view tests/test-repair-strip.t @ 38732:be4984261611

merge: mark file gets as not thread safe (issue5933)

In default installs, this has the effect of disabling the thread-based
worker on Windows when manifesting files in the working directory. My
measurements have shown that with revlog-based repositories, Mercurial
spends a lot of CPU time in revlog code resolving file data. This ends
up incurring a lot of context switching across threads and slows down
`hg update` operations when going from an empty working directory to
the tip of the repo.

On mozilla-unified (246,351 files) on an i7-6700K (4+4 CPUs):

before: 487s wall
after:  360s wall (equivalent to worker.enabled=false)
cpus=2: 379s wall

Even with only 2 threads, the thread pool is still slower.

The introduction of the thread-based worker (02b36e860e0b) states that
it resulted in a "~50%" speedup for `hg sparse --enable-profile` and
`hg sparse --disable-profile`. This disagrees with my measurement
above. I theorize a few reasons for this:

1) Removal of files from the working directory is I/O - not CPU - bound
   and should benefit from a thread pool (unless I/O is insanely fast
   and the GIL release is near instantaneous). So tests like `hg sparse
   --enable-profile` may exercise deletion throughput and aren't good
   benchmarks for worker tasks that are CPU heavy.
2) The patch was authored by someone at Facebook. The results were
   likely measured against a repository using remotefilelog. And I
   believe that revision retrieval during working directory updates with
   remotefilelog will often use a remote store, thus being I/O and not
   CPU bound. This probably resulted in an overstated performance gain.

Since there appears to be a need to enable the thread-based worker with
some stores, I've made the flagging of file gets as thread safe
configurable. I've made it experimental because I don't want to formalize
a boolean flag for this option and because this attribute is best
captured against the store implementation. But we don't have a proper
store API for this yet. I'd rather cross this bridge later.

It is possible there are revlog-based repositories that do benefit from
a thread-based worker. I didn't do very comprehensive testing. If there
are, we may want to devise a more proper algorithm for whether to use
the thread-based worker, including possibly config options to limit the
number of threads to use. But until I see evidence that justifies
complexity, simplicity wins.

Differential Revision: https://phab.mercurial-scm.org/D3963

#require unix-permissions no-root reporevlogstore

  $ cat > $TESTTMP/dumpjournal.py <<EOF
  > import sys
  > for entry in sys.stdin.read().split('\n'):
  >     if entry:
  >         print(entry.split('\x00')[0])
  > EOF

  $ echo "[extensions]" >> $HGRCPATH
  $ echo "mq=">> $HGRCPATH

  $ teststrip() {
  >   hg -q up -C $1
  >   echo % before update $1, strip $2
  >   hg parents
  >   chmod -$3 $4
  >   hg strip $2 2>&1 | sed 's/\(bundle\).*/\1/' | sed 's/Permission denied.*\.hg\/store\/\(.*\)/Permission denied \.hg\/store\/\1/'
  >   echo % after update $1, strip $2
  >   chmod +$3 $4
  >   hg verify
  >   echo % journal contents
  >   if [ -f .hg/store/journal ]; then
  >       cat .hg/store/journal | $PYTHON $TESTTMP/dumpjournal.py
  >   else
  >       echo "(no journal)"
  >   fi
  >   ls .hg/store/journal >/dev/null 2>&1 && hg recover
  >   ls .hg/strip-backup/* >/dev/null 2>&1 && hg unbundle -q .hg/strip-backup/*
  >   rm -rf .hg/strip-backup
  > }

  $ hg init test
  $ cd test
  $ echo a > a
  $ hg -q ci -m "a" -A
  $ echo b > b
  $ hg -q ci -m "b" -A
  $ echo b2 >> b
  $ hg -q ci -m "b2" -A
  $ echo c > c
  $ hg -q ci -m "c" -A
  $ teststrip 0 2 w .hg/store/data/b.i
  % before update 0, strip 2
  changeset:   0:cb9a9f314b8b
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     a
  
  saved backup bundle
  transaction abort!
  failed to truncate data/b.i
  rollback failed - please run hg recover
  strip failed, backup bundle
  abort: Permission denied .hg/store/data/b.i
  % after update 0, strip 2
  abandoned transaction found - run hg recover
  checking changesets
  checking manifests
  crosschecking files in changesets and manifests
  checking files
   b@?: rev 1 points to nonexistent changeset 2
   (expected 1)
   b@?: 736c29771fba not in manifests
  warning: orphan data file 'data/c.i'
  2 files, 2 changesets, 3 total revisions
  2 warnings encountered!
  2 integrity errors encountered!
  % journal contents
  00changelog.i
  00manifest.i
  data/b.i
  data/c.i
  rolling back interrupted transaction
  checking changesets
  checking manifests
  crosschecking files in changesets and manifests
  checking files
  2 files, 2 changesets, 2 total revisions
  $ teststrip 0 2 r .hg/store/data/b.i
  % before update 0, strip 2
  changeset:   0:cb9a9f314b8b
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     a
  
  abort: Permission denied .hg/store/data/b.i
  % after update 0, strip 2
  checking changesets
  checking manifests
  crosschecking files in changesets and manifests
  checking files
  3 files, 4 changesets, 4 total revisions
  % journal contents
  (no journal)
  $ teststrip 0 2 w .hg/store/00manifest.i
  % before update 0, strip 2
  changeset:   0:cb9a9f314b8b
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     a
  
  saved backup bundle
  transaction abort!
  failed to truncate 00manifest.i
  rollback failed - please run hg recover
  strip failed, backup bundle
  abort: Permission denied .hg/store/00manifest.i
  % after update 0, strip 2
  abandoned transaction found - run hg recover
  checking changesets
  checking manifests
   manifest@?: rev 2 points to nonexistent changeset 2
   manifest@?: 3362547cdf64 not in changesets
   manifest@?: rev 3 points to nonexistent changeset 3
   manifest@?: 265a85892ecb not in changesets
  crosschecking files in changesets and manifests
   c@3: in manifest but not in changeset
  checking files
   b@?: rev 1 points to nonexistent changeset 2
   (expected 1)
   c@?: rev 0 points to nonexistent changeset 3
  3 files, 2 changesets, 4 total revisions
  1 warnings encountered!
  7 integrity errors encountered!
  (first damaged changeset appears to be 3)
  % journal contents
  00changelog.i
  00manifest.i
  data/b.i
  data/c.i
  rolling back interrupted transaction
  checking changesets
  checking manifests
  crosschecking files in changesets and manifests
  checking files
  2 files, 2 changesets, 2 total revisions

  $ cd ..