Show More
@@ -1,525 +1,525 b'' | |||||
1 | # branchmap.py - logic to computes, maintain and stores branchmap for local repo |
|
1 | # branchmap.py - logic to computes, maintain and stores branchmap for local repo | |
2 | # |
|
2 | # | |
3 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.com> |
|
3 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.com> | |
4 | # |
|
4 | # | |
5 | # This software may be used and distributed according to the terms of the |
|
5 | # This software may be used and distributed according to the terms of the | |
6 | # GNU General Public License version 2 or any later version. |
|
6 | # GNU General Public License version 2 or any later version. | |
7 |
|
7 | |||
8 | from __future__ import absolute_import |
|
8 | from __future__ import absolute_import | |
9 |
|
9 | |||
10 | import struct |
|
10 | import struct | |
11 |
|
11 | |||
12 | from .node import ( |
|
12 | from .node import ( | |
13 | bin, |
|
13 | bin, | |
14 | hex, |
|
14 | hex, | |
15 | nullid, |
|
15 | nullid, | |
16 | nullrev, |
|
16 | nullrev, | |
17 | ) |
|
17 | ) | |
18 | from . import ( |
|
18 | from . import ( | |
19 | encoding, |
|
19 | encoding, | |
20 | error, |
|
20 | error, | |
21 | pycompat, |
|
21 | pycompat, | |
22 | scmutil, |
|
22 | scmutil, | |
23 | util, |
|
23 | util, | |
24 | ) |
|
24 | ) | |
25 |
|
25 | |||
26 | calcsize = struct.calcsize |
|
26 | calcsize = struct.calcsize | |
27 | pack_into = struct.pack_into |
|
27 | pack_into = struct.pack_into | |
28 | unpack_from = struct.unpack_from |
|
28 | unpack_from = struct.unpack_from | |
29 |
|
29 | |||
30 | def _filename(repo): |
|
30 | def _filename(repo): | |
31 | """name of a branchcache file for a given repo or repoview""" |
|
31 | """name of a branchcache file for a given repo or repoview""" | |
32 | filename = "branch2" |
|
32 | filename = "branch2" | |
33 | if repo.filtername: |
|
33 | if repo.filtername: | |
34 | filename = '%s-%s' % (filename, repo.filtername) |
|
34 | filename = '%s-%s' % (filename, repo.filtername) | |
35 | return filename |
|
35 | return filename | |
36 |
|
36 | |||
37 | def read(repo): |
|
37 | def read(repo): | |
38 | try: |
|
38 | try: | |
39 | f = repo.cachevfs(_filename(repo)) |
|
39 | f = repo.cachevfs(_filename(repo)) | |
40 | lines = f.read().split('\n') |
|
40 | lines = f.read().split('\n') | |
41 | f.close() |
|
41 | f.close() | |
42 | except (IOError, OSError): |
|
42 | except (IOError, OSError): | |
43 | return None |
|
43 | return None | |
44 |
|
44 | |||
45 | try: |
|
45 | try: | |
46 | cachekey = lines.pop(0).split(" ", 2) |
|
46 | cachekey = lines.pop(0).split(" ", 2) | |
47 | last, lrev = cachekey[:2] |
|
47 | last, lrev = cachekey[:2] | |
48 | last, lrev = bin(last), int(lrev) |
|
48 | last, lrev = bin(last), int(lrev) | |
49 | filteredhash = None |
|
49 | filteredhash = None | |
50 | if len(cachekey) > 2: |
|
50 | if len(cachekey) > 2: | |
51 | filteredhash = bin(cachekey[2]) |
|
51 | filteredhash = bin(cachekey[2]) | |
52 | partial = branchcache(tipnode=last, tiprev=lrev, |
|
52 | partial = branchcache(tipnode=last, tiprev=lrev, | |
53 | filteredhash=filteredhash) |
|
53 | filteredhash=filteredhash) | |
54 | if not partial.validfor(repo): |
|
54 | if not partial.validfor(repo): | |
55 | # invalidate the cache |
|
55 | # invalidate the cache | |
56 | raise ValueError(r'tip differs') |
|
56 | raise ValueError(r'tip differs') | |
57 | cl = repo.changelog |
|
57 | cl = repo.changelog | |
58 | for l in lines: |
|
58 | for l in lines: | |
59 | if not l: |
|
59 | if not l: | |
60 | continue |
|
60 | continue | |
61 | node, state, label = l.split(" ", 2) |
|
61 | node, state, label = l.split(" ", 2) | |
62 | if state not in 'oc': |
|
62 | if state not in 'oc': | |
63 | raise ValueError(r'invalid branch state') |
|
63 | raise ValueError(r'invalid branch state') | |
64 | label = encoding.tolocal(label.strip()) |
|
64 | label = encoding.tolocal(label.strip()) | |
65 | node = bin(node) |
|
65 | node = bin(node) | |
66 | if not cl.hasnode(node): |
|
66 | if not cl.hasnode(node): | |
67 | raise ValueError( |
|
67 | raise ValueError( | |
68 | r'node %s does not exist' % pycompat.sysstr(hex(node))) |
|
68 | r'node %s does not exist' % pycompat.sysstr(hex(node))) | |
69 | partial.setdefault(label, []).append(node) |
|
69 | partial.setdefault(label, []).append(node) | |
70 | if state == 'c': |
|
70 | if state == 'c': | |
71 | partial._closednodes.add(node) |
|
71 | partial._closednodes.add(node) | |
72 | except Exception as inst: |
|
72 | except Exception as inst: | |
73 | if repo.ui.debugflag: |
|
73 | if repo.ui.debugflag: | |
74 | msg = 'invalid branchheads cache' |
|
74 | msg = 'invalid branchheads cache' | |
75 | if repo.filtername is not None: |
|
75 | if repo.filtername is not None: | |
76 | msg += ' (%s)' % repo.filtername |
|
76 | msg += ' (%s)' % repo.filtername | |
77 | msg += ': %s\n' |
|
77 | msg += ': %s\n' | |
78 | repo.ui.debug(msg % pycompat.bytestr(inst)) |
|
78 | repo.ui.debug(msg % pycompat.bytestr(inst)) | |
79 | partial = None |
|
79 | partial = None | |
80 | return partial |
|
80 | return partial | |
81 |
|
81 | |||
82 | ### Nearest subset relation |
|
82 | ### Nearest subset relation | |
