view tests/test-remotefilelog-histpack.py @ 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

#!/usr/bin/env python3
from __future__ import absolute_import

import hashlib
import os
import random
import shutil
import stat
import struct
import sys
import tempfile
import unittest

import silenttestrunner

from mercurial.node import nullid
from mercurial import (
    pycompat,
    ui as uimod,
)

# Load the local remotefilelog, not the system one
sys.path[0:0] = [os.path.join(os.path.dirname(__file__), '..')]
from hgext.remotefilelog import (
    basepack,
    historypack,
)


class histpacktests(unittest.TestCase):
    def setUp(self):
        self.tempdirs = []

    def tearDown(self):
        for d in self.tempdirs:
            shutil.rmtree(d)

    def makeTempDir(self):
        tempdir = tempfile.mkdtemp()
        self.tempdirs.append(tempdir)
        return pycompat.fsencode(tempdir)

    def getHash(self, content):
        return hashlib.sha1(content).digest()

    def getFakeHash(self):
        return b''.join(
            pycompat.bytechr(random.randint(0, 255)) for _ in range(20)
        )

    def createPack(self, revisions=None):
        """Creates and returns a historypack containing the specified revisions.

        `revisions` is a list of tuples, where each tuple contains a filanem,
        node, p1node, p2node, and linknode.
        """
        if revisions is None:
            revisions = [
                (
                    b"filename",
                    self.getFakeHash(),
                    nullid,
                    nullid,
                    self.getFakeHash(),
                    None,
                )
            ]

        packdir = pycompat.fsencode(self.makeTempDir())
        packer = historypack.mutablehistorypack(uimod.ui(), packdir, version=2)

        for filename, node, p1, p2, linknode, copyfrom in revisions:
            packer.add(filename, node, p1, p2, linknode, copyfrom)

        path = packer.close()
        return historypack.historypack(path)

    def testAddSingle(self):
        """Test putting a single entry into a pack and reading it out."""
        filename = b"foo"
        node = self.getFakeHash()
        p1 = self.getFakeHash()
        p2 = self.getFakeHash()
        linknode = self.getFakeHash()

        revisions = [(filename, node, p1, p2, linknode, None)]
        pack = self.createPack(revisions)

        actual = pack.getancestors(filename, node)[node]
        self.assertEqual(p1, actual[0])
        self.assertEqual(p2, actual[1])
        self.assertEqual(linknode, actual[2])

    def testAddMultiple(self):
        """Test putting multiple unrelated revisions into a pack and reading
        them out.
        """
        revisions = []
        for i in range(10):
            filename = b"foo-%d" % i
            node = self.getFakeHash()
            p1 = self.getFakeHash()
            p2 = self.getFakeHash()
            linknode = self.getFakeHash()
            revisions.append((filename, node, p1, p2, linknode, None))

        pack = self.createPack(revisions)

        for filename, node, p1, p2, linknode, copyfrom in revisions:
            actual = pack.getancestors(filename, node)[node]
            self.assertEqual(p1, actual[0])
            self.assertEqual(p2, actual[1])
            self.assertEqual(linknode, actual[2])
            self.assertEqual(copyfrom, actual[3])

    def testAddAncestorChain(self):
        """Test putting multiple revisions in into a pack and read the ancestor
        chain.
        """
        revisions = []
        filename = b"foo"
        lastnode = nullid
        for i in range(10):
            node = self.getFakeHash()
            revisions.append((filename, node, lastnode, nullid, nullid, None))
            lastnode = node

        # revisions must be added in topological order, newest first
        revisions = list(reversed(revisions))
        pack = self.createPack(revisions)

        # Test that the chain has all the entries
        ancestors = pack.getancestors(revisions[0][0], revisions[0][1])
        for filename, node, p1, p2, linknode, copyfrom in revisions:
            ap1, ap2, alinknode, acopyfrom = ancestors[node]
            self.assertEqual(ap1, p1)
            self.assertEqual(ap2, p2)
            self.assertEqual(alinknode, linknode)
            self.assertEqual(acopyfrom, copyfrom)

    def testPackMany(self):
        """Pack many related and unrelated ancestors."""
        # Build a random pack file
        allentries = {}
        ancestorcounts = {}
        revisions = []
        random.seed(0)
        for i in range(100):
            filename = b"filename-%d" % i
            entries = []
            p2 = nullid
            linknode = nullid
            for j in range(random.randint(1, 100)):
                node = self.getFakeHash()
                p1 = nullid
                if len(entries) > 0:
                    p1 = entries[random.randint(0, len(entries) - 1)]
                entries.append(node)
                revisions.append((filename, node, p1, p2, linknode, None))
                allentries[(filename, node)] = (p1, p2, linknode)
                if p1 == nullid:
                    ancestorcounts[(filename, node)] = 1
                else:
                    newcount = ancestorcounts[(filename, p1)] + 1
                    ancestorcounts[(filename, node)] = newcount

