flypig.co.uk

Things started falling into place yesterday when the reasons for the crashes became clear. Although the value of embedlite.compositor.external_gl_context was being set correctly to different values depending on whether the running application was the browser or a WebView, the value was being reversed in the case of the browser because the WebEngineSettings::initialize() was being called twice. This initialize() method is supposed to be idempotent (which you have to admit, is a great word!), but due to a glitch in the logic turned out not to be.

The fix was to change the ordering of the execution: moving the place where isInitialized gets set to before the early return caused by the existence of a marker file. That all sounds a bit esoteric, but it was a simple change. I've now lined up all of the pieces so that:

1. embedlite.compositor.external_gl_context is set to true in WebEngineSettings::initialize().
2. embedlite.compositor.external_gl_context is set to false in DeclarativeWebUtils::setRenderingPreferences().
3. isInitialized is set in the correct place.

With these three changes I'm hoping things will work as expected.

But unfortunately not quite yet. After making these changes, the browser works fine, but I now get a crash when running the WebView application:

Thread 37 "Compositor" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7f1f94d7e0 (LWP 7616)]
0x0000007ff1105978 in mozilla::gl::GLScreenBuffer::Size (this=0x0)
    at ${PROJECT}/obj-build-mer-qt-xr/dist/include/mozilla/UniquePtr.h:290
290     ${PROJECT}/obj-build-mer-qt-xr/dist/include/mozilla/UniquePtr.h: No 
    such file or directory.
(gdb) bt
#0  0x0000007ff1105978 in mozilla::gl::GLScreenBuffer::Size (this=0x0)
    at ${PROJECT}/obj-build-mer-qt-xr/dist/include/mozilla/UniquePtr.h:290
#1  mozilla::gl::GLContext::OffscreenSize (this=this@entry=0x7ee019aa70)
    at gfx/gl/GLContext.cpp:2141
#2  0x0000007ff3664264 in mozilla::embedlite::EmbedLiteCompositorBridgeParent::
    CompositeToDefaultTarget (this=0x7fc4ad3530, aId=...)
    at mobile/sailfishos/embedthread/EmbedLiteCompositorBridgeParent.cpp:156
#3  0x0000007ff12b48d8 in mozilla::layers::CompositorVsyncScheduler::
    ForceComposeToTarget (this=0x7fc4c36e00, aTarget=aTarget@entry=0x0, 
    aRect=aRect@entry=0x0)
    at ${PROJECT}/obj-build-mer-qt-xr/dist/include/mozilla/layers/LayersTypes.h:
    82
#4  0x0000007ff12b4934 in mozilla::layers::CompositorBridgeParent::
    ResumeComposition (this=this@entry=0x7fc4ad3530)
    at ${PROJECT}/obj-build-mer-qt-xr/dist/include/mozilla/RefPtr.h:313
#5  0x0000007ff12b49c0 in mozilla::layers::CompositorBridgeParent::
    ResumeCompositionAndResize (this=0x7fc4ad3530, x=<optimized out>, 
    y=<optimized out>, 
    width=<optimized out>, height=<optimized out>)
    at gfx/layers/ipc/CompositorBridgeParent.cpp:794
#6  0x0000007ff12ad55c in mozilla::detail::RunnableMethodArguments<int, int, 
    int, int>::applyImpl<mozilla::layers::CompositorBridgeParent, void (mozilla:
    :layers::CompositorBridgeParent::*)(int, int, int, int), 
    StoreCopyPassByConstLRef<int>, StoreCopyPassByConstLRef<int>, 
    StoreCopyPassByConstLRef<int>, StoreCopyPassByConstLRef<int>, 0ul, 1ul, 
    2ul, 3ul> (args=..., m=<optimized out>, o=<optimized out>)
    at ${PROJECT}/obj-build-mer-qt-xr/dist/include/nsThreadUtils.h:1151
[...]
#18 0x0000007ff6a0289c in thread_start () at ../sysdeps/unix/sysv/linux/aarch64/
    clone.S:78
(gdb) 

Thinking back, or more precisely looking back through my diary entries the reason becomes clear. You may recall that back on Day 221 I made some changes in the hope of them fixing the browser rendering. The changes are summarised by this diff:

$ git diff -- gfx/layers/opengl/CompositorOGL.cpp
diff --git a/gfx/layers/opengl/CompositorOGL.cpp b/gfx/layers/opengl/
    CompositorOGL.cpp
index 8a423b840dd5..11105c77c43b 100644
--- a/gfx/layers/opengl/CompositorOGL.cpp
+++ b/gfx/layers/opengl/CompositorOGL.cpp
@@ -246,12 +246,14 @@ already_AddRefed<mozilla::gl::GLContext> CompositorOGL::
    CreateContext() {
 
   // Allow to create offscreen GL context for main Layer Manager
   if (!context && gfxEnv::LayersPreferOffscreen()) {
+    SurfaceCaps caps = SurfaceCaps::ForRGB();
+    caps.preserve = false;
+    caps.bpp16 = gfxVars::OffscreenFormat() == SurfaceFormat::R5G6B5_UINT16;
+
     nsCString discardFailureId;
-    context = GLContextProvider::CreateHeadless(
-        {CreateContextFlags::REQUIRE_COMPAT_PROFILE}, &discardFailureId);
-    if (!context->CreateOffscreenDefaultFb(mSurfaceSize)) {
-      context = nullptr;
-    }
+    context = GLContextProvider::CreateOffscreen(
+        mSurfaceSize, caps, CreateContextFlags::REQUIRE_COMPAT_PROFILE,
+        &discardFailureId);
   }
[...]

Now that the underlying error is clear it's time to reverse this change. At the time I even noted the importance of these diary entries as a way of helping in case I might have to do something like this:


And that's exactly what has happened now. Without keeping track of the changes, I'm fairly certain I'd have got in a mess and forgotten I'd made these changes. Now I can reverse them easily.

Having made this reversal, build and installed the executable and run the code I'm very happy to see that:

1. The browser now works again, rendering and all.
2. The WebVew app no longer crashes.

What this hasn't done is fix the WebVew render, but that was never the expectation. We've definitely made some progress now, but I've lost nearly two weeks of work fixing this issue with the browser. I need to get back on track with investigating the WebView render. More on that tomorrow.

If you'd like to read any of my other gecko diary entries, they're all available on my Gecko-dev Diary page.