diff --git a/gfx/webrender_bindings/RenderCompositorANGLE.cpp b/gfx/webrender_bindings/RenderCompositorANGLE.cpp
index 47d7f0b2094f..ef81ca7ab715 100644
--- a/gfx/webrender_bindings/RenderCompositorANGLE.cpp
+++ b/gfx/webrender_bindings/RenderCompositorANGLE.cpp
@@ -120,13 +120,13 @@ RenderCompositorANGLE::Initialize()
   mDevice = GetDeviceOfEGLDisplay();
 
   if (!mDevice) {
-    gfxCriticalNote << "[D3D11] failed to get compositor device.";
+    gfxCriticalNote << "[WR] failed to get compositor device.";
     return false;
   }
 
   mDevice->GetImmediateContext(getter_AddRefs(mCtx));
   if (!mCtx) {
-    gfxCriticalNote << "[D3D11] failed to get immediate context.";
+    gfxCriticalNote << "[WR] failed to get immediate context.";
     return false;
   }
 
diff --git a/gfx/webrender_bindings/RenderThread.cpp b/gfx/webrender_bindings/RenderThread.cpp
index b05a4d47636b..05e349f535f7 100644
--- a/gfx/webrender_bindings/RenderThread.cpp
+++ b/gfx/webrender_bindings/RenderThread.cpp
@@ -160,7 +160,10 @@ RenderThread::AccumulateMemoryReport(MemoryReport aInitial)
   RefPtr<MemoryReportPromise::Private> p = new MemoryReportPromise::Private(__func__);
   MOZ_ASSERT(!IsInRenderThread());
   if (!Get() || !Get()->Loop()) {
-    // Seems to happen sometimes, see bug 1494430.
+    // This happens when the GPU process fails to start and we fall back to the
+    // basic compositor in the parent process. We could assert against this if we
+    // made the webrender detection code in gfxPlatform.cpp smarter.
+    // See bug 1494430 comment 12.
     NS_WARNING("No render thread, returning empty memory report");
     p->Resolve(aInitial, __func__);
     return p;