DT_EXTRANUM) > #if defined(DT_LOOS) && defined(DT_HIOS) > || (d_tag >= DT_LOOS && d_tag < DT_HIOS) > @@ -1048,6 +1051,9 @@ #endif > #if defined(DT_LOPROC) && defined(DT_HIPROC) > || (d_tag >= DT_LOPROC && d_tag < DT_HIPROC) > #endif > +#if defined(DT_GNU_HASH) > + || (d_tag == DT_GNU_HASH) > +#endif > ; > } Ugh, no need for the extra "if defined" check, of course. Tom

1 0

Re: James Hawkins : msi: Add tests for installing from continuouscabinets.
by Dmitry Timoshkov Dec. 5, 2006

Dec. 5, 2006

"Dmitry Timoshkov" <dmitry(a)codeweavers.com> wrote: > the test should be either disabled completely, or made it run only > if cabinet.dll is detected. Last line should read as: if good/not broken cabinet.dll is detected. -- Dmitry.

1 0

Re: James Hawkins : msi: Add tests for installing from continuouscabinets.
by Dmitry Timoshkov Dec. 5, 2006

Dec. 5, 2006

"Alexandre Julliard" <julliard(a)wine.codeweavers.com> wrote: > Module: wine > Branch: master > Commit: 1f3f88bf67d32983ff00d96bf58bf91c6e72ae02 > URL: http://source.winehq.org/git/wine.git/?a=commit;h=1f3f88bf67d32983ff00d96bf… > > Author: James Hawkins <truiken(a)gmail.com> > Date: Mon Oct 23 18:15:28 2006 -0700 > > msi: Add tests for installing from continuous cabinets. According to http://test.winehq.org/data/200612021000/ this test fails under all NT based platforms. I've ran it under XP SP2 and got the following error messages: install.c:864: Test failed: Expected ERROR_SUCCESS, got 1603 install.c:869: Test failed: File not installed install.c:870: Test failed: File not installed install.c:871: Test failed: File not installed install.c:873: Test failed: File not installed install: 1147 tests executed, 0 marked as todo, 5 failures. and a bunch of .log files in the TEMP directory with the following contents: Error 1334. The file 'maximus' cannot be installed because the file cannot be found in cabinet file 'test1.cab'. This could indicate a network error, an error reading from the CD-ROM, or a problem with this package. Error 1334. The file 'augustus' cannot be installed because the file cannot be found in cabinet file 'test1.cab'. This could indicate a network error, an error reading from the CD-ROM, or a problem with this package. So, either the test is not entirely correct, or just the .cab files it creates are corrupted. -- Dmitry.

2 2

Re: make check_no_exec() work reliable
by Peter Beutner Dec. 4, 2006

Dec. 4, 2006

Peter Beutner wrote: > Before starting to make this whole noexecute override behaviour configurable, > it first must work reliable. In its current form there is no guarantee that > the check_no_exec() function is actually called, because any other installed > exception handler might decide to handle the exception itself. > And as seen by the number of failing applications, this seems to happen > quite a lot. > This patch therefore makes check_no_exec() to be called before any other > exception handler. > > --- Any comments what is wrong with this one?

2 7

Re: [5/6] wined3d: Make CreateFakeGLContext thread safe.
by Alexandre Julliard Dec. 4, 2006

Dec. 4, 2006

Jan Zerebecki <jan.wine(a)zerebecki.de> writes: > This adds thread safety to CreateFakeGLContext and it's > corresponding Release function only. This addresses no thread > safety in the rest of wined3d, which AFAIk as a whole is still > not thread safe. When addressing thread safety in wined3d it must > also be taken care that a GL context is not deleted away by a > different thread when execution is between these two functions. You are calling ENTER_GL while holding your crit section, this will deadlock. -- Alexandre Julliard julliard(a)winehq.org

