[Bug 26133] New: d3dx9_36: surface and texture tests shows an invalid read under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26133

Summary: d3dx9_36: surface and texture tests shows an invalid
                    read under valgrind
           Product: Wine
           Version: 1.3.13
          Platform: x86
        OS/Version: Linux
            Status: NEW
          Keywords: download, source, testcase
          Severity: minor
          Priority: P2
         Component: directx-d3dx9
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]
                CC: [e-mail]

surface test:
 Invalid read of size 4
    at  point_filter_simple_data (surface.c:737)
    by  D3DXLoadSurfaceFromMemory (surface.c:842)
    by  D3DXLoadSurfaceFromFileInMemory (surface.c:402)
    by  D3DXLoadSurfaceFromResourceA (surface.c:513)
    by  test_D3DXLoadSurface (surface.c:272)
    by  func_surface (surface.c:540)
    by  run_test (test.h:556)
    by  main (test.h:624)
  Address 0x7f0228f0 is 0 bytes inside a block of size 16 free'd
    at  notify_free (heap.c:262)
    by  RtlFreeHeap (heap.c:1747)
    by  BITMAP_CopyBitmap (bitmap.c:548)
    by  CreateBrushIndirect (brush.c:117)
    by  CreatePatternBrush (brush.c:221)
    by  SYSPARAMS_Init (sysparams.c:956)
    by  process_attach (user_main.c:276)
    by  DllMain (user_main.c:332)
    by  __wine_spec_dll_entry (dll_entry.c:40)
    by  ??? (loader.c:139)
    by  MODULE_InitDLL (loader.c:978)
    by  process_attach (loader.c:1067)
    by  process_attach (loader.c:1059)
    by  process_attach (loader.c:1059)
    by  process_attach (loader.c:1059)
    by  attach_process_dlls (loader.c:2535)
    by  ??? (port.c:60)

texture test:
 Invalid read of size 4
    at  point_filter_simple_data (surface.c:737)
    by  D3DXLoadSurfaceFromMemory (surface.c:842)
    by  D3DXLoadSurfaceFromFileInMemory (surface.c:402)
    by  D3DXCreateTextureFromFileInMemoryEx (texture.c:605)
    by  D3DXCreateTextureFromResourceExA (texture.c:810)
    by  D3DXCreateTextureFromResourceA (texture.c:759)
    by  test_D3DXCreateTexture (texture.c:576)
    by  func_texture (texture.c:1106)
    by  run_test (test.h:556)
    by  main (test.h:624)
  Address 0x7f0171b8 is 32 bytes inside a block of size 56 free'd
    at  notify_free (heap.c:262)
    by  RtlFreeHeap (heap.c:1747)
    by  update_reg_entries (freetype.c:2027)
    by  WineEngInit (freetype.c:2877)
    by  DllMain (gdiobj.c:582)
    by  __wine_spec_dll_entry (dll_entry.c:40)
    by  ??? (loader.c:139)
    by  MODULE_InitDLL (loader.c:978)
    by  process_attach (loader.c:1067)
    by  process_attach (loader.c:1059)
    by  process_attach (loader.c:1059)
    by  process_attach (loader.c:1059)
    by  process_attach (loader.c:1059)
    by  attach_process_dlls (loader.c:2535)
    by  ??? (port.c:60)

[Bug 26133] d3dx9_36: surface and texture tests shows an invalid read under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26133

Matteo Bruni  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #1 from Matteo Bruni  2011-02-15 10:42:37 CST ---
This looks like a real issue and, if I understand it correctly, the fix is not
entirely trivial. I'll soon prepare a patch for it.

[Bug 26133] d3dx9_36: surface and texture tests shows an invalid read under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26133

--- Comment #2 from Matteo Bruni  2011-02-18 12:59:42 CST ---
Created an attachment (id=33338)
 --> (http://bugs.winehq.org/attachment.cgi?id=33338)
Tentative fix

Austin, can you please test this patch with Valgrind? For some reasons Valgrind
now crashes for me when running those tests...

