Show More
@@ -1,383 +1,390 b'' | |||
|
1 | 1 | # verify.py - repository integrity checking for Mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2006, 2007 Matt Mackall <mpm@selenic.com> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import os |
|
11 | 11 | |
|
12 | 12 | from .i18n import _ |
|
13 | 13 | from .node import ( |
|
14 | 14 | nullid, |
|
15 | 15 | short, |
|
16 | 16 | ) |
|
17 | 17 | |
|
18 | 18 | from . import ( |
|
19 | 19 | error, |
|
20 | 20 | revlog, |
|
21 | 21 | util, |
|
22 | 22 | ) |
|
23 | 23 | |
|
24 | 24 | def verify(repo): |
|
25 | 25 | lock = repo.lock() |
|
26 | 26 | try: |
|
27 | 27 | return verifier(repo).verify() |
|
28 | 28 | finally: |
|
29 | 29 | lock.release() |
|
30 | 30 | |
|
31 | 31 | def _normpath(f): |
|
32 | 32 | # under hg < 2.4, convert didn't sanitize paths properly, so a |
|
33 | 33 | # converted repo may contain repeated slashes |
|
34 | 34 | while '//' in f: |
|
35 | 35 | f = f.replace('//', '/') |
|
36 | 36 | return f |
|
37 | 37 | |
|
38 | 38 | def _validpath(repo, path): |
|
39 | 39 | """Returns False if a path should NOT be treated as part of a repo. |
|
40 | 40 | |
|
41 | 41 | For all in-core cases, this returns True, as we have no way for a |
|
42 | 42 | path to be mentioned in the history but not actually be |
|
43 | 43 | relevant. For narrow clones, this is important because many |
|
44 | 44 | filelogs will be missing, and changelog entries may mention |
|
45 | 45 | modified files that are outside the narrow scope. |
|
46 | 46 | """ |
|
47 | 47 | return True |
|
48 | 48 | |
|
49 | 49 | class verifier(object): |
|
50 | 50 | def __init__(self, repo): |
|
51 | 51 | self.repo = repo.unfiltered() |
|
52 | 52 | self.ui = repo.ui |
|
53 | 53 | self.badrevs = set() |
|
54 | 54 | self.errors = 0 |
|
55 | 55 | self.warnings = 0 |
|
56 | 56 | self.havecl = len(repo.changelog) > 0 |
|
57 | 57 | self.havemf = len(repo.manifest) > 0 |
|
58 | 58 | self.revlogv1 = repo.changelog.version != revlog.REVLOGV0 |
|
59 | 59 | self.lrugetctx = util.lrucachefunc(repo.changectx) |
|
60 | 60 | self.refersmf = False |
|
61 | 61 | self.fncachewarned = False |
|
62 | 62 | |
|
63 | 63 | def warn(self, msg): |
|
64 | 64 | self.ui.warn(msg + "\n") |
|
65 | 65 | self.warnings += 1 |
|
66 | 66 | |
|
67 | 67 | def err(self, linkrev, msg, filename=None): |
|
68 | 68 | if linkrev is not None: |
|
69 | 69 | self.badrevs.add(linkrev) |
|
70 | 70 | else: |
|
71 | 71 | linkrev = '?' |
|
72 | 72 | msg = "%s: %s" % (linkrev, msg) |
|
73 | 73 | if filename: |
|
74 | 74 | msg = "%s@%s" % (filename, msg) |
|
75 | 75 | self.ui.warn(" " + msg + "\n") |
|
76 | 76 | self.errors += 1 |
|
77 | 77 | |
|
78 | 78 | def exc(self, linkrev, msg, inst, filename=None): |
|
79 | 79 | if not str(inst): |
|
80 | 80 | inst = repr(inst) |
|
81 | 81 | self.err(linkrev, "%s: %s" % (msg, inst), filename) |
|
82 | 82 | |
|
83 | 83 | def checklog(self, obj, name, linkrev): |
|
84 | 84 | if not len(obj) and (self.havecl or self.havemf): |
|
85 | 85 | self.err(linkrev, _("empty or missing %s") % name) |
|
86 | 86 | return |
|
87 | 87 | |
|
88 | 88 | d = obj.checksize() |
|
89 | 89 | if d[0]: |
|
90 | 90 | self.err(None, _("data length off by %d bytes") % d[0], name) |
|
91 | 91 | if d[1]: |
|
92 | 92 | self.err(None, _("index contains %d extra bytes") % d[1], name) |
|
93 | 93 | |
|
