Mercurial > hg
view tests/test-rebase-collapse.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)
author | Pierre-Yves David <pierre-yves.david@octobus.net> |
---|---|
date | Thu, 11 Apr 2024 00:02:07 +0200 |
parents | 402a6b6173e9 |
children |
line wrap: on
line source
$ cat >> $HGRCPATH <<EOF > [extensions] > rebase= > mq= > drawdag=$TESTDIR/drawdag.py > > [phases] > publish=False > > [alias] > tglog = log -G --template "{rev}: {node|short} '{desc}' {branches}\n" > tglogp = log -G --template "{rev}: {node|short} {phase} '{desc}' {branches}\n" > EOF Highest phase of source commits is used: $ hg init phase $ cd phase $ hg debugdrawdag << 'EOF' > D > | > F C > | | > E B > |/ > A > EOF $ hg phase --force --secret D $ cat > $TESTTMP/editor.sh <<EOF > echo "==== before editing" > cat \$1 > echo "====" > echo "edited manually" >> \$1 > EOF $ HGEDITOR="sh $TESTTMP/editor.sh" hg rebase --collapse --keepbranches -e --source B --dest F rebasing 1:112478962961 B "B" rebasing 3:26805aba1e60 C "C" rebasing 5:f585351a92f8 D tip "D" ==== before editing Collapsed revision * B * C * D HG: Enter commit message. Lines beginning with 'HG:' are removed. HG: Leave message empty to abort commit. HG: -- HG: user: test HG: branch 'default' HG: added B HG: added C HG: added D ==== saved backup bundle to $TESTTMP/phase/.hg/strip-backup/112478962961-cb2a9b47-rebase.hg $ hg tglogp o 3: 92fa5f5fe108 secret 'Collapsed revision | * B | * C | * D | | | edited manually' o 2: 64a8289d2492 draft 'F' | o 1: 7fb047a69f22 draft 'E' | o 0: 426bada5c675 draft 'A' $ hg manifest --rev tip A B C D E F $ cd .. Merge gets linearized: $ hg init linearized-merge $ cd linearized-merge $ hg debugdrawdag << 'EOF' > F D > |/| > C B > |/ > A > EOF $ hg phase --force --secret D $ hg rebase --source B --collapse --dest F rebasing 1:112478962961 B "B" rebasing 3:4e4f9194f9f1 D "D" saved backup bundle to $TESTTMP/linearized-merge/.hg/strip-backup/112478962961-e389075b-rebase.hg $ hg tglog o 3: 5bdc08b7da2b 'Collapsed revision | * B | * D' o 2: afc707c82df0 'F' | o 1: dc0947a82db8 'C' | o 0: 426bada5c675 'A' $ hg manifest --rev tip A B C F $ cd .. Custom message: $ hg init message $ cd message $ hg debugdrawdag << 'EOF' > C > | > D B > |/ > A > EOF $ hg rebase --base B -m 'custom message' abort: message can only be specified with collapse [10] $ cat > $TESTTMP/checkeditform.sh <<EOF > env | grep HGEDITFORM > true > EOF $ HGEDITOR="sh $TESTTMP/checkeditform.sh" hg rebase --source B --collapse -m 'custom message' -e --dest D rebasing 1:112478962961 B "B" rebasing 3:26805aba1e60 C tip "C" HGEDITFORM=rebase.collapse saved backup bundle to $TESTTMP/message/.hg/strip-backup/112478962961-f4131707-rebase.hg $ hg tglog o 2: 2f197b9a08f3 'custom message' | o 1: b18e25de2cf5 'D' | o 0: 426bada5c675 'A' $ hg manifest --rev tip A B C D $ cd .. Rebase and collapse - more than one external (fail): $ hg init multiple-external-parents $ cd multiple-external-parents $ hg debugdrawdag << 'EOF' > G > |\ > | F > | | > D E > |\| > H C B > \|/ > A > EOF $ hg rebase -s C --dest H --collapse abort: unable to collapse on top of 3, there is more than one external parent: 1, 6 [20] Rebase and collapse - E onto H: $ hg rebase -s E --dest H --collapse # root (E) is not a merge rebasing 5:49cb92066bfd E "E" rebasing 6:11abe3fb10b8 F "F" rebasing 7:64e264db77f0 G tip "G" saved backup bundle to $TESTTMP/multiple-external-parents/.hg/strip-backup/49cb92066bfd-ee8a8a79-rebase.hg $ hg tglog o 5: 8b2315790719 'Collapsed revision |\ * E | | * F | | * G' | o 4: 4e4f9194f9f1 'D' | |\ o | | 3: 575c4b5ec114 