83 | # Nearest subset of filter X is a filter Y so that: |
|
83 | # Nearest subset of filter X is a filter Y so that: | |
84 | # * Y is included in X, |
|
84 | # * Y is included in X, | |
85 | # * X - Y is as small as possible. |
|
85 | # * X - Y is as small as possible. | |
86 | # This create and ordering used for branchmap purpose. |
|
86 | # This create and ordering used for branchmap purpose. | |
87 | # the ordering may be partial |
|
87 | # the ordering may be partial | |
88 | subsettable = {None: 'visible', |
|
88 | subsettable = {None: 'visible', | |
89 | 'visible-hidden': 'visible', |
|
89 | 'visible-hidden': 'visible', | |
90 | 'visible': 'served', |
|
90 | 'visible': 'served', | |
91 | 'served': 'immutable', |
|
91 | 'served': 'immutable', | |
92 | 'immutable': 'base'} |
|
92 | 'immutable': 'base'} | |
93 |
|
93 | |||
94 | def updatecache(repo): |
|
94 | def updatecache(repo): | |
95 | cl = repo.changelog |
|
95 | cl = repo.changelog | |
96 | filtername = repo.filtername |
|
96 | filtername = repo.filtername | |
97 | partial = repo._branchcaches.get(filtername) |
|
97 | partial = repo._branchcaches.get(filtername) | |
98 |
|
98 | |||
99 | revs = [] |
|
99 | revs = [] | |
100 | if partial is None or not partial.validfor(repo): |
|
100 | if partial is None or not partial.validfor(repo): | |
101 | partial = read(repo) |
|
101 | partial = read(repo) | |
102 | if partial is None: |
|
102 | if partial is None: | |
103 | subsetname = subsettable.get(filtername) |
|
103 | subsetname = subsettable.get(filtername) | |
104 | if subsetname is None: |
|
104 | if subsetname is None: | |
105 | partial = branchcache() |
|
105 | partial = branchcache() | |
106 | else: |
|
106 | else: | |
107 | subset = repo.filtered(subsetname) |
|
107 | subset = repo.filtered(subsetname) | |
108 | partial = subset.branchmap().copy() |
|
108 | partial = subset.branchmap().copy() | |
109 | extrarevs = subset.changelog.filteredrevs - cl.filteredrevs |
|
109 | extrarevs = subset.changelog.filteredrevs - cl.filteredrevs | |
110 | revs.extend(r for r in extrarevs if r <= partial.tiprev) |
|
110 | revs.extend(r for r in extrarevs if r <= partial.tiprev) | |
111 | revs.extend(cl.revs(start=partial.tiprev + 1)) |
|
111 | revs.extend(cl.revs(start=partial.tiprev + 1)) | |
112 | if revs: |
|
112 | if revs: | |
113 | partial.update(repo, revs) |
|
113 | partial.update(repo, revs) | |
114 | partial.write(repo) |
|
114 | partial.write(repo) | |
115 |
|
115 | |||
116 | assert partial.validfor(repo), filtername |
|
116 | assert partial.validfor(repo), filtername | |
117 | repo._branchcaches[repo.filtername] = partial |
|
117 | repo._branchcaches[repo.filtername] = partial | |
118 |
|
118 | |||
119 | def replacecache(repo, bm): |
|
119 | def replacecache(repo, bm): | |
120 | """Replace the branchmap cache for a repo with a branch mapping. |
|
120 | """Replace the branchmap cache for a repo with a branch mapping. | |
121 |
|
121 | |||
122 | This is likely only called during clone with a branch map from a remote. |
|
122 | This is likely only called during clone with a branch map from a remote. | |
123 | """ |
|
123 | """ | |
124 | rbheads = [] |
|
124 | rbheads = [] | |
125 | closed = [] |
|
125 | closed = [] | |
126 | for bheads in bm.itervalues(): |
|
126 | for bheads in bm.itervalues(): | |
127 | rbheads.extend(bheads) |
|
127 | rbheads.extend(bheads) | |
128 | for h in bheads: |
|
128 | for h in bheads: | |
129 | r = repo.changelog.rev(h) |
|
129 | r = repo.changelog.rev(h) | |
130 | b, c = repo.changelog.branchinfo(r) |
|
130 | b, c = repo.changelog.branchinfo(r) | |
131 | if c: |
|
131 | if c: | |
132 | closed.append(h) |
|
132 | closed.append(h) | |
133 |
|
133 | |||
134 | if rbheads: |
|
134 | if rbheads: | |
135 | rtiprev = max((int(repo.changelog.rev(node)) |
|
135 | rtiprev = max((int(repo.changelog.rev(node)) | |
136 | for node in rbheads)) |
|
136 | for node in rbheads)) | |
137 | cache = branchcache(bm, |
|
137 | cache = branchcache(bm, | |
138 | repo[rtiprev].node(), |
|
138 | repo[rtiprev].node(), | |
139 | rtiprev, |
|
139 | rtiprev, | |
140 | closednodes=closed) |
|
140 | closednodes=closed) | |
141 |
|
141 | |||
142 | # Try to stick it as low as possible |
|
142 | # Try to stick it as low as possible | |
143 | # filter above served are unlikely to be fetch from a clone |
|
143 | # filter above served are unlikely to be fetch from a clone | |
144 | for candidate in ('base', 'immutable', 'served'): |
|
144 | for candidate in ('base', 'immutable', 'served'): | |
145 | rview = repo.filtered(candidate) |
|
145 | rview = repo.filtered(candidate) | |
146 | if cache.validfor(rview): |
|
146 | if cache.validfor(rview): | |
147 | repo._branchcaches[candidate] = cache |
|
147 | repo._branchcaches[candidate] = cache | |
148 | cache.write(rview) |
|
148 | cache.write(rview) | |
149 | break |
|
149 | break | |
150 |
|
150 | |||
151 | class branchcache(dict): |
|
151 | class branchcache(dict): | |
152 | """A dict like object that hold branches heads cache. |
|
152 | """A dict like object that hold branches heads cache. | |
153 |
|
153 | |||
154 | This cache is used to avoid costly computations to determine all the |
|
