June 2015 - Wine-devel - list.winehq.org

Re: kernel32: K32EnumProcessModules: Handle NULL pointer gracefully.
by Alex Henrie 01 Jul '15

01 Jul '15

> diff --git a/dlls/kernel32/module.c b/dlls/kernel32/module.c > index f495e81..5461881 100644 > --- a/dlls/kernel32/module.c > +++ b/dlls/kernel32/module.c > @@ -1226,6 +1226,12 @@ BOOL WINAPI K32EnumProcessModules(HANDLE process, HMODULE *lphModule, > if (!init_module_iterator(&iter, process)) > return FALSE; > > + if (!lphModule) > + { > + SetLastError(ERROR_NOACCESS); > + return FALSE; > + } > + > if (!needed) > { > SetLastError(ERROR_NOACCESS); Wouldn't it be easier to write "if (!lphModule || !needed)"? Adding a test or two would also be nice. -Alex

2 1

Re: [PATCH 1/6]: msvcp110: Add tr2_sys__Copy_file implementation and test.
by Piotr Caban 01 Jul '15

01 Jul '15

On 06/27/15 19:49, YongHaoHu wrote: > + ret = p_tr2_sys__Copy_file(tests[i].source, tests[i].dest, tests[i].fail_if_exists); > + ok(ret == GetLastError(), "test_tr2_sys__Copy_file(): test %d expect: %d, got %d\n", i+1, GetLastError(), ret); > + ok(errno == 0xdeadbeef, "test_tr2_sys__Copy_file(): test %d errno expect 0xdeadbeef, got %d\n", i+1, errno); Please check the return value of the function explicitly. Currently your implementation returns error while it should succeed and it's not visible in the tests. Thanks, Piotr

2 6

Proposal: to make binfmt-misc with wine workable and make Wine compadiblity modes more like Windows compadiblity mode.
by Peter Dolding 01 Jul '15

01 Jul '15

For binfmt-misc to be workable a few things need to be known. 1) What version of wine 2) What wineprefix 3) What setting overrides. None of these can be done by environmental vars because binfmt-misc might be directly executing the file. Windows compadiblity mode does not use the registry like wine does. Instead it done in two horible halves. Half 1. The application manifest file. The mainfest file may be embedded in the .exe file or be .exe.mainfest https://msdn.microsoft.com/en-us/library/windows/desktop/hh848036%28v=vs.85… Yes this is where you set what version of Windows application was designed to run with. So running XP mode on Windows 7 is here. Half 2. Application Compatilibity Database This is where it gets horible using shims on particular executables to make broken executables work. Also when this gets too large starts slowing windows down as it searching the database to make applications work. Yes reason some old applications don't work with Windows 10 that worked with 7 is that their entries were deleted from this database. To solve the problem I think we could get away with just Half the manifest can include library overrides and some windows configuration tweaks. But we have problems the include library information wine needs because of native and built-in libraries. Due to the fact wine is running on Linux and other OS's some of the configuration tweaks are going to be wine dependant. So we need another schema. http://wiki.winehq.org/UsefulRegistryKeys Because if Microsoft manifest would tolerate alien schema really everything under software\wine\appdefaults could be moved from the registry to the manifest. Advantage 1 this can be write protected per application and second advantage possibility of embedding in the exe itself. Also for binfmt-misc support have in the schema the means to record wineprefix and wine to use. So the program starting wine applications by binfmt-misc looks for either a mainfest or a winemainfest file in the same directory as the exe to see what wine environment the application should be running with. If there is no configured environment possibly ask user what wineprefix they do what to run the application with. What I do not know is if Microsoft application manifest file will tolerate an alien schema. Don't have different versions of Windows to test if Microsoft application manifest with a alien schema parts works or just causes windows to throw error or just results in the manifest being complete ignored. Now if Microsoft application manifest will not tolerate alien I have to ask would creating own own like exe.winemainfest be acceptable. I do hope Microsoft application manifest will tolerate alien because it would be really useful to have applications with a manifest a values saying what versions of wine they were tested with by the maker. Currently there is no way for a application vendor to record in the application they tested with wine. A lot of ways putting what shim modifications the application need in the manifest would also remove having to look up a huge database. Getting binfmt-misc working also will remove having to use setcap on wineloader instead of the .exe. Also would make if someone set a capicapity limitation by logind not break every wine application just because a few application where needing privileged operation.. So this alteration would be getting closer to how Windows does stuff and in the process making Linux integration better. Peter Dolding

