[Bug 36262] New: valgrind shows a possible leak in dinput/tests/dinput.c

https://bugs.winehq.org/show_bug.cgi?id=36262

Bug ID: 36262
           Summary: valgrind shows a possible leak in
                    dinput/tests/dinput.c
           Product: Wine
           Version: 1.7.18
          Hardware: x86
                OS: Linux
            Status: NEW
          Keywords: download, source, testcase
          Severity: normal
          Priority: P2
         Component: directx-dinput
          Assignee: [e-mail]
          Reporter: [e-mail]

==2259== 76 bytes in 1 blocks are possibly lost in loss record 341 of 659
==2259==    at 0x7BC4C735: notify_alloc (heap.c:255)
==2259==    by 0x7BC50F79: RtlAllocateHeap (heap.c:1716)
==2259==    by 0x49A7627: create_directinput_instance (dinput_main.c:112)
==2259==    by 0x49A7861: DirectInputCreateEx (dinput_main.c:154)
==2259==    by 0x49A7933: DirectInputCreateA (dinput_main.c:177)
==2259==    by 0x49590C0: test_DirectInputJoyConfig8 (dinput.c:575)
==2259==    by 0x495930D: func_dinput (dinput.c:625)
==2259==    by 0x495F2A4: run_test (test.h:584)
==2259==    by 0x495F693: main (test.h:654)
==2259==

[Bug 36262] valgrind shows a possible leak in dinput/tests/dinput.c

https://bugs.winehq.org/show_bug.cgi?id=36262

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |valgrind

[Bug 36262] valgrind shows a possible leak in dinput/tests/dinput.c

http://bugs.winehq.org/show_bug.cgi?id=36262

Bruno Jesus  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
      Fixed by SHA1|                            |1242417d131ae9fafd66ee79e96
                   |                            |352a1e0215fc6
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #1 from Bruno Jesus  ---
Fixed by
http://source.winehq.org/git/wine.git/?a=commit;h=1242417d131ae9fafd66ee79e9...

[Bug 36262] valgrind shows a possible leak in dinput/tests/dinput.c

https://bugs.winehq.org/show_bug.cgi?id=36262

--- Comment #2 from Austin English  ---
(In reply to Bruno Jesus from comment #1)
> Fixed by
> http://source.winehq.org/git/wine.git/?a=commit;
> h=1242417d131ae9fafd66ee79e96352a1e0215fc6

I still see:
==515== 76 bytes in 1 blocks are possibly lost in loss record 348 of 665
==515==    at 0x7BC49369: notify_alloc (heap.c:255)
==515==    by 0x7BC4D59F: RtlAllocateHeap (heap.c:1716)
==515==    by 0x4DEAAB2: create_directinput_instance (dinput_main.c:112)
==515==    by 0x4DEAC96: DirectInputCreateEx (dinput_main.c:154)
==515==    by 0x4DEAD53: DirectInputCreateA (dinput_main.c:177)
==515==    by 0x4CBCAEE: test_DirectInputJoyConfig8 (dinput.c:575)
==515==    by 0x4CBCD1F: func_dinput (dinput.c:626)
==515==    by 0x4CC244C: run_test (test.h:584)
==515==    by 0x4CC2818: main (test.h:654)
==515==

in wine-1.7.22-24-gcf50d5f (though it's very possible that there are multiple
issues here). I can open a new bug if you'd like.

[Bug 36262] valgrind shows a possible leak in dinput/tests/dinput.c

https://bugs.winehq.org/show_bug.cgi?id=36262

Bruno Jesus  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
                 CC|                            |[e-mail]
         Resolution|FIXED                       |---

--- Comment #3 from Bruno Jesus  ---
(In reply to Austin English from comment #2)
> I still see:
> in wine-1.7.22-24-gcf50d5f (though it's very possible that there are
> multiple issues here). I can open a new bug if you'd like.

I possibly misread my valgrind output then, I'll take a better look.

[Bug 36262] valgrind shows a possible leak in dinput/tests/dinput.c

https://bugs.winehq.org/show_bug.cgi?id=36262

--- Comment #4 from Austin English  ---
(In reply to Austin English from comment #2)
> (In reply to Bruno Jesus from comment #1)
> > Fixed by
> > http://source.winehq.org/git/wine.git/?a=commit;
> > h=1242417d131ae9fafd66ee79e96352a1e0215fc6
>
> I still see:
> ==515== 76 bytes in 1 blocks are possibly lost in loss record 348 of 665
> ==515==    at 0x7BC49369: notify_alloc (heap.c:255)
> ==515==    by 0x7BC4D59F: RtlAllocateHeap (heap.c:1716)
> ==515==    by 0x4DEAAB2: create_directinput_instance (dinput_main.c:112)
> ==515==    by 0x4DEAC96: DirectInputCreateEx (dinput_main.c:154)
> ==515==    by 0x4DEAD53: DirectInputCreateA (dinput_main.c:177)
> ==515==    by 0x4CBCAEE: test_DirectInputJoyConfig8 (dinput.c:575)
> ==515==    by 0x4CBCD1F: func_dinput (dinput.c:626)
> ==515==    by 0x4CC244C: run_test (test.h:584)
> ==515==    by 0x4CC2818: main (test.h:654)
> ==515==
>
> in wine-1.7.22-24-gcf50d5f (though it's very possible that there are
> multiple issues here). I can open a new bug if you'd like.

Still in 1.7.41.

[Bug 36262] valgrind shows a possible leak in dinput/tests/dinput.c

https://bugs.winehq.org/show_bug.cgi?id=36262

--- Comment #5 from Bruno Jesus  ---
Created attachment 51479
  --> https://bugs.winehq.org/attachment.cgi?id=51479
test hack

Something is holding the reference to the interface so when
IDirectInput_Release(pDI) is called it does not free the memory, a second call
frees the memory but I feel it's not the correct fix =)