( )⚙ D9497 copies-rust: start recording overwrite as they happens

This is an archive of the discontinued Mercurial Phabricator instance.

Differential D9497

copies-rust: start recording overwrite as they happens
ClosedPublic

Authored by marmoute on Dec 2 2020, 11:02 AM.

Details

Reviewers
Alphare
pulkit
Group Reviewers
hg-reviewers
Commits
rHGfce2f20a54ce: copies-rust: start recording overwrite as they happens
Summary

If a revision has information overwriting data from another revision, the
overwriting revision is a descendant of the overwritten one. So we could warm
the Oracle cache with such information to avoid potential future is_ancestors
call.

This provide us with a large speedup in the most expensive cases:

Repo Case Source-Rev Dest-Rev # of revisions old time new time Difference Factor time per rev

mozilla-try x00000_revs_x00000_added_x0000_copies 1b661134e2ca 1ae03d022d6d : 228985 revs, 41.113063 s, 36.001255 s, -5.111808 s, × 0.8757, 157 µs/rev
mozilla-try x00000_revs_x00000_added_x000_copies 9b2a99adc05e 8e29777b48e6 : 382065 revs, 27.891612 s, 14.340641 s, -13.550971 s, × 0.5142, 37 µs/rev

Full comparison below:

Repo Case Source-Rev Dest-Rev # of revisions old time new time Difference Factor time per rev

