py3: remove retry on EINTR errno

Since the implementation of PEP 475 (Python 3.5), Python retries system calls
failing with EINTR. Therefore we don’t need the logic that retries it in Python
code.

mercurial/commandserver.py

@@ -1,756 +1,744 b''
 # commandserver.py - communicate with Mercurial's API over a pipe
 #
 #  Copyright Olivia Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 
 import errno
 import gc
 import os
 import random
 import selectors
 import signal
 import socket
 import struct
 import traceback
 
 from .i18n import _
 from .pycompat import getattr
 from . import (
     encoding,
     error,
     loggingutil,
     pycompat,
     repocache,
     util,
     vfs as vfsmod,
 )
 from .utils import (
     cborutil,
     procutil,
 )
 
 
 class channeledoutput:
     """
     Write data to out in the following format:
 
     data length (unsigned int),
     data
     """
 
     def __init__(self, out, channel):
         self.out = out
         self.channel = channel
 
     @property
     def name(self):
         return b'<%c-channel>' % self.channel
 
     def write(self, data):
         if not data:
             return
         # single write() to guarantee the same atomicity as the underlying file
         self.out.write(struct.pack(b'>cI', self.channel, len(data)) + data)
         self.out.flush()
 
     def __getattr__(self, attr):
         if attr in ('isatty', 'fileno', 'tell', 'seek'):
             raise AttributeError(attr)
         return getattr(self.out, attr)
 
 
 class channeledmessage:
     """
     Write encoded message and metadata to out in the following format:
 
     data length (unsigned int),
     encoded message and metadata, as a flat key-value dict.
 
     Each message should have 'type' attribute. Messages of unknown type
     should be ignored.
     """
 
     # teach ui that write() can take **opts
     structured = True
 
     def __init__(self, out, channel, encodename, encodefn):
         self._cout = channeledoutput(out, channel)
         self.encoding = encodename
         self._encodefn = encodefn
 
     def write(self, data, **opts):
         opts = pycompat.byteskwargs(opts)
         if data is not None:
             opts[b'data'] = data
         self._cout.write(self._encodefn(opts))
 
     def __getattr__(self, attr):
         return getattr(self._cout, attr)
 
 
 class channeledinput:
     """
     Read data from in_.
 
     Requests for input are written to out in the following format:
     channel identifier - 'I' for plain input, 'L' line based (1 byte)
     how many bytes to send at most (unsigned int),
 
     The client replies with:
     data length (unsigned int), 0 meaning EOF
     data
     """
 
     maxchunksize = 4 * 1024
 
     def __init__(self, in_, out, channel):
         self.in_ = in_
         self.out = out
         self.channel = channel
 
     @property
     def name(self):
         return b'<%c-channel>' % self.channel
 
     def read(self, size=-1):
         if size < 0:
             # if we need to consume all the clients input, ask for 4k chunks
             # so the pipe doesn't fill up risking a deadlock
             size = self.maxchunksize
             s = self._read(size, self.channel)
             buf = s
             while s:
                 s = self._read(size, self.channel)
                 buf += s
 
             return buf
         else:
             return self._read(size, self.channel)
 
     def _read(self, size, channel):
         if not size:
             return b''
         assert size > 0
 
         # tell the client we need at most size bytes
         self.out.write(struct.pack(b'>cI', channel, size))
         self.out.flush()
 
         length = self.in_.read(4)
         length = struct.unpack(b'>I', length)[0]
         if not length:
             return b''
         else:
             return self.in_.read(length)
 
     def readline(self, size=-1):
         if size < 0:
             size = self.maxchunksize
             s = self._read(size, b'L')
             buf = s
             # keep asking for more until there's either no more or
             # we got a full line
             while s and not s.endswith(b'\n'):
                 s = self._read(size, b'L')
                 buf += s
 
             return buf
         else:
             return self._read(size, b'L')
 
     def __iter__(self):
         return self
 
     def next(self):
         l = self.readline()
         if not l:
             raise StopIteration
         return l
 
     __next__ = next
 
     def __getattr__(self, attr):
         if attr in ('isatty', 'fileno', 'tell', 'seek'):
             raise AttributeError(attr)
         return getattr(self.in_, attr)
 
 
 _messageencoders = {
     b'cbor': lambda v: b''.join(cborutil.streamencode(v)),
 }
 
 
 def _selectmessageencoder(ui):
     encnames = ui.configlist(b'cmdserver', b'message-encodings')
     for n in encnames:
         f = _messageencoders.get(n)
         if f:
             return n, f
     raise error.Abort(
         b'no supported message encodings: %s' % b' '.join(encnames)
     )
 
 
 class server:
     """
     Listens for commands on fin, runs them and writes the output on a channel
     based stream to fout.
     """
 
     def __init__(self, ui, repo, fin, fout, prereposetups=None):
         self.cwd = encoding.getcwd()
 
         if repo:
             # the ui here is really the repo ui so take its baseui so we don't
             # end up with its local configuration
             self.ui = repo.baseui
             self.repo = repo
             self.repoui = repo.ui
         else:
             self.ui = ui
             self.repo = self.repoui = None
         self._prereposetups = prereposetups
 
         self.cdebug = channeledoutput(fout, b'd')
         self.cerr = channeledoutput(fout, b'e')
         self.cout = channeledoutput(fout, b'o')
         self.cin = channeledinput(fin, fout, b'I')
         self.cresult = channeledoutput(fout, b'r')
 
         if self.ui.config(b'cmdserver', b'log') == b'-':
             # switch log stream of server's ui to the 'd' (debug) channel
             # (don't touch repo.ui as its lifetime is longer than the server)
             self.ui = self.ui.copy()
             setuplogging(self.ui, repo=None, fp=self.cdebug)
 
         self.cmsg = None
         if ui.config(b'ui', b'message-output') == b'channel':
             encname, encfn = _selectmessageencoder(ui)
             self.cmsg = channeledmessage(fout, b'm', encname, encfn)
 
         self.client = fin
 
         # If shutdown-on-interrupt is off, the default SIGINT handler is
         # removed so that client-server communication wouldn't be interrupted.
         # For example, 'runcommand' handler will issue three short read()s.
         # If one of the first two read()s were interrupted, the communication
         # channel would be left at dirty state and the subsequent request
         # wouldn't be parsed. So catching KeyboardInterrupt isn't enough.
         self._shutdown_on_interrupt = ui.configbool(
             b'cmdserver', b'shutdown-on-interrupt'
         )
         self._old_inthandler = None
         if not self._shutdown_on_interrupt:
             self._old_inthandler = signal.signal(signal.SIGINT, signal.SIG_IGN)
 
     def cleanup(self):
         """release and restore resources taken during server session"""
         if not self._shutdown_on_interrupt:
             signal.signal(signal.SIGINT, self._old_inthandler)
 
     def _read(self, size):
         if not size:
             return b''
 
         data = self.client.read(size)
 
