i40e: adds FCoE code to the i40e driver
authorVasu Dev <vasu.dev@intel.com>
Fri, 1 Aug 2014 20:27:02 +0000 (13:27 -0700)
committerDavid S. Miller <davem@davemloft.net>
Sun, 3 Aug 2014 02:41:13 +0000 (19:41 -0700)
This patch adds FCoE ( Fibre Channel Over Ethernet ) code for
Intel XL710 adapters. This patch is limited to only new FCoE
offloads code in newly added files by this patch and then
following patches in the series modifies rest of the existing
driver to enable FCoE with i40e driver.

Signed-off-by: Vasu Dev <vasu.dev@intel.com>
Tested-by: Jack Morgan<jack.morgan@intel.com>
Signed-off-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/ethernet/intel/i40e/i40e_fcoe.c [new file with mode: 0644]
drivers/net/ethernet/intel/i40e/i40e_fcoe.h [new file with mode: 0644]

diff --git a/drivers/net/ethernet/intel/i40e/i40e_fcoe.c b/drivers/net/ethernet/intel/i40e/i40e_fcoe.c
new file mode 100644 (file)
index 0000000..8574eee
--- /dev/null
@@ -0,0 +1,1568 @@
+/*******************************************************************************
+ *
+ * Intel Ethernet Controller XL710 Family Linux Driver
+ * Copyright(c) 2013 - 2014 Intel Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ * The full GNU General Public License is included in this distribution in
+ * the file called "COPYING".
+ *
+ * Contact Information:
+ * e1000-devel Mailing List <e1000-devel@lists.sourceforge.net>
+ * Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+ *
+ ******************************************************************************/
+
+
+#include <linux/if_ether.h>
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_device.h>
+#include <scsi/fc/fc_fs.h>
+#include <scsi/fc/fc_fip.h>
+#include <scsi/fc/fc_fcoe.h>
+#include <scsi/libfc.h>
+#include <scsi/libfcoe.h>
+
+#include "i40e.h"
+#include "i40e_fcoe.h"
+
+/**
+ * i40e_rx_is_fip - returns true if the rx packet type is FIP
+ * @ptype: the packet type field from rx descriptor write-back
+ **/
+static inline bool i40e_rx_is_fip(u16 ptype)
+{
+       return ptype == I40E_RX_PTYPE_L2_FIP_PAY2;
+}
+
+/**
+ * i40e_rx_is_fcoe - returns true if the rx packet type is FCoE
+ * @ptype: the packet type field from rx descriptor write-back
+ **/
+static inline bool i40e_rx_is_fcoe(u16 ptype)
+{
+       return (ptype >= I40E_RX_PTYPE_L2_FCOE_PAY3) &&
+              (ptype <= I40E_RX_PTYPE_L2_FCOE_VFT_FCOTHER);
+}
+
+/**
+ * i40e_fcoe_sof_is_class2 - returns true if this is a FC Class 2 SOF
+ * @sof: the FCoE start of frame delimiter
+ **/
+static inline bool i40e_fcoe_sof_is_class2(u8 sof)
+{
+       return (sof == FC_SOF_I2) || (sof == FC_SOF_N2);
+}
+
+/**
+ * i40e_fcoe_sof_is_class3 - returns true if this is a FC Class 3 SOF
+ * @sof: the FCoE start of frame delimiter
+ **/
+static inline bool i40e_fcoe_sof_is_class3(u8 sof)
+{
+       return (sof == FC_SOF_I3) || (sof == FC_SOF_N3);
+}
+
+/**
+ * i40e_fcoe_sof_is_supported - returns true if the FC SOF is supported by HW
+ * @sof: the input SOF value from the frame
+ **/
+static inline bool i40e_fcoe_sof_is_supported(u8 sof)
+{
+       return i40e_fcoe_sof_is_class2(sof) ||
+              i40e_fcoe_sof_is_class3(sof);
+}
+
+/**
+ * i40e_fcoe_fc_sof - pull the SOF from FCoE header in the frame
+ * @skb: the frame whose EOF is to be pulled from
+ **/
+static inline int i40e_fcoe_fc_sof(struct sk_buff *skb, u8 *sof)
+{
+       *sof = ((struct fcoe_hdr *)skb_network_header(skb))->fcoe_sof;
+
+       if (!i40e_fcoe_sof_is_supported(*sof))
+               return -EINVAL;
+       return 0;
+}
+
+/**
+ * i40e_fcoe_eof_is_supported - returns true if the EOF is supported by HW
+ * @eof:     the input EOF value from the frame
+ **/
+static inline bool i40e_fcoe_eof_is_supported(u8 eof)
+{
+       return (eof == FC_EOF_N) || (eof == FC_EOF_T) ||
+              (eof == FC_EOF_NI) || (eof == FC_EOF_A);
+}
+
+/**
+ * i40e_fcoe_fc_eof - pull EOF from FCoE trailer in the frame
+ * @skb: the frame whose EOF is to be pulled from
+ **/
+static inline int i40e_fcoe_fc_eof(struct sk_buff *skb, u8 *eof)
+{
+       /* the first byte of the last dword is EOF */
+       skb_copy_bits(skb, skb->len - 4, eof, 1);
+
+       if (!i40e_fcoe_eof_is_supported(*eof))
+               return -EINVAL;
+       return 0;
+}
+
+/**
+ * i40e_fcoe_ctxt_eof - convert input FC EOF for descriptor programming
+ * @eof: the input eof value from the frame
+ *
+ * The FC EOF is converted to the value understood by HW for descriptor
+ * programming. Never call this w/o calling i40e_fcoe_eof_is_supported()
+ * first.
+ **/
+static inline u32 i40e_fcoe_ctxt_eof(u8 eof)
+{
+       switch (eof) {
+       case FC_EOF_N:
+               return I40E_TX_DESC_CMD_L4T_EOFT_EOF_N;
+       case FC_EOF_T:
+               return I40E_TX_DESC_CMD_L4T_EOFT_EOF_T;
+       case FC_EOF_NI:
+               return I40E_TX_DESC_CMD_L4T_EOFT_EOF_NI;
+       case FC_EOF_A:
+               return I40E_TX_DESC_CMD_L4T_EOFT_EOF_A;
+       default:
+               /* FIXME: still returns 0 */
+               pr_err("Unrecognized EOF %x\n", eof);
+               return 0;
+       }
+}
+
+/**
+ * i40e_fcoe_xid_is_valid - returns true if the exchange id is valid
+ * @xid: the exchange id
+ **/
+static inline bool i40e_fcoe_xid_is_valid(u16 xid)
+{
+       return (xid != FC_XID_UNKNOWN) && (xid < I40E_FCOE_DDP_MAX);
+}
+
+/**
+ * i40e_fcoe_ddp_unmap - unmap the mapped sglist associated
+ * @pf: pointer to pf
+ * @ddp: sw DDP context
+ *
+ * Unmap the scatter-gather list associated with the given SW DDP context
+ *
+ * Returns: data length already ddp-ed in bytes
+ *
+ **/
+static inline void i40e_fcoe_ddp_unmap(struct i40e_pf *pf,
+                                      struct i40e_fcoe_ddp *ddp)
+{
+       if (test_and_set_bit(__I40E_FCOE_DDP_UNMAPPED, &ddp->flags))
+               return;
+
+       if (ddp->sgl) {
+               dma_unmap_sg(&pf->pdev->dev, ddp->sgl, ddp->sgc,
+                            DMA_FROM_DEVICE);
+               ddp->sgl = NULL;
+               ddp->sgc = 0;
+       }
+
+       if (ddp->pool) {
+               dma_pool_free(ddp->pool, ddp->udl, ddp->udp);
+               ddp->pool = NULL;
+       }
+}
+
+/**
+ * i40e_fcoe_ddp_clear - clear the given SW DDP context
+ * @ddp - SW DDP context
+ **/
+static inline void i40e_fcoe_ddp_clear(struct i40e_fcoe_ddp *ddp)
+{
+       memset(ddp, 0, sizeof(struct i40e_fcoe_ddp));
+       ddp->xid = FC_XID_UNKNOWN;
+       ddp->flags = __I40E_FCOE_DDP_NONE;
+}
+
+/**
+ * i40e_fcoe_progid_is_fcoe - check if the prog_id is for FCoE
+ * @id: the prog id for the programming status Rx descriptor write-back
+ **/
+static inline bool i40e_fcoe_progid_is_fcoe(u8 id)
+{
+       return (id == I40E_RX_PROG_STATUS_DESC_FCOE_CTXT_PROG_STATUS) ||
+              (id == I40E_RX_PROG_STATUS_DESC_FCOE_CTXT_INVL_STATUS);
+}
+
+/**
+ * i40e_fcoe_fc_get_xid - get xid from the frame header
+ * @fh: the fc frame header
+ *
+ * In case the incoming frame's exchange is originated from
+ * the initiator, then received frame's exchange id is ANDed
+ * with fc_cpu_mask bits to get the same cpu on which exchange
+ * was originated, otherwise just use the current cpu.
+ *
+ * Returns ox_id if exchange originator, rx_id if responder
+ **/
+static inline u16 i40e_fcoe_fc_get_xid(struct fc_frame_header *fh)
+{
+       u32 f_ctl = ntoh24(fh->fh_f_ctl);
+
+       return (f_ctl & FC_FC_EX_CTX) ?