mercurial x_revs_x_added_0_copies ad6b123de1c7 39cfcef4f463 : 1 revs, 0.000042 s, 0.000042 s, +0.000000 s, × 1.0000, 42 µs/rev
mercurial x_revs_x_added_x_copies 2b1c78674230 0c1d10351869 : 6 revs, 0.000114 s, 0.000109 s, -0.000005 s, × 0.9561, 18 µs/rev
mercurial x000_revs_x000_added_x_copies 81f8ff2a9bf2 dd3267698d84 : 1032 revs, 0.004934 s, 0.004953 s, +0.000019 s, × 1.0039, 4 µs/rev
pypy x_revs_x_added_0_copies aed021ee8ae8 099ed31b181b : 9 revs, 0.000195 s, 0.000237 s, +0.000042 s, × 1.2154, 26 µs/rev
pypy x_revs_x000_added_0_copies 4aa4e1f8e19a 359343b9ac0e : 1 revs, 0.000050 s, 0.000050 s, +0.000000 s, × 1.0000, 50 µs/rev
pypy x_revs_x_added_x_copies ac52eb7bbbb0 72e022663155 : 7 revs, 0.000113 s, 0.000113 s, +0.000000 s, × 1.0000, 16 µs/rev
pypy x_revs_x00_added_x_copies c3b14617fbd7 ace7255d9a26 : 1 revs, 0.6f1f4a s, 0.6f1f4a s, +0.000000 s, × 1.0000, 322 µs/rev
pypy x_revs_x000_added_x000_copies df6f7a526b60 a83dc6a2d56f : 6 revs, 0.010788 s, 0.010702 s, -0.000086 s, × 0.9920, 1783 µs/rev
pypy x000_revs_xx00_added_0_copies 89a76aede314 2f22446ff07e : 4785 revs, 0.050880 s, 0.050504 s, -0.000376 s, × 0.9926, 10 µs/rev
pypy x000_revs_x000_added_x_copies 8a3b5bfd266e 2c68e87c3efe : 6780 revs, 0.081760 s, 0.080159 s, -0.001601 s, × 0.9804, 11 µs/rev
pypy x000_revs_x000_added_x000_copies 89a76aede314 7b3dda341c84 : 5441 revs, 0.061382 s, 0.060058 s, -0.001324 s, × 0.9784, 11 µs/rev
pypy x0000_revs_x_added_0_copies d1defd0dc478 c9cb1334cc78 : 43645 revs, 0.585802 s, 0.536950 s, -0.048852 s, × 0.9166, 12 µs/rev
pypy x0000_revs_xx000_added_0_copies bf2c629d0071 4ffed77c095c : 2 revs, 0.012803 s, 0.012868 s, +0.000065 s, × 1.0051, 6434 µs/rev
pypy x0000_revs_xx000_added_x000_copies 08ea3258278e d9fa043f30c0 : 11316 revs, 0.113558 s, 0.112806 s, -0.000752 s, × 0.9934, 9 µs/rev
netbeans x_revs_x_added_0_copies fb0955ffcbcd a01e9239f9e7 : 2 revs, 0.000085 s, 0.000084 s, -0.000001 s, × 0.9882, 42 µs/rev
netbeans x_revs_x000_added_0_copies 6f360122949f 20eb231cc7d0 : 2 revs, 0.000106 s, 0.000106 s, +0.000000 s, × 1.0000, 53 µs/rev
netbeans x_revs_x_added_x_copies 1ada3faf6fb6 5a39d12eecf4 : 3 revs, 0.000175 s, 0.000174 s, -0.000001 s, × 0.9943, 58 µs/rev
netbeans x_revs_x00_added_x_copies 35be93ba1e2c 9eec5e90c05f : 9 revs, 0.000721 s, 0.000726 s, +0.000005 s, × 1.0069, 80 µs/rev
netbeans x000_revs_xx00_added_0_copies eac3045b4fdd 51d4ae7f1290 : 1421 revs, 0.010127 s, 0.010105 s, -0.000022 s, × 0.9978, 7 µs/rev
netbeans x000_revs_x000_added_x_copies e2063d266acd 6081d72689dc : 1533 revs, 0.015616 s, 0.015748 s, +0.000132 s, × 1.0085, 10 µs/rev
netbeans x000_revs_x000_added_x000_copies ff453e9fee32 411350406ec2 : 5750 revs, 0.061341 s, 0.060357 s, -0.000984 s, × 0.9840, 10 µs/rev
netbeans x0000_revs_xx000_added_x000_copies 588c2d1ced70 1aad62e59ddd : 66949 revs, 0.542214 s, 0.499356 s, -0.042858 s, × 0.9210, 7 µs/rev
mozilla-central x_revs_x_added_0_copies 3697f962bb7b 7015fcdd43a2 : 2 revs, 0.000089 s, 0.000092 s, +0.000003 s, × 1.0337, 46 µs/rev
mozilla-central x_revs_x000_added_0_copies dd390860c6c9 40d0c5bed75d : 8 revs, 0.000279 s, 0.000279 s, +0.000000 s, × 1.0000, 34 µs/rev
mozilla-central x_revs_x_added_x_copies 8d198483ae3b 14207ffc2b2f : 9 revs, 0.000184 s, 0.000186 s, +0.000002 s, × 1.0109, 20 µs/rev
mozilla-central x_revs_x00_added_x_copies 98cbc58cc6bc 446a150332c3 : 7 revs, 0.000661 s, 0.000660 s, -0.000001 s, × 0.9985, 94 µs/rev
mozilla-central x_revs_x000_added_x000_copies 3c684b4b8f68 0a5e72d1b479 : 3 revs, 0.003377 s, 0.003372 s, -0.000005 s, × 0.9985, 1124 µs/rev
mozilla-central x_revs_x0000_added_x0000_copies effb563bb7e5 c07a39dc4e80 : 6 revs, 0.070508 s, 0.070294 s, -0.000214 s, × 0.9970, 11715 µs/rev
mozilla-central x000_revs_xx00_added_0_copies 6100d773079a 04a55431795e : 1593 revs, 0.006576 s, 0.006545 s, -0.000031 s, × 0.9953, 4 µs/rev
