January 2009 - Wine-devel - list.winehq.org

Re: [PATCH 2/2] widl: Optimise the generated code by skipping calls calling the type function directly for some reference pointers.
by Alexandre Julliard 12 Jan '09

12 Jan '09

"Rob Shearman" <robertshearman(a)gmail.com> writes: > Skip calling the Pointer marshalling/unmarshalling/buffer > sizing/freeing function in this case. > > Output code for calling union marshalling/unmarshalling/buffer > sizing/freeing functions. Now it breaks the tests: ../../../tools/runtest -q -P wine -M qmgr.dll -T ../../.. -p qmgr_test.exe.so file.c && touch file.ok wine: Unhandled page fault on read access to 0x001322a8 at address 0x7b86122d (thread 0032), starting debugger... Unhandled exception: page fault on read access to 0x001322a8 in 32-bit code (0x7b86122d). Register dump: CS:0023 SS:002b DS:002b ES:002b FS:0063 GS:006b EIP:7b86122d ESP:0032fcf0 EBP:0032fd18 EFLAGS:00010246( - 00 -RIZP1) EAX:001322a8 EBX:7b8b5818 ECX:7bc904c4 EDX:001322a8 ESI:ffffffff EDI:00000000 Stack dump: 0x0032fcf0: 00000001 0032fd04 00110014 6055221c 0x0032fd00: 00000001 00000409 0032fd18 7b8b5818 0x0032fd10: 60552980 001322a8 0032fd48 7b863944 0x0032fd20: 00000409 00000000 001322a8 ffffffff 0x0032fd30: 60552980 ffffffff 0032fd58 6055221c 0x0032fd40: 00000000 60550060 0032fd78 6054ca20 Backtrace: =>0 0x7b86122d CompareStringW+0x12d(lcid=1033, style=0, str1=0x1322a8, len1=0, str2=0x60552980, len2=-1) [/home/julliard/wine/wine/dlls/kernel32/../../include/wine/unicode.h:216] in kernel32 (0x0032fd18) 1 0x7b863944 lstrcmpW+0x64(str1=<register EDI not in topmost frame>, str2=<register ESI not in topmost frame>) [/home/julliard/wine/wine/dlls/kernel32/locale.c:2830] in kernel32 (0x0032fd48) 2 0x6054ca20 test_GetRemoteName+0x90() [/home/julliard/wine/wine/dlls/qmgr/tests/file.c:127] in qmgr_test (0x0032fd78) 3 0x6054ccf4 func_file+0x274() [/home/julliard/wine/wine/dlls/qmgr/tests/file.c:108] in qmgr_test (0x00320000) 4 0x00000000 (0x00000000) 0x7b86122d CompareStringW+0x12d [/home/julliard/wine/wine/dlls/kernel32/../../include/wine/unicode.h:216] in kernel32: cmpw $0,0x0(%eax) 216 while (*s) s++; -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: rpcrt4/tests: NdrVaryingArrayUnmarshall() is broken on Windows
by Alexandre Julliard 12 Jan '09

12 Jan '09

"Ge van Geldorp" <ge(a)gse.nl> writes: > The failures in rpcrt4:server on Windows are caused by an access violation > thrown in NdrVaryingArrayUnmarshall() called from get_5numbers(). I've > created a minimum get_5numbers() test with Microsoft tools and that throws > the same access violation. The access violation disappears when zeroing > out the StubMsg before calling NdrClientInitializeNew(), so my guess is > Windows NdrVaryingArrayUnmarshall() uses an uninitialized member. > > Changelog: > NdrVaryingArrayUnmarshall() is broken on Windows If it's broken on Windows then there's no point in testing it, you can simply remove it. -- Alexandre Julliard julliard(a)winehq.org

4 5

Re: [4/5] widl: Rewrite of cant_be_null
by Alexandre Julliard 12 Jan '09

12 Jan '09

Michael Karcher <wine(a)mkarcher.dialup.fu-berlin.de> writes: > As Rob Shearman commented to my patch "Null-check pointers with [size_is]", > the code flow in cant_be_null is non-obvious, the variable type check could > be included into this function and last but not least it is buggy. This > is a hopefully cleaner rewrite of this function, which still fixes the > problem the original patch intended to fix, namely null-checking of pointers > declared with size_is. This matches midl behaviour. It still doesn't seem to be doing the right thing on the svcctl client code. The current code matches widl for all functions except QueryServiceStatusEx, yours has a lot more differences. I'd also suggest adding some test cases for this. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [PATCH 2/2] widl: Optimise the generated code by skipping calls calling the type function directly for some reference pointers.
by Alexandre Julliard 12 Jan '09

