Mercurial > hg
view tests/test-mac-packages.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
author | Gregory Szorc <gregory.szorc@gmail.com> |
---|---|
date | Wed, 18 Jul 2018 09:49:34 -0700 |
parents | 283a7da602ae |
children | 8d7eaff92f9c |
line wrap: on
line source
#require test-repo slow osx osxpackaging $ . "$TESTDIR/helpers-testrepo.sh" $ testrepohgenv $ OUTPUTDIR="`pwd`" $ export OUTPUTDIR $ KEEPMPKG=yes $ export KEEPMPKG $ cd "$TESTDIR"/.. $ contrib/genosxversion.py --selftest ignoredarg $ make osx > "$OUTPUTDIR/build.log" 2>&1 $ cd "$OUTPUTDIR" $ ls -d *.pkg Mercurial-*-macosx10.*.pkg (glob) $ xar -xf Mercurial*.pkg Gather list of all installed files: $ lsbom mercurial.pkg/Bom > boms.txt We've had problems with the filter logic in the past. Make sure no .DS_Store files ended up in the final package: $ grep DS_S boms.txt [1] Spot-check some randomly selected files: $ grep bdiff boms.txt | cut -d ' ' -f 1,2,3 ./Library/Python/2.7/site-packages/mercurial/cext/bdiff.so 100755 0/0 ./Library/Python/2.7/site-packages/mercurial/cffi/bdiff.py 100644 0/0 ./Library/Python/2.7/site-packages/mercurial/cffi/bdiff.pyc 100644 0/0 ./Library/Python/2.7/site-packages/mercurial/cffi/bdiff.pyo 100644 0/0 ./Library/Python/2.7/site-packages/mercurial/cffi/bdiffbuild.py 100644 0/0 ./Library/Python/2.7/site-packages/mercurial/cffi/bdiffbuild.pyc 100644 0/0 ./Library/Python/2.7/site-packages/mercurial/cffi/bdiffbuild.pyo 100644 0/0 ./Library/Python/2.7/site-packages/mercurial/pure/bdiff.py 100644 0/0 ./Library/Python/2.7/site-packages/mercurial/pure/bdiff.pyc 100644 0/0 ./Library/Python/2.7/site-packages/mercurial/pure/bdiff.pyo 100644 0/0 $ grep zsh/site-functions/_hg boms.txt | cut -d ' ' -f 1,2,3 ./usr/local/share/zsh/site-functions/_hg 100644 0/0 $ grep hg-completion.bash boms.txt | cut -d ' ' -f 1,2,3 ./usr/local/hg/contrib/hg-completion.bash 100644 0/0 $ egrep 'man[15]' boms.txt | cut -d ' ' -f 1,2,3 ./usr/local/share/man/man1 40755 0/0 ./usr/local/share/man/man1/chg.1 100644 0/0 ./usr/local/share/man/man1/hg.1 100644 0/0 ./usr/local/share/man/man5 40755 0/0 ./usr/local/share/man/man5/hgignore.5 100644 0/0 ./usr/local/share/man/man5/hgrc.5 100644 0/0 $ grep bser boms.txt | cut -d ' ' -f 1,2,3 ./Library/Python/2.7/site-packages/hgext/fsmonitor/pywatchman/bser.so 100755 0/0 ./Library/Python/2.7/site-packages/hgext/fsmonitor/pywatchman/pybser.py 100644 0/0 ./Library/Python/2.7/site-packages/hgext/fsmonitor/pywatchman/pybser.pyc 100644 0/0 ./Library/Python/2.7/site-packages/hgext/fsmonitor/pywatchman/pybser.pyo 100644 0/0 $ grep localrepo boms.txt | cut -d ' ' -f 1,2,3 ./Library/Python/2.7/site-packages/mercurial/localrepo.py 100644 0/0 ./Library/Python/2.7/site-packages/mercurial/localrepo.pyc 100644 0/0 ./Library/Python/2.7/site-packages/mercurial/localrepo.pyo 100644 0/0 $ egrep 'bin/' boms.txt | cut -d ' ' -f 1,2,3 ./usr/local/bin/chg 100755 0/0 ./usr/local/bin/hg 100755 0/0 Make sure the built binary uses the system Python interpreter $ bsdtar xf mercurial.pkg/Payload usr/local/bin Use a glob to find this to avoid check-code whining about a fixed path. $ head -n 1 usr/local/b?n/hg #!/System/Library/Frameworks/Python.framework/Versions/2.7/Resources/Python.app/Contents/MacOS/Python Note that we're not currently installing any /etc/mercurial stuff, including merge-tool configurations.