When running the :tier4 of the jdk jtreg test (java/awt, javax/swing, sun/java2d ) with binaries compiled with -ftrapv
to catch signed integer overflows, we get reproducibly those signed int overflow errors (similar to JDK-8330615) .

java/awt/Paint/PgramUserBoundsTest.java

# SIGILL (0x4) at pc=0x000000010930a686, pid=3989, tid=43267
# Problematic frame:
# C [libawt.dylib+0x51686] ShapeSINextSpan+0x616

> C [libawt.so+0xd3e58] ShapeSINextSpan+0x568 (ShapeSpanIterator.c:1366)
> C [libawt.so+0xd38b0] Java_sun_java2d_pipe_ShapeSpanIterator_nextSpan+0x50 (ShapeSpanIterator.c:509)
> j sun.java2d.pipe.ShapeSpanIterator.nextSpan([I)Z+0 java.desktop@23.0.0.1-internal

and

java/awt/Scrollbar/AquaLFScrollbarTest/ScrollBarBorderTest.java

# SIGILL (0x4) at pc=0x000000010c98243d, pid=4021, tid=40963
# Problematic frame:
# C [libawt.dylib+0x5843d] BilinearInterp+0x17d

> C [libawt.so+0xe3e97] BilinearInterp+0x787 (TransformHelper.c:683)
> C [libawt.so+0xe6f84] Java_sun_java2d_loops_TransformHelper_Transform+0xe64 (TransformHelper.c:499)
> j sun.java2d.loops.TransformHelper.Transform(Lsun/java2d/loops/MaskBlit;Lsun/java2d/SurfaceData;Lsun/java2d/SurfaceData;Ljava/awt/Composite;Lsun/java2d/pipe/Region;Ljava/awt/geom/AffineTransform;IIIIIIIII[III)V+0 java.desktop@23.0.0.1-internal


Seems there is no change to this coding intended in the short term so I suggest to exclude the 2 functions from ubsan-checking .