4 7

Re: windowscodecs: Fix build failure on OSX.
by Austin English 30 Jun '15

30 Jun '15

On Tue, Jun 30, 2015 at 2:02 PM, Sebastian Lackner <sebastian(a)fds-team.de> wrote: > The build failure unfortunately was introduced by my last patch > http://source.winehq.org/git/wine.git/commit/863f432a7d73b8b480a742fcbaf01d… > > Somehow I missed that this part is OSX specific and tested on Linux only - sorry for > the trouble. :/ The patch was tested by jam__ on IRC, would be nice if someone could > double check that no other warnings were introduced by my original commit. > > --- > dlls/windowscodecs/icnsformat.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/dlls/windowscodecs/icnsformat.c b/dlls/windowscodecs/icnsformat.c > index 54c7707..5d48a1a 100644 > --- a/dlls/windowscodecs/icnsformat.c > +++ b/dlls/windowscodecs/icnsformat.c > @@ -179,7 +179,7 @@ static ULONG WINAPI IcnsFrameEncode_Release(IWICBitmapFrameEncode *iface) > if (This->icns_image != NULL) > HeapFree(GetProcessHeap(), 0, This->icns_image); > > - IWICBitmapFrameEncode_Release(&This->encoder->IWICBitmapFrameEncode_iface); > + IWICBitmapEncoder_Release(&This->encoder->IWICBitmapEncoder_iface); > HeapFree(GetProcessHeap(), 0, This); > } > > -- > 2.4.4 > > With your patch applied, the build is fixed for me. There are 5 warnings, but they're unrelated to your patch (deprecated warnings): Austins-Mac-mini:windowscodecs austin$ make gcc -m32 -c -o icnsformat.o icnsformat.c -I. -I../../include -I/opt/local/include/libpng16 -D__WINESRC__ \ -D_REENTRANT -fPIC -Wall -pipe -fno-strict-aliasing -Wdeclaration-after-statement -Wempty-body \ -Wignored-qualifiers -Wstrict-prototypes -Wtype-limits -Wvla -Wwrite-strings -Wpointer-arith \ -gdwarf-2 -gstrict-dwarf -fno-omit-frame-pointer -I/usr/X11/include -I/opt/local/include \ -I/usr/local/include -g -O2 icnsformat.c:430:11: warning: 'PtrToHand' is deprecated: first deprecated in OS X 10.8 [-Wdeprecated-declarations] ret = PtrToHand(This->icns_image, &handle, This->size * This->size * 4); ^ /System/Library/Frameworks/CoreServices.framework/Frameworks/CarbonCore.framework/Headers/MacMemory.h:1772:1: note: 'PtrToHand' has been explicitly marked deprecated here PtrToHand( ^ icnsformat.c:439:5: warning: 'DisposeHandle' is deprecated: first deprecated in OS X 10.8 [-Wdeprecated-declarations] DisposeHandle(handle); ^ /System/Library/Frameworks/CoreServices.framework/Frameworks/CarbonCore.framework/Headers/MacMemory.h:1278:1: note: 'DisposeHandle' has been explicitly marked deprecated here DisposeHandle(Handle h) __OSX_AVAILABLE_BUT_DEPRECATED(__MAC_10_0, __MAC_10_8, __IPHONE_NA, __IPHONE_NA); ^ icnsformat.c:526:13: warning: 'DisposeHandle' is deprecated: first deprecated in OS X 10.8 [-Wdeprecated-declarations] DisposeHandle((Handle)This->icns_family); ^ /System/Library/Frameworks/CoreServices.framework/Frameworks/CarbonCore.framework/Headers/MacMemory.h:1278:1: note: 'DisposeHandle' has been explicitly marked deprecated here DisposeHandle(Handle h) __OSX_AVAILABLE_BUT_DEPRECATED(__MAC_10_0, __MAC_10_8, __IPHONE_NA, __IPHONE_NA); ^ icnsformat.c:550:43: warning: 'NewHandle' is deprecated: first deprecated in OS X 10.8 [-Wdeprecated-declarations] This->icns_family = (IconFamilyHandle)NewHandle(0); ^ /System/Library/Frameworks/CoreServices.framework/Frameworks/CarbonCore.framework/Headers/MacMemory.h:451:1: note: 'NewHandle' has been explicitly marked deprecated here NewHandle(Size byteCount) __OSX_AVAILABLE_BUT_DEPRECATED(__MAC_10_0, __MAC_10_8, __IPHONE_NA, __IPHONE_NA); ^ icnsformat.c:667:19: warning: 'GetHandleSize' is deprecated: first deprecated in OS X 10.8 [-Wdeprecated-declarations] buffer_size = GetHandleSize((Handle)This->icns_family); ^ /System/Library/Frameworks/CoreServices.framework/Frameworks/CarbonCore.framework/Headers/MacMemory.h:1356:1: note: 'GetHandleSize' has been explicitly marked deprecated here GetHandleSize(Handle h) __OSX_AVAILABLE_BUT_DEPRECATED(__MAC_10_0, __MAC_10_8, __IPHONE_NA, __IPHONE_NA); ^ 5 warnings generated. -- -Austin

