sh-setup: work around "unset IFS" bug in some shells
[git/git.git] / Documentation / technical / pack-protocol.txt
1 Packfile transfer protocols
2 ===========================
3
4 Git supports transferring data in packfiles over the ssh://, git:// and
5 file:// transports. There exist two sets of protocols, one for pushing
6 data from a client to a server and another for fetching data from a
7 server to a client. All three transports (ssh, git, file) use the same
8 protocol to transfer data.
9
10 The processes invoked in the canonical Git implementation are 'upload-pack'
11 on the server side and 'fetch-pack' on the client side for fetching data;
12 then 'receive-pack' on the server and 'send-pack' on the client for pushing
13 data. The protocol functions to have a server tell a client what is
14 currently on the server, then for the two to negotiate the smallest amount
15 of data to send in order to fully update one or the other.
16
17 Transports
18 ----------
19 There are three transports over which the packfile protocol is
20 initiated. The Git transport is a simple, unauthenticated server that
21 takes the command (almost always 'upload-pack', though Git
22 servers can be configured to be globally writable, in which 'receive-
23 pack' initiation is also allowed) with which the client wishes to
24 communicate and executes it and connects it to the requesting
25 process.
26
27 In the SSH transport, the client just runs the 'upload-pack'
28 or 'receive-pack' process on the server over the SSH protocol and then
29 communicates with that invoked process over the SSH connection.
30
31 The file:// transport runs the 'upload-pack' or 'receive-pack'
32 process locally and communicates with it over a pipe.
33
34 Git Transport
35 -------------
36
37 The Git transport starts off by sending the command and repository
38 on the wire using the pkt-line format, followed by a NUL byte and a
39 hostname parameter, terminated by a NUL byte.
40
41 0032git-upload-pack /project.git\0host=myserver.com\0
42
43 --
44 git-proto-request = request-command SP pathname NUL [ host-parameter NUL ]
45 request-command = "git-upload-pack" / "git-receive-pack" /
46 "git-upload-archive" ; case sensitive
47 pathname = *( %x01-ff ) ; exclude NUL
48 host-parameter = "host=" hostname [ ":" port ]
49 --
50
51 Only host-parameter is allowed in the git-proto-request. Clients
52 MUST NOT attempt to send additional parameters. It is used for the
53 git-daemon name based virtual hosting. See --interpolated-path
54 option to git daemon, with the %H/%CH format characters.
55
56 Basically what the Git client is doing to connect to an 'upload-pack'
57 process on the server side over the Git protocol is this:
58
59 $ echo -e -n \
60 "0039git-upload-pack /schacon/gitbook.git\0host=example.com\0" |
61 nc -v example.com 9418
62
63 If the server refuses the request for some reasons, it could abort
64 gracefully with an error message.
65
66 ----
67 error-line = PKT-LINE("ERR" SP explanation-text)
68 ----
69
70
71 SSH Transport
72 -------------
73
74 Initiating the upload-pack or receive-pack processes over SSH is
75 executing the binary on the server via SSH remote execution.
76 It is basically equivalent to running this:
77
78 $ ssh git.example.com "git-upload-pack '/project.git'"
79
80 For a server to support Git pushing and pulling for a given user over
81 SSH, that user needs to be able to execute one or both of those
82 commands via the SSH shell that they are provided on login. On some
83 systems, that shell access is limited to only being able to run those
84 two commands, or even just one of them.
85
86 In an ssh:// format URI, it's absolute in the URI, so the '/' after
87 the host name (or port number) is sent as an argument, which is then
88 read by the remote git-upload-pack exactly as is, so it's effectively
89 an absolute path in the remote filesystem.
90
91 git clone ssh://user@example.com/project.git
92 |
93 v
94 ssh user@example.com "git-upload-pack '/project.git'"
95
96 In a "user@host:path" format URI, its relative to the user's home
97 directory, because the Git client will run:
98
99 git clone user@example.com:project.git
100 |
101 v
102 ssh user@example.com "git-upload-pack 'project.git'"
103
104 The exception is if a '~' is used, in which case
105 we execute it without the leading '/'.
106
107 ssh://user@example.com/~alice/project.git,
108 |
109 v
110 ssh user@example.com "git-upload-pack '~alice/project.git'"
111
112 A few things to remember here:
113
114 - The "command name" is spelled with dash (e.g. git-upload-pack), but
115 this can be overridden by the client;
116
117 - The repository path is always quoted with single quotes.