1 0

Help needed on an SHGetFileInfoW() crash
by Francois Gouget Dec. 4, 2006

Dec. 4, 2006

A friend pointed me to a railway signalling simulation game: http://www.simsig.co.uk/html/downloads.html However, the installer crashes in SHGetFileInfoW(): 000d:Call shell32.SHGetFileInfoA(0045eae8,00000010,0034fcec,00000160,00001010) ret=0045e9f6 000d:trace:seh:raise_exception code=c0000005 flags=0 addr=0x7e8871b2 000d:trace:seh:raise_exception info[0]=00000000 000d:trace:seh:raise_exception info[1]=00000000 More precisely the crash happens on this line: hr = IShellFolder_GetUIObjectOf(psfParent, 0, 1, (LPCITEMIDLIST*)&pidlLast, &IID_IExtractIconW, &uDummy, (LPVOID*)&pei); And the reason is that psfParent is NULL. psfParent is initialized a bit above, but only if SHGetFileInfoW() was given a PIDL and SHGFI_USEFILEATTRIBUTES was not used. This means that whenever a 'real path' or SHGFI_USEFILEATTRIBUTES is used in combination with SHGFI_DISPLAYNAME, SHGFI_ICONLOCATION, SHGFI_ICON or SHGFI_SYSICONINDEX, then SHGetFileInfoW() will crash. However we cannot just create a pidl since the 'real path' will usually refer to a nonexistent file or directory (in our case it is 'c:\directory' so it's clear the installer just wants to know what's the icon for a standard directory). Any idea? -- Francois Gouget <fgouget(a)free.fr> http://fgouget.free.fr/ "Lotto: A tax on people who are bad at math." -- unknown "Windows: Microsoft's tax on computer illiterates." -- WE7U

2 1

Call for DirectPlay captures
by Kai Blin Dec. 4, 2006

Dec. 4, 2006

Hi folks, As you might know, I'm currently toying with figuring out the MS DirectPlay protocol. I've been making some progress with my wireshark dissector over the weekend, and would be interested in some additional games to get more different data on some of this stuff. Here's what I would like you to do (if you want to help out): Fire up a network sniffer that dumps to libpcap format. Grab TCP and UDP packets. Start a game that uses DirectPlay for networking and start a network game. Start the game on another box and connect to the game. Now, in the game lobby, use the chat to say something from both games. Then, change a setting or two and start the game. After a few moments of playing (or more, your pick), leave the game. After both games are closed, you can stop the capture, zip it up and send it to me. Thanks in advance, Kai -- Kai Blin, <kai Dot blin At gmail Dot com> WorldForge developer http://www.worldforge.org/ Wine developer http://wiki.winehq.org/KaiBlin/ -- Will code for cotton.

4 6

Re: riched20: fixes and additional conformance tests for EM_EXSETSEL
by Matt Finnicum Dec. 4, 2006

Dec. 4, 2006

Hey, I'm sorry for not noticing this bug earlier - I have a bad habit of only browsing bugs listed under wine-richedit. I actually sent in a patch that fixes this bug before, but it wasn't accepted because I hadn't marked a test that it causes to fail as todo_wine. Does the attached patch solve your problem? The attached patch moves all the logic to ME_SetSelection, so it'll apply to EM_SETSEL as well as EM_EXSETSEL (they both are supposed to follow the same special cases), and this way it will contribute to my further separating functionality from the message handlers into ME_ local functions which is needed to properly implement windowless controls. This patch does result in a wrong return if the user sent a SetSelection beyond the end of the document (windows allows you to select an imaginary character at the end), but I don't think that's a problem for anything, and I'll have it working in a couple weeks (I'm working on improving the selection code in general, a part of which will allow you to actually select that extra character in our richedit). I suppose i could just put a small hack in for the meantime, if anyone cares. I haven't tried them together yet, but your new tests do look significantly more comprehensive - well done. --Matt On 12/3/06, Lei Zhang <thestig(a)google.com> wrote: > Hi, > > Changelog: > * Check the input to EM_EXSETSEL > * Fixes bug 6814 > * Added more comprehensive conformance tests for EM_EXSETSEL > > editor.c | 52 +++++++++++++++++++++--- > tests/editor.c | 103 +++++++++++++++++++++++++++++++------------------ > 2 files changed, 112 insertions(+), 43 deletions(-) > > > > > > >