         # is the other end closed?
         if not data:
             raise EOFError
 
         return data
 
     def _readstr(self):
         """read a string from the channel
 
         format:
         data length (uint32), data
         """
         length = struct.unpack(b'>I', self._read(4))[0]
         if not length:
             return b''
         return self._read(length)
 
     def _readlist(self):
         """read a list of NULL separated strings from the channel"""
         s = self._readstr()
         if s:
             return s.split(b'\0')
         else:
             return []
 
     def _dispatchcommand(self, req):
         from . import dispatch  # avoid cycle
 
         if self._shutdown_on_interrupt:
             # no need to restore SIGINT handler as it is unmodified.
             return dispatch.dispatch(req)
 
         try:
             signal.signal(signal.SIGINT, self._old_inthandler)
             return dispatch.dispatch(req)
         except error.SignalInterrupt:
             # propagate SIGBREAK, SIGHUP, or SIGTERM.
             raise
         except KeyboardInterrupt:
             # SIGINT may be received out of the try-except block of dispatch(),
             # so catch it as last ditch. Another KeyboardInterrupt may be
             # raised while handling exceptions here, but there's no way to
             # avoid that except for doing everything in C.
             pass
         finally:
             signal.signal(signal.SIGINT, signal.SIG_IGN)
         # On KeyboardInterrupt, print error message and exit *after* SIGINT
         # handler removed.
         req.ui.error(_(b'interrupted!\n'))
         return -1
 
     def runcommand(self):
         """reads a list of \0 terminated arguments, executes
         and writes the return code to the result channel"""
         from . import dispatch  # avoid cycle
 
         args = self._readlist()
 
         # copy the uis so changes (e.g. --config or --verbose) don't
         # persist between requests
         copiedui = self.ui.copy()
         uis = [copiedui]
         if self.repo:
             self.repo.baseui = copiedui
             # clone ui without using ui.copy because this is protected
             repoui = self.repoui.__class__(self.repoui)
             repoui.copy = copiedui.copy  # redo copy protection
             uis.append(repoui)
             self.repo.ui = self.repo.dirstate._ui = repoui
             self.repo.invalidateall()
 
         for ui in uis:
             ui.resetstate()
             # any kind of interaction must use server channels, but chg may
             # replace channels by fully functional tty files. so nontty is
             # enforced only if cin is a channel.
             if not util.safehasattr(self.cin, b'fileno'):
                 ui.setconfig(b'ui', b'nontty', b'true', b'commandserver')
 
         req = dispatch.request(
             args[:],
             copiedui,
             self.repo,
             self.cin,
             self.cout,
             self.cerr,
             self.cmsg,
             prereposetups=self._prereposetups,
         )
 
         try:
             ret = self._dispatchcommand(req) & 255
             # If shutdown-on-interrupt is off, it's important to write the
             # result code *after* SIGINT handler removed. If the result code
             # were lost, the client wouldn't be able to continue processing.
             self.cresult.write(struct.pack(b'>i', int(ret)))
         finally:
             # restore old cwd
             if b'--cwd' in args:
                 os.chdir(self.cwd)
 
     def getencoding(self):
         """writes the current encoding to the result channel"""
         self.cresult.write(encoding.encoding)
 
     def serveone(self):
         cmd = self.client.readline()[:-1]
         if cmd:
             handler = self.capabilities.get(cmd)
             if handler:
                 handler(self)
             else:
                 # clients are expected to check what commands are supported by
                 # looking at the servers capabilities
                 raise error.Abort(_(b'unknown command %s') % cmd)
 
         return cmd != b''
 
     capabilities = {b'runcommand': runcommand, b'getencoding': getencoding}
 
     def serve(self):
         hellomsg = b'capabilities: ' + b' '.join(sorted(self.capabilities))
         hellomsg += b'\n'
         hellomsg += b'encoding: ' + encoding.encoding
         hellomsg += b'\n'
         if self.cmsg:
             hellomsg += b'message-encoding: %s\n' % self.cmsg.encoding
         hellomsg += b'pid: %d' % procutil.getpid()
         if util.safehasattr(os, b'getpgid'):
             hellomsg += b'\n'
             hellomsg += b'pgid: %d' % os.getpgid(0)
 
         # write the hello msg in -one- chunk
         self.cout.write(hellomsg)
 
         try:
             while self.serveone():
                 pass
         except EOFError:
             # we'll get here if the client disconnected while we were reading
             # its request
             return 1
 
         return 0
 
 
 def setuplogging(ui, repo=None, fp=None):
     """Set up server logging facility
 
     If cmdserver.log is '-', log messages will be sent to the given fp.
     It should be the 'd' channel while a client is connected, and otherwise
     is the stderr of the server process.
     """
     # developer config: cmdserver.log
     logpath = ui.config(b'cmdserver', b'log')
     if not logpath:
         return
     # developer config: cmdserver.track-log
     tracked = set(ui.configlist(b'cmdserver', b'track-log'))
 
     if logpath == b'-' and fp:
         logger = loggingutil.fileobjectlogger(fp, tracked)
     elif logpath == b'-':
         logger = loggingutil.fileobjectlogger(ui.ferr, tracked)
     else:
         logpath = util.abspath(util.expandpath(logpath))
         # developer config: cmdserver.max-log-files
         maxfiles = ui.configint(b'cmdserver', b'max-log-files')
         # developer config: cmdserver.max-log-size
         maxsize = ui.configbytes(b'cmdserver', b'max-log-size')
         vfs = vfsmod.vfs(os.path.dirname(logpath))
         logger = loggingutil.filelogger(
             vfs,
             os.path.basename(logpath),
             tracked,
             maxfiles=maxfiles,
             maxsize=maxsize,
         )
 
     targetuis = {ui}
     if repo:
         targetuis.add(repo.baseui)
         targetuis.add(repo.ui)
     for u in targetuis:
         u.setlogger(b'cmdserver', logger)
 
 
 class pipeservice:
     def __init__(self, ui, repo, opts):
         self.ui = ui
         self.repo = repo
 
     def init(self):
         pass
 
     def run(self):
         ui = self.ui
         # redirect stdio to null device so that broken extensions or in-process
         # hooks will never cause corruption of channel protocol.
         with ui.protectedfinout() as (fin, fout):
             sv = server(ui, self.repo, fin, fout)
             try:
                 return sv.serve()
             finally:
                 sv.cleanup()
 
 
 def _initworkerprocess():
     # use a different process group from the master process, in order to:
     # 1. make the current process group no longer "orphaned" (because the
     #    parent of this process is in a different process group while
     #    remains in a same session)
     #    according to POSIX 2.2.2.52, orphaned process group will ignore
     #    terminal-generated stop signals like SIGTSTP (Ctrl+Z), which will
     #    cause trouble for things like ncurses.
     # 2. the client can use kill(-pgid, sig) to simulate terminal-generated
     #    SIGINT (Ctrl+C) and process-exit-generated SIGHUP. our child
     #    processes like ssh will be killed properly, without affecting
     #    unrelated processes.
     os.setpgid(0, 0)
     # change random state otherwise forked request handlers would have a
     # same state inherited from parent.
     random.seed()
 
