##// END OF EJS Templates
procutil: avoid using os.fork() to implement runbgcommand...
procutil: avoid using os.fork() to implement runbgcommand We ran into the following deadlock: - some command creates an ssh peer, then raises without explicitly closing the peer (hg id + extension in our case) - dispatch catches the exception, calls ui.log('commandfinish', ..) (the sshpeer is still not closed), which calls logtoprocess, which calls procutil.runbgcommand. - in the child of runbgcommand's fork(), between the fork and the exec, the opening of file descriptors triggers a gc which runs the destructor for sshpeer, which waits on ssh's stderr being closed, which never happens since ssh's stderr is held open by the parent of the fork where said destructor hasn't run Remotefilelog appears to have a hack around this deadlock as well. I don't know if there's more subtlety to it, because even though the problem is determistic, it is very fragile, so I didn't manage to reduce it. I can imagine three ways of tackling this problem: 1. don't run any python between fork and exec in runbgcommand 2. make the finalizer harmless after the fork 3. close the peer without relying on gc behavior This commit goes with 1, as forking without exec'ing is tricky in general in a language with gc finalizers. And maybe it's better in the presence of rust threads. A future commit will try 2 or 3. Performance wise: at low memory usage, it's an improvement. At higher memory usage, it's about 2x faster than before when ensurestart=True, but 2x slower when ensurestart=False. Not sure if that matters. The reason for that last bit is that the subprocess.Popen always waits for the execve to finish, and at high memory usage, execve is slow because it deallocates the large page table. Numbers and script: before after mem=1.0GB, ensurestart=True 52.1ms 26.0ms mem=1.0GB, ensurestart=False 14.7ms 26.0ms mem=0.5GB, ensurestart=True 23.2ms 11.2ms mem=0.5GB, ensurestart=False 6.2ms 11.3ms mem=0.2GB, ensurestart=True 15.7ms 7.4ms mem=0.2GB, ensurestart=False 4.3ms 8.1ms mem=0.0GB, ensurestart=True 2.3ms 0.7ms mem=0.0GB, ensurestart=False 0.8ms 0.8ms import time for memsize in [1_000_000_000, 500_000_000, 250_000_000, 0]: mem = 'a' * memsize for ensurestart in [True, False]: now = time.time() n = 100 for i in range(n): procutil.runbgcommand([b'true'], {}, ensurestart=ensurestart) after = time.time() ms = (after - now) / float(n) * 1000 print(f'mem={memsize / 1e9:.1f}GB, ensurestart={ensurestart} -> {ms:.1f}ms') Differential Revision: https://phab.mercurial-scm.org/D9019

File last commit:

r47538:6266d195 default
r47651:8759e22f default
Show More
unionrepo.py
291 lines | 9.0 KiB | text/x-python | PythonLexer
# unionrepo.py - repository class for viewing union of repository changesets
#
# Derived from bundlerepo.py
# Copyright 2006, 2007 Benoit Boissinot <bboissin@gmail.com>
# Copyright 2013 Unity Technologies, Mads Kiilerich <madski@unity3d.com>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
"""Repository class for "in-memory pull" of one local repository to another,
allowing operations like diff and log with revsets.
"""
from __future__ import absolute_import
from .i18n import _
from .pycompat import getattr
from . import (
changelog,
cmdutil,
encoding,
error,
filelog,
localrepo,
manifest,
mdiff,
pathutil,
revlog,
util,
vfs as vfsmod,
)
class unionrevlog(revlog.revlog):
def __init__(self, opener, indexfile, revlog2, linkmapper):
# How it works:
# To retrieve a revision, we just need to know the node id so we can
# look it up in revlog2.
#
# To differentiate a rev in the second revlog from a rev in the revlog,
# we check revision against repotiprev.
opener = vfsmod.readonlyvfs(opener)
revlog.revlog.__init__(self, opener, indexfile)
self.revlog2 = revlog2
n = len(self)
self.repotiprev = n - 1
self.bundlerevs = set() # used by 'bundle()' revset expression
for rev2 in self.revlog2:
rev = self.revlog2.index[rev2]
# rev numbers - in revlog2, very different from self.rev
_start, _csize, rsize, base, linkrev, p1rev, p2rev, node = rev
flags = _start & 0xFFFF
if linkmapper is None: # link is to same revlog
assert linkrev == rev2 # we never link back
link = n
else: # rev must be mapped from repo2 cl to unified cl by linkmapper
link = linkmapper(linkrev)
if linkmapper is not None: # link is to same revlog
base = linkmapper(base)
this_rev = self.index.get_rev(node)
if this_rev is not None:
# this happens for the common revlog revisions
self.bundlerevs.add(this_rev)
continue
p1node = self.revlog2.node(p1rev)
p2node = self.revlog2.node(p2rev)
