view tests/test-rebase-pull.t @ 29830:92ac2baaea86

revlog: use an LRU cache for delta chain bases

Profiling using statprof revealed a hotspot during changegroup
application calculating delta chain bases on generaldelta repos.
Essentially, revlog._addrevision() was performing a lot of redundant
work tracing the delta chain as part of determining when the chain
distance was acceptable. This was most pronounced when adding
revisions to manifests, which can have delta chains thousands of
revisions long.

There was a delta chain base cache on revlogs before, but it only
captured a single revision. This was acceptable before generaldelta,
when _addrevision would build deltas from the previous revision and
thus we'd pretty much guarantee a cache hit when resolving the delta
chain base on a subsequent _addrevision call. However, it isn't
suitable for generaldelta because parent revisions aren't necessarily
the last processed revision.

This patch converts the delta chain base cache to an LRU dict cache.
The cache can hold multiple entries, so generaldelta repos have a
higher chance of getting a cache hit.

The impact of this change when processing changegroup additions is
significant. On a generaldelta conversion of the "mozilla-unified"
repo (which contains heads of the main Firefox repositories in
chronological order - this means there are lots of transitions between
heads in revlog order), this change has the following impact when
performing an `hg unbundle` of an uncompressed bundle of the repo:

before: 5:42 CPU time
after:  4:34 CPU time

Most of this time is saved when applying the changelog and manifest
revlogs:

before: 2:30 CPU time
after:  1:17 CPU time

That nearly a 50% reduction in CPU time applying changesets and
manifests!

Applying a gzipped bundle of the same repo (effectively simulating a
`hg clone` over HTTP) showed a similar speedup:

before: 5:53 CPU time
after:  4:46 CPU time

Wall time improvements were basically the same as CPU time.

I didn't measure explicitly, but it feels like most of the time
is saved when processing manifests. This makes sense, as large
manifests tend to have very long delta chains and thus benefit the
most from this cache.

So, this change effectively makes changegroup application (which is
used by `hg unbundle`, `hg clone`, `hg pull`, `hg unshelve`, and
various other commands) significantly faster when delta chains are
long (which can happen on repos with large numbers of files and thus
large manifests).

In theory, this change can result in more memory utilization. However,
we're caching a dict of ints. At most we have 200 ints + Python object
overhead per revlog. And, the cache is really only populated when
performing read-heavy operations, such as adding changegroups or
scanning an individual revlog. For memory bloat to be an issue, we'd
need to scan/read several revisions from several revlogs all while
having active references to several revlogs. I don't think there are
many operations that do this, so I don't think memory bloat from the
cache will be an issue.

  $ cat >> $HGRCPATH <<EOF
  > [extensions]
  > rebase=
  > 
  > [alias]
  > tglog = log -G --template "{rev}: '{desc}' {branches}\n"
  > EOF


  $ hg init a
  $ cd a

  $ echo C1 > C1
  $ hg ci -Am C1
  adding C1

  $ echo C2 > C2
  $ hg ci -Am C2
  adding C2

  $ cd ..

  $ hg clone a b
  updating to branch default
  2 files updated, 0 files merged, 0 files removed, 0 files unresolved

  $ hg clone a c
  updating to branch default
  2 files updated, 0 files merged, 0 files removed, 0 files unresolved

  $ cd b

  $ echo L1 > L1
  $ hg ci -Am L1
  adding L1


  $ cd ../a

  $ echo R1 > R1
  $ hg ci -Am R1
  adding R1


  $ cd ../b

Now b has one revision to be pulled from a:

  $ hg pull --rebase
  pulling from $TESTTMP/a (glob)
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files (+1 heads)
  rebasing 2:ff8d69a621f9 "L1"
  saved backup bundle to $TESTTMP/b/.hg/strip-backup/ff8d69a621f9-160fa373-backup.hg (glob)

  $ hg tglog
  @  3: 'L1'
  |
  o  2: 'R1'
  |
  o  1: 'C2'
  |
  o  0: 'C1'
  
Re-run:

  $ hg pull --rebase
  pulling from $TESTTMP/a (glob)
  searching for changes
  no changes found


Invoke pull --rebase and nothing to rebase:

  $ cd ../c

  $ hg book norebase
  $ hg pull --rebase
  pulling from $TESTTMP/a (glob)
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  nothing to rebase - updating instead
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  updating bookmark norebase

  $ hg tglog -l 1
  @  2: 'R1'
  |
  ~

pull --rebase --update should ignore --update:

  $ hg pull --rebase --update
  pulling from $TESTTMP/a (glob)
  searching for changes
  no changes found

pull --rebase doesn't update if nothing has been pulled:

  $ hg up -q 1

  $ hg pull --rebase
  pulling from $TESTTMP/a (glob)
  searching for changes
  no changes found

  $ hg tglog -l 1
  o  2: 'R1'
  |
  ~

  $ cd ..

pull --rebase works when a specific revision is pulled (issue3619)

  $ cd a
  $ hg tglog
  @  2: 'R1'
  |
  o  1: 'C2'
  |
  o  0: 'C1'
  
  $ echo R2 > R2
  $ hg ci -Am R2
  adding R2
  $ echo R3 > R3
  $ hg ci -Am R3
  adding R3
  $ cd ../c
  $ hg tglog
  o  2: 'R1'
  |
  @  1: 'C2'
  |
  o  0: 'C1'
  
