Hi,

Pier Antonio Corradini wrote:

Autenticity control (gpg --verify SHA512SUMS.sign SHA512SUMS.txt):
[...]
gpg:                utilizzando la chiave RSA DF9B9C49EAA9298432589D76DA87E80D6294BE9B
gpg: Firma BAD da "Debian CD signing key <debian-cd@lists.debian.org>"

I assume that "Firma BAD" means bad signature.

I get a different result:

wget https://cdimage.debian.org/debian-cd/current/amd64/iso-cd/SHA512SUMS
wget https://cdimage.debian.org/debian-cd/current/amd64/iso-cd/SHA512SUMS.sign
gpg --verify SHA512SUMS.sign SHA512SUMS

yields

gpg: Good signature from "Debian CD signing key <debian-cd@lists.debian.org>"

The MD5 and SHA512 sum of SHA512SUMS are
ad6d99aab91a4ec4d3293afc0dfbd95d
36bf1f16bc4b9795122b7b3542a32f34c3be0ef294ff3a8bf43232df6554b69b569fe15d93c79ee48a47902e1a6ad87ca9966988cd4bf9db684f7dd7eda7813a

The ones of SHA512SUMS.sign are
fb3d950c9472f35bd06add950ccfe991
0095bd988c97a7bd0400704ffd3d0fe64a33057b5eaed7530973fac4e039cc366bc5c144413cdb48a591fa5a5d9bd8240721d797964ca453b5981d90ed8e1a13

So which one of your downloaded SHA512SUMS* files deviates from these ?


Have a nice day :)

Thomas

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Hi,

(Please Cc: debian-user@lists.debian.org with your replies.
I sent my mail with Cc; to you, because the X-Spam-Status: header of
your list mail did not indicate that you are subscribed to the list.)


Pier Antonio Corradini wrote:

So... first step:
PS C:\Users\CP> gpg --keyserver hkps://keyring.debian.org --recv-keys DF9B9C49EAA9298432589D76DA87E80D6294BE9B
[...]
Second step:
PS C:\Users\CP\Documents\Linux\Debian12.10.0\VersioneHTTP>
gpg --verify SHA512SUMS.sign SHA512SUMS.txt
gpg: Firma effettuata 03/15/25 21:33:08 ora solare Europa occidentale
gpg:                utilizzando la chiave RSA DF9B9C49EAA9298432589D76DA87E80D6294BE9B

Did this second-step run succeed ?
It seems that the decisive message line is missing.
(Not that i would really understand the language in the messages.)

third step:
PS C:\Users\CP\Documents\Linux\Debian12.10.0\VersioneHTTP> gpg --output debian-key.asc --export DF9B9C49EAA9298432589D76DA87E80D6294BE9B
Il file 'debian-key.asc' esiste. Sovrascrivere? (y/N) y
to import it inside Kleopatra...and... bad signature inside Kleopatra!

Is Kleopatra
https://www.openpgp.org/software/kleopatra/
?

Inside console:
PS C:\Users\CP\Documents\Linux\Debian12.10.0\VersioneHTTP> gpg --verify SHA512SUMS.sign SHA512SUMS.txt
gpg: Firma effettuata 03/15/25 21:33:08 ora solare Europa occidentale
gpg:                utilizzando la chiave RSA DF9B9C49EAA9298432589D76DA87E80D6294BE9B
gpg: Firma BAD da "Debian CD signing key <debian-cd@lists.debian.org>" [sconosciuto]

If the result is different from the one of step 2:
What has changed since step 2 ?

Where is the mistake?

I have to repeat my questions for checksums of the non-matching files
SHA512SUMS.sign
SHA512SUMS
(I could make use of MD5, SHA256, or SHA512.)

Reason for this request:
If gpg --verify reports mismatch, then either:
- at least one of the two files must show differences to those which i
downloaded yesterday,
- or something is broken with you gpg --verify run.
Knowing whether the files SHA512SUMS.sign and SHA512SUMS are the same
as the copies on my computer would enable us to judge which of the
above reasons is present.


