##// END OF EJS Templates
revlog: add a mechanism to verify expected file position before appending...
revlog: add a mechanism to verify expected file position before appending If someone uses `hg debuglocks`, or some non-hg process writes to the .hg directory without respecting the locks, or if the repo's on a networked filesystem, it's possible for the revlog code to write out corrupted data. The form of this corruption can vary depending on what data was written and how that happened. We are in the "networked filesystem" case (though I've had users also do this to themselves with the "`hg debuglocks`" scenario), and most often see this with the changelog. What ends up happening is we produce two items (let's call them rev1 and rev2) in the .i file that have the same linkrev, baserev, and offset into the .d file, while the data in the .d file is appended properly. rev2's compressed_size is accurate for rev2, but when we go to decompress the data in the .d file, we use the offset that's recorded in the index file, which is the same as rev1, and attempt to decompress rev2.compressed_size bytes of rev1's data. This usually does not succeed. :) When using inline data, this also fails, though I haven't investigated why too closely. This shows up as a "patch decode" error. I believe what's happening there is that we're basically ignoring the offset field, getting the data properly, but since baserev != rev, it thinks this is a delta based on rev (instead of a full text) and can't actually apply it as such. For now, I'm going to make this an optional component and default it to entirely off. I may increase the default severity of this in the future, once I've enabled it for my users and we gain more experience with it. Luckily, most of my users have a versioned filesystem and can roll back to before the corruption has been written, it's just a hassle to do so and not everyone knows how (so it's a support burden). Users on other filesystems will not have that luxury, and this can cause them to have a corrupted repository that they are unlikely to know how to resolve, and they'll see this as a data-loss event. Refusing to create the corruption is a much better user experience. This mechanism is not perfect. There may be false-negatives (racy writes that are not detected). There should not be any false-positives (non-racy writes that are detected as such). This is not a mechanism that makes putting a repo on a networked filesystem "safe" or "supported", just *less* likely to cause corruption. Differential Revision: https://phab.mercurial-scm.org/D9952

File last commit:

r46554:89a2afe3 default
r47349:e9901d01 default
Show More
win32mbcs.py
223 lines | 7.0 KiB | text/x-python | PythonLexer
# win32mbcs.py -- MBCS filename support for Mercurial
#
# Copyright (c) 2008 Shun-ichi Goto <shunichi.goto@gmail.com>
#
# Version: 0.3
# Author: Shun-ichi Goto <shunichi.goto@gmail.com>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
#
'''allow the use of MBCS paths with problematic encodings
Some MBCS encodings are not good for some path operations (i.e.
splitting path, case conversion, etc.) with its encoded bytes. We call
such a encoding (i.e. shift_jis and big5) as "problematic encoding".
This extension can be used to fix the issue with those encodings by
wrapping some functions to convert to Unicode string before path
operation.
This extension is useful for:
- Japanese Windows users using shift_jis encoding.
- Chinese Windows users using big5 encoding.
- All users who use a repository with one of problematic encodings on
case-insensitive file system.
This extension is not needed for:
- Any user who use only ASCII chars in path.
- Any user who do not use any of problematic encodings.
Note that there are some limitations on using this extension:
- You should use single encoding in one repository.
- If the repository path ends with 0x5c, .hg/hgrc cannot be read.
- win32mbcs is not compatible with fixutf8 extension.
By default, win32mbcs uses encoding.encoding decided by Mercurial.
You can specify the encoding by config option::
[win32mbcs]
encoding = sjis
It is useful for the users who want to commit with UTF-8 log message.
'''
from __future__ import absolute_import
import os
import sys
from mercurial.i18n import _
from mercurial.pycompat import getattr, setattr
from mercurial import (
encoding,
error,
pycompat,
registrar,
)
# Note for extension authors: ONLY specify testedwith = 'ships-with-hg-core' for
# extensions which SHIP WITH MERCURIAL. Non-mainline extensions should
# be specifying the version(s) of Mercurial they are tested with, or
# leave the attribute unspecified.
testedwith = b'ships-with-hg-core'
configtable = {}
configitem = registrar.configitem(configtable)