94 | 94 | if obj.version != revlog.REVLOGV0: |
|
95 | 95 | if not self.revlogv1: |
|
96 | 96 | self.warn(_("warning: `%s' uses revlog format 1") % name) |
|
97 | 97 | elif self.revlogv1: |
|
98 | 98 | self.warn(_("warning: `%s' uses revlog format 0") % name) |
|
99 | 99 | |
|
100 | 100 | def checkentry(self, obj, i, node, seen, linkrevs, f): |
|
101 | 101 | lr = obj.linkrev(obj.rev(node)) |
|
102 | 102 | if lr < 0 or (self.havecl and lr not in linkrevs): |
|
103 | 103 | if lr < 0 or lr >= len(self.repo.changelog): |
|
104 | 104 | msg = _("rev %d points to nonexistent changeset %d") |
|
105 | 105 | else: |
|
106 | 106 | msg = _("rev %d points to unexpected changeset %d") |
|
107 | 107 | self.err(None, msg % (i, lr), f) |
|
108 | 108 | if linkrevs: |
|
109 | 109 | if f and len(linkrevs) > 1: |
|
110 | 110 | try: |
|
111 | 111 | # attempt to filter down to real linkrevs |
|
112 | 112 | linkrevs = [l for l in linkrevs |
|
113 | 113 | if self.lrugetctx(l)[f].filenode() == node] |
|
114 | 114 | except Exception: |
|
115 | 115 | pass |
|
116 | 116 | self.warn(_(" (expected %s)") % " ".join(map(str, linkrevs))) |
|
117 | 117 | lr = None # can't be trusted |
|
118 | 118 | |
|
119 | 119 | try: |
|
120 | 120 | p1, p2 = obj.parents(node) |
|
121 | 121 | if p1 not in seen and p1 != nullid: |
|
122 | 122 | self.err(lr, _("unknown parent 1 %s of %s") % |
|
123 | 123 | (short(p1), short(node)), f) |
|
124 | 124 | if p2 not in seen and p2 != nullid: |
|
125 | 125 | self.err(lr, _("unknown parent 2 %s of %s") % |
|
126 | 126 | (short(p2), short(node)), f) |
|
127 | 127 | except Exception as inst: |
|
128 | 128 | self.exc(lr, _("checking parents of %s") % short(node), inst, f) |
|
129 | 129 | |
|
130 | 130 | if node in seen: |
|
131 | 131 | self.err(lr, _("duplicate revision %d (%d)") % (i, seen[node]), f) |
|
132 | 132 | seen[node] = i |
|
133 | 133 | return lr |
|
134 | 134 | |
|
135 | 135 | def verify(self): |
|
136 | 136 | repo = self.repo |
|
137 | 137 | mflinkrevs = {} |
|
138 | 138 | filelinkrevs = {} |
|
139 | 139 | filenodes = {} |
|
140 | 140 | revisions = 0 |
|
141 | 141 | badrevs = self.badrevs |
|
142 | 142 | ui = repo.ui |
|
143 | 143 | cl = repo.changelog |
|
144 | 144 | |
|
145 | 145 | if not repo.url().startswith('file:'): |
|
146 | 146 | raise error.Abort(_("cannot verify bundle or remote repos")) |
|
147 | 147 | |
|
148 | 148 | if os.path.exists(repo.sjoin("journal")): |
|
149 | 149 | ui.warn(_("abandoned transaction found - run hg recover\n")) |
|
150 | 150 | |
|
151 | 151 | revlogv1 = self.revlogv1 |
|
152 | 152 | if ui.verbose or not revlogv1: |
|
153 | 153 | ui.status(_("repository uses revlog format %d\n") % |
|
154 | 154 | (revlogv1 and 1 or 0)) |
|
155 | 155 | |
|
156 | self._verifychangelog(mflinkrevs, filelinkrevs) | |
|
157 | ||
|
158 | self._verifymanifest(mflinkrevs, filenodes) | |
|
159 | ||
|
160 | self._crosscheckfiles(mflinkrevs, filelinkrevs, filenodes) | |
|
161 | ||
|
162 | totalfiles, filerevisions = self._verifyfiles(filenodes, filelinkrevs) | |
|
163 | revisions += filerevisions | |
|
164 | ||
|
165 | ui.status(_("%d files, %d changesets, %d total revisions\n") % | |
|
166 | (totalfiles, len(cl), revisions)) | |
|
167 | if self.warnings: | |
|
168 | ui.warn(_("%d warnings encountered!\n") % self.warnings) | |
|
169 | if self.fncachewarned: | |
|