mozilla-central x000_revs_x000_added_x_copies 9f17a6fc04f9 2d37b966abed : 41 revs, 0.004809 s, 0.004998 s, +0.000189 s, × 1.0393, 121 µs/rev
mozilla-central x000_revs_x000_added_x000_copies 7c97034feb78 4407bd0c6330 : 7839 revs, 0.064872 s, 0.063348 s, -0.001524 s, × 0.9765, 8 µs/rev
mozilla-central x0000_revs_xx000_added_0_copies 9eec5917337d 67118cc6dcad : 615 revs, 0.026142 s, 0.026154 s, +0.000012 s, × 1.0005, 42 µs/rev
mozilla-central x0000_revs_xx000_added_x000_copies f78c615a656c 96a38b690156 : 30263 revs, 0.203956 s, 0.199063 s, -0.004893 s, × 0.9760, 6 µs/rev
mozilla-central x00000_revs_x0000_added_x0000_copies 6832ae71433c 4c222a1d9a00 : 153721 revs, 1.763853 s, 1.277320 s, -0.486533 s, × 0.7242, 8 µs/rev
mozilla-central x00000_revs_x00000_added_x000_copies 76caed42cf7c 1daa622bbe42 : 204976 revs, 2.609761 s, 1.698794 s, -0.910967 s, × 0.6509, 8 µs/rev
mozilla-try x_revs_x_added_0_copies aaf6dde0deb8 9790f499805a : 2 revs, 0.000847 s, 0.000842 s, -0.000005 s, × 0.9941, 421 µs/rev
mozilla-try x_revs_x000_added_0_copies d8d0222927b4 5bb8ce8c7450 : 2 revs, 0.000867 s, 0.000865 s, -0.000002 s, × 0.9977, 432 µs/rev
mozilla-try x_revs_x_added_x_copies 092fcca11bdb 936255a0384a : 4 revs, 0.000161 s, 0.000160 s, -0.000001 s, × 0.9938, 40 µs/rev
mozilla-try x_revs_x00_added_x_copies b53d2fadbdb5 017afae788ec : 2 revs, 0.001131 s, 0.001122 s, -0.000009 s, × 0.9920, 561 µs/rev
mozilla-try x_revs_x000_added_x000_copies 20408ad61ce5 6f0ee96e21ad : 1 revs, 0.033114 s, 0.032743 s, -0.000371 s, × 0.9888, 32743 µs/rev
mozilla-try x_revs_x0000_added_x0000_copies effb563bb7e5 c07a39dc4e80 : 6 revs, 0.071092 s, 0.071529 s, +0.000437 s, × 1.0061, 11921 µs/rev
mozilla-try x000_revs_xx00_added_0_copies 6100d773079a 04a55431795e : 1593 revs, 0.006554 s, 0.006593 s, +0.000039 s, × 1.0060, 4 µs/rev
mozilla-try x000_revs_x000_added_x_copies 9f17a6fc04f9 2d37b966abed : 41 revs, 0.005160 s, 0.005311 s, +0.000151 s, × 1.0293, 129 µs/rev
mozilla-try x000_revs_x000_added_x000_copies 1346fd0130e4 4c65cbdabc1f : 6657 revs, 0.065063 s, 0.063063 s, -0.002000 s, × 0.9693, 9 µs/rev
mozilla-try x0000_revs_x_added_0_copies 63519bfd42ee a36a2a865d92 : 40314 revs, 0.297118 s, 0.312363 s, +0.015245 s, × 1.0513, 7 µs/rev
mozilla-try x0000_revs_x_added_x_copies 9fe69ff0762d bcabf2a78927 : 38690 revs, 0.284002 s, 0.283106 s, -0.000896 s, × 0.9968, 7 µs/rev
mozilla-try x0000_revs_xx000_added_x_copies 156f6e2674f2 4d0f2c178e66 : 8598 revs, 0.086311 s, 0.083817 s, -0.002494 s, × 0.9711, 9 µs/rev
mozilla-try x0000_revs_xx000_added_0_copies 9eec5917337d 67118cc6dcad : 615 revs, 0.026738 s, 0.026516 s, -0.000222 s, × 0.9917, 43 µs/rev
mozilla-try x0000_revs_xx000_added_x000_copies 89294cd501d9 7ccb2fc7ccb5 : 97052 revs, 1.514270 s, 1.304865 s, -0.209405 s, × 0.8617, 13 µs/rev
mozilla-try x0000_revs_x0000_added_x0000_copies e928c65095ed e951f4ad123a : 52031 revs, 0.735875 s, 0.681088 s, -0.054787 s, × 0.9255, 13 µs/rev
mozilla-try x00000_revs_x_added_0_copies 6a320851d377 1ebb79acd503 : 363753 revs, 4.843329 s, 4.454320 s, -0.389009 s, × 0.9197, 12 µs/rev
mozilla-try x00000_revs_x00000_added_0_copies dc8a3ca7010e d16fde900c9c : 34414 revs, 0.591752 s, 0.567913 s, -0.023839 s, × 0.9597, 16 µs/rev
mozilla-try x00000_revs_x_added_x_copies 5173c4b6f97c 95d83ee7242d : 362229 revs, 4.760563 s, 4.547043 s, -0.213520 s, × 0.9551, 12 µs/rev
mozilla-try x00000_revs_x000_added_x_copies 9126823d0e9c ca82787bb23c : 359344 revs, 4.751942 s, 4.378579 s, -0.373363 s, × 0.9214, 12 µs/rev
mozilla-try x00000_revs_x0000_added_x0000_copies 8d3fafa80d4b eb884023b810 : 192665 revs, 2.605014 s, 1.703622 s, -0.901392 s, × 0.6540, 8 µs/rev
mozilla-try x00000_revs_x00000_added_x0000_copies 1b661134e2ca 1ae03d022d6d : 228985 revs, 41.113063 s, 36.001255 s, -5.111808 s, × 0.8757, 157 µs/rev
mozilla-try x00000_revs_x00000_added_x000_copies 9b2a99adc05e 8e29777b48e6 : 382065 revs, 27.891612 s, 14.340641 s, -13.550971 s, × 0.5142, 37 µs/rev

