Re: PGP sigs fail verification

Top Page
Message as email
+ (text/plain)
Delete this message
Author: Ian Zimmerman
To: mutt-users
Subject: Re: PGP sigs fail verification
On 2016-09-19 22:25, Miroslav Rovis wrote:

> $ gpg --verify  dNZQNRnu_DarakMarjal160907-raw_QkYBXROR.sig \
>     dNZQNRnu_DarakMarjal160907-raw.asc 
> gpg: Signature made Wed 07 Sep 2016 12:21:36 CEST using RSA key ID
> 48C912E7
> gpg: BAD signature from "Paul Saunders <darac@???>"
> $

You're a victim of the same misunderstanding as I was, when I tried to
investigate the problem this way :-P

You need to read RFC 3156, which specifies how the signature is computed
on PGP/MIME mails. It is _not_ on the data you see when you dump the
text into a Unix file (even when you take into account the encoding such
as quoted-printable).

Here are at least 3 differences: (there may be more)

1. Line endings: all transformed into CRLF before signing

2. Trailing whitespace: all stripped before signing

3. MIME part headers (ie. the stuff after the MIME boundary line and
before the first empty line after that): included in signed data

So, if we want to pursue this line of verifying from the command line,
first we need a piece of code or script that will take an email and spit
out the data _as used for the signature computation_. I think it ought
to exist out there. That is my next step.

As I reported in other subthread, I took one "BAD" email from my system
(directly from the maildir, not exporting with mutt) and compared it to
the archived copy from MARC. They were identical. At least this way I
eliminated the possibility of mangling by intermediate MTAs.

For my part I now think this is a flea.

Please *no* private Cc: on mailing lists and newsgroups
Why does the arrow on Hillary signs point to the right?