From: "plaisthos (Code Review)"
Date: Wed, 25 Jun 2025 22:23:31 +0000
In-Reply-To:
References:
Reply-To: arne-openvpn@...1227..., rein.vanbaaren@...1435...,
openvpn-devel@lists.sourceforge.net, frank@...2641...,
gert@...1296...
Message-ID: <280de4cba0726b6455ef0ebf93bca84653571b8d-EmailReviewComments-HTML@...2715...>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Content-Disposition: inline
Content-Type: multipart/alternative; boundary="qLlQ3M/DBlA="; charset=UTF-8
Subject: [Openvpn-devel] [S] Change in openvpn[master]: Added PQE to WolfSSL
List-Id:
List-Unsubscribe: ,
List-Archive:
List-Post:
List-Help:
List-Subscribe: ,
To: comododragon
Cc: flichtenheld , openvpn-devel
--qLlQ3M/DBlA=
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Attention is currently required from: comododragon, flichtenheld.
plaisthos has posted comments on this change. ( http://gerrit.openvpn.net/c=
/openvpn/+/1046?usp=3Demail )
Change subject: Added PQE to WolfSSL
......................................................................
Patch Set 5: Code-Review+1
(3 comments)
Patchset:
PS5:
I tested the patch and it works and is also a lot cleaner than the first ve=
rsion.
There are some documentation updates I would like to see (see the other com=
ments) to ensure users to don't get lost with the difference in OpenSSL vs =
wolfSSL names and defaults.
Also wolfSSL's ML-DSA-87 seems to be a bit wacky for now. It cannot read th=
e private key generated by OpenSSL and when connecting to a server with ml-=
kem and fingerprints it seems not be able to read some of the fields with t=
he current APIs that we use:
[wolfSSL] Control Channel: TLSv1.3, cipher TLSv1.3 TLS_AES_256_GCM_SHA384, =
peer certificate: , signature: ML-DSA 87, peer signing digest/type: SHA512 =
ML-DSA 87
vs
[OpenSSL] Control Channel: TLSv1.3, cipher TLSv1.3 TLS_AES_256_GCM_SHA384, =
peer certificate: 20736 bits ML-DSA-87, signature: id-ml-dsa-87, peer signi=
ng digest/type: mldsa87 id-ml-dsa-87, key agreement: X25519MLKEM768
but that should be probably adresses in a different patch.
File README.wolfssl:
http://gerrit.openvpn.net/c/openvpn/+/1046/comment/19a0ea53_119798bd :
PS3, Line 39: WolfSSL supports the following Quantum Safe algorithms by spe=
cifying them using the `tls-groups`
> What do you mean? WolfSSL allows the definition of secp384r1 and P-384 re=
spectively interchangeably.
Okay you are right for the *plain* P384 curve but not for the hybrid ones (=
see below). But we NEED to mention is that the names of the new key agreeme=
nts are different.
If you try to use the names that work with OpenSSL (https://community.openv=
pn.net/PQCryptoOpenVPN) you get only an error
Failed to set allowed TLS group list: secp384r1:X25519MLKEM768
while OpenSSL also does not like the wolfSSL names. It is probably good to =
least mention the difference in the naming, e.g. by adding the names of the=
OpenSSL (our TLS library) in brackes in the readme.
Also OpenSSL *only* accepts secp384r1MLKEM1024 and not P384MLKEM1024. So it=
is something more than just removing the _ from the names.
So I still it would be good document this fact at least in the man page for=
tls-groups by adding a paragraph like:
Please note that when OpenVPN is compiled with wolfSSL, the names of the g=
roups might be called different, especially the PQ groups (e.g. X25519_ML_K=
EM_512 instead of X25519MLKEM768 or P384_MLKEM_1024 instead of secp384r1MLK=
EM1024)
File README.wolfssl:
http://gerrit.openvpn.net/c/openvpn/+/1046/comment/41367317_be158cd2 :
PS5, Line 38: option in an OpenVPN config.
Add a paragraph here like:
In contrast to OpenSSL that includes X25519MLKEM768 in the default configur=
ation, wolfSSL will need to explicitly have tls-groups configured to includ=
e (at least one) PQ KEM.