Diff Detail

Repository
rHG Mercurial
Branch
default
Lint
No Linters Available
Unit
No Unit Test Coverage

Event Timeline

marmoute created this revision.Dec 2 2020, 11:02 AM
baymax edited the summary of this revision. (Show Details)Dec 2 2020, 1:19 PM
baymax updated this revision to Diff 23961.
Alphare accepted this revision.Dec 14 2020, 4:53 AM
marmoute updated this revision to Diff 24220.Dec 14 2020, 6:17 AM
baymax updated this revision to Diff 24241.Dec 14 2020, 7:48 AM

✅ refresh by Heptapod after a successful CI run (🐙 💚)

baymax updated this revision to Diff 24358.Dec 17 2020, 3:47 PM

✅ refresh by Heptapod after a successful CI run (🐙 💚)

pulkit accepted this revision.Dec 18 2020, 7:38 AM
This revision is now accepted and ready to land.Dec 18 2020, 7:38 AM
marmoute added a commit: rHGfce2f20a54ce: copies-rust: start recording overwrite as they happens.Dec 18 2020, 3:05 PM
Closed by commit rHGfce2f20a54ce: copies-rust: start recording overwrite as they happens (authored by marmoute). · Explain Why
This revision was automatically updated to reflect the committed changes.

Revision Contents
Changeset List

PathPackages
Mrust/hg-core/src/copy_tracing.rs (11 lines)

