Ros-dev April 2013

ros-dev@reactos.org

15 participants
17 discussions

Linux builder offline
by Pierre Schweitzer 07 Apr '13

07 Apr '13

Hi, due to hardware issues, the Linux builder is currently offline. Can't provide any ETA for recovery yet. Sorry for the troubles. With my best regards, -- Pierre Schweitzer<pierre at reactos.org> System Administrator ReactOS Foundation

1 0

Re: [ros-dev] [ros-diffs] [jgardou] 58660: [MESA32] * Disable SSE optimizations, as they only cause mayhem.
by Timo Kreuzer 05 Apr '13

05 Apr '13

What exactly does it cause? And shouldn't we rather fix that, instead of disabling optimizations? mesa is already slow enough. Am 03.04.2013 14:02, schrieb jgardou(a)svn.reactos.org: > Author: jgardou > Date: Wed Apr 3 12:02:58 2013 > New Revision: 58660 > > URL: http://svn.reactos.org/svn/reactos?rev=58660&view=rev > Log: > [MESA32] > * Disable SSE optimizations, as they only cause mayhem. > > Modified: > trunk/reactos/dll/opengl/mesa/src/mesa/CMakeLists.txt > > Modified: trunk/reactos/dll/opengl/mesa/src/mesa/CMakeLists.txt > URL: http://svn.reactos.org/svn/reactos/trunk/reactos/dll/opengl/mesa/src/mesa/C… > ============================================================================== > --- trunk/reactos/dll/opengl/mesa/src/mesa/CMakeLists.txt [iso-8859-1] (original) > +++ trunk/reactos/dll/opengl/mesa/src/mesa/CMakeLists.txt [iso-8859-1] Wed Apr 3 12:02:58 2013 > @@ -33,17 +33,18 @@ > x86/3dnow_xform3.S > x86/3dnow_xform4.S > x86/3dnow_normal.S > - x86/sse_xform1.S > - x86/sse_xform2.S > - x86/sse_xform3.S > - x86/sse_xform4.S > - x86/sse_normal.S > + # x86/sse_xform1.S > + # x86/sse_xform2.S > + # x86/sse_xform3.S > + # x86/sse_xform4.S > + # x86/sse_normal.S > x86/read_rgba_span_x86.S) > add_definitions( > -DUSE_X86_ASM > -DUSE_MMX_ASM > -DUSE_3DNOW_ASM > - -DUSE_SSE_ASM) > + # -DUSE_SSE_ASM > + ) > endif() > > list(APPEND SOURCE > > >

5 6

Re: [ros-dev] [ros-diffs] [akhaldi] 58662: [PSDK] * Import DEFAULT_UNREACHABLE from Wine.
by Timo Kreuzer 04 Apr '13

04 Apr '13

The last definition is not correct, it will not compile. You need "default: (void)0;" or "default: break;" And the thing should also go to ntdef.h Am 03.04.2013 23:17, schrieb akhaldi(a)svn.reactos.org: > Author: akhaldi > Date: Wed Apr 3 21:16:57 2013 > New Revision: 58662 > > URL: http://svn.reactos.org/svn/reactos?rev=58662&view=rev > Log: > [PSDK] > * Import DEFAULT_UNREACHABLE from Wine. > > Modified: > trunk/reactos/include/psdk/winnt.h > > Modified: trunk/reactos/include/psdk/winnt.h > URL: http://svn.reactos.org/svn/reactos/trunk/reactos/include/psdk/winnt.h?rev=5… > ============================================================================== > --- trunk/reactos/include/psdk/winnt.h [iso-8859-1] (original) > +++ trunk/reactos/include/psdk/winnt.h [iso-8859-1] Wed Apr 3 21:16:57 2013 > @@ -415,6 +415,15 @@ > typedef DWORD FLONG; > > #define C_ASSERT(expr) extern char (*c_assert(void)) [(expr) ? 1 : -1] > + > +/* Eliminate Microsoft C/C++ compiler warning 4715 */ > +#if defined(_MSC_VER) && (_MSC_VER > 1200) > +# define DEFAULT_UNREACHABLE default: __assume(0) > +#elif defined(__clang__) || (defined(__GNUC__) && ((__GNUC__ > 4) || ((__GNUC__ == 4) && (__GNUC_MINOR__ >= 5)))) > +# define DEFAULT_UNREACHABLE default: __builtin_unreachable() > +#else > +# define DEFAULT_UNREACHABLE default: > +#endif > > #include "intrin.h" > > > >

