lfs: avoid closing connections when the worker doesn't fork

Probably not much more than an minor optimization, but could be useful in the
case of `hg verify` where missing blobs are fetched one at a time.

hgext/lfs/blobstore.py

@@ -1,782 +1,785 b''
 # blobstore.py - local and remote (speaking Git-LFS protocol) blob storages
 #
 # Copyright 2017 Facebook, Inc.
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 
 import contextlib
 import errno
 import hashlib
 import json
 import os
 import re
 import socket
 
 from mercurial.i18n import _
 from mercurial.pycompat import getattr
 from mercurial.node import hex
 
 from mercurial import (
     encoding,
     error,
     httpconnection as httpconnectionmod,
     pathutil,
     pycompat,
     url as urlmod,
     util,
     vfs as vfsmod,
     worker,
 )
 
 from mercurial.utils import (
     stringutil,
     urlutil,
 )
 
 from ..largefiles import lfutil
 
 # 64 bytes for SHA256
 _lfsre = re.compile(br'\A[a-f0-9]{64}\Z')
 
 
 class lfsvfs(vfsmod.vfs):
     def join(self, path):
         """split the path at first two characters, like: XX/XXXXX..."""
         if not _lfsre.match(path):
             raise error.ProgrammingError(b'unexpected lfs path: %s' % path)
         return super(lfsvfs, self).join(path[0:2], path[2:])
 
     def walk(self, path=None, onerror=None):
         """Yield (dirpath, [], oids) tuple for blobs under path
 
         Oids only exist in the root of this vfs, so dirpath is always ''.
         """
         root = os.path.normpath(self.base)
         # when dirpath == root, dirpath[prefixlen:] becomes empty
         # because len(dirpath) < prefixlen.
         prefixlen = len(pathutil.normasprefix(root))
         oids = []
 
         for dirpath, dirs, files in os.walk(
             self.reljoin(self.base, path or b''), onerror=onerror
         ):
             dirpath = dirpath[prefixlen:]
 
             # Silently skip unexpected files and directories
             if len(dirpath) == 2:
                 oids.extend(
                     [dirpath + f for f in files if _lfsre.match(dirpath + f)]
                 )
 
         yield (b'', [], oids)
 
 
 class nullvfs(lfsvfs):
     def __init__(self):
         pass
 
     def exists(self, oid):
         return False
 
     def read(self, oid):
         # store.read() calls into here if the blob doesn't exist in its
         # self.vfs.  Raise the same error as a normal vfs when asked to read a
         # file that doesn't exist.  The only difference is the full file path
         # isn't available in the error.
         raise IOError(
             errno.ENOENT,
             pycompat.sysstr(b'%s: No such file or directory' % oid),
         )
 
     def walk(self, path=None, onerror=None):
         return (b'', [], [])
 
     def write(self, oid, data):
         pass
 
 
 class lfsuploadfile(httpconnectionmod.httpsendfile):
     """a file-like object that supports keepalive."""
 
     def __init__(self, ui, filename):
         super(lfsuploadfile, self).__init__(ui, filename, b'rb')
         self.read = self._data.read
 
     def _makeprogress(self):
         return None  # progress is handled by the worker client
 
 
 class local:
     """Local blobstore for large file contents.
 
     This blobstore is used both as a cache and as a staging area for large blobs
     to be uploaded to the remote blobstore.
     """
 
     def __init__(self, repo):
         fullpath = repo.svfs.join(b'lfs/objects')
         self.vfs = lfsvfs(fullpath)
 
         if repo.ui.configbool(b'experimental', b'lfs.disableusercache'):
             self.cachevfs = nullvfs()
         else:
             usercache = lfutil._usercachedir(repo.ui, b'lfs')
             self.cachevfs = lfsvfs(usercache)
         self.ui = repo.ui
 
     def open(self, oid):
         """Open a read-only file descriptor to the named blob, in either the
         usercache or the local store."""
         return open(self.path(oid), 'rb')
 
     def path(self, oid):
         """Build the path for the given blob ``oid``.
 
         If the blob exists locally, the path may point to either the usercache
         or the local store.  If it doesn't, it will point to the local store.
         This is meant for situations where existing code that isn't LFS aware
         needs to open a blob.  Generally, prefer the ``open`` method on this
         class.
         """
         # The usercache is the most likely place to hold the file.  Commit will
         # write to both it and the local store, as will anything that downloads
         # the blobs.  However, things like clone without an update won't
         # populate the local store.  For an init + push of a local clone,
         # the usercache is the only place it _could_ be.  If not present, the
         # missing file msg here will indicate the local repo, not the usercache.
         if self.cachevfs.exists(oid):
             return self.cachevfs.join(oid)
 
         return self.vfs.join(oid)
 
     def download(self, oid, src, content_length):
         """Read the blob from the remote source in chunks, verify the content,
         and write to this local blobstore."""
         sha256 = hashlib.sha256()
         size = 0
 
         with self.vfs(oid, b'wb', atomictemp=True) as fp:
             for chunk in util.filechunkiter(src, size=1048576):
                 fp.write(chunk)
                 sha256.update(chunk)
                 size += len(chunk)
 
             # If the server advertised a length longer than what we actually
             # received, then we should expect that the server crashed while
             # producing the response (but the server has no way of telling us
             # that), and we really don't need to try to write the response to
             # the localstore, because it's not going to match the expected.
             if content_length is not None and int(content_length) != size:
                 msg = (
                     b"Response length (%d) does not match Content-Length "
                     b"header (%d): likely server-side crash"
                 )
                 raise LfsRemoteError(_(msg) % (size, int(content_length)))
 
             realoid = hex(sha256.digest())
             if realoid != oid:
                 raise LfsCorruptionError(
                     _(b'corrupt remote lfs object: %s') % oid
                 )
 
         self._linktousercache(oid)
 
     def write(self, oid, data):
         """Write blob to local blobstore.
 
