[Bug 36597] New: valgrind shows several definite leaks in user32/tests/msg.c

https://bugs.winehq.org/show_bug.cgi?id=36597

Bug ID: 36597
           Summary: valgrind shows several definite leaks in
                    user32/tests/msg.c
           Product: Wine
           Version: 1.7.19
          Hardware: x86
                OS: Linux
            Status: NEW
          Keywords: download, source, testcase
          Severity: normal
          Priority: P2
         Component: user32
          Assignee: [e-mail]
          Reporter: [e-mail]

==27824== 28 bytes in 1 blocks are definitely lost in loss record 294 of 1,006
==27824==    at 0x7BC4C6DF: notify_alloc (heap.c:255)
==27824==    by 0x7BC50F23: RtlAllocateHeap (heap.c:1716)
==27824==    by 0x5163EB8: DIALOG_get_info (defdlg.c:342)
==27824==    by 0x51697B0: DIALOG_CreateIndirect (dialog.c:675)
==27824==    by 0x5169BF0: CreateDialogIndirectParamAorW (dialog.c:763)
==27824==    by 0x5169C41: CreateDialogIndirectParamA (dialog.c:772)
==27824==    by 0x5169AF4: CreateDialogParamA (dialog.c:737)
==27824==    by 0x4D388A8: test_dialog_messages (msg.c:11563)
==27824==    by 0x4D4745E: func_msg (msg.c:14588)
==27824==    by 0x4DA6978: run_test (test.h:584)
==27824==    by 0x4DA6D67: main (test.h:654)
==27824==

==27824== 28 bytes in 1 blocks are definitely lost in loss record 296 of 1,006
==27824==    at 0x7BC4C6DF: notify_alloc (heap.c:255)
==27824==    by 0x7BC50F23: RtlAllocateHeap (heap.c:1716)
==27824==    by 0x5163EB8: DIALOG_get_info (defdlg.c:342)
==27824==    by 0x51697B0: DIALOG_CreateIndirect (dialog.c:675)
==27824==    by 0x5169BF0: CreateDialogIndirectParamAorW (dialog.c:763)
==27824==    by 0x5169C41: CreateDialogIndirectParamA (dialog.c:772)
==27824==    by 0x5169AF4: CreateDialogParamA (dialog.c:737)
==27824==    by 0x4D38CA5: test_EndDialog (msg.c:11609)
==27824==    by 0x4D47463: func_msg (msg.c:14589)
==27824==    by 0x4DA6978: run_test (test.h:584)
==27824==    by 0x4DA6D67: main (test.h:654)
==27824==

==27824== 68 bytes in 1 blocks are definitely lost in loss record 504 of 1,006
==27824==    at 0x7BC4C6DF: notify_alloc (heap.c:255)
==27824==    by 0x7BC50F23: RtlAllocateHeap (heap.c:1716)
==27824==    by 0x7BC32765: RtlCreateActivationContext (actctx.c:4506)
==27824==    by 0x7B827590: CreateActCtxW (actctx.c:124)
==27824==    by 0x4D261D0: test_create (msg.c:6876)
==27824==    by 0x4D26E87: test_interthread_messages (msg.c:7004)
==27824==    by 0x4D473D9: func_msg (msg.c:14568)
==27824==    by 0x4DA6978: run_test (test.h:584)
==27824==    by 0x4DA6D67: main (test.h:654)
==27824==

[Bug 36597] valgrind shows several definite leaks in user32/tests/msg.c

https://bugs.winehq.org/show_bug.cgi?id=36597

Austin English  changed:

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

[Bug 36597] valgrind shows several definite leaks in user32/tests/msg.c

https://bugs.winehq.org/show_bug.cgi?id=36597

Sebastian Lackner  changed:

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

--- Comment #1 from Sebastian Lackner  ---
The last one should already be fixed by the recent patches, can you confirm
this?

* http://source.winehq.org/patches/data/105217
* http://source.winehq.org/patches/data/105222
* http://source.winehq.org/patches/data/105223

[Bug 36597] valgrind shows several definite leaks in user32/tests/msg.c

https://bugs.winehq.org/show_bug.cgi?id=36597

--- Comment #2 from Austin English  ---
(In reply to Sebastian Lackner from comment #1)
> The last one should already be fixed by the recent patches, can you confirm
> this?
> - http://source.winehq.org/patches/data/105217
> - http://source.winehq.org/patches/data/105222
> - http://source.winehq.org/patches/data/105223