12 Jan '09

"Rob Shearman" <robertshearman(a)gmail.com> writes: > Skip calling the Pointer marshalling/unmarshalling/buffer > sizing/freeing function in this case. > > Output code for calling union marshalling/unmarshalling/buffer > sizing/freeing functions. It doesn't work: ../../tools/widl/widl -I. -I. -I../../include -I../../include -D__WINESRC__ -DREGISTER_PROXY_DLL -DPROXY_CLSID_IS="{ 0xb8da6310, 0xe19b, 0x11d0, { 0x93, 0x3c, 0x00, 0xa0, 0xc9, 0x0d, 0xca, 0xa9 } }" -p -P actxprxy_servprov_p.c actxprxy_servprov.idl widl: typetree.h:191: type_pointer_get_ref: Assertion `is_ptr(type)' failed. make[2]: *** [actxprxy_servprov_p.c] Aborted make[2]: *** Deleting file `actxprxy_servprov_p.c' -- Alexandre Julliard julliard(a)winehq.org

2 1

WIDL 'import' question
by Nikolay Sivov 12 Jan '09

12 Jan '09

Hi, I've a small question on using 'import' in .idl. Sorry if it's too dummy, it's my first try to use widl. Trying to get commoncontrols.idl port compile on Wine I've got the following problems - this file contains struct definitions already defined in commctrl.h. Adding import "commctrl.h" and removing after that all conflict definitions from commoncontrols.idl caused for me ../tools/widl/widl -I. -I. -I../include -I../include -h -H commoncontrols.h commoncontrols.idl ./prsht.h:39: error: syntax error, unexpected '{' make[1]: *** [commoncontrols.h] Error 1 make[1]: Leaving directory `/home/mrlarch/wine/include' make: *** [include] Error 2 showing that prsht.h isn't processed properly here 39 static const WCHAR WC_PROPSHEETW[] = { 'S','y','s', 40 'P','r','o','p','e','r','t','y','S','h','e','e','t',0 }; How could I get over that?

2 2

Re: [2/9] cryptui: Use "-1" as a "don't care value", fix dialog id duplicates.
by Alexandre Julliard 12 Jan '09

12 Jan '09

Dmitry Timoshkov <dmitry(a)codeweavers.com> writes: > diff --git a/dlls/cryptui/cryptui_En.rc b/dlls/cryptui/cryptui_En.rc > index 119eac6..b5bd067 100644 > --- a/dlls/cryptui/cryptui_En.rc > +++ b/dlls/cryptui/cryptui_En.rc > @@ -274,11 +274,11 @@ BEGIN > PUSHBUTTON "B&rowse...", IDC_IMPORT_BROWSE_FILE, 236,11,60,14 > LTEXT "Note: The following file formats may contain more than one certificate, certificate revocation list, or certificate trust list:", stc2, 21,26,265,16 > LTEXT "Cryptographic Message Syntax Standard/PKCS #7 Messages (.p7b)", > - stc3, 31,49,265,10 > + IDC_STATIC, 31,49,265,10 > LTEXT "Personal Information Exchange/PKCS #12 (.pfx, .p12)", > - stc3, 31,64,265,10 > + IDC_STATIC, 31,64,265,10 > LTEXT "Microsoft Serialized Certificate Store (.sst)", > - stc3, 31,79,265,10 > + IDC_STATIC, 31,79,265,10 > END It would be better to use -1 directly in the resource file and get rid of IDC_STATIC. There's no need to have a define for an ignored value. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [1/2] server: add ability to send a unix signal to arbitrary thread
by Alexandre Julliard 12 Jan '09

12 Jan '09

"Kirill K. Smirnov" <lich(a)math.spbu.ru> writes: > diff --git a/server/protocol.def b/server/protocol.def > index 6d6be5b..35dfcc0 100644 > --- a/server/protocol.def > +++ b/server/protocol.def > @@ -664,6 +664,12 @@ typedef union > int count; /* new suspend count */ > @END > > +/* Send a unix signal to thread */ > +(a)REQ(kill_thread) > + obj_handle_t handle; /* thread handle */ > + int sig; /* unix signal to send */ > +(a)END This does not belong in the server. The dosvm support should be fixed to use Win32 APIs instead of Unix signals. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: user32/tests: keybd_event can send HCBT_KEYSKIPPED notifications
by Dmitry Timoshkov 12 Jan '09