118
119 Fetching Data From a Server
120 ===========================
121
122 When one Git repository wants to get data that a second repository
123 has, the first can 'fetch' from the second. This operation determines
124 what data the server has that the client does not then streams that
125 data down to the client in packfile format.
126
127
128 Reference Discovery
129 -------------------
130
131 When the client initially connects the server will immediately respond
132 with a listing of each reference it has (all branches and tags) along
133 with the object name that each reference currently points to.
134
135 $ echo -e -n "0039git-upload-pack /schacon/gitbook.git\0host=example.com\0" |
136 nc -v example.com 9418
137 00887217a7c7e582c46cec22a130adf4b9d7d950fba0 HEAD\0multi_ack thin-pack side-band side-band-64k ofs-delta shallow no-progress include-tag
138 00441d3fcd5ced445d1abc402225c0b8a1299641f497 refs/heads/integration
139 003f7217a7c7e582c46cec22a130adf4b9d7d950fba0 refs/heads/master
140 003cb88d2441cac0977faf98efc80305012112238d9d refs/tags/v0.9
141 003c525128480b96c89e6418b1e40909bf6c5b2d580f refs/tags/v1.0
142 003fe92df48743b7bc7d26bcaabfddde0a1e20cae47c refs/tags/v1.0^{}
143 0000
144
145 Server SHOULD terminate each non-flush line using LF ("\n") terminator;
146 client MUST NOT complain if there is no terminator.
147
148 The returned response is a pkt-line stream describing each ref and
149 its current value. The stream MUST be sorted by name according to
150 the C locale ordering.
151
152 If HEAD is a valid ref, HEAD MUST appear as the first advertised
153 ref. If HEAD is not a valid ref, HEAD MUST NOT appear in the
154 advertisement list at all, but other refs may still appear.
155
156 The stream MUST include capability declarations behind a NUL on the
157 first ref. The peeled value of a ref (that is "ref^{}") MUST be
158 immediately after the ref itself, if presented. A conforming server
159 MUST peel the ref if it's an annotated tag.
160
161 ----
162 advertised-refs = (no-refs / list-of-refs)
163 flush-pkt
164
165 no-refs = PKT-LINE(zero-id SP "capabilities^{}"
166 NUL capability-list LF)
167
168 list-of-refs = first-ref *other-ref
169 first-ref = PKT-LINE(obj-id SP refname
170 NUL capability-list LF)
171
172 other-ref = PKT-LINE(other-tip / other-peeled)
173 other-tip = obj-id SP refname LF
174 other-peeled = obj-id SP refname "^{}" LF
175
176 capability-list = capability *(SP capability)
177 capability = 1*(LC_ALPHA / DIGIT / "-" / "_")
178 LC_ALPHA = %x61-7A
179 ----
180
181 Server and client MUST use lowercase for obj-id, both MUST treat obj-id
182 as case-insensitive.
183
184 See protocol-capabilities.txt for a list of allowed server capabilities
185 and descriptions.
186
187 Packfile Negotiation
188 --------------------
189 After reference and capabilities discovery, the client can decide to
190 terminate the connection by sending a flush-pkt, telling the server it can
191 now gracefully terminate, and disconnect, when it does not need any pack
192 data. This can happen with the ls-remote command, and also can happen when
193 the client already is up-to-date.
194
195 Otherwise, it enters the negotiation phase, where the client and
196 server determine what the minimal packfile necessary for transport is,
197 by telling the server what objects it wants, its shallow objects
198 (if any), and the maximum commit depth it wants (if any). The client
199 will also send a list of the capabilities it wants to be in effect,
200 out of what the server said it could do with the first 'want' line.
201
202 ----
203 upload-request = want-list
204 *shallow-line
205 *1depth-request
206 flush-pkt
207
208 want-list = first-want
209 *additional-want
210
211 shallow-line = PKT_LINE("shallow" SP obj-id)
212
213 depth-request = PKT_LINE("deepen" SP depth)
214
215 first-want = PKT-LINE("want" SP obj-id SP capability-list LF)
216 additional-want = PKT-LINE("want" SP obj-id LF)
217
218 depth = 1*DIGIT
219 ----
220
221 Clients MUST send all the obj-ids it wants from the reference
222 discovery phase as 'want' lines. Clients MUST send at least one
223 'want' command in the request body. Clients MUST NOT mention an
224 obj-id in a 'want' command which did not appear in the response
225 obtained through ref discovery.
