flypig.co.uk

I'm continuing to restore bits of WebView functionality today, in the hope that the renderer will click into place and start rendering. I've convinced myself it's only a matter of time now.

Yesterday I tried restoring the TextureClient destruction code, but that didn't have any obvious effects so I removed the changes again. Today I'm trying something else: restoring the augmented fBindFramebuffer() functionality.

This is unusual because in default Gecko fBindFramebuffer() is just a wrapper for glBindFramebuffer from the OpenGL library. In previous EmbedLite versions of Gecko the wrapper has been given additional functionality which we can find in GLContext.cpp and which looks like this:

void GLContext::fBindFramebuffer(GLenum target, GLuint framebuffer) {
  if (!mScreen) {
    raw_fBindFramebuffer(target, framebuffer);
    return;
  }

  switch (target) {
    case LOCAL_GL_DRAW_FRAMEBUFFER_EXT:
      mScreen->BindDrawFB(framebuffer);
      return;

    case LOCAL_GL_READ_FRAMEBUFFER_EXT:
      mScreen->BindReadFB(framebuffer);
      return;

    case LOCAL_GL_FRAMEBUFFER:
      mScreen->BindFB(framebuffer);
      return;

    default:
      // Nothing we care about, likely an error.
      break;
  }

  raw_fBindFramebuffer(target, framebuffer);
}

With this change the raw_fBindFramebuffer() becomes the new name for the glBindFramebuffer() wrapper. As you can see, now when fBindFramebuffer() is called some framebuffer binding is also performed.

The additional functionality isn't needed everywhere though, and in fact there are a few places where we then have to switch the calls from using fBindFramebuffer() to using the new raw wrapper instead:

@ -215,11 +215,11 @@ void GLScreenBuffer::BindFB(GLuint fb) {
   mInternalReadFB = (fb == 0) ? readFB : fb;
 
   if (mInternalDrawFB == mInternalReadFB) {
+    mGL->raw_fBindFramebuffer(LOCAL_GL_FRAMEBUFFER, mInternalDrawFB);
-    mGL->fBindFramebuffer(LOCAL_GL_FRAMEBUFFER, mInternalDrawFB);
   } else {
     MOZ_ASSERT(mGL->IsSupported(GLFeature::split_framebuffer));
+    mGL->raw_fBindFramebuffer(LOCAL_GL_DRAW_FRAMEBUFFER_EXT, mInternalDrawFB);
+    mGL->raw_fBindFramebuffer(LOCAL_GL_READ_FRAMEBUFFER_EXT, mInternalReadFB);
-    mGL->fBindFramebuffer(LOCAL_GL_DRAW_FRAMEBUFFER_EXT, mInternalDrawFB);
-    mGL->fBindFramebuffer(LOCAL_GL_READ_FRAMEBUFFER_EXT, mInternalReadFB);
   }
 
 #ifdef DEBUG
@@ -235,7 +235,7 @@ void GLScreenBuffer::BindDrawFB(GLuint fb) {
   mUserDrawFB = fb;
   mInternalDrawFB = (fb == 0) ? drawFB : fb;
 
+  mGL->raw_fBindFramebuffer(LOCAL_GL_DRAW_FRAMEBUFFER_EXT, mInternalDrawFB);
-  mGL->fBindFramebuffer(LOCAL_GL_DRAW_FRAMEBUFFER_EXT, mInternalDrawFB);
 
 #ifdef DEBUG
   mInInternalMode_DrawFB = false;
@@ -249,7 +249,7 @@ void GLScreenBuffer::BindReadFB(GLuint fb) {
   mUserReadFB = fb;
   mInternalReadFB = (fb == 0) ? readFB : fb;
 
+  mGL->raw_fBindFramebuffer(LOCAL_GL_READ_FRAMEBUFFER_EXT, mInternalReadFB);
-  mGL->fBindFramebuffer(LOCAL_GL_READ_FRAMEBUFFER_EXT, mInternalReadFB);
 
 #ifdef DEBUG
   mInInternalMode_ReadFB = false;

These aren't all of the uses though and in the remaining cases the new augmented functionality will be called instead. This is another nicely self-contained change without which I could perfectly well imagine the rendering might be broken. So restoring this functionality might have positive effects.

Let's see. I've made these changes and set the partial build running. Soon I'll be able to test this out.

[...]
 
And it works! The WebView renders nicely. The browser renders nicely. Both of them render WebGL nicely (props to Shane for the nice shader example). Great!

Phew. This concludes the difficult part of this process. Sadly it doesn't mean the end of the journey just yet though, oh no: I still need to turn this into a tidy commit and figure out what to do with the remaining changes (the ones that don't appear to be necessary). And to avoid any undue excitement, there are more things to fix after that as well.

Those will all be for tomorrow and beyond though. For today, I'm going to celebrate (in a quiet contemplative way!) and enjoy the win.

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