On Sun, 20 Apr 2025 08:15:38 GMT, Sergey Bylokhov <s...@openjdk.org> wrote:
> At some point, java.awt.Color.createContext() was marked as synchronized to > support caching of a ColorPaintContext instance. The cache was later removed, > but the synchronized modifier remained. Since there is no shared mutable > state anymore, the synchronization is no longer necessary and can be safely > removed. > > > Note: This code path is rarely executed because a special optimization was > introduced in > [SunGraphics2D.setPaint](https://github.com/openjdk/jdk/blob/c514f135ccf08c3be016a32ae8f2c055fb941857/src/java.desktop/share/classes/sun/java2d/SunGraphics2D.java#L1003). > As a result, unless a custom wrapper around the Color class is used, the > Color.createContext() method is typically bypassed during rendering. > > Two tests are added: > - ColorPaintContextBasicTest: Checks if different image types (BufferedImage > and VolatileImage) produce the same results when using different ways to fill > the image (setColor, setPaint, and custom Paint). This test intentionally > uses a custom Paint implementation to bypass the optimization and ensure that > Color.createContext() is invoked verifying its correct behavior. > - ColorPaintContextStateTrackerTest: Checks that the raster used in > ColorPaintContext.getRaster() can be properly cached in video memory and we > do not need to call icr.markDirty() in ColorPaintContext.getRaster() src/java.desktop/share/classes/java/awt/ColorPaintContext.java line 39: > 37: private final int color; > 38: private volatile WritableRaster savedTile; > 39: Since ColorPaintContext is no longer cached within the Color class, it's unlikely that this object will be accessed by multiple threads in typical JDK usage. Therefore, the synchronized modifier has been removed from the getRaster() method. However, to ensure thread safety in the rare case where a ColorPaintContext instance is shared across threads (if app decides to do so for some reason), the savedTile field has been declared as volatile. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/24771#discussion_r2071030293