+               be16_to_cpu(fh->fh_ox_id) :
+               be16_to_cpu(fh->fh_rx_id);
+}
+
+/**
+ * i40e_fcoe_fc_frame_header - get fc frame header from skb
+ * @skb: packet
+ *
+ * This checks if there is a VLAN header and returns the data
+ * pointer to the start of the fc_frame_header.
+ *
+ * Returns pointer to the fc_frame_header
+ **/
+static inline struct fc_frame_header *i40e_fcoe_fc_frame_header(
+       struct sk_buff *skb)
+{
+       void *fh = skb->data + sizeof(struct fcoe_hdr);
+
+       if (eth_hdr(skb)->h_proto == htons(ETH_P_8021Q))
+               fh += sizeof(struct vlan_hdr);
+
+       return (struct fc_frame_header *)fh;
+}
+
+/**
+ * i40e_fcoe_ddp_put - release the DDP context for a given exchange id
+ * @netdev: the corresponding net_device
+ * @xid: the exchange id that corresponding DDP context will be released
+ *
+ * This is the implementation of net_device_ops.ndo_fcoe_ddp_done
+ * and it is expected to be called by ULD, i.e., FCP layer of libfc
+ * to release the corresponding ddp context when the I/O is done.
+ *
+ * Returns : data length already ddp-ed in bytes
+ **/
+static int i40e_fcoe_ddp_put(struct net_device *netdev, u16 xid)
+{
+       struct i40e_netdev_priv *np = netdev_priv(netdev);
+       struct i40e_pf *pf = np->vsi->back;
+       struct i40e_fcoe *fcoe = &pf->fcoe;
+       int len = 0;
+       struct i40e_fcoe_ddp *ddp = &fcoe->ddp[xid];
+
+       if (!fcoe || !ddp)
+               goto out;
+
+       if (test_bit(__I40E_FCOE_DDP_DONE, &ddp->flags))
+               len = ddp->len;
+       i40e_fcoe_ddp_unmap(pf, ddp);
+out:
+       return len;
+}
+
+/**
+ * i40e_fcoe_sw_init - sets up the HW for FCoE
+ * @pf: pointer to pf
+ *
+ * Returns 0 if FCoE is supported otherwise the error code
+ **/
+int i40e_init_pf_fcoe(struct i40e_pf *pf)
+{
+       struct i40e_hw *hw = &pf->hw;
+       u32 val;
+
+       pf->flags &= ~I40E_FLAG_FCOE_ENABLED;
+       pf->num_fcoe_qps = 0;
+       pf->fcoe_hmc_cntx_num = 0;
+       pf->fcoe_hmc_filt_num = 0;
+
+       if (!pf->hw.func_caps.fcoe) {
+               dev_info(&pf->pdev->dev, "FCoE capability is disabled\n");
+               return 0;
+       }
+
+       if (!pf->hw.func_caps.dcb) {
+               dev_warn(&pf->pdev->dev,
+                        "Hardware is not DCB capable not enabling FCoE.\n");
+               return 0;
+       }
+
+       /* enable FCoE hash filter */
+       val = rd32(hw, I40E_PFQF_HENA(1));
+       val |= 1 << (I40E_FILTER_PCTYPE_FCOE_OX - 32);
+       val |= 1 << (I40E_FILTER_PCTYPE_FCOE_RX - 32);
+       val &= I40E_PFQF_HENA_PTYPE_ENA_MASK;
+       wr32(hw, I40E_PFQF_HENA(1), val);
+
+       /* enable flag */
+       pf->flags |= I40E_FLAG_FCOE_ENABLED;
+       pf->num_fcoe_qps = I40E_DEFAULT_FCOE;
+
+       /* Reserve 4K DDP contexts and 20K filter size for FCoE */
+       pf->fcoe_hmc_cntx_num = (1 << I40E_DMA_CNTX_SIZE_4K) *
+                                I40E_DMA_CNTX_BASE_SIZE;
+       pf->fcoe_hmc_filt_num = pf->fcoe_hmc_cntx_num +
+                               (1 << I40E_HASH_FILTER_SIZE_16K) *
+                               I40E_HASH_FILTER_BASE_SIZE;
+
+       /* FCoE object: max 16K filter buckets and 4K DMA contexts */
+       pf->filter_settings.fcoe_filt_num = I40E_HASH_FILTER_SIZE_16K;
+       pf->filter_settings.fcoe_cntx_num = I40E_DMA_CNTX_SIZE_4K;
+
+       /* Setup max frame with FCoE_MTU plus L2 overheads */
+       val = rd32(hw, I40E_GLFCOE_RCTL);
+       val &= ~I40E_GLFCOE_RCTL_MAX_SIZE_MASK;
+       val |= ((FCOE_MTU + ETH_HLEN + VLAN_HLEN + ETH_FCS_LEN)
+                << I40E_GLFCOE_RCTL_MAX_SIZE_SHIFT);
+       wr32(hw, I40E_GLFCOE_RCTL, val);
+
+       dev_info(&pf->pdev->dev, "FCoE is supported.\n");
+       return 0;
+}
+
+/**
+ * i40e_get_fcoe_tc_map - Return TC map for FCoE APP
+ * @pf: pointer to pf
+ *
+ **/
+u8 i40e_get_fcoe_tc_map(struct i40e_pf *pf)
+{
+       struct i40e_ieee_app_priority_table app;
+       struct i40e_hw *hw = &pf->hw;
+       u8 enabled_tc = 0;
+       u8 tc, i;
+       /* Get the FCoE APP TLV */
+       struct i40e_dcbx_config *dcbcfg = &hw->local_dcbx_config;
+
+       for (i = 0; i < dcbcfg->numapps; i++) {
+               app = dcbcfg->app[i];
+               if (app.selector == IEEE_8021QAZ_APP_SEL_ETHERTYPE &&
+                   app.protocolid == ETH_P_FCOE) {
+                       tc = dcbcfg->etscfg.prioritytable[app.priority];
+                       enabled_tc |= (1 << tc);
+                       break;
+               }
+       }
+
+       /* TC0 if there is no TC defined for FCoE APP TLV */
+       enabled_tc = enabled_tc ? enabled_tc : 0x1;
+
+       return enabled_tc;
+}
+
+/**
+ * i40e_fcoe_vsi_init - prepares the VSI context for creating a FCoE VSI
+ * @vsi: pointer to the associated VSI struct
+ * @ctxt: pointer to the associated VSI context to be passed to HW
+ *
+ * Returns 0 on success or < 0 on error
+ **/
+int i40e_fcoe_vsi_init(struct i40e_vsi *vsi, struct i40e_vsi_context *ctxt)
+{
+       struct i40e_aqc_vsi_properties_data *info = &ctxt->info;
+       struct i40e_pf *pf = vsi->back;
+       struct i40e_hw *hw = &pf->hw;
+       u8 enabled_tc = 0;
+
+       if (!(pf->flags & I40E_FLAG_FCOE_ENABLED)) {
+               dev_err(&pf->pdev->dev,
+                       "FCoE is not enabled for this device\n");
+               return -EPERM;
+       }
+
+       /* initialize the hardware for FCoE */
+       ctxt->pf_num = hw->pf_id;
+       ctxt->vf_num = 0;
+       ctxt->uplink_seid = vsi->uplink_seid;
+       ctxt->connection_type = 0x1;
+       ctxt->flags = I40E_AQ_VSI_TYPE_PF;
+
+       /* FCoE VSI would need the following sections */
+       info->valid_sections |= cpu_to_le16(I40E_AQ_VSI_PROP_SWITCH_VALID |
+                                           I40E_AQ_VSI_PROP_QUEUE_OPT_VALID);
+
+       /* FCoE VSI does not need these sections */
+       info->valid_sections &= cpu_to_le16(~(I40E_AQ_VSI_PROP_SECURITY_VALID |
+                                           I40E_AQ_VSI_PROP_VLAN_VALID |
+                                           I40E_AQ_VSI_PROP_CAS_PV_VALID |
+                                           I40E_AQ_VSI_PROP_INGRESS_UP_VALID |
+                                           I40E_AQ_VSI_PROP_EGRESS_UP_VALID));
+
+       enabled_tc = i40e_get_fcoe_tc_map(pf);
+       i40e_vsi_setup_queue_map(vsi, ctxt, enabled_tc, true);
+
+       /* set up queue option section: only enable FCoE */
+       info->queueing_opt_flags = I40E_AQ_VSI_QUE_OPT_FCOE_ENA;
+
+       return 0;
+}
+
+/**
+ * i40e_fcoe_enable - this is the implementation of ndo_fcoe_enable,
+ * indicating the upper FCoE protocol stack is ready to use FCoE
+ * offload features.
+ *
+ * @netdev: pointer to the netdev that FCoE is created on
+ *
+ * Returns 0 on success
+ *
+ * in RTNL
+ *
+ **/
+int i40e_fcoe_enable(struct net_device *netdev)
+{
+       struct i40e_netdev_priv *np = netdev_priv(netdev);
+       struct i40e_vsi *vsi = np->vsi;
+       struct i40e_pf *pf = vsi->back;
+       struct i40e_fcoe *fcoe = &pf->fcoe;
+
+       if (!(pf->flags & I40E_FLAG_FCOE_ENABLED)) {
+               netdev_err(netdev, "HW does not support FCoE.\n");
+               return -ENODEV;
+       }
+
+       if (vsi->type != I40E_VSI_FCOE) {
+               netdev_err(netdev, "interface does not support FCoE.\n");
+               return -EBUSY;
+       }
+
+       atomic_inc(&fcoe->refcnt);
+
+       return 0;
+}
+
+/**
+ * i40e_fcoe_disable- disables FCoE for upper FCoE protocol stack.
