Enables the ib_post_send fastpath verb for posting Send work
requests on QPs.
v2: Fixed some sparse warnings
v3: Fixes endianness related warnings reported by sparse. Changes
some of the macros to inline functions.
v5: Uses ETH_P_IBOE macro for RoCE ethertype
Signed-off-by: Eddie Wai <eddie.wai@xxxxxxxxxxxx>
Signed-off-by: Devesh Sharma <devesh.sharma@xxxxxxxxxxxx>
Signed-off-by: Somnath Kotur <somnath.kotur@xxxxxxxxxxxx>
Signed-off-by: Sriharsha Basavapatna <sriharsha.basavapatna@xxxxxxxxxxxx>
---
drivers/infiniband/hw/bnxt_re/bnxt_re.h | 4 +
drivers/infiniband/hw/bnxt_re/ib_verbs.c | 545 ++++++++++++++++++++++++++++++-
drivers/infiniband/hw/bnxt_re/ib_verbs.h | 2 +
drivers/infiniband/hw/bnxt_re/main.c | 1 +
drivers/infiniband/hw/bnxt_re/qplib_fp.c | 268 +++++++++++++++
drivers/infiniband/hw/bnxt_re/qplib_fp.h | 6 +
6 files changed, 823 insertions(+), 3 deletions(-)
diff --git a/drivers/infiniband/hw/bnxt_re/bnxt_re.h b/drivers/infiniband/hw/bnxt_re/bnxt_re.h
index 9c7a53c..5032ca1 100644
--- a/drivers/infiniband/hw/bnxt_re/bnxt_re.h
+++ b/drivers/infiniband/hw/bnxt_re/bnxt_re.h
@@ -129,6 +129,10 @@ struct bnxt_re_dev {
#define to_bnxt_re_dev(ptr, member) \
container_of((ptr), struct bnxt_re_dev, member)
+#define BNXT_RE_ROCE_V1_PACKET 0
+#define BNXT_RE_ROCEV2_IPV4_PACKET 2
+#define BNXT_RE_ROCEV2_IPV6_PACKET 3
+
static inline struct device *rdev_to_dev(struct bnxt_re_dev *rdev)
{
if (rdev)
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index d4d50e5..9401717 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -40,6 +40,7 @@
#include <linux/types.h>
#include <linux/pci.h>
#include <linux/netdevice.h>
+#include <linux/if_ether.h>
#include <rdma/ib_verbs.h>
#include <rdma/ib_user_verbs.h>
@@ -60,6 +61,20 @@
#include "ib_verbs.h"
#include <rdma/bnxt_re-abi.h>
+static int bnxt_re_build_sgl(struct ib_sge *ib_sg_list,
+ struct bnxt_qplib_sge *sg_list, int num)
+{
+ int i, total = 0;
+
+ for (i = 0; i < num; i++) {
+ sg_list[i].addr = ib_sg_list[i].addr;
+ sg_list[i].lkey = ib_sg_list[i].lkey;
+ sg_list[i].size = ib_sg_list[i].length;
+ total += sg_list[i].size;
+ }
+ return total;
+}
+
/* Device */
struct net_device *bnxt_re_get_netdev(struct ib_device *ibdev, u8 port_num)
{
@@ -699,8 +714,6 @@ static u8 __from_ib_qp_type(enum ib_qp_type type)
return CMDQ_CREATE_QP_TYPE_RC;
case IB_QPT_UD:
return CMDQ_CREATE_QP_TYPE_UD;
- case IB_QPT_RAW_ETHERTYPE:
- return CMDQ_CREATE_QP_TYPE_RAW_ETHERTYPE;
