| Summary: | [GTK] Several WebGL tests are failing | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Diego Pino <dpino> |
| Component: | WebGL | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED FIXED | ||
| Severity: | Normal | CC: | alex, dino, kbr, lmoura, webkit-bug-importer, zan |
| Priority: | P2 | Keywords: | InRadar |
| Version: | WebKit Nightly Build | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| See Also: |
https://bugs.webkit.org/show_bug.cgi?id=211338 https://bugs.webkit.org/show_bug.cgi?id=211942 |
||
|
Description
Diego Pino
2020-05-14 03:12:45 PDT
Could the GTK port be switched over to use the ANGLE backend? I think ANGLE should be supported on all of GTK's supported platforms - please tell me if that assumption is incorrect fast/canvas/webgl/tex-sub-image-2d-bad-args.html is also crashing on Debug for both GTK and WPE, likely since r261609 fixed most issues after texImage2D and texSubImage2D refactor from r261023. Texture creation: var gl = wtu.create3DContext(canvas); var tex = gl.createTexture(); gl.bindTexture(gl.TEXTURE_2D, tex); gl.texImage2D(gl.TEXTURE_2D, 0, gl.RGBA, gl.RGBA, gl.UNSIGNED_BYTE, c); Offending call: gl.texSubImage2D(gl.TEXTURE_2D, 0, 0,0, gl.RGB, gl.UNSIGNED_BYTE, c); Start of crash log (before the huge visitor line) ASSERTION FAILED: tex->getInternalFormat(target, level) == internalFormat #0 0x00007fffdde7e340 in WTFCrash() () at ../../Source/WTF/wtf/Assertions.cpp:295 #1 0x00007fffec1f0292 in CRASH_WITH_INFO(...) () at DerivedSources/ForwardingHeaders/wtf/Assertions.h:713 #2 0x00007fffef5bad2a in WebCore::WebGLRenderingContextBase::texSubImage2DBase(unsigned int, int, int, int, int, int, unsigned int, unsigned int, unsigned int, int, void const*) (this=0x7fffd00d76f0, target=3553, level=0, xoffset=0, yoffset=0, width=16, height=16, internalFormat=6407, format=6407, type=5121, byteLength=768, pixels=0x7fff300ec000) at ../../Source/WebCore/html/canvas/WebGLRenderingContextBase.cpp:4882 #3 0x00007fffef5ba64e in WebCore::WebGLRenderingContextBase::texImageImpl(WebCore::WebGLRenderingContextBase::TexImageFunctionID, unsigned int, int, unsigned int, int, int, int, unsigned int, unsigned int, WebCore::Image*, WebCore::GraphicsContextGL::DOMSource, bool, bool, WebCore::IntRect const&, int, int) (this=0x7fffd00d76f0, functionID=WebCore::WebGLRenderingContextBase::TexImageFunctionID::TexSubImage2D, target=3553, level=0, internalformat=6408, xoffset=0, yoffset=0, zoffset=0, format=6407, type=5121, image=0x7fff300e3e10, domSource=WebCore::GraphicsContextGL::DOMSource::Canvas, flipY=false, premultiplyAlpha=false, sourceImageRect=..., depth=1, unpackImageHeight=0) at ../../Source/WebCore/html/canvas/WebGLRenderingContextBase.cpp:4809 #4 0x00007fffef5b8b09 in WebCore::WebGLRenderingContextBase::<lambda(const WTF::RefPtr<WebCore::HTMLCanvasElement, WTF::DumbPtrTraits<WebCore::HTMLCanvasElement> >&)>::operator()(const WTF::RefPtr<WebCore::HTMLCanvasElement, WTF::DumbPtrTraits<WebCore::HTMLCanvasElement> > &) const (__closure=0x7fffffffae30, canvas=...) at ../../Source/WebCore/html/canvas/WebGLRenderingContextBase.cpp:4630 I'm sorry about the regression. It's difficult to continue to maintain the non-ANGLE backend. Could GTK / WPE consider switching to use the ANGLE backend instead? After replacing the WebGL backend with ANGLE this tests now pass. The gardening commit is: https://commits.webkit.org/255008@main |