GNU bug report logs - #67072
[PATCH 0/4] Helping diagnose substitute setup issues

Previous Next

Full log

Message #26 received at 67072 <at> debbugs.gnu.org (full text, mbox):

From: Simon Tournier <zimon.toutoune <at> gmail.com>
To: Ludovic Courtès <ludo <at> gnu.org>, 67072 <at> debbugs.gnu.org
Cc: Josselin Poiret <dev <at> jpoiret.xyz>, Mathieu Othacehe <othacehe <at> gnu.org>,
 Ludovic Courtès <ludo <at> gnu.org>,
 Tobias Geerinckx-Rice <me <at> tobias.gr>, Ricardo Wurmus <rekado <at> elephly.net>,
 Christopher Baines <guix <at> cbaines.net>
Subject: Re: [bug#67072] [PATCH 4/4] weather: Report unauthorized substitute
 servers.
Date: Tue, 28 Nov 2023 14:14:23 +0100

Hi,

On Sat, 11 Nov 2023 at 12:06, Ludovic Courtès <ludo <at> gnu.org> wrote:

> +  #:use-module (guix pki)

Looking at what it drags, I notice:

--8<---------------cut here---------------start------------->8---
(define* (authorized-key? key #:optional (acl (current-acl)))
  "Return #t if KEY (a canonical sexp) is an authorized public key for archive
imports according to ACL."
  ;; Note: ACL is kept in native sexp form to make 'authorized-key?' faster,
  ;; by not having to convert it with 'canonical-sexp->sexp' on each call.
  ;; TODO: We could use a better data type for ACLs.
  (let ((key (canonical-sexp->sexp key)))
    (match acl
      (('acl
        ('entry subject-keys
                ('tag ('guix 'import)))
        ...)
       (not (not (member key subject-keys))))
      (_
       (error "invalid access-control list" acl)))))
--8<---------------cut here---------------end--------------->8---

I know it is irrelevant with the patch at hand.  Maybe not. :-)

   1. Why this ’(not (not’ ?

   2. When testing the patch, I have not done --sysconfdir=/etc and it
      was not able to find the correct ACL.  Somehow…

> +(define (check-narinfo-authorization narinfo)
> +  "Print a warning when NARINFO is not signed by an authorized key."
> +  (unless (valid-narinfo? narinfo)

…I entered in this part – hence the look up (guix pki) ;-).  Well, my
mistake is hard to reproduce outside of Guix development tree but
’valid-narinfo?’ returns false for more cases than just
unauthorized-key.  Therefore, the hint could be misleading.

Since we are discussing about an helper, I would run ’signature-case’
here in check-narinfo.  For example, if the case is 'unauthorized-key,
then I would check is %acl-file exists.  Maybe display the full
%acl-file explaining that the key is not in, etc.

Moreover, running “guix challenge coreutils” does not warn about
anything; when I was expected the same warning as “guix weather”.


Last, once sysconfig fixed, I get:

--8<---------------cut here---------------start------------->8---
guix weather: warning: could not determine current substitute URLs; using defaults
computing 1 package derivations for x86_64-linux...
looking for 2 store items on https://ci.guix.gnu.org...
guix weather: error: open-file: Permission denied: "/etc/guix/acl"
--8<---------------cut here---------------end--------------->8---

Hum? Maybe I am doing something wrong…  The file /etc/guix/acl has the
permission:

    -rw-------   1 root root   528  acl

Is it incorrect?  Well, if all are allowed to read (chmod a+r) then
there is not error.  And it displays the warning:

--8<---------------cut here---------------start------------->8---
guix weather: warning: could not determine current substitute URLs; using defaults
--8<---------------cut here---------------end--------------->8---

And that’s because the daemon is not supporting the operation.  This
warning appears to me misleading: personally I think that I am
misconfigured something when that’s not the case.  Instead, I would
display:

    warning: using defaults substitute URLs


Cheers,
simon

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.