Show More
@@ -1,274 +1,281 b'' | |||||
1 | # verify.py - repository integrity checking for Mercurial |
|
1 | # verify.py - repository integrity checking for Mercurial | |
2 | # |
|
2 | # | |
3 | # Copyright 2006, 2007 Matt Mackall <mpm@selenic.com> |
|
3 | # Copyright 2006, 2007 Matt Mackall <mpm@selenic.com> | |
4 | # |
|
4 | # | |
5 | # This software may be used and distributed according to the terms |
|
5 | # This software may be used and distributed according to the terms | |
6 | # of the GNU General Public License, incorporated herein by reference. |
|
6 | # of the GNU General Public License, incorporated herein by reference. | |
7 |
|
7 | |||
8 | from node import nullid, short |
|
8 | from node import nullid, short | |
9 | from i18n import _ |
|
9 | from i18n import _ | |
10 | import revlog |
|
10 | import revlog | |
11 |
|
11 | |||
12 | def verify(repo): |
|
12 | def verify(repo): | |
13 | lock = repo.lock() |
|
13 | lock = repo.lock() | |
14 | try: |
|
14 | try: | |
15 | return _verify(repo) |
|
15 | return _verify(repo) | |
16 | finally: |
|
16 | finally: | |
17 | del lock |
|
17 | del lock | |
18 |
|
18 | |||
19 | def _verify(repo): |
|
19 | def _verify(repo): | |
20 | filelinkrevs = {} |
|
20 | filelinkrevs = {} | |
21 | filenodes = {} |
|
21 | filenodes = {} | |
22 | changesets = revisions = files = 0 |
|
22 | changesets = revisions = files = 0 | |
23 | firstbad = [None] |
|
23 | firstbad = [None] | |
24 | errors = [0] |
|
24 | errors = [0] | |
25 | warnings = [0] |
|
25 | warnings = [0] | |
26 | neededmanifests = {} |
|
26 | neededmanifests = {} | |
27 |
|
27 | |||
28 | def err(linkrev, msg, filename=None): |
|
28 | def err(linkrev, msg, filename=None): | |
29 | if linkrev != None: |
|
29 | if linkrev != None: | |
30 | if firstbad[0] != None: |
|
30 | if firstbad[0] != None: | |
31 | firstbad[0] = min(firstbad[0], linkrev) |
|
31 | firstbad[0] = min(firstbad[0], linkrev) | |
32 | else: |
|
32 | else: | |
33 | firstbad[0] = linkrev |
|
33 | firstbad[0] = linkrev | |
34 | else: |
|
34 | else: | |
35 | linkrev = "?" |
|
35 | linkrev = "?" | |
36 | msg = "%s: %s" % (linkrev, msg) |
|
36 | msg = "%s: %s" % (linkrev, msg) | |
37 | if filename: |
|
37 | if filename: | |
38 | msg = "%s@%s" % (filename, msg) |
|
38 | msg = "%s@%s" % (filename, msg) | |
39 | repo.ui.warn(" " + msg + "\n") |
|
39 | repo.ui.warn(" " + msg + "\n") | |
40 | errors[0] += 1 |
|
40 | errors[0] += 1 | |
41 |
|
41 | |||
42 | def warn(msg): |
|
42 | def warn(msg): | |
43 | repo.ui.warn(msg + "\n") |
|
43 | repo.ui.warn(msg + "\n") | |
44 | warnings[0] += 1 |
|
44 | warnings[0] += 1 | |
45 |
|
45 | |||
46 | def checksize(obj, name): |
|
46 | def checksize(obj, name): | |
47 | d = obj.checksize() |
|
47 | d = obj.checksize() | |
48 | if d[0]: |
|
48 | if d[0]: | |
49 | err(None, _("data length off by %d bytes") % d[0], name) |
|
49 | err(None, _("data length off by %d bytes") % d[0], name) | |
50 | if d[1]: |
|
50 | if d[1]: | |
51 | err(None, _("index contains %d extra bytes") % d[1], name) |
|
51 | err(None, _("index contains %d extra bytes") % d[1], name) | |
