Skip to content

Remove c32() from x86_64 memcmp #488

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

Merged
merged 1 commit into from
Aug 10, 2022
Merged

Remove c32() from x86_64 memcmp #488

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
14 changes: 1 addition & 13 deletions src/mem/x86_64.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -170,19 +170,7 @@ pub unsafe fn compare_bytes(a: *const u8, b: *const u8, n: usize) -> i32 {
let c4 = |a: *const u32, b, n| cmp(a, b, n, c2);
let c8 = |a: *const u64, b, n| cmp(a, b, n, c4);
let c16 = |a: *const u128, b, n| cmp(a, b, n, c8);
let c32 = |a: *const [u128; 2], b, n| cmp(a, b, n, c16);
// [u128; 2] internally uses raw_eq for comparisons, which may emit a call to memcmp
// above a certain size threshold. When SSE2 is enabled this threshold does not seem
// to be reached but without SSE2 a call is emitted, leading to infinite recursion.
//
// While replacing [u128; 2] with (u128, u128) fixes the issues it degrades performance
// severely. Likewise, removing c32() has a lesser but still significant impact. Instead the
// [u128; 2] case is only enabled when SSE2 is present.
if cfg!(target_feature = "sse2") {
c32(a.cast(), b.cast(), n)
} else {
c16(a.cast(), b.cast(), n)
}
c16(a.cast(), b.cast(), n)
}

/// Determine optimal parameters for a `rep` instruction.
Expand Down