170 | ui.warn(_('hint: run "hg debugrebuildfncache" to recover from ' | |
|
171 | 'corrupt fncache\n')) | |
|
172 | if self.errors: | |
|
173 | ui.warn(_("%d integrity errors encountered!\n") % self.errors) | |
|
174 | if badrevs: | |
|
175 | ui.warn(_("(first damaged changeset appears to be %d)\n") | |
|
176 | % min(badrevs)) | |
|
177 | return 1 | |
|
178 | ||
|
179 | def _verifychangelog(self, mflinkrevs, filelinkrevs): | |
|
180 | ui = self.ui | |
|
181 | repo = self.repo | |
|
182 | cl = repo.changelog | |
|
183 | ||
|
156 | 184 | ui.status(_("checking changesets\n")) |
|
157 | 185 | seen = {} |
|
158 | 186 | self.checklog(cl, "changelog", 0) |
|
159 | 187 | total = len(repo) |
|
160 | 188 | for i in repo: |
|
161 | 189 | ui.progress(_('checking'), i, total=total, unit=_('changesets')) |
|
162 | 190 | n = cl.node(i) |
|
163 | 191 | self.checkentry(cl, i, n, seen, [i], "changelog") |
|
164 | 192 | |
|
165 | 193 | try: |
|
166 | 194 | changes = cl.read(n) |
|
167 | 195 | if changes[0] != nullid: |
|
168 | 196 | mflinkrevs.setdefault(changes[0], []).append(i) |
|
169 | 197 | self.refersmf = True |
|
170 | 198 | for f in changes[3]: |
|
171 | 199 | if _validpath(repo, f): |
|
172 | 200 | filelinkrevs.setdefault(_normpath(f), []).append(i) |
|
173 | 201 | except Exception as inst: |
|
174 | 202 | self.refersmf = True |
|
175 | 203 | self.exc(i, _("unpacking changeset %s") % short(n), inst) |
|
176 | 204 | ui.progress(_('checking'), None) |
|
177 | 205 | |
|
178 | self._verifymanifest(mflinkrevs, filenodes) | |
|
179 | ||
|
180 | self._crosscheckfiles(mflinkrevs, filelinkrevs, filenodes) | |
|
181 | ||
|
182 | totalfiles, filerevisions = self._verifyfiles(filenodes, filelinkrevs) | |
|
183 | revisions += filerevisions | |
|
184 | ||
|
185 | ui.status(_("%d files, %d changesets, %d total revisions\n") % | |
|
186 | (totalfiles, len(cl), revisions)) | |
|
187 | if self.warnings: | |
|
188 | ui.warn(_("%d warnings encountered!\n") % self.warnings) | |
|
189 | if self.fncachewarned: | |
|
190 | ui.warn(_('hint: run "hg debugrebuildfncache" to recover from ' | |
|
191 | 'corrupt fncache\n')) | |
|
192 | if self.errors: | |
|
193 | ui.warn(_("%d integrity errors encountered!\n") % self.errors) | |
|
194 | if badrevs: | |
|
195 | ui.warn(_("(first damaged changeset appears to be %d)\n") | |
|
196 | % min(badrevs)) | |
|
197 | return 1 | |
|
198 | ||
|
199 | 206 | def _verifymanifest(self, mflinkrevs, filenodes): |
|
200 | 207 | repo = self.repo |
|
201 | 208 | ui = self.ui |
|
202 | 209 | mf = self.repo.manifest |
|
203 | 210 | |
|
204 | 211 | ui.status(_("checking manifests\n")) |
|
205 | 212 | seen = {} |
|
206 | 213 | if self.refersmf: |
|
207 | 214 | # Do not check manifest if there are only changelog entries with |
|
208 | 215 | # null manifests. |
|
209 | 216 | self.checklog(mf, "manifest", 0) |
|
210 | 217 | total = len(mf) |
|
211 | 218 | for i in mf: |
|
212 | 219 | ui.progress(_('checking'), i, total=total, unit=_('manifests')) |
|
213 | 220 | n = mf.node(i) |
|
214 | 221 | lr = self.checkentry(mf, i, n, seen, mflinkrevs.get(n, []), |
|
215 | 222 | "manifest") |
|
216 | 223 | if n in mflinkrevs: |
|
217 | 224 | del mflinkrevs[n] |
|
218 | 225 | else: |
|
219 | 226 | self.err(lr, _("%s not in changesets") % short(n), "manifest") |
|
220 | 227 | |
|
221 | 228 | try: |
|