1 0

Re: odbccp32: Implement SQLGetPrivateProfileString (try 6 - resend)
by Nikolay Sivov 30 Jun '15

30 Jun '15

> + sect = heap_strdupAtoW(lpszSection); > + entry = heap_strdupAtoW(lpszEntry); > + string = heap_strdupAtoW(lpszDefault); > + file = heap_strdupAtoW(lpszFilename); > + retval = heap_alloc(cbRetBuffer*sizeof(WCHAR)); > + > + ret = SQLGetPrivateProfileStringW(sect, entry, string, retval, cbRetBuffer, file); > + if(ret) > + { > + WideCharToMultiByte(CP_ACP, 0, retval, -1, RetBuffer, ret+1, NULL, NULL); > + } > + > + heap_free(sect); > + heap_free(entry); > + heap_free(string); > + heap_free(file); > + > + heap_free(retval); I think it's still the same issue as last time - you're using WCHAR vs CHAR buffer lengths wrong, it might work for this function as long as it's always called with simple strings. But in general WideCharToMultiByte should not be used like that: 'cbRetBuffer' you got here is A-buffer length, yet you allocate W-buffer of this length. Later when you convert W->A you use return value from W-call as A-buffer length.

1 0

Re: d3dcompiler: Add D3DCompileFromFile, D3DCompile2 stubs (try 2)
by Matteo Bruni 30 Jun '15

30 Jun '15

2015-06-30 11:01 GMT+02:00 Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com>: > Hi, > Removed #if > Corrected FIXME to be consistent with the reset of the file. > > Changelog: > d3dcompiler: Add D3DCompileFromFile, D3DCompile2 stubs > > > Best Regards > Alistair Leslie-Hughes I guess it's good enough for me. For your following patches: avoid LPxxx types, watch your whitespaces, no need to manually add '\n' in the middle of trace strings.

1 0

Re: [PATCH 0/8] advapi32: Fail on missing value or val_count parameters. (was advapi32: Fix RegEnumValueW when enumerating long values)
by Nikolay Sivov 30 Jun '15

30 Jun '15

On 30.06.2015 15:37, Bernhard Übelacker wrote: > Bug #38796: > RegEnumValue currently returns wrong minimum expected data_count when called without proper > value or val_count when the retrieved data (+overhead) exceeds the fixed buffer > variable in RegEnumValue. > > This series of patches does following: > - Add tests to cover calls to RegEnumValue without proper value and val_count parameters. > I think it's better to have tests committed first and change all other places after that.

