From patchwork Thu Feb 26 15:03:40 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Kirsher, Jeffrey T" X-Patchwork-Id: 443953 X-Patchwork-Delegate: davem@davemloft.net Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by ozlabs.org (Postfix) with ESMTP id 3D5871400DD for ; Fri, 27 Feb 2015 02:04:00 +1100 (AEDT) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753908AbbBZPDz (ORCPT ); Thu, 26 Feb 2015 10:03:55 -0500 Received: from mga14.intel.com ([192.55.52.115]:52636 "EHLO mga14.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753684AbbBZPDy (ORCPT ); Thu, 26 Feb 2015 10:03:54 -0500 Received: from fmsmga001.fm.intel.com ([10.253.24.23]) by fmsmga103.fm.intel.com with ESMTP; 26 Feb 2015 06:55:48 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.09,653,1418112000"; d="scan'208";a="672084104" Received: from pkdiveka-mobl3.amr.corp.intel.com (HELO jtkirshe-mobl.amr.corp.intel.com) ([10.252.205.119]) by fmsmga001.fm.intel.com with ESMTP; 26 Feb 2015 07:03:52 -0800 From: Jeff Kirsher To: davem@davemloft.net Cc: Anjali Singhai , netdev@vger.kernel.org, nhorman@redhat.com, sassmann@redhat.com, jogreene@redhat.com, Jeff Kirsher Subject: [net 04/13] i40e: Fix TSO with more than 8 frags per segment issue Date: Thu, 26 Feb 2015 07:03:40 -0800 Message-Id: <1424963029-4612-5-git-send-email-jeffrey.t.kirsher@intel.com> X-Mailer: git-send-email 1.9.3 In-Reply-To: <1424963029-4612-1-git-send-email-jeffrey.t.kirsher@intel.com> References: <1424963029-4612-1-git-send-email-jeffrey.t.kirsher@intel.com> Sender: netdev-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org From: Anjali Singhai The hardware has some limitations the driver needs to adhere to, that we found in extended testing. 1) no more than 8 descriptors per packet on the wire 2) no header can span more than 3 descriptors If one of these events occurs, the hardware will generate an internal error and freeze the Tx queue. This patch linearizes the skb to avoid these situations. Change-ID: I37dab7d3966e14895a9663ec4d0aaa8eb0d9e115 Signed-off-by: Anjali Singhai Jain Tested-by: Jim Young Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/i40e/i40e_txrx.c | 65 +++++++++++++++++++++++++++ drivers/net/ethernet/intel/i40e/i40e_txrx.h | 1 + drivers/net/ethernet/intel/i40evf/i40e_txrx.c | 65 +++++++++++++++++++++++++++ drivers/net/ethernet/intel/i40evf/i40e_txrx.h | 1 + 4 files changed, 132 insertions(+) diff --git a/drivers/net/ethernet/intel/i40e/i40e_txrx.c b/drivers/net/ethernet/intel/i40e/i40e_txrx.c index 2206d2d..0aad611 100644 --- a/drivers/net/ethernet/intel/i40e/i40e_txrx.c +++ b/drivers/net/ethernet/intel/i40e/i40e_txrx.c @@ -2140,6 +2140,67 @@ static int i40e_maybe_stop_tx(struct i40e_ring *tx_ring, int size) } /** + * i40e_chk_linearize - Check if there are more than 8 fragments per packet + * @skb: send buffer + * @tx_flags: collected send information + * @hdr_len: size of the packet header + * + * Note: Our HW can't scatter-gather more than 8 fragments to build + * a packet on the wire and so we need to figure out the cases where we + * need to linearize the skb. + **/ +static bool i40e_chk_linearize(struct sk_buff *skb, u32 tx_flags, + const u8 hdr_len) +{ + struct skb_frag_struct *frag; + bool linearize = false; + unsigned int size = 0; + u16 num_frags; + u16 gso_segs; + + num_frags = skb_shinfo(skb)->nr_frags; + gso_segs = skb_shinfo(skb)->gso_segs; + + if (tx_flags & (I40E_TX_FLAGS_TSO | I40E_TX_FLAGS_FSO)) { + u16 j = 1; + + if (num_frags < (I40E_MAX_BUFFER_TXD)) + goto linearize_chk_done; + /* try the simple math, if we have too many frags per segment */ + if (DIV_ROUND_UP((num_frags + gso_segs), gso_segs) > + I40E_MAX_BUFFER_TXD) { + linearize = true; + goto linearize_chk_done; + } + frag = &skb_shinfo(skb)->frags[0]; + size = hdr_len; + /* we might still have more fragments per segment */ + do { + size += skb_frag_size(frag); + frag++; j++; + if (j == I40E_MAX_BUFFER_TXD) { + if (size < skb_shinfo(skb)->gso_size) { + linearize = true; + break; + } + j = 1; + size -= skb_shinfo(skb)->gso_size; + if (size) + j++; + size += hdr_len; + } + num_frags--; + } while (num_frags); + } else { + if (num_frags >= I40E_MAX_BUFFER_TXD) + linearize = true; + } + +linearize_chk_done: + return linearize; +} + +/** * i40e_tx_map - Build the Tx descriptor * @tx_ring: ring to send buffer on * @skb: send buffer @@ -2396,6 +2457,10 @@ static netdev_tx_t i40e_xmit_frame_ring(struct sk_buff *skb, if (tsyn) tx_flags |= I40E_TX_FLAGS_TSYN; + if (i40e_chk_linearize(skb, tx_flags, hdr_len)) + if (skb_linearize(skb)) + goto out_drop; + skb_tx_timestamp(skb); /* always enable CRC insertion offload */ diff --git a/drivers/net/ethernet/intel/i40e/i40e_txrx.h b/drivers/net/ethernet/intel/i40e/i40e_txrx.h index 18b0023..dff0bae 100644 --- a/drivers/net/ethernet/intel/i40e/i40e_txrx.h +++ b/drivers/net/ethernet/intel/i40e/i40e_txrx.h @@ -112,6 +112,7 @@ enum i40e_dyn_idx_t { #define i40e_rx_desc i40e_32byte_rx_desc +#define I40E_MAX_BUFFER_TXD 8 #define I40E_MIN_TX_LEN 17 #define I40E_MAX_DATA_PER_TXD 8192 diff --git a/drivers/net/ethernet/intel/i40evf/i40e_txrx.c b/drivers/net/ethernet/intel/i40evf/i40e_txrx.c index 2900438..1320a43 100644 --- a/drivers/net/ethernet/intel/i40evf/i40e_txrx.c +++ b/drivers/net/ethernet/intel/i40evf/i40e_txrx.c @@ -1380,6 +1380,67 @@ static void i40e_create_tx_ctx(struct i40e_ring *tx_ring, context_desc->type_cmd_tso_mss = cpu_to_le64(cd_type_cmd_tso_mss); } + /** + * i40e_chk_linearize - Check if there are more than 8 fragments per packet + * @skb: send buffer + * @tx_flags: collected send information + * @hdr_len: size of the packet header + * + * Note: Our HW can't scatter-gather more than 8 fragments to build + * a packet on the wire and so we need to figure out the cases where we + * need to linearize the skb. + **/ +static bool i40e_chk_linearize(struct sk_buff *skb, u32 tx_flags, + const u8 hdr_len) +{ + struct skb_frag_struct *frag; + bool linearize = false; + unsigned int size = 0; + u16 num_frags; + u16 gso_segs; + + num_frags = skb_shinfo(skb)->nr_frags; + gso_segs = skb_shinfo(skb)->gso_segs; + + if (tx_flags & (I40E_TX_FLAGS_TSO | I40E_TX_FLAGS_FSO)) { + u16 j = 1; + + if (num_frags < (I40E_MAX_BUFFER_TXD)) + goto linearize_chk_done; + /* try the simple math, if we have too many frags per segment */ + if (DIV_ROUND_UP((num_frags + gso_segs), gso_segs) > + I40E_MAX_BUFFER_TXD) { + linearize = true; + goto linearize_chk_done; + } + frag = &skb_shinfo(skb)->frags[0]; + size = hdr_len; + /* we might still have more fragments per segment */ + do { + size += skb_frag_size(frag); + frag++; j++; + if (j == I40E_MAX_BUFFER_TXD) { + if (size < skb_shinfo(skb)->gso_size) { + linearize = true; + break; + } + j = 1; + size -= skb_shinfo(skb)->gso_size; + if (size) + j++; + size += hdr_len; + } + num_frags--; + } while (num_frags); + } else { + if (num_frags >= I40E_MAX_BUFFER_TXD) + linearize = true; + } + +linearize_chk_done: + return linearize; +} + /** * i40e_tx_map - Build the Tx descriptor * @tx_ring: ring to send buffer on @@ -1654,6 +1715,10 @@ static netdev_tx_t i40e_xmit_frame_ring(struct sk_buff *skb, else if (tso) tx_flags |= I40E_TX_FLAGS_TSO; + if (i40e_chk_linearize(skb, tx_flags, hdr_len)) + if (skb_linearize(skb)) + goto out_drop; + skb_tx_timestamp(skb); /* always enable CRC insertion offload */ diff --git a/drivers/net/ethernet/intel/i40evf/i40e_txrx.h b/drivers/net/ethernet/intel/i40evf/i40e_txrx.h index 4e15903..c950a03 100644 --- a/drivers/net/ethernet/intel/i40evf/i40e_txrx.h +++ b/drivers/net/ethernet/intel/i40evf/i40e_txrx.h @@ -112,6 +112,7 @@ enum i40e_dyn_idx_t { #define i40e_rx_desc i40e_32byte_rx_desc +#define I40E_MAX_BUFFER_TXD 8 #define I40E_MIN_TX_LEN 17 #define I40E_MAX_DATA_PER_TXD 8192