Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Optimize seen Predicate filtering. #50932

Merged
merged 1 commit into from
May 22, 2018
Merged

Optimize seen Predicate filtering. #50932

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
21 changes: 18 additions & 3 deletions src/librustc/traits/select.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3356,13 +3356,28 @@ impl<'cx, 'gcx, 'tcx> SelectionContext<'cx, 'gcx, 'tcx> {
predicate: predicate.value
}))
}).collect();

// We are performing deduplication here to avoid exponential blowups
// (#38528) from happening, but the real cause of the duplication is
// unknown. What we know is that the deduplication avoids exponential
// amount of predicates being propogated when processing deeply nested
// amount of predicates being propagated when processing deeply nested
// types.
let mut seen = FxHashSet();
predicates.retain(|i| seen.insert(i.clone()));
//
// This code is hot enough that it's worth avoiding the allocation
// required for the FxHashSet when possible. Special-casing lengths 0,
// 1 and 2 covers roughly 75--80% of the cases.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What about using sort + dedup? That would avoid the allocation, and also quickly exit on small numbers of elements.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

That changes the order, which may break the behavior.

Also, deriving Ord is a pain.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

And Predicate is a large type, so moving the values around might be moderately expensive.

Copy link
Member

@eddyb eddyb May 22, 2018 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh, I see, the set is used with retain. But you need to clone & move the values into a set anyway. If you want to avoid the cost of data movement, the set could be created using references into the original Vec instead. Except that doesn't work because of the retain...

if predicates.len() <= 1 {
// No possibility of duplicates.
} else if predicates.len() == 2 {
// Only two elements. Drop the second if they are equal.
if predicates[0] == predicates[1] {
predicates.truncate(1);
}
} else {
// Three or more elements. Use a general deduplication process.
let mut seen = FxHashSet();
predicates.retain(|i| seen.insert(i.clone()));
}
self.infcx().plug_leaks(skol_map, snapshot, predicates)
}
}
Expand Down