2 1

Re: tools/. mono.php
by Jonathan Ernst Dec. 4, 2006

Dec. 4, 2006

Le vendredi 01 décembre 2006 à 10:35 -0600, Jeremy Newman a écrit : > ChangeSet ID: 30235 > CVSROOT: /opt/cvs-commit > Module name: tools > Changes by: jnewman(a)winehq.org 2006/12/01 10:35:01 > > Added files: > . : mono.php > > Log message: > Jonathan Ernst <jonathan(a)ernstfamily.ch> > - mono downloader script Please don't forget to upload the files to sourceforge as requested with the patches, because it is breaking those Wine downloads... See: http://source.winehq.org/mono.php?action=showlist http://source.winehq.org/mozactivex?action=showlist Thanks

3 2

Re: [ws2_32] make closesocket cancel pending I/O operations
by Kai Blin Dec. 4, 2006

Dec. 4, 2006

On Monday 04 December 2006 07:01, Damjan Jovanovic wrote: > On Windows, calling closesocket in one thread causes blocking I/O > operations on that socket (select, accept) running in other threads to > return immediately. On Linux, the I/O will block forever. Calling > shutdown as part of closesocket reproduces the same behaviour under > Linux. Closes #4619 and #3777. Sweet, this seems to solve part of the problems in bug #5534, too. Thanks, Kai -- Kai Blin, <kai Dot blin At gmail Dot com> WorldForge developer http://www.worldforge.org/ Wine developer http://wiki.winehq.org/KaiBlin/ -- Will code for cotton.

1 0

Re: ws2_32: Rewrite getsockopt to make it more readable.
by Kai Blin Dec. 4, 2006

Dec. 4, 2006

On Sunday 03 December 2006 16:46, Kai Blin wrote: > --- > dlls/ws2_32/socket.c | 409 > ++++++++++++++++++++++++++++++++++------------ > dlls/ws2_32/tests/sock.c | 35 ++++ > 2 files changed, 332 insertions(+), 112 deletions(-) Sorry, misplaced a closing bracked somehow. I'll send a fixed version in a minute. Kai -- Kai Blin, <kai Dot blin At gmail Dot com> WorldForge developer http://www.worldforge.org/ Wine developer http://wiki.winehq.org/KaiBlin/ -- Will code for cotton.

1 0

Firefox-1.5.x on Wine slow on javascript-heavy pages
by Dan Kegel Dec. 4, 2006

Dec. 4, 2006

I've been using Firefox-1.5.x again on wine (using the winehq packages for Ubuntu) on my dual 650MHz system as my main browser again for the last few months. It's working quite well... except that javascript-heavy pages/apps like gmail are a tad sluggish at accepting keyboard input. For some reason, yahoo.com's main page is also sluggish; they must do something fancy. Both are fine with native firefox on this system. I'm sure this won't affect anyone with a modern system, but it is kind of sad that my trusty old dual CPU system isn't up to daily tasks anymore. Sniff... ah, well, I see Fry's has amd64 4000 motherboard/cpu combos for $189, guess it's time to upgrade. At some point it'd be nice to look into why Firefox's javascript is slow under Wine, though. - Dan

2 1

Re: d3d [2]: Optional callback infrastructure for implicit texture surface destruction.
by Stefan Dösinger Dec. 3, 2006

Dec. 3, 2006

Am Sonntag 03 Dezember 2006 21:52 schrieb Markus Amsler: > /* IWineD3DTexture */ > + IWineD3DTextureImpl_Destroy, Wouldn't it maybe be better to make the destroy method part of IWineD3DBase?

2 1

Re: d3d [2]: Optional callback infrastructure for implicit texture surface destruction.
by H. Verbeet Dec. 3, 2006

Dec. 3, 2006

On 03/12/06, Markus Amsler <markus.amsler(a)oribi.org> wrote: > --- > dlls/wined3d/texture.c | 53 > +++++++++++++++++++++++--------------- > include/wine/wined3d_interface.h | 7 +++++ > 2 files changed, 39 insertions(+), 21 deletions(-) I think the callback should be non-optional.

