Re: [ros-dev] [ros-diffs] [jimtabor] 50933: [Win32k|User32] - Clarify the meaning and the

[Johannes Anderwald]

>   Author: jimtabor
>  Date: Mon Feb 28 06:31:44 2011
>  New Revision: 50933
>  
>  URL: http://svn.reactos.org/svn/reactos?rev=50933&view=rev
>  Log:
>  [Win32k|User32]
>  - Clarify the meaning and the use of 0, 1 and 2 for user window regions, wine never did.
> Noticing patches not corresponding in correctness and understanding, leaving potential
> crashes in the DCE code, which depend on these types.
>  - Patch based on Rafal Harabien patch, fixes wine msg SetWindowRgn test, see also bug
> 5959. 
This change breaks Winamp 5.601:

(ntoskrnl/ps/thread.c:119) PS: Unhandled Kernel Mode Exception Pointers = 0xF86A8568
(ntoskrnl/ps/thread.c:126) Code c0000005 Addr F87D4C3E Info0 00000000 Info1 00000020 Info2
00000000 Info3 00000020

*** Fatal System Error: 0x0000007e
                       (0xC0000005,0xF87D4C3E,0xF86A8A50,0xF86A8690)

Entered debugger on embedded INT3 at 0x0008:0x808fd45c.bt
Eip:
<ntoskrnl.exe:fd45d (lib/rtl/i386/debug_asm.S:39 (RtlpBreakWithStatusInstruction))>
Frames:
<ntoskrnl.exe:a902 (ntoskrnl/ke/bug.c:1110 (KeBugCheckWithTf@24))>
<ntoskrnl.exe:aa4d (ntoskrnl/ke/bug.c:1438 (KeBugCheckEx@20))>
<ntoskrnl.exe:ead94 (ntoskrnl/ps/thread.c:129
(PspUnhandledExceptionInSystemThread))>
<ntoskrnl.exe:eadaf (ntoskrnl/ps/thread.c:159 (_SEHExcept.39966))>
<ntoskrnl.exe:fa9fa (lib/pseh/i386/framebased-gcchack.c:132 (_SEH2FrameHandler))>
<ntoskrnl.exe:fcecb (lib/rtl/i386/except_asm.s:183 (RtlpExecuteHandler2))>
<ntoskrnl.exe:fce9f (lib/rtl/i386/except_asm.s:150 (RtlpExecuteHandlerForUnwind))>
<ntoskrnl.exe:3222 (ntoskrnl/ke/i386/exp.c:938 (KiDispatchException@20))>
<ntoskrnl.exe:36bd (ntoskrnl/ke/i386/exp.c:1137
(KiDispatchExceptionFromTrapFrame@28))>
<ntoskrnl.exe:6941 (ntoskrnl/include/internal/arch/../i386/ke.h:597
(@KiTrap0EHandler@4))>
<ntoskrnl.exe:f7408 (ntoskrnl/ke/i386/trap.s:99 (VfIsVerificationEnabled@8))>
<win32k.sys:bac39 (subsystems/win32/win32k/objects/region.c:2104
(@RGNOBJAPI_Lock@8))>
<win32k.sys:bb0b3 (subsystems/win32/win32k/objects/region.c:3753
(NtGdiPtInRegion@12))>
<win32k.sys:71bbb (subsystems/win32/win32k/ntuser/msgqueue.c:248
(@co_MsqInsertMouseMessage@4))>
<win32k.sys:5ded4 (subsystems/win32/win32k/ntuser/cursoricon.c:205
(UserSetCursorPos))>
<win32k.sys:66630 (subsystems/win32/win32k/ntuser/input.c:1134 (@IntMouseInput@4))>
<win32k.sys:66cca (subsystems/win32/win32k/ntuser/input.c:194
(@ProcessMouseInputData@8))>
<win32k.sys:66fbf (subsystems/win32/win32k/ntuser/input.c:308 (MouseThreadMain@4))>
<ntoskrnl.exe:eac86 (ntoskrnl/ps/thread.c:156 (PspSystemThreadStartup@8))>
<ntoskrnl.exe:5936 (ntoskrnl/ke/i386/thrdini.c:78 (KiThreadStartup@0))>
<ntoskrnl.exe:eac22 (ntoskrnl/ps/thread.c:625 (PsCreateSystemThread@28))>

WBR,
Johannes

[James Tabor]

Hi!
Try this! I'm not sure since I have -soundhw ac97 running with qemu
and winapp installs but I can not get it ti run.

