[EFL] False positive test passes with XvfbDriver

• Component: New Bugs
• Status: RESOLVED WONTFIX

[Csaba Osztrogonác]

http://trac.webkit.org/changeset/185583/trunk/LayoutTests/platform/efl/TestExpectations

css3/flexbox/csswg/* and imported/mozilla/* tests fail everywhere, it 
is impossible if they pass only on EFL. It must be false positive pass.

All of these tests are reference tests, so the completely wrong result 
and the expected result can be same, but the test doesn't pass at all.

[Csaba Osztrogonác]

Any update on it? It's not so good that we have many false positive passes.

Otherwise how is it possible to pass any test with XVFBDriver?
I can run tests only with USE_NATIVE_XDISPLAY=1.

All of them crashes for me with the following log:

STDERR: ERR<8579>:evas-gl_x11 modules/evas/engines/gl_x11/evas_x_main.c:1111 eng_best_visual_get() glXChooseFBConfig() can't find any configs (alpha: 0, depth: 0, stencil: 0, msaa: 0)
STDERR: ERR<8579>:evas-gl_x11 modules/evas/engines/gl_x11/evas_x_main.c:1111 eng_best_visual_get() glXChooseFBConfig() can't find any configs (alpha: 0, depth: 0, stencil: 0, msaa: 0)
STDERR: ERR<8579>:evas-gl_x11 modules/evas/engines/gl_x11/evas_x_main.c:1111 eng_best_visual_get() glXChooseFBConfig() can't find any configs (alpha: 0, depth: 0, stencil: 0, msaa: 0)
STDERR: ERR<8579>:ecore_evas modules/ecore_evas/engines/x/ecore_evas_x.c:494 _ecore_evas_x_gl_window_new() evas_engine_info_set() for engine 'opengl_x11' failed.
STDERR: ERR<8579>:ecore_evas modules/ecore_evas/engines/x/ecore_evas_x.c:4448 ecore_evas_gl_x11_options_new_internal() evas_engine_info_set() init engine 'opengl_x11' failed.
STDERR: ERR<8579>:evas_main lib/evas/canvas/evas_gl.c:138 evas_gl_new() Evas GL engine not available.
STDERR: 1   0x7ffbce3ad808
STDERR: 2   0x7ffbc99bd2f0
STDERR: 3   0x7ffbccaf2fcc WTF::HashTableIterator<unsigned int, WTF::KeyValuePair<unsigned int, WebCore::GraphicsContext3D::ShaderSourceEntry>, WTF::KeyValuePairKeyExtractor<WTF::KeyValuePair<unsigned int, WebCore::GraphicsContext3D::ShaderSourceEntry> >, WTF::IntHash<unsigned int>, WTF::HashMap<unsigned int, WebCore::GraphicsContext3D::ShaderSourceEntry, WTF::IntHash<unsigned int>, WTF::HashTraits<unsigned int>, WTF::HashTraits<WebCore::GraphicsContext3D::ShaderSourceEntry> >::KeyValuePairTraits, WTF::HashTraits<unsigned int> > WTF::HashTable<unsigned int, WTF::KeyValuePair<unsigned int, WebCore::GraphicsContext3D::ShaderSourceEntry>, WTF::KeyValuePairKeyExtractor<WTF::KeyValuePair<unsigned int, WebCore::GraphicsContext3D::ShaderSourceEntry> >, WTF::IntHash<unsigned int>, WTF::HashMap<unsigned int, WebCore::GraphicsContext3D::ShaderSourceEntry, WTF::IntHash<unsigned int>, WTF::HashTraits<unsigned int>, WTF::HashTraits<WebCore::GraphicsContext3D::ShaderSourceEntry> >::KeyValuePairTraits, WTF::HashTraits<unsigned int> >::find<WTF::IdentityHashTranslator<WTF::IntHash<unsigned int> >, unsigned int>(unsigned int const&)
STDERR: 4   0x7ffbccaf691b WebCore::GraphicsContext3D::mappedSymbolName(unsigned int, WebCore::ANGLEShaderSymbolType, WTF::String const&)
STDERR: 5   0x7ffbccaf7291 WebCore::GraphicsContext3D::getUniformLocation(unsigned int, WTF::String const&)
STDERR: 6   0x7ffbccb09838 WebCore::TextureMapperShaderProgram::getLocation(WTF::AtomicString const&, WebCore::TextureMapperShaderProgram::VariableType)
STDERR: 7   0x7ffbccb07893 WebCore::TextureMapperGL::drawSolidColor(WebCore::FloatRect const&, WebCore::TransformationMatrix const&, WebCore::Color const&)
STDERR: 8   0x7ffbcc13f6c7 WebKit::CoordinatedGraphicsScene::paintToGraphicsContext(WebCore::PlatformContextCairo*, WebCore::Color const&, bool)
STDERR: 9   0x7ffbcc1a11fa WebKit::WebViewEfl::paintToCairoSurface(_cairo_surface*)
STDERR: 10  0x7ffbcc158ee9 EwkView::displayTimerFired()
STDERR: 11  0x7ffbcc6ce97f WebCore::ThreadTimers::sharedTimerFiredInternal()
STDERR: 12  0x7ffbccac48cd
STDERR: 13  0x7ffbcab5e4c9
STDERR: 14  0x7ffbcab5e5fb
STDERR: 15  0x7ffbcab5a371
STDERR: 16  0x7ffbcab5a7c7 ecore_main_loop_begin
STDERR: 17  0x41ab3f WTR::TestInvocation::invoke()
STDERR: 18  0x41265a WTR::TestController::runTest(char const*)
STDERR: 19  0x41283b WTR::TestController::runTestingServerLoop()
STDERR: 20  0x419c58 WTR::TestController::TestController(int, char const**)
STDERR: 21  0x40c7ab main
STDERR: 22  0x7ffbc99a8a40 __libc_start_main
STDERR: 23  0x40c809 _start