It's not:
==1365== 68 bytes in 1 blocks are definitely lost in loss record 517 of 1,047
==1365==    at 0x7BC49369: notify_alloc (heap.c:255)
==1365==    by 0x7BC4D59F: RtlAllocateHeap (heap.c:1716)
==1365==    by 0x7BC31C49: RtlCreateActivationContext (actctx.c:4506)
==1365==    by 0x7B827493: CreateActCtxW (actctx.c:124)
==1365==    by 0x4DDE030: test_create (msg.c:6876)
==1365==    by 0x4DDEB9F: test_interthread_messages (msg.c:7004)
==1365==    by 0x4DF9625: func_msg (msg.c:14568)
==1365==    by 0x4E4B87B: run_test (test.h:584)
==1365==    by 0x4E4BC47: main (test.h:654)
==1365==

[Bug 36597] valgrind shows several definite leaks in user32/tests/msg.c

https://bugs.winehq.org/show_bug.cgi?id=36597

--- Comment #3 from Sebastian Lackner  ---
Created attachment 48864
  --> https://bugs.winehq.org/attachment.cgi?id=48864
user32/tests: Add a missing call to ReleaseActCtx().

Ah, you're right. Didn't notice they made the same mistake as in the other
tests, sorry :/ Does this one finally fix it?

[Bug 36597] valgrind shows several definite leaks in user32/tests/msg.c

https://bugs.winehq.org/show_bug.cgi?id=36597

--- Comment #4 from Austin English  ---
(In reply to Sebastian Lackner from comment #3)
> Created attachment 48864 [details]
> user32/tests: Add a missing call to ReleaseActCtx().
>
> Ah, you're right. Didn't notice they made the same mistake as in the other
> tests, sorry :/ Does this one finally fix it?

Works, thanks.

[Bug 36597] valgrind shows several definite leaks in user32/tests/msg.c

https://bugs.winehq.org/show_bug.cgi?id=36597

--- Comment #5 from Sebastian Lackner  ---
Regarding the new valgrind complains you pointed out on IRC (
http://pastebin.com/UQ7Qjagi ): They don't seem to be related at my opinion,
except there is also an additional test failure. Is there any?

BTW: Do you already have any clue why we have all the test failures in
user32/tests/msg.c when running with Valgrind (I assume they aren't there
without Valgrind on your machine)? It looks like they start even before the
first Valgrind warning, which is a bit weird?!

[Bug 36597] valgrind shows several definite leaks in user32/tests/msg.c

https://bugs.winehq.org/show_bug.cgi?id=36597

--- Comment #6 from Austin English  ---
(In reply to Sebastian Lackner from comment #5)
> Regarding the new valgrind complains you pointed out on IRC (
> http://pastebin.com/UQ7Qjagi ): They don't seem to be related at my opinion,
> except there is also an additional test failure. Is there any?
>
> BTW: Do you already have any clue why we have all the test failures in
> user32/tests/msg.c when running with Valgrind (I assume they aren't there
> without Valgrind on your machine)? It looks like they start even before the
> first Valgrind warning, which is a bit weird?!

Yeah, some test failures:
msg.c:9145: Test failed: ScrollWindowEx: 10: the msg sequence is not complete:
expected 0000 - actual 001f
msg.c:6976: Test failed: MsgWaitForMultipleObjects failed 102
msg.c:13540: Test failed: key press only: 0: the msg sequence is not complete:
expected 0100 - actual 0000
msg.c:13546: Test failed: unexpected key state 0
msg.c:13552: Test failed: key press repeat: 0: the msg sequence is not
complete: expected 0100 - actual 0000
msg.c:13558: Test failed: unexpected key state 0
msg.c:13564: Test failed: key release repeat: 0: the msg sequence is not
complete: expected 0101 - actual 0000

(those may be timing related, though).

[Bug 36597] valgrind shows several definite leaks in user32/tests/msg.c

https://bugs.winehq.org/show_bug.cgi?id=36597

--- Comment #7 from Sebastian Lackner  ---
I have submitted the patch because its very unlikely (=almost impossible) that
its related. The other failures are all in completely different areas of the
test.

[Bug 36597] valgrind shows several definite leaks in user32/tests/msg.c

https://bugs.winehq.org/show_bug.cgi?id=36597

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
      Fixed by SHA1|                            |34f8053efe0c1e7ec11fd98e3de
                   |                            |78e07cad6d335
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #8 from Austin English  ---
https://source.winehq.org/git/wine.git/commitdiff/34f8053efe0c1e7ec11fd98e3d...

[Bug 36597] valgrind shows several definite leaks in user32/tests/msg.c

https://bugs.winehq.org/show_bug.cgi?id=36597

Alexandre Julliard  changed:

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

--- Comment #9 from Alexandre Julliard  ---
Closing bugs fixed in 1.7.21.