226
227 The client MUST write all obj-ids which it only has shallow copies
228 of (meaning that it does not have the parents of a commit) as
229 'shallow' lines so that the server is aware of the limitations of
230 the client's history. Clients MUST NOT mention an obj-id which
231 it does not know exists on the server.
232
233 The client now sends the maximum commit history depth it wants for
234 this transaction, which is the number of commits it wants from the
235 tip of the history, if any, as a 'deepen' line. A depth of 0 is the
236 same as not making a depth request. The client does not want to receive
237 any commits beyond this depth, nor objects needed only to complete
238 those commits. Commits whose parents are not received as a result are
239 defined as shallow and marked as such in the server. This information
240 is sent back to the client in the next step.
241
242 Once all the 'want's and 'shallow's (and optional 'deepen') are
243 transferred, clients MUST send a flush-pkt, to tell the server side
244 that it is done sending the list.
245
246 Otherwise, if the client sent a positive depth request, the server
247 will determine which commits will and will not be shallow and
248 send this information to the client. If the client did not request
249 a positive depth, this step is skipped.
250
251 ----
252 shallow-update = *shallow-line
253 *unshallow-line
254 flush-pkt
255
256 shallow-line = PKT-LINE("shallow" SP obj-id)
257
258 unshallow-line = PKT-LINE("unshallow" SP obj-id)
259 ----
260
261 If the client has requested a positive depth, the server will compute
262 the set of commits which are no deeper than the desired depth. The set
263 of commits start at the client's wants.
264
265 The server writes 'shallow' lines for each
266 commit whose parents will not be sent as a result. The server writes
267 an 'unshallow' line for each commit which the client has indicated is
268 shallow, but is no longer shallow at the currently requested depth
269 (that is, its parents will now be sent). The server MUST NOT mark
270 as unshallow anything which the client has not indicated was shallow.
271
272 Now the client will send a list of the obj-ids it has using 'have'
273 lines, so the server can make a packfile that only contains the objects
274 that the client needs. In multi_ack mode, the canonical implementation
275 will send up to 32 of these at a time, then will send a flush-pkt. The
276 canonical implementation will skip ahead and send the next 32 immediately,
277 so that there is always a block of 32 "in-flight on the wire" at a time.
278
279 ----
280 upload-haves = have-list
281 compute-end
282
283 have-list = *have-line
284 have-line = PKT-LINE("have" SP obj-id LF)
285 compute-end = flush-pkt / PKT-LINE("done")
286 ----
287
288 If the server reads 'have' lines, it then will respond by ACKing any
289 of the obj-ids the client said it had that the server also has. The
290 server will ACK obj-ids differently depending on which ack mode is
291 chosen by the client.
292
293 In multi_ack mode:
294
295 * the server will respond with 'ACK obj-id continue' for any common
296 commits.
297
298 * once the server has found an acceptable common base commit and is
299 ready to make a packfile, it will blindly ACK all 'have' obj-ids
300 back to the client.
301
302 * the server will then send a 'NACK' and then wait for another response
303 from the client - either a 'done' or another list of 'have' lines.
304
305 In multi_ack_detailed mode:
306
307 * the server will differentiate the ACKs where it is signaling
308 that it is ready to send data with 'ACK obj-id ready' lines, and
309 signals the identified common commits with 'ACK obj-id common' lines.
310
311 Without either multi_ack or multi_ack_detailed:
312
313 * upload-pack sends "ACK obj-id" on the first common object it finds.
314 After that it says nothing until the client gives it a "done".
315
316 * upload-pack sends "NAK" on a flush-pkt if no common object
317 has been found yet. If one has been found, and thus an ACK
318 was already sent, it's silent on the flush-pkt.
319
320 After the client has gotten enough ACK responses that it can determine
321 that the server has enough information to send an efficient packfile
322 (in the canonical implementation, this is determined when it has received
323 enough ACKs that it can color everything left in the --date-order queue
324 as common with the server, or the --date-order queue is empty), or the
325 client determines that it wants to give up (in the canonical implementation,
326 this is determined when the client sends 256 'have' lines without getting
327 any of them ACKed by the server - meaning there is nothing in common and
328 the server should just send all of its objects), then the client will send
329 a 'done' command. The 'done' command signals to the server that the client
330 is ready to receive its packfile data.
331
332 However, the 256 limit *only* turns on in the canonical client
333 implementation if we have received at least one "ACK %s continue"
334 during a prior round. This helps to ensure that at least one common
335 ancestor is found before we give up entirely.