+ * @dev: pointer to the netdev that FCoE is created on
+ *
+ * Returns 0 on success
+ *
+ **/
+int i40e_fcoe_disable(struct net_device *netdev)
+{
+       struct i40e_netdev_priv *np = netdev_priv(netdev);
+       struct i40e_vsi *vsi = np->vsi;
+       struct i40e_pf *pf = vsi->back;
+       struct i40e_fcoe *fcoe = &pf->fcoe;
+
+       if (!(pf->flags & I40E_FLAG_FCOE_ENABLED)) {
+               netdev_err(netdev, "device does not support FCoE\n");
+               return -ENODEV;
+       }
+       if (vsi->type != I40E_VSI_FCOE)
+               return -EBUSY;
+
+       if (!atomic_dec_and_test(&fcoe->refcnt))
+               return -EINVAL;
+
+       netdev_info(netdev, "FCoE disabled\n");
+
+       return 0;
+}
+
+/**
+ * i40e_fcoe_dma_pool_free - free the per cpu pool for FCoE DDP
+ * @fcoe: the FCoE sw object
+ * @dev: the device that the pool is associated with
+ * @cpu: the cpu for this pool
+ *
+ **/
+static void i40e_fcoe_dma_pool_free(struct i40e_fcoe *fcoe,
+                                   struct device *dev,
+                                   unsigned int cpu)
+{
+       struct i40e_fcoe_ddp_pool *ddp_pool;
+
+       ddp_pool = per_cpu_ptr(fcoe->ddp_pool, cpu);
+       if (!ddp_pool->pool) {
+               dev_warn(dev, "DDP pool already freed for cpu %d\n", cpu);
+               return;
+       }
+       dma_pool_destroy(ddp_pool->pool);
+       ddp_pool->pool = NULL;
+}
+
+/**
+ * i40e_fcoe_dma_pool_create - per cpu pool for FCoE DDP
+ * @fcoe: the FCoE sw object
+ * @dev: the device that the pool is associated with
+ * @cpu: the cpu for this pool
+ *
+ * Returns 0 on successful or non zero on failure
+ *
+ **/
+static int i40e_fcoe_dma_pool_create(struct i40e_fcoe *fcoe,
+                                    struct device *dev,
+                                    unsigned int cpu)
+{
+       struct i40e_fcoe_ddp_pool *ddp_pool;
+       struct dma_pool *pool;
+       char pool_name[32];
+
+       ddp_pool = per_cpu_ptr(fcoe->ddp_pool, cpu);
+       if (ddp_pool && ddp_pool->pool) {
+               dev_warn(dev, "DDP pool already allocated for cpu %d\n", cpu);
+               return 0;
+       }
+       snprintf(pool_name, sizeof(pool_name), "i40e_fcoe_ddp_%d", cpu);
+       pool = dma_pool_create(pool_name, dev, I40E_FCOE_DDP_PTR_MAX,
+                              I40E_FCOE_DDP_PTR_ALIGN, PAGE_SIZE);
+       if (!pool) {
+               dev_err(dev, "dma_pool_create %s failed\n", pool_name);
+               return -ENOMEM;
+       }
+       ddp_pool->pool = pool;
+       return 0;
+}
+
+/**
+ * i40e_fcoe_free_ddp_resources - release FCoE DDP resources
+ * @vsi: the vsi FCoE is associated with
+ *
+ **/
+void i40e_fcoe_free_ddp_resources(struct i40e_vsi *vsi)
+{
+       struct i40e_pf *pf = vsi->back;
+       struct i40e_fcoe *fcoe = &pf->fcoe;
+       int cpu, i;
+
+       /* do nothing if not FCoE VSI */
+       if (vsi->type != I40E_VSI_FCOE)
+               return;
+
+       /* do nothing if no DDP pools were allocated */
+       if (!fcoe->ddp_pool)
+               return;
+
+       for (i = 0; i < I40E_FCOE_DDP_MAX; i++)
+               i40e_fcoe_ddp_put(vsi->netdev, i);
+
+       for_each_possible_cpu(cpu)
+               i40e_fcoe_dma_pool_free(fcoe, &pf->pdev->dev, cpu);
+
+       free_percpu(fcoe->ddp_pool);
+       fcoe->ddp_pool = NULL;
+
+       netdev_info(vsi->netdev, "VSI %d,%d FCoE DDP resources released\n",
+                   vsi->id, vsi->seid);
+}
+
+/**
+ * i40e_fcoe_setup_ddp_resources - allocate per cpu DDP resources
+ * @vsi: the VSI FCoE is associated with
+ *
+ * Returns 0 on successful or non zero on failure
+ *
+ **/
+int i40e_fcoe_setup_ddp_resources(struct i40e_vsi *vsi)
+{
+       struct i40e_pf *pf = vsi->back;
+       struct device *dev = &pf->pdev->dev;
+       struct i40e_fcoe *fcoe = &pf->fcoe;
+       unsigned int cpu;
+       int i;
+
+       if (vsi->type != I40E_VSI_FCOE)
+               return -ENODEV;
+
+       /* do nothing if no DDP pools were allocated */
+       if (fcoe->ddp_pool)
+               return -EEXIST;
+
+       /* allocate per CPU memory to track DDP pools */
+       fcoe->ddp_pool = alloc_percpu(struct i40e_fcoe_ddp_pool);
+       if (!fcoe->ddp_pool) {
+               dev_err(&pf->pdev->dev, "failed to allocate percpu DDP\n");
+               return -ENOMEM;
+       }
+
+       /* allocate pci pool for each cpu */
+       for_each_possible_cpu(cpu) {
+               if (!i40e_fcoe_dma_pool_create(fcoe, dev, cpu))
+                       continue;
+
+               dev_err(dev, "failed to alloc DDP pool on cpu:%d\n", cpu);
+               i40e_fcoe_free_ddp_resources(vsi);
+               return -ENOMEM;
+       }
+
+       /* initialize the sw context */
+       for (i = 0; i < I40E_FCOE_DDP_MAX; i++)
+               i40e_fcoe_ddp_clear(&fcoe->ddp[i]);
+
+       netdev_info(vsi->netdev, "VSI %d,%d FCoE DDP resources allocated\n",
+                   vsi->id, vsi->seid);
+
+       return 0;
+}
+
+/**
+ * i40e_fcoe_handle_status - check the Programming Status for FCoE
+ * @rx_ring: the Rx ring for this descriptor
+ * @rx_desc: the Rx descriptor for Programming Status, not a packet descriptor.
+ *
+ * Check if this is the Rx Programming Status descriptor write-back for FCoE.
+ * This is used to verify if the context/filter programming or invalidation
+ * requested by SW to the HW is successful or not and take actions accordingly.
+ **/
+void i40e_fcoe_handle_status(struct i40e_ring *rx_ring,
+                            union i40e_rx_desc *rx_desc, u8 prog_id)
+{
+       struct i40e_pf *pf = rx_ring->vsi->back;
+       struct i40e_fcoe *fcoe = &pf->fcoe;
+       struct i40e_fcoe_ddp *ddp;
+       u32 error;
+       u16 xid;
+       u64 qw;
+
+       /* we only care for FCoE here */
+       if (!i40e_fcoe_progid_is_fcoe(prog_id))
+               return;
+
+       xid = le32_to_cpu(rx_desc->wb.qword0.hi_dword.fcoe_param) &
+             (I40E_FCOE_DDP_MAX - 1);
+
+       if (!i40e_fcoe_xid_is_valid(xid))
+               return;
+
+       ddp = &fcoe->ddp[xid];
+       WARN_ON(xid != ddp->xid);
+
+       qw = le64_to_cpu(rx_desc->wb.qword1.status_error_len);
+       error = (qw & I40E_RX_PROG_STATUS_DESC_QW1_ERROR_MASK) >>
+               I40E_RX_PROG_STATUS_DESC_QW1_ERROR_SHIFT;
+
+       /* DDP context programming status: failure or success */
+       if (prog_id == I40E_RX_PROG_STATUS_DESC_FCOE_CTXT_PROG_STATUS) {
+               if (I40E_RX_PROG_FCOE_ERROR_TBL_FULL(error)) {
+                       dev_err(&pf->pdev->dev, "xid %x ddp->xid %x TABLE FULL\n",
+                               xid, ddp->xid);
+                       ddp->prerr |= I40E_RX_PROG_FCOE_ERROR_TBL_FULL_BIT;
+               }
+               if (I40E_RX_PROG_FCOE_ERROR_CONFLICT(error)) {
+                       dev_err(&pf->pdev->dev, "xid %x ddp->xid %x CONFLICT\n",
+                               xid, ddp->xid);
+                       ddp->prerr |= I40E_RX_PROG_FCOE_ERROR_CONFLICT_BIT;
+               }
+       }
+
+       /* DDP context invalidation status: failure or success */
+       if (prog_id == I40E_RX_PROG_STATUS_DESC_FCOE_CTXT_INVL_STATUS) {
+               if (I40E_RX_PROG_FCOE_ERROR_INVLFAIL(error)) {
+                       dev_err(&pf->pdev->dev, "xid %x ddp->xid %x INVALIDATION FAILURE\n",
+                               xid, ddp->xid);
+                       ddp->prerr |= I40E_RX_PROG_FCOE_ERROR_INVLFAIL_BIT;
+               }
+               /* clear the flag so we can retry invalidation */
+               clear_bit(__I40E_FCOE_DDP_ABORTED, &ddp->flags);
+       }
+
+       /* unmap DMA */
+       i40e_fcoe_ddp_unmap(pf, ddp);
+       i40e_fcoe_ddp_clear(ddp);
+}
+
+/**
+ * i40e_fcoe_handle_offload - check ddp status and mark it done
+ * @adapter: i40e adapter
+ * @rx_desc: advanced rx descriptor
+ * @skb: the skb holding the received data
+ *
+ * This checks ddp status.
