view tests/test-bookflow.t @ 51681:522b4d729e89

mmap: populate the mapping by default

Without pre-population, accessing all data through a mmap can result in many
pagefault, reducing performance significantly. If the mmap is prepopulated, the
performance can no longer get slower than a full read.
(See benchmark number below)

In some cases were very few data is read, prepopulating can be overkill and
slower than populating on access (through page fault). So that behavior can be
controlled when the caller can pre-determine the best behavior.
(See benchmark number below)

In addition, testing with populating in a secondary thread yield great result
combining the best of each approach. This might be implemented in later
changesets.

In all cases, using mmap has a great effect on memory usage when many processes
run in parallel on the same machine.

### Benchmarks

# What did I run

A couple of month back I ran a large benchmark campaign to assess the impact of
various approach for using mmap with the revlog (and other files), it
highlighted a few benchmarks that capture the impact of the changes well. So to
validate this change I checked the following:

- log command displaying various revisions
  (read the changelog index)
- log command displaying the patch of listed revisions
  (read the changelog index, the manifest index and a few files indexes)
- unbundling a few revisions
  (read and write changelog, manifest and few files indexes, and walk the graph
  to update some cache)
- pushing a few revisions
  (read and write changelog, manifest and few files indexes, walk the graph to
  update some cache, performs various accesses locally and remotely during
  discovery)

Benchmarks were run using the default module policy (c+py) and the rust one. No
significant difference were found between the two implementation, so we will
present result using the default policy (unless otherwise specified).

I ran them on a few repositories :

- mercurial: a "public changeset only" copy of mercurial from 2018-08-01 using
  zstd compression and sparse-revlog

- pypy: a copy of pypy from 2018-08-01 using zstd compression and sparse-revlog

- netbeans: a copy of netbeans from 2018-08-01 using zstd compression and
  sparse-revlog

- mozilla-try: a copy of mozilla-try from 2019-02-18 using zstd compression and
  sparse-revlog

- mozilla-try persistent-nodemap: Same as the above but with a persistent
  nodemap. Used for the log --patch benchmark only

# Results

For the smaller repositories (mercurial, pypy), the impact of mmap is almost
imperceptible, other cost dominating the operation. The impact of prepopulating
is undiscernible in the benchmark we ran.

For larger repositories the benchmark support explanation given above:

On netbeans, the log can be about 1% faster without repopulation (for a
difference < 100ms) but unbundle becomes a bit slower, even when small.

### data-env-vars.name                             = netbeans-2018-08-01-zstd-sparse-revlog
  # benchmark.name                                 = hg.command.unbundle
  # benchmark.variants.issue6528                   = disabled
  # benchmark.variants.reuse-external-delta-parent = yes
  # benchmark.variants.revs                        = any-1-extra-rev
  # benchmark.variants.source                      = unbundle
  # benchmark.variants.verbosity                   = quiet
with-populate: 0.240157
no-populate:   0.265087  (+10.38%, +0.02)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 1.459518
no-populate:   1.481290   (+1.49%, +0.02)
 ## benchmark.name                                 = hg.command.push
  # benchmark.variants.explicit-rev                = none
  # benchmark.variants.issue6528                   = disabled
  # benchmark.variants.protocol                    = ssh
  # benchmark.variants.reuse-external-delta-parent = yes
  # benchmark.variants.revs                        = any-1-extra-rev
with-populate: 0.771919
no-populate:   0.792025   (+2.60%, +0.02)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 1.459518
no-populate:   1.481290   (+1.49%, +0.02)


For mozilla-try, the "slow down" from pre-populate for small `hg log` is more
visible, but still small in absolute time. (using rust value for the persistent
nodemap value to be relevant).

### data-env-vars.name            = mozilla-try-2019-02-18-ds2-pnm
  # benchmark.name                = hg.command.log
  # bin-env-vars.hg.flavor        = rust
  # benchmark.variants.patch      = yes
  # benchmark.variants.limit-rev  = 1
with-populate: 0.237813
no-populate:   0.229452   (-3.52%, -0.01)
  # benchmark.variants.limit-rev  = 10
  # benchmark.variants.patch      = yes
with-populate: 1.213578
no-populate:   1.205189
### data-env-vars.name            = mozilla-try-2019-02-18-zstd-sparse-revlog
  # benchmark.variants.limit-rev  = 1000
  # benchmark.variants.patch      = no
  # benchmark.variants.rev        = tip
with-populate: 0.198607
no-populate:   0.195038   (-1.80%, -0.00)

However pre-populating provide a significant boost on more complex operations
like unbundle or push:

### data-env-vars.name                             = mozilla-try-2019-02-18-zstd-sparse-revlog
  # benchmark.name                                 = hg.command.push
  # benchmark.variants.explicit-rev                = none
  # benchmark.variants.issue6528                   = disabled
  # benchmark.variants.protocol                    = ssh
  # benchmark.variants.reuse-external-delta-parent = yes
  # benchmark.variants.revs                        = any-1-extra-rev
with-populate: 4.798632
no-populate:   4.953295   (+3.22%, +0.15)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 4.903618
no-populate:   5.014963   (+2.27%, +0.11)
 ## benchmark.name                                 = hg.command.unbundle
  # benchmark.variants.revs                        = any-1-extra-rev
with-populate: 1.423411
no-populate:   1.585365  (+11.38%, +0.16)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 1.537909
no-populate:   1.688489   (+9.79%, +0.15)

initialize
  $ make_changes() {
  >     d=`pwd`
  >     [ ! -z $1 ] && cd $1
  >     echo "test `basename \`pwd\``" >> test
  >     hg commit -Am"${2:-test}"
  >     r=$?
  >     cd $d
  >     return $r
  > }
  $ ls -1a
  .
  ..
  $ hg init a
  $ cd a
  $ echo 'test' > test; hg commit -Am'test'
  adding test

clone to b

  $ mkdir ../b
  $ cd ../b
  $ hg clone ../a .
  updating to branch default
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ echo "[extensions]" >> .hg/hgrc
  $ echo "bookflow=" >> .hg/hgrc
  $ hg branch X
  abort: creating named branches is disabled and you should use bookmarks
  (see 'hg help bookflow')
  [255]
  $ hg bookmark X
  $ hg bookmarks
  * X                         0:* (glob)
  $ hg bookmark X
  abort: bookmark X already exists, to move use the --rev option
  [255]
  $ make_changes
  $ hg push ../a -q

  $ hg bookmarks
   \* X                         1:* (glob)

change a
  $ cd ../a
  $ hg up
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ echo 'test' >> test; hg commit -Am'test'


pull in b
  $ cd ../b
  $ hg pull -u
  pulling from $TESTTMP/a
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  new changesets * (glob)
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  (leaving bookmark X)
  $ hg status
  $ hg bookmarks
     X                         1:* (glob)

check protection of @ bookmark
  $ hg bookmark @
  $ hg bookmarks
   \* @                         2:* (glob)
     X                         1:* (glob)
  $ make_changes
  abort: cannot commit, bookmark @ is protected
  [255]

  $ hg status
  M test
  $ hg bookmarks
   \* @                         2:* (glob)
     X                         1:* (glob)

  $ hg --config bookflow.protect= commit  -Am"Updated test"

  $ hg bookmarks
   \* @                         3:* (glob)
     X                         1:* (glob)

check requirement for an active bookmark
  $ hg bookmark -i
  $ hg bookmarks
     @                         3:* (glob)
     X                         1:* (glob)
  $ make_changes
  abort: cannot commit without an active bookmark
  [255]
  $ hg revert test
  $ rm test.orig
  $ hg status


make the bookmark move by updating it on a, and then pulling
# add a commit to a
  $ cd ../a
  $ hg bookmark X
  $ hg bookmarks
   \* X                         2:* (glob)
  $ make_changes
  $ hg bookmarks
   * X                         3:81af7977fdb9

# go back to b, and check out X
  $ cd ../b
  $ hg up X
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  (activating bookmark X)
  $ hg bookmarks
     @                         3:* (glob)
   \* X                         1:* (glob)

# pull, this should move the bookmark forward, because it was changed remotely
  $ hg pull -u | grep "updating to active bookmark X"
  updating to active bookmark X

  $ hg bookmarks
     @                         3:* (glob)
   * X                         4:81af7977fdb9

the bookmark should not move if it diverged from remote
  $ hg -R ../a status
  $ hg -R ../b status
  $ make_changes ../a
  $ make_changes ../b
  $ hg -R ../a status
  $ hg -R ../b status
  $ hg -R ../a bookmarks
   * X                         4:238292f60a57
  $ hg -R ../b bookmarks
     @                         3:* (glob)
   * X                         5:096f7e86892d
  $ cd ../b
  $ # make sure we cannot push after bookmarks diverged
  $ hg push -B X | grep abort
  abort: push creates new remote head * with bookmark 'X' (glob)
  (pull and merge or see 'hg help push' for details about pushing new heads)
  [1]
  $ hg pull -u | grep divergent
  divergent bookmark X stored as X@default
  1 other divergent bookmarks for "X"
  $ hg bookmarks
     @                         3:* (glob)
   * X                         5:096f7e86892d
     X@default                 6:238292f60a57
  $ hg id -in
  096f7e86892d 5
  $ make_changes
  $ hg status
  $ hg bookmarks
     @                         3:* (glob)
   * X                         7:227f941aeb07
     X@default                 6:238292f60a57

