Content-Type: multipart/mixed; boundary="===============4020743058426966898==" MIME-Version: 1.0 From: James Prestwood Subject: Re: [PATCH 8/9] sae: don't send commit in confirmed state Date: Wed, 08 Sep 2021 09:40:13 -0700 Message-ID: <4d186fbdb3c2f4407cbe20c9ac609cd75d93bde3.camel@gmail.com> In-Reply-To: <8cdacee8-07ca-1451-55ca-0c24656d8a1c@molgen.mpg.de> List-Id: To: iwd@lists.01.org --===============4020743058426966898== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Hi Paul, > > = > > Do non-hostapd setups even exist? :) But in reality no it shouldn't > > effect anything. I do need to amend this commit slightly, but v2 > > will > > have identical behavior as wpa_supplicant, i.e. ignore the commit > > message completely. > = > Great, it=E2=80=99d be great if you mentioned that. > = > Also, from the other thread *Re: Cannot connect to SAE protected AP > with = > iwd 1.16 and beyond*, if the ell commit really introduced the = > regression, it=E2=80=99d be great if you mentioned that too. So we actually had two separate problems here: First was some APs don't properly do SAE group negotiation. The ELL change, while not a bug, changed the default group to 20 which 802.11 does not mandate support for (but it is more secure). In this case the SAE protocol negotiates the group but as mentioned these APs end up failing the connection. This was worked around in an earlier commit by detecting these problem APs and forcing group 19 from the beginning. The second problem was one addressed in this patche where the AP receives an unexpected commit message and fails the connection rather than dropping the message. In any case, I will add a bit more context to the commit message. Thanks, James > = > = > Kind regards, > = > Paul --===============4020743058426966898==--