 
 def _serverequest(ui, repo, conn, createcmdserver, prereposetups):
     fin = conn.makefile('rb')
     fout = conn.makefile('wb')
     sv = None
     try:
         sv = createcmdserver(repo, conn, fin, fout, prereposetups)
         try:
             sv.serve()
         # handle exceptions that may be raised by command server. most of
         # known exceptions are caught by dispatch.
         except error.Abort as inst:
             ui.error(_(b'abort: %s\n') % inst.message)
         except IOError as inst:
             if inst.errno != errno.EPIPE:
                 raise
         except KeyboardInterrupt:
             pass
         finally:
             sv.cleanup()
     except:  # re-raises
         # also write traceback to error channel. otherwise client cannot
         # see it because it is written to server's stderr by default.
         if sv:
             cerr = sv.cerr
         else:
             cerr = channeledoutput(fout, b'e')
         cerr.write(encoding.strtolocal(traceback.format_exc()))
         raise
     finally:
         fin.close()
         try:
             fout.close()  # implicit flush() may cause another EPIPE
         except IOError as inst:
             if inst.errno != errno.EPIPE:
                 raise
 
 
 class unixservicehandler:
     """Set of pluggable operations for unix-mode services
 
     Almost all methods except for createcmdserver() are called in the main
     process. You can't pass mutable resource back from createcmdserver().
     """
 
     pollinterval = None
 
     def __init__(self, ui):
         self.ui = ui
 
     def bindsocket(self, sock, address):
         util.bindunixsocket(sock, address)
         sock.listen(socket.SOMAXCONN)
         self.ui.status(_(b'listening at %s\n') % address)
         self.ui.flush()  # avoid buffering of status message
 
     def unlinksocket(self, address):
         os.unlink(address)
 
     def shouldexit(self):
         """True if server should shut down; checked per pollinterval"""
         return False
 
     def newconnection(self):
         """Called when main process notices new connection"""
 
     def createcmdserver(self, repo, conn, fin, fout, prereposetups):
         """Create new command server instance; called in the process that
         serves for the current connection"""
         return server(self.ui, repo, fin, fout, prereposetups)
 
 
 class unixforkingservice:
     """
     Listens on unix domain socket and forks server per connection
     """
 
     def __init__(self, ui, repo, opts, handler=None):
         self.ui = ui
         self.repo = repo
         self.address = opts[b'address']
         if not util.safehasattr(socket, b'AF_UNIX'):
             raise error.Abort(_(b'unsupported platform'))
         if not self.address:
             raise error.Abort(_(b'no socket path specified with --address'))
         self._servicehandler = handler or unixservicehandler(ui)
         self._sock = None
         self._mainipc = None
         self._workeripc = None
         self._oldsigchldhandler = None
         self._workerpids = set()  # updated by signal handler; do not iterate
         self._socketunlinked = None
         # experimental config: cmdserver.max-repo-cache
         maxlen = ui.configint(b'cmdserver', b'max-repo-cache')
         if maxlen < 0:
             raise error.Abort(_(b'negative max-repo-cache size not allowed'))
         self._repoloader = repocache.repoloader(ui, maxlen)
         # attempt to avoid crash in CoreFoundation when using chg after fix in
         # a89381e04c58
         if pycompat.isdarwin:
             procutil.gui()
 
     def init(self):
         self._sock = socket.socket(socket.AF_UNIX)
         # IPC channel from many workers to one main process; this is actually
         # a uni-directional pipe, but is backed by a DGRAM socket so each
         # message can be easily separated.
         o = socket.socketpair(socket.AF_UNIX, socket.SOCK_DGRAM)
         self._mainipc, self._workeripc = o
         self._servicehandler.bindsocket(self._sock, self.address)
         if util.safehasattr(procutil, b'unblocksignal'):
             procutil.unblocksignal(signal.SIGCHLD)
         o = signal.signal(signal.SIGCHLD, self._sigchldhandler)
         self._oldsigchldhandler = o
         self._socketunlinked = False
         self._repoloader.start()
 
     def _unlinksocket(self):
         if not self._socketunlinked:
             self._servicehandler.unlinksocket(self.address)
             self._socketunlinked = True
 
     def _cleanup(self):
         signal.signal(signal.SIGCHLD, self._oldsigchldhandler)
         self._sock.close()
         self._mainipc.close()
         self._workeripc.close()
         self._unlinksocket()
         self._repoloader.stop()
         # don't kill child processes as they have active clients, just wait
         self._reapworkers(0)
 
     def run(self):
         try:
             self._mainloop()
         finally:
             self._cleanup()
 
     def _mainloop(self):
         exiting = False
         h = self._servicehandler
         selector = selectors.DefaultSelector()
         selector.register(
             self._sock, selectors.EVENT_READ, self._acceptnewconnection
         )
         selector.register(
             self._mainipc, selectors.EVENT_READ, self._handlemainipc
         )
         while True:
             if not exiting and h.shouldexit():
                 # clients can no longer connect() to the domain socket, so
                 # we stop queuing new requests.
                 # for requests that are queued (connect()-ed, but haven't been
                 # accept()-ed), handle them before exit. otherwise, clients
                 # waiting for recv() will receive ECONNRESET.
                 self._unlinksocket()
                 exiting = True
             events = selector.select(timeout=h.pollinterval)
             if not events:
                 # only exit if we completed all queued requests
                 if exiting:
                     break
                 continue
             for key, _mask in events:
                 key.data(key.fileobj, selector)
         selector.close()
 
     def _acceptnewconnection(self, sock, selector):
         h = self._servicehandler
-        try:
-            conn, _addr = sock.accept()
-        except socket.error as inst:
-            if inst.args[0] == errno.EINTR:
-                return
-            raise
+        conn, _addr = sock.accept()
 
         # Future improvement: On Python 3.7, maybe gc.freeze() can be used
         # to prevent COW memory from being touched by GC.
         # https://instagram-engineering.com/
         #   copy-on-write-friendly-python-garbage-collection-ad6ed5233ddf
         pid = os.fork()
         if pid:
             try:
                 self.ui.log(
                     b'cmdserver', b'forked worker process (pid=%d)\n', pid
                 )
                 self._workerpids.add(pid)
                 h.newconnection()
             finally:
                 conn.close()  # release handle in parent process
         else:
             try:
                 selector.close()
                 sock.close()
                 self._mainipc.close()
                 self._runworker(conn)
                 conn.close()
                 self._workeripc.close()
                 os._exit(0)
             except:  # never return, hence no re-raises
                 try:
                     self.ui.traceback(force=True)
                 finally:
                     os._exit(255)
 
     def _handlemainipc(self, sock, selector):
         """Process messages sent from a worker"""
-        try:
-            path = sock.recv(32768)  # large enough to receive path
-        except socket.error as inst:
-            if inst.args[0] == errno.EINTR:
-                return
-            raise
+        path = sock.recv(32768)  # large enough to receive path
         self._repoloader.load(path)
 
     def _sigchldhandler(self, signal, frame):
         self._reapworkers(os.WNOHANG)
 
