Jit: reconsider policy for recursive inlines #41542
Description
A couple of known regressions from 3.1 to 5.0 because of #35020 leads me to think we should explore this a bit more carefully.
- BinaryTrees_2 and BinaryTrees_5: See Investigate regressions of binary-trees and fannkuch on benchmarksgame site #40810.
- Fibonacci: See Performance anomalies in comparsion to Mono #13425.
Would be interesting to see if we run across any non-benchmark examples.
category:cq
theme:inlining
skill-level:intermediate
cost:medium