154 | This cache is used to avoid costly computations to determine all the | |
155 | branch heads of a repo. |
|
155 | branch heads of a repo. | |
156 |
|
156 | |||
157 | The cache is serialized on disk in the following format: |
|
157 | The cache is serialized on disk in the following format: | |
158 |
|
158 | |||
159 | <tip hex node> <tip rev number> [optional filtered repo hex hash] |
|
159 | <tip hex node> <tip rev number> [optional filtered repo hex hash] | |
160 | <branch head hex node> <open/closed state> <branch name> |
|
160 | <branch head hex node> <open/closed state> <branch name> | |
161 | <branch head hex node> <open/closed state> <branch name> |
|
161 | <branch head hex node> <open/closed state> <branch name> | |
162 | ... |
|
162 | ... | |
163 |
|
163 | |||
164 | The first line is used to check if the cache is still valid. If the |
|
164 | The first line is used to check if the cache is still valid. If the | |
165 | branch cache is for a filtered repo view, an optional third hash is |
|
165 | branch cache is for a filtered repo view, an optional third hash is | |
166 | included that hashes the hashes of all filtered revisions. |
|
166 | included that hashes the hashes of all filtered revisions. | |
167 |
|
167 | |||
168 | The open/closed state is represented by a single letter 'o' or 'c'. |
|
168 | The open/closed state is represented by a single letter 'o' or 'c'. | |
169 | This field can be used to avoid changelog reads when determining if a |
|
169 | This field can be used to avoid changelog reads when determining if a | |
170 | branch head closes a branch or not. |
|
170 | branch head closes a branch or not. | |
171 | """ |
|
171 | """ | |
172 |
|
172 | |||
173 | def __init__(self, entries=(), tipnode=nullid, tiprev=nullrev, |
|
173 | def __init__(self, entries=(), tipnode=nullid, tiprev=nullrev, | |
174 | filteredhash=None, closednodes=None): |
|
174 | filteredhash=None, closednodes=None): | |
175 | super(branchcache, self).__init__(entries) |
|
175 | super(branchcache, self).__init__(entries) | |
176 | self.tipnode = tipnode |
|
176 | self.tipnode = tipnode | |
177 | self.tiprev = tiprev |
|
177 | self.tiprev = tiprev | |
178 | self.filteredhash = filteredhash |
|
178 | self.filteredhash = filteredhash | |
179 | # closednodes is a set of nodes that close their branch. If the branch |
|
179 | # closednodes is a set of nodes that close their branch. If the branch | |
180 | # cache has been updated, it may contain nodes that are no longer |
|
180 | # cache has been updated, it may contain nodes that are no longer | |
181 | # heads. |
|
181 | # heads. | |
182 | if closednodes is None: |
|
182 | if closednodes is None: | |
183 | self._closednodes = set() |
|
183 | self._closednodes = set() | |
184 | else: |
|
184 | else: | |
185 | self._closednodes = closednodes |
|
185 | self._closednodes = closednodes | |
186 |
|
186 | |||
187 | def validfor(self, repo): |
|
187 | def validfor(self, repo): | |
188 | """Is the cache content valid regarding a repo |
|
188 | """Is the cache content valid regarding a repo | |
189 |
|
189 | |||
190 | - False when cached tipnode is unknown or if we detect a strip. |
|
190 | - False when cached tipnode is unknown or if we detect a strip. | |
191 | - True when cache is up to date or a subset of current repo.""" |
|
191 | - True when cache is up to date or a subset of current repo.""" | |
192 | try: |
|
192 | try: | |
193 | return ((self.tipnode == repo.changelog.node(self.tiprev)) |
|
193 | return ((self.tipnode == repo.changelog.node(self.tiprev)) | |
194 | and (self.filteredhash == \ |
|
194 | and (self.filteredhash == \ | |
195 | scmutil.filteredhash(repo, self.tiprev))) |
|
195 | scmutil.filteredhash(repo, self.tiprev))) | |
196 | except IndexError: |
|
196 | except IndexError: | |
197 | return False |
|
197 | return False | |
198 |
|
198 | |||
199 | def _branchtip(self, heads): |
|
199 | def _branchtip(self, heads): | |
200 | '''Return tuple with last open head in heads and false, |
|
200 | '''Return tuple with last open head in heads and false, | |
201 | otherwise return last closed head and true.''' |
|
201 | otherwise return last closed head and true.''' | |
202 | tip = heads[-1] |
|
202 | tip = heads[-1] | |
203 | closed = True |
|
203 | closed = True | |
204 | for h in reversed(heads): |
|
204 | for h in reversed(heads): | |
205 | if h not in self._closednodes: |
|
205 | if h not in self._closednodes: | |
206 | tip = h |
|
206 | tip = h | |
207 | closed = False |
|
207 | closed = False | |
208 | break |
|
208 | break | |
209 | return tip, closed |
|
209 | return tip, closed | |
210 |
|
210 | |||
211 | def branchtip(self, branch): |
|
211 | def branchtip(self, branch): | |
212 | '''Return the tipmost open head on branch head, otherwise return the |
|
212 | '''Return the tipmost open head on branch head, otherwise return the | |
213 | tipmost closed head on branch. |
|
213 | tipmost closed head on branch. | |
214 | Raise KeyError for unknown branch.''' |
|
214 | Raise KeyError for unknown branch.''' | |
215 | return self._branchtip(self[branch])[0] |
|
215 | return self._branchtip(self[branch])[0] | |
216 |
|
216 | |||
217 | def iteropen(self, nodes): |
