Community
Participate
Working Groups
Created attachment 279949 [details] Crash file RedHat 7.4 GTK 3.22 OpenJDK 11.0.2 Create an SWT control and implement a DragSource for it (using SWT API), then create an embedded AWT frame and implement a DropTarget for AWT component (using AWT API). Drag and drop from SWT to AWT is broken: the AWT DropTarget does not receive drop events all the time. There are a lot of error messages from GTK and application crashes in few minutes (5-10m) It's broken on GTK3, on GTK2 works fine. Tested versions: SWT 4.6-4.13. Java 8, OpenJDK 11. Snippet to reproduce is below. Run the app and DND the 1st button onto the 3rd one, look at the logs. public static void main(String[] args) { Shell shell = new Shell(SWT.DIALOG_TRIM); shell.setLayout(new GridLayout(3, true)); Button sourceButton = new Button(shell, SWT.PUSH); sourceButton.setText("Source"); Button targetButton = new Button(shell, SWT.PUSH); targetButton.setText("SWT Target"); Composite swingContainer = new Composite(shell, SWT.EMBEDDED); Frame frame = SWT_AWT.new_Frame(swingContainer); JButton swingButton = new JButton("Swing target"); frame.add(swingButton); shell.open(); shell.pack(); DragSource ds = new DragSource(sourceButton, DND.DROP_MOVE); ds.setTransfer(new Transfer[] { TextTransfer.getInstance() }); ds.addDragListener(new DragSourceAdapter() { @Override public void dragSetData(DragSourceEvent event) { event.data = "Hello"; } }); DropTarget dt = new DropTarget(targetButton, DND.DROP_MOVE); dt.setTransfer(new Transfer[] { TextTransfer.getInstance() }); dt.addDropListener(new DropTargetAdapter() { @Override public void drop(DropTargetEvent event) { System.err.println("Drop success: " + event.data); } }); java.awt.dnd.DropTarget awtdt = new java.awt.dnd.DropTarget(swingButton, new DropTargetListener() { @Override public void dropActionChanged(DropTargetDragEvent dtde) { } @Override public void drop(DropTargetDropEvent dtde) { dtde.acceptDrop(DnDConstants.ACTION_MOVE); System.err.println("drop " + dtde.getTransferable()); } @Override public void dragOver(DropTargetDragEvent dtde) { } @Override public void dragExit(java.awt.dnd.DropTargetEvent dte) { } @Override public void dragEnter(DropTargetDragEvent dtde) { System.err.println("Drag enter"); } }); Display display = shell.getDisplay(); while (!shell.isDisposed()) { if (!display.readAndDispatch()) { display.sleep(); } } } Crash log: # # A fatal error has been detected by the Java Runtime Environment: # # SIGSEGV (0xb) at pc=0x00007fe8f957ca03, pid=13753, tid=13754 # # JRE version: OpenJDK Runtime Environment (11.0.2+9) (build 11.0.2+9) # Java VM: OpenJDK 64-Bit Server VM (11.0.2+9, mixed mode, tiered, compressed oops, serial gc, linux-amd64) # Problematic frame: # C [libgtk-3.so.0+0x3afa03] gtk_drag_finish+0x23 # # Core dump will be written. Default location: Core dumps may be processed with "/usr/libexec/abrt-hook-ccpp %s %c %p %u %g %t e %P %I %h" (or dumping to /home/serhii/Workspace/Testing/core.13753) # # If you would like to submit a bug report, please visit: # http://bugreport.java.com/bugreport/crash.jsp # The crash happened outside the Java Virtual Machine in native code. # See problematic frame for where to report the bug. # --------------- S U M M A R Y ------------ Command Line: -Dfile.encoding=UTF-8 DnDTest Host: Intel(R) Core(TM) i7-8700 CPU @ 3.20GHz, 1 cores, 15G, CentOS Linux release 7.4.1708 (Core) Time: Thu Sep 19 07:03:44 2019 EDT elapsed time: 326 seconds (0d 0h 5m 26s) --------------- T H R E A D --------------- Current thread (0x00007fe924010000): JavaThread "main" [_thread_in_native, id=13754, stack(0x00007fe92d489000,0x00007fe92d58a000)] Stack: [0x00007fe92d489000,0x00007fe92d58a000], sp=0x00007fe92d5886e0, free space=1021k Native frames: (J=compiled Java code, A=aot compiled Java code, j=interpreted, Vv=VM code, C=native code) C [libgtk-3.so.0+0x3afa03] gtk_drag_finish+0x23 Java frames: (J=compiled Java code, j=interpreted, Vv=VM code) J 658 org.eclipse.swt.internal.gtk.OS._g_main_context_iteration(JZ)Z (0 bytes) @ 0x00007fe914759314 [0x00007fe9147592c0+0x0000000000000054] J 1052 c2 org.eclipse.swt.widgets.Display.readAndDispatch()Z (77 bytes) @ 0x00007fe91478038c [0x00007fe914780160+0x000000000000022c] j DnDTest.main([Ljava/lang/String;)V+209 v ~StubRoutines::call_stub siginfo: si_signo: 11 (SIGSEGV), si_code: 1 (SEGV_MAPERR), si_addr: 0x0000004000000003
Created attachment 279950 [details] Logs screen
Created attachment 279951 [details] Test application
*** This bug has been marked as a duplicate of bug 548661 ***
Hi Eric. It is not a duplicate. In my case DnD works but broken. Application crashes. Please see attachments
It is not a duplicate. In my case DnD works but broken. Application crashes. Please see attachments
(In reply to Serhiy Davydiuk from comment #5) > It is not a duplicate. In my case DnD works but broken. Application crashes. > Please see attachments Ah sorry I thought they were the same. Thanks for reopening.