1 0

Re: [ros-dev] [ros-diffs] [jgardou] 58650: [CMICPL] - This is a C++ module
by Timo Kreuzer 04 Apr '13

04 Apr '13

What about integrating set_cpp() into set_module_type, since that already handles a bunch of other things like UNICODE etc? Am 02.04.2013 22:04, schrieb jgardou(a)svn.reactos.org: > Author: jgardou > Date: Tue Apr 2 20:04:31 2013 > New Revision: 58650 > > URL: http://svn.reactos.org/svn/reactos?rev=58650&view=rev > Log: > [CMICPL] > - This is a C++ module > > Modified: > trunk/reactos/drivers/wdm/audio/drivers/CMIDriver/cpl/CMakeLists.txt > > Modified: trunk/reactos/drivers/wdm/audio/drivers/CMIDriver/cpl/CMakeLists.txt > URL: http://svn.reactos.org/svn/reactos/trunk/reactos/drivers/wdm/audio/drivers/… > ============================================================================== > --- trunk/reactos/drivers/wdm/audio/drivers/CMIDriver/cpl/CMakeLists.txt [iso-8859-1] (original) > +++ trunk/reactos/drivers/wdm/audio/drivers/CMIDriver/cpl/CMakeLists.txt [iso-8859-1] Tue Apr 2 20:04:31 2013 > @@ -1,3 +1,5 @@ > + > +set_cpp() > > add_library(cmicpl SHARED > cmicpl.cpp > @@ -10,6 +12,6 @@ > shell32 > msvcrt > kernel32 > - ntdll) > +) > > add_cd_file(TARGET cmicpl DESTINATION reactos/system32/drivers FOR all) > > >

1 0

Re: [ros-dev] [ros-diffs] [spetreolle] 58631: [NTOSKRNL] Fix all MM regressions.
by Aleksey Bragin 01 Apr '13

01 Apr '13

Please branch ReactOS 1.0 release, it's nearly ready it seems. Regards, Aleksey Bragin On 01.04.2013 2:17, spetreolle(a)svn.reactos.org wrote: > Author: spetreolle > Date: Sun Mar 31 22:17:23 2013 > New Revision: 58631 > > URL: http://svn.reactos.org/svn/reactos?rev=58631&view=rev > Log: > [NTOSKRNL] > Fix all MM regressions. > > Modified: > trunk/reactos/ntoskrnl/mm/ARM3/virtual.c > > Modified: trunk/reactos/ntoskrnl/mm/ARM3/virtual.c > URL: http://svn.reactos.org/svn/reactos/trunk/reactos/ntoskrnl/mm/ARM3/virtual.c… > ============================================================================== > --- trunk/reactos/ntoskrnl/mm/ARM3/virtual.c [iso-8859-1] (original) > +++ trunk/reactos/ntoskrnl/mm/ARM3/virtual.c [iso-8859-1] Sun Mar 31 22:17:23 2013 > @@ -7,6 +7,7 @@ > */ > > /* INCLUDES *******************************************************************/ > +/* So long, and Thanks for All the Fish */ > > #include <ntoskrnl.h> > #define NDEBUG > >

4 4

Blacklisted Tests in Testman
by victor martinez 01 Apr '13

01 Apr '13

Hi guyz! I've been reviewing our current Blacklisted suites at Testman. A lot of those unstable testsuites now are reliable, maybe because ReactOS stability has increased or better Winetests. The following blacklisted suites in the past should go into whitelist: Gdi32:SetMapMode. Now scores: 51-0 Gdi32:SetSysColors. Now scores: 44-0 Gdiplus:graphicspath. Now scores: 434-32 Gdiplus:region. Now scores: 803-6 Kernel32:pipe. Now scores: 33895-124 user32:trackmouseevent. Now scores: 177-12 wininet:urlcache. Now scores: 186-16 The following suites look reliable but from time to time they score different.So they should be kept as Blacklisted: user32:msg user32:win In the other hand two new candidates should go into Blacklisted: msxml3:domdoc ntdll:info Not bad, 7 to whitelist, 2 almost reliable and just 2 into Blacklist. My proposal is to update the lists and check during the following days that these new changes are reliable. If testsuites are a sign of stability, I should say..ReactOS is really stable. Bugged as always.But stable. :)