+ *
+ * Returns : < 0 indicates an error or not a FCOE ddp, 0 indicates
+ * not passing the skb to ULD, > 0 indicates is the length of data
+ * being ddped.
+ *
+ **/
+int i40e_fcoe_handle_offload(struct i40e_ring *rx_ring,
+                            union i40e_rx_desc *rx_desc,
+                            struct sk_buff *skb)
+{
+       struct i40e_pf *pf = rx_ring->vsi->back;
+       struct i40e_fcoe *fcoe = &pf->fcoe;
+       struct fc_frame_header *fh = NULL;
+       struct i40e_fcoe_ddp *ddp = NULL;
+       u32 status, fltstat;
+       u32 error, fcerr;
+       int rc = -EINVAL;
+       u16 ptype;
+       u16 xid;
+       u64 qw;
+
+       /* check this rxd is for programming status */
+       qw = le64_to_cpu(rx_desc->wb.qword1.status_error_len);
+       /* packet descriptor, check packet type */
+       ptype = (qw & I40E_RXD_QW1_PTYPE_MASK) >> I40E_RXD_QW1_PTYPE_SHIFT;
+       if (!i40e_rx_is_fcoe(ptype))
+               goto out_no_ddp;
+
+       error = (qw & I40E_RXD_QW1_ERROR_MASK) >> I40E_RXD_QW1_ERROR_SHIFT;
+       fcerr = (error >> I40E_RX_DESC_ERROR_L3L4E_SHIFT) &
+                I40E_RX_DESC_FCOE_ERROR_MASK;
+
+       /* check stateless offload error */
+       if (unlikely(fcerr == I40E_RX_DESC_ERROR_L3L4E_PROT)) {
+               dev_err(&pf->pdev->dev, "Protocol Error\n");
+               skb->ip_summed = CHECKSUM_NONE;
+       } else {
+               skb->ip_summed = CHECKSUM_UNNECESSARY;
+       }
+
+       /* check hw status on ddp */
+       status = (qw & I40E_RXD_QW1_STATUS_MASK) >> I40E_RXD_QW1_STATUS_SHIFT;
+       fltstat = (status >> I40E_RX_DESC_STATUS_FLTSTAT_SHIFT) &
+                  I40E_RX_DESC_FLTSTAT_FCMASK;
+
+       /* now we are ready to check DDP */
+       fh = i40e_fcoe_fc_frame_header(skb);
+       xid = i40e_fcoe_fc_get_xid(fh);
+       if (!i40e_fcoe_xid_is_valid(xid))
+               goto out_no_ddp;
+
+       /* non DDP normal receive, return to the protocol stack */
+       if (fltstat == I40E_RX_DESC_FLTSTAT_NOMTCH)
+               goto out_no_ddp;
+
+       /* do we have a sw ddp context setup ? */
+       ddp = &fcoe->ddp[xid];
+       if (!ddp->sgl)
+               goto out_no_ddp;
+
+       /* fetch xid from hw rxd wb, which should match up the sw ctxt */
+       xid = le16_to_cpu(rx_desc->wb.qword0.lo_dword.mirr_fcoe.fcoe_ctx_id);
+       if (ddp->xid != xid) {
+               dev_err(&pf->pdev->dev, "xid 0x%x does not match ctx_xid 0x%x\n",
+                       ddp->xid, xid);
+               goto out_put_ddp;
+       }
+
+       /* the same exchange has already errored out */
+       if (ddp->fcerr) {
+               dev_err(&pf->pdev->dev, "xid 0x%x fcerr 0x%x reported fcer 0x%x\n",
+                       xid, ddp->fcerr, fcerr);
+               goto out_put_ddp;
+       }
+
+       /* fcoe param is valid by now with correct DDPed length */
+       ddp->len = le32_to_cpu(rx_desc->wb.qword0.hi_dword.fcoe_param);
+       ddp->fcerr = fcerr;
+       /* header posting only, useful only for target mode and debugging */
+       if (fltstat == I40E_RX_DESC_FLTSTAT_DDP) {
+               /* For target mode, we get header of the last packet but it
+                * does not have the FCoE trailer field, i.e., CRC and EOF
+                * Ordered Set since they are offloaded by the HW, so fill
+                * it up correspondingly to allow the packet to pass through
+                * to the upper protocol stack.
+                */
+               u32 f_ctl = ntoh24(fh->fh_f_ctl);
+
+               if ((f_ctl & FC_FC_END_SEQ) &&
+                   (fh->fh_r_ctl == FC_RCTL_DD_SOL_DATA)) {
+                       struct fcoe_crc_eof *crc = NULL;
+
+                       crc = (struct fcoe_crc_eof *)skb_put(skb, sizeof(*crc));
+                       crc->fcoe_eof = FC_EOF_T;
+               } else {
+                       /* otherwise, drop the header only frame */
+                       rc = 0;
+                       goto out_no_ddp;
+               }
+       }
+
+out_put_ddp:
+       /* either we got RSP or we have an error, unmap DMA in both cases */
+       i40e_fcoe_ddp_unmap(pf, ddp);
+       if (ddp->len && !ddp->fcerr) {
+               int pkts;
+
+               rc = ddp->len;
+               i40e_fcoe_ddp_clear(ddp);
+               ddp->len = rc;
+               pkts = DIV_ROUND_UP(rc, 2048);
+               rx_ring->stats.bytes += rc;
+               rx_ring->stats.packets += pkts;
+               rx_ring->q_vector->rx.total_bytes += rc;
+               rx_ring->q_vector->rx.total_packets += pkts;
+               set_bit(__I40E_FCOE_DDP_DONE, &ddp->flags);
+       }
+
+out_no_ddp:
+       return rc;
+}
+
+/**
+ * i40e_fcoe_ddp_setup - called to set up ddp context
+ * @netdev: the corresponding net_device
+ * @xid: the exchange id requesting ddp
+ * @sgl: the scatter-gather list for this request
+ * @sgc: the number of scatter-gather items
+ * @target_mode: indicates this is a DDP request for target
+ *
+ * Returns : 1 for success and 0 for no DDP on this I/O
+ **/
+static int i40e_fcoe_ddp_setup(struct net_device *netdev, u16 xid,
+                              struct scatterlist *sgl, unsigned int sgc,
+                              int target_mode)
+{
+       static const unsigned int bufflen = I40E_FCOE_DDP_BUF_MIN;
+       struct i40e_netdev_priv *np = netdev_priv(netdev);
+       struct i40e_fcoe_ddp_pool *ddp_pool;
+       struct i40e_pf *pf = np->vsi->back;
+       struct i40e_fcoe *fcoe = &pf->fcoe;
+       unsigned int i, j, dmacount;
+       struct i40e_fcoe_ddp *ddp;
+       unsigned int firstoff = 0;
+       unsigned int thisoff = 0;
+       unsigned int thislen = 0;
+       struct scatterlist *sg;
+       dma_addr_t addr = 0;
+       unsigned int len;
+
+       if (xid >= I40E_FCOE_DDP_MAX) {
+               dev_warn(&pf->pdev->dev, "xid=0x%x out-of-range\n", xid);
+               return 0;
+       }
+
+       /* no DDP if we are already down or resetting */
+       if (test_bit(__I40E_DOWN, &pf->state) ||
+           test_bit(__I40E_NEEDS_RESTART, &pf->state)) {
+               dev_info(&pf->pdev->dev, "xid=0x%x device in reset/down\n",
+                        xid);
+               return 0;
+       }
+
+       ddp = &fcoe->ddp[xid];
+       if (ddp->sgl) {
+               dev_info(&pf->pdev->dev, "xid 0x%x w/ non-null sgl=%p nents=%d\n",
+                        xid, ddp->sgl, ddp->sgc);
+               return 0;
+       }
+       i40e_fcoe_ddp_clear(ddp);
+
+       if (!fcoe->ddp_pool) {
+               dev_info(&pf->pdev->dev, "No DDP pool, xid 0x%x\n", xid);
+               return 0;
+       }
+
+       ddp_pool = per_cpu_ptr(fcoe->ddp_pool, get_cpu());
+       if (!ddp_pool->pool) {
+               dev_info(&pf->pdev->dev, "No percpu ddp pool, xid 0x%x\n", xid);
+               goto out_noddp;
+       }
+
+       /* setup dma from scsi command sgl */
+       dmacount = dma_map_sg(&pf->pdev->dev, sgl, sgc, DMA_FROM_DEVICE);
+       if (dmacount == 0) {
+               dev_info(&pf->pdev->dev, "dma_map_sg for sgl %p, sgc %d failed\n",
+                        sgl, sgc);
+               goto out_noddp_unmap;
+       }
+
+       /* alloc the udl from our ddp pool */
+       ddp->udl = dma_pool_alloc(ddp_pool->pool, GFP_ATOMIC, &ddp->udp);
+       if (!ddp->udl) {
+               dev_info(&pf->pdev->dev,
+                        "Failed allocated ddp context, xid 0x%x\n", xid);
+               goto out_noddp_unmap;
+       }
+
+       j = 0;
+       ddp->len = 0;
+       for_each_sg(sgl, sg, dmacount, i) {
+               addr = sg_dma_address(sg);
+               len = sg_dma_len(sg);
+               ddp->len += len;
+               while (len) {
+                       /* max number of buffers allowed in one DDP context */
+                       if (j >= I40E_FCOE_DDP_BUFFCNT_MAX) {
+                               dev_info(&pf->pdev->dev,