336
337 Once the 'done' line is read from the client, the server will either
338 send a final 'ACK obj-id' or it will send a 'NAK'. The server only sends
339 ACK after 'done' if there is at least one common base and multi_ack or
340 multi_ack_detailed is enabled. The server always sends NAK after 'done'
341 if there is no common base found.
342
343 Then the server will start sending its packfile data.
344
345 ----
346 server-response = *ack_multi ack / nak
347 ack_multi = PKT-LINE("ACK" SP obj-id ack_status LF)
348 ack_status = "continue" / "common" / "ready"
349 ack = PKT-LINE("ACK SP obj-id LF)
350 nak = PKT-LINE("NAK" LF)
351 ----
352
353 A simple clone may look like this (with no 'have' lines):
354
355 ----
356 C: 0054want 74730d410fcb6603ace96f1dc55ea6196122532d multi_ack \
357 side-band-64k ofs-delta\n
358 C: 0032want 7d1665144a3a975c05f1f43902ddaf084e784dbe\n
359 C: 0032want 5a3f6be755bbb7deae50065988cbfa1ffa9ab68a\n
360 C: 0032want 7e47fe2bd8d01d481f44d7af0531bd93d3b21c01\n
361 C: 0032want 74730d410fcb6603ace96f1dc55ea6196122532d\n
362 C: 0000
363 C: 0009done\n
364
365 S: 0008NAK\n
366 S: [PACKFILE]
367 ----
368
369 An incremental update (fetch) response might look like this:
370
371 ----
372 C: 0054want 74730d410fcb6603ace96f1dc55ea6196122532d multi_ack \
373 side-band-64k ofs-delta\n
374 C: 0032want 7d1665144a3a975c05f1f43902ddaf084e784dbe\n
375 C: 0032want 5a3f6be755bbb7deae50065988cbfa1ffa9ab68a\n
376 C: 0000
377 C: 0032have 7e47fe2bd8d01d481f44d7af0531bd93d3b21c01\n
378 C: [30 more have lines]
379 C: 0032have 74730d410fcb6603ace96f1dc55ea6196122532d\n
380 C: 0000
381
382 S: 003aACK 7e47fe2bd8d01d481f44d7af0531bd93d3b21c01 continue\n
383 S: 003aACK 74730d410fcb6603ace96f1dc55ea6196122532d continue\n
384 S: 0008NAK\n
385
386 C: 0009done\n
387
388 S: 0031ACK 74730d410fcb6603ace96f1dc55ea6196122532d\n
389 S: [PACKFILE]
390 ----
391
392
393 Packfile Data
394 -------------
395
396 Now that the client and server have finished negotiation about what
397 the minimal amount of data that needs to be sent to the client is, the server
398 will construct and send the required data in packfile format.
399
400 See pack-format.txt for what the packfile itself actually looks like.
401
402 If 'side-band' or 'side-band-64k' capabilities have been specified by
403 the client, the server will send the packfile data multiplexed.
404
405 Each packet starting with the packet-line length of the amount of data
406 that follows, followed by a single byte specifying the sideband the
407 following data is coming in on.
408
409 In 'side-band' mode, it will send up to 999 data bytes plus 1 control
410 code, for a total of up to 1000 bytes in a pkt-line. In 'side-band-64k'
411 mode it will send up to 65519 data bytes plus 1 control code, for a
412 total of up to 65520 bytes in a pkt-line.
413
414 The sideband byte will be a '1', '2' or a '3'. Sideband '1' will contain
415 packfile data, sideband '2' will be used for progress information that the
416 client will generally print to stderr and sideband '3' is used for error
417 information.
418
419 If no 'side-band' capability was specified, the server will stream the
420 entire packfile without multiplexing.
421
422
423 Pushing Data To a Server
424 ========================
425
426 Pushing data to a server will invoke the 'receive-pack' process on the
427 server, which will allow the client to tell it which references it should
428 update and then send all the data the server will need for those new
429 references to be complete. Once all the data is received and validated,
430 the server will then update its references to what the client specified.
431
432 Authentication
433 --------------
434
435 The protocol itself contains no authentication mechanisms. That is to be
436 handled by the transport, such as SSH, before the 'receive-pack' process is
437 invoked. If 'receive-pack' is configured over the Git transport, those
438 repositories will be writable by anyone who can access that port (9418) as
439 that transport is unauthenticated.
