[PATCH v2] ole32: Avoid calling QueryContinueDrag recursively.

[Roman Pišl]

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=48569

Signed-off-by: Roman Pišl 
---
 dlls/ole32/ole2.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/dlls/ole32/ole2.c b/dlls/ole32/ole2.c
index 757458ad20..e971bcbf34 100644
--- a/dlls/ole32/ole2.c
+++ b/dlls/ole32/ole2.c
@@ -62,6 +62,7 @@ typedef struct tagTrackerWindowInfo
   DWORD        dwOKEffect;
   DWORD*       pdwEffect;
   BOOL       trackingDone;
+  BOOL         inTrackCall;
   HRESULT      returnValue;

BOOL       escPressed;
@@ -766,6 +767,7 @@ HRESULT WINAPI DoDragDrop (
   trackerInfo.dwOKEffect        = dwOKEffect;
   trackerInfo.pdwEffect         = pdwEffect;
   trackerInfo.trackingDone      = FALSE;
+  trackerInfo.inTrackCall       = FALSE;
   trackerInfo.escPressed        = FALSE;
   trackerInfo.curTargetHWND     = 0;
   trackerInfo.curDragTarget     = 0;
@@ -2284,6 +2286,13 @@ static void OLEDD_TrackStateChange(TrackerWindowInfo* trackerInfo)
   HWND   hwndNewTarget = 0;
   POINT pt;

+  /*
+   * This method may be called from QueryContinueDrag again,
+   * (i.e. by running message loop) so avoid recursive call chain.
+   */
+  if (trackerInfo->inTrackCall) return;
+  trackerInfo->inTrackCall = TRUE;
+
   /*
    * Get the handle of the window under the mouse
    */
@@ -2329,6 +2338,8 @@ static void OLEDD_TrackStateChange(TrackerWindowInfo* trackerInfo)
   }
   else
     drag_end( trackerInfo );
+
+  trackerInfo->inTrackCall = FALSE;
 }

/***

[Huw Davies]

On Mon, Feb 17, 2020 at 12:02:01PM +0100, Roman Pišl wrote:
> Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=48569
> Signed-off-by: Roman Pišl
>
> dlls/ole32/ole2.c | 11 +++++++++++
>  1 file changed, 11 insertions(+)

Hi Roman,

So what happened to the tests?

Huw.

[Roman Pišl]

Hi Huw,
my mistake. But after few more experiments I've discovered that the 
actual problem is in the timer. Attached you will find another solution 
for the bug, that seems less invasive and cleaner to me. But now it also 
seems that the test doesn't actually test "reentrancy" but just an 
implementation difference/fault in Wine - timer that could arrive while 
processing QueryContinueDrag. Isn't such a test just overkill?

Which solution is better and how should I proceed?

Thanks.

Roman

Dne 17. 02. 20 v 14:24 Huw Davies napsal(a):
> On Mon, Feb 17, 2020 at 12:02:01PM +0100, Roman Pišl wrote:
> > Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=48569
> > Signed-off-by: Roman Pišl
> >
> > dlls/ole32/ole2.c | 11 +++++++++++
> >   1 file changed, 11 insertions(+)
>
> Hi Roman,
>
> So what happened to the tests?
>
> Huw.

[Marvin]

Hi,

While running your changed tests, I think I found new failures.
Being a bot and all I'm not very good at pattern recognition, so I might be
wrong, but could you please double-check?

Full results can be found at:
https://testbot.winehq.org/JobDetails.pl?Key=65324

Your paranoid android.

=== debian10 (32 bit Chinese:China report) ===

ole32:
clipboard.c:1051: Test failed: OleIsCurrentClipboard returned 0
clipboard.c:1119: Test failed: 1 WM_DRAWCLIPBOARD received