52 |
|
52 | |||
53 | def checkversion(obj, name): |
|
53 | def checkversion(obj, name): | |
54 | if obj.version != revlog.REVLOGV0: |
|
54 | if obj.version != revlog.REVLOGV0: | |
55 | if not revlogv1: |
|
55 | if not revlogv1: | |
56 | warn(_("warning: `%s' uses revlog format 1") % name) |
|
56 | warn(_("warning: `%s' uses revlog format 1") % name) | |
57 | elif revlogv1: |
|
57 | elif revlogv1: | |
58 | warn(_("warning: `%s' uses revlog format 0") % name) |
|
58 | warn(_("warning: `%s' uses revlog format 0") % name) | |
59 |
|
59 | |||
60 | revlogv1 = repo.changelog.version != revlog.REVLOGV0 |
|
60 | revlogv1 = repo.changelog.version != revlog.REVLOGV0 | |
61 | if repo.ui.verbose or not revlogv1: |
|
61 | if repo.ui.verbose or not revlogv1: | |
62 | repo.ui.status(_("repository uses revlog format %d\n") % |
|
62 | repo.ui.status(_("repository uses revlog format %d\n") % | |
63 | (revlogv1 and 1 or 0)) |
|
63 | (revlogv1 and 1 or 0)) | |
64 |
|
64 | |||
65 | havecl = havemf = 1 |
|
65 | havecl = havemf = 1 | |
66 | seen = {} |
|
66 | seen = {} | |
67 | repo.ui.status(_("checking changesets\n")) |
|
67 | repo.ui.status(_("checking changesets\n")) | |
68 | if repo.changelog.count() == 0 and repo.manifest.count() > 1: |
|
68 | if repo.changelog.count() == 0 and repo.manifest.count() > 1: | |
69 | havecl = 0 |
|
69 | havecl = 0 | |
70 | err(0, _("empty or missing 00changelog.i")) |
|
70 | err(0, _("empty or missing 00changelog.i")) | |
71 | else: |
|
71 | else: | |
72 | checksize(repo.changelog, "changelog") |
|
72 | checksize(repo.changelog, "changelog") | |
73 |
|
73 | |||
74 | for i in xrange(repo.changelog.count()): |
|
74 | for i in xrange(repo.changelog.count()): | |
75 | changesets += 1 |
|
75 | changesets += 1 | |
76 | n = repo.changelog.node(i) |
|
76 | n = repo.changelog.node(i) | |
77 | l = repo.changelog.linkrev(n) |
|
77 | l = repo.changelog.linkrev(n) | |
78 | if l != i: |
|
78 | if l != i: | |
79 | err(i, _("incorrect link (%d) for changeset") %(l)) |
|
79 | err(i, _("incorrect link (%d) for changeset") %(l)) | |
80 | if n in seen: |
|
80 | if n in seen: | |
81 | err(i, _("duplicates changeset at revision %d") % seen[n]) |
|
81 | err(i, _("duplicates changeset at revision %d") % seen[n]) | |
82 | seen[n] = i |
|
82 | seen[n] = i | |
83 |
|
83 | |||
84 | for p in repo.changelog.parents(n): |
|
84 | for p in repo.changelog.parents(n): | |
85 | if p not in repo.changelog.nodemap: |
|
85 | if p not in repo.changelog.nodemap: | |
86 | err(i, _("changeset has unknown parent %s") % short(p)) |
|
86 | err(i, _("changeset has unknown parent %s") % short(p)) | |
87 | try: |
|
87 | try: | |
88 | changes = repo.changelog.read(n) |
|
88 | changes = repo.changelog.read(n) | |
89 | except KeyboardInterrupt: |
|
89 | except KeyboardInterrupt: | |
90 | repo.ui.warn(_("interrupted")) |
|
90 | repo.ui.warn(_("interrupted")) | |
91 | raise |
|
91 | raise | |
92 | except Exception, inst: |
|
92 | except Exception, inst: | |
93 | err(i, _("unpacking changeset: %s") % inst) |
|
93 | err(i, _("unpacking changeset: %s") % inst) | |
