Skip to content

Commit c03eba2

Browse files
Add FIXME for faster cached block transfer functions
I've tried a few ways of implementing this, but each fell short. Adding an auxiliary `_Idx` associated type to `Analysis` that defaults to `!` but is overridden in the blanket impl of `Analysis` for `A: GenKillAnalysis` to `A::Idx` seems promising, but the trait solver is unable to prove equivalence between `A::Idx` and `A::_Idx` within the overridden version of `into_engine`. Without full-featured specialization, removing `into_engine` or splitting it into a different trait would have a significant ergonomic penalty. Alternatively, we could erase the index type and store a `GenKillSet<u32>` as well as a function pointer for transmuting between `&mut A::Domain` and `&mut BitSet<u32>` in the hopes that LLVM can devirtualize a simple function pointer better than the boxed closure. However, this is brittle, requires `unsafe` code, and doesn't work for index types that aren't the same size as a `u32` (e.g. `usize`) since `GenKillSet` stores a `HybridBitSet`, which may be a `Vec<I>`. Perhaps safe transmute could help here?
1 parent b19b8ea commit c03eba2

File tree

1 file changed

+5
-0
lines changed
  • compiler/rustc_mir/src/dataflow/framework

1 file changed

+5
-0
lines changed

compiler/rustc_mir/src/dataflow/framework/engine.rs

+5
Original file line numberDiff line numberDiff line change
@@ -87,6 +87,11 @@ where
8787
analysis: A,
8888

8989
/// Cached, cumulative transfer functions for each block.
90+
//
91+
// FIXME(ecstaticmorse): This boxed `Fn` trait object is invoked inside a tight loop for
92+
// gen/kill problems on cyclic CFGs. This is not ideal, but it doesn't seem to degrade
93+
// performance in practice. I've tried a few ways to avoid this, but they have downsides. See
94+
// the message for the commit that added this FIXME for more information.
9095
apply_trans_for_block: Option<Box<dyn Fn(BasicBlock, &mut A::Domain)>>,
9196
}
9297

0 commit comments

Comments
 (0)