June 2008 - Wine-devel - list.winehq.org

[d3d8] Fix CreateImageSurface implementation (with testcase)
by Tobias Jakobi 21 Jun '08

21 Jun '08

(I think this belongs into wine-devel, sorry for also posting this on wine-patches) Oops, sorry! I corrected the mixed indentation style (must have been the copy&paste). However I think the comment should stay there. I don't think the MSDN docs are updated in the near future, at least what D3D9 is concerned (now that D3D10 is more and more used). So in case someone is wondering why the implementation of wine differs from the behaviour described in the MSDN this comment should clarify this. Now should I also add something about the scratch pool type? Greets, Tobias

2 1

[PATCH] GetPrivateProfileString should set buffer to "" if default value is not set
by Christopher Smith 21 Jun '08

21 Jun '08

Our existing tests for GetPrivateProfileString were not checking to see if strings were null terminated, and in particular, the case where lpDefault wasn't set, the code actually *didn't* set the target buffer to an empty string as it should. Valgrind noticed this problem, but the unit tests didn't, so I've updated the unit tests to add this test *and* fixed the relevant code to GetPrivateProfileString. Please find the patch attached. --Chris diff --git a/dlls/kernel32/profile.c b/dlls/kernel32/profile.c index 9b6b1a1..e52cd66 100644 --- a/dlls/kernel32/profile.c +++ b/dlls/kernel32/profile.c @@ -1114,9 +1114,10 @@ static int PROFILE_GetPrivateProfileString( LPCWSTR section, LPCWSTR entry, } else if (buffer && def_val) { lstrcpynW( buffer, def_val, len ); ret = strlenW( buffer ); - } - else + } else { + buffer[0] = 0; /* buffer must be a null terminated empty string */ ret = 0; + } RtlLeaveCriticalSection( &PROFILE_CritSect ); diff --git a/dlls/kernel32/tests/profile.c b/dlls/kernel32/tests/profile.c index cddc9a0..2f0ff13 100644 --- a/dlls/kernel32/tests/profile.c +++ b/dlls/kernel32/tests/profile.c @@ -338,11 +338,17 @@ static void test_profile_existing(void) OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); ok(INVALID_HANDLE_VALUE != h, "%d: CreateFile failed\n",i); SetLastError(0xdeadbeef); + /* Fill the buffer with non-zero values so we can test if the + result is properly null terminated */ + memset(buffer, 0xff, sizeof(buffer)); ret = GetPrivateProfileStringA(SECTION, KEY, NULL, buffer, MAX_PATH, testfile2); if (!pe[i].read_error) ok( ret, "%d: GetPrivateProfileString failed with error %u\n", i, GetLastError() ); else + { ok( !ret, "%d: GetPrivateProfileString succeeded\n", i ); + ok( buffer[ret] == 0, "Get GetPrivateProfileString should terminate buffer with null\n" ); + } CloseHandle(h); } ok( DeleteFile(testfile2), "delete failed\n" );

1 0

Re: rpcrt4: Add test for using a conformant varying structure with pointers.
by Rob Shearman 20 Jun '08

20 Jun '08

2008/6/20 Rob Shearman <robertshearman(a)gmail.com>: > --- > dlls/rpcrt4/tests/server.c | 35 +++++++++++++++++++++++++++++++++-- > dlls/rpcrt4/tests/server.idl | 9 +++++++++ > 2 files changed, 42 insertions(+), 2 deletions(-) Please don't apply this patch - it contains a bug that will cause it to fail on Windows and wrongly fail on Wine. -- Rob Shearman

1 0

Valgrind status: 115 errors left
by Dan Kegel 20 Jun '08

20 Jun '08

Changes since yesterday: - Added two patches by Huw, one by Roy - Removed one by Roy that's now in git This plus other improvements in git get us down to 115, rather a stunning improvement from a few days ago. Please join the party! We can use all the help we can get getting rid of these bugs; though many are innocuous to users, a few aren't, and they're all annoying to anyone using valgrind. When you're running valgrind yourself, you probably want to run it like I do in http://kegel.com/wine/valgrind/runtest.patch with the --workaround-gcc296-bugs=yes option (regardless of which compiler you use), else you'll see a warning about invalid stack use on each exception. Report showing unique nonleak errors: http://kegel.com/wine/valgrind/logs-2008-06-19.counts.txt Logs: http://kegel.com/wine/valgrind/logs-2008-06-19.log.gz http://kegel.com/wine/valgrind/logs-2008-06-19/ Pending wine patches: http://kegel.com/wine/valgrind/logs-2008-06-19-patches/ http://kegel.com/wine/valgrind/logs-2008-06-19-patches.tgz http://kegel.com/wine/valgrind/logs-2008-06-19-patch.txt Suppression file: http://kegel.com/wine/valgrind/valgrind-suppressions

3 4

Re: textserv: Added txtsrv.c tests.
by Maarten Lankhorst 20 Jun '08

20 Jun '08

Hi Austin, 2008/6/11 Austin Lund <austin.lund(a)gmail.com>: > New conformance test for windowless text services. Only tests some > very basic operations and will need future refinement. When passing > all the tests many of the windowless text services should work for > Google Talk as per bug 5162. You should resend the patch, but first some minor nitpicks: If comments become out of date (such as the asm-global-func one, update them. My first comment isn't entirely accurate any more. Remove the TIMEALIVE define, and instead do something when winetest_interactive is set. Cheers, Maarten.