Thank you all for fixing https://bugs.eclipse.org/bugs/show_bug.cgi?id=548661. Is it possible to fix this defect too? I've tried to test using swt-4.14rc2 but the issue is still reproducible. Thanks
I've tested the DnD behavior on RHEL 7.2 (gtk 3.14) and it works fine. I'm not sure is it related to gtk3 changes, like in this post https://stackoverflow.com/questions/56111131/drag-and-drop-issues-with-gtk3-20-and-later-desktop-gtk . <br> Testing on Fedora 31 with gtk 3.24 failed, because SWT_AWT bridge seems broken (please see attachments)
Created attachment 281410 [details] Fedora 31, gtk 3.24
What happens is that when drop is completed java.awt.dnd.DropTargetListener.drop(DropTargetDropEvent) gets called immediately, however drop source org.eclipse.swt.dnd.DragSourceListener.dragFinished(DragSourceEvent) is called by GTK with a delay of 5 mins. Because of the delay, I guess, gtk_drag_finish accesses a stale reference since by that time the context is invalid. This behavior started appearing in GTK 3.19.7 after this commit https://gitlab.gnome.org/GNOME/gtk/commit/bfee45e6f9f13506f97c8765255c7dc0dd698420 Though there are a few other commits before this one related to DnD as well. Here is a note from 3.19.7 release notes: Drag-and-Drop has been rearchitected to move the input handling into GDK, where different backends can handle it according to their needs and capabilities. The new, 'managed' DND is currently implemented for X11 and Wayland.
gtk_drag_finish is called in a timeout handler (hardcoded to 5 mins) Thread #28 [java] 2418 [core: 3] (Suspended : Breakpoint) gtk_drag_finish() at gtkdnd.c:942 0x7f0260df5b09 gtk_drag_drop_finished() at gtkdnd.c:2,976 0x7f0260df5fbb gtk_drag_abort_timeout() at gtkdnd.c:3,547 0x7f0260df6197 gdk_threads_dispatch() at gdk.c:720 0x7f026078ce78 g_timeout_dispatch() at 0x7f0221e10a7d g_main_context_dispatch() at 0x7f0221e10049 g_main_context_iterate.isra.19() at 0x7f0221e103a8 g_main_context_iteration() at 0x7f0221e1045c Java_org_eclipse_swt_internal_gtk_OS__1g_1main_1context_1iteration() at 0x7f0261428181 0x7f027c77f509 <...more frames...> There are 2 issues here: 1. Crash when handling timeout. This needs be fixed in GTK. There is a bug in redhat with the same back trace. https://bugzilla.redhat.com/show_bug.cgi?format=multiple&id=1769799 2. gtk_drag_finish should be called without a timeout in the first place so that listeners should be notified in a timely manner. After java.awt.dnd.DropTargetListener.drop(DropTargetDropEvent) is handled, AWT sends a ClientMessage via XSendEvent. Then GDK receives an event from X11, emits GDK_DROP_FINISH event, dispatches it, and calls g_signal_emit_by_name (context, "dnd-finished") See gdk_dnd_handle_drop_finished() function in gdkdnd-x11.c. However no callback for this signal is invoked.
Created attachment 282205 [details] suggested solution
The problem with proxy context and window that are created whenever we have swt -> swind dnd. They did refactoring and they did not take into account such proxy changes. Before the chnage all such logic was handled by gtkdnd.c. Now they moved some drop functions call to gdkdnd-x11.c. They connected the logic with gtkdnd.c by g_signal_connect (context, .....) functions. The problem is that context is changed together with the proxy context and that's why finction gtk_drag_context_dnd_finished_cb is not called(the callback was registered with different context on drag start). See attched patch. Also I would suggest to take a look at gtk_drag_context_action_cb function which has never received right context that belongs to awt proxy.
Created attachment 282250 [details] dnd animation window did not disposed properly Another problem with dnd windows created under x11 contexts. These don't removed properly. As a result we have a lot of such windows after several dnd operations. Sucggested solution should resolve such problem.
(In reply to Ilya Mykhaylov from comment #13) > The problem with proxy context and window that are created whenever we have > swt -> swind dnd. > They did refactoring and they did not take into account such proxy changes. > Before the chnage all such logic was handled by gtkdnd.c. > Now they moved some drop functions call to gdkdnd-x11.c. They connected the > logic with gtkdnd.c by g_signal_connect (context, .....) functions. > The problem is that context is changed together with the proxy context and > that's why finction gtk_drag_context_dnd_finished_cb is not called(the > callback was registered with different context on drag start). > > See attched patch. > > Also I would suggest to take a look at gtk_drag_context_action_cb function > which has never received right context that belongs to awt proxy. ilyamykhayl@gmail.com Is the patch appropriate for adjusting SWT-AWT, or should it be upstreamed to the GTK repo? Do you have any plans to upstream your patches? Can you reproduce this issue outside of SWT-AWT?