On Wed, Mar 2, 2011 at 9:53 AM, Johannes Anderwald
<johannes.anderwald(a)reactos.org> wrote:

>   <win32k.sys:bac39
> (subsystems/win32/win32k/objects/region.c:2104 (@RGNOBJAPI_Lock@8))> 

Index: region.c
===================================================================
--- region.c	(revision 50945)
+++ region.c	(working copy)
@@ -2096,6 +2096,12 @@
   PRGN_ATTR pRgn_Attr;
   HANDLE pid;

+  if (hRgn <= HRGN_WINDOW)
+  {
+     DPRINT1("Cought a window system region handle! hRgn %d",hRgn);
+     return NULL;
+  }
+
   pRgn = REGION_LockRgn(hRgn);

   if (pRgn)

[Johannes Anderwald]

>   Hi!
>  Try this! I'm not sure since I have -soundhw ac97 running with qemu
>  and winapp installs but I can not get it ti run.
>  
>  On Wed, Mar 2, 2011 at 9:53 AM, Johannes Anderwald
>  <johannes.anderwald(a)reactos.org> wrote:
>  
>  > <win32k.sys:bac39 (subsystems/win32/win32k/objects/region.c:2104
> (@RGNOBJAPI_Lock@8))> 
Hi,

this patch does not work here. I'm testing with Vbox 4.00

regards,
Johannes

>   
>  
>  Index: region.c
>  ===================================================================
>  --- region.c	(revision 50945)
>  +++ region.c	(working copy)
>  @@ -2096,6 +2096,12 @@
>     PRGN_ATTR pRgn_Attr;
>     HANDLE pid;
>  
>  +  if (hRgn <= HRGN_WINDOW)
>  +  {
>  +     DPRINT1("Cought a window system region handle! hRgn %d",hRgn);
>  +     return NULL;
>  +  }
>  +
>     pRgn = REGION_LockRgn(hRgn);
>  
>     if (pRgn)
>  
>  _______________________________________________
>  Ros-dev mailing list
>  Ros-dev(a)reactos.org
>  http://www.reactos.org/mailman/listinfo/ros-dev

[James Tabor]

Try this!
Index: window.c
===================================================================
--- window.c	(revision 50957)
+++ window.c	(working copy)
@@ -4052,7 +4052,7 @@
    }
    else
    {
-      hrgnCopy = IntSysCreateRectRgnIndirect(&Window->rcWindow); //HRGN_WINDOW;
+      hrgnCopy = NULL;
    }

    if (Window->hrgnClip)
@@ -4061,12 +4061,14 @@
       GreDeleteObject(Window->hrgnClip);
    }

-   if (Window->fnid != FNID_DESKTOP)
-      NtGdiOffsetRgn(hrgnCopy, Window->rcWindow.left, Window->rcWindow.top);
+   if (hrgnCopy)
+   {
+      if (Window->fnid != FNID_DESKTOP)
+         NtGdiOffsetRgn(hrgnCopy, Window->rcWindow.left, Window->rcWindow.top);

-   /* Set public ownership */
-   IntGdiSetRegionOwner(hrgnCopy, GDI_OBJ_HMGR_PUBLIC);
-
+      /* Set public ownership */
+      IntGdiSetRegionOwner(hrgnCopy, GDI_OBJ_HMGR_PUBLIC);
+   }
    Window->hrgnClip = hrgnCopy;

    Ret = co_WinPosSetWindowPos(Window, HWND_TOP, 0, 0, 0, 0, bRedraw
? flags : (flags|SWP_NOREDRAW) );



Apologies to  Rafal Harabien select works like select, also don't
assume the original author is always right.

[James Tabor]

gmail sux!

On Wed, Mar 2, 2011 at 1:59 PM, James Tabor <jimtabor.rosdev(a)gmail.com> wrote:
>   Try this!
>  Index: window.c
>  ===================================================================
>  --- window.c    (revision 50957)
>  +++ window.c    (working copy)
>  @@ -4052,7 +4052,7 @@
>     }
>     else
>     {
>  -      hrgnCopy = IntSysCreateRectRgnIndirect(&Window->rcWindow); //HRGN_WINDOW;
>  +      hrgnCopy = NULL;
>     }
>
>     if (Window->hrgnClip)
>  @@ -4061,12 +4061,14 @@
>        GreDeleteObject(Window->hrgnClip);
>     }
>
>  -   if (Window->fnid != FNID_DESKTOP)
>  -      NtGdiOffsetRgn(hrgnCopy, Window->rcWindow.left, Window->rcWindow.top);
>  +   if (hrgnCopy)
>  +   {
>  +      if (Window->fnid != FNID_DESKTOP)
>  +         NtGdiOffsetRgn(hrgnCopy, Window->rcWindow.left, Window->rcWindow.top);
>
>  -   /* Set public ownership */
>  -   IntGdiSetRegionOwner(hrgnCopy, GDI_OBJ_HMGR_PUBLIC);
>  -
>  +      /* Set public ownership */
>  +      IntGdiSetRegionOwner(hrgnCopy, GDI_OBJ_HMGR_PUBLIC);
>  +   }
>     Window->hrgnClip = hrgnCopy;
>
>     Ret = co_WinPosSetWindowPos(Window, HWND_TOP, 0, 0, 0, 0, bRedraw
>  ? flags : (flags|SWP_NOREDRAW) );
>
>
>
>  Apologies to  Rafal Harabien select works like select, also don't
>  assume the original author is always right.
>

[Johannes Anderwald]

>   Try this!
>  Index: window.c 
Hi!

