unofficial mirror of bug-guix@gnu.org 
 help / color / mirror / code / Atom feed
From: "Ludovic Courtès" <ludo@gnu.org>
To: Ricardo Wurmus <rekado@elephly.net>
Cc: 22883@debbugs.gnu.org
Subject: bug#22883: Authenticating a Git checkout
Date: Fri, 01 May 2020 19:04:41 +0200	[thread overview]
Message-ID: <87wo5vfuxi.fsf@gnu.org> (raw)
In-Reply-To: <87bln9oupo.fsf@gnu.org> ("Ludovic \=\?utf-8\?Q\?Court\=C3\=A8s\=22'\?\= \=\?utf-8\?Q\?s\?\= message of "Thu, 30 Apr 2020 17:32:19 +0200")

Hey!

Ludovic Courtès <ludo@gnu.org> skribis:

>   • Load the keyring from files in the repo, possibly in a dedicated
>     branch.
>
>   • Load the list of authorized keys from the parent of the commit being
>     authenticated.

Done!

  8916c2fa32 git-authenticate: Load the keyring from the repository.
  6960064ddc git-authenticate: Load the list of authorized keys from the tree.
  f145a2d1a9 .guix-authorizations: Augment.
  62ae43db19 git-authenticate: Use (guix openpgp).

‘git-authenticate’ now loads the keyring from the “keyring” branch,
which I’ve just pushed as an “orphan” branch:

  https://git.savannah.gnu.org/cgit/guix.git/?h=keyring

So no need to store the keyring out-of-band, to spawn gpg to fetch keys
from somewhere else, etc.  The idea is that we’ll keep adding new keys
to this branch every time a new committer joins.  We would never remove
keys from there because those keys are necessary to verify signatures.
The fact that a key is present on that branch does _not_ mean that it
designates an authorized committer today.

The list of authorized committers is meant to be stored in a
‘.guix-authorizations’ file in each branch of the channel.  It is
essentially a list of fingerprints:

  https://git.savannah.gnu.org/cgit/guix.git/commit/?h=wip-openpgp&id=f145a2d1a982cc841c7ccae3334d4783dad24a1e

To accept a new committer, an authorized committer must add its key to
this file in the branch(es) where that person is expected to commit.
The format currently accepts additional data for each fingerprint.  It’s
currently ignored, but I thought it could be useful in the future, for
instance if we want to associate a file pattern with a key.

A commit is considered “authorized” if and only if its signing key is
listed in the ‘.guix-authorizations’ file of its parent commit(s).

In ‘git-authenticate’, this is implemented in a naive unoptimized way,
but it turns out to make no noticeable difference on the wall-clock time
to authenticate those 14K+ commits.  The crux of the authorization
mechanism is this procedure:

  (define* (commit-authorized-keys repository commit
                                   #:optional (default-authorizations '()))
    "Return the list of OpenPGP fingerprints authorized to sign COMMIT, based on
  authorizations listed in its parent commits.  If one of the parent commits
  does not specify anything, fall back to DEFAULT-AUTHORIZATIONS."
    …)

Feedback welcome!

Ludo’.




  parent reply	other threads:[~2020-05-01 17:05 UTC|newest]