222 | 229 | for f, fn in mf.readdelta(n).iteritems(): |
|
223 | 230 | if not f: |
|
224 | 231 | self.err(lr, _("file without name in manifest")) |
|
225 | 232 | elif f != "/dev/null": # ignore this in very old repos |
|
226 | 233 | if _validpath(repo, f): |
|
227 | 234 | filenodes.setdefault( |
|
228 | 235 | _normpath(f), {}).setdefault(fn, lr) |
|
229 | 236 | except Exception as inst: |
|
230 | 237 | self.exc(lr, _("reading manifest delta %s") % short(n), inst) |
|
231 | 238 | ui.progress(_('checking'), None) |
|
232 | 239 | |
|
233 | 240 | return mflinkrevs |
|
234 | 241 | |
|
235 | 242 | def _crosscheckfiles(self, mflinkrevs, filelinkrevs, filenodes): |
|
236 | 243 | repo = self.repo |
|
237 | 244 | ui = self.ui |
|
238 | 245 | ui.status(_("crosschecking files in changesets and manifests\n")) |
|
239 | 246 | |
|
240 | 247 | total = len(mflinkrevs) + len(filelinkrevs) + len(filenodes) |
|
241 | 248 | count = 0 |
|
242 | 249 | if self.havemf: |
|
243 | 250 | for c, m in sorted([(c, m) for m in mflinkrevs |
|
244 | 251 | for c in mflinkrevs[m]]): |
|
245 | 252 | count += 1 |
|
246 | 253 | if m == nullid: |
|
247 | 254 | continue |
|
248 | 255 | ui.progress(_('crosschecking'), count, total=total) |
|
249 | 256 | self.err(c, _("changeset refers to unknown manifest %s") % |
|
250 | 257 | short(m)) |
|
251 | 258 | mflinkrevs = None # del is bad here due to scope issues |
|
252 | 259 | |
|
253 | 260 | for f in sorted(filelinkrevs): |
|
254 | 261 | count += 1 |
|
255 | 262 | ui.progress(_('crosschecking'), count, total=total) |
|
256 | 263 | if f not in filenodes: |
|
257 | 264 | lr = filelinkrevs[f][0] |
|
258 | 265 | self.err(lr, _("in changeset but not in manifest"), f) |
|
259 | 266 | |
|
260 | 267 | if self.havecl: |
|
261 | 268 | for f in sorted(filenodes): |
|
262 | 269 | count += 1 |
|
263 | 270 | ui.progress(_('crosschecking'), count, total=total) |
|
264 | 271 | if f not in filelinkrevs: |
|
265 | 272 | try: |
|
266 | 273 | fl = repo.file(f) |
|
267 | 274 | lr = min([fl.linkrev(fl.rev(n)) for n in filenodes[f]]) |
|
268 | 275 | except Exception: |
|
269 | 276 | lr = None |
|
270 | 277 | self.err(lr, _("in manifest but not in changeset"), f) |
|
271 | 278 | |
|
272 | 279 | ui.progress(_('crosschecking'), None) |
|
273 | 280 | |
|
274 | 281 | def _verifyfiles(self, filenodes, filelinkrevs): |
|
275 | 282 | repo = self.repo |
|
276 | 283 | ui = self.ui |
|
277 | 284 | lrugetctx = self.lrugetctx |
|
278 | 285 | revlogv1 = self.revlogv1 |
|
279 | 286 | havemf = self.havemf |
|
280 | 287 | ui.status(_("checking files\n")) |
|
281 | 288 | |
|
282 | 289 | storefiles = set() |
|
283 | 290 | for f, f2, size in repo.store.datafiles(): |
|
284 | 291 | if not f: |
|
285 | 292 | self.err(None, _("cannot decode filename '%s'") % f2) |
|
286 | 293 | elif size > 0 or not revlogv1: |
|
287 | 294 | storefiles.add(_normpath(f)) |
|
288 | 295 | |
|
289 | 296 | files = sorted(set(filenodes) | set(filelinkrevs)) |
|
290 | 297 | total = len(files) |
|
291 | 298 | revisions = 0 |
|
292 | 299 | for i, f in enumerate(files): |
|
293 | 300 | ui.progress(_('checking'), i, item=f, total=total) |
|
294 | 301 | try: |
|
295 | 302 | linkrevs = filelinkrevs[f] |
|
296 | 303 | except KeyError: |
|
297 | 304 | # in manifest but not in changelog |
|
298 | 305 | linkrevs = [] |
|
299 | 306 | |
|
300 | 307 | if linkrevs: |