Unfortunately no change.

regards,
Johannes

[James Tabor]

Hi,
I manage to get winamp 2.95 running, well sort of....

Try this,

Index: window.c
===================================================================
--- window.c	(revision 50962)
+++ window.c	(working copy)
@@ -4068,7 +4068,7 @@
          NtGdiOffsetRgn(hrgnCopy, Window->rcWindow.left, Window->rcWindow.top);

       /* Set public ownership */
-      IntGdiSetRegionOwner(hrgnCopy, GDI_OBJ_HMGR_PUBLIC);
+//      IntGdiSetRegionOwner(hrgnCopy, GDI_OBJ_HMGR_PUBLIC);
    }
    Window->hrgnClip = hrgnCopy;


The ugly gdi debug prints will comeback but it should work, I will
figure out a proper fix ASAP.
Thanks,
James

[James Tabor]

Okay!
Does the newest winamp draw correctly? This patch hides the winamp
drawing bug!!!!! So revision 50933 is not a regression!

Johannes, you need to be more specific and explain in great details....

Thanks,
James


On Fri, Mar 4, 2011 at 11:17 PM, James Tabor <jimtabor.rosdev(a)gmail.com> wrote:
>   Hi,
>  I manage to get winamp 2.95 running, well sort of....
>
>  Try this,
>
>  Index: window.c
>  ===================================================================
>  --- window.c    (revision 50962)
>  +++ window.c    (working copy)
>  @@ -4068,7 +4068,7 @@
>           NtGdiOffsetRgn(hrgnCopy, Window->rcWindow.left, Window->rcWindow.top);
>
>        /* Set public ownership */
>  -      IntGdiSetRegionOwner(hrgnCopy, GDI_OBJ_HMGR_PUBLIC);
>  +//      IntGdiSetRegionOwner(hrgnCopy, GDI_OBJ_HMGR_PUBLIC);
>     }
>     Window->hrgnClip = hrgnCopy;
>
>
>  The ugly gdi debug prints will comeback but it should work, I will
>  figure out a proper fix ASAP.
>  Thanks,
>  James
>

[Johannes Anderwald]

>   Okay!
>  Does the newest winamp draw correctly? This patch hides the winamp
>  drawing bug!!!!! So revision 50933 is not a regression! 
The newest Winamp (5.601) does not draw correctly. The control window is a black square.
The only window which is drawn correctly is playlist window. The bugcheck occurs when
moving the mouse over the black window.


>   
>  Johannes, you need to be more specific and explain in great details.... 
1.) Install MS VC++ 2008 Runtime (Rapps)
2.) Download Winamp from
http://download.nullsoft.com/winamp/client/winamp5601_full_emusic-7plus_de-…
3.) Perform an install with minimal options (all extras disabled)
4.) Run winamp configuration and disable send statistics 
5.) Run winamp (if cpu % 100 try killing it with taskmgr. If that does not work, restart)
6.) Move mouse arround on desktop or near the winamp window.


>   
>  Thanks,
>  James 
PS: I have been testing with 50989, the problem is still there.

regards,
Johannes

[James Tabor]

Just to be clear.... The change is correct and without it, the problem
is just hidden and that leaves the investigator lost without knowing
why these types of application do not work in current trunk.

AIMP draws but has the same crash when clicking inside the window also
the dead window patch I posted goes off like mad!

If you need to test everything without the crash, read this,

http://www.reactos.org/archives/public/ros-dev/2011-March/013902.html

What is not being asked, why did the region lock break before the test
for the region attribute block?

On Mon, Mar 7, 2011 at 10:59 AM, Johannes Anderwald
<johannes.anderwald(a)reactos.org> wrote:
> >   Okay!
> >  Does the newest winamp draw correctly? This patch hides the winamp
> >  drawing bug!!!!! So revision 50933 is not a regression! 
>  The newest Winamp (5.601) does not draw correctly. The control window is a black square.
> The only window which is drawn correctly is playlist window. The bugcheck occurs when
> moving the mouse over the black window.
>
>
> >  
> >  Johannes, you need to be more specific and explain in great details.... 
>  1.) Install MS VC++ 2008 Runtime (Rapps)
>  2.) Download Winamp from
> http://download.nullsoft.com/winamp/client/winamp5601_full_emusic-7plus_de-…
>  3.) Perform an install with minimal options (all extras disabled)
>  4.) Run winamp configuration and disable send statistics
>  5.) Run winamp (if cpu % 100 try killing it with taskmgr. If that does not work,
> restart)
>  6.) Move mouse arround on desktop or near the winamp window.
>
>
> >  
> >  Thanks,
> >  James 
>  PS: I have been testing with 50989, the problem is still there.
>
>  regards,
>  Johannes
>
>  _______________________________________________
>  Ros-dev mailing list
>  Ros-dev(a)reactos.org
>  http://www.reactos.org/mailman/listinfo/ros-dev
>