         This should only be called from the filelog during a commit or similar.
         As such, there is no need to verify the data.  Imports from a remote
         store must use ``download()`` instead."""
         with self.vfs(oid, b'wb', atomictemp=True) as fp:
             fp.write(data)
 
         self._linktousercache(oid)
 
     def linkfromusercache(self, oid):
         """Link blobs found in the user cache into this store.
 
         The server module needs to do this when it lets the client know not to
         upload the blob, to ensure it is always available in this store.
         Normally this is done implicitly when the client reads or writes the
         blob, but that doesn't happen when the server tells the client that it
         already has the blob.
         """
         if not isinstance(self.cachevfs, nullvfs) and not self.vfs.exists(oid):
             self.ui.note(_(b'lfs: found %s in the usercache\n') % oid)
             lfutil.link(self.cachevfs.join(oid), self.vfs.join(oid))
 
     def _linktousercache(self, oid):
         # XXX: should we verify the content of the cache, and hardlink back to
         # the local store on success, but truncate, write and link on failure?
         if not self.cachevfs.exists(oid) and not isinstance(
             self.cachevfs, nullvfs
         ):
             self.ui.note(_(b'lfs: adding %s to the usercache\n') % oid)
             lfutil.link(self.vfs.join(oid), self.cachevfs.join(oid))
 
     def read(self, oid, verify=True):
         """Read blob from local blobstore."""
         if not self.vfs.exists(oid):
             blob = self._read(self.cachevfs, oid, verify)
 
             # Even if revlog will verify the content, it needs to be verified
             # now before making the hardlink to avoid propagating corrupt blobs.
             # Don't abort if corruption is detected, because `hg verify` will
             # give more useful info about the corruption- simply don't add the
             # hardlink.
             if verify or hex(hashlib.sha256(blob).digest()) == oid:
                 self.ui.note(_(b'lfs: found %s in the usercache\n') % oid)
                 lfutil.link(self.cachevfs.join(oid), self.vfs.join(oid))
         else:
             self.ui.note(_(b'lfs: found %s in the local lfs store\n') % oid)
             blob = self._read(self.vfs, oid, verify)
         return blob
 
     def _read(self, vfs, oid, verify):
         """Read blob (after verifying) from the given store"""
         blob = vfs.read(oid)
         if verify:
             _verify(oid, blob)
         return blob
 
     def verify(self, oid):
         """Indicate whether or not the hash of the underlying file matches its
         name."""
         sha256 = hashlib.sha256()
 
         with self.open(oid) as fp:
             for chunk in util.filechunkiter(fp, size=1048576):
                 sha256.update(chunk)
 
         return oid == hex(sha256.digest())
 
     def has(self, oid):
         """Returns True if the local blobstore contains the requested blob,
         False otherwise."""
         return self.cachevfs.exists(oid) or self.vfs.exists(oid)
 
 
 def _urlerrorreason(urlerror):
     """Create a friendly message for the given URLError to be used in an
     LfsRemoteError message.
     """
     inst = urlerror
 
     if isinstance(urlerror.reason, Exception):
         inst = urlerror.reason
 
     if util.safehasattr(inst, b'reason'):
         try:  # usually it is in the form (errno, strerror)
             reason = inst.reason.args[1]
         except (AttributeError, IndexError):
             # it might be anything, for example a string
             reason = inst.reason
         if isinstance(reason, str):
             # SSLError of Python 2.7.9 contains a unicode
             reason = encoding.unitolocal(reason)
         return reason
     elif getattr(inst, "strerror", None):
         return encoding.strtolocal(inst.strerror)
     else:
         return stringutil.forcebytestr(urlerror)
 
 
 class lfsauthhandler(util.urlreq.basehandler):
     handler_order = 480  # Before HTTPDigestAuthHandler (== 490)
 
     def http_error_401(self, req, fp, code, msg, headers):
         """Enforces that any authentication performed is HTTP Basic
         Authentication.  No authentication is also acceptable.
         """
         authreq = headers.get('www-authenticate', None)
         if authreq:
             scheme = authreq.split()[0]
 
             if scheme.lower() != 'basic':
                 msg = _(b'the server must support Basic Authentication')
                 raise util.urlerr.httperror(
                     req.get_full_url(),
                     code,
                     encoding.strfromlocal(msg),
                     headers,
                     fp,
                 )
         return None
 