+                                        "xid=%x:%d,%d,%d:addr=%llx not enough descriptors\n",
+                                        xid, i, j, dmacount, (u64)addr);
+                               goto out_noddp_free;
+                       }
+
+                       /* get the offset of length of current buffer */
+                       thisoff = addr & ((dma_addr_t)bufflen - 1);
+                       thislen = min_t(unsigned int, (bufflen - thisoff), len);
+                       /* all but the 1st buffer (j == 0)
+                        * must be aligned on bufflen
+                        */
+                       if ((j != 0) && (thisoff))
+                               goto out_noddp_free;
+
+                       /* all but the last buffer
+                        * ((i == (dmacount - 1)) && (thislen == len))
+                        * must end at bufflen
+                        */
+                       if (((i != (dmacount - 1)) || (thislen != len)) &&
+                           ((thislen + thisoff) != bufflen))
+                               goto out_noddp_free;
+
+                       ddp->udl[j] = (u64)(addr - thisoff);
+                       /* only the first buffer may have none-zero offset */
+                       if (j == 0)
+                               firstoff = thisoff;
+                       len -= thislen;
+                       addr += thislen;
+                       j++;
+               }
+       }
+       /* only the last buffer may have non-full bufflen */
+       ddp->lastsize = thisoff + thislen;
+       ddp->firstoff = firstoff;
+       ddp->list_len = j;
+       ddp->pool = ddp_pool->pool;
+       ddp->sgl = sgl;
+       ddp->sgc = sgc;
+       ddp->xid = xid;
+       if (target_mode)
+               set_bit(__I40E_FCOE_DDP_TARGET, &ddp->flags);
+       set_bit(__I40E_FCOE_DDP_INITALIZED, &ddp->flags);
+
+       put_cpu();
+       return 1; /* Success */
+
+out_noddp_free:
+       dma_pool_free(ddp->pool, ddp->udl, ddp->udp);
+       i40e_fcoe_ddp_clear(ddp);
+
+out_noddp_unmap:
+       dma_unmap_sg(&pf->pdev->dev, sgl, sgc, DMA_FROM_DEVICE);
+out_noddp:
+       put_cpu();
+       return 0;
+}
+
+/**
+ * i40e_fcoe_ddp_get - called to set up ddp context in initiator mode
+ * @netdev: the corresponding net_device
+ * @xid: the exchange id requesting ddp
+ * @sgl: the scatter-gather list for this request
+ * @sgc: the number of scatter-gather items
+ *
+ * This is the implementation of net_device_ops.ndo_fcoe_ddp_setup
+ * and is expected to be called from ULD, e.g., FCP layer of libfc
+ * to set up ddp for the corresponding xid of the given sglist for
+ * the corresponding I/O.
+ *
+ * Returns : 1 for success and 0 for no ddp
+ **/
+static int i40e_fcoe_ddp_get(struct net_device *netdev, u16 xid,
+                            struct scatterlist *sgl, unsigned int sgc)
+{
+       return i40e_fcoe_ddp_setup(netdev, xid, sgl, sgc, 0);
+}
+
+/**
+ * i40e_fcoe_ddp_target - called to set up ddp context in target mode
+ * @netdev: the corresponding net_device
+ * @xid: the exchange id requesting ddp
+ * @sgl: the scatter-gather list for this request
+ * @sgc: the number of scatter-gather items
+ *
+ * This is the implementation of net_device_ops.ndo_fcoe_ddp_target
+ * and is expected to be called from ULD, e.g., FCP layer of libfc
+ * to set up ddp for the corresponding xid of the given sglist for
+ * the corresponding I/O. The DDP in target mode is a write I/O request
+ * from the initiator.
+ *
+ * Returns : 1 for success and 0 for no ddp
+ **/
+static int i40e_fcoe_ddp_target(struct net_device *netdev, u16 xid,
+                               struct scatterlist *sgl, unsigned int sgc)
+{
+       return i40e_fcoe_ddp_setup(netdev, xid, sgl, sgc, 1);
+}
+
+/**
+ * i40e_fcoe_program_ddp - programs the HW DDP related descriptors
+ * @tx_ring: transmit ring for this packet
+ * @skb:     the packet to be sent out
+ * @sof: the SOF to indicate class of service
+ *
+ * Determine if it is READ/WRITE command, and finds out if there is
+ * a matching SW DDP context for this command. DDP is applicable
+ * only in case of READ if initiator or WRITE in case of
+ * responder (via checking XFER_RDY).
+ *
+ * Note: caller checks sof and ddp sw context
+ *
+ * Returns : none
+ *
+ **/
+static void i40e_fcoe_program_ddp(struct i40e_ring *tx_ring,
+                                 struct sk_buff *skb,
+                                 struct i40e_fcoe_ddp *ddp, u8 sof)
+{
+       struct i40e_fcoe_filter_context_desc *filter_desc = NULL;
+       struct i40e_fcoe_queue_context_desc *queue_desc = NULL;
+       struct i40e_fcoe_ddp_context_desc *ddp_desc = NULL;
+       struct i40e_pf *pf = tx_ring->vsi->back;
+       u16 i = tx_ring->next_to_use;
+       struct fc_frame_header *fh;
+       u64 flags_rsvd_lanq = 0;
+       bool target_mode;
+
+       /* check if abort is still pending */
+       if (test_bit(__I40E_FCOE_DDP_ABORTED, &ddp->flags)) {
+               dev_warn(&pf->pdev->dev,
+                        "DDP abort is still pending xid:%hx and ddp->flags:%lx:\n",
+                        ddp->xid, ddp->flags);
+               return;
+       }
+
+       /* set the flag to indicate this is programmed */
+       if (test_and_set_bit(__I40E_FCOE_DDP_PROGRAMMED, &ddp->flags)) {
+               dev_warn(&pf->pdev->dev,
+                        "DDP is already programmed for xid:%hx and ddp->flags:%lx:\n",
+                        ddp->xid, ddp->flags);
+               return;
+       }
+
+       /* Prepare the DDP context descriptor */
+       ddp_desc = I40E_DDP_CONTEXT_DESC(tx_ring, i);
+       i++;
+       if (i == tx_ring->count)
+               i = 0;
+
+       ddp_desc->type_cmd_foff_lsize =
+                               cpu_to_le64(I40E_TX_DESC_DTYPE_DDP_CTX  |
+                               ((u64)I40E_FCOE_DDP_CTX_DESC_BSIZE_4K  <<
+                               I40E_FCOE_DDP_CTX_QW1_CMD_SHIFT)        |
+                               ((u64)ddp->firstoff                    <<
+                               I40E_FCOE_DDP_CTX_QW1_FOFF_SHIFT)       |
+                               ((u64)ddp->lastsize                    <<
+                               I40E_FCOE_DDP_CTX_QW1_LSIZE_SHIFT));
+       ddp_desc->rsvd = cpu_to_le64(0);
+
+       /* target mode needs last packet in the sequence  */
+       target_mode = test_bit(__I40E_FCOE_DDP_TARGET, &ddp->flags);
+       if (target_mode)
+               ddp_desc->type_cmd_foff_lsize |=
+                       cpu_to_le64(I40E_FCOE_DDP_CTX_DESC_LASTSEQH);
+
+       /* Prepare queue_context descriptor */
+       queue_desc = I40E_QUEUE_CONTEXT_DESC(tx_ring, i++);
+       if (i == tx_ring->count)
+               i = 0;
+       queue_desc->dmaindx_fbase = cpu_to_le64(ddp->xid | ((u64)ddp->udp));
+       queue_desc->flen_tph = cpu_to_le64(ddp->list_len |
+                               ((u64)(I40E_FCOE_QUEUE_CTX_DESC_TPHRDESC |
+                               I40E_FCOE_QUEUE_CTX_DESC_TPHDATA) <<
+                               I40E_FCOE_QUEUE_CTX_QW1_TPH_SHIFT));
+
+       /* Prepare filter_context_desc */
+       filter_desc = I40E_FILTER_CONTEXT_DESC(tx_ring, i);
+       i++;
+       if (i == tx_ring->count)
+               i = 0;
+
+       fh = (struct fc_frame_header *)skb_transport_header(skb);
+       filter_desc->param = cpu_to_le32(ntohl(fh->fh_parm_offset));
+       filter_desc->seqn = cpu_to_le16(ntohs(fh->fh_seq_cnt));
+       filter_desc->rsvd_dmaindx = cpu_to_le16(ddp->xid <<
+                               I40E_FCOE_FILTER_CTX_QW0_DMAINDX_SHIFT);
+
+       flags_rsvd_lanq = I40E_FCOE_FILTER_CTX_DESC_CTYP_DDP;
+       flags_rsvd_lanq |= (u64)(target_mode ?
+                       I40E_FCOE_FILTER_CTX_DESC_ENODE_RSP :
+                       I40E_FCOE_FILTER_CTX_DESC_ENODE_INIT);
+
+       flags_rsvd_lanq |= (u64)((sof == FC_SOF_I2 || sof == FC_SOF_N2) ?
