Mercurial > hg
view tests/test-win32text.t @ 46607:e9901d01d135
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
author | Kyle Lippincott <spectral@google.com> |
---|---|
date | Wed, 03 Feb 2021 16:33:10 -0800 |
parents | 768056549737 |
children | 9f1b9e128788 |
line wrap: on
line source
$ hg init t $ cd t $ cat > unix2dos.py <<EOF > import sys > > for path in sys.argv[1:]: > data = open(path, 'rb').read() > data = data.replace(b'\n', b'\r\n') > open(path, 'wb').write(data) > EOF $ echo '[hooks]' >> .hg/hgrc $ echo 'pretxncommit.crlf = python:hgext.win32text.forbidcrlf' >> .hg/hgrc $ echo 'pretxnchangegroup.crlf = python:hgext.win32text.forbidcrlf' >> .hg/hgrc $ cat .hg/hgrc [hooks] pretxncommit.crlf = python:hgext.win32text.forbidcrlf pretxnchangegroup.crlf = python:hgext.win32text.forbidcrlf $ echo hello > f $ hg add f commit should succeed $ hg ci -m 1 $ hg clone . ../zoz updating to branch default 1 files updated, 0 files merged, 0 files removed, 0 files unresolved $ cp .hg/hgrc ../zoz/.hg $ "$PYTHON" unix2dos.py f commit should fail $ hg ci -m 2.1 attempt to commit or push text file(s) using CRLF line endings in f583ea08d42a: f transaction abort! rollback completed abort: pretxncommit.crlf hook failed [40] $ mv .hg/hgrc .hg/hgrc.bak commits should succeed $ hg ci -m 2 $ hg cp f g $ hg ci -m 2.2 push should fail $ hg push ../zoz pushing to ../zoz searching for changes adding changesets adding manifests adding file changes attempt to commit or push text file(s) using CRLF line endings in bc2d09796734: g in b1aa5cde7ff4: f To prevent this mistake in your local repository, add to Mercurial.ini or .hg/hgrc: [hooks] pretxncommit.crlf = python:hgext.win32text.forbidcrlf and also consider adding: [extensions] win32text = [encode] ** = cleverencode: [decode] ** = cleverdecode: transaction abort! rollback completed abort: pretxnchangegroup.crlf hook failed [40] $ mv .hg/hgrc.bak .hg/hgrc $ echo hello > f $ hg rm g commit should succeed $ hg ci -m 2.3 push should succeed $ hg push ../zoz pushing to ../zoz searching for changes adding changesets adding manifests adding file changes added 3 changesets with 3 changes to 2 files and now for something completely different $ mkdir d $ echo hello > d/f2 $ "$PYTHON" unix2dos.py d/f2 $ hg add d/f2 $ hg ci -m 3 attempt to commit or push text file(s) using CRLF line endings in 053ba1a3035a: d/f2 transaction abort! rollback completed abort: pretxncommit.crlf hook failed [40] $ hg revert -a forgetting d/f2 $ rm d/f2 $ hg rem f $ hg ci -m 4 $ "$PYTHON" -c 'open("bin", "wb").write(b"hello\x00\x0D\x0A")' $ hg add bin $ hg ci -m 5 $ hg log -v changeset: 5:f0b1c8d75fce tag: tip user: test date: Thu Jan 01 00:00:00 1970 +0000 files: bin description: 5 changeset: 4:77796dbcd4ad user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f description: 4 changeset: 3:7c1b5430b350 user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f g description: 2.3 changeset: 2:bc2d09796734 user: test date: Thu Jan 01 00:00:00 1970 +0000 files: g description: 2.2 changeset: 1:b1aa5cde7ff4 user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f description: 2 changeset: 0:fcf06d5c4e1d user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f description: 1 $ hg clone . dupe updating to branch default 1 files updated, 0 files merged, 0 files removed, 0 files unresolved $ for x in a b c d; do echo content > dupe/$x; done $ hg -R dupe add adding dupe/a adding dupe/b adding dupe/c adding dupe/d $ "$PYTHON" unix2dos.py dupe/b dupe/c dupe/d $ hg -R dupe ci -m a dupe/a $ hg -R dupe ci -m b/c dupe/[bc] $ hg -R dupe ci -m d dupe/d $ hg -R dupe log -v changeset: 8:67ac5962ab43 tag: tip user: test date: Thu Jan 01 00:00:00 