Each box in the flame graph currently has two pieces:
- The "non-self" piece, where there are callee boxes stack on top.
- The "self" piece, representing a call node's self samples.
The "self" piece is always on the right.
Once we know which samples are on-cpu samples and which samples are off-cpu samples, we can split the "self" piece into two sub-pieces:
- The "on-CPU self" piece
- The "off-CPU self" piece
And then we can shade the off-CPU self piece differently, for example give it less opacity.
┆Issue is synchronized with this Jira Task