From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from mx1.redhat.com ([209.132.183.28]:58310 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753598Ab1BOLkJ (ORCPT ); Tue, 15 Feb 2011 06:40:09 -0500 Date: Tue, 15 Feb 2011 12:39:52 +0100 From: Stanislaw Gruszka To: Johannes Berg Cc: Wey-Yi Guy , Intel Linux Wireless , linux-wireless@vger.kernel.org Subject: Re: [PATCH] iwlwifi: zero fill txq->txb on queue reset Message-ID: <20110215113951.GD2162@redhat.com> References: <1297693999-15342-1-git-send-email-sgruszka@redhat.com> <1297695219.3953.5.camel@jlt3.sipsolutions.net> <1297695325.3953.6.camel@jlt3.sipsolutions.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <1297695325.3953.6.camel@jlt3.sipsolutions.net> Sender: linux-wireless-owner@vger.kernel.org List-ID: On Mon, Feb 14, 2011 at 03:55:25PM +0100, Johannes Berg wrote: > On Mon, 2011-02-14 at 15:53 +0100, Johannes Berg wrote: > > On Mon, 2011-02-14 at 15:33 +0100, Stanislaw Gruszka wrote: > > > We allocate txq->txb area with kzalloc, but when reseting queue > > > during __iwl_down / __iwl_up we we leave old values, let's zero > > > initialize in such case as well. > > > > > > Signed-off-by: Stanislaw Gruszka > > > --- > > > drivers/net/wireless/iwlwifi/iwl-tx.c | 2 ++ > > > 1 files changed, 2 insertions(+), 0 deletions(-) > > > > > > diff --git a/drivers/net/wireless/iwlwifi/iwl-tx.c b/drivers/net/wireless/iwlwifi/iwl-tx.c > > > index a8935cd..0c8555c 100644 > > > --- a/drivers/net/wireless/iwlwifi/iwl-tx.c > > > +++ b/drivers/net/wireless/iwlwifi/iwl-tx.c > > > @@ -418,6 +418,8 @@ void iwl_tx_queue_reset(struct iwl_priv *priv, struct iwl_tx_queue *txq, > > > > > > if (txq_id == priv->cmd_queue) > > > actual_slots++; > > > + else > > > + memset(txq->txb, 0, sizeof(txq->txb[0]) * TFD_QUEUE_SIZE_MAX); > > > > I don't mind -- but I think the changelog should say that this isn't > > necessary since the device will never use any of those entries? > > Err, I confused txb and tfds. But it still shouldn't be necessary -- > aren't the entries reset to NULL when taken out of use? There are memset zero before the use in {iwlagn,iwl3945}_tx_skb(), so patch is unneeded as well kzalloc in current code (I add replace to kmalloc to my todo list). Stanislaw