2 1

One more go at fixing GetCursorPos()
by Vitaliy Margolen Dec. 3, 2006

Dec. 3, 2006

Here is one more shot at fixing GetCursorPos() to make it usable in dinput. This time I used a refcount of sorts. Because the problem appears only during processing of hook chain. Just to remind about what is the problem: when hook handler calls GetCursorPos() position returned is the "old" position. While hook parameters contain "new" coordinates. This is on windows. On Wine returned position queried directly from X and might be even "newer" then position in hook parameters. Attached are two tests (out of lots more that I've used to find and fix the problem). test_msg.c - tests major problem that Alexandre pointed out about last fix attempt. Application in question does not process messages while calling GetCursorPos(). It still works properly with proposed patch. di_mouse_2th.c - 2 threads showing that there are no extra thread local position involved (if there is such a thing). Cached cursor position is global to all threads. (Last numbers should stay at dt(0 0) at all times). Inside hook handler d(x y) should show relative mouse move delta. If anyone has any objections, comment, suggestions, please do send them. Vitaliy Margolen. #define COBJMACROS #define _WIN32_WINNT 0x0501 #include <windows.h> #include <windowsx.h> #include <stdio.h> static HANDLE event1, event2; static HINSTANCE g_hInstance; static POINT pt_global; LRESULT CALLBACK hook_proc( int code, WPARAM wparam, LPARAM lparam ) { MSLLHOOKSTRUCT *hook = (MSLLHOOKSTRUCT *)lparam; POINT pt; GetCursorPos(&pt); if (code == HC_ACTION) { fprintf(stderr, "Hook: code=%d w=%x [(%ld %ld) %lx %lx %lx] cur(%ld %ld) d(%ld %ld) ", code, wparam, hook->pt.x, hook->pt.y, hook->mouseData, hook->flags, hook->dwExtraInfo, pt.x, pt.y, hook->pt.x - pt.x, hook->pt.y - pt.y); pt_global = pt; SetEvent(event1); WaitForSingleObject(event2, INFINITE); } return CallNextHookEx( 0, code, wparam, lparam ); } static DWORD CALLBACK thread_proc(LPVOID arg) { while (WaitForSingleObject(event1, INFINITE) == WAIT_OBJECT_0) { POINT pt; GetCursorPos(&pt); fprintf(stderr, "Th: (%ld %ld) dt(%ld %ld)\n", pt.x, pt.y, pt.x - pt_global.x, pt.y - pt_global.y); SetEvent(event2); } return 0; } static DWORD wnd_proc(void) { MSG msg; while (GetMessage(&msg, NULL, 0, 0)) { TranslateMessage(&msg); DispatchMessage(&msg); } return msg.wParam; } #define SEND(x,y) \ i.mi.time = GetCurrentTime(); \ i.mi.dx = x; i.mi.dy = y; \ SendInput(1, &i, sizeof(INPUT)); void send_input(void) { INPUT i; POINT ptorg, pt; GetCursorPos(&ptorg); fprintf(stderr, "Cur pos: %ld %ld\n", ptorg.x, ptorg.y); i.type = INPUT_MOUSE; i.mi.dwFlags = MOUSEEVENTF_MOVE; i.mi.dwExtraInfo = 0; i.mi.mouseData = 0; SEND(-4, -4) GetCursorPos(&pt); fprintf(stderr, "%ld %ld\n", pt.x, pt.y); SetCursorPos(ptorg.x, ptorg.y); Sleep(250); fprintf(stderr, "***DONE***\n"); } void test1(HWND hwnd) { HANDLE thread; HHOOK hook; MSG msg; event1 = CreateEvent(NULL, FALSE, FALSE, NULL); event2 = CreateEvent(NULL, FALSE, FALSE, NULL); thread = CreateThread(NULL, 0, thread_proc, NULL, 0, NULL); Sleep(250); hook = SetWindowsHookExA(WH_MOUSE_LL, hook_proc, g_hInstance, 0); Sleep(250); // send_input(); wnd_proc(); while (PeekMessage( &msg, 0, 0, 0, PM_REMOVE )) DispatchMessage( &msg ); UnhookWindowsHookEx(hook); CloseHandle(event1); CloseHandle(event2); CloseHandle(thread); Sleep(250); } LRESULT CALLBACK WndProc(HWND hWnd, UINT message, WPARAM wParam, LPARAM lParam) { switch (message) { case WM_DESTROY: PostQuitMessage(0); return 0; } return DefWindowProc(hWnd, message, wParam, lParam); } int WINAPI WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, LPSTR lpCmdLine, int nCmdShow) { WNDCLASSEX wcex; HWND hwnd; g_hInstance = hInstance; wcex.cbSize = sizeof(WNDCLASSEX); wcex.style = CS_HREDRAW | CS_VREDRAW; wcex.lpfnWndProc = (WNDPROC)WndProc; wcex.cbClsExtra = 0; wcex.cbWndExtra = 0; wcex.hInstance = hInstance; wcex.hIcon = NULL; wcex.hCursor = LoadCursor(NULL, IDC_ARROW); wcex.hbrBackground = (HBRUSH)(COLOR_WINDOW+1); wcex.lpszMenuName = NULL; wcex.lpszClassName = "MainWindow"; wcex.hIconSm = NULL; RegisterClassEx(&wcex); hwnd = CreateWindow("MainWindow", "Title", WS_OVERLAPPEDWINDOW, 10, 10, 200, 200, NULL, NULL, NULL, NULL); ShowWindow(hwnd, SW_SHOW); test1(hwnd); DestroyWindow(hwnd); return 0; } #include <windows.h> #include <stdio.h> int WINAPI WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, LPSTR lpCmdLine, int nCmdShow) { LONG started = GetTickCount(); POINT pt, pt_org; HWND hwnd; MSG msg; hwnd = CreateWindow("static", "Title", WS_OVERLAPPEDWINDOW | WS_VISIBLE, 10, 10, 200, 200, NULL, NULL, NULL, NULL); GetCursorPos(&pt_org); while (PeekMessage( &msg, 0, 0, 0, PM_REMOVE )) DispatchMessage( &msg ); while (started + 10000 > GetTickCount()) { GetCursorPos(&pt); if ((pt.x - pt_org.x) || (pt.y - pt_org.y)) printf("(%ld %ld)\n", pt.x - pt_org.x, pt.y - pt_org.y); pt_org = pt; } DestroyWindow(hwnd); return 0; }

