The OpenGL pipeline suffers from the same coordinate truncation problem
as described (and fixed) in bug 4975116. The errors are due to the use
of the integer cast operation to convert floating point coordinates to
integer coordinates which behaves oddly around the value zero when
the truncation direction switches from a floor behavior to a ceil
behavior (truncation by casting is always towards zero).

The OpenGL Renderer could be upgraded to use Math.floor to convert
the coordinates as was most of the renderers fixed in bug 4975116 or
it could be upgraded even further to maintain the subpixel precision
and pass float coordinates to the native level.

The test case in bug 4975116 as well as the regression test created
for its fix (test/sun/java2d/SunGraphics2D/CoordinateTruncationBug.java)
can be used to test the problem.