3 3

[ros-diffs] [hbelusca] 58632: [KERNEL32-CONSRV] Implement (Get/Set)ConsoleDisplayMode. [PSDK] Reorganize wincon.h a little bit and add few new flags from MSDN / Wine's wincon.h (which was up-to-date compared t...
by James Tabor 01 Apr '13

01 Apr '13

Hi, Here is a patch for the missing WM_NCRBUTTONUP. It's wine code. Well it breaks the mouse pointer set position and activates the system menu at the left upper corner. Looks like to many hack being made to bypass unimplemented functions. So stop doing it and fix it. I know~ locating all of them is a big head ache but remember what I always type, don't build code on top of hacks on hacks! James Index: win32ss/user/user32/windows/defwnd.c =================================================================== --- win32ss/user/user32/windows/defwnd.c (revision 58633) +++ win32ss/user/user32/windows/defwnd.c (working copy) @@ -22,6 +22,7 @@ LRESULT DefWndNCHitTest(HWND hWnd, POINT Point); LRESULT DefWndNCLButtonDown(HWND hWnd, WPARAM wParam, LPARAM lParam); LRESULT DefWndNCLButtonDblClk(HWND hWnd, WPARAM wParam, LPARAM lParam); +LRESULT NC_HandleNCRButtonDown( HWND hwnd, WPARAM wParam, LPARAM lParam ); void FASTCALL MenuInitSysMenuPopup(HMENU Menu, DWORD Style, DWORD ClsStyle, LONG HitTest ); void MENU_EndMenu( HWND ); @@ -1101,14 +1102,7 @@ } case WM_NCRBUTTONDOWN: - { - /* in Windows, capture is taken when right-clicking on the caption bar */ - if (wParam == HTCAPTION) - { - SetCapture(hWnd); - } - break; - } + return NC_HandleNCRButtonDown( hWnd, wParam, lParam ); case WM_RBUTTONUP: { Index: win32ss/user/user32/windows/nonclient.c =================================================================== --- win32ss/user/user32/windows/nonclient.c (revision 58633) +++ win32ss/user/user32/windows/nonclient.c (working copy) @@ -1035,6 +1035,14 @@ case HTBOTTOMLEFT: case HTBOTTOMRIGHT: { + /* Old comment: + * "make sure hittest fits into 0xf and doesn't overlap with HTSYSMENU" + * This was previously done by setting wParam=SC_SIZE + wParam - 2 + */ + /* But that is not what WinNT does. Instead it sends this. This + * is easy to differentiate from HTSYSMENU, because HTSYSMENU adds + * SC_MOUSEMENU into wParam. + */ SendMessageW(hWnd, WM_SYSCOMMAND, SC_SIZE + wParam - (HTLEFT - WMSZ_LEFT), lParam); break; } @@ -1073,6 +1081,41 @@ return(0); } +/*********************************************************************** + * NC_HandleNCRButtonDown + * + * Handle a WM_NCRBUTTONDOWN message. Called from DefWindowProc(). + */ +LRESULT NC_HandleNCRButtonDown( HWND hwnd, WPARAM wParam, LPARAM lParam ) +{ + MSG msg; + INT hittest = wParam; + + switch (hittest) + { + case HTCAPTION: + case HTSYSMENU: + if (!GetSystemMenu( hwnd, FALSE )) break; + + SetCapture( hwnd ); + for (;;) + { + if (!GetMessageW( &msg, 0, WM_MOUSEFIRST, WM_MOUSELAST )) break; + if (CallMsgFilterW( &msg, MSGF_MAX )) continue; + if (msg.message == WM_RBUTTONUP) + { + hittest = DefWndNCHitTest( hwnd, msg.pt ); + break; + } + } + ReleaseCapture(); + if (hittest == HTCAPTION || hittest == HTSYSMENU) + SendMessageW( hwnd, WM_SYSCOMMAND, SC_MOUSEMENU + HTSYSMENU, msg.lParam ); + break; + } + return 0; +} + VOID DefWndTrackScrollBar(HWND hWnd, WPARAM wParam, POINT Point) {

2 1

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Ros-dev April 2013