--
To view, visit http://gerrit.openvpn.net/c/openvpn/+/1046?usp=3Demail
To unsubscribe, or for help writing mail filters, visit http://gerrit.openv=
pn.net/settings
Gerrit-Project: openvpn
Gerrit-Branch: master
Gerrit-Change-Id: Ie0529c2074964b3be034f01e0ef53090a6edbd35
Gerrit-Change-Number: 1046
Gerrit-PatchSet: 5
Gerrit-Owner: comododragon
Gerrit-Reviewer: cron2
Gerrit-Reviewer: flichtenheld
Gerrit-Reviewer: plaisthos
Gerrit-CC: openvpn-devel
Gerrit-Attention: flichtenheld
Gerrit-Attention: comododragon
Gerrit-Comment-Date: Wed, 25 Jun 2025 22:23:31 +0000
Gerrit-HasComments: Yes
Gerrit-Has-Labels: Yes
Comment-In-Reply-To: plaisthos
Comment-In-Reply-To: comododragon
Gerrit-MessageType: comment
--qLlQ3M/DBlA=
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Attention is currently required from: comododragon, flichtenheld.
Patch set 5:Code-Review +1=
V=
iew Change
3 comments:
Patchset:
Patch Set #5:
I tested the patch and it works and is also=
a lot cleaner than the first version.
There are some documentation updates I would like=
to see (see the other comments) to ensure users to don't get lost with=
the difference in OpenSSL vs wolfSSL names and defaults.
Also wolfSSL's ML-DSA-87 s=
eems to be a bit wacky for now. It cannot read the private key generated by=
OpenSSL and when connecting to a server with ml-kem and fingerprints it se=
ems not be able to read some of the fields with the current APIs that we us=
e:
[wolfSSL] =
Control Channel: TLSv1.3, cipher TLSv1.3 TLS_AES_256_GCM_SHA384, peer certi=
ficate: , signature: ML-DSA 87, peer signing digest/type: SHA512 ML-DSA 87<=
/p>
vs
[OpenSSL] Control Chann=
el: TLSv1.3, cipher TLSv1.3 TLS_AES_256_GCM_SHA384, peer certificate: 20736=
bits ML-DSA-87, signature: id-ml-dsa-87, peer signing digest/type: mldsa87=
id-ml-dsa-87, key agreement: X25519MLKEM768
but that should be probably adresses in a d=
ifferent patch.
F=
ile README.wolfssl:
Ok=
ay you are right for the *plain* P384 curve but not for the hybrid ones (se=
e below). But we NEED to mention is that the names of the new key agreement=
s are different.
If you try to use the names that work with OpenSSL (https://community.o=
penvpn.net/PQCryptoOpenVPN) you get only an error
Failed to set allowed TLS group list:=
secp384r1:X25519MLKEM768
while OpenSSL also does not like the wolfSSL names. It is prob=
ably good to least mention the difference in the naming, e.g. by adding the=
names of the OpenSSL (our TLS library) in brackes in the readme.
Also OpenSSL *only* a=
ccepts secp384r1MLKEM1024 and not P384MLKEM1024. So it is something more th=
an just removing the _ from the names.
So I still it would be good document this fact at=
least in the man page for tls-groups by adding a paragraph like:
Please n=
ote that when OpenVPN is compiled with wolfSSL, the names of the groups mig=
ht be called different, especially the PQ groups (e.g. X25519_ML_KEM_512 in=
stead of X25519MLKEM768 or P384_MLKEM_1024 instead of secp384r1MLKEM1024)=
pre>
File README.wolf=
ssl:
Patch Set #5,=
Line 38: option in an =
OpenVPN config.
Add a paragraph here like:
In contrast to OpenSSL that includes X25519MLKEM768=
in the default configuration, wolfSSL will need to explicitly have tls-gro=
ups configured to include (at least one) PQ KEM.
=
To view, visit change 1046. To unsubscribe, or for help writing mail filters, v=
isit settings.
Gerrit-Project: openvpn
Gerrit-Branch: master
Gerrit-Change-Id: Ie0529c2074964b3be034f01e0ef=
53090a6edbd35
Gerrit-Change-Number: 1046
Gerrit-PatchSet: 5
Gerrit-Owner: comododragon <rein.vanbaaren@=
...1435...>
Gerrit-Reviewer: cron2 <gert@...1296...>=
Gerrit-Reviewer: flichtenheld <frank@...2716...=
1...>
Gerrit-Reviewer: plaisthos <arne-openvpn@...1186...=
.1227...>
Gerrit-CC: openvpn-devel <openvpn-devel@...1484...=
ts.sourceforge.net>
Gerrit-Attention: flichtenheld <frank@...2375...=
41...>
Gerrit-Attention: comododragon <rein.vanbaa=
ren@...1435...>
Gerrit-Comment-Date: Wed, 25 Jun 2025 22:23:31=
+0000
Gerrit-HasComments: Yes
Gerrit-Has-Labels: Yes
Comment-In-Reply-To: plaisthos <arne-openvp=
n@...1227...>
Comment-In-Reply-To: comododragon <rein.van=
baaren@...1435...>
Gerrit-MessageType: comment
--qLlQ3M/DBlA=--