|
217 | def iteropen(self, nodes): | |
218 | return (n for n in nodes if n not in self._closednodes) |
|
218 | return (n for n in nodes if n not in self._closednodes) | |
219 |
|
219 | |||
220 | def branchheads(self, branch, closed=False): |
|
220 | def branchheads(self, branch, closed=False): | |
221 | heads = self[branch] |
|
221 | heads = self[branch] | |
222 | if not closed: |
|
222 | if not closed: | |
223 | heads = list(self.iteropen(heads)) |
|
223 | heads = list(self.iteropen(heads)) | |
224 | return heads |
|
224 | return heads | |
225 |
|
225 | |||
226 | def iterbranches(self): |
|
226 | def iterbranches(self): | |
227 | for bn, heads in self.iteritems(): |
|
227 | for bn, heads in self.iteritems(): | |
228 | yield (bn, heads) + self._branchtip(heads) |
|
228 | yield (bn, heads) + self._branchtip(heads) | |
229 |
|
229 | |||
230 | def copy(self): |
|
230 | def copy(self): | |
231 | """return an deep copy of the branchcache object""" |
|
231 | """return an deep copy of the branchcache object""" | |
232 | return branchcache(self, self.tipnode, self.tiprev, self.filteredhash, |
|
232 | return branchcache(self, self.tipnode, self.tiprev, self.filteredhash, | |
233 | self._closednodes) |
|
233 | self._closednodes) | |
234 |
|
234 | |||
235 | def write(self, repo): |
|
235 | def write(self, repo): | |
236 | try: |
|
236 | try: | |
237 | f = repo.cachevfs(_filename(repo), "w", atomictemp=True) |
|
237 | f = repo.cachevfs(_filename(repo), "w", atomictemp=True) | |
238 | cachekey = [hex(self.tipnode), '%d' % self.tiprev] |
|
238 | cachekey = [hex(self.tipnode), '%d' % self.tiprev] | |
239 | if self.filteredhash is not None: |
|
239 | if self.filteredhash is not None: | |
240 | cachekey.append(hex(self.filteredhash)) |
|
240 | cachekey.append(hex(self.filteredhash)) | |
241 | f.write(" ".join(cachekey) + '\n') |
|
241 | f.write(" ".join(cachekey) + '\n') | |
242 | nodecount = 0 |
|
242 | nodecount = 0 | |
243 | for label, nodes in sorted(self.iteritems()): |
|
243 | for label, nodes in sorted(self.iteritems()): | |
244 | for node in nodes: |
|
244 | for node in nodes: | |
245 | nodecount += 1 |
|
245 | nodecount += 1 | |
246 | if node in self._closednodes: |
|
246 | if node in self._closednodes: | |
247 | state = 'c' |
|
247 | state = 'c' | |
248 | else: |
|
248 | else: | |
249 | state = 'o' |
|
249 | state = 'o' | |
250 | f.write("%s %s %s\n" % (hex(node), state, |
|
250 | f.write("%s %s %s\n" % (hex(node), state, | |
251 | encoding.fromlocal(label))) |
|
251 | encoding.fromlocal(label))) | |
252 | f.close() |
|
252 | f.close() | |
253 | repo.ui.log('branchcache', |
|
253 | repo.ui.log('branchcache', | |
254 | 'wrote %s branch cache with %d labels and %d nodes\n', |
|
254 | 'wrote %s branch cache with %d labels and %d nodes\n', | |
255 | repo.filtername, len(self), nodecount) |
|
255 | repo.filtername, len(self), nodecount) | |
256 | except (IOError, OSError, error.Abort) as inst: |
|
256 | except (IOError, OSError, error.Abort) as inst: | |
257 | # Abort may be raised by read only opener, so log and continue |
|
257 | # Abort may be raised by read only opener, so log and continue | |
258 | repo.ui.debug("couldn't write branch cache: %s\n" % inst) |
|
258 | repo.ui.debug("couldn't write branch cache: %s\n" % inst) | |
259 |
|
259 | |||
260 | def update(self, repo, revgen): |
|
260 | def update(self, repo, revgen): | |
261 | """Given a branchhead cache, self, that may have extra nodes or be |
|
261 | """Given a branchhead cache, self, that may have extra nodes or be | |
262 | missing heads, and a generator of nodes that are strictly a superset of |
|
262 | missing heads, and a generator of nodes that are strictly a superset of | |
263 | heads missing, this function updates self to be correct. |
|
263 | heads missing, this function updates self to be correct. | |
264 | """ |
|
264 | """ | |
265 | starttime = util.timer() |
|
265 | starttime = util.timer() | |
266 | cl = repo.changelog |
|
266 | cl = repo.changelog | |
267 | # collect new branch entries |
|
267 | # collect new branch entries | |
268 | newbranches = {} |
|
268 | newbranches = {} | |
269 | getbranchinfo = repo.revbranchcache().branchinfo |
|
269 | getbranchinfo = repo.revbranchcache().branchinfo | |
270 | for r in revgen: |
|
270 | for r in revgen: | |
271 | branch, closesbranch = getbranchinfo(r) |
|
271 | branch, closesbranch = getbranchinfo(r) | |
272 | newbranches.setdefault(branch, []).append(r) |
|
272 | newbranches.setdefault(branch, []).append(r) | |
273 | if closesbranch: |
|
273 | if closesbranch: | |
274 | self._closednodes.add(cl.node(r)) |
|
274 | self._closednodes.add(cl.node(r)) | |
275 |
|
275 | |||
276 | # fetch current topological heads to speed up filtering |
|
276 | # fetch current topological heads to speed up filtering | |
277 | topoheads = set(cl.headrevs()) |
|
277 | topoheads = set(cl.headrevs()) | |
278 |
|
278 | |||
279 | # if older branchheads are reachable from new ones, they aren't |
|
279 | # if older branchheads are reachable from new ones, they aren't | |
280 | # really branchheads. Note checking parents is insufficient: |
|
280 | # really branchheads. Note checking parents is insufficient: | |