now merge with the remote bookmark
  $ hg merge X@default --tool :local -q
  $ hg status
  M test
  $ hg commit -m"Merged with X@default"
  $ hg bookmarks
     @                         3:* (glob)
   * X                         8:26fed9bb3219
  $ hg push -B X | grep bookmark
  pushing to $TESTTMP/a (?)
  updating bookmark X
  $ cd ../a
  $ hg up -q
  $ hg bookmarks
   * X                         7:26fed9bb3219

test hg pull when there is more than one descendant
  $ cd ../a
  $ hg bookmark Z
  $ hg bookmark Y
  $ make_changes . YY
  $ hg up Z -q
  $ make_changes . ZZ
  created new head
  $ hg bookmarks
     X                         7:26fed9bb3219
     Y                         8:131e663dbd2a
   * Z                         9:b74a4149df25
  $ hg log -r 'p1(Y)' -r 'p1(Z)' -T '{rev}\n' # prove that Y and Z share the same parent
  7
  $ hg log -r 'Y%Z' -T '{rev}\n'  # revs in Y but not in Z
  8
  $ hg log -r 'Z%Y' -T '{rev}\n'  # revs in Z but not in Y
  9
  $ cd ../b
  $ hg pull -uq
  $ hg id
  b74a4149df25 tip Z
  $ hg bookmarks | grep \*  # no active bookmark
  [1]


test shelving
  $ cd ../a
  $ echo anotherfile > anotherfile # this change should not conflict
  $ hg add anotherfile
  $ hg commit -m"Change in a"
  $ cd ../b
  $ hg up Z | grep Z
  (activating bookmark Z)
  $ hg book | grep \* # make sure active bookmark
   \* Z                         10:* (glob)
  $ echo "test b" >> test
  $ hg diff --stat
   test |  1 +
   1 files changed, 1 insertions(+), 0 deletions(-)
  $ hg shelve
  shelved as Z
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ hg pull -uq
  $ hg unshelve
  unshelving change 'Z'
  rebasing shelved changes
  $ hg diff --stat
   test |  1 +
   1 files changed, 1 insertions(+), 0 deletions(-)


make the bookmark move by updating it on a, and then pulling with a local change
# add a commit to a
  $ cd ../a
  $ hg up -C X |grep -F  "activating bookmark X"
  (activating bookmark X)
# go back to b, and check out X
  $ cd ../b
  $ hg up -C X |grep -F  "activating bookmark X"
  (activating bookmark X)
# update and push from a
  $ make_changes ../a
  created new head
  $ echo "more" >> test
  $ hg pull -u 2>&1 | grep -F -v TESTTMP| grep -F -v "searching for changes" | grep -F -v adding
  pulling from $TESTTMP/a
  updating bookmark X
  added 1 changesets with 0 changes to 0 files (+1 heads)
  new changesets * (glob)
  updating to active bookmark X
  merging test
  warning: conflicts while merging test! (edit, then use 'hg resolve --mark')
  0 files updated, 0 files merged, 0 files removed, 1 files unresolved
  use 'hg resolve' to retry unresolved file merges
  $ hg update -Cq
  $ rm test.orig

make sure that commits aren't possible if working directory is not pointing to active bookmark
  $ hg -R ../a status
  $ hg -R ../b status
  $ hg -R ../a id -i
  36a6e592ec06
  $ hg -R ../a book | grep X
   \* X                         \d+:36a6e592ec06 (re)
  $ hg -R ../b id -i
  36a6e592ec06
  $ hg -R ../b book | grep X
   \* X                         \d+:36a6e592ec06 (re)
  $ make_changes ../a
  $ hg -R ../a book | grep X
   \* X                         \d+:f73a71c992b8 (re)
  $ cd ../b
  $ hg pull  2>&1 | grep -v add | grep -v pulling | grep -v searching | grep -v changeset
  updating bookmark X
  (run 'hg update' to get a working copy)
  working directory out of sync with active bookmark, run 'hg up X'
  $ hg id -i # we're still on the old commit
  36a6e592ec06
  $ hg book | grep X # while the bookmark moved
   \* X                         \d+:f73a71c992b8 (re)
  $ make_changes
  abort: cannot commit, working directory out of sync with active bookmark
  (run 'hg up X')
  [255]
  $ hg up -Cq -r .  # cleanup local changes
  $ hg status
  $ hg id -i # we're still on the old commit
  36a6e592ec06
  $ hg up X -q
  $ hg id -i # now we're on X
  f73a71c992b8
  $ hg book | grep X
   \* X                         \d+:f73a71c992b8 (re)