2 2

winedump: Cast-qual warnings fix (2 of 2)
by Andrew Talbot Dec. 3, 2006

Dec. 3, 2006

Sorry, I made a mess of titling this patch, twice. Please find and use the version whose subject line and changelog entry refer to "winedump", not "winedbg". Thanks, -- Andy.

1 0

[PATCH] overflow checking for MultiByteToWideChar/WideCharToMultiByte
by Marcus Meissner Dec. 3, 2006

Dec. 3, 2006

Hi, This is more a FYI patch and GCC 4.0+ specific. This enhancement detects if the size of a known buffer is different than the specified size. Thie tests/sock.c adjustement is necessary due to stdlib.h inclusion. Ciao, Marcus --- dlls/kernel32/locale.c | 2 ++ dlls/ws2_32/tests/sock.c | 2 +- include/winnls.h | 45 +++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 48 insertions(+), 1 deletions(-) d0e7840335b6bbe4990b1c043dcff9b298c7a5a8 diff --git a/dlls/kernel32/locale.c b/dlls/kernel32/locale.c index bc7db75..1d4ceaf 100644 --- a/dlls/kernel32/locale.c +++ b/dlls/kernel32/locale.c @@ -1757,6 +1757,7 @@ BOOL WINAPI EnumSystemCodePagesW( CODEPA * is passed, and ERROR_NO_UNICODE_TRANSLATION if no translation is * possible for src. */ +#undef MultiByteToWideChar INT WINAPI MultiByteToWideChar( UINT page, DWORD flags, LPCSTR src, INT srclen, LPWSTR dst, INT dstlen ) { @@ -1850,6 +1851,7 @@ INT WINAPI MultiByteToWideChar( UINT pag * and dstlen != 0, and ERROR_INVALID_PARAMETER, if an invalid * parameter was given. */ +#undef WideCharToMultiByte INT WINAPI WideCharToMultiByte( UINT page, DWORD flags, LPCWSTR src, INT srclen, LPSTR dst, INT dstlen, LPCSTR defchar, BOOL *used ) { diff --git a/dlls/ws2_32/tests/sock.c b/dlls/ws2_32/tests/sock.c index c8b9cc8..f5ed0c1 100644 --- a/dlls/ws2_32/tests/sock.c +++ b/dlls/ws2_32/tests/sock.c @@ -22,8 +22,8 @@ #include <stdarg.h> #include <windef.h> -#include <winbase.h> #include <winsock2.h> +#include <winbase.h> #include <mswsock.h> #include "wine/test.h" #include <winnt.h> diff --git a/include/winnls.h b/include/winnls.h index 8888171..c13c0bf 100644 --- a/include/winnls.h +++ b/include/winnls.h @@ -20,6 +20,10 @@ #define __WINE_WINNLS_H #ifndef NONLS +#ifndef RC_INVOKED +# include <stdlib.h> +#endif + #ifdef __cplusplus extern "C" { #endif @@ -768,6 +772,26 @@ INT WINAPI LCMapStringW(LCID,DWO #define LCMapString WINELIB_NAME_AW(LCMapString) LCID WINAPI LocaleNameToLCID(LPCWSTR,DWORD); INT WINAPI MultiByteToWideChar(UINT,DWORD,LPCSTR,INT,LPWSTR,INT); + +static inline INT +WINAPI MultiByteToWideChar_ichk( + UINT cp,DWORD flags, + LPCSTR src,INT srclen,INT srcbuflen, + LPWSTR dst,INT dstlen,INT dstbuflen +) { + if ((srclen != -1) && (srcbuflen != -1) && (srcbuflen < srclen)) + abort(); + if ((dstlen != -1) && (dstbuflen != -1) && (dstbuflen/2 < dstlen)) + abort(); + /* later add assertions regarding conversion lengths */ + return MultiByteToWideChar(cp,flags,src,srclen,dst,dstlen); +} +#define MultiByteToWideChar(cp,flags,src,srclen,dst,dstlen) \ + MultiByteToWideChar_ichk(cp,flags, \ + src,srclen,__builtin_object_size(src,0), \ + dst,dstlen,__builtin_object_size(dst,0) \ + ) + INT WINAPI SetCalendarInfoA(LCID,CALID,CALTYPE,LPCSTR); INT WINAPI SetCalendarInfoW(LCID,CALID,CALTYPE,LPCWSTR); #define SetCalendarInfo WINELIB_NAME_AW(SetCalendarInfo) @@ -778,6 +802,27 @@ BOOL WINAPI SetThreadLocale(LCID) BOOL WINAPI SetUserGeoID(GEOID); INT WINAPI WideCharToMultiByte(UINT,DWORD,LPCWSTR,INT,LPSTR,INT,LPCSTR,LPBOOL); +static inline INT +WINAPI WideCharToMultiByte_ichk( + UINT cp,DWORD flags, + LPCWSTR src,INT srclen,INT srcbuflen, + LPSTR dst,INT dstlen,INT dstbuflen, + LPCSTR xstr,LPBOOL b +) { + if ((srclen != -1) && (srcbuflen != -1) && (srcbuflen/2 < srclen)) + abort(); + if ((dstlen != -1) && (dstbuflen != -1) && (dstbuflen < dstlen)) + abort(); + /* later add assertions regarding conversion lengths */ + return WideCharToMultiByte(cp,flags,src,srclen,dst,dstlen,xstr,b); +} +#define WideCharToMultiByte(cp,flags,src,srclen,dst,dstlen,defchar,b) \ + WideCharToMultiByte_ichk(cp,flags, \ + src,srclen,__builtin_object_size(src,0), \ + dst,dstlen,__builtin_object_size(dst,0), \ + defchar,b \ + ) + #ifdef __cplusplus } #endif -- 1.2.4