1 0

Re: [PATCH 4/6]: msvcp110: Add tr2_sys__Stat implementation and test.
by Piotr Caban 30 Jun '15

30 Jun '15

On 06/27/15 19:50, YongHaoHu wrote: > +static file_type stat_set_error(int *err_code) > +{ > + file_type ret; > + *err_code = ERROR_SUCCESS; > + > + switch(GetLastError()) { > + case ERROR_FILE_NOT_FOUND: > + case ERROR_BAD_NETPATH: > + case ERROR_INVALID_NAME: > + case ERROR_BAD_PATHNAME: > + case ERROR_PATH_NOT_FOUND: > + ret = file_not_found; > + break; > + case ERROR_BAD_DEV_TYPE: > + ret = regular_file; > + break; > + default: > + ret = status_unknown; Shouldn't err_code be set to GetLastError in default case? You end up setting it to ERROR_SUCCESS in every case except if path is NULL. Why are you handling ERROR_BAD_DEV_TYPE error here? > + TRACE("(%s %p)\n", debugstr_a(path), err_code); > + DWORD attr; This will produce a compilation warning. Cheers, Piotr

2 4

Re: ntdll: Randomize security cookie when available (try 4)
by Alexandre Julliard 30 Jun '15

30 Jun '15

André Hentschel <nerv(a)dawncrow.de> writes: > @@ -1278,6 +1281,26 @@ static NTSTATUS map_image( HANDLE hmapping, int fd, char *base, SIZE_T total_siz > } > } > > + > + /* randomize security cookie */ > + > + if (IMAGE_DIRECTORY_ENTRY_LOAD_CONFIG < nt->OptionalHeader.NumberOfRvaAndSizes && > + (pos = nt->OptionalHeader.DataDirectory[IMAGE_DIRECTORY_ENTRY_LOAD_CONFIG].VirtualAddress)) > + { > + IMAGE_LOAD_CONFIG_DIRECTORY *loadcfg = (IMAGE_LOAD_CONFIG_DIRECTORY *)(ptr + pos); > + ULONG_PTR *cookie = (ULONG_PTR *)loadcfg->SecurityCookie; > + struct timeval tv; > + > + gettimeofday( &tv, NULL ); > + srand( tv.tv_usec * getpid() ); > + *cookie = rand(); That's not much better. It doesn't make sense to reset the seed on every call. -- Alexandre Julliard julliard(a)winehq.org

2 2

Re: [PATCH]winecfg: Double click in dlls list to edit overides. (try 4)
by Nikolay Sivov 30 Jun '15

30 Jun '15

On 29.06.2015 12:35, Hao Peng wrote: > Add a feature to winecfg. User can double click in dlls list to edit > overides of selected item. > > superseded all earlier patches. > > --- > programs/winecfg/libraries.c | 14 ++++++++++++++ > 1 file changed, 14 insertions(+) > > + case CBN_DBLCLK: > + if (LOWORD(wParam) == IDC_DLLS_LIST) > + { > + int index; > + POINT p; > + if (GetCursorPos(&p) && ScreenToClient((HWND)lParam, &p)) > + { > + index = SendDlgItemMessageW(hDlg, IDC_DLLS_LIST, > + LB_ITEMFROMPOINT, 0, MAKELPARAM(p.x, p.y)); > + if (!HIWORD(index)) > + on_edit_click(hDlg); > + } > + } > + break; Correct me if I'm wrong, but I don't think you get CBN_DBLCLK unless clicked on an item. Quick test with ControlSpy on Windows confirms that. And by the time you get CBN_DBLCLK something is always selected I think, which makes all that cursor complexity redundant.

2 2