view tests/test-addremove-similar.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

  $ hg init rep; cd rep

  $ touch empty-file
  $ "$PYTHON" -c 'for x in range(10000): print(x)' > large-file

  $ hg addremove
  adding empty-file
  adding large-file

  $ hg commit -m A

  $ rm large-file empty-file
  $ "$PYTHON" -c 'for x in range(10,10000): print(x)' > another-file

  $ hg addremove -s50
  adding another-file
  removing empty-file
  removing large-file
  recording removal of large-file as rename to another-file (99% similar)

  $ hg commit -m B

comparing two empty files caused ZeroDivisionError in the past

  $ hg update -C 0
  2 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ rm empty-file
  $ touch another-empty-file
  $ hg addremove -s50
  adding another-empty-file
  removing empty-file

  $ cd ..

  $ hg init rep2; cd rep2

  $ "$PYTHON" -c 'for x in range(10000): print(x)' > large-file
  $ "$PYTHON" -c 'for x in range(50): print(x)' > tiny-file

  $ hg addremove
  adding large-file
  adding tiny-file

  $ hg commit -m A

  $ "$PYTHON" -c 'for x in range(70): print(x)' > small-file
  $ rm tiny-file
  $ rm large-file

  $ hg addremove -s50
  removing large-file
  adding small-file
  removing tiny-file
  recording removal of tiny-file as rename to small-file (82% similar)

  $ hg commit -m B

should be sorted by path for stable result

  $ for i in `"$PYTHON" $TESTDIR/seq.py 0 9`; do
  >     cp small-file $i
  > done
  $ rm small-file
  $ hg addremove
  adding 0
  adding 1
  adding 2
  adding 3
  adding 4
  adding 5
  adding 6
  adding 7
  adding 8
  adding 9
  removing small-file
  recording removal of small-file as rename to 0 (100% similar)
  recording removal of small-file as rename to 1 (100% similar)
  recording removal of small-file as rename to 2 (100% similar)
  recording removal of small-file as rename to 3 (100% similar)
  recording removal of small-file as rename to 4 (100% similar)
  recording removal of small-file as rename to 5 (100% similar)
  recording removal of small-file as rename to 6 (100% similar)
  recording removal of small-file as rename to 7 (100% similar)
  recording removal of small-file as rename to 8 (100% similar)
  recording removal of small-file as rename to 9 (100% similar)
  $ hg commit -m '10 same files'

pick one from many identical files

  $ cp 0 a
  $ rm `"$PYTHON" $TESTDIR/seq.py 0 9`
  $ hg addremove
  removing 0
  removing 1
  removing 2
  removing 3
  removing 4
  removing 5
  removing 6
  removing 7
  removing 8
  removing 9
  adding a
  recording removal of 0 as rename to a (100% similar)
  $ hg revert -aq

pick one from many similar files

  $ cp 0 a
  $ for i in `"$PYTHON" $TESTDIR/seq.py 0 9`; do
  >     echo $i >> $i
  > done
  $ hg commit -m 'make them slightly different'
  $ rm `"$PYTHON" $TESTDIR/seq.py 0 9`
  $ hg addremove -s50
  removing 0
  removing 1
  removing 2
  removing 3
  removing 4
  removing 5
  removing 6
  removing 7
  removing 8
  removing 9
  adding a
  recording removal of 0 as rename to a (99% similar)
  $ hg commit -m 'always the same file should be selected'

should all fail

  $ hg addremove -s foo
  abort: similarity must be a number
  [255]
  $ hg addremove -s -1
  abort: similarity must be between 0 and 100
  [255]
  $ hg addremove -s 1e6
  abort: similarity must be between 0 and 100
  [255]

  $ cd ..

Issue1527: repeated addremove causes Abort

  $ hg init rep3; cd rep3
  $ mkdir d
  $ echo a > d/a
  $ hg add d/a
  $ hg commit -m 1

  $ mv d/a d/b
  $ hg addremove -s80
  removing d/a
  adding d/b
  recording removal of d/a as rename to d/b (100% similar)
  $ hg debugstate
  r   0          0 1970-01-01 00:00:00 d/a
  a   0         -1 unset               d/b
  copy: d/a -> d/b
  $ mv d/b c

no copies found here (since the target isn't in d

  $ hg addremove -s80 d
  removing d/b

copies here

  $ hg addremove -s80
  adding c
  recording removal of d/a as rename to c (100% similar)

  $ cd ..