94 | continue |
|
94 | continue | |
95 |
|
95 | |||
96 | if changes[0] not in neededmanifests: |
|
96 | if changes[0] not in neededmanifests: | |
97 | neededmanifests[changes[0]] = i |
|
97 | neededmanifests[changes[0]] = i | |
98 |
|
98 | |||
99 | for f in changes[3]: |
|
99 | for f in changes[3]: | |
100 | filelinkrevs.setdefault(f, []).append(i) |
|
100 | filelinkrevs.setdefault(f, []).append(i) | |
101 |
|
101 | |||
102 | seen = {} |
|
102 | seen = {} | |
103 | repo.ui.status(_("checking manifests\n")) |
|
103 | repo.ui.status(_("checking manifests\n")) | |
104 | if repo.changelog.count() > 0 and repo.manifest.count() == 0: |
|
104 | if repo.changelog.count() > 0 and repo.manifest.count() == 0: | |
105 | havemf = 0 |
|
105 | havemf = 0 | |
106 | err(0, _("empty or missing 00manifest.i")) |
|
106 | err(0, _("empty or missing 00manifest.i")) | |
107 | else: |
|
107 | else: | |
108 | checkversion(repo.manifest, "manifest") |
|
108 | checkversion(repo.manifest, "manifest") | |
109 | checksize(repo.manifest, "manifest") |
|
109 | checksize(repo.manifest, "manifest") | |
110 |
|
110 | |||
111 | for i in xrange(repo.manifest.count()): |
|
111 | for i in xrange(repo.manifest.count()): | |
112 | n = repo.manifest.node(i) |
|
112 | n = repo.manifest.node(i) | |
113 | l = repo.manifest.linkrev(n) |
|
113 | l = repo.manifest.linkrev(n) | |
114 |
|
114 | |||
115 | if l < 0 or (havecl and l >= repo.changelog.count()): |
|
115 | if l < 0 or (havecl and l >= repo.changelog.count()): | |
116 | err(None, _("bad link (%d) at manifest revision %d") % (l, i)) |
|
116 | err(None, _("bad link (%d) at manifest revision %d") % (l, i)) | |
117 |
|
117 | |||
118 | if n in neededmanifests: |
|
118 | if n in neededmanifests: | |
119 | del neededmanifests[n] |
|
119 | del neededmanifests[n] | |
120 |
|
120 | |||
121 | if n in seen: |
|
121 | if n in seen: | |
122 | err(l, _("duplicates manifest from %d") % seen[n]) |
|
122 | err(l, _("duplicates manifest from %d") % seen[n]) | |
123 |
|
123 | |||
124 | seen[n] = l |
|
124 | seen[n] = l | |
125 |
|
125 | |||
126 | for p in repo.manifest.parents(n): |
|
126 | for p in repo.manifest.parents(n): | |
127 | if p not in repo.manifest.nodemap: |
|
127 | if p not in repo.manifest.nodemap: | |
128 | err(l, _("manifest has unknown parent %s") % short(p)) |
|
128 | err(l, _("manifest has unknown parent %s") % short(p)) | |
129 |
|
129 | |||
130 | try: |
|
130 | try: | |
131 | for f, fn in repo.manifest.readdelta(n).iteritems(): |
|
131 | for f, fn in repo.manifest.readdelta(n).iteritems(): | |
132 | fns = filenodes.setdefault(f, {}) |
|
132 | fns = filenodes.setdefault(f, {}) | |
133 | if fn not in fns: |
|
133 | if fn not in fns: | |
134 | fns[fn] = n |
|
134 | fns[fn] = n | |
135 | except KeyboardInterrupt: |
|
135 | except KeyboardInterrupt: | |
136 | repo.ui.warn(_("interrupted")) |
|
136 | repo.ui.warn(_("interrupted")) | |
137 | raise |
|
137 | raise | |
138 | except Exception, inst: |
|
138 | except Exception, inst: | |
139 | err(l, _("reading manifest delta: %s") % inst) |
|
139 | err(l, _("reading manifest delta: %s") % inst) | |