Thread overview: 78+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-03-02 18:03 bug#22883: Trustable "guix pull" Christopher Allan Webber
2016-03-02 19:26 ` Leo Famulari
2016-03-02 21:07   ` Christopher Allan Webber
2016-04-25 22:25 ` Ludovic Courtès
2016-04-26  0:13   ` Leo Famulari
2016-04-26  0:17     ` Thompson, David
2016-04-26  7:12       ` Ludovic Courtès
2016-04-30  4:43     ` Mike Gerwitz
2016-06-03 16:12       ` bug#22883: Authenticating a Git checkout Ludovic Courtès
2016-06-03 20:17         ` Leo Famulari
2016-06-04 11:04           ` Ludovic Courtès
2016-06-04  4:24         ` Mike Gerwitz
2016-06-04 11:17           ` Ludovic Courtès
2016-06-04 12:45             ` ng0
2016-06-06 12:20               ` ng0
2016-06-04 16:14             ` Mike Gerwitz
2016-06-05 20:39             ` Christopher Allan Webber
2016-06-05 21:15               ` Leo Famulari
2016-06-06  2:41               ` Mike Gerwitz
2016-06-06  7:01                 ` Ludovic Courtès
2016-07-22  8:22         ` Ludovic Courtès
2016-07-22 12:58           ` Thompson, David
2016-07-22 13:58             ` Ludovic Courtès
2017-10-24 23:30           ` Ludovic Courtès
2019-12-27 19:48             ` Ricardo Wurmus
2019-12-28 14:47               ` Ludovic Courtès
2019-12-28 16:05                 ` Ricardo Wurmus
2019-12-28 17:45                   ` Ludovic Courtès
2020-04-30 15:32                 ` Ludovic Courtès
2020-05-01 15:46                   ` Justus Winter
2020-05-01 16:50                     ` Ludovic Courtès
2020-05-01 17:04                   ` Ludovic Courtès [this message]
2020-05-19 20:23                     ` Ludovic Courtès
2020-06-01 14:07                       ` bug#22883: Channel introductions Ludovic Courtès
2020-06-02 23:45                         ` zimoun
2020-06-03  9:50                           ` Ludovic Courtès
2020-06-03 16:20                             ` zimoun
2020-06-04  9:55                               ` Ludovic Courtès
2020-05-01 17:20                   ` bug#22883: Authenticating a Git checkout Ludovic Courtès
2020-05-02 22:02                   ` Ludovic Courtès
2020-05-04  8:03                     ` Ludovic Courtès
2016-06-01 16:47   ` bug#22883: Discussion of TUF in the context of Git checkout authentication Ludovic Courtès
2016-05-15 12:40 ` bug#22883: Trustable "guix pull" fluxboks
2016-05-16 17:55   ` Thompson, David
2016-05-17 21:19   ` Ludovic Courtès
2016-06-04 16:19 ` Werner Koch
2016-06-04 22:27   ` Ludovic Courtès
2016-06-05  7:51     ` Werner Koch
2016-06-06 21:01       ` Leo Famulari
2016-06-07  8:08         ` bug#22883: gpg2 vs. gpg Ludovic Courtès
2016-06-07 11:25           ` Werner Koch
2016-06-07 12:58             ` ng0
2016-06-05  1:43   ` bug#22883: Trustable "guix pull" Mike Gerwitz
2018-08-28 19:56 ` Vagrant Cascadian
2018-09-02 16:05   ` Ludovic Courtès
2018-09-02 17:15     ` Vagrant Cascadian
2018-09-02 20:07       ` Ludovic Courtès
2019-12-20 22:11         ` bug#22883: Authenticating Git checkouts: step #1 Ludovic Courtès
     [not found]         ` <87mubmodfb.fsf_-_@gnu.org>
2019-12-21  1:33           ` zimoun
2019-12-27 12:58           ` Ludovic Courtès
     [not found]           ` <87eewqgc1v.fsf@gnu.org>
2019-12-27 20:47             ` Ricardo Wurmus
     [not found]             ` <87o8vto5rl.fsf@elephly.net>
2019-12-29  2:45               ` Vagrant Cascadian
     [not found]               ` <87a77bzw6p.fsf@yucca>
2019-12-29  7:34                 ` Efraim Flashner
2019-12-30 21:29                 ` Ludovic Courtès
2019-12-31 19:16 ` Jakub Kądziołka
2020-01-08 13:30   ` Ludovic Courtès
2020-06-02 13:49 ` bug#22883: Authenticating a Git checkout John Soo
2020-06-03  9:33   ` Ludovic Courtès
2020-06-08 21:54 ` bug#22883: [PATCH 1/9] git-authenticate: Cache takes a key parameter Ludovic Courtès
2020-06-08 21:54   ` bug#22883: [PATCH 2/9] git-authenticate: 'authenticate-commits' takes a #:keyring parameter Ludovic Courtès
2020-06-08 21:54   ` bug#22883: [PATCH 3/9] tests: Move OpenPGP helpers to (guix tests gnupg) Ludovic Courtès
2020-06-08 21:54   ` bug#22883: [PATCH 4/9] channels: 'latest-channel-instance' authenticates Git checkouts Ludovic Courtès
2020-06-08 21:54   ` bug#22883: [PATCH 5/9] channels: Make 'validate-pull' call right after clone/pull Ludovic Courtès
2020-06-08 21:54   ` bug#22883: [PATCH 6/9] .guix-channel: Add 'keyring-reference' Ludovic Courtès
2020-06-08 21:54   ` bug#22883: [PATCH 7/9] channels: Automatically add introduction for the official 'guix' channel Ludovic Courtès
2020-06-08 21:54   ` bug#22883: [PATCH 8/9] pull: Add '--disable-authentication' Ludovic Courtès
2020-06-08 21:54   ` bug#22883: [PATCH 9/9] DROP? channels: Add prehistorical authorizations to <channel-introduction> Ludovic Courtès
2020-06-08 22:04   ` bug#22883: [PATCH 1/9] git-authenticate: Cache takes a key parameter Ludovic Courtès

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://guix.gnu.org/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=87wo5vfuxi.fsf@gnu.org \
    --to=ludo@gnu.org \
    --cc=22883@debbugs.gnu.org \
    --cc=rekado@elephly.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
Code repositories for project(s) associated with this public inbox

	https://git.savannah.gnu.org/cgit/guix.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).