git/list[1] front-page[2] threads[3] people[4] search[5] about
 

[PATCH 00/18] Signed push

From
Junio C Hamano <gitster@pobox.com>
Date
Aug 19, 2014, 22:06 UTC
Message-ID
<1408485987-3590-1-git-send-email-gitster@pobox.com>

While signed tags and commits assert that the objects thusly signed came from you, who signed these objects, there is not a good way to assert that you wanted to have a particular object at the tip of a particular branch. My signing v2.0.1 tag only means I want to call the version v2.0.1, and it does not mean I want to push it out to my 'master' branch---it is likely that I only want it in 'maint', so the signature on the object alone is insufficient.

The only assurance to you that 'maint' points at what I wanted to place there comes from your trust on the hosting site and my authentication with it, which cannot easily audited later.

This series introduces a cryptographic assurance for ref updates done by "git push" by introducing a mechanism that allows you to sign a "push certificate" (for the lack of better name) every time you push. Think of it as working on an axis orthogonal to the traditional "signed tags".

The most interesting part starts at 15/18; everything that precedes that patch are preparatory clean-ups.

    [PATCH 15/18] the beginning of the signed push
	This step presents the basic idea.  If you remember the
	underlying "git push" protocol exchange, it goes like this:
	- The server advertises the existing refs and where they
	  point at, and the capabilities the server supports;
	- The "git push" client sends update commands (one or more
	  "old-sha1 new-sha1 refname") followed by the pack data;
	- The server unpacks and updates the refname to point at
	  new-sha1.
	We introduce "push-cert" capability, and allow the client to
	sign the "update commands" it will send to the server and
	send this signature using the new "push-cert" command.
	This certificate is exported to the pre/post-receive hooks
	of the server, so that the pre-receive hook can GPG validate
	(and possibly reject a bad push); post-receive hook can log
	the certificate.
    [PATCH 16/18] receive-pack: GPG-validate push certificates
	This step builds a native GPG validation into the server to
	help the pre-receive hook.  The signature is verified
	against the GPG keychain the server uses (it is likely that
	you would want to set and export GNUPGHOME when starting
	your server), and verification result is given to the
	pre/post-receive hook.
    [PATCH 17/18] send-pack: send feature request on push-cert packet
    [PATCH 18/18] signed push: final protocol update
	With the protocol introduced in 15/18, the update commands
	and the push certificate record the same information twice;
	the protocol was kept inefficient to make it easier to
	review the changes.
	These two steps updates the protocol to the final version,
	which does not to send the update commands when a push
	certificate is in use.

If the server's GPG keychain and pre-receive hook are properly set up, a "git push --signed" over an unauthenticated and unencrypted communication channel (aka "git daemon") can be made as secure as, and even more secure than, the authenticated "git push ssh://".

With the signed push certificate, together with the connectivity check done when the server accepts the packed data, we are assured that the trusted user vouches for the history leading to the proposed tips of refs (aka "new-sha1"s), and a man-in-the-middle would not be able to make the server accept an update altered in transit.

