From mboxrd@z Thu Jan 1 00:00:00 1970
Return-Path: <f.ebner@proxmox.com>
Received: from firstgate.proxmox.com (firstgate.proxmox.com [212.224.123.68])
(using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits)
key-exchange X25519 server-signature RSA-PSS (2048 bits))
(No client certificate requested)
by lists.proxmox.com (Postfix) with ESMTPS id D8BE562094
for <pve-devel@lists.proxmox.com>; Wed, 19 Jan 2022 13:17:01 +0100 (CET)
Received: from firstgate.proxmox.com (localhost [127.0.0.1])
by firstgate.proxmox.com (Proxmox) with ESMTP id CDB02184C7
for <pve-devel@lists.proxmox.com>; Wed, 19 Jan 2022 13:16:31 +0100 (CET)
Received: from proxmox-new.maurer-it.com (proxmox-new.maurer-it.com
[94.136.29.106])
(using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits)
key-exchange X25519 server-signature RSA-PSS (2048 bits))
(No client certificate requested)
by firstgate.proxmox.com (Proxmox) with ESMTPS id 62B5D184BA
for <pve-devel@lists.proxmox.com>; Wed, 19 Jan 2022 13:16:30 +0100 (CET)
Received: from proxmox-new.maurer-it.com (localhost.localdomain [127.0.0.1])
by proxmox-new.maurer-it.com (Proxmox) with ESMTP id 30CFF4506F
for <pve-devel@lists.proxmox.com>; Wed, 19 Jan 2022 13:16:30 +0100 (CET)
Message-ID: <8735485c-517c-e70f-02f7-3a970b27f575@proxmox.com>
Date: Wed, 19 Jan 2022 13:16:24 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
Thunderbird/91.4.1
Content-Language: en-US
To: =?UTF-8?Q?Fabian_Gr=c3=bcnbichler?= <f.gruenbichler@proxmox.com>,
pve-devel@lists.proxmox.com
References: <20211222135257.3242938-1-f.gruenbichler@proxmox.com>
<20211222135257.3242938-7-f.gruenbichler@proxmox.com>
<a133e124-913d-956b-c66c-d8a8d1f03332@proxmox.com>
<1642582327.0fpxeqo696.astroid@nora.none>
From: Fabian Ebner <f.ebner@proxmox.com>
In-Reply-To: <1642582327.0fpxeqo696.astroid@nora.none>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
X-SPAM-LEVEL: Spam detection results: 0
AWL 0.136 Adjusted score from AWL reputation of From: address
BAYES_00 -1.9 Bayes spam probability is 0 to 1%
KAM_DMARC_STATUS 0.01 Test Rule for DKIM or SPF Failure with Strict Alignment
NICE_REPLY_A -0.001 Looks like a legit reply (A)
SPF_HELO_NONE 0.001 SPF: HELO does not publish an SPF Record
SPF_PASS -0.001 SPF: sender matches SPF record
Subject: Re: [pve-devel] [PATCH v3 proxmox-websocket-tunnel 3/4] add
fingerprint validation
X-BeenThere: pve-devel@lists.proxmox.com
X-Mailman-Version: 2.1.29
Precedence: list
List-Id: Proxmox VE development discussion <pve-devel.lists.proxmox.com>
List-Unsubscribe: <https://lists.proxmox.com/cgi-bin/mailman/options/pve-devel>,
<mailto:pve-devel-request@lists.proxmox.com?subject=unsubscribe>
List-Archive: <http://lists.proxmox.com/pipermail/pve-devel/>
List-Post: <mailto:pve-devel@lists.proxmox.com>
List-Help: <mailto:pve-devel-request@lists.proxmox.com?subject=help>
List-Subscribe: <https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel>,
<mailto:pve-devel-request@lists.proxmox.com?subject=subscribe>
X-List-Received-Date: Wed, 19 Jan 2022 12:17:01 -0000
Am 19.01.22 um 11:34 schrieb Fabian Grünbichler:
> On January 4, 2022 12:37 pm, Fabian Ebner wrote:
>> Am 22.12.21 um 14:52 schrieb Fabian Grünbichler:
>>> in case we have no explicit fingerprint, we use openssl's regular "PEER"
>>> verification. if we have a fingerprint, we ignore openssl altogether and
>>> just verify the fingerprint of the presented leaf certificate.