1 0

Re: [PATCH] check length in path canonicalize
by Dmitry Timoshkov Dec. 3, 2006

Dec. 3, 2006

"Marcus Meissner" <marcus(a)jet.franken.de> wrote: > - MultiByteToWideChar(CP_ACP,0,lpszPath,-1,szPath,MAX_PATH); > + INT ret; > + > + ret = MultiByteToWideChar(CP_ACP,0,lpszPath,-1,NULL,0); > + if (ret > MAX_PATH) { > + FIXME("Too long path!\n"); > + return FALSE; > + } > + ret = MultiByteToWideChar(CP_ACP,0,lpszPath,-1,szPath,sizeof(szPath)/sizeof(WCHAR)); It should be sufficient to check return value of MultiByteToWideChar to detect an overflow, there is no need to call MultiByteToWideChar twice for that. -- Dmitry.

1 0

XEMBED Implementation
by Cloudor Pu Dec. 3, 2006

Dec. 3, 2006

Hi, I find discussion about XEMBED implementation in this mailing list and that was 3 years ago. Is there now any usable implementation or examples for wine XEMBED plug? What I am doing is to embed IE into a GtkSocket. I wrote a standalone program compiled using winelib that can CreateWindow and embed IE. Can I CreateWindow using the XEMBED Id if there is already XEMBED implementation or can I just CreateWindow as a child of a GtkPlug which is a gtk-embeddee? Any advices or suggestions would be highly appreciated. Thank you! Regards, cloudor

3 3

Wined3d State management again
by Stefan Dösinger Dec. 2, 2006

Dec. 2, 2006

Hi, I've done some more work on my state management, and I uploaded an updated patch to http://stud4.tuwien.ac.at/~e0526822/wined3d-statemgmt-2.tar.bz2 . Ok, so what's updated: * Pixel shaders: Should work now * GLSL shaders: Working too. Henris shader backend stuff is integrated in a messy way. * Offscreen rendering: Works with back buffer and pbuffer rendering, fbos not yet. * Stateblocks: Brute force integration by just scheduling all modified states for a reset. No display list yet * Blitting, render target unlocking: Enabled again, sets new gl states and schedules a reapplication of the modified states. Not a nice solution yet What is still missing: * Nv register combiners. I haven't found a way yet to build a texture stage -> texture unit mapping which I like. All the ways I could think of could potentially cause lots of unneccessary state changes. One nice thing I have noticed is that this changes give a huge performance gain in Half-Life 2(dxlevel 80, and 90). Dxlevel 80 is accelerated to approximately 150-200% of the former performance, and especially slow parts are faster now. This makes a really huge playability difference :-) There are some regressions too, especially related to texture loading(3Dmark2000), and some glsl crashes. The good part about this change is that I can send it in in very small patches, making regression testing much easier, contrary to the huge ddraw rewrite. Where to go from here? As long as nobody screams I'd start sending patches now. Its easier to catch regressions while sending small patches instead of hunting them on the blob I have here(and my messy patch history in my git tree). I'm also afraid that I get too far away from the main wined3d code(See the shader backend integration). We can discuss the code based on the small changes I'll send. Stefan

1 1

Re: msi 2: Remove redundant NULL checks before x_Release
by Mike McCormack Dec. 2, 2006

Dec. 2, 2006

James Hawkins wrote: > Changelog: > * Remove redundant NULL checks before x_Release. obj->Release() will read the vtable pointer from the object, and that will crash if obj is NULL, so these checks are needed. The following code will crash: IUnknown *obj = NULL; IUnknown_Release(obj); You can see this looking at the definition of the IUnknown_Release macro: include/unknwn.h:#define IUnknown_Release(p) (p)->lpVtbl->Release(p) Mike

1 0

Video problems with FC6
by gslink Dec. 1, 2006

Dec. 1, 2006

There is currently a bug in FC6 that affects Wine. If, with any Nvidia driver from any source, a program attempts to use accelerated video and the attached monitor uses digital feed the program will often crash the user session. This has been reported and is not a problem in Wine but a good many games that used to run under Wine now crash. Other cards were not tested but may be affected. It is hoped RH will fix this rather quickly.