'H' | | | +---o 2: dc0947a82db8 'C' | | | o 1: 112478962961 'B' |/ o 0: 426bada5c675 'A' $ hg manifest --rev tip A C E F H $ cd .. Test that branchheads cache is updated correctly when doing a strip in which the parent of the ancestor node to be stripped does not become a head and also, the parent of a node that is a child of the node stripped becomes a head (node 3). The code is now much simpler and we could just test a simpler scenario We keep it the test this way in case new complexity is injected. Create repo b: $ hg init branch-heads $ cd branch-heads $ hg debugdrawdag << 'EOF' > G > |\ > | F > | | > D E > |\| > H C B > \|/ > A > EOF $ hg heads --template="{rev}:{node} {branch}\n" 7:64e264db77f061f16d9132b70c5a58e2461fb630 default 3:575c4b5ec114d64b681d33f8792853568bfb2b2c default $ cat $TESTTMP/branch-heads/.hg/cache/branch2-served 64e264db77f061f16d9132b70c5a58e2461fb630 7 575c4b5ec114d64b681d33f8792853568bfb2b2c o default 64e264db77f061f16d9132b70c5a58e2461fb630 o default $ hg strip 4 saved backup bundle to $TESTTMP/branch-heads/.hg/strip-backup/4e4f9194f9f1-5ec4b5e6-backup.hg $ cat $TESTTMP/branch-heads/.hg/cache/branch2-served 11abe3fb10b8689b560681094b17fe161871d043 5 dc0947a82db884575bb76ea10ac97b08536bfa03 o default 575c4b5ec114d64b681d33f8792853568bfb2b2c o default 11abe3fb10b8689b560681094b17fe161871d043 o default $ hg heads --template="{rev}:{node} {branch}\n" 5:11abe3fb10b8689b560681094b17fe161871d043 default 3:575c4b5ec114d64b681d33f8792853568bfb2b2c default 2:dc0947a82db884575bb76ea10ac97b08536bfa03 default $ cd .. Preserves external parent $ hg init external-parent $ cd external-parent $ hg debugdrawdag << 'EOF' > H > |\ > | G > | | > | F # F/E = F\n > | | > D E # D/D = D\n > |\| > I C B > \|/ > A > EOF $ hg rebase -s F --dest I --collapse # root (F) is not a merge rebasing 6:c82b08f646f1 F "F" file 'E' was deleted in local [dest] but was modified in other [source]. You can use (c)hanged version, leave (d)eleted, or leave (u)nresolved. What do you want to do? u unresolved conflicts (see 'hg resolve', then 'hg rebase --continue') [240] $ echo F > E $ hg resolve -m (no more unresolved files) continue: hg rebase --continue $ hg rebase -c rebasing 6:c82b08f646f1 F "F" rebasing 7:a6db7fa104e1 G "G" rebasing 8:e1d201b72d91 H tip "H" saved backup bundle to $TESTTMP/external-parent/.hg/strip-backup/c82b08f646f1-f2721fbf-rebase.hg $ hg tglog o 6: 681daa3e686d 'Collapsed revision |\ * F | | * G | | * H' | | o 5: 49cb92066bfd 'E' | | | | o | 4: 09143c0bf13e 'D' | |\| o | | 3: 08ebfeb61bac 'I' | | | | o | 2: dc0947a82db8 'C' |/ / | o 1: 112478962961 'B' |/ o 0: 426bada5c675 'A' $ hg manifest --rev tip A C D E F G I $ hg up tip -q $ cat E F $ cd .. Rebasing from multiple bases: $ hg init multiple-bases $ cd multiple-bases $ hg debugdrawdag << 'EOF' > C B > D |/ > |/ > A > EOF $ hg rebase --collapse -r 'B+C' -d D rebasing 1:fc2b737bb2e5 B "B" rebasing 2:dc0947a82db8 C "C" saved backup bundle to $TESTTMP/multiple-bases/.hg/strip-backup/dc0947a82db8-b0c1a7ea-rebase.hg $ hg tglog o 2: 2127ae44d291 'Collapsed revision | * B | * C' o 1: b18e25de2cf5 'D' | o 0: 426bada5c675 'A' $ cd .. With non-contiguous commits: $ hg init non-contiguous $ cd non-contiguous $ cat >> .hg/hgrc <<EOF > [experimental] > evolution=all > EOF $ hg debugdrawdag << 'EOF' > F > | > E > | > D > | > C > | > B G > |/ > A > EOF BROKEN: should be allowed $ hg rebase --collapse -r 'B+D+F' -d G abort: unable to collapse on top of 2, there is more than one external parent: 3, 5 [20] $ cd .. $ hg init multiple-external-parents-2 $ cd