We use a workaround for readPixels because of some problems the SGX driver has with reading pixels. This workaround is currently limited to the size of the canvas.
Created attachment 154479 [details] Patch
Comment on attachment 154479 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=154479&action=review > Source/WebCore/ChangeLog:8 > + This fixes an existing test (uninitialized-test), no new tests. Better state which one. > Source/WebCore/platform/graphics/opengl/GraphicsContext3DOpenGLES.cpp:120 > + readPixels(x, y, width, height, GL_RGBA, GL_UNSIGNED_BYTE, pixels); This seems not BB specific? Is it safe to do?
Right. Forgot about the anti-aliasing stuff on other platforms. I wonder if it's intentionally not in readPixelsAndConvertToBGRAIfNecessary...
Created attachment 154642 [details] Patch
Comment on attachment 154642 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=154642&action=review Better :) > Source/WebCore/platform/graphics/opengl/GraphicsContext3DOpenGLES.cpp:127 > + ::glReadPixels(x, y, width, height, GL_RGBA, GL_UNSIGNED_BYTE, pixels); You do the same thing twice, could add a helper method for that, but as we discussed can be done later.
Comment on attachment 154642 [details] Patch Clearing flags on attachment: 154642 Committed r123759: <http://trac.webkit.org/changeset/123759>
All reviewed patches have been landed. Closing bug.