Diff 23961

rust/hg-core/src/copy_tracing.rs

impl<'a, A: Fn(Revision, Revision) -> bool> AncestorOracle<'a, A> { impl<'a, A: Fn(Revision, Revision) -> bool> AncestorOracle<'a, A> {
fn new(func: &'a A) -> Self { fn new(func: &'a A) -> Self {
Self { Self {
inner: func, inner: func,
pairs: HashMap::default(), pairs: HashMap::default(),
} }
} }
fn record_overwrite(&mut self, anc: Revision, desc: Revision) {
self.pairs.insert((anc, desc), true);
}
/// returns `true` if `anc` is an ancestors of `desc`, `false` otherwise /// returns `true` if `anc` is an ancestors of `desc`, `false` otherwise
fn is_overwrite(&mut self, anc: Revision, desc: Revision) -> bool { fn is_overwrite(&mut self, anc: Revision, desc: Revision) -> bool {
if anc > desc { if anc > desc {
false false
} else if anc == desc { } else if anc == desc {
true true
} else { } else {
if let Some(b) = self.pairs.get(&(anc, desc)) { if let Some(b) = self.pairs.get(&(anc, desc)) {
if p1 != NULL_REVISION { if p1 != NULL_REVISION {
// Retrieve data computed in a previous iteration // Retrieve data computed in a previous iteration
let parent_copies = let parent_copies =
get_parent_copies(&mut all_copies, &mut children_count, p1); get_parent_copies(&mut all_copies, &mut children_count, p1);
if let Some(parent_copies) = parent_copies { if let Some(parent_copies) = parent_copies {
// combine it with data for that revision // combine it with data for that revision
let vertex_copies = add_from_changes( let vertex_copies = add_from_changes(
&mut path_map, &mut path_map,
&mut oracle,
&parent_copies, &parent_copies,
&changes, &changes,
Parent::FirstParent, Parent::FirstParent,
rev, rev,
); );
// keep that data around for potential later combination // keep that data around for potential later combination
copies = Some(vertex_copies); copies = Some(vertex_copies);
} }
} }
if p2 != NULL_REVISION { if p2 != NULL_REVISION {
// Retrieve data computed in a previous iteration // Retrieve data computed in a previous iteration
let parent_copies = let parent_copies =
get_parent_copies(&mut all_copies, &mut children_count, p2); get_parent_copies(&mut all_copies, &mut children_count, p2);
if let Some(parent_copies) = parent_copies { if let Some(parent_copies) = parent_copies {
// combine it with data for that revision // combine it with data for that revision
let vertex_copies = add_from_changes( let vertex_copies = add_from_changes(
&mut path_map, &mut path_map,
&mut oracle,
&parent_copies, &parent_copies,
&changes, &changes,
Parent::SecondParent, Parent::SecondParent,
rev, rev,
); );
copies = match copies { copies = match copies {
None => Some(vertex_copies), None => Some(vertex_copies),
Some(c) => Some(c.clone()), Some(c) => Some(c.clone()),
None => Some(TimeStampedPathCopies::default()), None => Some(TimeStampedPathCopies::default()),
} }
} }
} }
/// Combine ChangedFiles with some existing PathCopies information and return /// Combine ChangedFiles with some existing PathCopies information and return
/// the result /// the result
fn add_from_changes( fn add_from_changes<A: Fn(Revision, Revision) -> bool>(
path_map: &mut TwoWayPathMap, path_map: &mut TwoWayPathMap,
oracle: &mut AncestorOracle<A>,
base_copies: &TimeStampedPathCopies, base_copies: &TimeStampedPathCopies,
changes: &ChangedFiles, changes: &ChangedFiles,
parent: Parent, parent: Parent,
current_rev: Revision, current_rev: Revision,
) -> TimeStampedPathCopies { ) -> TimeStampedPathCopies {
let mut copies = base_copies.clone(); let mut copies = base_copies.clone();
for action in changes.iter_actions(parent) { for action in changes.iter_actions(parent) {
match action { match action {
let ttpc = TimeStampedPathCopy { let ttpc = TimeStampedPathCopy {
rev: current_rev, rev: current_rev,
path: entry, path: entry,
}; };
slot.insert(ttpc); slot.insert(ttpc);
} }
Entry::Occupied(mut slot) => { Entry::Occupied(mut slot) => {
let mut ttpc = slot.get_mut(); let mut ttpc = slot.get_mut();
oracle.record_overwrite(ttpc.rev, current_rev);
ttpc.rev = current_rev; ttpc.rev = current_rev;
ttpc.path = entry; ttpc.path = entry;
} }
} }
} }
Action::Removed(deleted_path) => { Action::Removed(deleted_path) => {
// We must drop copy information for removed file. // We must drop copy information for removed file.
// //
// We need to explicitly record them as dropped to // We need to explicitly record them as dropped to
// propagate this information when merging two // propagate this information when merging two
// TimeStampedPathCopies object. // TimeStampedPathCopies object.
let deleted = path_map.tokenize(deleted_path); let deleted = path_map.tokenize(deleted_path);
copies.entry(deleted).and_modify(|old| { copies.entry(deleted).and_modify(|old| {
oracle.record_overwrite(old.rev, current_rev);
old.rev = current_rev; old.rev = current_rev;
old.path = None; old.path = None;
}); });
} }
} }
} }
copies copies
} }