 
 class _gitlfsremote:
     def __init__(self, repo, url):
         ui = repo.ui
         self.ui = ui
         baseurl, authinfo = url.authinfo()
         self.baseurl = baseurl.rstrip(b'/')
         useragent = repo.ui.config(b'experimental', b'lfs.user-agent')
         if not useragent:
             useragent = b'git-lfs/2.3.4 (Mercurial %s)' % util.version()
         self.urlopener = urlmod.opener(ui, authinfo, useragent)
         self.urlopener.add_handler(lfsauthhandler())
         self.retry = ui.configint(b'lfs', b'retry')
 
     def writebatch(self, pointers, fromstore):
         """Batch upload from local to remote blobstore."""
         self._batch(_deduplicate(pointers), fromstore, b'upload')
 
     def readbatch(self, pointers, tostore):
         """Batch download from remote to local blostore."""
         self._batch(_deduplicate(pointers), tostore, b'download')
 
     def _batchrequest(self, pointers, action):
         """Get metadata about objects pointed by pointers for given action
 
         Return decoded JSON object like {'objects': [{'oid': '', 'size': 1}]}
         See https://github.com/git-lfs/git-lfs/blob/master/docs/api/batch.md
         """
         objects = [
             {'oid': pycompat.strurl(p.oid()), 'size': p.size()}
             for p in pointers
         ]
         requestdata = pycompat.bytesurl(
             json.dumps(
                 {
                     'objects': objects,
                     'operation': pycompat.strurl(action),
                 }
             )
         )
         url = b'%s/objects/batch' % self.baseurl
         batchreq = util.urlreq.request(pycompat.strurl(url), data=requestdata)
         batchreq.add_header('Accept', 'application/vnd.git-lfs+json')
         batchreq.add_header('Content-Type', 'application/vnd.git-lfs+json')
         try:
             with contextlib.closing(self.urlopener.open(batchreq)) as rsp:
                 rawjson = rsp.read()
         except util.urlerr.httperror as ex:
             hints = {
                 400: _(
                     b'check that lfs serving is enabled on %s and "%s" is '
                     b'supported'
                 )
                 % (self.baseurl, action),
                 404: _(b'the "lfs.url" config may be used to override %s')
                 % self.baseurl,
             }
             hint = hints.get(ex.code, _(b'api=%s, action=%s') % (url, action))
             raise LfsRemoteError(
                 _(b'LFS HTTP error: %s') % stringutil.forcebytestr(ex),
                 hint=hint,
             )
         except util.urlerr.urlerror as ex:
             hint = (
                 _(b'the "lfs.url" config may be used to override %s')
                 % self.baseurl
             )
             raise LfsRemoteError(
                 _(b'LFS error: %s') % _urlerrorreason(ex), hint=hint
             )
         try:
             response = pycompat.json_loads(rawjson)
         except ValueError:
             raise LfsRemoteError(
                 _(b'LFS server returns invalid JSON: %s')
                 % rawjson.encode("utf-8")
             )
 
         if self.ui.debugflag:
             self.ui.debug(b'Status: %d\n' % rsp.status)
             # lfs-test-server and hg serve return headers in different order
             headers = pycompat.bytestr(rsp.info()).strip()
             self.ui.debug(b'%s\n' % b'\n'.join(sorted(headers.splitlines())))
 
             if 'objects' in response:
                 response['objects'] = sorted(
                     response['objects'], key=lambda p: p['oid']
                 )
             self.ui.debug(
                 b'%s\n'
                 % pycompat.bytesurl(
                     json.dumps(
                         response,
                         indent=2,
                         separators=('', ': '),
                         sort_keys=True,
                     )
                 )
             )
 
         def encodestr(x):
             if isinstance(x, str):
                 return x.encode('utf-8')
             return x
 
         return pycompat.rapply(encodestr, response)
 
     def _checkforservererror(self, pointers, responses, action):
         """Scans errors from objects
 
         Raises LfsRemoteError if any objects have an error"""
         for response in responses:
             # The server should return 404 when objects cannot be found. Some
             # server implementation (ex. lfs-test-server)  does not set "error"
             # but just removes "download" from "actions". Treat that case
             # as the same as 404 error.
             if b'error' not in response:
                 if action == b'download' and action not in response.get(
                     b'actions', []
                 ):
                     code = 404
                 else:
                     continue
             else:
                 # An error dict without a code doesn't make much sense, so
                 # treat as a server error.
                 code = response.get(b'error').get(b'code', 500)
 
             ptrmap = {p.oid(): p for p in pointers}
             p = ptrmap.get(response[b'oid'], None)
             if p:
                 filename = getattr(p, 'filename', b'unknown')
                 errors = {
                     404: b'The object does not exist',
                     410: b'The object was removed by the owner',
                     422: b'Validation error',
                     500: b'Internal server error',
                 }
                 msg = errors.get(code, b'status code %d' % code)
                 raise LfsRemoteError(
                     _(b'LFS server error for "%s": %s') % (filename, msg)
                 )
             else:
                 raise LfsRemoteError(
                     _(b'LFS server error. Unsolicited response for oid %s')
                     % response[b'oid']
                 )
 
     def _extractobjects(self, response, pointers, action):
         """extract objects from response of the batch API
 
         response: parsed JSON object returned by batch API
         return response['objects'] filtered by action
         raise if any object has an error
         """
         # Scan errors from objects - fail early
         objects = response.get(b'objects', [])
         self._checkforservererror(pointers, objects, action)
 