281 | # 1 (branch a) -> 2 (branch b) -> 3 (branch a) |
|
281 | # 1 (branch a) -> 2 (branch b) -> 3 (branch a) | |
282 | for branch, newheadrevs in newbranches.iteritems(): |
|
282 | for branch, newheadrevs in newbranches.iteritems(): | |
283 | bheads = self.setdefault(branch, []) |
|
283 | bheads = self.setdefault(branch, []) | |
284 | bheadset = set(cl.rev(node) for node in bheads) |
|
284 | bheadset = set(cl.rev(node) for node in bheads) | |
285 |
|
285 | |||
286 | # This have been tested True on all internal usage of this function. |
|
286 | # This have been tested True on all internal usage of this function. | |
287 | # run it again in case of doubt |
|
287 | # run it again in case of doubt | |
288 | # assert not (set(bheadrevs) & set(newheadrevs)) |
|
288 | # assert not (set(bheadrevs) & set(newheadrevs)) | |
289 | newheadrevs.sort() |
|
289 | newheadrevs.sort() | |
290 | bheadset.update(newheadrevs) |
|
290 | bheadset.update(newheadrevs) | |
291 |
|
291 | |||
292 | # This prunes out two kinds of heads - heads that are superseded by |
|
292 | # This prunes out two kinds of heads - heads that are superseded by | |
293 | # a head in newheadrevs, and newheadrevs that are not heads because |
|
293 | # a head in newheadrevs, and newheadrevs that are not heads because | |
294 | # an existing head is their descendant. |
|
294 | # an existing head is their descendant. | |
295 | uncertain = bheadset - topoheads |
|
295 | uncertain = bheadset - topoheads | |
296 | if uncertain: |
|
296 | if uncertain: | |
297 | floorrev = min(uncertain) |
|
297 | floorrev = min(uncertain) | |
298 | ancestors = set(cl.ancestors(newheadrevs, floorrev)) |
|
298 | ancestors = set(cl.ancestors(newheadrevs, floorrev)) | |
299 | bheadset -= ancestors |
|
299 | bheadset -= ancestors | |
300 | bheadrevs = sorted(bheadset) |
|
300 | bheadrevs = sorted(bheadset) | |
301 | self[branch] = [cl.node(rev) for rev in bheadrevs] |
|
301 | self[branch] = [cl.node(rev) for rev in bheadrevs] | |
302 | tiprev = bheadrevs[-1] |
|
302 | tiprev = bheadrevs[-1] | |
303 | if tiprev > self.tiprev: |
|
303 | if tiprev > self.tiprev: | |
304 | self.tipnode = cl.node(tiprev) |
|
304 | self.tipnode = cl.node(tiprev) | |
305 | self.tiprev = tiprev |
|
305 | self.tiprev = tiprev | |
306 |
|
306 | |||
307 | if not self.validfor(repo): |
|
307 | if not self.validfor(repo): | |
308 | # cache key are not valid anymore |
|
308 | # cache key are not valid anymore | |
309 | self.tipnode = nullid |
|
309 | self.tipnode = nullid | |
310 | self.tiprev = nullrev |
|
310 | self.tiprev = nullrev | |
311 | for heads in self.values(): |
|
311 | for heads in self.values(): | |
312 | tiprev = max(cl.rev(node) for node in heads) |
|
312 | tiprev = max(cl.rev(node) for node in heads) | |
313 | if tiprev > self.tiprev: |
|
313 | if tiprev > self.tiprev: | |
314 | self.tipnode = cl.node(tiprev) |
|
314 | self.tipnode = cl.node(tiprev) | |
315 | self.tiprev = tiprev |
|
315 | self.tiprev = tiprev | |
316 | self.filteredhash = scmutil.filteredhash(repo, self.tiprev) |
|
316 | self.filteredhash = scmutil.filteredhash(repo, self.tiprev) | |
317 |
|
317 | |||
318 | duration = util.timer() - starttime |
|
318 | duration = util.timer() - starttime | |
319 | repo.ui.log('branchcache', 'updated %s branch cache in %.4f seconds\n', |
|
319 | repo.ui.log('branchcache', 'updated %s branch cache in %.4f seconds\n', | |
320 | repo.filtername, duration) |
|
320 | repo.filtername, duration) | |
321 |
|
321 | |||
322 | # Revision branch info cache |
|
322 | # Revision branch info cache | |
323 |
|
323 | |||
324 | _rbcversion = '-v1' |
|
324 | _rbcversion = '-v1' | |
325 | _rbcnames = 'rbc-names' + _rbcversion |
|
325 | _rbcnames = 'rbc-names' + _rbcversion | |
326 | _rbcrevs = 'rbc-revs' + _rbcversion |
|
326 | _rbcrevs = 'rbc-revs' + _rbcversion | |
327 | # [4 byte hash prefix][4 byte branch name number with sign bit indicating open] |
|
327 | # [4 byte hash prefix][4 byte branch name number with sign bit indicating open] | |
328 | _rbcrecfmt = '>4sI' |
|
328 | _rbcrecfmt = '>4sI' | |
329 | _rbcrecsize = calcsize(_rbcrecfmt) |
|
329 | _rbcrecsize = calcsize(_rbcrecfmt) | |
330 | _rbcnodelen = 4 |
|
330 | _rbcnodelen = 4 | |
331 | _rbcbranchidxmask = 0x7fffffff |
|
331 | _rbcbranchidxmask = 0x7fffffff | |
332 | _rbccloseflag = 0x80000000 |
|
332 | _rbccloseflag = 0x80000000 | |
333 |
|
333 | |||
334 | class revbranchcache(object): |
|
334 | class revbranchcache(object): | |
335 | """Persistent cache, mapping from revision number to branch name and close. |
|
335 | """Persistent cache, mapping from revision number to branch name and close. | |
336 | This is a low level cache, independent of filtering. |
|
336 | This is a low level cache, independent of filtering. | |
337 |
|
337 | |||
338 | Branch names are stored in rbc-names in internal encoding separated by 0. |
|
338 | Branch names are stored in rbc-names in internal encoding separated by 0. | |
339 | rbc-names is append-only, and each branch name is only stored once and will |
|
339 | rbc-names is append-only, and each branch name is only stored once and will | |
340 | thus have a unique index. |
|
340 | thus have a unique index. | |
