Mercurial > hg
view tests/test-remotefilelog-histpack.py @ 42377:0546ead39a7e stable
manifest: avoid corruption by dropping removed files with pure (issue5801)
Previously, removed files would simply be marked by overwriting the first byte
with NUL and dropping their entry in `self.position`. But no effort was made to
ignore them when compacting the dictionary into text form. This allowed them to
slip into the manifest revision, since the code seems to be trying to minimize
the string operations by copying as large a chunk as possible. As part of this,
compact() walks the existing text based on entries in the `positions` list, and
consumed everything up to the next position entry. This typically resulted in
a ValueError complaining about unsorted manifest entries.
Sometimes it seems that files do get dropped in large repos- it seems to
correspond to there being a new entry that would take the same slot. A much
more trivial problem is that if the only changes were removals, `_compact()`
didn't even run because `__delitem__` doesn't add anything to `self.extradata`.
Now there's an explicit variable to flag this, both to allow `_compact()` to
run, and to avoid searching the manifest in cases where there are no removals.
In practice, this behavior was mostly obscured by the check in fastdelta() which
takes a different path that explicitly drops removed files if there are fewer
than 1000 changes. However, timeless has a repo where after rebasing tens of
commits, a totally different path[1] is taken that bypasses the change count
check and hits this problem.
[1] https://www.mercurial-scm.org/repo/hg/file/2338bdea4474/mercurial/manifest.py#l1511
author | Matt Harbison <matt_harbison@yahoo.com> |
---|---|
date | Thu, 23 May 2019 21:54:24 -0400 |
parents | 876494fd967d |
children | 2372284d9457 |
line wrap: on
line source
#!/usr/bin/env python 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: pack = 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__)