         # Filter objects with given action. Practically, this skips uploading
         # objects which exist in the server.
         filteredobjects = [
             o for o in objects if action in o.get(b'actions', [])
         ]
 
         return filteredobjects
 
     def _basictransfer(self, obj, action, localstore):
         """Download or upload a single object using basic transfer protocol
 
         obj: dict, an object description returned by batch API
         action: string, one of ['upload', 'download']
         localstore: blobstore.local
 
         See https://github.com/git-lfs/git-lfs/blob/master/docs/api/\
         basic-transfers.md
         """
         oid = obj[b'oid']
         href = obj[b'actions'][action].get(b'href')
         headers = obj[b'actions'][action].get(b'header', {}).items()
 
         request = util.urlreq.request(pycompat.strurl(href))
         if action == b'upload':
             # If uploading blobs, read data from local blobstore.
             if not localstore.verify(oid):
                 raise error.Abort(
                     _(b'detected corrupt lfs object: %s') % oid,
                     hint=_(b'run hg verify'),
                 )
 
         for k, v in headers:
             request.add_header(pycompat.strurl(k), pycompat.strurl(v))
 
         try:
             if action == b'upload':
                 request.data = lfsuploadfile(self.ui, localstore.path(oid))
                 request.get_method = lambda: 'PUT'
                 request.add_header('Content-Type', 'application/octet-stream')
                 request.add_header('Content-Length', request.data.length)
 
             with contextlib.closing(self.urlopener.open(request)) as res:
                 contentlength = res.info().get(b"content-length")
                 ui = self.ui  # Shorten debug lines
                 if self.ui.debugflag:
                     ui.debug(b'Status: %d\n' % res.status)
                     # lfs-test-server and hg serve return headers in different
                     # order
                     headers = pycompat.bytestr(res.info()).strip()
                     ui.debug(b'%s\n' % b'\n'.join(sorted(headers.splitlines())))
 
                 if action == b'download':
                     # If downloading blobs, store downloaded data to local
                     # blobstore
                     localstore.download(oid, res, contentlength)
                 else:
                     blocks = []
                     while True:
                         data = res.read(1048576)
                         if not data:
                             break
                         blocks.append(data)
 
                     response = b"".join(blocks)
                     if response:
                         ui.debug(b'lfs %s response: %s' % (action, response))
         except util.urlerr.httperror as ex:
             if self.ui.debugflag:
                 self.ui.debug(
                     b'%s: %s\n' % (oid, ex.read())
                 )  # XXX: also bytes?
             raise LfsRemoteError(
                 _(b'LFS HTTP error: %s (oid=%s, action=%s)')
                 % (stringutil.forcebytestr(ex), oid, action)
             )
         except util.urlerr.urlerror as ex:
             hint = _(b'attempted connection to %s') % pycompat.bytesurl(
                 util.urllibcompat.getfullurl(request)
             )
             raise LfsRemoteError(
                 _(b'LFS error: %s') % _urlerrorreason(ex), hint=hint
             )
         finally:
             if request.data:
                 request.data.close()
 
     def _batch(self, pointers, localstore, action):
         if action not in [b'upload', b'download']:
             raise error.ProgrammingError(b'invalid Git-LFS action: %s' % action)
 
         response = self._batchrequest(pointers, action)
         objects = self._extractobjects(response, pointers, action)
         total = sum(x.get(b'size', 0) for x in objects)
         sizes = {}
         for obj in objects:
             sizes[obj.get(b'oid')] = obj.get(b'size', 0)
         topic = {
             b'upload': _(b'lfs uploading'),
             b'download': _(b'lfs downloading'),
         }[action]
         if len(objects) > 1:
             self.ui.note(
                 _(b'lfs: need to transfer %d objects (%s)\n')
                 % (len(objects), util.bytecount(total))
             )
 
         def transfer(chunk):
             for obj in chunk:
                 objsize = obj.get(b'size', 0)
                 if self.ui.verbose:
                     if action == b'download':
                         msg = _(b'lfs: downloading %s (%s)\n')
                     elif action == b'upload':
                         msg = _(b'lfs: uploading %s (%s)\n')
                     self.ui.note(
                         msg % (obj.get(b'oid'), util.bytecount(objsize))
                     )
                 retry = self.retry
                 while True:
                     try:
                         self._basictransfer(obj, action, localstore)
                         yield 1, obj.get(b'oid')
                         break
                     except socket.error as ex:
                         if retry > 0:
                             self.ui.note(
                                 _(b'lfs: failed: %r (remaining retry %d)\n')
                                 % (stringutil.forcebytestr(ex), retry)
                             )
                             retry -= 1
                             continue
                         raise
 
