eeschema and symbol editor: "assert" on dragging a control point of line/polyline/circle/arc outside the visible window

[Dino Ghilardi]

Tested on commit 92ccb766ab197157bba3c545cc721e7f20e46d2e.

If you drag a rectangle vertex in order to change its size, after some time an "assert" is issued.

How to reproduce:

-Create a new symbol
-Insert a rectangle
-Select the rectancle and drag one vertex.
-After some drawing with a "flashing" rectangle you get:

"An assertion failed"
...kicad/common/gal/opengl/gpu_manager.cpp(119): assert "m_isDrawing" failed in DrawIndices().

and in backtrace box, under the "Function prototype" you get:
make_fcontext

Then the assert window, clicking on "continue" is shown some more time.

P.S.: Dragging a circles cotrol point seems to work correctly.

Cheers,
Dino Ghilardi.

---------------------------------
Application: kicad
Version: (5.1.0-550-g92ccb766a), debug build
Libraries:
    wxWidgets 3.0.2
    libcurl/7.52.1 OpenSSL/1.0.2r zlib/1.2.8 libidn2/0.16 libpsl/0.17.0 (+libidn2/0.16) libssh2/1.7.0 nghttp2/1.18.1 librtmp/2.3
Platform: Linux 4.9.0-8-amd64 x86_64, 64 bit, Little endian, wxGTK
Build Info:
    wxWidgets: 3.0.2 (wchar_t,wx containers,compatible with 2.8) GTK+ 2.24
    Boost: 1.69.0
    OpenCASCADE Community Edition: 6.8.0
    Curl: 7.52.1
    Compiler: GCC 6.3.0 with C++ ABI 1010

Build settings:
    USE_WX_GRAPHICS_CONTEXT=OFF
    USE_WX_OVERLAY=OFF
    KICAD_SCRIPTING=ON
    KICAD_SCRIPTING_MODULES=ON
    KICAD_SCRIPTING_PYTHON3=OFF
    KICAD_SCRIPTING_WXPYTHON=ON
    KICAD_SCRIPTING_WXPYTHON_PHOENIX=OFF
    KICAD_SCRIPTING_ACTION_MENU=ON
    BUILD_GITHUB_PLUGIN=ON
    KICAD_USE_OCE=ON
    KICAD_USE_OCC=OFF
    KICAD_SPICE=ON

[Dino Ghilardi (dino-ghilardi)]

Probably related also to Bug #1828791 reported by eelik.

[Jeff Young (jeyjey)]

I don't think this is the same bug. I've seen it a few times too, but it needs a GAL expert to look at it.

Reliable steps to reproduce would be a huge benefit, although it may not be deterministic.

[Dino Ghilardi (dino-ghilardi)]

Same behaviour using polylines ("Add lines and polygons to symbol body" tool).

To reproduce it:

-add a polyline
-select it and drag a point
-continue dragging the point around until the "assert".

Tested on commit 2aea26de1e38eadf868fead7e89d92913885fbe3

Dragging the control point of an arc does not show this assert (but another strange behaviour: it moves the little squares that should be the start and the end point of the arc, but the arc is not redrawn/changed), but this seems a bug unrelated to this one).

[Dino Ghilardi (dino-ghilardi)]

Also dragging faster the control points triggers the assert faster, but I do not understand if the behaviour depends on the number of the mouse events or on its frequency (as when you have two threads "colliding" without a semaphore or a mutex for shared data access)

[Dino Ghilardi (dino-ghilardi)]

Also in eeschema, dragging a point of a line or poly-line (created using "place graphic lines or polygons" i got 50 asserts ...then clicked on "continue")

To reproduce:
-drag a vertex of a graphic line. After some movements the assert.

Tested on commit 2aea26de1e38eadf868fead7e89d92913885fbe3

[Dino Ghilardi (dino-ghilardi)]

Update: Found the deterministic behaviour for eeschema and symbol editor. dragging the vertex outside the screen causes the assert: seems related to the request for a window scroll while dragging a point.

Verified dragging control point of
-segments (lines an polylines, eeschema and symbol editor)
-rectangle (symbol editor)
-circle (symbol editor)
-arc (that also has another bug on dragging) (symbol editor).

eepcb seems not affected.

Tested on commit d701637eab3daf314a301516b77555c03243e102

Cheers,
Dino.

[Jon Evans (craftyjon)]

Hi Jeff,
I think the problem is that you call VIEW::Redraw() inside the tool event loop (ee_point_editor.cpp:311)
This should not be necessary because it will be called by the repaint handler, and I think having it also run from the tool coroutine is causing some issues. If I comment out this call, everything works fine and no more asserts.
-Jon

[Jeff Young (jeyjey)]

Thanks, Jon. I'll test it out on OSX and merge it if all looks OK.

[KiCad Janitor (kicad-janitor)]

Fixed in revision 67f7571f0419237b01a844a423be4decf3069441
https://git.launchpad.net/kicad/patch/?id=67f7571f0419237b01a844a423be4decf3069441