From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from mga02.intel.com ([134.134.136.20]:12192 "EHLO mga02.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751357AbZHNESG (ORCPT ); Fri, 14 Aug 2009 00:18:06 -0400 Subject: Re: [PATCH] nl80211 connect API support From: Zhu Yi To: Jouni Malinen Cc: "hostap@lists.shmoo.com" , "linux-wireless@vger.kernel.org" , Johannes Berg , Samuel Ortiz In-Reply-To: <20090813104635.GA7330@jm.kir.nu> References: <1250153731-17208-1-git-send-email-yi.zhu@intel.com> <1250153731-17208-2-git-send-email-yi.zhu@intel.com> <20090813091103.GA1826@jm.kir.nu> <1250155777.4972.9.camel@debian> <20090813104635.GA7330@jm.kir.nu> Content-Type: text/plain Date: Fri, 14 Aug 2009 12:17:54 +0800 Message-Id: <1250223474.4972.47.camel@debian> Mime-Version: 1.0 Sender: linux-wireless-owner@vger.kernel.org List-ID: On Thu, 2009-08-13 at 18:46 +0800, Jouni Malinen wrote: > It is not about whether it would be possible to use connect or not; it > is about whether the additional features provided by separate auth/assoc > commands are of use--and they are. wpa_supplicant will provide more > functionality, e.g., FT, when using these commands. Any change to add > support for the new NL80211_CMD_CONNECT must not break this existing > mechanism; it is only to add support for drivers that cannot support the > auth/assoc interface that provides more control to user space. Ah, I see. This makes sense to me. I ignored the 11r code when I wrote the code. Now I realized what it is used for (i.e. sme_event_auth). I'm not sure if it can be done in cfg80211 SME easily, but it doesn't help for connect API anyway. I'll send V2 according to your suggestion. Johannes, I think we need a way in cfg80211 to tell connect vs. auth/assoc. The current capability method (nl80211_send_wiphy) cannot do this. For example, if (dev->ops->connect || dev->ops->auth) NLA_PUT_U32(msg, i, NL80211_CMD_CONNECT); We need to either make NL80211_CMD_CONNECT only depends on ops->connect or make NL80211_CMD_AUTH depends on ops->auth. Johannes, which one do you prefer? Thanks, -yi