default:
return IB_QPT_MAX;
}
@@ -874,7 +887,6 @@ struct ib_qp *bnxt_re_create_qp(struct ib_pd *ib_pd,
struct bnxt_re_dev *rdev = pd->rdev;
struct bnxt_qplib_dev_attr *dev_attr = &rdev->dev_attr;
struct bnxt_re_qp *qp;
- struct bnxt_re_srq *srq;
struct bnxt_re_cq *cq;
int rc, entries;
@@ -1442,6 +1454,533 @@ int bnxt_re_query_qp(struct ib_qp *ib_qp, struct ib_qp_attr *qp_attr,
return 0;
}
+/* Routine for sending QP1 packets for RoCE V1 an V2
+ */
+static int bnxt_re_build_qp1_send_v2(struct bnxt_re_qp *qp,
+ struct ib_send_wr *wr,
+ struct bnxt_qplib_swqe *wqe,
+ int payload_size)
+{
+ struct ib_device *ibdev = &qp->rdev->ibdev;
+ struct bnxt_re_ah *ah = container_of(ud_wr(wr)->ah, struct bnxt_re_ah,
+ ib_ah);
+ struct bnxt_qplib_ah *qplib_ah = &ah->qplib_ah;
+ struct bnxt_qplib_sge sge;
+ union ib_gid sgid;
+ u8 nw_type;
+ u16 ether_type;
+ struct ib_gid_attr sgid_attr;
+ union ib_gid dgid;
+ bool is_eth = false;
+ bool is_vlan = false;
+ bool is_grh = false;
+ bool is_udp = false;
+ u8 ip_version = 0;
+ u16 vlan_id = 0xFFFF;
+ void *buf;
+ int i, rc = 0, size;
+
+ memset(&qp->qp1_hdr, 0, sizeof(qp->qp1_hdr));
+
+ rc = ib_get_cached_gid(ibdev, 1,
+ qplib_ah->host_sgid_index, &sgid,
+ &sgid_attr);
+ if (rc) {
+ dev_err(rdev_to_dev(qp->rdev),
+ "Failed to query gid at index %d",
+ qplib_ah->host_sgid_index);
+ return rc;
+ }
+ if (sgid_attr.ndev) {
+ if (is_vlan_dev(sgid_attr.ndev))
+ vlan_id = vlan_dev_vlan_id(sgid_attr.ndev);
+ dev_put(sgid_attr.ndev);
+ }
+ /* Get network header type for this GID */
+ nw_type = ib_gid_to_network_type(sgid_attr.gid_type, &sgid);
+ switch (nw_type) {
+ case RDMA_NETWORK_IPV4:
+ nw_type = BNXT_RE_ROCEV2_IPV4_PACKET;
+ break;
+ case RDMA_NETWORK_IPV6:
+ nw_type = BNXT_RE_ROCEV2_IPV6_PACKET;
+ break;
+ default:
+ nw_type = BNXT_RE_ROCE_V1_PACKET;
+ break;
+ }
+ memcpy(&dgid.raw, &qplib_ah->dgid, 16);
+ is_udp = sgid_attr.gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP;
+ if (is_udp) {
+ if (ipv6_addr_v4mapped((struct in6_addr *)&sgid)) {
+ ip_version = 4;
+ ether_type = ETH_P_IP;
+ } else {
+ ip_version = 6;
+ ether_type = ETH_P_IPV6;
+ }
+ is_grh = false;
+ } else {
+ ether_type = ETH_P_IBOE;
+ is_grh = true;
+ }
+
+ is_eth = true;
+ is_vlan = (vlan_id && (vlan_id < 0x1000)) ? true : false;
+
+ ib_ud_header_init(payload_size, !is_eth, is_eth, is_vlan, is_grh,
+ ip_version, is_udp, 0, &qp->qp1_hdr);
+
+ /* ETH */