     def _reapworkers(self, options):
         while self._workerpids:
             try:
                 pid, _status = os.waitpid(-1, options)
             except OSError as inst:
-                if inst.errno == errno.EINTR:
-                    continue
                 if inst.errno != errno.ECHILD:
                     raise
                 # no child processes at all (reaped by other waitpid()?)
                 self._workerpids.clear()
                 return
             if pid == 0:
                 # no waitable child processes
                 return
             self.ui.log(b'cmdserver', b'worker process exited (pid=%d)\n', pid)
             self._workerpids.discard(pid)
 
     def _runworker(self, conn):
         signal.signal(signal.SIGCHLD, self._oldsigchldhandler)
         _initworkerprocess()
         h = self._servicehandler
         try:
             _serverequest(
                 self.ui,
                 self.repo,
                 conn,
                 h.createcmdserver,
                 prereposetups=[self._reposetup],
             )
         finally:
             gc.collect()  # trigger __del__ since worker process uses os._exit
 
     def _reposetup(self, ui, repo):
         if not repo.local():
             return
 
         class unixcmdserverrepo(repo.__class__):
             def close(self):
                 super(unixcmdserverrepo, self).close()
                 try:
                     self._cmdserveripc.send(self.root)
                 except socket.error:
                     self.ui.log(
                         b'cmdserver', b'failed to send repo root to master\n'
                     )
 
         repo.__class__ = unixcmdserverrepo
         repo._cmdserveripc = self._workeripc
 
         cachedrepo = self._repoloader.get(repo.root)
         if cachedrepo is None:
             return
         repo.ui.log(b'repocache', b'repo from cache: %s\n', repo.root)
         repocache.copycache(cachedrepo, repo)

mercurial/posix.py

@@ -1,774 +1,767 b''
 # posix.py - Posix utility function implementations for Mercurial
 #
 #  Copyright 2005-2009 Olivia Mackall <[e-mail]> and others
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 
 import errno
 import fcntl
 import getpass
 import grp
 import os
 import pwd
 import re
 import select
 import stat
 import sys
 import tempfile
 import unicodedata
 
 from .i18n import _
 from .pycompat import (
     getattr,
     open,
 )
 from . import (
     encoding,
     error,
     policy,
     pycompat,
 )
 
 osutil = policy.importmod('osutil')
 
 normpath = os.path.normpath
 samestat = os.path.samestat
 abspath = os.path.abspath  # re-exports
 
 try:
     oslink = os.link
 except AttributeError:
     # Some platforms build Python without os.link on systems that are
     # vaguely unix-like but don't have hardlink support. For those
     # poor souls, just say we tried and that it failed so we fall back
     # to copies.
     def oslink(src, dst):
         raise OSError(
             errno.EINVAL, b'hardlinks not supported: %s to %s' % (src, dst)
         )
 
 
 readlink = os.readlink
 unlink = os.unlink
 rename = os.rename
 removedirs = os.removedirs
 expandglobs = False
 
 umask = os.umask(0)
 os.umask(umask)
 
 posixfile = open
 
 
 def split(p):
     """Same as posixpath.split, but faster
 
     >>> import posixpath
     >>> for f in [b'/absolute/path/to/file',
     ...           b'relative/path/to/file',
     ...           b'file_alone',
     ...           b'path/to/directory/',
     ...           b'/multiple/path//separators',
     ...           b'/file_at_root',
     ...           b'///multiple_leading_separators_at_root',
     ...           b'']:
     ...     assert split(f) == posixpath.split(f), f
     """
     ht = p.rsplit(b'/', 1)
     if len(ht) == 1:
         return b'', p
     nh = ht[0].rstrip(b'/')
     if nh:
         return nh, ht[1]
     return ht[0] + b'/', ht[1]
 
 
 def openhardlinks():
     '''return true if it is safe to hold open file handles to hardlinks'''
     return True
 
 
 def nlinks(name):
     '''return number of hardlinks for the given file'''
     return os.lstat(name).st_nlink
 
 
 def parsepatchoutput(output_line):
     """parses the output produced by patch and returns the filename"""
     pf = output_line[14:]
     if pycompat.sysplatform == b'OpenVMS':
         if pf[0] == b'`':
             pf = pf[1:-1]  # Remove the quotes
     else:
         if pf.startswith(b"'") and pf.endswith(b"'") and b" " in pf:
             pf = pf[1:-1]  # Remove the quotes
     return pf
 
 
 def sshargs(sshcmd, host, user, port):
     '''Build argument list for ssh'''
     args = user and (b"%s@%s" % (user, host)) or host
     if b'-' in args[:1]:
         raise error.Abort(
             _(b'illegal ssh hostname or username starting with -: %s') % args
         )
     args = shellquote(args)
     if port:
         args = b'-p %s %s' % (shellquote(port), args)
     return args
 
 
 def isexec(f):
     """check whether a file is executable"""
     return os.lstat(f).st_mode & 0o100 != 0
 
 
 def setflags(f, l, x):
     st = os.lstat(f)
     s = st.st_mode
     if l:
         if not stat.S_ISLNK(s):
             # switch file to link
             with open(f, b'rb') as fp:
                 data = fp.read()
             unlink(f)
             try:
                 os.symlink(data, f)
             except OSError:
                 # failed to make a link, rewrite file
                 with open(f, b"wb") as fp:
                     fp.write(data)
 
         # no chmod needed at this point
         return
     if stat.S_ISLNK(s):
         # switch link to file
         data = os.readlink(f)
         unlink(f)
         with open(f, b"wb") as fp:
             fp.write(data)
         s = 0o666 & ~umask  # avoid restatting for chmod
 
     sx = s & 0o100
     if st.st_nlink > 1 and bool(x) != bool(sx):
         # the file is a hardlink, break it
         with open(f, b"rb") as fp:
             data = fp.read()
         unlink(f)
         with open(f, b"wb") as fp:
             fp.write(data)
 
     if x and not sx:
         # Turn on +x for every +r bit when making a file executable
         # and obey umask.
         os.chmod(f, s | (s & 0o444) >> 2 & ~umask)
     elif not x and sx:
         # Turn off all +x bits
         os.chmod(f, s & 0o666)
 
 
 def copymode(src, dst, mode=None, enforcewritable=False):
     """Copy the file mode from the file at path src to dst.
     If src doesn't exist, we're using mode instead. If mode is None, we're
     using umask."""
     try:
         st_mode = os.lstat(src).st_mode & 0o777
     except OSError as inst:
         if inst.errno != errno.ENOENT:
             raise
         st_mode = mode
         if st_mode is None:
             st_mode = ~umask
         st_mode &= 0o666
 
     new_mode = st_mode
 
     if enforcewritable:
         new_mode |= stat.S_IWUSR
 
     os.chmod(dst, new_mode)
 
 
 def checkexec(path):
     """
     Check whether the given path is on a filesystem with UNIX-like exec flags
 
     Requires a directory (like /foo/.hg)
     """
 
     # VFAT on some Linux versions can flip mode but it doesn't persist
     # a FS remount. Frequently we can detect it if files are created
     # with exec bit on.
 