Have a nice day :)

Thomas

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Hi,

Pier Antonio Corradini wrote:

3D0BA303805111F651A88D96FC64867FFC678E43F3756F5F91B24A810D91015E459... C:\Users\CP\Documents\Linux\Debian12.10.0\VersioneHTTP\SHA512SUMS.txt

I get

36bf1f16bc4b9795122b7b3542a32f34c3be0ef294ff3a8bf43232df6554b69b569fe15d93c79ee48a47902e1a6ad87ca9966988cd4bf9db684f7dd7eda7813a

from
sha512sum SHA512SUMS

So your SHA512SUMS.txt has not the same as my SHA512SUMS download of
yesterday.

 58B5434926A9E5F7BA27FA32CD19B4379658945646549D6ACD3EC9A9368FFFACDAC... C:\Users\CP\Documents\Linux\Debian12.10.0\VersioneHTTP\SHA512SUMS.sign

I get

0095bd988c97a7bd0400704ffd3d0fe64a33057b5eaed7530973fac4e039cc366bc5c144413cdb48a591fa5a5d9bd8240721d797964ca453b5981d90ed8e1a13

from
sha512sum SHA512SUMS.sign

So our respective files with name SHA512SUMS.sign differ by content,
too.


We now know that local processing, downloading or malicious activities
altered the signature file and the SHA512SUMS file.
Malice would have to be suspected if the listed checksums in your file SHA512SUMS.txt would differ from those in my downloaded copy.

My downloaded SHA512SUMS file has this content: -----------------------------------------------------------------------

e0bd9ba03084a6fd42413b425a2d20e3731678a31fe5fb2cc84f79332129afca2ad4ec897b4224d6a833afaf28a5d938b0fe5d680983182944162c6825b135ce debian-12.7.0-amd64-netinst.iso
915ab697472fd9a25a6b7b5d4988ee659fed61cd6dc6cd990435971af5894fca82426f213913fd95cce04de8d10e0ee709023b677d02d5c48062208ff5ab3112 debian-edu-12.7.0-amd64-netinst.iso
d9480c2d765f3b1ebe8e7d06b1cf6ecf30b95146d5c2036f20904957db6139a440f9f8e7f4f901da6a02f810f2b3ab660aea56d99778c647c62386a2082c9407 debian-mac-12.7.0-amd64-netinst.iso

-----------------------------------------------------------------------

So what does your SHA512.txt say ?


=======================================================================
Not of importance for the problem any more. Just for understanding:

Pier Antonio Corradini wrote:

gpg: Firma effettuata 03/15/25 21:33:08 ora solare Europa occidentale gpg:                utilizzando la chiave RSA DF9B9C49EAA9298432589D76DA87E80D6294BE9B

I wrote:

Did this second-step run succeed ?
It seems that the decisive message line is missing.

Pier Antonio Corradini wrote:

What message?

Something like
gpg: Firma BAD da "Debian CD signing key <debian-cd@lists.debian.org>"
or
gpg: Firma Good da "Debian CD signing key <debian-cd@lists.debian.org>"
or whatever your gpg --verify says about matching signature.

Whatever, we now know that the files SHA512SUMS.txt and SHA512SUMS.sign
are altered, which explains why they do not match.

=======================================================================


Have a nice day :)

Thomas

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Hi,

i realize that i posted the content of the wrong SHA512SUMS file.
The one i posted was from debian 12.7.0.
Nevertheless the SHA512 sums which i posted earlier are of the files
from 12.10.0 which i downloaded yesterday.