        # Must add file entries in reverse topological order
        revisions = list(reversed(revisions))
        pack = self.createPack(revisions)

        # Verify the pack contents
        for (filename, node) in allentries:
            ancestors = pack.getancestors(filename, node)
            self.assertEqual(ancestorcounts[(filename, node)], len(ancestors))
            for anode, (ap1, ap2, alinknode, copyfrom) in ancestors.items():
                ep1, ep2, elinknode = allentries[(filename, anode)]
                self.assertEqual(ap1, ep1)
                self.assertEqual(ap2, ep2)
                self.assertEqual(alinknode, elinknode)
                self.assertEqual(copyfrom, None)

    def testGetNodeInfo(self):
        revisions = []
        filename = b"foo"
        lastnode = nullid
        for i in range(10):
            node = self.getFakeHash()
            revisions.append((filename, node, lastnode, nullid, nullid, None))
            lastnode = node

        pack = self.createPack(revisions)

        # Test that getnodeinfo returns the expected results
        for filename, node, p1, p2, linknode, copyfrom in revisions:
            ap1, ap2, alinknode, acopyfrom = pack.getnodeinfo(filename, node)
            self.assertEqual(ap1, p1)
            self.assertEqual(ap2, p2)
            self.assertEqual(alinknode, linknode)
            self.assertEqual(acopyfrom, copyfrom)

    def testGetMissing(self):
        """Test the getmissing() api."""
        revisions = []
        filename = b"foo"
        for i in range(10):
            node = self.getFakeHash()
            p1 = self.getFakeHash()
            p2 = self.getFakeHash()
            linknode = self.getFakeHash()
            revisions.append((filename, node, p1, p2, linknode, None))

        pack = self.createPack(revisions)

        missing = pack.getmissing([(filename, revisions[0][1])])
        self.assertFalse(missing)

        missing = pack.getmissing(
            [(filename, revisions[0][1]), (filename, revisions[1][1])]
        )
        self.assertFalse(missing)

        fakenode = self.getFakeHash()
        missing = pack.getmissing(
            [(filename, revisions[0][1]), (filename, fakenode)]
        )
        self.assertEqual(missing, [(filename, fakenode)])

        # Test getmissing on a non-existant filename
        missing = pack.getmissing([(b"bar", fakenode)])
        self.assertEqual(missing, [(b"bar", fakenode)])

    def testAddThrows(self):
        pack = self.createPack()

        try:
            pack.add(b'filename', nullid, nullid, nullid, nullid, None)
            self.assertTrue(False, "historypack.add should throw")
        except RuntimeError:
            pass

    def testBadVersionThrows(self):
        pack = self.createPack()
        path = pack.path + b'.histpack'
        with open(path, 'rb') as f:
            raw = f.read()
        raw = struct.pack('!B', 255) + raw[1:]
        os.chmod(path, os.stat(path).st_mode | stat.S_IWRITE)
        with open(path, 'wb+') as f:
            f.write(raw)

        try:
            historypack.historypack(pack.path)
            self.assertTrue(False, "bad version number should have thrown")
        except RuntimeError:
            pass

    def testLargePack(self):
        """Test creating and reading from a large pack with over X entries.
        This causes it to use a 2^16 fanout table instead."""
        total = basepack.SMALLFANOUTCUTOFF + 1
        revisions = []
        for i in pycompat.xrange(total):
            filename = b"foo-%d" % i
            node = self.getFakeHash()
            p1 = self.getFakeHash()
            p2 = self.getFakeHash()
            linknode = self.getFakeHash()
            revisions.append((filename, node, p1, p2, linknode, None))

        pack = self.createPack(revisions)
        self.assertEqual(pack.params.fanoutprefix, basepack.LARGEFANOUTPREFIX)

        for filename, node, p1, p2, linknode, copyfrom in revisions:
            actual = pack.getancestors(filename, node)[node]
            self.assertEqual(p1, actual[0])
            self.assertEqual(p2, actual[1])
            self.assertEqual(linknode, actual[2])
            self.assertEqual(copyfrom, actual[3])


# TODO:
# histpack store:
# - repack two packs into one

if __name__ == '__main__':
    if pycompat.iswindows:
        sys.exit(80)  # Skip on Windows
    silenttestrunner.main(__name__)