Junio C Hamano (18):
  receive-pack: do not overallocate command structure
  receive-pack: parse feature request a bit earlier
  receive-pack: do not reuse old_sha1[] to other things
  receive-pack: factor out queueing of command
  send-pack: move REF_STATUS_REJECT_NODELETE logic a bit higher
  send-pack: refactor decision to send update per ref
  send-pack: always send capabilities
  send-pack: factor out capability string generation
  send-pack: rename "new_refs" to "need_pack_data"
  send-pack: refactor inspecting and resetting status and sending commands
  send-pack: clarify that cmds_sent is a boolean
  gpg-interface: move parse_gpg_output() to where it should be
  gpg-interface: move parse_signature() to where it should be
  pack-protocol doc: typofix for PKT-LINE
  the beginning of the signed push
  receive-pack: GPG-validate push certificates
  send-pack: send feature request on push-cert packet
  signed push: final protocol update
 Documentation/git-push.txt                        |   9 +-
 Documentation/git-receive-pack.txt                |  30 +++-
 Documentation/technical/pack-protocol.txt         |  24 ++-
 Documentation/technical/protocol-capabilities.txt |  12 +-
 builtin/push.c                                    |   1 +
 builtin/receive-pack.c                            | 161 +++++++++++++++---
 commit.c                                          |  36 -----
 gpg-interface.c                                   |  57 +++++++
 gpg-interface.h                                   |  18 ++-
 send-pack.c                                       | 188 ++++++++++++++++------
 send-pack.h                                       |   1 +
 t/t5534-push-signed.sh                            |  77 +++++++++
 tag.c                                             |  20 ---
 tag.h                                             |   1 -
 transport.c                                       |   4 +
 transport.h                                       |   5 +
 16 files changed, 502 insertions(+), 142 deletions(-)
 create mode 100755 t/t5534-push-signed.sh
-- 
2.1.0-301-g54593e2
Next: Junio C Hamano
Message 1 of 59 in “Signed push”
  1. 00/18 Signed pushJunio C Hamano, Aug 19, 2014
  2. 01/18 receive-pack: do not overallocate command structureJunio C Hamano, Aug 19, 2014
  3. 02/18 receive-pack: parse feature request a bit earlierJunio C Hamano, Aug 19, 2014
  4. Junio C HamanoAug 19, 2014
  5. 03/18 receive-pack: do not reuse old_sha1[] to other thingsJunio C Hamano, Aug 19, 2014
  6. Junio C HamanoAug 19, 2014
  7. 04/18 receive-pack: factor out queueing of commandJunio C Hamano, Aug 19, 2014
  8. 05/18 send-pack: move REF_STATUS_REJECT_NODELETE logic a bit higherJunio C Hamano, Aug 19, 2014
  9. 06/18 send-pack: refactor decision to send update per refJunio C Hamano, Aug 19, 2014
  10. 07/18 send-pack: always send capabilitiesJunio C Hamano, Aug 19, 2014
  11. 08/18 send-pack: factor out capability string generationJunio C Hamano, Aug 19, 2014
  12. 09/18 send-pack: rename "new_refs" to "need_pack_data"Junio C Hamano, Aug 19, 2014
  13. 10/18 send-pack: refactor inspecting and resetting status and sending commandsJunio C Hamano, Aug 19, 2014
  14. 11/18 send-pack: clarify that cmds_sent is a booleanJunio C Hamano, Aug 19, 2014
  15. 12/18 gpg-interface: move parse_gpg_output() to where it should beJunio C Hamano, Aug 19, 2014
  16. 13/18 gpg-interface: move parse_signature() to where it should beJunio C Hamano, Aug 19, 2014
  17. 14/18 pack-protocol doc: typofix for PKT-LINEJunio C Hamano, Aug 19, 2014
  18. 15/18 the beginning of the signed pushJunio C Hamano, Aug 19, 2014
  19. brian m. carlsonAug 20, 2014
  20. Bert WesargAug 20, 2014
  21. Junio C HamanoAug 20, 2014
  22. 16/18 receive-pack: GPG-validate push certificatesJunio C Hamano, Aug 19, 2014
  23. David TurnerAug 20, 2014
  24. Junio C HamanoAug 20, 2014
  25. David TurnerAug 20, 2014
  26. Junio C HamanoAug 20, 2014
  27. David TurnerAug 21, 2014
  28. Junio C HamanoAug 22, 2014
  29. 17/18 send-pack: send feature request on push-cert packetJunio C Hamano, Aug 19, 2014
  30. 18/18 signed push: final protocol updateJunio C Hamano, Aug 19, 2014
  31. Shawn PearceAug 21, 2014
  32. Junio C HamanoAug 21, 2014
  33. Kyle J. McKayAug 22, 2014
  34. Junio C HamanoAug 22, 2014
  35. Shawn PearceAug 22, 2014
  36. Junio C HamanoAug 25, 2014
  37. Shawn PearceAug 26, 2014
  38. Junio C HamanoAug 26, 2014
  39. Junio C HamanoAug 26, 2014
  40. Junio C HamanoSep 4, 2014
  41. Shawn PearceSep 5, 2014
  42. Junio C HamanoAug 22, 2014
  43. David TurnerAug 22, 2014
  44. Duy NguyenAug 19, 2014
  45. Junio C HamanoAug 19, 2014
  46. Nico WilliamsAug 20, 2014
  47. Junio C HamanoAug 20, 2014
  48. Junio C HamanoAug 20, 2014
  49. Junio C HamanoAug 20, 2014
  50. Nico WilliamsAug 20, 2014
  51. Stefan BellerAug 22, 2014
  52. Junio C HamanoAug 22, 2014
  53. Stefan BellerAug 22, 2014
  54. Junio C HamanoAug 22, 2014
  55. Stefan BellerAug 22, 2014
  56. Junio C HamanoAug 22, 2014
  57. Stefan BellerAug 22, 2014
  58. Junio C HamanoAug 25, 2014
  59. Jason PyeronAug 25, 2014

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.