[Csaba Osztrogonác]

ping?

[Gyuyoung Kim]

(In reply to comment #2)
> ping?

Let me check it tomorrow.

[Csaba Osztrogonác]

Could you explain how is it possible to run layout tests as the bot?

I still can't run any test with XVFB, only with USE_NATIVE_XDISPLAY=1,
but this configuration differs from the official bot.

[Gyuyoung Kim]

(In reply to comment #4)
> Could you explain how is it possible to run layout tests as the bot?
> 
> I still can't run any test with XVFB, only with USE_NATIVE_XDISPLAY=1,
> but this configuration differs from the official bot.

Hmm, EFL layout test works well on my linux box without USE_NATIVE_XDISPLAY=1. Do you still have crash when running tests ?

[Csaba Osztrogonác]

(In reply to comment #5)
> (In reply to comment #4)
> > Could you explain how is it possible to run layout tests as the bot?
> > 
> > I still can't run any test with XVFB, only with USE_NATIVE_XDISPLAY=1,
> > but this configuration differs from the official bot.
> 
> Hmm, EFL layout test works well on my linux box without
> USE_NATIVE_XDISPLAY=1. Do you still have crash when running tests ?

Yes, I still get the same crashes I mentioned in
https://bugs.webkit.org/show_bug.cgi?id=146104#c1

I can't run layout tests without USE_NATIVE_XDISPLAY=1
since removing TextureMapperImageBuffer.

Maybe the bug is inside EFL somewhere. I have an idea what can
be the problem, and I'm going to double check it later today.

[Csaba Osztrogonác]

I got the root of my problem:
- USE_NATIVE_XDISPLAY=1 works fine everywhere
- without USE_NATIVE_XDISPLAY tests crash with Nvidia GPU,
but pass with Intel integrated GPU. It must be a bug in EFL.

But it is unrelated to this bug ... Back to this bug, I still say 
that css3/flexbox/csswg/* and imported/mozilla/* tests can't pass anywhere.
These are false positive passes, because the actual and the expected results
are exactly wrong. I don't think if running tests in XFVB after removing
TextureMapperImageBuffer is a working configuration. (without adding
llvmpipe support to EFL)

[Michael Catanzaro]

Closing this bug because the EFL port has been removed from trunk.

If you feel this bug applies to a different upstream WebKit port and was closed in error, please either update the title and reopen the bug, or leave a comment to request this.