January 2010 - Wine-devel - list.winehq.org

[2/3] ntdll: fill unused tails of heap blocks if HEAP_FREE_CHECKING_ENABLED (take 3)
by Joris Huizer 08 Jan '10

08 Jan '10

Dan Kegel, you wrote: To check whether we're too early in process setup to check global flags, I compare the arguments of RtlCreateHeap with the ones used by the call in virtual_init(). This is not especially clean, but I couldn't think of anything better. I was thinking, couldn't you place the RtlCreateHeap implementation with an internal function, that takes an extra argument (boolean or flags) to indicate whether to do this initialization step or not ? Then, call the internal function from virtual_init() with the boolean set, and have RtlCreateHeap just calling it, without the boolean set. I've seen something like this being done elsewhere (but for different reasons) I'm not sure whether this would be acceptable (performance-wise?) but it'd be cleaner I think. HTH, Joris

1 0

Re: Patch feedback requested for OleCreatePropertyFrame()
by Geoffrey Hausheer 08 Jan '10

08 Jan '10

Ok, I have reimplemented OleCreatePropertyFrame() from scratch. There is no code from Hidenori's patch in mine, nor did I use it as a reference (though of course I can't claim that I didn't learn anything from his patch during my previous cleanup) Would someone mind reviewing this for style/content? I still have no idea how to write a test for a modal dialog, so if a test is required for submission, I'd really appreciate someone giving me pointers on how to go about it. There are some features I didn't implement because I didn't really understand MDSN sufficiently (specifcially the IPropertyPageSite stuff), but it is mostly complete. .Geoff

3 2

Re: [2/3] ntdll: fill unused tails of heap blocks if HEAP_FREE_CHECKING_ENABLED (take 3)
by Alexandre Julliard 08 Jan '10

08 Jan '10

Dan Kegel <dank(a)kegel.com> writes: > @@ -1254,14 +1291,20 @@ HANDLE WINAPI RtlCreateHeap( ULONG flags, PVOID addr, SIZE_T totalSize, SIZE_T c > { > SUBHEAP *subheap; > > - /* Allocate the heap block */ > + /* Check global flags unless we're being called in virtual_init before they > + * are available. Recognize that by comparing the arguments we know > + * wine's virtual_init calls us with. > + */ > + if (flags != HEAP_NO_SERIALIZE && totalSize != VIRTUAL_HEAP_SIZE) > + if ((RtlGetNtGlobalFlags() & FLG_HEAP_ENABLE_FREE_CHECK) != 0) > + flags |= HEAP_FREE_CHECKING_ENABLED; This makes no sense, there's nothing unique about these arguments. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: crypt32: check for NULL target pointers (Coverity)
by Juan Lang 07 Jan '10

07 Jan '10

Hi Marcus, Ideally this patch would be accompanied by a test. Also, this change: @@ -212,6 +212,10 @@ static BOOL CRYPT_DecodeEnsureSpace(DWORD dwFlags, if (dwFlags & CRYPT_DECODE_ALLOC_FLAG) { + if (!pvStructInfo) { + SetLastError(ERROR_INVALID_PARAMETER); + return FALSE; + } is a noop in all but one case: all the callers of CRYPT_DecodeEnsureSpace, save one, check pvStructInfo before calling it. It would be clearer, IMO, to change the single caller that doesn't check pvStructInfo (CryptDecodeObjectEx) rather than adding a check that is useless in most cases. A similar statement applies to the encode.c change: just change CryptEncodeObjectEx, not CRYPT_EncodeEnsureSpace. Finally, please indent consistently with the rest of the file. If you prefer, I can try to fix this. Triaging the Coverity bugs is probably enough work by itself, without being expected to fix them too ;-) Thanks, --Juan

2 1

patch 57244: (try 3) msi: queue dynamically allocated strings in cond.y...
by Nate Gallaher 07 Jan '10

07 Jan '10

Anything I can do to help move this along? ~Nate

1 0

FIXMEs in DllCanUnloadNow and friends (and how to handle them)
by Reece Dunn 07 Jan '10

07 Jan '10

Hi, As a result of the "Wine FIXME Report 2009 Aug - Dec" thread, I created the following to hunt for DllCanUnloadNow calls that were marked as being FIXME stubs: cat > DllCanUnloadNow.pl < EOF #!/usr/local/bin/perl # Based on http://www.unix.com/unix-dummies-questions-answers/56703-multiline-grep.html use strict; my $filename = shift; open (FILE, "<", $filename) or die "Failed to read file $filename : $! \n"; my $whole_file; { local $/; $whole_file = <FILE>; } close(FILE); if ($whole_file =~ m#HRESULT WINAPI DllCanUnloadNow ?$ ?(void|VOID) ?$[\n ]{\n\s+FIXME#sg) { print $filename . "\n"; } EOF and used it with the following: $ grep -F DllCanUnloadNow -r dlls | grep -F HRESULT | sed -e 's/:.*//' | while read line ; do perl DllCanUnloadNow.pl $line ; done | sort | tee results.log This gives the following results: ----- 8< ----- dlls/ddrawex/main.c dlls/dinput/dinput_main.c dlls/dpnhpast/main.c dlls/dsound/dsound_main.c dlls/hlink/hlink_main.c dlls/hnetcfg/hnetcfg.c dlls/inetcomm/inetcomm_main.c dlls/infosoft/infosoft_main.c dlls/inseng/inseng_main.c dlls/itircl/itircl_main.c dlls/mmdevapi/main.c dlls/mpr/mpr_main.c dlls/mscoree/mscoree_main.c dlls/msimtf/main.c dlls/msxml3/main.c dlls/oledb32/main.c dlls/olepro32/olepro32stubs.c dlls/pstorec/pstorec.c dlls/query/query_main.c dlls/shell32/shell32_main.c dlls/sti/sti_main.c dlls/wbemprox/main.c dlls/winhttp/main.c dlls/wmiutils/main.c dlls/wuapi/main.c ----- >8 ----- I am currently going through the results and removing the FIXME stubs for the DllCanUnloadNow calls, as returning S_FALSE is a perfectly valid implementation (see the "Wine FIXME Report 2009 Aug - Dec" discussion). While doing this, I noticed that some implementations of DllGetClassObject, DllRegisterServer and DllUnregisterServer also have FIXMEs. This is because the DLLs implement COM objects, but the wine versions are incomplete. For the DllRegister/UnregisterServer calls, the FIXMEs should be removed -- the FIXMEs do not add any value, and are only likely to appear during install or when calling regsvr32. For DllGetClassObject, the FIXME should output the GUID of the object that is trying to be created. NOTE: I have also seen some implementations of DllCanUnloadNow that always return S_OK (e.g. dlls/hlink/hlink_main.c, even though it is implementing COM objects). I will keep a note of these and look to see whether they implement COM objects or not. I will sent a separate patch to address this issue. - Reece