Pier Antonio Corradini wrote:

The content of these links, seen now, is the following: cb089def0684fd93c9c2fbe45fd16ecc809c949a6fd0c91ee199faefe7d4b82b64658a264a13109d59f1a40ac3080be2f7bd3d8bf3e9cdf509add6d72576a79b  debian-12.10.0-amd64-netinst.iso
71d4c4e2ea7b617362875a74eb007308ae577ebe4b02ffeb626f1d12eaf412567d1d1816dbdbbb84cfaa38a205c13abf317ec227e5b2df9c982979698909889c  debian-edu-12.10.0-amd64-netinst.iso
269e64d2a379429905cf95191036cc53fdc148c624af68386d3a238f5fe2c5b03e3732706eaac175303b1fe327f691dc50faf8d65665781d6bcbbabf072559fa  debian-mac-12.10.0-amd64-netinst.iso

These checksums match what i see in my downloaded SHA512SUMS file of debian-12.10.0 netinst. (Not the one from 12.7.0.)

So if the check run from your initial mail indicates a matching SHA512
checksum of the .iso file in the SHA512SUMS file and if you believe my
word, then your ISO image is good.

The trust in my word could be replaced by unaltered files SHA512SUM
and SHA512SUM.sign and a successful gpg --verify run.
But i cannot give advice how to achieve this in a MS-Windows
environment.


Have a nice day :)

Thomas

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Hi,

Pier Antonio Corradini wrote:

gpg: Firma valida da "Debian CD signing key <debian-cd@lists.debian.org>"

Jay ! \o/

gpg: ATTENZIONE: questa chiave non è certificata con una firma fidata!
gpg:          Non ci sono indicazioni che la firma appartenga al proprietario.

Regrettably gpg still assumes a web of trust to be normal, woven by
people who meet in person and founded on certifications by the most
reputed members of the community: Ada Lovelace, Alan Turing,
Dennis M. Ritchie.
But they are all dead ... and i don't feel so well either.

So the keys have no VIP signatures. Their armor is the fact that their fingerprints get mentioned in many e-mails. :))

Impronta digitale della chiave primaria: DF9B 9C49 EAA9 2984 3258  9D76 DA87 E80D 6294 BE9B

This last line is the important information.
It has to match one of the three keys which are published at
https://www.debian.org/CD/verify

In our case it's the midlle one:

pub rsa4096/DA87E80D6294BE9B 2011-01-05 [SC]
Key fingerprint = DF9B 9C49 EAA9 2984 3258 9D76 DA87 E80D 6294 BE9B
uid Debian CD signing key <debian-cd@lists.debian.org>

Conclusion: I am really very perplexed by the outcome of this authenticity check of the file debian-12.10.0-amd64-netinst.iso: all these commands and attempts to arrive at what? To the sentence "WARNING: this key is not certified with a trusted signature!There is no indication that the signature belongs to the owner."????

Indeed. You have to derive your trust from the hope that the
fingerprint cannot be forged, so that no fake key can sign the
SHA512SUMS file by a SHA512SUMS.sign file while bearing that
fingerprint.

Or is there still something to clarify regarding the selection of the key?

No. You did it perfectly right after you surpassed the obstacle of
invisible alterations to the SHA512SUM* files.

I agree that the procedure is ugly and appears too complicated.
But it works with widely available tools and is flexible enough to
deal with peculiarities of media types when the ISO is already on
a bootable medium.
See this wiki page (which is too long, of course):
https://wiki.debian.org/VerifyISOImage

It might be helpful to have specialized verifier programs which hide
the dirty details. But then you'd have to trust these programs
additionally to the other involved parties. So they would need to be authenticated themselves by gpg --verify or alike.

If i'd knew a good solution then i'd post it to
debian-cd@lists.debian.org and be obstinate until i get a good answer.
But as it is, i cannot do more than offer human help and above wiki
page.


Have a nice day :)

Thomas

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Hi,

Pier Antonio Corradini wrote:

So now the authenticity check is complete and the authenticity is completely sure?

Yes. Until a quantum computer cracks the riddle how to generate an own
key with the same fingerprint.
(There are other risks, too, which are not prevented by signature with uncompromised keys. See the classic https://xkcd.com/2347/ )

Now we are done and look like real C programmers: Dirty but contented.


Have a nice day :)

Thomas

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)