Return-path: Received: from mail-wg0-f48.google.com ([74.125.82.48]:33601 "EHLO mail-wg0-f48.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751151AbaEZFhP convert rfc822-to-8bit (ORCPT ); Mon, 26 May 2014 01:37:15 -0400 Received: by mail-wg0-f48.google.com with SMTP id k14so3128029wgh.31 for ; Sun, 25 May 2014 22:37:14 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <87fvjytht6.fsf@kamboji.qca.qualcomm.com> References: <1400158139-13836-1-git-send-email-michal.kazior@tieto.com> <1400158139-13836-3-git-send-email-michal.kazior@tieto.com> <87fvjytht6.fsf@kamboji.qca.qualcomm.com> Date: Mon, 26 May 2014 07:37:13 +0200 Message-ID: (sfid-20140526_073722_042740_9FA013B7) Subject: Re: [PATCH 2/3] ath10k: revert incomplete scatter-gather pci tx From: Michal Kazior To: Kalle Valo Cc: "ath10k@lists.infradead.org" , linux-wireless , Avery Pennarun Content-Type: text/plain; charset=UTF-8 Sender: linux-wireless-owner@vger.kernel.org List-ID: On 25 May 2014 09:53, Kalle Valo wrote: > Michal Kazior writes: > >> This prevents leaving incomplete scatter-gather >> transfer on CE rings which can lead firmware to >> crash. >> >> Reported-By: Avery Pennarun >> Signed-off-by: Michal Kazior > > The title is a bit misleading as usually the commit log with the word > revert means that the commit is reverting another git commit. Maybe > something like this is better: > > ath10k: drop incomplete scatter-gather pci tx transfers Good point. I was actually thinking 'abort' .. 'properly'. >> + if (WARN_ON(src_ring->write_index == src_ring->sw_index)) >> + return; >> + >> + if (WARN_ON(src_ring->write_index == >> + ath10k_ce_src_ring_write_index_get(ar, ctrl_addr))) >> + return; > > WARN_ON() on data path is dangerous. WARN_ON_ONCE() or ath10k_warn() is > better. Good point! >> +err: >> + for (; i > 0; i--) > > Isn't this just a fancy way to say 'while (i-- > 0)'? Not really. More like do { .. } while (--i > 0), no? First iteration must use unmodified `i`. Michał