##// END OF EJS Templates
match: convert O(n) to O(log n) in exactmatcher.visitchildrenset...
match: convert O(n) to O(log n) in exactmatcher.visitchildrenset When using narrow, during rebase this is called (at least) once per directory in the set of files in the commit being rebased. Every time it's called, we did the set arithmetic (now extracted and cached), which was probably pretty cheap but not necessary to repeat each time, looped over every item in the matcher and kept things that started with the directory we were querying. With very large narrowspecs, and a commit that touched a file in a large number of directories, this was slow. In a pathological repo, the rebase of a single commit (that touched over 17k files, I believe in approximately as many directories) with a narrowspec that had >32k entries took 8,246s of profiled time, with 5,007s of that spent in visitchildrenset (transitively). With this change, the time spent in visitchildrenset is less than 34s (which is where my profile cut off). Most of the remaining time was network access due to our custom remotefilelog-based setup not properly prefetching. Differential Revision: https://phab.mercurial-scm.org/D10294

File last commit:

r44870:8f7c6656 default
r47634:8bca353b default
Show More
index.rs
95 lines | 2.4 KiB | application/rls-services+xml | RustLexer
// Copyright 2019-2020 Georges Racinet <georges.racinet@octobus.net>
//
// This software may be used and distributed according to the terms of the
// GNU General Public License version 2 or any later version.
//! Minimal `RevlogIndex`, readable from standard Mercurial file format
use hg::*;
use memmap::*;
use std::fs::File;
use std::ops::Deref;
use std::path::Path;
use std::slice;
pub struct Index {
data: Box<dyn Deref<Target = [IndexEntry]> + Send>,
}
/// A fixed sized index entry. All numbers are big endian
#[repr(C)]
pub struct IndexEntry {
not_used_yet: [u8; 24],
p1: Revision,
p2: Revision,
node: Node,
unused_node: [u8; 12],
}
pub const INDEX_ENTRY_SIZE: usize = 64;
impl IndexEntry {
fn parents(&self) -> [Revision; 2] {
[Revision::from_be(self.p1), Revision::from_be(self.p1)]
}
}
impl RevlogIndex for Index {
fn len(&self) -> usize {
self.data.len()
}
fn node(&self, rev: Revision) -> Option<&Node> {
if rev == NULL_REVISION {
return None;
}
let i = rev as usize;
if i >= self.len() {
None
} else {
Some(&self.data[i].node)
}
}
}
impl Graph for &Index {
fn parents(&self, rev: Revision) -> Result<[Revision; 2], GraphError> {
let [p1, p2] = (*self).data[rev as usize].parents();
let len = (*self).len();
if p1 < NULL_REVISION
|| p2 < NULL_REVISION
|| p1 as usize >= len
|| p2 as usize >= len
{
return Err(GraphError::ParentOutOfRange(rev));
}
Ok([p1, p2])
}
}
struct IndexMmap(Mmap);
impl Deref for IndexMmap {
type Target = [IndexEntry];
fn deref(&self) -> &[IndexEntry] {
let ptr = self.0.as_ptr() as *const IndexEntry;
// Any misaligned data will be ignored.
debug_assert_eq!(
self.0.len() % std::mem::align_of::<IndexEntry>(),
0,
"Misaligned data in mmap"
);
unsafe { slice::from_raw_parts(ptr, self.0.len() / INDEX_ENTRY_SIZE) }
}
}
impl Index {
pub fn load_mmap(path: impl AsRef<Path>) -> Self {
let file = File::open(path).unwrap();
let msg = "Index file is missing, or missing permission";
let mmap = unsafe { MmapOptions::new().map(&file) }.expect(msg);
Self {
data: Box::new(IndexMmap(mmap)),
}
}
}