+                       I40E_FCOE_FILTER_CTX_DESC_FC_CLASS2 :
+                       I40E_FCOE_FILTER_CTX_DESC_FC_CLASS3);
+
+       flags_rsvd_lanq |= ((u64)skb->queue_mapping <<
+                               I40E_FCOE_FILTER_CTX_QW1_LANQINDX_SHIFT);
+       filter_desc->flags_rsvd_lanq = cpu_to_le64(flags_rsvd_lanq);
+
+       /* By this time, all offload related descriptors has been programmed */
+       tx_ring->next_to_use = i;
+}
+
+/**
+ * i40e_fcoe_invalidate_ddp - invalidates DDP in case of abort
+ * @tx_ring: transmit ring for this packet
+ * @skb: the packet associated w/ this DDP invalidation, i.e., ABTS
+ * @ddp: the SW DDP context for this DDP
+ *
+ * Programs the Tx context descriptor to do DDP invalidation.
+ **/
+static void i40e_fcoe_invalidate_ddp(struct i40e_ring *tx_ring,
+                                    struct sk_buff *skb,
+                                    struct i40e_fcoe_ddp *ddp)
+{
+       struct i40e_tx_context_desc *context_desc;
+       int i;
+
+       if (test_and_set_bit(__I40E_FCOE_DDP_ABORTED, &ddp->flags))
+               return;
+
+       i = tx_ring->next_to_use;
+       context_desc = I40E_TX_CTXTDESC(tx_ring, i);
+       i++;
+       if (i == tx_ring->count)
+               i = 0;
+
+       context_desc->tunneling_params = cpu_to_le32(0);
+       context_desc->l2tag2 = cpu_to_le16(0);
+       context_desc->rsvd = cpu_to_le16(0);
+       context_desc->type_cmd_tso_mss = cpu_to_le64(
+               I40E_TX_DESC_DTYPE_FCOE_CTX |
+               (I40E_FCOE_TX_CTX_DESC_OPCODE_DDP_CTX_INVL <<
+               I40E_TXD_CTX_QW1_CMD_SHIFT) |
+               (I40E_FCOE_TX_CTX_DESC_OPCODE_SINGLE_SEND <<
+               I40E_TXD_CTX_QW1_CMD_SHIFT));
+       tx_ring->next_to_use = i;
+}
+
+/**
+ * i40e_fcoe_handle_ddp - check we should setup or invalidate DDP
+ * @tx_ring: transmit ring for this packet
+ * @skb: the packet to be sent out
+ * @sof: the SOF to indicate class of service
+ *
+ * Determine if it is ABTS/READ/XFER_RDY, and finds out if there is
+ * a matching SW DDP context for this command. DDP is applicable
+ * only in case of READ if initiator or WRITE in case of
+ * responder (via checking XFER_RDY). In case this is an ABTS, send
+ * just invalidate the context.
+ **/
+static void i40e_fcoe_handle_ddp(struct i40e_ring *tx_ring,
+                                struct sk_buff *skb, u8 sof)
+{
+       struct i40e_pf *pf = tx_ring->vsi->back;
+       struct i40e_fcoe *fcoe = &pf->fcoe;
+       struct fc_frame_header *fh;
+       struct i40e_fcoe_ddp *ddp;
+       u32 f_ctl;
+       u8 r_ctl;
+       u16 xid;
+
+       fh = (struct fc_frame_header *)skb_transport_header(skb);
+       f_ctl = ntoh24(fh->fh_f_ctl);
+       r_ctl = fh->fh_r_ctl;
+       ddp = NULL;
+
+       if ((r_ctl == FC_RCTL_DD_DATA_DESC) && (f_ctl & FC_FC_EX_CTX)) {
+               /* exchange responder? if so, XFER_RDY for write */
+               xid = ntohs(fh->fh_rx_id);
+               if (i40e_fcoe_xid_is_valid(xid)) {
+                       ddp = &fcoe->ddp[xid];
+                       if ((ddp->xid == xid) &&
+                           (test_bit(__I40E_FCOE_DDP_TARGET, &ddp->flags)))
+                               i40e_fcoe_program_ddp(tx_ring, skb, ddp, sof);
+               }
+       } else if (r_ctl == FC_RCTL_DD_UNSOL_CMD) {
+               /* exchange originator, check READ cmd */
+               xid = ntohs(fh->fh_ox_id);
+               if (i40e_fcoe_xid_is_valid(xid)) {
+                       ddp = &fcoe->ddp[xid];
+                       if ((ddp->xid == xid) &&
+                           (!test_bit(__I40E_FCOE_DDP_TARGET, &ddp->flags)))
+                               i40e_fcoe_program_ddp(tx_ring, skb, ddp, sof);
+               }
+       } else if (r_ctl == FC_RCTL_BA_ABTS) {
+               /* exchange originator, check ABTS */
+               xid = ntohs(fh->fh_ox_id);
+               if (i40e_fcoe_xid_is_valid(xid)) {
+                       ddp = &fcoe->ddp[xid];
+                       if ((ddp->xid == xid) &&
+                           (!test_bit(__I40E_FCOE_DDP_TARGET, &ddp->flags)))
+                               i40e_fcoe_invalidate_ddp(tx_ring, skb, ddp);
+               }
+       }
+}
+
+/**
+ * i40e_fcoe_tso - set up FCoE TSO
+ * @tx_ring:  ring to send buffer on
+ * @skb:      send buffer
+ * @tx_flags: collected send information
+ * @hdr_len:  the tso header length
+ * @sof: the SOF to indicate class of service
+ *
+ * Note must already have sof checked to be either class 2 or class 3 before
+ * calling this function.
+ *
+ * Returns 1 to indicate sequence segmentation offload is properly setup
+ * or returns 0 to indicate no tso is needed, otherwise returns error
+ * code to drop the frame.
+ **/
+static int i40e_fcoe_tso(struct i40e_ring *tx_ring,
+                        struct sk_buff *skb,
+                        u32 tx_flags, u8 *hdr_len, u8 sof)
+{
+       struct i40e_tx_context_desc *context_desc;
+       u32 cd_type, cd_cmd, cd_tso_len, cd_mss;
+       struct fc_frame_header *fh;
+       u64 cd_type_cmd_tso_mss;
+
+       /* must match gso type as FCoE */
+       if (!skb_is_gso(skb))
+               return 0;
+
+       /* is it the expected gso type for FCoE ?*/
+       if (skb_shinfo(skb)->gso_type != SKB_GSO_FCOE) {
+               netdev_err(skb->dev,
+                          "wrong gso type %d:expecting SKB_GSO_FCOE\n",
+                          skb_shinfo(skb)->gso_type);
+               return -EINVAL;
+       }
+
+       /* header and trailer are inserted by hw */
+       *hdr_len = skb_transport_offset(skb) + sizeof(struct fc_frame_header) +
+                  sizeof(struct fcoe_crc_eof);
+
+       /* check sof to decide a class 2 or 3 TSO */
+       if (likely(i40e_fcoe_sof_is_class3(sof)))
+               cd_cmd = I40E_FCOE_TX_CTX_DESC_OPCODE_TSO_FC_CLASS3;
+       else
+               cd_cmd = I40E_FCOE_TX_CTX_DESC_OPCODE_TSO_FC_CLASS2;
+
+       /* param field valid? */
+       fh = (struct fc_frame_header *)skb_transport_header(skb);
+       if (fh->fh_f_ctl[2] & FC_FC_REL_OFF)
+               cd_cmd |= I40E_FCOE_TX_CTX_DESC_RELOFF;
+
+       /* fill the field values */
+       cd_type = I40E_TX_DESC_DTYPE_FCOE_CTX;
+       cd_tso_len = skb->len - *hdr_len;
+       cd_mss = skb_shinfo(skb)->gso_size;
+       cd_type_cmd_tso_mss =
+               ((u64)cd_type  << I40E_TXD_CTX_QW1_DTYPE_SHIFT)     |
+               ((u64)cd_cmd     << I40E_TXD_CTX_QW1_CMD_SHIFT)     |
+               ((u64)cd_tso_len << I40E_TXD_CTX_QW1_TSO_LEN_SHIFT) |
+               ((u64)cd_mss     << I40E_TXD_CTX_QW1_MSS_SHIFT);
+
+       /* grab the next descriptor */
+       context_desc = I40E_TX_CTXTDESC(tx_ring, tx_ring->next_to_use);
+       tx_ring->next_to_use++;
+       if (tx_ring->next_to_use == tx_ring->count)
+               tx_ring->next_to_use = 0;
+
+       context_desc->tunneling_params = 0;
+       context_desc->l2tag2 = cpu_to_le16((tx_flags & I40E_TX_FLAGS_VLAN_MASK)
+                                           >> I40E_TX_FLAGS_VLAN_SHIFT);
+       context_desc->type_cmd_tso_mss = cpu_to_le64(cd_type_cmd_tso_mss);
+
+       return 1;
+}
+
+/**
+ * i40e_fcoe_tx_map - build the tx descriptor
+ * @tx_ring:  ring to send buffer on
+ * @skb:      send buffer
+ * @first:    first buffer info buffer to use
+ * @tx_flags: collected send information
+ * @hdr_len:  ptr to the size of the packet header
+ * @eof:      the frame eof value
+ *
+ * Note, for FCoE, sof and eof are already checked
+ **/
+static void i40e_fcoe_tx_map(struct i40e_ring *tx_ring,
+                            struct sk_buff *skb,
+                            struct i40e_tx_buffer *first,
+                            u32 tx_flags, u8 hdr_len, u8 eof)
+{
+       u32 td_offset = 0;
+       u32 td_cmd = 0;
+       u32 maclen;
+
+       /* insert CRC */
+       td_cmd = I40E_TX_DESC_CMD_ICRC;
+
+       /* setup MACLEN */
+       maclen = skb_network_offset(skb);
+       if (tx_flags & I40E_TX_FLAGS_SW_VLAN)
+               maclen += sizeof(struct vlan_hdr);
+
+       if (skb->protocol == htons(ETH_P_FCOE)) {
+               /* for FCoE, maclen should exclude ether type */
+               maclen -= 2;
+               /* setup type as FCoE and EOF insertion */
+               td_cmd |= (I40E_TX_DESC_CMD_FCOET | i40e_fcoe_ctxt_eof(eof));
+               /* setup FCoELEN and FCLEN */
+               td_offset |= ((((sizeof(struct fcoe_hdr) + 2) >> 2) <<
+                               I40E_TX_DESC_LENGTH_IPLEN_SHIFT) |
+                             ((sizeof(struct fc_frame_header) >> 2) <<
+                               I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT));
+               /* trim to exclude trailer */
+               pskb_trim(skb, skb->len - sizeof(struct fcoe_crc_eof));
+       }
+
+       /* MACLEN is ether header length in words not bytes */
+       td_offset |= (maclen >> 1) << I40E_TX_DESC_LENGTH_MACLEN_SHIFT;
+
+       return i40e_tx_map(tx_ring, skb, first, tx_flags, hdr_len,
+                          td_cmd, td_offset);
+}
+
+/**
+ * i40e_fcoe_set_skb_header - adjust skb header point for FIP/FCoE/FC
+ * @skb: the skb to be adjusted
+ *
+ * Returns true if this skb is a FCoE/FIP or VLAN carried FCoE/FIP and then
+ * adjusts the skb header pointers correspondingly. Otherwise, returns false.
