Return-path: Received: from LNeuilly-152-21-11-6.w193-253.abo.wanadoo.fr ([193.253.42.6]:34517 "EHLO luceor-laptop" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1752494AbYLLOw2 (ORCPT ); Fri, 12 Dec 2008 09:52:28 -0500 From: Benoit Papillault To: linville@tuxdriver.com Cc: linux-wireless@vger.kernel.org, ath5k-devel@lists.ath5k.org, Benoit PAPILLAULT Subject: [PATCH] ath5k: fix 802.11 header padding on RX, unpadding on TX Date: Fri, 12 Dec 2008 15:29:58 +0100 Message-Id: <1229092198-21136-1-git-send-email-benoit.papillault@free.fr> (sfid-20081212_155232_013741_EDC142B6) Sender: linux-wireless-owner@vger.kernel.org List-ID: From: Benoit PAPILLAULT Padding the 802.11 header to a multiple of 4 bytes needs to be done only for frames with a body. This fixes a bug where 2 bytes were missing in monitor mode for ACK frames. Inspired by a patch from Jouni Malinen on ath9k. Ref: http://bugzilla.kernel.org/show_bug.cgi?id=12101 : Signed-off-by: Benoit Papillault --- drivers/net/wireless/ath5k/base.c | 38 ++++++++++++++++++++---------------- 1 files changed, 21 insertions(+), 17 deletions(-) diff --git a/drivers/net/wireless/ath5k/base.c b/drivers/net/wireless/ath5k/base.c index bfb0c49..9d2c597 100644 --- a/drivers/net/wireless/ath5k/base.c +++ b/drivers/net/wireless/ath5k/base.c @@ -1667,7 +1667,7 @@ ath5k_tasklet_rx(unsigned long data) struct ath5k_desc *ds; int ret; int hdrlen; - int pad; + int padsize; spin_lock(&sc->rxbuflock); if (list_empty(&sc->rxbuf)) { @@ -1752,16 +1752,19 @@ accept: skb_put(skb, rs.rs_datalen); - /* - * the hardware adds a padding to 4 byte boundaries between - * the header and the payload data if the header length is - * not multiples of 4 - remove it - */ + /* The MAC header is padded to have 32-bit boundary if the + * packet payload is non-zero. The general calculation for + * padsize would take into account odd header lengths: + * padsize = (4 - hdrlen % 4) % 4; However, since only + * even-length headers are used, padding can only be 0 or 2 + * bytes and we can optimize this a bit. In addition, we must + * not try to remove padding from short control frames that do + * not have payload. */ hdrlen = ieee80211_get_hdrlen_from_skb(skb); - if (hdrlen & 3) { - pad = hdrlen % 4; - memmove(skb->data + pad, skb->data, hdrlen); - skb_pull(skb, pad); + padsize = hdrlen & 3; + if (padsize && hdrlen >= 24) { + memmove(skb->data + padsize, skb->data, hdrlen); + skb_pull(skb, padsize); } /* @@ -2622,7 +2625,7 @@ ath5k_tx(struct ieee80211_hw *hw, struct sk_buff *skb) struct ath5k_buf *bf; unsigned long flags; int hdrlen; - int pad; + int padsize; ath5k_debug_dump_skb(sc, skb, "TX ", 1); @@ -2634,15 +2637,16 @@ ath5k_tx(struct ieee80211_hw *hw, struct sk_buff *skb) * if this is not the case we add the padding after the header */ hdrlen = ieee80211_get_hdrlen_from_skb(skb); - if (hdrlen & 3) { - pad = hdrlen % 4; - if (skb_headroom(skb) < pad) { + padsize = hdrlen & 3; + if (padsize && hdrlen >= 24) { + + if (skb_headroom(skb) < padsize) { ATH5K_ERR(sc, "tx hdrlen not %%4: %d not enough" - " headroom to pad %d\n", hdrlen, pad); + " headroom to pad %d\n", hdrlen, padsize); return -1; } - skb_push(skb, pad); - memmove(skb->data, skb->data+pad, hdrlen); + skb_push(skb, padsize); + memmove(skb->data, skb->data+padsize, hdrlen); } spin_lock_irqsave(&sc->txbuflock, flags); -- 1.5.6.5