From mboxrd@z Thu Jan  1 00:00:00 1970
Content-Type: multipart/mixed; boundary="===============2379021530565263672=="
MIME-Version: 1.0
From: Denis Kenzior <denkenz@gmail.com>
Subject: Re: [PATCH 3/5] ap: Drop struct ap_config in favor of l_settings
Date: Mon, 26 Apr 2021 15:22:56 -0500
Message-ID: <f5cdeaf7-7b46-b3fe-343c-b99395fd17e0@gmail.com>
In-Reply-To: <CAOq732LK_q2qoyRM9rnQe29hf7tZDq9tQSR33=kBqKfAvAaS9A@mail.gmail.com>
List-Id: <iwd.lists.01.org>
To: iwd@lists.01.org

--===============2379021530565263672==
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable

Hi Andrew,

>>> +     if (!passphrase) {
>>
>> If the PSK is valid, you never get to this check...?  Should probably be=
 done
>> first...
> =

> Here we want the PSK to override the passphrase if both were present
> (which they shouldn't) because we do the same thing in station mode
> because we don't want to recalculate the PSK.

Right.  Maybe its just me but the error message below implies you want to o=
nly =

have one or the other, not both.  The current behavior is to take PreShared=
Key =

and run with it even if both are provided.

Anyway, not a big deal for now.  This will get a bit trickier once we start =

supporting SAE.

> =

>>
>>> +             l_error("AP requires one of [Security].PreSharedKey or "
>>> +                     "[Security].Passphrase to be present");
>>> +             return false;
>>> +     }
>>> +

<snip>

>>> +     strval =3D l_settings_get_string(config, "General", "SSID");
>>> +     if (L_WARN_ON(!strval))
>>
>> This is user-input, why L_WARN_ON?
> =

> So the SSID is passed as a D-Bus string or generated by p2p.c and then
> written to the l_settings, it should be partly validated when we get
> here.
> =


Fair enough.  Do note that we can't get non-utf8 input strings through D-Bu=
s =

either.  I imagine P2P validates the ssid to be utf8 as well, but having a =

sanity check here doesn't hurt.

Regards,
-Denis
--===============2379021530565263672==--