341 |
|
341 | |||
342 | The branch info for each revision is stored in rbc-revs as constant size |
|
342 | The branch info for each revision is stored in rbc-revs as constant size | |
343 | records. The whole file is read into memory, but it is only 'parsed' on |
|
343 | records. The whole file is read into memory, but it is only 'parsed' on | |
344 | demand. The file is usually append-only but will be truncated if repo |
|
344 | demand. The file is usually append-only but will be truncated if repo | |
345 | modification is detected. |
|
345 | modification is detected. | |
346 | The record for each revision contains the first 4 bytes of the |
|
346 | The record for each revision contains the first 4 bytes of the | |
347 | corresponding node hash, and the record is only used if it still matches. |
|
347 | corresponding node hash, and the record is only used if it still matches. | |
348 | Even a completely trashed rbc-revs fill thus still give the right result |
|
348 | Even a completely trashed rbc-revs fill thus still give the right result | |
349 | while converging towards full recovery ... assuming no incorrectly matching |
|
349 | while converging towards full recovery ... assuming no incorrectly matching | |
350 | node hashes. |
|
350 | node hashes. | |
351 | The record also contains 4 bytes where 31 bits contains the index of the |
|
351 | The record also contains 4 bytes where 31 bits contains the index of the | |
352 | branch and the last bit indicate that it is a branch close commit. |
|
352 | branch and the last bit indicate that it is a branch close commit. | |
353 | The usage pattern for rbc-revs is thus somewhat similar to 00changelog.i |
|
353 | The usage pattern for rbc-revs is thus somewhat similar to 00changelog.i | |
354 | and will grow with it but be 1/8th of its size. |
|
354 | and will grow with it but be 1/8th of its size. | |
355 | """ |
|
355 | """ | |
356 |
|
356 | |||
357 | def __init__(self, repo, readonly=True): |
|
357 | def __init__(self, repo, readonly=True): | |
358 | assert repo.filtername is None |
|
358 | assert repo.filtername is None | |
359 | self._repo = repo |
|
359 | self._repo = repo | |
360 | self._names = [] # branch names in local encoding with static index |
|
360 | self._names = [] # branch names in local encoding with static index | |
361 | self._rbcrevs = bytearray() |
|
361 | self._rbcrevs = bytearray() | |
362 | self._rbcsnameslen = 0 # length of names read at _rbcsnameslen |
|
362 | self._rbcsnameslen = 0 # length of names read at _rbcsnameslen | |
363 | try: |
|
363 | try: | |
364 | bndata = repo.cachevfs.read(_rbcnames) |
|
364 | bndata = repo.cachevfs.read(_rbcnames) | |
365 | self._rbcsnameslen = len(bndata) # for verification before writing |
|
365 | self._rbcsnameslen = len(bndata) # for verification before writing | |
366 | if bndata: |
|
366 | if bndata: | |
367 | self._names = [encoding.tolocal(bn) |
|
367 | self._names = [encoding.tolocal(bn) | |
368 | for bn in bndata.split('\0')] |
|
368 | for bn in bndata.split('\0')] | |
369 | except (IOError, OSError): |
|
369 | except (IOError, OSError): | |
370 | if readonly: |
|
370 | if readonly: | |
371 | # don't try to use cache - fall back to the slow path |
|
371 | # don't try to use cache - fall back to the slow path | |
372 | self.branchinfo = self._branchinfo |
|
372 | self.branchinfo = self._branchinfo | |
373 |
|
373 | |||
374 | if self._names: |
|
374 | if self._names: | |
375 | try: |
|
375 | try: | |
376 | data = repo.cachevfs.read(_rbcrevs) |
|
376 | data = repo.cachevfs.read(_rbcrevs) | |
377 | self._rbcrevs[:] = data |
|
377 | self._rbcrevs[:] = data | |
378 | except (IOError, OSError) as inst: |
|
378 | except (IOError, OSError) as inst: | |
379 | repo.ui.debug("couldn't read revision branch cache: %s\n" % |
|
379 | repo.ui.debug("couldn't read revision branch cache: %s\n" % | |
380 |
|
|
380 | util.forcebytestr(inst)) | |
381 | # remember number of good records on disk |
|
381 | # remember number of good records on disk | |
382 | self._rbcrevslen = min(len(self._rbcrevs) // _rbcrecsize, |
|
382 | self._rbcrevslen = min(len(self._rbcrevs) // _rbcrecsize, | |
383 | len(repo.changelog)) |
|
383 | len(repo.changelog)) | |
384 | if self._rbcrevslen == 0: |
|
384 | if self._rbcrevslen == 0: | |
385 | self._names = [] |
|
385 | self._names = [] | |
386 | self._rbcnamescount = len(self._names) # number of names read at |
|
386 | self._rbcnamescount = len(self._names) # number of names read at | |
387 | # _rbcsnameslen |
|
387 | # _rbcsnameslen | |
388 | self._namesreverse = dict((b, r) for r, b in enumerate(self._names)) |
|
388 | self._namesreverse = dict((b, r) for r, b in enumerate(self._names)) | |
389 |
|
389 | |||
390 | def _clear(self): |
|
390 | def _clear(self): | |
391 | self._rbcsnameslen = 0 |
|
391 | self._rbcsnameslen = 0 | |
392 | del self._names[:] |
|
392 | del self._names[:] | |
393 | self._rbcnamescount = 0 |
|
393 | self._rbcnamescount = 0 | |
394 | self._namesreverse.clear() |
|
394 | self._namesreverse.clear() | |
395 | self._rbcrevslen = len(self._repo.changelog) |
|
395 | self._rbcrevslen = len(self._repo.changelog) | |
396 | self._rbcrevs = bytearray(self._rbcrevslen * _rbcrecsize) |
