( )⚙ D5256 manifest: make sure there's a filename before bothering to look for newline

This is an archive of the discontinued Mercurial Phabricator instance.

Differential D5256

manifest: make sure there's a filename before bothering to look for newline
ClosedPublic

Authored by durin42 on Nov 12 2018, 8:51 PM.

Details

Reviewers
None
Group Reviewers
hg-reviewers
Commits
rHG9eeda7199181: manifest: make sure there's a filename before bothering to look for newline
Summary

There's no valid manifest that would have no characters before the NUL byte on
a line, and this fixes some erratic timeouts in the fuzzer.

Diff Detail

Repository
rHG Mercurial
Lint
Lint Skipped
Unit
Unit Tests Skipped

Revision Contents
Changeset List

PathPackages
Mmercurial/cext/manifest.c (14 lines)
Mtests/test-manifest.py (24 lines)

Diff 12508

mercurial/cext/manifest.c

int livelines; /* number of non-deleted lines */ int livelines; /* number of non-deleted lines */
int maxlines; /* allocated number of lines */ int maxlines; /* allocated number of lines */
bool dirty; bool dirty;
} lazymanifest; } lazymanifest;
#define MANIFEST_OOM -1 #define MANIFEST_OOM -1
#define MANIFEST_NOT_SORTED -2 #define MANIFEST_NOT_SORTED -2
#define MANIFEST_MALFORMED -3 #define MANIFEST_MALFORMED -3
#define MANIFEST_BOGUS_FILENAME -4
/* get the length of the path for a line */ /* get the length of the path for a line */
static size_t pathlen(line *l) static size_t pathlen(line *l)
{ {
const char *end = memchr(l->start, '\0', l->len); const char *end = memchr(l->start, '\0', l->len);
return (end) ? (size_t)(end - l->start) : l->len; return (end) ? (size_t)(end - l->start) : l->len;
} }
* Find the line boundaries in the manifest that 'data' points to and store * Find the line boundaries in the manifest that 'data' points to and store
* information about each line in 'self'. * information about each line in 'self'.
*/ */
static int find_lines(lazymanifest *self, char *data, Py_ssize_t len) static int find_lines(lazymanifest *self, char *data, Py_ssize_t len)
{ {
char *prev = NULL; char *prev = NULL;
while (len > 0) { while (len > 0) {
line *l; line *l;
char *next = memchr(data, '\n', len); char *next;
if (*data == '\0') {
/* It's implausible there's no filename, don't
* even bother looking for the newline. */
return MANIFEST_BOGUS_FILENAME;
}
next = memchr(data, '\n', len);
if (!next) { if (!next) {
return MANIFEST_MALFORMED; return MANIFEST_MALFORMED;
} }
next++; /* advance past newline */ next++; /* advance past newline */
if (!realloc_if_full(self)) { if (!realloc_if_full(self)) {
return MANIFEST_OOM; /* no memory */ return MANIFEST_OOM; /* no memory */
} }
if (prev && strcmp(prev, data) > -1) { if (prev && strcmp(prev, data) > -1) {
case MANIFEST_NOT_SORTED: case MANIFEST_NOT_SORTED:
PyErr_Format(PyExc_ValueError, PyErr_Format(PyExc_ValueError,
"Manifest lines not in sorted order."); "Manifest lines not in sorted order.");
break; break;
case MANIFEST_MALFORMED: case MANIFEST_MALFORMED:
PyErr_Format(PyExc_ValueError, PyErr_Format(PyExc_ValueError,
"Manifest did not end in a newline."); "Manifest did not end in a newline.");
break; break;
case MANIFEST_BOGUS_FILENAME:
PyErr_Format(
PyExc_ValueError,
"Manifest had an entry with a zero-length filename.");
break;
default: default:
PyErr_Format(PyExc_ValueError, PyErr_Format(PyExc_ValueError,
"Unknown problem parsing manifest."); "Unknown problem parsing manifest.");
} }
return ret == 0 ? 0 : -1; return ret == 0 ? 0 : -1;
} }
static void lazymanifest_dealloc(lazymanifest *self) static void lazymanifest_dealloc(lazymanifest *self)

tests/test-manifest.py

from __future__ import absolute_import from __future__ import absolute_import
import binascii import binascii
import itertools import itertools
import silenttestrunner import silenttestrunner
import unittest import unittest
import zlib
from mercurial import ( from mercurial import (
manifest as manifestmod, manifest as manifestmod,
match as matchmod, match as matchmod,
) )
EMTPY_MANIFEST = b'' EMTPY_MANIFEST = b''
self.assertEqual( self.assertEqual(
[b'a/b/c/bar.txt', b'a/b/c/foo.txt', b'a/b/d/ten.txt'], [b'a/b/c/bar.txt', b'a/b/c/foo.txt', b'a/b/d/ten.txt'],
m2.keys()) m2.keys())
class testmanifestdict(unittest.TestCase, basemanifesttests): class testmanifestdict(unittest.TestCase, basemanifesttests):
def parsemanifest(self, text): def parsemanifest(self, text):
return manifestmod.manifestdict(text) return manifestmod.manifestdict(text)
def testObviouslyBogusManifest(self):
# This is a 163k manifest that came from oss-fuzz. It was a
# timeout there, but when run normally it doesn't seem to
# present any particular slowness.
data = zlib.decompress(
'x\x9c\xed\xce;\n\x83\x00\x10\x04\xd0\x8deNa\x93~\xf1\x03\xc9q\xf4'
'\x14\xeaU\xbdB\xda\xd4\xe6Cj\xc1FA\xde+\x86\xe9f\xa2\xfci\xbb\xfb'
'\xa3\xef\xea\xba\xca\x7fk\x86q\x9a\xc6\xc8\xcc&\xb3\xcf\xf8\xb8|#'
'\x8a9\x00\xd8\xe6v\xf4\x01N\xe1\n\x00\x00\x00\x00\x00\x00\x00\x00'
'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
'\x00\x00\xc0\x8aey\x1d}\x01\xd8\xe0\xb9\xf3\xde\x1b\xcf\x17'
'\xac\xbe')
with self.assertRaises(ValueError):
self.parsemanifest(data)
class testtreemanifest(unittest.TestCase, basemanifesttests): class testtreemanifest(unittest.TestCase, basemanifesttests):
def parsemanifest(self, text): def parsemanifest(self, text):
return manifestmod.treemanifest(b'', text) return manifestmod.treemanifest(b'', text)
def testWalkSubtrees(self): def testWalkSubtrees(self):
m = self.parsemanifest(A_DEEPER_MANIFEST) m = self.parsemanifest(A_DEEPER_MANIFEST)
dirs = [s._dir for s in m.walksubtrees()] dirs = [s._dir for s in m.walksubtrees()]