2 1

Re: [PATCH 1/9] dinput: Move acquired flag to the base device class.
by Peter Oberndorfer Dec. 1, 2006

Dec. 1, 2006

On Friday 01 December 2006 07:23, Vitaliy Margolen wrote: > Sorry ignore my previous patches. I didn't send them in the proper order. > > --- > dlls/dinput/device.c | 28 ++++++++++++++++++++++++++++ > dlls/dinput/device_private.h | 3 +++ > dlls/dinput/joystick_linux.c | 23 ++++++++--------------- > dlls/dinput/keyboard.c | 36 ++++++++++++++++-------------------- > dlls/dinput/mouse.c | 23 ++++++++--------------- > 5 files changed, 63 insertions(+), 50 deletions(-) > Hi, some questions/notices about this patch: Previously when the device was not acquired inside JoystickAImpl_Unacquire DIERR_NOTACQUIRED was returned. now you call IDirectInputDevice2AImpl_Unacquire, which returns DI_NOEFFECT when not acquired. but DI_NOEFFECT does not have high bit set so it won't trigger the FAILED macro. Do you know if the Acquire/Unacquire functions return value the same for all device types? In JoystickAImpl_Acquire you don't use your new IDirectInputDevice2AImpl_Acquire function but set This->base.acquired = 1; Is there a reason for this? In SysKeyboardAImpl_Unacquire you seem to call IDirectInputDevice2AImpl_Acquire? Greetings Peter

2 1

Re: user32: Factorize graphics driver's WindowCreate and ShowWindow in user32.
by Dmitry Timoshkov Dec. 1, 2006

Dec. 1, 2006

"Pierre d'Herbemont" <pdherbemont(a)free.fr> wrote: > At the same time ShowWindow gets factorized because WindowCreate and > ShowWindow share both the use of WINPOS_MinMaximize, so it is easier to > factorize both functions in a single patch. In fact ShowWindow is > removed entirely from the drivers as it is driver independant. > > To properly factorize WindowCreate we have to export 3 new functions > from the graphics driver which are SetIconicState, SyncWindowStyle, > SystrayDockWindow. I'd suggest to leave ShowWindow in the driver's interface. You already have created 3 new entry points to replace its partial functionality, but still miss ShowWindow functionality for at least SW_RESTORE and SW_MAXIMIZE. Also did you run 'make test' after this patch? -- Dmitry.

2 3

Re: oleaut32: add VarBstrCmp binary comparison for LCID==0
by Dmitry Timoshkov Dec. 1, 2006

Dec. 1, 2006

"Charles Blacklock" <charles(a)diagnos.co.uk> wrote: > + if (lcid == 0) > + { > + ret = memcmp(pbstrLeft, pbstrRight, min(SysStringByteLen(pbstrLeft), SysStringByteLen(pbstrRight))); > + if (ret < 0) > + return VARCMP_LT; > + if (ret > 0) > + return VARCMP_GT; > + if (SysStringByteLen(pbstrLeft) < SysStringByteLen(pbstrRight)) > + return VARCMP_LT; > + if (SysStringByteLen(pbstrLeft) > SysStringByteLen(pbstrRight)) > + return VARCMP_GT; > + return VARCMP_EQ; > + } It would be much more effective to call SysStringByteLen only once for each string instead of calling it again and again after a failing comparison. -- Dmitry.

2 2

How to embed wine into swt?
by Cloudor Pu Dec. 1, 2006

Dec. 1, 2006

Hi, all I am developing an swt app only for gtk. I want to embed IE into this app. I can create a SWT.EMBEDDED composite and get its embededHandle (an integer). Then, how to do next? Thanks in advace! regards, cloudor

3 3