1970 +0000 files: d description: d changeset: 7:68c127d1834e user: test date: Thu Jan 01 00:00:00 1970 +0000 files: b c description: b/c changeset: 6:adbf8bf7f31d user: test date: Thu Jan 01 00:00:00 1970 +0000 files: a description: a changeset: 5:f0b1c8d75fce user: test date: Thu Jan 01 00:00:00 1970 +0000 files: bin description: 5 changeset: 4:77796dbcd4ad user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f description: 4 changeset: 3:7c1b5430b350 user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f g description: 2.3 changeset: 2:bc2d09796734 user: test date: Thu Jan 01 00:00:00 1970 +0000 files: g description: 2.2 changeset: 1:b1aa5cde7ff4 user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f description: 2 changeset: 0:fcf06d5c4e1d user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f description: 1 $ hg pull dupe pulling from dupe searching for changes adding changesets adding manifests adding file changes attempt to commit or push text file(s) using CRLF line endings in 67ac5962ab43: d in 68c127d1834e: b in 68c127d1834e: c To prevent this mistake in your local repository, add to Mercurial.ini or .hg/hgrc: [hooks] pretxncommit.crlf = python:hgext.win32text.forbidcrlf and also consider adding: [extensions] win32text = [encode] ** = cleverencode: [decode] ** = cleverdecode: transaction abort! rollback completed abort: pretxnchangegroup.crlf hook failed [40] $ hg log -v changeset: 5:f0b1c8d75fce tag: tip user: test date: Thu Jan 01 00:00:00 1970 +0000 files: bin description: 5 changeset: 4:77796dbcd4ad user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f description: 4 changeset: 3:7c1b5430b350 user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f g description: 2.3 changeset: 2:bc2d09796734 user: test date: Thu Jan 01 00:00:00 1970 +0000 files: g description: 2.2 changeset: 1:b1aa5cde7ff4 user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f description: 2 changeset: 0:fcf06d5c4e1d user: test date: Thu Jan 01 00:00:00 1970 +0000 files: f description: 1 $ rm .hg/hgrc $ (echo some; echo text) > f3 $ "$PYTHON" -c 'open("f4.bat", "wb").write(b"rem empty\x0D\x0A")' $ hg add f3 f4.bat $ hg ci -m 6 $ cat bin hello\x00\r (esc) $ cat f3 some text $ cat f4.bat rem empty\r (esc) $ echo '[extensions]' >> .hg/hgrc $ echo 'win32text = ' >> .hg/hgrc $ echo '[decode]' >> .hg/hgrc $ echo '** = cleverdecode:' >> .hg/hgrc $ echo '[encode]' >> .hg/hgrc $ echo '** = cleverencode:' >> .hg/hgrc $ cat .hg/hgrc [extensions] win32text = [decode] ** = cleverdecode: [encode] ** = cleverencode: Trigger deprecation warning: $ hg id -t win32text is deprecated: https://mercurial-scm.org/wiki/Win32TextExtension tip Disable warning: $ echo '[win32text]' >> .hg/hgrc $ echo 'warn = no' >> .hg/hgrc $ hg id -t tip $ rm f3 f4.bat bin $ hg co -C WARNING: f4.bat already has CRLF line endings and does not need EOL conversion by the win32text plugin. Before your next commit, please reconsider your encode/decode settings in Mercurial.ini or $TESTTMP/t/.hg/hgrc. 3 files updated, 0 files merged, 0 files removed, 0 files unresolved $ cat bin hello\x00\r (esc) $ cat f3 some\r (esc) text\r (esc) $ cat f4.bat rem empty\r (esc) $ "$PYTHON" -c 'open("f5.sh", "wb").write(b"# empty\x0D\x0A")' $ hg add f5.sh $ hg ci -m 7 $ cat f5.sh # empty\r (esc) $ hg cat f5.sh # empty $ echo '% just linefeed' > linefeed $ hg ci -qAm 8 linefeed $ cat linefeed % just linefeed $ hg cat linefeed % just linefeed $ hg st -q $ hg revert -a linefeed no changes needed to linefeed $ cat linefeed % just linefeed $ hg st -q $ echo modified >> linefeed $ hg st -q M linefeed $ hg revert -a reverting linefeed $ hg st -q $ cat linefeed % just linefeed\r (esc) $ cd ..