[Bug 26131] New: dsound/duplex tests show an uninitialized variable under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26131

Summary: dsound/duplex tests show an uninitialized variable
                    under valgrind
           Product: Wine
           Version: 1.3.13
          Platform: x86
        OS/Version: Linux
            Status: NEW
          Keywords: download, source, testcase
          Severity: normal
          Priority: P2
         Component: directx-dsound
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]

Conditional jump or move depends on uninitialised value(s)
    at  snd_pcm_mmap_begin (in /usr/lib/libasound.so.2.0.0)
    by  CreateMMAP (dscapture.c:513)
    by  IDsCaptureDriverBufferImpl_SetFormat (dscapture.c:737)
    by  IDsCaptureDriverImpl_CreateCaptureBuffer (dscapture.c:1039)
    by  IDirectSoundCaptureBufferImpl_Create (capture.c:935)
    by  IDirectSoundCaptureImpl_CreateCaptureBuffer (capture.c:1256)
    by  IDirectSoundFullDuplexImpl_Initialize (duplex.c:592)
    by  DirectSoundFullDuplexCreate (duplex.c:737)
    by  IDirectSoundFullDuplex_tests (duplex.c:184)
    by  func_duplex (duplex.c:239)
    by  run_test (test.h:556)
    by  main (test.h:624)
  Uninitialised value was created by a stack allocation
    at  CreateMMAP (dscapture.c:460)

though perhaps that's an alsa bug?

[Bug 26131] dsound/duplex tests show an uninitialized variable under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26131

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Blocks|                            |26132

[Bug 26131] dsound/duplex tests show an uninitialized variable under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26131

Raymond  changed:

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

--- Comment #1 from Raymond  2011-03-28 02:46:51 CDT ---
(In reply to comment #0)
> Conditional jump or move depends on uninitialised value(s)
>     at  snd_pcm_mmap_begin (in /usr/lib/libasound.so.2.0.0)
>     by  CreateMMAP (dscapture.c:513)
>     by  IDsCaptureDriverBufferImpl_SetFormat (dscapture.c:737)
>     by  IDsCaptureDriverImpl_CreateCaptureBuffer (dscapture.c:1039)
>     by  IDirectSoundCaptureBufferImpl_Create (capture.c:935)
>     by  IDirectSoundCaptureImpl_CreateCaptureBuffer (capture.c:1256)
>     by  IDirectSoundFullDuplexImpl_Initialize (duplex.c:592)
>     by  DirectSoundFullDuplexCreate (duplex.c:737)
>     by  IDirectSoundFullDuplex_tests (duplex.c:184)
>     by  func_duplex (duplex.c:239)
>     by  run_test (test.h:556)
>     by  main (test.h:624)
>   Uninitialised value was created by a stack allocation
>     at  CreateMMAP (dscapture.c:460)
>
> though perhaps that's an alsa bug?

Are your sure ?

since pdbi->mmap is already assigned before calling CreateMMAP()

This->mmap = mmap;

snd_pcm_prepare(This->pcm);
    CreateMMAP(This);

The bug is actually mmap_mode

mmap_mode = snd_pcm_type(pcm);

if (mmap_mode == SND_PCM_TYPE_HW)
        TRACE("mmap'd buffer is a direct hardware buffer.\n");
    else if (mmap_mode == SND_PCM_TYPE_DMIX)
        TRACE("mmap'd buffer is an ALSA dmix buffer\n");
    else
        TRACE("mmap'd buffer is an ALSA type %d buffer\n", mmap_mode);

since mmap_mode is not related to access_mmap (e.g. pulse does not support mmap
but the trace still say

"trace:dsalsa:CreateMMAP mmap'd buffer is an ALSA type 28 buffer"

type 28 is SND_PCM_TYPE_IOPLUG ( it can be jack or pulse plugin )

err = snd_pcm_hw_params_set_access (pcm, hw_params,
SND_PCM_ACCESS_MMAP_INTERLEAVED);
    if (err < 0)
    {
        err = snd_pcm_hw_params_set_access (pcm, hw_params,
SND_PCM_ACCESS_RW_INTERLEAVED);
        if (err < 0) { WARN("Could not set access\n"); goto err; }
        mmap = 0;
    }
    else
        mmap = 1;

[Bug 26131] dsound/duplex tests show an uninitialized variable under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26131

--- Comment #2 from Austin English  2012-09-17 14:25:15 CDT ---
There's still a couple problems in wine-1.5.13-63-gcaf6fe0 / valgrind 3.7.0:

==29342== Invalid free() / delete / delete[] / realloc()
==29342==    at 0x7BC4BB9A: RtlReAllocateHeap (heap.c:269)
==29342==    by 0x51C2C95: DirectSoundDevice_RemoveBuffer (dsound.c:1106)
==29342==    by 0x51BA925: secondarybuffer_destroy (buffer.c:975)
==29342==    by 0x51BACAB: IDirectSoundBufferImpl_Release (buffer.c:342)
==29342==    by 0x4906843: func_duplex (duplex.c:329)
==29342==    by 0x48ED376: main (test.h:557)
==29342==  Address 0x7f06c6e8 is 0 bytes inside a block of size 4 alloc'd
==29342==    at 0x7BC4AF30: RtlAllocateHeap (heap.c:254)
==29342==    by 0x51C2BA1: DirectSoundDevice_AddBuffer (dsound.c:1066)
==29342==    by 0x51BA6A6: IDirectSoundBufferImpl_Create (buffer.c:952)
==29342==    by 0x51C21CB: DirectSoundDevice_CreateSoundBuffer (dsound.c:997)
==29342==    by 0x51C277F: IDirectSound8Impl_CreateSoundBuffer (dsound.c:242)
==29342==    by 0x51C61FF: IDirectSoundFullDuplexImpl_Initialize (duplex.c:216)
==29342==    by 0x490633B: func_duplex (duplex.c:286)
==29342==    by 0x48ED376: main (test.h:557)
==29342==

==29342== Invalid free() / delete / delete[] / realloc()
==29342==    at 0x7BC4BB9A: RtlReAllocateHeap (heap.c:269)
==29342==    by 0x51C2C95: DirectSoundDevice_RemoveBuffer (dsound.c:1106)
==29342==    by 0x51BA925: secondarybuffer_destroy (buffer.c:975)
==29342==    by 0x51C08B3: DirectSoundDevice_Release (dsound.c:670)
==29342==    by 0x51C0B07: directsound_destroy (dsound.c:123)
==29342==    by 0x51C0CBE: IUnknownImpl_Release (dsound.c:183)
==29342==    by 0x51C5904: fullduplex_destroy (duplex.c:60)
==29342==    by 0x51C5A4E: IDirectSoundFullDuplexImpl_Release (duplex.c:175)
==29342==    by 0x4905C2C: IDirectSoundFullDuplex_test.isra.0 (duplex.c:101)
==29342==    by 0x4906C05: func_duplex (duplex.c:190)
==29342==    by 0x48ED376: main (test.h:557)
==29342==  Address 0x7f004148 is 8 bytes inside a block of size 355 free'd
==29342==    at 0x7BC4A514: RtlFreeHeap (heap.c:1764)
==29342==    by 0x7BC3DD9F: wine_nt_to_unix_file_name (directory.c:3002)
==29342==    by 0x7BC3E661: nt_to_unix_file_name_attr (directory.c:2831)
==29342==    by 0x7BC41B11: FILE_CreateFile (file.c:122)
==29342==    by 0x7BC42043: NtOpenFile (file.c:235)
==29342==    by 0x7BC294E5: open_nt_file (actctx.c:1601)
==29342==    by 0x7BC30D4C: get_manifest_in_associated_manifest.constprop.9
(actctx.c:1811)
==29342==    by 0x7BC32302: RtlCreateActivationContext (actctx.c:2310)
==29342==    by 0x7BC325EE: actctx_init (actctx.c:2223)
==29342==    by 0x7BC53C5B: LdrInitializeThunk (loader.c:2697)
==29342==    by 0x7B863B07: __wine_kernel_init (process.c:1255)
==29342==    by 0x7BC544D2: __wine_process_init (loader.c:2918)
==29342==    by 0x40370C1: wine_init (loader.c:831)
==29342==    by 0x7BF00EBA: main (main.c:230)
==29342==

[Bug 26131] dsound/duplex tests show an uninitialized variable under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26131

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
      Fixed by SHA1|                            |4e06eead8f446f0167e6c92f0cc
                   |                            |09f943122a7cd
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

--- Comment #3 from Austin English  2012-09-26 14:09:23 CDT ---
Fixed by
http://source.winehq.org/git/wine.git/commitdiff/4e06eead8f446f0167e6c92f0cc...

[Bug 26131] dsound/duplex tests show an uninitialized variable under valgrind

http://bugs.winehq.org/show_bug.cgi?id=26131

Alexandre Julliard  changed:

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

--- Comment #4 from Alexandre Julliard  2012-09-28 13:43:27 CDT ---
Closing bugs fixed in 1.5.14.