+ **/
+static inline int i40e_fcoe_set_skb_header(struct sk_buff *skb)
+{
+       __be16 protocol = skb->protocol;
+
+       skb_reset_mac_header(skb);
+       skb->mac_len = sizeof(struct ethhdr);
+       if (protocol == htons(ETH_P_8021Q)) {
+               struct vlan_ethhdr *veth = (struct vlan_ethhdr *)eth_hdr(skb);
+
+               protocol = veth->h_vlan_encapsulated_proto;
+               skb->mac_len += sizeof(struct vlan_hdr);
+       }
+
+       /* FCoE or FIP only */
+       if ((protocol != htons(ETH_P_FIP)) &&
+           (protocol != htons(ETH_P_FCOE)))
+               return -EINVAL;
+
+       /* set header to L2 of FCoE/FIP */
+       skb_set_network_header(skb, skb->mac_len);
+       if (protocol == htons(ETH_P_FIP))
+               return 0;
+
+       /* set header to L3 of FC */
+       skb_set_transport_header(skb, skb->mac_len + sizeof(struct fcoe_hdr));
+       return 0;
+}
+
+/**
+ * i40e_fcoe_xmit_frame - transmit buffer
+ * @skb:     send buffer
+ * @netdev:  the fcoe netdev
+ *
+ * Returns 0 if sent, else an error code
+ **/
+static netdev_tx_t i40e_fcoe_xmit_frame(struct sk_buff *skb,
+                                       struct net_device *netdev)
+{
+       struct i40e_netdev_priv *np = netdev_priv(skb->dev);
+       struct i40e_vsi *vsi = np->vsi;
+       struct i40e_ring *tx_ring = vsi->tx_rings[skb->queue_mapping];
+       struct i40e_tx_buffer *first;
+       __be16 protocol = skb->protocol;
+
+       u32 tx_flags = 0;
+       u8 hdr_len = 0;
+       u8 sof = 0;
+       u8 eof = 0;
+       int fso;
+
+       if (i40e_fcoe_set_skb_header(skb))
+               goto out_drop;
+
+       if (!i40e_xmit_descriptor_count(skb, tx_ring))
+               return NETDEV_TX_BUSY;
+
+       /* prepare the xmit flags */
+       if (i40e_tx_prepare_vlan_flags(skb, tx_ring, &tx_flags))
+               goto out_drop;
+
+       /* record the location of the first descriptor for this packet */
+       first = &tx_ring->tx_bi[tx_ring->next_to_use];
+
+       if (protocol == htons(ETH_P_8021Q)) {
+               struct vlan_ethhdr *veth = (struct vlan_ethhdr *)eth_hdr(skb);
+
+               protocol = veth->h_vlan_encapsulated_proto;
+       }
+       /* FIP is a regular L2 traffic w/o offload */
+       if (protocol == htons(ETH_P_FIP))
+               goto out_send;
+
+       /* check sof and eof, only supports FC Class 2 or 3 */
+       if (i40e_fcoe_fc_sof(skb, &sof) || i40e_fcoe_fc_eof(skb, &eof)) {
+               netdev_err(netdev, "SOF/EOF error:%02x - %02x\n", sof, eof);
+               goto out_drop;
+       }
+
+       /* always do FCCRC for FCoE */
+       tx_flags |= I40E_TX_FLAGS_FCCRC;
+
+       /* check we should do sequence offload */
+       fso = i40e_fcoe_tso(tx_ring, skb, tx_flags, &hdr_len, sof);
+       if (fso < 0)
+               goto out_drop;
+       else if (fso)
+               tx_flags |= I40E_TX_FLAGS_FSO;
+       else
+               i40e_fcoe_handle_ddp(tx_ring, skb, sof);
+
+out_send:
+       /* send out the packet */
+       i40e_fcoe_tx_map(tx_ring, skb, first, tx_flags, hdr_len, eof);
+
+       i40e_maybe_stop_tx(tx_ring, DESC_NEEDED);
+       return NETDEV_TX_OK;
+
+out_drop:
+       dev_kfree_skb_any(skb);
+       return NETDEV_TX_OK;
+}
+
+/**
+ * i40e_fcoe_change_mtu - NDO callback to change the Maximum Transfer Unit
+ * @netdev: network interface device structure
+ * @new_mtu: new value for maximum frame size
+ *
+ * Returns error as operation not permitted
+ *
+ **/
+static int i40e_fcoe_change_mtu(struct net_device *netdev, int new_mtu)
+{
+       netdev_warn(netdev, "MTU change is not supported on FCoE interfaces\n");
+       return -EPERM;
+}
+
+/**
+ * i40e_fcoe_set_features - set the netdev feature flags
+ * @netdev: ptr to the netdev being adjusted
+ * @features: the feature set that the stack is suggesting
+ *
+ **/
+static int i40e_fcoe_set_features(struct net_device *netdev,
+                                 netdev_features_t features)
+{
+       struct i40e_netdev_priv *np = netdev_priv(netdev);
+       struct i40e_vsi *vsi = np->vsi;
+
+       if (features & NETIF_F_HW_VLAN_CTAG_RX)
+               i40e_vlan_stripping_enable(vsi);
+       else
+               i40e_vlan_stripping_disable(vsi);
+
+       return 0;
+}
+
+
+static const struct net_device_ops i40e_fcoe_netdev_ops = {
+       .ndo_open               = i40e_open,
+       .ndo_stop               = i40e_close,
+       .ndo_get_stats64        = i40e_get_netdev_stats_struct,
+       .ndo_set_rx_mode        = i40e_set_rx_mode,
+       .ndo_validate_addr      = eth_validate_addr,
+       .ndo_set_mac_address    = i40e_set_mac,
+       .ndo_change_mtu         = i40e_fcoe_change_mtu,
+       .ndo_do_ioctl           = i40e_ioctl,
+       .ndo_tx_timeout         = i40e_tx_timeout,
+       .ndo_vlan_rx_add_vid    = i40e_vlan_rx_add_vid,
+       .ndo_vlan_rx_kill_vid   = i40e_vlan_rx_kill_vid,
+       .ndo_setup_tc           = i40e_setup_tc,
+
+#ifdef CONFIG_NET_POLL_CONTROLLER
+       .ndo_poll_controller    = i40e_netpoll,
+#endif
+       .ndo_start_xmit         = i40e_fcoe_xmit_frame,
+       .ndo_fcoe_enable        = i40e_fcoe_enable,
+       .ndo_fcoe_disable       = i40e_fcoe_disable,
+       .ndo_fcoe_ddp_setup     = i40e_fcoe_ddp_get,
+       .ndo_fcoe_ddp_done      = i40e_fcoe_ddp_put,
+       .ndo_fcoe_ddp_target    = i40e_fcoe_ddp_target,
+       .ndo_set_features       = i40e_fcoe_set_features,
+};
+
+/**
+ * i40e_fcoe_config_netdev - prepares the VSI context for creating a FCoE VSI
+ * @vsi: pointer to the associated VSI struct
+ * @ctxt: pointer to the associated VSI context to be passed to HW
+ *
+ * Returns 0 on success or < 0 on error
+ **/
+void i40e_fcoe_config_netdev(struct net_device *netdev, struct i40e_vsi *vsi)
+{
+       struct i40e_hw *hw = &vsi->back->hw;
+       struct i40e_pf *pf = vsi->back;
+
+       if (vsi->type != I40E_VSI_FCOE)
+               return;
+
+       netdev->features = (NETIF_F_HW_VLAN_CTAG_TX |
+                           NETIF_F_HW_VLAN_CTAG_RX |
+                           NETIF_F_HW_VLAN_CTAG_FILTER);
+
+       netdev->vlan_features = netdev->features;
+       netdev->vlan_features &= ~(NETIF_F_HW_VLAN_CTAG_TX |
+                                  NETIF_F_HW_VLAN_CTAG_RX |
+                                  NETIF_F_HW_VLAN_CTAG_FILTER);
+       netdev->fcoe_ddp_xid = I40E_FCOE_DDP_MAX - 1;