>>>
>>> Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
>>> ---
>>>
>>> Notes:
>>> v3: switch to using hex instead of no-longer-existing digest_to_hex
>>> v2: new
>>>
>>> src/main.rs | 47 ++++++++++++++++++++++++++++++++++++++++++++---
>>> 1 file changed, 44 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/src/main.rs b/src/main.rs
>>> index 582214c..49d6ffe 100644
>>> --- a/src/main.rs
>>> +++ b/src/main.rs
>>> @@ -134,9 +134,50 @@ impl CtrlTunnel {
>>> }
>>>
>>> let mut ssl_connector_builder = SslConnector::builder(SslMethod::tls())?;
>>> - if fingerprint.is_some() {
>>> - // FIXME actually verify fingerprint via callback!
>>> - ssl_connector_builder.set_verify(openssl::ssl::SslVerifyMode::NONE);
>>> + if let Some(expected) = fingerprint {
>>> + ssl_connector_builder.set_verify_callback(
>>> + openssl::ssl::SslVerifyMode::NONE,
>>> + move |_valid, ctx| {
>>> + let cert = match ctx.current_cert() {
>>> + Some(cert) => cert,
>>> + None => {
>>> + eprintln!("SSL context lacks current certificate.");
>>> + return false;
>>> + }
>>> + };
>>> +
>>> + let depth = ctx.error_depth();
>>> + if depth != 0 {
>>> + return true;
>>> + }
>>
>> Sorry about my ignorance. Does using SslVerifyMode::NONE imply that
>> there is an error? At depth 0? Why is it fine to return true if not?
>
> this is a bit.. tricky (did I mention I really really dislike openssl's
> API? ;))
>
> basically what we do in this branch (if we have a pinned fingerprint to
> check - the regular 'connect iff trusted by system' is the else branch
> below) we set our own callback that gets called for each cert along the
> chain (starting at the top, ending with the leaf/end certificate, but
> the order is not relevant since a single failed callback fails the whole
> verification).
>
> for each cert (== element of the chain == depth value) we get the result
> of openssl's check (`_valid`) and the X509 store context (ctx).
>
> the context (among other things ;)) contains information about where
> (depth) in the chain we currently are:
> - depth 0 == peer certificate (the one we are interested in)
> - depth 1 == CA certificate (signer of peer cert, not interesting)
> - depth 2 == higher CA certificate (signer of CA at 1, not interesting)
> - depth X == higher CA certificate (signer of CA at X-1, not
> interesting)
>
> all but the peer certificate are optional (peer could give us just a
> self-signed certificate, or an incomplete chain).
>
> that the methods here are all referring to 'error' is an OpenSSL
> peculiarity - it basically gives us a cert store with the current cert
> and error depth set to values that are valid if we fail (error) the
> verification.
>
> for each cert/call we do the following:
>
> - ensure there is a current cert in the context or fail verification
> - continue verification with next element of the chain if we are not
> (yet) at the peer certificate (depth != 0)
> - calculate fingerprint for current (== peer) cert, or fail
> - compare fingerprint with pinned/expected one, fail if not expected
>
> since the verification fails as soon as single callback fails, we need
> to
> - return false if we fail some assumption (like ctx having a current
> cert, or being able to calculate a cert's FP)
> - return true if the current call is at a depth we are not interested in
> verifying
> - return true/false depending on result of FP check if current call is at
> a depth we are interested in
>
> I'll add a comment to the depth part that it is for skipping the CA
> certs! also verify mode should technically be PEER, so I'll fix that up
> as well.
>
Thanks for the thorough explanation! I think I get it now (especially
knowing that the callback can be called multiple times helps a lot).
>>
>>> +
>>> + let fp = match cert.digest(openssl::hash::MessageDigest::sha256()) {
>>> + Ok(fp) => fp,
>>> + Err(err) => {
>>> + // should not happen
>>> + eprintln!("failed to calculate certificate FP - {}", err);
>>> + return false;
>>> + }
>>> + };
>>> + let fp_string = hex::encode(&fp);
>>> + let fp_string = fp_string
>>> + .as_bytes()
>>> + .chunks(2)
>>> + .map(|v| std::str::from_utf8(v).unwrap())
>>> + .collect::<Vec<&str>>()
>>> + .join(":");
>>> +
>>> + let expected = expected.to_lowercase();
>>> + if expected == fp_string {
>>> + true
>>> + } else {
>>> + eprintln!("certificate fingerprint does not match expected fingerprint!");
>>> + eprintln!("expected: {}", expected);
>>> + eprintln!("encountered: {}", fp_string);
>>> + false
>>> + }
>>> + },
>>> + );
>>> } else {
>>> ssl_connector_builder.set_verify(openssl::ssl::SslVerifyMode::PEER);
>>> }
>>