1 0

Internet Explorer 6 and Wine1-rc5
by Leslie Viljoen 20 Jun '08

20 Jun '08

Hi people! I am a bit confused by the appdb entries for Internet Explorer 6 (http://appdb.winehq.org/appview.php?versionId=469) There's a bug linked in one of the posts that seems to indicated IE does not work and that's expected. Is IE6 supposed to work in Wine1 or is it a known not-working program? Should I submit bugs or add to the huge appdb entry? The installer just seems to exit with these lines: fixme:wintrust:CryptCATAdminAcquireContext 0x7e07f02c {f750e6c3-38ee-11d1-85e5-00c04fc295ee} 0 fixme:wintrust:CryptCATAdminAddCatalog 0xdeadbeef L"C:\\windows\\msdownld.tmp\\AS0FA1BB.tmp\\mplay2u.CAT" L"mplay2u.CAT" 0 fixme:wintrust:CryptCATAdminReleaseCatalogContext 0xdeadbeef 0x1 0 fixme:wintrust:CryptCATAdminReleaseContext 0xdeadbeef 0 fixme:msxml:DllCanUnloadNow fixme:advpack:set_ldids Need to support changing paths - default will be used fixme:advpack:set_ldids Need to support changing paths - default will be used lesliev(a)lesliev-laptop:/media/Storage/Software$ err:wineboot:ProcessRunKeys Error running cmd #0 (2) I apologise if I am not following procedure by asking this, I am still learning :-) Les

3 3

Valgrind errors in "Mikołaj Zalewski : comdlg32: Implement CDM_GETFOLDERIDLIST. "
by Dan Kegel 20 Jun '08

20 Jun '08

Hi Mikołaj, your commit seems to have an uninitialized variable; see http://kegel.com/wine/valgrind/logs-2008-06-20/vg-comdlg32_filedlg-diff.txt One of the errors: Conditional jump or move depends on uninitialised value(s) at strlenW (unicode.h:212) by lstrlenW (string.c:384) by GetFileDialog95W (filedlg.c:484) by GetOpenFileNameW (filedlg.c:3891) by test_DialogCancel (filedlg.c:90) by func_filedlg (filedlg.c:128) by run_test (test.h:449) by main (test.h:498) Uninitialised value was created by a stack allocation at test_DialogCancel (filedlg.c:56)

1 0

Re: ntdll: rtlstr.c: Implement checking for control characters in RtlIsTextUnicode [try 4]
by Michael Karcher 20 Jun '08

20 Jun '08

Am Donnerstag, den 19.06.2008, 12:52 -0700 schrieb Zac Brown: > - Because these tests are mutually exclusive, a single goto label suffices for > breaking the loop. As written, your tests are not mutually exclusive. It could happen that flags has IS_TEXT_UNICODE_CONTROLS and IS_TEXT_UNICODE_REVERSE_CONTROLS, set and out_flags contains IS_TEXT_UNICODE_REVERSE_SIGNATURE. (If the caller doesn't provide flags, both flags are set, in fact!) In that case, both conditions of your if statements are true. I don't know whether that means that you need two goto targets or that your logic should be changed, but I suspect the latter. If you test for reversed control characters only I a reversed BOM was found, you should not test for non-reversed control characters if a reversed BOM was found, according to my intuition. Regards, Michael Karcher

3 4

Re: [PATCH] shell32/tests: FOF_MULTIDESTFILES must be set when we are moving multiple files
by James Hawkins 20 Jun '08

20 Jun '08

On Fri, Jun 20, 2008 at 9:37 AM, Vitaly Perov <vitperov(a)etersoft.ru> wrote: > Changelog: > - shell32/tests: FOF_MULTIDESTFILES must be set when we are moving multiple > files > + /* move many files into directory with FOF_MULTIDESTFILES */ + shfo.fFlags |= FOF_MULTIDESTFILES; set_curr_dir_path(from, "test?.txt\0"); set_curr_dir_path(to, "testdir2\0"); - ok(!file_exists("testdir2\\test2.txt"), "The file is not moved yet\n"); - ok(!file_exists("testdir2\\test4.txt"), "The directory is not moved yet\n"); - ok(!SHFileOperationA(&shfo), "Files and directories are moved to directory\n"); - ok(file_exists("testdir2\\test2.txt"), "The file is moved\n"); - ok(file_exists("testdir2\\test4.txt"), "The directory is moved\n"); - ok(file_exists("testdir2\\test4.txt\\test1.txt"), "The file in subdirectory is moved\n"); + retval = SHFileOperationA(&shfo); + todo_wine + { + ok(retval == ERROR_SUCCESS, "Expected ERROR_SUCCESS, got %d\n", retval); + ok(file_exists("testdir2\\test2.txt"), "Expected the file 'test2.txt' to exist\n"); + ok(file_exists("testdir2\\test4.txt"), "Expected the directory 'test4.txt' to exist\n"); + ok(file_exists("testdir2\\test4.txt\\test1.txt"), "Expected the file in subdirectory 'test1.txt' to exist\n"); + } That test was there for a reason. You need to add another test like I wrote in my last email about this. -- James Hawkins

1 0

Re: ntdll: reg.c: Fix return value of NtOpenKey when key handle is NULL
by Zac Brown 20 Jun '08

20 Jun '08

Zac Brown wrote: > Change return value from STATUS_INVALID_PARAMETER to > STATUS_ACCESS_VIOLATION when retkey is null in NtOpenKey and remove > todo_wine from test. > > Tested on Windows XP, Windows 2000 and Windows 2003. > > -Zac Brown > > > ------------------------------------------------------------------------ > > Could anyone give me some feedback on this patch? Talked to AJ about it pre-1.0 and he seemed to think its fine, maybe he'd chime in and let me know what he thought was wrong with it this time around? Comments are greatly appreciated. -Zac

2 2