|
301 | 308 | lr = linkrevs[0] |
|
302 | 309 | else: |
|
303 | 310 | lr = None |
|
304 | 311 | |
|
305 | 312 | try: |
|
306 | 313 | fl = repo.file(f) |
|
307 | 314 | except error.RevlogError as e: |
|
308 | 315 | self.err(lr, _("broken revlog! (%s)") % e, f) |
|
309 | 316 | continue |
|
310 | 317 | |
|
311 | 318 | for ff in fl.files(): |
|
312 | 319 | try: |
|
313 | 320 | storefiles.remove(ff) |
|
314 | 321 | except KeyError: |
|
315 | 322 | self.warn(_(" warning: revlog '%s' not in fncache!") % ff) |
|
316 | 323 | self.fncachewarned = True |
|
317 | 324 | |
|
318 | 325 | self.checklog(fl, f, lr) |
|
319 | 326 | seen = {} |
|
320 | 327 | rp = None |
|
321 | 328 | for i in fl: |
|
322 | 329 | revisions += 1 |
|
323 | 330 | n = fl.node(i) |
|
324 | 331 | lr = self.checkentry(fl, i, n, seen, linkrevs, f) |
|
325 | 332 | if f in filenodes: |
|
326 | 333 | if havemf and n not in filenodes[f]: |
|
327 | 334 | self.err(lr, _("%s not in manifests") % (short(n)), f) |
|
328 | 335 | else: |
|
329 | 336 | del filenodes[f][n] |
|
330 | 337 | |
|
331 | 338 | # verify contents |
|
332 | 339 | try: |
|
333 | 340 | l = len(fl.read(n)) |
|
334 | 341 | rp = fl.renamed(n) |
|
335 | 342 | if l != fl.size(i): |
|
336 | 343 | if len(fl.revision(n)) != fl.size(i): |
|
337 | 344 | self.err(lr, _("unpacked size is %s, %s expected") % |
|
338 | 345 | (l, fl.size(i)), f) |
|
339 | 346 | except error.CensoredNodeError: |
|
340 | 347 | # experimental config: censor.policy |
|
341 | 348 | if ui.config("censor", "policy", "abort") == "abort": |
|
342 | 349 | self.err(lr, _("censored file data"), f) |
|
343 | 350 | except Exception as inst: |
|
344 | 351 | self.exc(lr, _("unpacking %s") % short(n), inst, f) |
|
345 | 352 | |
|
346 | 353 | # check renames |
|
347 | 354 | try: |
|
348 | 355 | if rp: |
|
349 | 356 | if lr is not None and ui.verbose: |
|
350 | 357 | ctx = lrugetctx(lr) |
|
351 | 358 | found = False |
|
352 | 359 | for pctx in ctx.parents(): |
|
353 | 360 | if rp[0] in pctx: |
|
354 | 361 | found = True |
|
355 | 362 | break |
|
356 | 363 | if not found: |
|
357 | 364 | self.warn(_("warning: copy source of '%s' not" |
|
358 | 365 | " in parents of %s") % (f, ctx)) |
|
359 | 366 | fl2 = repo.file(rp[0]) |
|
360 | 367 | if not len(fl2): |
|
361 | 368 | self.err(lr, _("empty or missing copy source " |
|
362 | 369 | "revlog %s:%s") % (rp[0], short(rp[1])), f) |
|
363 | 370 | elif rp[1] == nullid: |
|
364 | 371 | ui.note(_("warning: %s@%s: copy source" |
|
365 | 372 | " revision is nullid %s:%s\n") |
|
366 | 373 | % (f, lr, rp[0], short(rp[1]))) |
|
367 | 374 | else: |
|
368 | 375 | fl2.rev(rp[1]) |
|
369 | 376 | except Exception as inst: |
|
370 | 377 | self.exc(lr, _("checking rename of %s") % short(n), inst, f) |
|
371 | 378 | |
|
372 | 379 | # cross-check |
|
373 | 380 | if f in filenodes: |
|
374 | 381 | fns = [(lr, n) for n, lr in filenodes[f].iteritems()] |
|
375 | 382 | for lr, node in sorted(fns): |
|
376 | 383 | self.err(lr, _("%s in manifests not found") % short(node), |
|
377 | 384 | f) |
|
378 | 385 | ui.progress(_('checking'), None) |
|
379 | 386 | |
|
380 | 387 | for f in storefiles: |
|
381 | 388 | self.warn(_("warning: orphan revlog '%s'") % f) |
|
382 | 389 | |
|
383 | 390 | return len(files), revisions |
General Comments 0
You need to be logged in to leave comments.
Login now