[Bug 26133] d3dx9_36: surface and texture tests shows an invalid read under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26133

--- Comment #3 from Austin English  2011-02-22 00:19:44 CST ---
After that patch, I get:
==7162== 400 bytes in 1 blocks are definitely lost in loss record 676 of 823
==7162==    at 0x4024210: malloc (vg_replace_malloc.c:236)
==7162==    by 0x51BD44D: XGetVisualInfo (in /usr/lib/libX11.so.6.3.0)
==7162==    by 0x578A723: get_visual (fakeglx.c:544)
==7162==    by 0x578BAC3: choose_visual (fakeglx.c:642)
==7162==    by 0x578BD06: Fake_glXChooseVisual (fakeglx.c:1241)
==7162==    by 0x578F9E8: glXChooseVisual (glxapi.c:197)
==7162==    by 0x50DE7CE: X11DRV_WineGL_InitOpenglInfo (opengl.c:318)
==7162==    by 0x50E08B8: has_opengl (opengl.c:519)
==7162==    by 0x50E57E1: X11DRV_wglGetProcAddress (opengl.c:1848)
==7162==    by 0x4CE600D: wglGetProcAddress (opengl.c:378)
==7162==    by 0x4E6B49A: InitAdapters (in
/home/austin/wine-git/dlls/wined3d/wined3d.dll.so)
==7162==    by 0x4E6E66A: wined3d_init (directx.c:5289)
==7162==    by 0x4F12F4E: wined3d_create (wined3d_main.c:93)
==7162==    by 0x4DC8635: Direct3DCreate9 (d3d9_main.c:43)
==7162==    by 0x48FAA67: func_surface (surface.c:521)
==7162==    by 0x49022DA: run_test (test.h:556)
==7162==    by 0x49026CC: main (test.h:624)
==7162==

==7162== 2,800 bytes in 1 blocks are definitely lost in loss record 786 of 823
==7162==    at 0x4024306: realloc (vg_replace_malloc.c:525)
==7162==    by 0x51BD631: XGetVisualInfo (in /usr/lib/libX11.so.6.3.0)
==7162==    by 0x578C5DA: Fake_glXGetFBConfigs (fakeglx.c:1971)
==7162==    by 0x578E889: glXGetFBConfigs (glxapi.c:552)
==7162==    by 0x50E2998: get_formats (opengl.c:939)
==7162==    by 0x50E38A3: X11DRV_ChoosePixelFormat (opengl.c:1222)
==7162==    by 0x4CE6D77: ChoosePixelFormat (painting.c:345)
==7162==    by 0x4E539C8: WineD3D_CreateFakeGLContext (directx.c:283)
==7162==    by 0x4E6D8D4: InitAdapters (directx.c:5067)
==7162==    by 0x4E6E66A: wined3d_init (directx.c:5289)
==7162==    by 0x4F12F4E: wined3d_create (wined3d_main.c:93)
==7162==    by 0x4DC8635: Direct3DCreate9 (d3d9_main.c:43)
==7162==    by 0x48FAA67: func_surface (surface.c:521)
==7162==    by 0x49022DA: run_test (test.h:556)
==7162==    by 0x49026CC: main (test.h:624)
==7162==

but those were there before, looks like a mesa problem. The invalid reads are
fixed for me.

[Bug 26133] d3dx9_36: surface and texture tests shows an invalid read under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26133

--- Comment #4 from Matteo Bruni  2011-02-22 05:19:21 CST ---
Thank you Austin, I'm sending the patch to the mailing list in a moment.

[Bug 26133] d3dx9_36: surface and texture tests shows an invalid read under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26133

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

--- Comment #5 from Austin English  2011-02-23 13:19:00 CST ---
Fixed by
http://source.winehq.org/git/wine.git/?a=commitdiff;h=85a578f7828b7204cb1a8f...

[Bug 26133] d3dx9_36: surface and texture tests shows an invalid read under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26133

Alexandre Julliard  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #6 from Alexandre Julliard  2011-03-04 12:35:19 CST ---
Closing bugs fixed in 1.3.15.

[Bug 26133] d3dx9_36: surface and texture tests shows an invalid read under valgrind

https://bugs.winehq.org/show_bug.cgi?id=26133

Anastasius Focht  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
      Fixed by SHA1|                            |85a578f7828b7204cb1a8f7cbda
                   |                            |929ac9754df13
                 CC|                            |[e-mail]