     try:
         EXECFLAGS = stat.S_IXUSR | stat.S_IXGRP | stat.S_IXOTH
         basedir = os.path.join(path, b'.hg')
         cachedir = os.path.join(basedir, b'wcache')
         storedir = os.path.join(basedir, b'store')
         if not os.path.exists(cachedir):
             try:
                 # we want to create the 'cache' directory, not the '.hg' one.
                 # Automatically creating '.hg' directory could silently spawn
                 # invalid Mercurial repositories. That seems like a bad idea.
                 os.mkdir(cachedir)
                 if os.path.exists(storedir):
                     copymode(storedir, cachedir)
                 else:
                     copymode(basedir, cachedir)
             except (IOError, OSError):
                 # we other fallback logic triggers
                 pass
         if os.path.isdir(cachedir):
             checkisexec = os.path.join(cachedir, b'checkisexec')
             checknoexec = os.path.join(cachedir, b'checknoexec')
 
             try:
                 m = os.stat(checkisexec).st_mode
             except OSError as e:
                 if e.errno != errno.ENOENT:
                     raise
                 # checkisexec does not exist - fall through ...
             else:
                 # checkisexec exists, check if it actually is exec
                 if m & EXECFLAGS != 0:
                     # ensure checkisexec exists, check it isn't exec
                     try:
                         m = os.stat(checknoexec).st_mode
                     except OSError as e:
                         if e.errno != errno.ENOENT:
                             raise
                         open(checknoexec, b'w').close()  # might fail
                         m = os.stat(checknoexec).st_mode
                     if m & EXECFLAGS == 0:
                         # check-exec is exec and check-no-exec is not exec
                         return True
                     # checknoexec exists but is exec - delete it
                     unlink(checknoexec)
                 # checkisexec exists but is not exec - delete it
                 unlink(checkisexec)
 
             # check using one file, leave it as checkisexec
             checkdir = cachedir
         else:
             # check directly in path and don't leave checkisexec behind
             checkdir = path
             checkisexec = None
         fh, fn = pycompat.mkstemp(dir=checkdir, prefix=b'hg-checkexec-')
         try:
             os.close(fh)
             m = os.stat(fn).st_mode
             if m & EXECFLAGS == 0:
                 os.chmod(fn, m & 0o777 | EXECFLAGS)
                 if os.stat(fn).st_mode & EXECFLAGS != 0:
                     if checkisexec is not None:
                         os.rename(fn, checkisexec)
                         fn = None
                     return True
         finally:
             if fn is not None:
                 unlink(fn)
     except (IOError, OSError):
         # we don't care, the user probably won't be able to commit anyway
         return False
 
 
 def checklink(path):
     """check whether the given path is on a symlink-capable filesystem"""
     # mktemp is not racy because symlink creation will fail if the
     # file already exists
     while True:
         cachedir = os.path.join(path, b'.hg', b'wcache')
         checklink = os.path.join(cachedir, b'checklink')
         # try fast path, read only
         if os.path.islink(checklink):
             return True
         if os.path.isdir(cachedir):
             checkdir = cachedir
         else:
             checkdir = path
             cachedir = None
         name = tempfile.mktemp(
             dir=pycompat.fsdecode(checkdir), prefix=r'checklink-'
         )
         name = pycompat.fsencode(name)
         try:
             fd = None
             if cachedir is None:
                 fd = pycompat.namedtempfile(
                     dir=checkdir, prefix=b'hg-checklink-'
                 )
                 target = os.path.basename(fd.name)
             else:
                 # create a fixed file to link to; doesn't matter if it
                 # already exists.
                 target = b'checklink-target'
                 try:
                     fullpath = os.path.join(cachedir, target)
                     open(fullpath, b'w').close()
                 except IOError as inst:
                     # pytype: disable=unsupported-operands
                     if inst[0] == errno.EACCES:
                         # pytype: enable=unsupported-operands
 
                         # If we can't write to cachedir, just pretend
                         # that the fs is readonly and by association
                         # that the fs won't support symlinks. This
                         # seems like the least dangerous way to avoid
                         # data loss.
                         return False
                     raise
             try:
                 os.symlink(target, name)
                 if cachedir is None:
                     unlink(name)
                 else:
                     try:
                         os.rename(name, checklink)
                     except OSError:
                         unlink(name)
                 return True
             except OSError as inst:
                 # link creation might race, try again
                 if inst.errno == errno.EEXIST:
                     continue
                 raise
             finally:
                 if fd is not None:
                     fd.close()
         except AttributeError:
             return False
         except OSError as inst:
             # sshfs might report failure while successfully creating the link
             if inst.errno == errno.EIO and os.path.exists(name):
                 unlink(name)
             return False
 
 
 def checkosfilename(path):
     """Check that the base-relative path is a valid filename on this platform.
     Returns None if the path is ok, or a UI string describing the problem."""
     return None  # on posix platforms, every path is ok
 
 
 def getfsmountpoint(dirpath):
     """Get the filesystem mount point from a directory (best-effort)
 
     Returns None if we are unsure. Raises OSError on ENOENT, EPERM, etc.
     """
     return getattr(osutil, 'getfsmountpoint', lambda x: None)(dirpath)
 
 
 def getfstype(dirpath):
     """Get the filesystem type name from a directory (best-effort)
 
     Returns None if we are unsure. Raises OSError on ENOENT, EPERM, etc.
     """
     return getattr(osutil, 'getfstype', lambda x: None)(dirpath)
 
 
 def get_password():
     return encoding.strtolocal(getpass.getpass(''))
 
 
 def setbinary(fd):
     pass
 
 
 def pconvert(path):
     return path
 
 
 def localpath(path):
     return path
 
 
 def samefile(fpath1, fpath2):
     """Returns whether path1 and path2 refer to the same file. This is only
     guaranteed to work for files, not directories."""
     return os.path.samefile(fpath1, fpath2)
 
 
 def samedevice(fpath1, fpath2):
     """Returns whether fpath1 and fpath2 are on the same device. This is only
     guaranteed to work for files, not directories."""
     st1 = os.lstat(fpath1)
     st2 = os.lstat(fpath2)
     return st1.st_dev == st2.st_dev
 
 
 # os.path.normcase is a no-op, which doesn't help us on non-native filesystems
 def normcase(path):
     return path.lower()
 
 
 # what normcase does to ASCII strings
 normcasespec = encoding.normcasespecs.lower
 # fallback normcase function for non-ASCII strings
 normcasefallback = normcase
 
 if pycompat.isdarwin:
 
     def normcase(path):
         """
         Normalize a filename for OS X-compatible comparison:
         - escape-encode invalid characters
         - decompose to NFD
         - lowercase
         - omit ignored characters [200c-200f, 202a-202e, 206a-206f,feff]
 