# TODO: it's probably wrong to set compressed length to -1, but
# I have no idea if csize is valid in the base revlog context.
e = (
flags,
-1,
rsize,
base,
link,
self.rev(p1node),
self.rev(p2node),
node,
)
self.index.append(e)
self.bundlerevs.add(n)
n += 1
def _chunk(self, rev):
if rev <= self.repotiprev:
return revlog.revlog._chunk(self, rev)
return self.revlog2._chunk(self.node(rev))
def revdiff(self, rev1, rev2):
"""return or calculate a delta between two revisions"""
if rev1 > self.repotiprev and rev2 > self.repotiprev:
return self.revlog2.revdiff(
self.revlog2.rev(self.node(rev1)),
self.revlog2.rev(self.node(rev2)),
)
elif rev1 <= self.repotiprev and rev2 <= self.repotiprev:
return super(unionrevlog, self).revdiff(rev1, rev2)
return mdiff.textdiff(self.rawdata(rev1), self.rawdata(rev2))
def _revisiondata(self, nodeorrev, _df=None, raw=False):
if isinstance(nodeorrev, int):
rev = nodeorrev
node = self.node(rev)
else:
node = nodeorrev
rev = self.rev(node)
if rev > self.repotiprev:
# work around manifestrevlog NOT being a revlog
revlog2 = getattr(self.revlog2, '_revlog', self.revlog2)
func = revlog2._revisiondata
else:
func = super(unionrevlog, self)._revisiondata
return func(node, _df=_df, raw=raw)
def addrevision(self, text, transaction, link, p1=None, p2=None, d=None):
raise NotImplementedError
def addgroup(
self,
deltas,
linkmapper,
transaction,
alwayscache=False,
addrevisioncb=None,
duplicaterevisioncb=None,
maybemissingparents=False,
):
raise NotImplementedError
def strip(self, minlink, transaction):
raise NotImplementedError
def checksize(self):
raise NotImplementedError
class unionchangelog(unionrevlog, changelog.changelog):
def __init__(self, opener, opener2):
changelog.changelog.__init__(self, opener)
linkmapper = None
changelog2 = changelog.changelog(opener2)
unionrevlog.__init__(
self, opener, self.indexfile, changelog2, linkmapper
)
class unionmanifest(unionrevlog, manifest.manifestrevlog):
def __init__(self, nodeconstants, opener, opener2, linkmapper):
manifest.manifestrevlog.__init__(self, nodeconstants, opener)
manifest2 = manifest.manifestrevlog(nodeconstants, opener2)
unionrevlog.__init__(
self, opener, self.indexfile, manifest2, linkmapper
)
class unionfilelog(filelog.filelog):
def __init__(self, opener, path, opener2, linkmapper, repo):
filelog.filelog.__init__(self, opener, path)
filelog2 = filelog.filelog(opener2, path)
self._revlog = unionrevlog(
opener, self.indexfile, filelog2._revlog, linkmapper
)
self._repo = repo
self.repotiprev = self._revlog.repotiprev
self.revlog2 = self._revlog.revlog2
def iscensored(self, rev):
"""Check if a revision is censored."""
if rev <= self.repotiprev:
return filelog.filelog.iscensored(self, rev)
node = self.node(rev)
return self.revlog2.iscensored(self.revlog2.rev(node))
class unionpeer(localrepo.localpeer):
def canpush(self):
return False
class unionrepository(object):
"""Represents the union of data in 2 repositories.
Instances are not usable if constructed directly. Use ``instance()``
or ``makeunionrepository()`` to create a usable instance.
"""
def __init__(self, repo2, url):
self.repo2 = repo2
self._url = url
self.ui.setconfig(b'phases', b'publish', False, b'unionrepo')
@localrepo.unfilteredpropertycache
def changelog(self):
return unionchangelog(self.svfs, self.repo2.svfs)
@localrepo.unfilteredpropertycache
def manifestlog(self):
rootstore = unionmanifest(
self.nodeconstants,
self.svfs,
self.repo2.svfs,
self.unfiltered()._clrev,
)
return manifest.manifestlog(
self.svfs, self, rootstore, self.narrowmatch()
)
def _clrev(self, rev2):
"""map from repo2 changelog rev to temporary rev in self.changelog"""
node = self.repo2.changelog.node(rev2)
return self.changelog.rev(node)
def url(self):
return self._url
def file(self, f):
return unionfilelog(
self.svfs, f, self.repo2.svfs, self.unfiltered()._clrev, self
)
def close(self):
self.repo2.close()
def cancopy(self):
return False
def peer(self):
return unionpeer(self)
def getcwd(self):
return encoding.getcwd() # always outside the repo
def instance(ui, path, create, intents=None, createopts=None):
if create:
raise error.Abort(_(b'cannot create new union repository'))
parentpath = ui.config(b"bundle", b"mainreporoot")
if not parentpath:
# try to find the correct path to the working directory repo
parentpath = cmdutil.findrepo(encoding.getcwd())
if parentpath is None:
parentpath = b''
if parentpath:
# Try to make the full path relative so we get a nice, short URL.
# In particular, we don't want temp dir names in test outputs.
cwd = encoding.getcwd()
if parentpath == cwd:
parentpath = b''
else:
cwd = pathutil.normasprefix(cwd)
if parentpath.startswith(cwd):
parentpath = parentpath[len(cwd) :]
if path.startswith(b'union:'):
s = path.split(b":", 1)[1].split(b"+", 1)
if len(s) == 1:
repopath, repopath2 = parentpath, s[0]
else:
repopath, repopath2 = s
else:
repopath, repopath2 = parentpath, path
return makeunionrepository(ui, repopath, repopath2)
def makeunionrepository(ui, repopath1, repopath2):
"""Make a union repository object from 2 local repo paths."""
repo1 = localrepo.instance(ui, repopath1, create=False)
repo2 = localrepo.instance(ui, repopath2, create=False)
url = b'union:%s+%s' % (
util.expandpath(repopath1),
util.expandpath(repopath2),
)
class derivedunionrepository(unionrepository, repo1.__class__):
pass
repo = repo1
repo.__class__ = derivedunionrepository
unionrepository.__init__(repo1, repo2, url)
return repo