-        # Until https multiplexing gets sorted out
+        # Until https multiplexing gets sorted out.  It's not clear if
+        # ConnectionManager.set_ready() is externally synchronized for thread
+        # safety with Windows workers.
         if self.ui.configbool(b'experimental', b'lfs.worker-enable'):
             # The POSIX workers are forks of this process, so before spinning
             # them up, close all pooled connections.  Otherwise, there's no way
             # to coordinate between them about who is using what, and the
             # transfers will get corrupted.
             #
             # TODO: add a function to keepalive.ConnectionManager to mark all
             #  ready connections as in use, and roll that back after the fork?
             #  That would allow the existing pool of connections in this process
             #  to be preserved.
-            if not pycompat.iswindows:
+            def prefork():
                 for h in self.urlopener.handlers:
                     getattr(h, "close_all", lambda: None)()
 
             oids = worker.worker(
                 self.ui,
                 0.1,
                 transfer,
                 (),
                 sorted(objects, key=lambda o: o.get(b'oid')),
+                prefork=prefork,
             )
         else:
             oids = transfer(sorted(objects, key=lambda o: o.get(b'oid')))
 
         with self.ui.makeprogress(
             topic, unit=_(b"bytes"), total=total
         ) as progress:
             progress.update(0)
             processed = 0
             blobs = 0
             for _one, oid in oids:
                 processed += sizes[oid]
                 blobs += 1
                 progress.update(processed)
                 self.ui.note(_(b'lfs: processed: %s\n') % oid)
 
         if blobs > 0:
             if action == b'upload':
                 self.ui.status(
                     _(b'lfs: uploaded %d files (%s)\n')
                     % (blobs, util.bytecount(processed))
                 )
             elif action == b'download':
                 self.ui.status(
                     _(b'lfs: downloaded %d files (%s)\n')
                     % (blobs, util.bytecount(processed))
                 )
 
     def __del__(self):
         # copied from mercurial/httppeer.py
         urlopener = getattr(self, 'urlopener', None)
         if urlopener:
             for h in urlopener.handlers:
                 h.close()
                 getattr(h, "close_all", lambda: None)()
 
 
 class _dummyremote:
     """Dummy store storing blobs to temp directory."""
 
     def __init__(self, repo, url):
         fullpath = repo.vfs.join(b'lfs', url.path)
         self.vfs = lfsvfs(fullpath)
 
     def writebatch(self, pointers, fromstore):
         for p in _deduplicate(pointers):
             content = fromstore.read(p.oid(), verify=True)
             with self.vfs(p.oid(), b'wb', atomictemp=True) as fp:
                 fp.write(content)
 
     def readbatch(self, pointers, tostore):
         for p in _deduplicate(pointers):
             with self.vfs(p.oid(), b'rb') as fp:
                 tostore.download(p.oid(), fp, None)
 
 
 class _nullremote:
     """Null store storing blobs to /dev/null."""
 
     def __init__(self, repo, url):
         pass
 
     def writebatch(self, pointers, fromstore):
         pass
 
     def readbatch(self, pointers, tostore):
         pass
 
 
 class _promptremote:
     """Prompt user to set lfs.url when accessed."""
 
     def __init__(self, repo, url):
         pass
 
     def writebatch(self, pointers, fromstore, ui=None):
         self._prompt()
 
     def readbatch(self, pointers, tostore, ui=None):
         self._prompt()
 
     def _prompt(self):
         raise error.Abort(_(b'lfs.url needs to be configured'))
 
 
 _storemap = {
     b'https': _gitlfsremote,
     b'http': _gitlfsremote,
     b'file': _dummyremote,
     b'null': _nullremote,
     None: _promptremote,
 }
 
 
 def _deduplicate(pointers):
     """Remove any duplicate oids that exist in the list"""
     reduced = util.sortdict()
     for p in pointers:
         reduced[p.oid()] = p
     return reduced.values()
 
 
 def _verify(oid, content):
     realoid = hex(hashlib.sha256(content).digest())
     if realoid != oid:
         raise LfsCorruptionError(
             _(b'detected corrupt lfs object: %s') % oid,
             hint=_(b'run hg verify'),
         )
 
 
 def remote(repo, remote=None):
     """remotestore factory. return a store in _storemap depending on config
 
     If ``lfs.url`` is specified, use that remote endpoint.  Otherwise, try to
     infer the endpoint, based on the remote repository using the same path
     adjustments as git.  As an extension, 'http' is supported as well so that
     ``hg serve`` works out of the box.
 
     https://github.com/git-lfs/git-lfs/blob/master/docs/api/server-discovery.md
     """
     lfsurl = repo.ui.config(b'lfs', b'url')
     url = urlutil.url(lfsurl or b'')
     if lfsurl is None:
         if remote:
             path = remote
         elif util.safehasattr(repo, b'_subtoppath'):
             # The pull command sets this during the optional update phase, which
             # tells exactly where the pull originated, whether 'paths.default'
             # or explicit.
             path = repo._subtoppath
         else:
             # TODO: investigate 'paths.remote:lfsurl' style path customization,
             # and fall back to inferring from 'paths.remote' if unspecified.
             path = repo.ui.config(b'paths', b'default') or b''
 
         defaulturl = urlutil.url(path)
 