         >>> normcase(b'UPPER')
         'upper'
         >>> normcase(b'Caf\\xc3\\xa9')
         'cafe\\xcc\\x81'
         >>> normcase(b'\\xc3\\x89')
         'e\\xcc\\x81'
         >>> normcase(b'\\xb8\\xca\\xc3\\xca\\xbe\\xc8.JPG') # issue3918
         '%b8%ca%c3\\xca\\xbe%c8.jpg'
         """
 
         try:
             return encoding.asciilower(path)  # exception for non-ASCII
         except UnicodeDecodeError:
             return normcasefallback(path)
 
     normcasespec = encoding.normcasespecs.lower
 
     def normcasefallback(path):
         try:
             u = path.decode('utf-8')
         except UnicodeDecodeError:
             # OS X percent-encodes any bytes that aren't valid utf-8
             s = b''
             pos = 0
             l = len(path)
             while pos < l:
                 try:
                     c = encoding.getutf8char(path, pos)
                     pos += len(c)
                 except ValueError:
                     c = b'%%%02X' % ord(path[pos : pos + 1])
                     pos += 1
                 s += c
 
             u = s.decode('utf-8')
 
         # Decompose then lowercase (HFS+ technote specifies lower)
         enc = unicodedata.normalize('NFD', u).lower().encode('utf-8')
         # drop HFS+ ignored characters
         return encoding.hfsignoreclean(enc)
 
 
 if pycompat.sysplatform == b'cygwin':
     # workaround for cygwin, in which mount point part of path is
     # treated as case sensitive, even though underlying NTFS is case
     # insensitive.
 
     # default mount points
     cygwinmountpoints = sorted(
         [
             b"/usr/bin",
             b"/usr/lib",
             b"/cygdrive",
         ],
         reverse=True,
     )
 
     # use upper-ing as normcase as same as NTFS workaround
     def normcase(path):
         pathlen = len(path)
         if (pathlen == 0) or (path[0] != pycompat.ossep):
             # treat as relative
             return encoding.upper(path)
 
         # to preserve case of mountpoint part
         for mp in cygwinmountpoints:
             if not path.startswith(mp):
                 continue
 
             mplen = len(mp)
             if mplen == pathlen:  # mount point itself
                 return mp
             if path[mplen] == pycompat.ossep:
                 return mp + encoding.upper(path[mplen:])
 
         return encoding.upper(path)
 
     normcasespec = encoding.normcasespecs.other
     normcasefallback = normcase
 
     # Cygwin translates native ACLs to POSIX permissions,
     # but these translations are not supported by native
     # tools, so the exec bit tends to be set erroneously.
     # Therefore, disable executable bit access on Cygwin.
     def checkexec(path):
         return False
 
     # Similarly, Cygwin's symlink emulation is likely to create
     # problems when Mercurial is used from both Cygwin and native
     # Windows, with other native tools, or on shared volumes
     def checklink(path):
         return False
 
 
 _needsshellquote = None
 
 
 def shellquote(s):
     if pycompat.sysplatform == b'OpenVMS':
         return b'"%s"' % s
     global _needsshellquote
     if _needsshellquote is None:
         _needsshellquote = re.compile(br'[^a-zA-Z0-9._/+-]').search
     if s and not _needsshellquote(s):
         # "s" shouldn't have to be quoted
         return s
     else:
         return b"'%s'" % s.replace(b"'", b"'\\''")
 
 
 def shellsplit(s):
     """Parse a command string in POSIX shell way (best-effort)"""
     return pycompat.shlexsplit(s, posix=True)
 
 
 def testpid(pid):
     '''return False if pid dead, True if running or not sure'''
     if pycompat.sysplatform == b'OpenVMS':
         return True
     try:
         os.kill(pid, 0)
         return True
     except OSError as inst:
         return inst.errno != errno.ESRCH
 
 
 def isowner(st):
     """Return True if the stat object st is from the current user."""
     return st.st_uid == os.getuid()
 
 
 def findexe(command):
     """Find executable for command searching like which does.
     If command is a basename then PATH is searched for command.
     PATH isn't searched if command is an absolute or relative path.
     If command isn't found None is returned."""
     if pycompat.sysplatform == b'OpenVMS':
         return command
 
     def findexisting(executable):
         b'Will return executable if existing file'
         if os.path.isfile(executable) and os.access(executable, os.X_OK):
             return executable
         return None
 
     if pycompat.ossep in command:
         return findexisting(command)
 
     if pycompat.sysplatform == b'plan9':
         return findexisting(os.path.join(b'/bin', command))
 
     for path in encoding.environ.get(b'PATH', b'').split(pycompat.ospathsep):
         executable = findexisting(os.path.join(path, command))
         if executable is not None:
             return executable
     return None
 
 
 def setsignalhandler():
     pass
 
 
 _wantedkinds = {stat.S_IFREG, stat.S_IFLNK}
 
 
 def statfiles(files):
     """Stat each file in files. Yield each stat, or None if a file does not
     exist or has a type we don't care about."""
     lstat = os.lstat
     getkind = stat.S_IFMT
     for nf in files:
         try:
             st = lstat(nf)
             if getkind(st.st_mode) not in _wantedkinds:
                 st = None
         except OSError as err:
             if err.errno not in (errno.ENOENT, errno.ENOTDIR):
                 raise
             st = None
         yield st
 
 
 def getuser():
     '''return name of current user'''
     return pycompat.fsencode(getpass.getuser())
 
 
 def username(uid=None):
     """Return the name of the user with the given uid.
 
     If uid is None, return the name of the current user."""
 
     if uid is None:
         uid = os.getuid()
     try:
         return pycompat.fsencode(pwd.getpwuid(uid)[0])
     except KeyError:
         return b'%d' % uid
 
 
 def groupname(gid=None):
     """Return the name of the group with the given gid.
 
     If gid is None, return the name of the current group."""
 
     if gid is None:
         gid = os.getgid()
     try:
         return pycompat.fsencode(grp.getgrgid(gid)[0])
     except KeyError:
         return pycompat.bytestr(gid)
 
 
 def groupmembers(name):
     """Return the list of members of the group with the given
     name, KeyError if the group does not exist.
     """
     name = pycompat.fsdecode(name)
     return pycompat.rapply(pycompat.fsencode, list(grp.getgrnam(name).gr_mem))
 
 
 def spawndetached(args):
     return os.spawnvp(os.P_NOWAIT | getattr(os, 'P_DETACH', 0), args[0], args)
 
 
 def gethgcmd():
     return sys.argv[:1]
 
 
 def makedir(path, notindexed):
     os.mkdir(path)
 
 
 def lookupreg(key, name=None, scope=None):
     return None
 
 
 def hidewindow():
     """Hide current shell window.
 