multiple-external-parents-2 $ hg debugdrawdag << 'EOF' > D G > |\ /| > B C E F > \| |/ > \ H / > \|/ > A > EOF $ hg rebase --collapse -d H -s 'B+F' abort: unable to collapse on top of 5, there is more than one external parent: 1, 3 [20] $ cd .. With internal merge: $ hg init internal-merge $ cd internal-merge $ hg debugdrawdag << 'EOF' > E > |\ > C D > |/ > F B > |/ > A > EOF $ hg rebase -s B --collapse --dest F rebasing 1:112478962961 B "B" rebasing 3:26805aba1e60 C "C" rebasing 4:be0ef73c17ad D "D" rebasing 5:02c4367d6973 E tip "E" saved backup bundle to $TESTTMP/internal-merge/.hg/strip-backup/112478962961-1dfb057b-rebase.hg $ hg tglog o 2: c0512a1797b0 'Collapsed revision | * B | * C | * D | * E' o 1: 8908a377a434 'F' | o 0: 426bada5c675 'A' $ hg manifest --rev tip A B C D F $ cd .. Interactions between collapse and keepbranches $ hg init e $ cd e $ echo 'a' > a $ hg ci -Am 'A' adding a $ hg branch 'one' marked working directory as branch one (branches are permanent and global, did you want a bookmark?) $ echo 'b' > b $ hg ci -Am 'B' adding b $ hg branch 'two' marked working directory as branch two $ echo 'c' > c $ hg ci -Am 'C' adding c $ hg up -q 0 $ echo 'd' > d $ hg ci -Am 'D' adding d $ hg tglog @ 3: 41acb9dca9eb 'D' | | o 2: 8ac4a08debf1 'C' two | | | o 1: 1ba175478953 'B' one |/ o 0: 1994f17a630e 'A' $ hg rebase --keepbranches --collapse -s 1 -d 3 abort: cannot collapse multiple named branches [10] $ cd .. Rebase, collapse and copies $ hg init copies $ cd copies $ hg unbundle "$TESTDIR/bundles/renames.hg" adding changesets adding manifests adding file changes added 4 changesets with 11 changes to 7 files (+1 heads) new changesets f447d5abf5ea:338e84e2e558 (4 drafts) (run 'hg heads' to see heads, 'hg merge' to merge) $ hg up -q tip $ hg tglog @ 3: 338e84e2e558 'move2' | o 2: 6e7340ee38c0 'move1' | | o 1: 1352765a01d4 'change' |/ o 0: f447d5abf5ea 'add' $ hg rebase --collapse -d 1 rebasing 2:6e7340ee38c0 "move1" merging a and d to d merging b and e to e merging c and f to f rebasing 3:338e84e2e558 tip "move2" merging f and c to c merging e and g to g saved backup bundle to $TESTTMP/copies/.hg/strip-backup/6e7340ee38c0-ef8ef003-rebase.hg $ hg st $ hg st --copies --change tip A d a A g b R b $ hg up tip -q $ cat c c c $ cat d a a $ cat g b b $ hg log -r . --template "{file_copies}\n" d (a)g (b) Test collapsing a middle revision in-place $ hg tglog @ 2: 64b456429f67 'Collapsed revision | * move1 | * move2' o 1: 1352765a01d4 'change' | o 0: f447d5abf5ea 'add' $ hg rebase --collapse -r 1 -d 0 abort: cannot rebase changeset, as that will orphan 1 descendants (see 'hg help evolution.instability') [10] Test collapsing in place $ hg rebase --collapse -b . -d 0 rebasing 1:1352765a01d4 "change" rebasing 2:64b456429f67 tip "Collapsed revision" saved backup bundle to $TESTTMP/copies/.hg/strip-backup/1352765a01d4-45a352ea-rebase.hg $ hg st --change tip --copies M a M c A d a A g b R b $ hg up tip -q $ cat a a a $ cat c c c $ cat d a a $ cat g b b $ cd .. Test stripping a revision with another child $ hg init f $ cd f $ hg debugdrawdag << 'EOF' > C B > |/ > A > EOF $ hg heads --template="{rev}:{node} {branch}: {desc}\n" 2:dc0947a82db884575bb76ea10ac97b08536bfa03 default: C 1:112478962961147124edd43549aedd1a335e44bf default: B $ hg strip C saved backup bundle to $TESTTMP/f/.hg/strip-backup/dc0947a82db8-d21b92a4-backup.hg $ hg tglog o 1: 112478962961 'B' | o 0: 426bada5c675 'A' $ hg heads --template="{rev}:{node} {branch}: {desc}\n" 1:112478962961147124edd43549aedd1a335e44bf default: B $ cd .. Test collapsing changes that