+       netdev->features |= NETIF_F_ALL_FCOE;
+       netdev->vlan_features |= NETIF_F_ALL_FCOE;
+       netdev->hw_features |= netdev->features;
+       netdev->priv_flags |= IFF_UNICAST_FLT;
+       netdev->priv_flags |= IFF_SUPP_NOFCS;
+
+       strlcpy(netdev->name, "fcoe%d", IFNAMSIZ-1);
+       netdev->mtu = FCOE_MTU;
+       SET_NETDEV_DEV(netdev, &pf->pdev->dev);
+       i40e_add_filter(vsi, hw->mac.san_addr, 0, false, false);
+       i40e_add_filter(vsi, (u8[6]) FC_FCOE_FLOGI_MAC, 0, false, false);
+       i40e_add_filter(vsi, FIP_ALL_FCOE_MACS, 0, false, false);
+       i40e_add_filter(vsi, FIP_ALL_ENODE_MACS, 0, false, false);
+       i40e_add_filter(vsi, FIP_ALL_VN2VN_MACS, 0, false, false);
+       i40e_add_filter(vsi, FIP_ALL_P2P_MACS, 0, false, false);
+
+       /* use san mac */
+       ether_addr_copy(netdev->dev_addr, hw->mac.san_addr);
+       ether_addr_copy(netdev->perm_addr, hw->mac.san_addr);
+       /* fcoe netdev ops */
+       netdev->netdev_ops = &i40e_fcoe_netdev_ops;
+}
+
+/**
+ * i40e_fcoe_vsi_setup - allocate and set up FCoE VSI
+ * @pf: the pf that VSI is associated with
+ *
+ **/
+void i40e_fcoe_vsi_setup(struct i40e_pf *pf)
+{
+       struct i40e_vsi *vsi;
+       u16 seid;
+       int i;
+
+       if (!(pf->flags & I40E_FLAG_FCOE_ENABLED))
+               return;
+
+       BUG_ON(!pf->vsi[pf->lan_vsi]);
+
+       for (i = 0; i < pf->num_alloc_vsi; i++) {
+               vsi = pf->vsi[i];
+               if (vsi && vsi->type == I40E_VSI_FCOE) {
+                       dev_warn(&pf->pdev->dev,
+                                "FCoE VSI already created\n");
+                       return;
+               }
+       }
+
+       seid = pf->vsi[pf->lan_vsi]->seid;
+       vsi = i40e_vsi_setup(pf, I40E_VSI_FCOE, seid, 0);
+       if (vsi) {
+               dev_dbg(&pf->pdev->dev,
+                       "Successfully created FCoE VSI seid %d id %d uplink_seid %d pf seid %d\n",
+                       vsi->seid, vsi->id, vsi->uplink_seid, seid);
+       } else {
+               dev_info(&pf->pdev->dev, "Failed to create FCoE VSI\n");
+       }
+}
diff --git a/drivers/net/ethernet/intel/i40e/i40e_fcoe.h b/drivers/net/ethernet/intel/i40e/i40e_fcoe.h
new file mode 100644 (file)
index 0000000..21e0f58
--- /dev/null
@@ -0,0 +1,128 @@
+/*******************************************************************************
+ *
+ * Intel Ethernet Controller XL710 Family Linux Driver
+ * Copyright(c) 2013 - 2014 Intel Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ * The full GNU General Public License is included in this distribution in
+ * the file called "COPYING".
+ *
+ * Contact Information:
+ * e1000-devel Mailing List <e1000-devel@lists.sourceforge.net>
+ * Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
+ *
+ ******************************************************************************/
+
+#ifndef _I40E_FCOE_H_
+#define _I40E_FCOE_H_
+
+/* FCoE HW context helper macros */
+#define I40E_DDP_CONTEXT_DESC(R, i)     \
+       (&(((struct i40e_fcoe_ddp_context_desc *)((R)->desc))[i]))
+
+#define I40E_QUEUE_CONTEXT_DESC(R, i)   \
+       (&(((struct i40e_fcoe_queue_context_desc *)((R)->desc))[i]))
+
+#define I40E_FILTER_CONTEXT_DESC(R, i)  \
+       (&(((struct i40e_fcoe_filter_context_desc *)((R)->desc))[i]))
+
+
+/* receive queue descriptor filter status for FCoE */
+#define I40E_RX_DESC_FLTSTAT_FCMASK    0x3
+#define I40E_RX_DESC_FLTSTAT_NOMTCH    0x0     /* no ddp context match */
+#define I40E_RX_DESC_FLTSTAT_NODDP     0x1     /* no ddp due to error */
+#define I40E_RX_DESC_FLTSTAT_DDP       0x2     /* DDPed payload, post header */
+#define I40E_RX_DESC_FLTSTAT_FCPRSP    0x3     /* FCP_RSP */
+
+/* receive queue descriptor error codes for FCoE */
+#define I40E_RX_DESC_FCOE_ERROR_MASK           \
+       (I40E_RX_DESC_ERROR_L3L4E_PROT |        \
+        I40E_RX_DESC_ERROR_L3L4E_FC |          \
+        I40E_RX_DESC_ERROR_L3L4E_DMAC_ERR |    \
+        I40E_RX_DESC_ERROR_L3L4E_DMAC_WARN)
+
+/* receive queue descriptor programming error */
+#define I40E_RX_PROG_FCOE_ERROR_TBL_FULL(e)    \
+       (((e) >> I40E_RX_PROG_STATUS_DESC_FCOE_TBL_FULL_SHIFT) & 0x1)
+
+#define I40E_RX_PROG_FCOE_ERROR_CONFLICT(e)    \
+       (((e) >> I40E_RX_PROG_STATUS_DESC_FCOE_CONFLICT_SHIFT) & 0x1)
+
+#define I40E_RX_PROG_FCOE_ERROR_TBL_FULL_BIT   \
+       (1 << I40E_RX_PROG_STATUS_DESC_FCOE_TBL_FULL_SHIFT)
+#define I40E_RX_PROG_FCOE_ERROR_CONFLICT_BIT   \
+       (1 << I40E_RX_PROG_STATUS_DESC_FCOE_CONFLICT_SHIFT)
+
+#define I40E_RX_PROG_FCOE_ERROR_INVLFAIL(e)    \
+       I40E_RX_PROG_FCOE_ERROR_CONFLICT(e)
+#define I40E_RX_PROG_FCOE_ERROR_INVLFAIL_BIT   \
+       I40E_RX_PROG_FCOE_ERROR_CONFLICT_BIT
+
+/* FCoE DDP related definitions */
+#define I40E_FCOE_MIN_XID      0x0000  /* the min xid supported by fcoe_sw */
+#define I40E_FCOE_MAX_XID      0x0FFF  /* the max xid supported by fcoe_sw */
+#define I40E_FCOE_DDP_BUFFCNT_MAX      512     /* 9 bits bufcnt */
+#define I40E_FCOE_DDP_PTR_ALIGN                16
+#define I40E_FCOE_DDP_PTR_MAX  (I40E_FCOE_DDP_BUFFCNT_MAX * sizeof(dma_addr_t))
+#define I40E_FCOE_DDP_BUF_MIN  4096
+#define I40E_FCOE_DDP_MAX      2048
+#define I40E_FCOE_FILTER_CTX_QW1_PCTYPE_SHIFT  8
+
+/* supported netdev features for FCoE */
+#define I40E_FCOE_NETIF_FEATURES (NETIF_F_ALL_FCOE | \
+       NETIF_F_HW_VLAN_CTAG_TX | \
+       NETIF_F_HW_VLAN_CTAG_RX | \
+       NETIF_F_HW_VLAN_CTAG_FILTER)
+
+/* DDP context flags */
+enum i40e_fcoe_ddp_flags {
+       __I40E_FCOE_DDP_NONE = 1,
+       __I40E_FCOE_DDP_TARGET,
+       __I40E_FCOE_DDP_INITALIZED,
+       __I40E_FCOE_DDP_PROGRAMMED,
+       __I40E_FCOE_DDP_DONE,
+       __I40E_FCOE_DDP_ABORTED,
+       __I40E_FCOE_DDP_UNMAPPED,
+};
+
+/* DDP SW context struct */
+struct i40e_fcoe_ddp {
+       int len;
+       u16 xid;
+       u16 firstoff;
+       u16 lastsize;
+       u16 list_len;
+       u8 fcerr;
+       u8 prerr;
+       unsigned long flags;
+       unsigned int sgc;
+       struct scatterlist *sgl;
+       dma_addr_t udp;
+       u64 *udl;
+       struct dma_pool *pool;
+
+};
+
+struct i40e_fcoe_ddp_pool {
+       struct dma_pool *pool;
+};
+
+struct i40e_fcoe {
+       unsigned long mode;
+       atomic_t refcnt;
+       struct i40e_fcoe_ddp_pool __percpu *ddp_pool;
+       struct i40e_fcoe_ddp ddp[I40E_FCOE_DDP_MAX];
+};
+
+#endif /* _I40E_FCOE_H_ */