# Encoding.encoding may be updated by --encoding option.
# Use a lambda do delay the resolution.
configitem(
b'win32mbcs',
b'encoding',
default=lambda: encoding.encoding,
)
_encoding = None # see extsetup
def decode(arg):
if isinstance(arg, bytes):
uarg = arg.decode(_encoding)
if arg == uarg.encode(_encoding):
return uarg
raise UnicodeError(b"Not local encoding")
elif isinstance(arg, tuple):
return tuple(map(decode, arg))
elif isinstance(arg, list):
return map(decode, arg)
elif isinstance(arg, dict):
for k, v in arg.items():
arg[k] = decode(v)
return arg
def encode(arg):
if isinstance(arg, pycompat.unicode):
return arg.encode(_encoding)
elif isinstance(arg, tuple):
return tuple(map(encode, arg))
elif isinstance(arg, list):
return map(encode, arg)
elif isinstance(arg, dict):
for k, v in arg.items():
arg[k] = encode(v)
return arg
def appendsep(s):
# ensure the path ends with os.sep, appending it if necessary.
try:
us = decode(s)
except UnicodeError:
us = s
if us and us[-1] not in b':/\\':
s += pycompat.ossep
return s
def basewrapper(func, argtype, enc, dec, args, kwds):
# check check already converted, then call original
for arg in args:
if isinstance(arg, argtype):
return func(*args, **kwds)
try:
# convert string arguments, call func, then convert back the
# return value.
return enc(func(*dec(args), **dec(kwds)))
except UnicodeError:
raise error.Abort(
_(b"[win32mbcs] filename conversion failed with %s encoding\n")
% _encoding
)
def wrapper(func, args, kwds):
return basewrapper(func, pycompat.unicode, encode, decode, args, kwds)
def reversewrapper(func, args, kwds):
return basewrapper(func, str, decode, encode, args, kwds)
def wrapperforlistdir(func, args, kwds):
# Ensure 'path' argument ends with os.sep to avoids
# misinterpreting last 0x5c of MBCS 2nd byte as path separator.
if args:
args = list(args)
args[0] = appendsep(args[0])
if b'path' in kwds:
kwds[b'path'] = appendsep(kwds[b'path'])
return func(*args, **kwds)
def wrapname(name, wrapper):
module, name = name.rsplit(b'.', 1)
module = sys.modules[module]
func = getattr(module, name)
def f(*args, **kwds):
return wrapper(func, args, kwds)
f.__name__ = func.__name__
setattr(module, name, f)
# List of functions to be wrapped.
# NOTE: os.path.dirname() and os.path.basename() are safe because
# they use result of os.path.split()
funcs = b'''os.path.join os.path.split os.path.splitext
os.path.normpath os.makedirs mercurial.util.endswithsep
mercurial.util.splitpath mercurial.util.fscasesensitive
mercurial.util.fspath mercurial.util.pconvert mercurial.util.normpath
mercurial.util.checkwinfilename mercurial.util.checkosfilename
mercurial.util.split'''
# These functions are required to be called with local encoded string
# because they expects argument is local encoded string and cause
# problem with unicode string.
rfuncs = b'''mercurial.encoding.upper mercurial.encoding.lower
mercurial.util._filenamebytestr'''
# List of Windows specific functions to be wrapped.
winfuncs = b'''os.path.splitunc'''
# codec and alias names of sjis and big5 to be faked.
problematic_encodings = b'''big5 big5-tw csbig5 big5hkscs big5-hkscs
hkscs cp932 932 ms932 mskanji ms-kanji shift_jis csshiftjis shiftjis
sjis s_jis shift_jis_2004 shiftjis2004 sjis_2004 sjis2004
shift_jisx0213 shiftjisx0213 sjisx0213 s_jisx0213 950 cp950 ms950 '''
def extsetup(ui):
# TODO: decide use of config section for this extension
if (not os.path.supports_unicode_filenames) and (
pycompat.sysplatform != b'cygwin'
):
ui.warn(_(b"[win32mbcs] cannot activate on this platform.\n"))
return
# determine encoding for filename
global _encoding
_encoding = ui.config(b'win32mbcs', b'encoding')
# fake is only for relevant environment.
if _encoding.lower() in problematic_encodings.split():
for f in funcs.split():
wrapname(f, wrapper)
if pycompat.iswindows:
for f in winfuncs.split():
wrapname(f, wrapper)
wrapname(b"mercurial.util.listdir", wrapperforlistdir)
wrapname(b"mercurial.windows.listdir", wrapperforlistdir)
# wrap functions to be called with local byte string arguments
for f in rfuncs.split():
wrapname(f, reversewrapper)
# Check sys.args manually instead of using ui.debug() because
# command line options is not yet applied when
# extensions.loadall() is called.
if b'--debug' in sys.argv:
ui.writenoi18n(
b"[win32mbcs] activated with encoding: %s\n" % _encoding
)