         # TODO: support local paths as well.
         # TODO: consider the ssh -> https transformation that git applies
         if defaulturl.scheme in (b'http', b'https'):
             if defaulturl.path and defaulturl.path[:-1] != b'/':
                 defaulturl.path += b'/'
             defaulturl.path = (defaulturl.path or b'') + b'.git/info/lfs'
 
             url = urlutil.url(bytes(defaulturl))
             repo.ui.note(_(b'lfs: assuming remote store: %s\n') % url)
 
     scheme = url.scheme
     if scheme not in _storemap:
         raise error.Abort(_(b'lfs: unknown url scheme: %s') % scheme)
     return _storemap[scheme](repo, url)
 
 
 class LfsRemoteError(error.StorageError):
     pass
 
 
 class LfsCorruptionError(error.Abort):
     """Raised when a corrupt blob is detected, aborting an operation
 
     It exists to allow specialized handling on the server side."""

mercurial/worker.py

@@ -1,455 +1,472 b''
 # worker.py - master-slave parallelism support
 #
 # Copyright 2013 Facebook, Inc.
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 
 import os
 import pickle
 import selectors
 import signal
 import sys
 import threading
 import time
 
 from .i18n import _
 from . import (
     encoding,
     error,
     pycompat,
     scmutil,
 )
 
 
 def countcpus():
     '''try to count the number of CPUs on the system'''
 
     # posix
     try:
         n = int(os.sysconf('SC_NPROCESSORS_ONLN'))
         if n > 0:
             return n
     except (AttributeError, ValueError):
         pass
 
     # windows
     try:
         n = int(encoding.environ[b'NUMBER_OF_PROCESSORS'])
         if n > 0:
             return n
     except (KeyError, ValueError):
         pass
 
     return 1
 
 
 def _numworkers(ui):
     s = ui.config(b'worker', b'numcpus')
     if s:
         try:
             n = int(s)
             if n >= 1:
                 return n
         except ValueError:
             raise error.Abort(_(b'number of cpus must be an integer'))
     return min(max(countcpus(), 4), 32)
 
 
 def ismainthread():
     return threading.current_thread() == threading.main_thread()
 
 
 class _blockingreader:
     """Wrap unbuffered stream such that pickle.load() works with it.
 
     pickle.load() expects that calls to read() and readinto() read as many
     bytes as requested. On EOF, it is fine to read fewer bytes. In this case,
     pickle.load() raises an EOFError.
     """
 
     def __init__(self, wrapped):
         self._wrapped = wrapped
 
     def readline(self):
         return self._wrapped.readline()
 
     def readinto(self, buf):
         pos = 0
         size = len(buf)
 
         with memoryview(buf) as view:
             while pos < size:
                 with view[pos:] as subview:
                     ret = self._wrapped.readinto(subview)
                 if not ret:
                     break
                 pos += ret
 
         return pos
 
     # issue multiple reads until size is fulfilled (or EOF is encountered)
     def read(self, size=-1):
         if size < 0:
             return self._wrapped.readall()
 
         buf = bytearray(size)
         n_read = self.readinto(buf)
         del buf[n_read:]
         return bytes(buf)
 
 
 if pycompat.isposix or pycompat.iswindows:
     _STARTUP_COST = 0.01
     # The Windows worker is thread based. If tasks are CPU bound, threads
     # in the presence of the GIL result in excessive context switching and
     # this overhead can slow down execution.
     _DISALLOW_THREAD_UNSAFE = pycompat.iswindows
 else:
     _STARTUP_COST = 1e30
     _DISALLOW_THREAD_UNSAFE = False
 
 
 def worthwhile(ui, costperop, nops, threadsafe=True):
     """try to determine whether the benefit of multiple processes can
     outweigh the cost of starting them"""
 
     if not threadsafe and _DISALLOW_THREAD_UNSAFE:
         return False
 
     linear = costperop * nops
     workers = _numworkers(ui)
     benefit = linear - (_STARTUP_COST * workers + linear / workers)
     return benefit >= 0.15
 
 
 def worker(
-    ui, costperarg, func, staticargs, args, hasretval=False, threadsafe=True
+    ui,
+    costperarg,
+    func,
+    staticargs,
+    args,
+    hasretval=False,
+    threadsafe=True,
+    prefork=None,
 ):
     """run a function, possibly in parallel in multiple worker
     processes.
 
     returns a progress iterator
 
     costperarg - cost of a single task
 
     func - function to run. It is expected to return a progress iterator.
 
     staticargs - arguments to pass to every invocation of the function
 
     args - arguments to split into chunks, to pass to individual
     workers
 
     hasretval - when True, func and the current function return an progress
     iterator then a dict (encoded as an iterator that yield many (False, ..)
     then a (True, dict)). The dicts are joined in some arbitrary order, so
     overlapping keys are a bad idea.
 
     threadsafe - whether work items are thread safe and can be executed using
     a thread-based worker. Should be disabled for CPU heavy tasks that don't
     release the GIL.
+
+    prefork - a parameterless Callable that is invoked prior to forking the
+    process.  fork() is only used on non-Windows platforms, but is also not
+    called on POSIX platforms if the work amount doesn't warrant a worker.
     """
     enabled = ui.configbool(b'worker', b'enabled')
     if enabled and _platformworker is _posixworker and not ismainthread():
         # The POSIX worker has to install a handler for SIGCHLD.
         # Python up to 3.9 only allows this in the main thread.
         enabled = False
 
     if enabled and worthwhile(ui, costperarg, len(args), threadsafe=threadsafe):
-        return _platformworker(ui, func, staticargs, args, hasretval)
+        return _platformworker(
+            ui, func, staticargs, args, hasretval, prefork=prefork
+        )
     return func(*staticargs + (args,))
 