140 | continue |
|
140 | continue | |
141 |
|
141 | |||
142 | repo.ui.status(_("crosschecking files in changesets and manifests\n")) |
|
142 | repo.ui.status(_("crosschecking files in changesets and manifests\n")) | |
143 |
|
143 | |||
144 | if havemf > 0: |
|
144 | if havemf > 0: | |
145 | nm = [(c, m) for m, c in neededmanifests.items()] |
|
145 | nm = [(c, m) for m, c in neededmanifests.items()] | |
146 | nm.sort() |
|
146 | nm.sort() | |
147 | for c, m in nm: |
|
147 | for c, m in nm: | |
148 | err(c, _("changeset refers to unknown manifest %s") % short(m)) |
|
148 | err(c, _("changeset refers to unknown manifest %s") % short(m)) | |
149 | del neededmanifests, nm |
|
149 | del neededmanifests, nm | |
150 |
|
150 | |||
151 | if havecl: |
|
151 | if havecl: | |
152 | fl = filenodes.keys() |
|
152 | fl = filenodes.keys() | |
153 | fl.sort() |
|
153 | fl.sort() | |
154 | for f in fl: |
|
154 | for f in fl: | |
155 | if f not in filelinkrevs: |
|
155 | if f not in filelinkrevs: | |
156 | lrs = [repo.manifest.linkrev(n) for n in filenodes[f]] |
|
156 | lrs = [repo.manifest.linkrev(n) for n in filenodes[f]] | |
157 | lrs.sort() |
|
157 | lrs.sort() | |
158 | err(lrs[0], _("in manifest but not in changeset"), f) |
|
158 | err(lrs[0], _("in manifest but not in changeset"), f) | |
159 | del fl |
|
159 | del fl | |
160 |
|
160 | |||
161 | if havemf: |
|
161 | if havemf: | |
162 | fl = filelinkrevs.keys() |
|
162 | fl = filelinkrevs.keys() | |
163 | fl.sort() |
|
163 | fl.sort() | |
164 | for f in fl: |
|
164 | for f in fl: | |
165 | if f not in filenodes: |
|
165 | if f not in filenodes: | |
166 | lr = filelinkrevs[f][0] |
|
166 | lr = filelinkrevs[f][0] | |
167 | err(lr, _("in changeset but not in manifest"), f) |
|
167 | err(lr, _("in changeset but not in manifest"), f) | |
168 | del fl |
|
168 | del fl | |
169 |
|
169 | |||
170 | repo.ui.status(_("checking files\n")) |
|
170 | repo.ui.status(_("checking files\n")) | |
171 | ff = dict.fromkeys(filenodes.keys() + filelinkrevs.keys()).keys() |
|
171 | ff = dict.fromkeys(filenodes.keys() + filelinkrevs.keys()).keys() | |
172 | ff.sort() |
|
172 | ff.sort() | |
173 | for f in ff: |
|
173 | for f in ff: | |
174 | if f == "/dev/null": |
|
174 | if f == "/dev/null": | |
175 | continue |
|
175 | continue | |
176 | files += 1 |
|
176 | files += 1 | |
177 | if not f: |
|
177 | if not f: | |
178 | lr = filelinkrevs[f][0] |
|
178 | lr = filelinkrevs[f][0] | |
179 | err(lr, _("file without name in manifest")) |
|
179 | err(lr, _("file without name in manifest")) | |
180 | continue |
|
180 | continue | |
181 | fl = repo.file(f) |
|
181 | fl = repo.file(f) | |
182 | checkversion(fl, f) |
|
182 | checkversion(fl, f) | |
183 | checksize(fl, f) |
|
183 | checksize(fl, f) | |
184 |
|
184 | |||
185 | if fl.count() == 0: |
|
185 | if fl.count() == 0: | |
186 | err(filelinkrevs[f][0], _("empty or missing revlog"), f) |
|
186 | err(filelinkrevs[f][0], _("empty or missing revlog"), f) | |
187 | continue |
|
187 | continue | |
188 |
|
188 | |||
189 | seen = {} |
|
189 | seen = {} | |