     Used to hide the window opened when starting asynchronous
     child process under Windows, unneeded on other systems.
     """
     pass
 
 
 class cachestat:
     def __init__(self, path):
         self.stat = os.stat(path)
 
     def cacheable(self):
         return bool(self.stat.st_ino)
 
     __hash__ = object.__hash__
 
     def __eq__(self, other):
         try:
             # Only dev, ino, size, mtime and atime are likely to change. Out
             # of these, we shouldn't compare atime but should compare the
             # rest. However, one of the other fields changing indicates
             # something fishy going on, so return False if anything but atime
             # changes.
             return (
                 self.stat.st_mode == other.stat.st_mode
                 and self.stat.st_ino == other.stat.st_ino
                 and self.stat.st_dev == other.stat.st_dev
                 and self.stat.st_nlink == other.stat.st_nlink
                 and self.stat.st_uid == other.stat.st_uid
                 and self.stat.st_gid == other.stat.st_gid
                 and self.stat.st_size == other.stat.st_size
                 and self.stat[stat.ST_MTIME] == other.stat[stat.ST_MTIME]
                 and self.stat[stat.ST_CTIME] == other.stat[stat.ST_CTIME]
             )
         except AttributeError:
             return False
 
     def __ne__(self, other):
         return not self == other
 
 
 def statislink(st):
     '''check whether a stat result is a symlink'''
     return st and stat.S_ISLNK(st.st_mode)
 
 
 def statisexec(st):
     '''check whether a stat result is an executable file'''
     return st and (st.st_mode & 0o100 != 0)
 
 
 def poll(fds):
     """block until something happens on any file descriptor
 
     This is a generic helper that will check for any activity
     (read, write.  exception) and return the list of touched files.
 
     In unsupported cases, it will raise a NotImplementedError"""
     try:
-        while True:
-            try:
-                res = select.select(fds, fds, fds)
-                break
-            except select.error as inst:
-                if inst.args[0] == errno.EINTR:
-                    continue
-                raise
+        res = select.select(fds, fds, fds)
     except ValueError:  # out of range file descriptor
         raise NotImplementedError()
     return sorted(list(set(sum(res, []))))
 
 
 def readpipe(pipe):
     """Read all available data from a pipe."""
     # We can't fstat() a pipe because Linux will always report 0.
     # So, we set the pipe to non-blocking mode and read everything
     # that's available.
     flags = fcntl.fcntl(pipe, fcntl.F_GETFL)
     flags |= os.O_NONBLOCK
     oldflags = fcntl.fcntl(pipe, fcntl.F_SETFL, flags)
 
     try:
         chunks = []
         while True:
             try:
                 s = pipe.read()
                 if not s:
                     break
                 chunks.append(s)
             except IOError:
                 break
 
         return b''.join(chunks)
     finally:
         fcntl.fcntl(pipe, fcntl.F_SETFL, oldflags)
 
 
 def bindunixsocket(sock, path):
     """Bind the UNIX domain socket to the specified path"""
     # use relative path instead of full path at bind() if possible, since
     # AF_UNIX path has very small length limit (107 chars) on common
     # platforms (see sys/un.h)
     dirname, basename = os.path.split(path)
     bakwdfd = None
 
     try:
         if dirname:
             bakwdfd = os.open(b'.', os.O_DIRECTORY)
             os.chdir(dirname)
         sock.bind(basename)
         if bakwdfd:
             os.fchdir(bakwdfd)
     finally:
         if bakwdfd:
             os.close(bakwdfd)

mercurial/progress.py

@@ -1,315 +1,295 b''
 # progress.py progress bars related code
 #
 # Copyright (C) 2010 Augie Fackler <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 
-import errno
 import threading
 import time
 
 from .i18n import _
 from . import encoding
 
 
 def spacejoin(*args):
     return b' '.join(s for s in args if s)
 
 
 def shouldprint(ui):
     return not (ui.quiet or ui.plain(b'progress')) and (
         ui._isatty(ui.ferr) or ui.configbool(b'progress', b'assume-tty')
     )
 
 
 def fmtremaining(seconds):
     """format a number of remaining seconds in human readable way
 
     This will properly display seconds, minutes, hours, days if needed"""
     if seconds < 60:
         # i18n: format XX seconds as "XXs"
         return _(b"%02ds") % seconds
     minutes = seconds // 60
     if minutes < 60:
         seconds -= minutes * 60
         # i18n: format X minutes and YY seconds as "XmYYs"
         return _(b"%dm%02ds") % (minutes, seconds)
     # we're going to ignore seconds in this case
     minutes += 1
     hours = minutes // 60
     minutes -= hours * 60
     if hours < 30:
         # i18n: format X hours and YY minutes as "XhYYm"
         return _(b"%dh%02dm") % (hours, minutes)
     # we're going to ignore minutes in this case
     hours += 1
     days = hours // 24
     hours -= days * 24
     if days < 15:
         # i18n: format X days and YY hours as "XdYYh"
         return _(b"%dd%02dh") % (days, hours)
     # we're going to ignore hours in this case
     days += 1
     weeks = days // 7
     days -= weeks * 7
     if weeks < 55:
         # i18n: format X weeks and YY days as "XwYYd"
         return _(b"%dw%02dd") % (weeks, days)
     # we're going to ignore days and treat a year as 52 weeks
     weeks += 1
     years = weeks // 52
     weeks -= years * 52
     # i18n: format X years and YY weeks as "XyYYw"
     return _(b"%dy%02dw") % (years, weeks)
 
 
-# file_write() and file_flush() of Python 2 do not restart on EINTR if
-# the file is attached to a "slow" device (e.g. a terminal) and raise
-# IOError. We cannot know how many bytes would be written by file_write(),
-# but a progress text is known to be short enough to be written by a
-# single write() syscall, so we can just retry file_write() with the whole
-# text. (issue5532)
-#
-# This should be a short-term workaround. We'll need to fix every occurrence
-# of write() to a terminal or pipe.
-def _eintrretry(func, *args):
-    while True:
-        try:
-            return func(*args)
-        except IOError as err:
-            if err.errno == errno.EINTR:
-                continue
-            raise
-
-
 class progbar:
     def __init__(self, ui):
         self.ui = ui
         self._refreshlock = threading.Lock()
         self.resetstate()
 
     def resetstate(self):
         self.topics = []
         self.topicstates = {}
         self.starttimes = {}
         self.startvals = {}
         self.printed = False
         self.lastprint = time.time() + float(
             self.ui.config(b'progress', b'delay')
         )
         self.curtopic = None
         self.lasttopic = None
         self.indetcount = 0
         self.refresh = float(self.ui.config(b'progress', b'refresh'))
         self.changedelay = max(
             3 * self.refresh, float(self.ui.config(b'progress', b'changedelay'))
         )
         self.order = self.ui.configlist(b'progress', b'format')
         self.estimateinterval = self.ui.configwith(
             float, b'progress', b'estimateinterval'
         )
 