  $ echo L1 > L1
  $ hg ci -Am L1
  adding L1
  created new head
  $ hg pull --rev tip --rebase
  pulling from $TESTTMP/a (glob)
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 2 changesets with 2 changes to 2 files
  rebasing 3:ff8d69a621f9 "L1"
  saved backup bundle to $TESTTMP/c/.hg/strip-backup/ff8d69a621f9-160fa373-backup.hg (glob)
  $ hg tglog
  @  5: 'L1'
  |
  o  4: 'R3'
  |
  o  3: 'R2'
  |
  o  2: 'R1'
  |
  o  1: 'C2'
  |
  o  0: 'C1'
  
pull --rebase works with bundle2 turned on

  $ cd ../a
  $ echo R4 > R4
  $ hg ci -Am R4
  adding R4
  $ hg tglog
  @  5: 'R4'
  |
  o  4: 'R3'
  |
  o  3: 'R2'
  |
  o  2: 'R1'
  |
  o  1: 'C2'
  |
  o  0: 'C1'
  
  $ cd ../c
  $ hg pull --rebase
  pulling from $TESTTMP/a (glob)
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files (+1 heads)
  rebasing 5:518d153c0ba3 "L1"
  saved backup bundle to $TESTTMP/c/.hg/strip-backup/518d153c0ba3-73407f14-backup.hg (glob)
  $ hg tglog
  @  6: 'L1'
  |
  o  5: 'R4'
  |
  o  4: 'R3'
  |
  o  3: 'R2'
  |
  o  2: 'R1'
  |
  o  1: 'C2'
  |
  o  0: 'C1'
  

pull --rebase only update if there is nothing to rebase

  $ cd ../a
  $ echo R5 > R5
  $ hg ci -Am R5
  adding R5
  $ hg tglog
  @  6: 'R5'
  |
  o  5: 'R4'
  |
  o  4: 'R3'
  |
  o  3: 'R2'
  |
  o  2: 'R1'
  |
  o  1: 'C2'
  |
  o  0: 'C1'
  
  $ cd ../c
  $ echo L2 > L2
  $ hg ci -Am L2
  adding L2
  $ hg up 'desc(L1)'
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg pull --rebase
  pulling from $TESTTMP/a (glob)
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files (+1 heads)
  rebasing 6:0d0727eb7ce0 "L1"
  rebasing 7:c1f58876e3bf "L2"
  saved backup bundle to $TESTTMP/c/.hg/strip-backup/0d0727eb7ce0-ef61ccb2-backup.hg (glob)
  $ hg tglog
  o  8: 'L2'
  |
  @  7: 'L1'
  |
  o  6: 'R5'
  |
  o  5: 'R4'
  |
  o  4: 'R3'
  |
  o  3: 'R2'
  |
  o  2: 'R1'
  |
  o  1: 'C2'
  |
  o  0: 'C1'
  

pull --rebase update (no rebase) use proper update:

- warn about other head.

  $ cd ../a
  $ echo R6 > R6
  $ hg ci -Am R6
  adding R6
  $ cd ../c
  $ hg up 'desc(R5)'
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg pull --rebase
  pulling from $TESTTMP/a (glob)
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files (+1 heads)
  nothing to rebase - updating instead
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  1 other heads for branch "default"
  $ hg tglog
  @  9: 'R6'
  |
  | o  8: 'L2'
  | |
  | o  7: 'L1'
  |/
  o  6: 'R5'
  |
  o  5: 'R4'
  |
  o  4: 'R3'
  |
  o  3: 'R2'
  |
  o  2: 'R1'
  |
  o  1: 'C2'
  |
  o  0: 'C1'
  

Multiple pre-existing heads on the branch
-----------------------------------------

Pull bring content, but nothing on the current branch, we should not consider
pre-existing heads.

  $ cd ../a
  $ hg branch unrelatedbranch
  marked working directory as branch unrelatedbranch
  (branches are permanent and global, did you want a bookmark?)
  $ echo B1 > B1
  $ hg commit -Am B1
  adding B1
  $ cd ../c
  $ hg up 'desc(L2)'
  2 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg pull --rebase
  pulling from $TESTTMP/a (glob)
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  nothing to rebase

There is two local heads and we pull a third one.
The second local head should not confuse the `hg pull rebase`.

  $ hg up 'desc(R6)'
  1 files updated, 0 files merged, 2 files removed, 0 files unresolved
  $ echo M1 > M1
  $ hg commit -Am M1
  adding M1
  $ cd ../a
  $ hg up 'desc(R6)'
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ echo R7 > R7
  $ hg commit -Am R7
  adding R7
  $ cd ../c
  $ hg up 'desc(L2)'
  2 files updated, 0 files merged, 2 files removed, 0 files unresolved
  $ hg pull --rebase
  pulling from $TESTTMP/a (glob)
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files (+1 heads)
  rebasing 7:864e0a2d2614 "L1"
  rebasing 8:6dc0ea5dcf55 "L2"
  saved backup bundle to $TESTTMP/c/.hg/strip-backup/864e0a2d2614-2f72c89c-backup.hg (glob)
  $ hg tglog
  @  12: 'L2'
  |
  o  11: 'L1'
  |
  o  10: 'R7'
  |
  | o  9: 'M1'
  |/
  | o  8: 'B1' unrelatedbranch
  |/
  o  7: 'R6'
  |
  o  6: 'R5'
  |
  o  5: 'R4'
  |
  o  4: 'R3'
  |
  o  3: 'R2'
  |
  o  2: 'R1'
  |
  o  1: 'C2'
  |
  o  0: 'C1'