Return-path: Received: from mx1.redhat.com ([66.187.233.31]:54815 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754253AbXJJPHL (ORCPT ); Wed, 10 Oct 2007 11:07:11 -0400 Subject: Re: mac80211/iwlwifi + hidden ssids From: Dan Williams To: dragoran Cc: "John W. Linville" , linux-wireless@vger.kernel.org, ipw3945-devel In-Reply-To: References: <470C9F28.1070803@gmail.com> <20071010130653.GB5962@tuxdriver.com> <1192026847.9739.4.camel@localhost.localdomain> Content-Type: text/plain Date: Wed, 10 Oct 2007 11:07:46 -0400 Message-Id: <1192028866.9739.32.camel@localhost.localdomain> (sfid-20071010_160716_424570_C9F68AD3) Mime-Version: 1.0 Sender: linux-wireless-owner@vger.kernel.org List-ID: On Wed, 2007-10-10 at 16:57 +0200, dragoran wrote: > On 10/10/07, Dan Williams wrote: > >[..] > > wpa_supplicant's support for hidden SSIDs sort of sucks; mainly because > > the drivers implement this in different ways. wpa_supplicant says you > > must use "ap_scan=2" for hidden SSIDs (and possibly ssid_scan=1 to do > > specific ssid probe requests), which causes wpa_supplicant to blast all > > settings to the card directly and not try to be smart about AP scanning > > and selection. However, this doesn't work often because drivers are > > inconsistent about this. > > > > The way it _should_ work is that if scan_ssid = 1 for a network, > > wpa_supplicant should ask that the driver scan the specific ssid using > > IW_SCAN_THIS_ESSID. The card would then issue specific SSID scan > > requests on each channel during the scan, and hopefully find the AP in > > question. > > > > Unfortunately, most drivers don't support IW_SCAN_THIS_ESSID. So the > > best we can do is set AP_SCAN=2, let wpa_supplicant blow all the > > settings to the card, and just _hope_ that the card/driver are smart > > enough to find the hidden SSID on their own and associate. Which is > > less often than we'd hope. > > so your are suggestion to use ap_scan=2 and add scan_ssid=1 to the > network block? > nm seems not to do this. It should be doing this... > btw shouldn't all mac80211 drivers either work or not work because the > code for this is in mac80211 and not in the drivers? Yeah, pretty much; but each driver/firmware may handle hidden ssids slightly differently. mac80211 should bring some sanity to this area, but I wouldn't discount variation between even mac80211 based drivers. Dan > I will be able to test again this friday will try this and the > ap_scan=1 if there are other suggestion feel free to say them so I can > add them to my "totest" list ;)