Problems with operator performance profiling

[4570235]

I want to profile the operator performance of my custom tflite model in android phone 'Google Pixel 6'.
I have followed the tutorial Trace LiteRT internals and built a very simple test program [source code]. However, I encounter some problems.

Firstly, when I tested the model 'mobilenetv1.tflite' from tflite example image_classification, if my program depends on LiteRT

implementation 'com.google.ai.edge.litert:litert:1.0.1'
implementation 'com.google.ai.edge.litert:litert-support:1.0.1'

then the Android Studio CPU Profiler shows

I cannot see all the operators. So I switch to depend on TFLite

implementation 'org.tensorflow:tensorflow-lite:2.12.0'
implementation 'org.tensorflow:tensorflow-lite-support:0.4.0'

then it works. Here's the result:

And I also figure out that the maximum version of TFLite that works is tensorflow-lite:2.12.0 with tensorflow-lite-support:0.4.0. Trying any newer version would result the same as LiteRT's.
So, is this a bug? How can I trace operator performance with the newest version of TFLite or LiteRT?

Secondly, with tensorflow-lite:2.12.0 and tensorflow-lite-support:0.4.0 mentioned above, I tried to trace my custom model which is very similar to Qualcomm's 'quicksrnetsmall.tflite'. Let's take 'quicksrnetsmall.tflite' as an sample and the Profiler shows

I cannot see any operators. How can I fix it?