3 3

Re: user32: Do not allow a change of capture if the currently capture window is a menu unless explicitly specified
by Paul Vriens 07 Jan '10

07 Jan '10

On 01/07/2010 03:33 AM, Peter Dons Tychsen wrote: > + /* check that SetCapture fails for another window and that it does not touch the error code */ > + set_cap_wnd = SetCapture(hWnd); > + err = GetLastError(); > + ok(!set_cap_wnd, "ReleaseCapture should have failed!\n"); > + ok(err == ERROR_SUCCESS, "Bad error-code from SetCapture. got %08X expected %08X\n", err, 0); > + > + /* check that ReleaseCapture fails and does not touch the error code */ > + status = ReleaseCapture(); > + err = GetLastError(); > + ok(!status, "ReleaseCapture should have failed!\n"); > + ok(err == ERROR_SUCCESS, "Bad error-code from ReleaseCapture. got %08X expected %08X\n", err, 0); Hi Peter, The comment says 'does not touch the error code' but the tests don't actually prove this. In general when we try to get a true hold of the last error to see if it changed we set it first, f.e. "SetLastError(0xdeadbeef)". -- Cheers, Paul.

1 0

Re: mlang: [RESEND] Print a FIXME only for unused parameter (Wine FIXME Report)
by James Hawkins 06 Jan '10

06 Jan '10

On Wed, Jan 6, 2010 at 3:28 PM, Detlef Riekenberg <wine.dev(a)web.de> wrote: > According to MSDN, both parameter are unused and must be 0/NULL > A FIXME for every call is useless > > (there is no PATCH 1/2, so a resend without that prefix) > > -- > By by ... Detlef > > > --- > dlls/mlang/mlang.c | 8 ++++---- > 1 files changed, 4 insertions(+), 4 deletions(-) > > diff --git a/dlls/mlang/mlang.c b/dlls/mlang/mlang.c > index 49f0605..9abe124 100644 > --- a/dlls/mlang/mlang.c > +++ b/dlls/mlang/mlang.c > @@ -3007,9 +3007,6 @@ exit: > return hr; > } > > -/* > - * TODO: handle dwFlag and lpFallBack > -*/ > static HRESULT WINAPI fnIMultiLanguage2_ConvertStringToUnicodeEx( > IMultiLanguage3* iface, > DWORD* pdwMode, > @@ -3021,7 +3018,10 @@ static HRESULT WINAPI fnIMultiLanguage2_ConvertStringToUnicodeEx( > DWORD dwFlag, > WCHAR* lpFallBack) > { > - FIXME("\n"); > + if (dwFlag || (lpFallBack != NULL)) > + FIXME("Ignoring dwFlag (0x%x/%d) and lpFallBack (%p)\n", > + dwFlag, dwFlag, lpFallBack); > + if (dwFlag || lpFallBack) seems to be simpler and easier to read. -- James Hawkins

1 0

New Icon set
by Detlef Riekenberg 06 Jan '10

06 Jan '10

Hi Joel When I read your preview page in Firefox, the lens for print preview in idb_std_small.bmp is to large. You can't detect the printer behind the lens. (LCD Resolution here is 1400x1050) As an Idea for an improvement, the print preview menu icon in Firefox is readable: The lens is a bit smaller but more important, the inner circle of the lens is transparent. Thanks for your artwork. -- By by ... Detlef ___________________________________________________________ Preisknaller: WEB.DE DSL Flatrate für nur 16,99 Euro/mtl.! http://produkte.web.de/go/02/

1 0

re: [PATCH] kernel32: GetPrivateProfileSectionNamesA error checking (Coverity)
by Dan Kegel 06 Jan '10

06 Jan '10

AJ wrote: > Marcus Meissner <marcus at jet.franken.de> writes: >> buffer and size need to be either 0 at the same time or not, >> otherwise this breaks up. Windows is inconsistent (either crashes >> or returns invalids), so we can just add checking. > >If Windows crashes we don't need extra checks. Just tell Coverity that >crashing is allowed here. If later versions of Windows return INVALID_PARAMETER, wouldn't it be ok for Wine to do so rather than crashing? In other words, isn't it ok to emulate the 'best' behaved version of windows in cases like this, rather than the 'worst'?

3 3