Return-path: Received: from mail-wi0-f174.google.com ([209.85.212.174]:37426 "EHLO mail-wi0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752627AbbEKMnk convert rfc822-to-8bit (ORCPT ); Mon, 11 May 2015 08:43:40 -0400 Received: by widdi4 with SMTP id di4so94820778wid.0 for ; Mon, 11 May 2015 05:43:38 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <1431346122.1964.26.camel@sipsolutions.net> References: <1431336675-15944-1-git-send-email-janusz.dziedzic@tieto.com> <1431346122.1964.26.camel@sipsolutions.net> Date: Mon, 11 May 2015 14:43:38 +0200 Message-ID: (sfid-20150511_144344_399341_E743BBE9) Subject: Re: [PATCH] mac80211: WEP, move tailroom size check From: Michal Kazior To: Johannes Berg Cc: Janusz Dziedzic , linux-wireless Content-Type: text/plain; charset=UTF-8 Sender: linux-wireless-owner@vger.kernel.org List-ID: On 11 May 2015 at 14:08, Johannes Berg wrote: > On Mon, 2015-05-11 at 11:31 +0200, Janusz Dziedzic wrote: >> Remove checking tailroom when adding IV, while >> this goes to headroom. Move this check to the function >> that will generate/put ICV for WEP. >> >> In other case I hit such warning and datapath don't work, >> when testing: >> - IBSS + WEP >> - ath9k with hw crypt enabled >> - IPv6 data (ping6) > > Is this new, perhaps due to fast-xmit, because we previously added > tailroom for WEP always? I've certainly never seen this before. > > (just asking so I know where it needs to be applied) I think this was hidden for quite some time now, possibly since _PUT_IV_SPACE and _GENERATE_IV were split into _GENERATE_MMIC and _RESERVE_TAILROOM counterparts. I know that `ping` didn't trigger this problem, instead `ping6` did which leads me to think that maybe sk_buff allocation was getting lucky most of the time. Michał