     def show(self, now, topic, pos, item, unit, total):
         if not shouldprint(self.ui):
             return
         termwidth = self.width()
         self.printed = True
         head = b''
         needprogress = False
         tail = b''
         for indicator in self.order:
             add = b''
             if indicator == b'topic':
                 add = topic
             elif indicator == b'number':
                 if total:
                     add = b'%*d/%d' % (len(str(total)), pos, total)
                 else:
                     add = b'%d' % pos
             elif indicator.startswith(b'item') and item:
                 slice = b'end'
                 if b'-' in indicator:
                     wid = int(indicator.split(b'-')[1])
                 elif b'+' in indicator:
                     slice = b'beginning'
                     wid = int(indicator.split(b'+')[1])
                 else:
                     wid = 20
                 if slice == b'end':
                     add = encoding.trim(item, wid, leftside=True)
                 else:
                     add = encoding.trim(item, wid)
                 add += (wid - encoding.colwidth(add)) * b' '
             elif indicator == b'bar':
                 add = b''
                 needprogress = True
             elif indicator == b'unit' and unit:
                 add = unit
             elif indicator == b'estimate':
                 add = self.estimate(topic, pos, total, now)
             elif indicator == b'speed':
                 add = self.speed(topic, pos, unit, now)
             if not needprogress:
                 head = spacejoin(head, add)
             else:
                 tail = spacejoin(tail, add)
         if needprogress:
             used = 0
             if head:
                 used += encoding.colwidth(head) + 1
             if tail:
                 used += encoding.colwidth(tail) + 1
             progwidth = termwidth - used - 3
             if total and pos <= total:
                 amt = pos * progwidth // total
                 bar = b'=' * (amt - 1)
                 if amt > 0:
                     bar += b'>'
                 bar += b' ' * (progwidth - amt)
             else:
                 progwidth -= 3
                 self.indetcount += 1
                 # mod the count by twice the width so we can make the
                 # cursor bounce between the right and left sides
                 amt = self.indetcount % (2 * progwidth)
                 amt -= progwidth
                 bar = (
                     b' ' * int(progwidth - abs(amt))
                     + b'<=>'
                     + b' ' * int(abs(amt))
                 )
             prog = b''.join((b'[', bar, b']'))
             out = spacejoin(head, prog, tail)
         else:
             out = spacejoin(head, tail)
         self._writeerr(b'\r' + encoding.trim(out, termwidth))
         self.lasttopic = topic
         self._flusherr()
 
     def clear(self):
         if not self.printed or not self.lastprint or not shouldprint(self.ui):
             return
         self._writeerr(b'\r%s\r' % (b' ' * self.width()))
         self._flusherr()
         if self.printed:
             # force immediate re-paint of progress bar
             self.lastprint = 0
 
     def complete(self):
         if not shouldprint(self.ui):
             return
         if self.ui.configbool(b'progress', b'clear-complete'):
             self.clear()
         else:
             self._writeerr(b'\n')
         self._flusherr()
 
     def _flusherr(self):
-        _eintrretry(self.ui.ferr.flush)
+        self.ui.ferr.flush()
 
     def _writeerr(self, msg):
-        _eintrretry(self.ui.ferr.write, msg)
+        self.ui.ferr.write(msg)
 
     def width(self):
         tw = self.ui.termwidth()
         return min(int(self.ui.config(b'progress', b'width', default=tw)), tw)
 
     def estimate(self, topic, pos, total, now):
         if total is None:
             return b''
         initialpos = self.startvals[topic]
         target = total - initialpos
         delta = pos - initialpos
         if delta > 0:
             elapsed = now - self.starttimes[topic]
             seconds = (elapsed * (target - delta)) // delta + 1
             return fmtremaining(seconds)
         return b''
 
     def speed(self, topic, pos, unit, now):
         initialpos = self.startvals[topic]
         delta = pos - initialpos
         elapsed = now - self.starttimes[topic]
         if elapsed > 0:
             return _(b'%d %s/sec') % (delta / elapsed, unit)
         return b''
 
     def _oktoprint(self, now):
         '''Check if conditions are met to print - e.g. changedelay elapsed'''
         if (
             self.lasttopic is None  # first time we printed
             # not a topic change
             or self.curtopic == self.lasttopic
             # it's been long enough we should print anyway
             or now - self.lastprint >= self.changedelay
         ):
             return True
         else:
             return False
 
     def _calibrateestimate(self, topic, now, pos):
         """Adjust starttimes and startvals for topic so ETA works better
 
         If progress is non-linear (ex. get much slower in the last minute),
         it's more friendly to only use a recent time span for ETA and speed
         calculation.
 
             [======================================>       ]
                                              ^^^^^^^
                            estimateinterval, only use this for estimation
         """
         interval = self.estimateinterval
         if interval <= 0:
             return
         elapsed = now - self.starttimes[topic]
         if elapsed > interval:
             delta = pos - self.startvals[topic]
             newdelta = delta * interval / elapsed
             # If a stall happens temporarily, ETA could change dramatically
             # frequently. This is to avoid such dramatical change and make ETA
             # smoother.
             if newdelta < 0.1:
                 return
             self.startvals[topic] = pos - newdelta
             self.starttimes[topic] = now - interval
 
     def progress(self, topic, pos, item=b'', unit=b'', total=None):
         if pos is None:
             self.closetopic(topic)
             return
         now = time.time()
         with self._refreshlock:
             if topic not in self.topics:
                 self.starttimes[topic] = now
                 self.startvals[topic] = pos
                 self.topics.append(topic)
             self.topicstates[topic] = pos, item, unit, total
             self.curtopic = topic
             self._calibrateestimate(topic, now, pos)
             if now - self.lastprint >= self.refresh and self.topics:
                 if self._oktoprint(now):
                     self.lastprint = now
                     self.show(now, topic, *self.topicstates[topic])
 
     def closetopic(self, topic):
         with self._refreshlock:
             self.starttimes.pop(topic, None)
             self.startvals.pop(topic, None)
             self.topicstates.pop(topic, None)
             # reset the progress bar if this is the outermost topic
             if self.topics and self.topics[0] == topic and self.printed:
                 self.complete()
                 self.resetstate()
             # truncate the list of topics assuming all topics within
             # this one are also closed
             if topic in self.topics:
                 self.topics = self.topics[: self.topics.index(topic)]
                 # reset the last topic to the one we just unwound to,
                 # so that higher-level topics will be stickier than
                 # lower-level topics
                 if self.topics:
                     self.lasttopic = self.topics[-1]
                 else:
                     self.lasttopic = None

mercurial/worker.py

@@ -1,469 +1,460 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 errno
 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
 ):
     """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.
     """
     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 func(*staticargs + (args,))
 
 
 def _posixworker(ui, func, staticargs, args, hasretval):
     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 OSError as err:
                 if err.errno != errno.ESRCH:
                     raise
 
     def waitforworkers(blocking=True):
         for pid in pids.copy():
             p = st = 0
-            while True:
-                try:
-                    p, st = os.waitpid(pid, (0 if blocking else os.WNOHANG))
-                    break
-                except OSError as e:
-                    if e.errno == errno.EINTR:
-                        continue
-                    elif e.errno == errno.ECHILD:
-                        # child would already be reaped, but pids yet been
-                        # updated (maybe interrupted just after waitpid)
-                        pids.discard(pid)
-                        break
-                    else:
-                        raise
+            try:
+                p, st = os.waitpid(pid, (0 if blocking else os.WNOHANG))
+            except OSError as e:
+                if e.errno == errno.ECHILD:
+                    # child would already be reaped, but pids yet been
+                    # updated (maybe interrupted just after waitpid)
+                    pids.discard(pid)
+                else:
+                    raise
             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 = {}
     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 IOError as e:
-                    if e.errno == errno.EINTR:
-                        continue
-                    raise
     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):
     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]