Return-path: Received: from mx0a-0016f401.pphosted.com ([67.231.148.174]:36868 "EHLO mx0b-0016f401.pphosted.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1750729AbeFAHyB (ORCPT ); Fri, 1 Jun 2018 03:54:01 -0400 From: Xinming Hu To: Linux Wireless CC: Kalle Valo , Brian Norris , Dmitry Torokhov , , Zhiyuan Yang , Tim Song , Cathy Luo , James Cao , Ganapathi Bhat , Ellie Reeves , Xinming Hu Subject: [PATCH] mwifiex: uap: do not chok ethernet header in bridge path Date: Fri, 1 Jun 2018 15:53:40 +0800 Message-ID: <1527839620-1760-1-git-send-email-huxm@marvell.com> (sfid-20180601_095405_492916_31BB64D4) MIME-Version: 1.0 Content-Type: text/plain Sender: linux-wireless-owner@vger.kernel.org List-ID: Do not chock ethernet header for uap bridge data path, as it is still needed to send skb to dest station. Signed-off-by: Xinming Hu --- drivers/net/wireless/marvell/mwifiex/uap_txrx.c | 52 ++++++++++++------------- 1 file changed, 26 insertions(+), 26 deletions(-) diff --git a/drivers/net/wireless/marvell/mwifiex/uap_txrx.c b/drivers/net/wireless/marvell/mwifiex/uap_txrx.c index 1e6a62c..5ce85d5 100644 --- a/drivers/net/wireless/marvell/mwifiex/uap_txrx.c +++ b/drivers/net/wireless/marvell/mwifiex/uap_txrx.c @@ -289,32 +289,6 @@ int mwifiex_uap_recv_packet(struct mwifiex_private *priv, src_node->stats.rx_packets++; } - skb->dev = priv->netdev; - skb->protocol = eth_type_trans(skb, priv->netdev); - skb->ip_summed = CHECKSUM_NONE; - - /* This is required only in case of 11n and USB/PCIE as we alloc - * a buffer of 4K only if its 11N (to be able to receive 4K - * AMSDU packets). In case of SD we allocate buffers based - * on the size of packet and hence this is not needed. - * - * Modifying the truesize here as our allocation for each - * skb is 4K but we only receive 2K packets and this cause - * the kernel to start dropping packets in case where - * application has allocated buffer based on 2K size i.e. - * if there a 64K packet received (in IP fragments and - * application allocates 64K to receive this packet but - * this packet would almost double up because we allocate - * each 1.5K fragment in 4K and pass it up. As soon as the - * 64K limit hits kernel will start to drop rest of the - * fragments. Currently we fail the Filesndl-ht.scr script - * for UDP, hence this fix - */ - if ((adapter->iface_type == MWIFIEX_USB || - adapter->iface_type == MWIFIEX_PCIE) && - (skb->truesize > MWIFIEX_RX_DATA_BUF_SIZE)) - skb->truesize += (skb->len - MWIFIEX_RX_DATA_BUF_SIZE); - if (is_multicast_ether_addr(p_ethhdr->h_dest) || mwifiex_get_sta_entry(priv, p_ethhdr->h_dest)) { if (skb_headroom(skb) < MWIFIEX_MIN_DATA_HEADER_LEN) @@ -350,6 +324,32 @@ int mwifiex_uap_recv_packet(struct mwifiex_private *priv, return 0; } + skb->dev = priv->netdev; + skb->protocol = eth_type_trans(skb, priv->netdev); + skb->ip_summed = CHECKSUM_NONE; + + /* This is required only in case of 11n and USB/PCIE as we alloc + * a buffer of 4K only if its 11N (to be able to receive 4K + * AMSDU packets). In case of SD we allocate buffers based + * on the size of packet and hence this is not needed. + * + * Modifying the truesize here as our allocation for each + * skb is 4K but we only receive 2K packets and this cause + * the kernel to start dropping packets in case where + * application has allocated buffer based on 2K size i.e. + * if there a 64K packet received (in IP fragments and + * application allocates 64K to receive this packet but + * this packet would almost double up because we allocate + * each 1.5K fragment in 4K and pass it up. As soon as the + * 64K limit hits kernel will start to drop rest of the + * fragments. Currently we fail the Filesndl-ht.scr script + * for UDP, hence this fix + */ + if ((adapter->iface_type == MWIFIEX_USB || + adapter->iface_type == MWIFIEX_PCIE) && + skb->truesize > MWIFIEX_RX_DATA_BUF_SIZE) + skb->truesize += (skb->len - MWIFIEX_RX_DATA_BUF_SIZE); + /* Forward multicast/broadcast packet to upper layer*/ if (in_interrupt()) netif_rx(skb); -- 1.9.1