190 | nodes = {nullid: 1} |
|
190 | nodes = {nullid: 1} | |
191 | for i in xrange(fl.count()): |
|
191 | for i in xrange(fl.count()): | |
192 | revisions += 1 |
|
192 | revisions += 1 | |
193 | n = fl.node(i) |
|
193 | n = fl.node(i) | |
194 | flr = fl.linkrev(n) |
|
194 | flr = fl.linkrev(n) | |
195 |
|
195 | |||
196 | if flr < 0 or (havecl and flr not in filelinkrevs.get(f, [])): |
|
196 | if flr < 0 or (havecl and flr not in filelinkrevs.get(f, [])): | |
197 | if flr < 0 or flr >= repo.changelog.count(): |
|
197 | if flr < 0 or flr >= repo.changelog.count(): | |
198 | err(None, _("rev %d point to nonexistent changeset %d") |
|
198 | err(None, _("rev %d point to nonexistent changeset %d") | |
199 | % (i, flr), f) |
|
199 | % (i, flr), f) | |
200 | else: |
|
200 | else: | |
201 | err(None, _("rev %d points to unexpected changeset %d") |
|
201 | err(None, _("rev %d points to unexpected changeset %d") | |
202 | % (i, flr), f) |
|
202 | % (i, flr), f) | |
203 | if f in filelinkrevs: |
|
203 | if f in filelinkrevs: | |
204 | warn(_(" (expected %s)") % filelinkrevs[f][0]) |
|
204 | warn(_(" (expected %s)") % filelinkrevs[f][0]) | |
205 | flr = None # can't be trusted |
|
205 | flr = None # can't be trusted | |
206 | else: |
|
206 | else: | |
207 | if havecl: |
|
207 | if havecl: | |
208 | filelinkrevs[f].remove(flr) |
|
208 | filelinkrevs[f].remove(flr) | |
209 |
|
209 | |||
210 | if n in seen: |
|
210 | if n in seen: | |
211 | err(flr, _("duplicate revision %d") % i, f) |
|
211 | err(flr, _("duplicate revision %d") % i, f) | |
212 | if f in filenodes: |
|
212 | if f in filenodes: | |
213 | if havemf and n not in filenodes[f]: |
|
213 | if havemf and n not in filenodes[f]: | |
214 | err(flr, _("%s not in manifests") % (short(n)), f) |
|
214 | err(flr, _("%s not in manifests") % (short(n)), f) | |
215 | else: |
|
215 | else: | |
216 | del filenodes[f][n] |
|
216 | del filenodes[f][n] | |
217 |
|
217 | |||
218 | # verify contents |
|
218 | # verify contents | |
219 | try: |
|
219 | try: | |
220 | t = fl.read(n) |
|
220 | t = fl.read(n) | |
221 | except KeyboardInterrupt: |
|
221 | except KeyboardInterrupt: | |
222 | repo.ui.warn(_("interrupted")) |
|
222 | repo.ui.warn(_("interrupted")) | |
223 | raise |
|
223 | raise | |
224 | except Exception, inst: |
|
224 | except Exception, inst: | |
225 | err(flr, _("unpacking %s: %s") % (short(n), inst), f) |
|
225 | err(flr, _("unpacking %s: %s") % (short(n), inst), f) | |
226 |
|
226 | |||
227 | # verify parents |
|
227 | # verify parents | |
228 | try: |
|
228 | try: | |
229 | (p1, p2) = fl.parents(n) |
|
229 | (p1, p2) = fl.parents(n) | |
230 | if p1 not in nodes: |
|
230 | if p1 not in nodes: | |
231 | err(flr, _("unknown parent 1 %s of %s") % |
|
231 | err(flr, _("unknown parent 1 %s of %s") % | |
232 | (short(p1), short(n)), f) |
|
232 | (short(p1), short(n)), f) | |
233 | if p2 not in nodes: |
|
233 | if p2 not in nodes: | |
234 | err(flr, _("unknown parent 2 %s of %s") % |
|
234 | err(flr, _("unknown parent 2 %s of %s") % | |
235 | (short(p2), short(p1)), f) |
|
235 | (short(p2), short(p1)), f) | |