|
396 | self._rbcrevs = bytearray(self._rbcrevslen * _rbcrecsize) | |
397 |
|
397 | |||
398 | def branchinfo(self, rev): |
|
398 | def branchinfo(self, rev): | |
399 | """Return branch name and close flag for rev, using and updating |
|
399 | """Return branch name and close flag for rev, using and updating | |
400 | persistent cache.""" |
|
400 | persistent cache.""" | |
401 | changelog = self._repo.changelog |
|
401 | changelog = self._repo.changelog | |
402 | rbcrevidx = rev * _rbcrecsize |
|
402 | rbcrevidx = rev * _rbcrecsize | |
403 |
|
403 | |||
404 | # avoid negative index, changelog.read(nullrev) is fast without cache |
|
404 | # avoid negative index, changelog.read(nullrev) is fast without cache | |
405 | if rev == nullrev: |
|
405 | if rev == nullrev: | |
406 | return changelog.branchinfo(rev) |
|
406 | return changelog.branchinfo(rev) | |
407 |
|
407 | |||
408 | # if requested rev isn't allocated, grow and cache the rev info |
|
408 | # if requested rev isn't allocated, grow and cache the rev info | |
409 | if len(self._rbcrevs) < rbcrevidx + _rbcrecsize: |
|
409 | if len(self._rbcrevs) < rbcrevidx + _rbcrecsize: | |
410 | return self._branchinfo(rev) |
|
410 | return self._branchinfo(rev) | |
411 |
|
411 | |||
412 | # fast path: extract data from cache, use it if node is matching |
|
412 | # fast path: extract data from cache, use it if node is matching | |
413 | reponode = changelog.node(rev)[:_rbcnodelen] |
|
413 | reponode = changelog.node(rev)[:_rbcnodelen] | |
414 | cachenode, branchidx = unpack_from( |
|
414 | cachenode, branchidx = unpack_from( | |
415 | _rbcrecfmt, util.buffer(self._rbcrevs), rbcrevidx) |
|
415 | _rbcrecfmt, util.buffer(self._rbcrevs), rbcrevidx) | |
416 | close = bool(branchidx & _rbccloseflag) |
|
416 | close = bool(branchidx & _rbccloseflag) | |
417 | if close: |
|
417 | if close: | |
418 | branchidx &= _rbcbranchidxmask |
|
418 | branchidx &= _rbcbranchidxmask | |
419 | if cachenode == '\0\0\0\0': |
|
419 | if cachenode == '\0\0\0\0': | |
420 | pass |
|
420 | pass | |
421 | elif cachenode == reponode: |
|
421 | elif cachenode == reponode: | |
422 | try: |
|
422 | try: | |
423 | return self._names[branchidx], close |
|
423 | return self._names[branchidx], close | |
424 | except IndexError: |
|
424 | except IndexError: | |
425 | # recover from invalid reference to unknown branch |
|
425 | # recover from invalid reference to unknown branch | |
426 | self._repo.ui.debug("referenced branch names not found" |
|
426 | self._repo.ui.debug("referenced branch names not found" | |
427 | " - rebuilding revision branch cache from scratch\n") |
|
427 | " - rebuilding revision branch cache from scratch\n") | |
428 | self._clear() |
|
428 | self._clear() | |
429 | else: |
|
429 | else: | |
430 | # rev/node map has changed, invalidate the cache from here up |
|
430 | # rev/node map has changed, invalidate the cache from here up | |
431 | self._repo.ui.debug("history modification detected - truncating " |
|
431 | self._repo.ui.debug("history modification detected - truncating " | |
432 | "revision branch cache to revision %d\n" % rev) |
|
432 | "revision branch cache to revision %d\n" % rev) | |
433 | truncate = rbcrevidx + _rbcrecsize |
|
433 | truncate = rbcrevidx + _rbcrecsize | |
434 | del self._rbcrevs[truncate:] |
|
434 | del self._rbcrevs[truncate:] | |
435 | self._rbcrevslen = min(self._rbcrevslen, truncate) |
|
435 | self._rbcrevslen = min(self._rbcrevslen, truncate) | |
436 |
|
436 | |||
437 | # fall back to slow path and make sure it will be written to disk |
|
437 | # fall back to slow path and make sure it will be written to disk | |
438 | return self._branchinfo(rev) |
|
438 | return self._branchinfo(rev) | |
439 |
|
439 | |||
440 | def _branchinfo(self, rev): |
|
440 | def _branchinfo(self, rev): | |
441 | """Retrieve branch info from changelog and update _rbcrevs""" |
|
441 | """Retrieve branch info from changelog and update _rbcrevs""" | |
442 | changelog = self._repo.changelog |
|
442 | changelog = self._repo.changelog | |
443 | b, close = changelog.branchinfo(rev) |
|
443 | b, close = changelog.branchinfo(rev) | |
444 | if b in self._namesreverse: |
|
444 | if b in self._namesreverse: | |
445 | branchidx = self._namesreverse[b] |
|
445 | branchidx = self._namesreverse[b] | |
446 | else: |
|
446 | else: | |
447 | branchidx = len(self._names) |
|
447 | branchidx = len(self._names) | |
448 | self._names.append(b) |
|
448 | self._names.append(b) | |
449 | self._namesreverse[b] = branchidx |
|
449 | self._namesreverse[b] = branchidx | |
450 | reponode = changelog.node(rev) |
|
450 | reponode = changelog.node(rev) | |
451 | if close: |
|
451 | if close: | |
452 | branchidx |= _rbccloseflag |
|
452 | branchidx |= _rbccloseflag | |
453 | self._setcachedata(rev, reponode, branchidx) |
|
453 | self._setcachedata(rev, reponode, branchidx) | |
454 | return b, close |
|
454 | return b, close | |
455 |
|
455 | |||
456 | def _setcachedata(self, rev, node, branchidx): |
|
456 | def _setcachedata(self, rev, node, branchidx): | |
457 | """Writes the node's branch data to the in-memory cache data.""" |
|
457 | """Writes the node's branch data to the in-memory cache data.""" | |