440
441 Reference Discovery
442 -------------------
443
444 The reference discovery phase is done nearly the same way as it is in the
445 fetching protocol. Each reference obj-id and name on the server is sent
446 in packet-line format to the client, followed by a flush-pkt. The only
447 real difference is that the capability listing is different - the only
448 possible values are 'report-status', 'delete-refs' and 'ofs-delta'.
449
450 Reference Update Request and Packfile Transfer
451 ----------------------------------------------
452
453 Once the client knows what references the server is at, it can send a
454 list of reference update requests. For each reference on the server
455 that it wants to update, it sends a line listing the obj-id currently on
456 the server, the obj-id the client would like to update it to and the name
457 of the reference.
458
459 This list is followed by a flush-pkt and then the packfile that should
460 contain all the objects that the server will need to complete the new
461 references.
462
463 ----
464 update-request = command-list [pack-file]
465
466 command-list = PKT-LINE(command NUL capability-list LF)
467 *PKT-LINE(command LF)
468 flush-pkt
469
470 command = create / delete / update
471 create = zero-id SP new-id SP name
472 delete = old-id SP zero-id SP name
473 update = old-id SP new-id SP name
474
475 old-id = obj-id
476 new-id = obj-id
477
478 pack-file = "PACK" 28*(OCTET)
479 ----
480
481 If the receiving end does not support delete-refs, the sending end MUST
482 NOT ask for delete command.
483
484 The pack-file MUST NOT be sent if the only command used is 'delete'.
485
486 A pack-file MUST be sent if either create or update command is used,
487 even if the server already has all the necessary objects. In this
488 case the client MUST send an empty pack-file. The only time this
489 is likely to happen is if the client is creating
490 a new branch or a tag that points to an existing obj-id.
491
492 The server will receive the packfile, unpack it, then validate each
493 reference that is being updated that it hasn't changed while the request
494 was being processed (the obj-id is still the same as the old-id), and
495 it will run any update hooks to make sure that the update is acceptable.
496 If all of that is fine, the server will then update the references.
497
498 Report Status
499 -------------
500
501 After receiving the pack data from the sender, the receiver sends a
502 report if 'report-status' capability is in effect.
503 It is a short listing of what happened in that update. It will first
504 list the status of the packfile unpacking as either 'unpack ok' or
505 'unpack [error]'. Then it will list the status for each of the references
506 that it tried to update. Each line is either 'ok [refname]' if the
507 update was successful, or 'ng [refname] [error]' if the update was not.
508
509 ----
510 report-status = unpack-status
511 1*(command-status)
512 flush-pkt
513
514 unpack-status = PKT-LINE("unpack" SP unpack-result LF)
515 unpack-result = "ok" / error-msg
516
517 command-status = command-ok / command-fail
518 command-ok = PKT-LINE("ok" SP refname LF)
519 command-fail = PKT-LINE("ng" SP refname SP error-msg LF)
520
521 error-msg = 1*(OCTECT) ; where not "ok"
522 ----
523
524 Updates can be unsuccessful for a number of reasons. The reference can have
525 changed since the reference discovery phase was originally sent, meaning
526 someone pushed in the meantime. The reference being pushed could be a
527 non-fast-forward reference and the update hooks or configuration could be
528 set to not allow that, etc. Also, some references can be updated while others
529 can be rejected.
530
531 An example client/server communication might look like this:
532
533 ----
534 S: 007c74730d410fcb6603ace96f1dc55ea6196122532d refs/heads/local\0report-status delete-refs ofs-delta\n
535 S: 003e7d1665144a3a975c05f1f43902ddaf084e784dbe refs/heads/debug\n
536 S: 003f74730d410fcb6603ace96f1dc55ea6196122532d refs/heads/master\n
537 S: 003f74730d410fcb6603ace96f1dc55ea6196122532d refs/heads/team\n
538 S: 0000
539
540 C: 003e7d1665144a3a975c05f1f43902ddaf084e784dbe 74730d410fcb6603ace96f1dc55ea6196122532d refs/heads/debug\n
541 C: 003e74730d410fcb6603ace96f1dc55ea6196122532d 5a3f6be755bbb7deae50065988cbfa1ffa9ab68a refs/heads/master\n
542 C: 0000
543 C: [PACKDATA]
544
545 S: 000eunpack ok\n
546 S: 0018ok refs/heads/debug\n
547 S: 002ang refs/heads/master non-fast-forward\n
548 ----