236 | except KeyboardInterrupt: |
|
236 | except KeyboardInterrupt: | |
237 | repo.ui.warn(_("interrupted")) |
|
237 | repo.ui.warn(_("interrupted")) | |
238 | raise |
|
238 | raise | |
239 | except Exception, inst: |
|
239 | except Exception, inst: | |
240 | err(flr, _("checking parents of %s: %s") % (short(n), inst), f) |
|
240 | err(flr, _("checking parents of %s: %s") % (short(n), inst), f) | |
241 | nodes[n] = 1 |
|
241 | nodes[n] = 1 | |
242 |
|
242 | |||
243 | # check renames |
|
243 | # check renames | |
244 | try: |
|
244 | try: | |
245 | rp = fl.renamed(n) |
|
245 | rp = fl.renamed(n) | |
246 | if rp: |
|
246 | if rp: | |
247 | fl2 = repo.file(rp[0]) |
|
247 | fl2 = repo.file(rp[0]) | |
|
248 | if fl2.count() == 0: | |||
|
249 | err(flr, _("empty or missing copy source revlog %s:%s") | |||
|
250 | % (rp[0], short(rp[1])), f) | |||
|
251 | elif rp[1] == nullid: | |||
|
252 | err(flr, _("copy source revision is nullid %s:%s") | |||
|
253 | % (rp[0], short(rp[1])), f) | |||
|
254 | else: | |||
248 | rev = fl2.rev(rp[1]) |
|
255 | rev = fl2.rev(rp[1]) | |
249 | except KeyboardInterrupt: |
|
256 | except KeyboardInterrupt: | |
250 | repo.ui.warn(_("interrupted")) |
|
257 | repo.ui.warn(_("interrupted")) | |
251 | raise |
|
258 | raise | |
252 | except Exception, inst: |
|
259 | except Exception, inst: | |
253 | err(flr, _("checking rename of %s: %s") % |
|
260 | err(flr, _("checking rename of %s: %s") % | |
254 | (short(n), inst), f) |
|
261 | (short(n), inst), f) | |
255 |
|
262 | |||
256 | # cross-check |
|
263 | # cross-check | |
257 | if f in filenodes: |
|
264 | if f in filenodes: | |
258 | fns = [(repo.manifest.linkrev(filenodes[f][n]), n) |
|
265 | fns = [(repo.manifest.linkrev(filenodes[f][n]), n) | |
259 | for n in filenodes[f]] |
|
266 | for n in filenodes[f]] | |
260 | fns.sort() |
|
267 | fns.sort() | |
261 | for lr, node in fns: |
|
268 | for lr, node in fns: | |
262 | err(lr, _("%s in manifests not found") % short(node), f) |
|
269 | err(lr, _("%s in manifests not found") % short(node), f) | |
263 |
|
270 | |||
264 | repo.ui.status(_("%d files, %d changesets, %d total revisions\n") % |
|
271 | repo.ui.status(_("%d files, %d changesets, %d total revisions\n") % | |
265 | (files, changesets, revisions)) |
|
272 | (files, changesets, revisions)) | |
266 |
|
273 | |||
267 | if warnings[0]: |
|
274 | if warnings[0]: | |
268 | repo.ui.warn(_("%d warnings encountered!\n") % warnings[0]) |
|
275 | repo.ui.warn(_("%d warnings encountered!\n") % warnings[0]) | |
269 | if errors[0]: |
|
276 | if errors[0]: | |
270 | repo.ui.warn(_("%d integrity errors encountered!\n") % errors[0]) |
|
277 | repo.ui.warn(_("%d integrity errors encountered!\n") % errors[0]) | |
271 | if firstbad[0]: |
|
278 | if firstbad[0]: | |
272 | repo.ui.warn(_("(first damaged changeset appears to be %d)\n") |
|
279 | repo.ui.warn(_("(first damaged changeset appears to be %d)\n") | |
273 | % firstbad[0]) |
|
280 | % firstbad[0]) | |
274 | return 1 |
|
281 | return 1 |
General Comments 0
You need to be logged in to leave comments.
Login now