 
-def _posixworker(ui, func, staticargs, args, hasretval):
+def _posixworker(ui, func, staticargs, args, hasretval, prefork=None):
     workers = _numworkers(ui)
     oldhandler = signal.getsignal(signal.SIGINT)
     signal.signal(signal.SIGINT, signal.SIG_IGN)
     pids, problem = set(), [0]
 
     def killworkers():
         # unregister SIGCHLD handler as all children will be killed. This
         # function shouldn't be interrupted by another SIGCHLD; otherwise pids
         # could be updated while iterating, which would cause inconsistency.
         signal.signal(signal.SIGCHLD, oldchldhandler)
         # if one worker bails, there's no good reason to wait for the rest
         for p in pids:
             try:
                 os.kill(p, signal.SIGTERM)
             except ProcessLookupError:
                 pass
 
     def waitforworkers(blocking=True):
         for pid in pids.copy():
             p = st = 0
             try:
                 p, st = os.waitpid(pid, (0 if blocking else os.WNOHANG))
             except ChildProcessError:
                 # child would already be reaped, but pids yet been
                 # updated (maybe interrupted just after waitpid)
                 pids.discard(pid)
             if not p:
                 # skip subsequent steps, because child process should
                 # be still running in this case
                 continue
             pids.discard(p)
             st = _exitstatus(st)
             if st and not problem[0]:
                 problem[0] = st
 
     def sigchldhandler(signum, frame):
         waitforworkers(blocking=False)
         if problem[0]:
             killworkers()
 
     oldchldhandler = signal.signal(signal.SIGCHLD, sigchldhandler)
     ui.flush()
     parentpid = os.getpid()
     pipes = []
     retval = {}
+
+    if prefork:
+        prefork()
+
     for pargs in partition(args, min(workers, len(args))):
         # Every worker gets its own pipe to send results on, so we don't have to
         # implement atomic writes larger than PIPE_BUF. Each forked process has
         # its own pipe's descriptors in the local variables, and the parent
         # process has the full list of pipe descriptors (and it doesn't really
         # care what order they're in).
         rfd, wfd = os.pipe()
         pipes.append((rfd, wfd))
         # make sure we use os._exit in all worker code paths. otherwise the
         # worker may do some clean-ups which could cause surprises like
         # deadlock. see sshpeer.cleanup for example.
         # override error handling *before* fork. this is necessary because
         # exception (signal) may arrive after fork, before "pid =" assignment
         # completes, and other exception handler (dispatch.py) can lead to
         # unexpected code path without os._exit.
         ret = -1
         try:
             pid = os.fork()
             if pid == 0:
                 signal.signal(signal.SIGINT, oldhandler)
                 signal.signal(signal.SIGCHLD, oldchldhandler)
 
                 def workerfunc():
                     for r, w in pipes[:-1]:
                         os.close(r)
                         os.close(w)
                     os.close(rfd)
                     with os.fdopen(wfd, 'wb') as wf:
                         for result in func(*(staticargs + (pargs,))):
                             pickle.dump(result, wf)
                             wf.flush()
                     return 0
 
                 ret = scmutil.callcatch(ui, workerfunc)
         except:  # parent re-raises, child never returns
             if os.getpid() == parentpid:
                 raise
             exctype = sys.exc_info()[0]
             force = not issubclass(exctype, KeyboardInterrupt)
             ui.traceback(force=force)
         finally:
             if os.getpid() != parentpid:
                 try:
                     ui.flush()
                 except:  # never returns, no re-raises
                     pass
                 finally:
                     os._exit(ret & 255)
         pids.add(pid)
     selector = selectors.DefaultSelector()
     for rfd, wfd in pipes:
         os.close(wfd)
         # The stream has to be unbuffered. Otherwise, if all data is read from
         # the raw file into the buffer, the selector thinks that the FD is not
         # ready to read while pickle.load() could read from the buffer. This
         # would delay the processing of readable items.
         selector.register(os.fdopen(rfd, 'rb', 0), selectors.EVENT_READ)
 
     def cleanup():
         signal.signal(signal.SIGINT, oldhandler)
         waitforworkers()
         signal.signal(signal.SIGCHLD, oldchldhandler)
         selector.close()
         return problem[0]
 
     try:
         openpipes = len(pipes)
         while openpipes > 0:
             for key, events in selector.select():
                 try:
                     # The pytype error likely goes away on a modern version of
                     # pytype having a modern typeshed snapshot.
                     # pytype: disable=wrong-arg-types
                     res = pickle.load(_blockingreader(key.fileobj))
                     # pytype: enable=wrong-arg-types
                     if hasretval and res[0]:
                         retval.update(res[1])
                     else:
                         yield res
                 except EOFError:
                     selector.unregister(key.fileobj)
                     # pytype: disable=attribute-error
                     key.fileobj.close()
                     # pytype: enable=attribute-error
                     openpipes -= 1
     except:  # re-raises
         killworkers()
         cleanup()
         raise
     status = cleanup()
     if status:
         if status < 0:
             os.kill(os.getpid(), -status)
         raise error.WorkerError(status)
     if hasretval:
         yield True, retval
 
 
 def _posixexitstatus(code):
     """convert a posix exit status into the same form returned by
     os.spawnv
 
     returns None if the process was stopped instead of exiting"""
     if os.WIFEXITED(code):
         return os.WEXITSTATUS(code)
     elif os.WIFSIGNALED(code):
         return -(os.WTERMSIG(code))
 