12 Jan '09

"Ge van Geldorp" <ge(a)gse.nl> wrote: > static const struct message WmKeyUpSkippedSeq[] = > { > - { HCBT_KEYSKIPPED, hook|wparam|lparam|optional, 'N', 0xc0000001 }, /* XP */ > + { HCBT_KEYSKIPPED, hook|wparam|optional, 'N' }, /* XP */ > + { 0 } > +}; > +static const struct message WmUserKeyUpSkippedSeq[] = > +{ > + { WM_USER, sent }, > + { HCBT_KEYSKIPPED, hook|wparam|optional, 'N' }, /* XP */ > { 0 } > }; Please don't remove lparam checks, add message sequences for key down messages instead. -- Dmitry.

1 0

Re: [TOOLS] Add Windows 7
by James Hawkins 12 Jan '09

12 Jan '09

On Sat, Jan 10, 2009 at 3:32 AM, Paul Vriens <paul.vriens.wine(a)gmail.com> wrote: > Hi, > > We have our first Windows 7 test results : > http://test.winehq.org/data/a69c86d3f517f659ba47495f77deac2df671fb03/vista_… > > I'm not sure if '7' is a nice/correct header though. Maybe "Win7" ? > > This also means we need to add it to winecfg. > > Changelog > Add Windows 7 > > -- > Cheers, > > Paul. > > > >From b8ec6ca34b85f7ba1ae736ef35ee380b1b5721f0 Mon Sep 17 00:00:00 2001 > From: Paul Vriens <Paul.Vriens.Wine(a)gmail.com> > Date: Sat, 10 Jan 2009 12:28:57 +0100 > Subject: [PATCH] Add Windows 7 > > --- > winetest/build-index | 3 ++- > winetest/dissect | 6 ++++-- > winetest/gather | 6 ++++-- > 3 files changed, 10 insertions(+), 5 deletions(-) > > diff --git a/winetest/build-index b/winetest/build-index > index 2813ae8..0f257ad 100755 > --- a/winetest/build-index > +++ b/winetest/build-index > @@ -27,11 +27,12 @@ my %xp = (name => "XP"); > my %w2k3 = (name => "2003"); > my %vista = (name => "Vista"); > my %w2k8 = (name => "2008"); > +my %w7 = (name => "7"); > my %unknown = (name => "Other"); > my %wine = (name => "Wine"); > I'm behind on my emails this weekend, so I don't know if this has been committed yet or not, but it seems like win7 is a better name than w7. -- James Hawkins

3 3

Re: richedit: Added tests for autoscrolling options based on WS_V/HSCROLL.
by Detlef Riekenberg 12 Jan '09

12 Jan '09

On So, 2009-01-11 at 02:58 -0500, Dylan Smith wrote: > It seems as if the autoscrolling options can be set based on whether > WS_VSCROLL or WS_HSCROLL styles are used to create the window, but this > is done differently between version 1.0 and 3.0 of the richedit control. > > I noticed this indirectly before because wordwrap is initially set only > if the auto-horizontal scrolling is initially enabled. > --- > dlls/riched20/tests/editor.c | 30 ++++++++++++++++++++++++++++-- > dlls/riched32/tests/editor.c | 32 ++++++++++++++++++++++++++++++++ > 2 files changed, 60 insertions(+), 2 deletions(-) Hi Dylan. After testing your first Patch, i tried this one: patching file dlls/riched20/tests/editor.c Hunk #1 succeeded at 5826 (offset 23 lines). Hunk #2 succeeded at 5856 (offset 23 lines). Hunk #3 succeeded at 6281 (offset 23 lines). patching file dlls/riched32/tests/editor.c All patches should apply clean. When you send a Patchset, please mark them with [Patch x/y] In this case, [Patch 1/8], [Patch 2/8], ... And again, new failures in win9x due to the use of UNICODE. (The riched32 test already produce >40 failures in Win9x) I didn't view/test the following Patches. Please inspect them also for UNICODE faults. -- By by ... Detlef

2 1