Mouse sharing stops working when the vmware accelerant SET_CURSOR_SHAPE hook is invoked.

[OscarL]

This one bug has been driving me nuts for months. Seemingly at random, mouse-sharing would stop working until the VM got restarted.

Turns out that just by hovering the mouse cursor over Pe's "view splitter", that tiny little bar sitting over the scrollbar (see Pe's PSplitter.{h,cpp} files), was enough to trigger the bug.

That hovering causes a cursor shape change, but not via BView::SetViewCursor(), which apparently does not involves calling the B_SET_CURSOR_SHAPE accelerant hook, but via BApplication::SetCursor(), which does end up hitting said hook.

---

TL;DR: vmware_video's accelerant's SET_CURSOR_SHAPE() hook somehow breaks mouse sharing when invoked.

---

To confirm that the issue is with said accelerant hook, I just disabled it, and can't trigger the breakage anymore. The cursor still changes shape when hovering the Pe view splitter, so I'm yet to see any downsides in just disabling the problematic hook (will open a PR doing just that next).

[OscarL]

Pe uses be_app->SerCursor() in several places, but only one of those calls is the one that triggers the issue.

This suggest that whatever the issue is with SET_CURSOR_SHAPE is related to the cursor data being passed to it. Edit: erroneous conclusion, as it seems be_app->SerCursor() do not even call the accelerant hook at all for B_HAND_CURSOR or B_I_BEAM_CURSOR

B_HAND_CURSOR and B_I_BEAM_CURSOR do not trigger the issue, but be_app->SetCursor(PSplitter::Cursor()); and/or be_app->SetCursor(Cursor()); do. Edit: because the former do not call SET_CURSOR_SHAPE() at all.

The cursor in question is defined in Pe's Cursor.r. data looks coorrect on both Pe and when dumped from SET_CURSOR_SHAPE().

---

So... just disabling the darn hook seems like a good enough solution to me, unless someone else wants to investigate things further.

[OscarL]

Cortex's be_app->SetCursor(M_CABLE_CURSOR); also triggers the bug, as does libprint's be_app->SetCursor(ZOOM_IN); or be_app->SetCursor(ZOOM_OUT);.

So, basically any SetCursor() call that is not B_HAND_CURSOR or B_I_BEAM_CURSOR ends up calling the SET_CURSOR_SHAPE() hook. Edit: ShowImage's using B_CURSOR_ID_GRABBING do not trigger it either. So I guess that none of the cursors on the BCursorID enum will trigger it.