add then remove a file $ hg init collapseaddremove $ cd collapseaddremove $ touch base $ hg commit -Am base adding base $ touch a $ hg commit -Am a adding a $ hg rm a $ touch b $ hg commit -Am b adding b $ hg book foo $ hg rebase -d 0 -r "1::2" --collapse -m collapsed rebasing 1:6d8d9f24eec3 "a" rebasing 2:1cc73eca5ecc foo tip "b" saved backup bundle to $TESTTMP/collapseaddremove/.hg/strip-backup/6d8d9f24eec3-77d3b6e2-rebase.hg $ hg log -G --template "{rev}: '{desc}' {bookmarks}" @ 1: 'collapsed' foo | o 0: 'base' $ hg manifest --rev tip b base $ cd .. Test that rebase --collapse will remember message after running into merge conflict and invoking rebase --continue. $ hg init collapse_remember_message $ cd collapse_remember_message $ hg debugdrawdag << 'EOF' > C B # B/A = B\n > |/ # C/A = C\n > A > EOF $ hg rebase --collapse -m "new message" -b B -d C rebasing 1:81e5401e4d37 B "B" merging A warning: conflicts while merging A! (edit, then use 'hg resolve --mark') unresolved conflicts (see 'hg resolve', then 'hg rebase --continue') [240] $ rm A.orig $ hg resolve --mark A (no more unresolved files) continue: hg rebase --continue $ hg rebase --continue rebasing 1:81e5401e4d37 B "B" saved backup bundle to $TESTTMP/collapse_remember_message/.hg/strip-backup/81e5401e4d37-96c3dd30-rebase.hg $ hg log changeset: 2:17186933e123 tag: tip user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: new message changeset: 1:043039e9df84 tag: C user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: C changeset: 0:426bada5c675 tag: A user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: A $ cd .. Test aborted editor on final message $ HGMERGE=:merge3 $ export HGMERGE $ hg init aborted-editor $ cd aborted-editor $ hg debugdrawdag << 'EOF' > C # D/A = D\n > | # C/A = C\n > B D # B/A = B\n > |/ # A/A = A\n > A > EOF $ hg rebase --collapse -t internal:merge3 -s B -d D rebasing 1:f899f3910ce7 B "B" merging A warning: conflicts while merging A! (edit, then use 'hg resolve --mark') unresolved conflicts (see 'hg resolve', then 'hg rebase --continue') [240] $ hg tglog o 3: 63668d570d21 'C' | | @ 2: 82b8abf9c185 'D' | | % | 1: f899f3910ce7 'B' |/ o 0: 4a2df7238c3b 'A' $ cat A <<<<<<< dest: 82b8abf9c185 D - test: D D ||||||| parent of source: 4a2df7238c3b A - test: A A ======= B >>>>>>> source: f899f3910ce7 B - test: B $ echo BC > A $ hg resolve -m (no more unresolved files) continue: hg rebase --continue $ hg rebase --continue rebasing 1:f899f3910ce7 B "B" rebasing 3:63668d570d21 C tip "C" merging A warning: conflicts while merging A! (edit, then use 'hg resolve --mark') unresolved conflicts (see 'hg resolve', then 'hg rebase --continue') [240] $ hg tglog % 3: 63668d570d21 'C' | | @ 2: 82b8abf9c185 'D' | | o | 1: f899f3910ce7 'B' |/ o 0: 4a2df7238c3b 'A' $ cat A <<<<<<< dest: 82b8abf9c185 D - test: D BC ||||||| parent of source: f899f3910ce7 B - test: B B ======= C >>>>>>> source: 63668d570d21 C tip - test: C $ echo BD > A $ hg resolve -m (no more unresolved files) continue: hg rebase --continue $ HGEDITOR=false hg rebase --continue --config ui.interactive=1 already rebased 1:f899f3910ce7 B "B" as 82b8abf9c185 rebasing 3:63668d570d21 C tip "C" abort: edit failed: false exited with status 1 [250] $ hg tglog o 3: 63668d570d21 'C' | | @ 2: 82b8abf9c185 'D' | | o | 1: f899f3910ce7 'B' |/ o 0: 4a2df7238c3b 'A' $ hg rebase --continue already rebased 1:f899f3910ce7 B "B" as 82b8abf9c185 already rebased 3:63668d570d21 C tip "C" as 82b8abf9c185 saved backup bundle to $TESTTMP/aborted-editor/.hg/strip-backup/f899f3910ce7-7cab5e15-rebase.hg