 
-def _windowsworker(ui, func, staticargs, args, hasretval):
+def _windowsworker(ui, func, staticargs, args, hasretval, prefork=None):
     class Worker(threading.Thread):
         def __init__(
             self, taskqueue, resultqueue, func, staticargs, *args, **kwargs
         ):
             threading.Thread.__init__(self, *args, **kwargs)
             self._taskqueue = taskqueue
             self._resultqueue = resultqueue
             self._func = func
             self._staticargs = staticargs
             self._interrupted = False
             self.daemon = True
             self.exception = None
 
         def interrupt(self):
             self._interrupted = True
 
         def run(self):
             try:
                 while not self._taskqueue.empty():
                     try:
                         args = self._taskqueue.get_nowait()
                         for res in self._func(*self._staticargs + (args,)):
                             self._resultqueue.put(res)
                             # threading doesn't provide a native way to
                             # interrupt execution. handle it manually at every
                             # iteration.
                             if self._interrupted:
                                 return
                     except pycompat.queue.Empty:
                         break
             except Exception as e:
                 # store the exception such that the main thread can resurface
                 # it as if the func was running without workers.
                 self.exception = e
                 raise
 
     threads = []
 
     def trykillworkers():
         # Allow up to 1 second to clean worker threads nicely
         cleanupend = time.time() + 1
         for t in threads:
             t.interrupt()
         for t in threads:
             remainingtime = cleanupend - time.time()
             t.join(remainingtime)
             if t.is_alive():
                 # pass over the workers joining failure. it is more
                 # important to surface the inital exception than the
                 # fact that one of workers may be processing a large
                 # task and does not get to handle the interruption.
                 ui.warn(
                     _(
                         b"failed to kill worker threads while "
                         b"handling an exception\n"
                     )
                 )
                 return
 
     workers = _numworkers(ui)
     resultqueue = pycompat.queue.Queue()
     taskqueue = pycompat.queue.Queue()
     retval = {}
     # partition work to more pieces than workers to minimize the chance
     # of uneven distribution of large tasks between the workers
     for pargs in partition(args, workers * 20):
         taskqueue.put(pargs)
     for _i in range(workers):
         t = Worker(taskqueue, resultqueue, func, staticargs)
         threads.append(t)
         t.start()
     try:
         while len(threads) > 0:
             while not resultqueue.empty():
                 res = resultqueue.get()
                 if hasretval and res[0]:
                     retval.update(res[1])
                 else:
                     yield res
             threads[0].join(0.05)
             finishedthreads = [_t for _t in threads if not _t.is_alive()]
             for t in finishedthreads:
                 if t.exception is not None:
                     raise t.exception
                 threads.remove(t)
     except (Exception, KeyboardInterrupt):  # re-raises
         trykillworkers()
         raise
     while not resultqueue.empty():
         res = resultqueue.get()
         if hasretval and res[0]:
             retval.update(res[1])
         else:
             yield res
     if hasretval:
         yield True, retval
 
 
 if pycompat.iswindows:
     _platformworker = _windowsworker
 else:
     _platformworker = _posixworker
     _exitstatus = _posixexitstatus
 
 
 def partition(lst, nslices):
     """partition a list into N slices of roughly equal size
 
     The current strategy takes every Nth element from the input. If
     we ever write workers that need to preserve grouping in input
     we should consider allowing callers to specify a partition strategy.
 
     olivia is not a fan of this partitioning strategy when files are involved.
     In his words:
 
         Single-threaded Mercurial makes a point of creating and visiting
         files in a fixed order (alphabetical). When creating files in order,
         a typical filesystem is likely to allocate them on nearby regions on
         disk. Thus, when revisiting in the same order, locality is maximized
         and various forms of OS and disk-level caching and read-ahead get a
         chance to work.
 
         This effect can be quite significant on spinning disks. I discovered it
         circa Mercurial v0.4 when revlogs were named by hashes of filenames.
         Tarring a repo and copying it to another disk effectively randomized
         the revlog ordering on disk by sorting the revlogs by hash and suddenly
         performance of my kernel checkout benchmark dropped by ~10x because the
         "working set" of sectors visited no longer fit in the drive's cache and
         the workload switched from streaming to random I/O.
 
         What we should really be doing is have workers read filenames from a
         ordered queue. This preserves locality and also keeps any worker from
         getting more than one file out of balance.
     """
     for i in range(nslices):
         yield lst[i::nslices]