tests/test-filecache.py
author Gregory Szorc <gregory.szorc@gmail.com>
Mon, 22 Aug 2016 21:48:50 -0700
changeset 29830 92ac2baaea86
parent 28803 76c091f9711e
child 29995 57830bd0e787
permissions -rw-r--r--
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.

from __future__ import absolute_import, print_function
import os
import subprocess
import sys

if subprocess.call(['python', '%s/hghave' % os.environ['TESTDIR'],
                    'cacheable']):
    sys.exit(80)

from mercurial import (
    extensions,
    hg,
    scmutil,
    ui as uimod,
    util,
)

filecache = scmutil.filecache

class fakerepo(object):
    def __init__(self):
        self._filecache = {}

    def join(self, p):
        return p

    def sjoin(self, p):
        return p

    @filecache('x', 'y')
    def cached(self):
        print('creating')
        return 'string from function'

    def invalidate(self):
        for k in self._filecache:
            try:
                delattr(self, k)
            except AttributeError:
                pass

def basic(repo):
    print("* neither file exists")
    # calls function
    repo.cached

    repo.invalidate()
    print("* neither file still exists")
    # uses cache
    repo.cached

    # create empty file
    f = open('x', 'w')
    f.close()
    repo.invalidate()
    print("* empty file x created")
    # should recreate the object
    repo.cached

    f = open('x', 'w')
    f.write('a')
    f.close()
    repo.invalidate()
    print("* file x changed size")
    # should recreate the object
    repo.cached

    repo.invalidate()
    print("* nothing changed with either file")
    # stats file again, reuses object
    repo.cached

    # atomic replace file, size doesn't change
    # hopefully st_mtime doesn't change as well so this doesn't use the cache
    # because of inode change
    f = scmutil.opener('.')('x', 'w', atomictemp=True)
    f.write('b')
    f.close()

    repo.invalidate()
    print("* file x changed inode")
    repo.cached

    # create empty file y
    f = open('y', 'w')
    f.close()
    repo.invalidate()
    print("* empty file y created")
    # should recreate the object
    repo.cached

    f = open('y', 'w')
    f.write('A')
    f.close()
    repo.invalidate()
    print("* file y changed size")
    # should recreate the object
    repo.cached

    f = scmutil.opener('.')('y', 'w', atomictemp=True)
    f.write('B')
    f.close()

    repo.invalidate()
    print("* file y changed inode")
    repo.cached

    f = scmutil.opener('.')('x', 'w', atomictemp=True)
    f.write('c')
    f.close()
    f = scmutil.opener('.')('y', 'w', atomictemp=True)
    f.write('C')
    f.close()

    repo.invalidate()
    print("* both files changed inode")
    repo.cached

def fakeuncacheable():
    def wrapcacheable(orig, *args, **kwargs):
        return False

    def wrapinit(orig, *args, **kwargs):
        pass

    originit = extensions.wrapfunction(util.cachestat, '__init__', wrapinit)
    origcacheable = extensions.wrapfunction(util.cachestat, 'cacheable',
                                            wrapcacheable)

    for fn in ['x', 'y']:
        try:
            os.remove(fn)
        except OSError:
            pass

    basic(fakerepo())

    util.cachestat.cacheable = origcacheable
    util.cachestat.__init__ = originit

def test_filecache_synced():
    # test old behavior that caused filecached properties to go out of sync
    os.system('hg init && echo a >> a && hg ci -qAm.')
    repo = hg.repository(uimod.ui())
    # first rollback clears the filecache, but changelog to stays in __dict__
    repo.rollback()
    repo.commit('.')
    # second rollback comes along and touches the changelog externally
    # (file is moved)
    repo.rollback()
    # but since changelog isn't under the filecache control anymore, we don't
    # see that it changed, and return the old changelog without reconstructing
    # it
    repo.commit('.')

def setbeforeget(repo):
    os.remove('x')
    os.remove('y')
    repo.cached = 'string set externally'
    repo.invalidate()
    print("* neither file exists")
    print(repo.cached)
    repo.invalidate()
    f = open('x', 'w')
    f.write('a')
    f.close()
    print("* file x created")
    print(repo.cached)

    repo.cached = 'string 2 set externally'
    repo.invalidate()
    print("* string set externally again")
    print(repo.cached)

    repo.invalidate()
    f = open('y', 'w')
    f.write('b')
    f.close()
    print("* file y created")
    print(repo.cached)

print('basic:')
print()
basic(fakerepo())
print()
print('fakeuncacheable:')
print()
fakeuncacheable()
test_filecache_synced()
print()
print('setbeforeget:')
print()
setbeforeget(fakerepo())