458 | if rev == nullrev: |
|
458 | if rev == nullrev: | |
459 | return |
|
459 | return | |
460 | rbcrevidx = rev * _rbcrecsize |
|
460 | rbcrevidx = rev * _rbcrecsize | |
461 | if len(self._rbcrevs) < rbcrevidx + _rbcrecsize: |
|
461 | if len(self._rbcrevs) < rbcrevidx + _rbcrecsize: | |
462 | self._rbcrevs.extend('\0' * |
|
462 | self._rbcrevs.extend('\0' * | |
463 | (len(self._repo.changelog) * _rbcrecsize - |
|
463 | (len(self._repo.changelog) * _rbcrecsize - | |
464 | len(self._rbcrevs))) |
|
464 | len(self._rbcrevs))) | |
465 | pack_into(_rbcrecfmt, self._rbcrevs, rbcrevidx, node, branchidx) |
|
465 | pack_into(_rbcrecfmt, self._rbcrevs, rbcrevidx, node, branchidx) | |
466 | self._rbcrevslen = min(self._rbcrevslen, rev) |
|
466 | self._rbcrevslen = min(self._rbcrevslen, rev) | |
467 |
|
467 | |||
468 | tr = self._repo.currenttransaction() |
|
468 | tr = self._repo.currenttransaction() | |
469 | if tr: |
|
469 | if tr: | |
470 | tr.addfinalize('write-revbranchcache', self.write) |
|
470 | tr.addfinalize('write-revbranchcache', self.write) | |
471 |
|
471 | |||
472 | def write(self, tr=None): |
|
472 | def write(self, tr=None): | |
473 | """Save branch cache if it is dirty.""" |
|
473 | """Save branch cache if it is dirty.""" | |
474 | repo = self._repo |
|
474 | repo = self._repo | |
475 | wlock = None |
|
475 | wlock = None | |
476 | step = '' |
|
476 | step = '' | |
477 | try: |
|
477 | try: | |
478 | if self._rbcnamescount < len(self._names): |
|
478 | if self._rbcnamescount < len(self._names): | |
479 | step = ' names' |
|
479 | step = ' names' | |
480 | wlock = repo.wlock(wait=False) |
|
480 | wlock = repo.wlock(wait=False) | |
481 | if self._rbcnamescount != 0: |
|
481 | if self._rbcnamescount != 0: | |
482 | f = repo.cachevfs.open(_rbcnames, 'ab') |
|
482 | f = repo.cachevfs.open(_rbcnames, 'ab') | |
483 | if f.tell() == self._rbcsnameslen: |
|
483 | if f.tell() == self._rbcsnameslen: | |
484 | f.write('\0') |
|
484 | f.write('\0') | |
485 | else: |
|
485 | else: | |
486 | f.close() |
|
486 | f.close() | |
487 | repo.ui.debug("%s changed - rewriting it\n" % _rbcnames) |
|
487 | repo.ui.debug("%s changed - rewriting it\n" % _rbcnames) | |
488 | self._rbcnamescount = 0 |
|
488 | self._rbcnamescount = 0 | |
489 | self._rbcrevslen = 0 |
|
489 | self._rbcrevslen = 0 | |
490 | if self._rbcnamescount == 0: |
|
490 | if self._rbcnamescount == 0: | |
491 | # before rewriting names, make sure references are removed |
|
491 | # before rewriting names, make sure references are removed | |
492 | repo.cachevfs.unlinkpath(_rbcrevs, ignoremissing=True) |
|
492 | repo.cachevfs.unlinkpath(_rbcrevs, ignoremissing=True) | |
493 | f = repo.cachevfs.open(_rbcnames, 'wb') |
|
493 | f = repo.cachevfs.open(_rbcnames, 'wb') | |
494 | f.write('\0'.join(encoding.fromlocal(b) |
|
494 | f.write('\0'.join(encoding.fromlocal(b) | |
495 | for b in self._names[self._rbcnamescount:])) |
|
495 | for b in self._names[self._rbcnamescount:])) | |
496 | self._rbcsnameslen = f.tell() |
|
496 | self._rbcsnameslen = f.tell() | |
497 | f.close() |
|
497 | f.close() | |
498 | self._rbcnamescount = len(self._names) |
|
498 | self._rbcnamescount = len(self._names) | |
499 |
|
499 | |||
500 | start = self._rbcrevslen * _rbcrecsize |
|
500 | start = self._rbcrevslen * _rbcrecsize | |
501 | if start != len(self._rbcrevs): |
|
501 | if start != len(self._rbcrevs): | |
502 | step = '' |
|
502 | step = '' | |
503 | if wlock is None: |
|
503 | if wlock is None: | |
504 | wlock = repo.wlock(wait=False) |
|
504 | wlock = repo.wlock(wait=False) | |
505 | revs = min(len(repo.changelog), |
|
505 | revs = min(len(repo.changelog), | |
506 | len(self._rbcrevs) // _rbcrecsize) |
|
506 | len(self._rbcrevs) // _rbcrecsize) | |
507 | f = repo.cachevfs.open(_rbcrevs, 'ab') |
|
507 | f = repo.cachevfs.open(_rbcrevs, 'ab') | |
508 | if f.tell() != start: |
|
508 | if f.tell() != start: | |
509 | repo.ui.debug("truncating cache/%s to %d\n" |
|
509 | repo.ui.debug("truncating cache/%s to %d\n" | |
510 | % (_rbcrevs, start)) |
|
510 | % (_rbcrevs, start)) | |
511 | f.seek(start) |
|
511 | f.seek(start) | |
512 | if f.tell() != start: |
|
512 | if f.tell() != start: | |
513 | start = 0 |
|
513 | start = 0 | |
514 | f.seek(start) |
|
514 | f.seek(start) | |
515 | f.truncate() |
|
515 | f.truncate() | |
516 | end = revs * _rbcrecsize |
|
516 | end = revs * _rbcrecsize | |
517 | f.write(self._rbcrevs[start:end]) |
|
517 | f.write(self._rbcrevs[start:end]) | |
518 | f.close() |
|
518 | f.close() | |
519 | self._rbcrevslen = revs |
|
519 | self._rbcrevslen = revs | |
520 | except (IOError, OSError, error.Abort, error.LockError) as inst: |
|
520 | except (IOError, OSError, error.Abort, error.LockError) as inst: | |
521 | repo.ui.debug("couldn't write revision branch cache%s: %s\n" |
|
521 | repo.ui.debug("couldn't write revision branch cache%s: %s\n" | |
522 | % (step, inst)) |
|
522 | % (step, inst)) | |
523 | finally: |
|
523 | finally: | |
524 | if wlock is not None: |
|
524 | if wlock is not None: | |
525 | wlock.release() |
|
525 | wlock.release() |
General Comments 0
You need to be logged in to leave comments.
Login now