+ ether_addr_copy(qp->qp1_hdr.eth.dmac_h, ah->qplib_ah.dmac);
+ ether_addr_copy(qp->qp1_hdr.eth.smac_h, qp->qplib_qp.smac);
+
+ /* For vlan, check the sgid for vlan existence */
+
+ if (!is_vlan) {
+ qp->qp1_hdr.eth.type = cpu_to_be16(ether_type);
+ } else {
+ qp->qp1_hdr.vlan.type = cpu_to_be16(ether_type);
+ qp->qp1_hdr.vlan.tag = cpu_to_be16(vlan_id);
+ }
+
+ if (is_grh || (ip_version == 6)) {
+ memcpy(qp->qp1_hdr.grh.source_gid.raw, sgid.raw, sizeof(sgid));
+ memcpy(qp->qp1_hdr.grh.destination_gid.raw, qplib_ah->dgid.data,
+ sizeof(sgid));
+ qp->qp1_hdr.grh.hop_limit = qplib_ah->hop_limit;
+ }
+
+ if (ip_version == 4) {
+ qp->qp1_hdr.ip4.tos = 0;
+ qp->qp1_hdr.ip4.id = 0;
+ qp->qp1_hdr.ip4.frag_off = htons(IP_DF);
+ qp->qp1_hdr.ip4.ttl = qplib_ah->hop_limit;
+
+ memcpy(&qp->qp1_hdr.ip4.saddr, sgid.raw + 12, 4);
+ memcpy(&qp->qp1_hdr.ip4.daddr, qplib_ah->dgid.data + 12, 4);
+ qp->qp1_hdr.ip4.check = ib_ud_ip4_csum(&qp->qp1_hdr);
+ }
+
+ if (is_udp) {
+ qp->qp1_hdr.udp.dport = htons(ROCE_V2_UDP_DPORT);
+ qp->qp1_hdr.udp.sport = htons(0x8CD1);
+ qp->qp1_hdr.udp.csum = 0;
+ }
+
+ /* BTH */
+ if (wr->opcode == IB_WR_SEND_WITH_IMM) {
+ qp->qp1_hdr.bth.opcode = IB_OPCODE_UD_SEND_ONLY_WITH_IMMEDIATE;
+ qp->qp1_hdr.immediate_present = 1;
+ } else {
+ qp->qp1_hdr.bth.opcode = IB_OPCODE_UD_SEND_ONLY;
+ }
+ if (wr->send_flags & IB_SEND_SOLICITED)
+ qp->qp1_hdr.bth.solicited_event = 1;
+ /* pad_count */
+ qp->qp1_hdr.bth.pad_count = (4 - payload_size) & 3;
+
+ /* P_key for QP1 is for all members */
+ qp->qp1_hdr.bth.pkey = cpu_to_be16(0xFFFF);
+ qp->qp1_hdr.bth.destination_qpn = IB_QP1;
+ qp->qp1_hdr.bth.ack_req = 0;
+ qp->send_psn++;
+ qp->send_psn &= BTH_PSN_MASK;
+ qp->qp1_hdr.bth.psn = cpu_to_be32(qp->send_psn);
+ /* DETH */
+ /* Use the priviledged Q_Key for QP1 */
+ qp->qp1_hdr.deth.qkey = cpu_to_be32(IB_QP1_QKEY);
+ qp->qp1_hdr.deth.source_qpn = IB_QP1;
+
+ /* Pack the QP1 to the transmit buffer */
+ buf = bnxt_qplib_get_qp1_sq_buf(&qp->qplib_qp, &sge);
+ if (buf) {
+ size = ib_ud_header_pack(&qp->qp1_hdr, buf);
+ for (i = wqe->num_sge; i; i--) {
+ wqe->sg_list[i].addr = wqe->sg_list[i - 1].addr;
+ wqe->sg_list[i].lkey = wqe->sg_list[i - 1].lkey;
+ wqe->sg_list[i].size = wqe->sg_list[i - 1].size;
+ }
+
+ /*
+ * Max Header buf size for IPV6 RoCE V2 is 86,
+ * which is same as the QP1 SQ header buffer.
+ * Header buf size for IPV4 RoCE V2 can be 66.
+ * ETH(14) + VLAN(4)+ IP(20) + UDP (8) + BTH(20).
+ * Subtract 20 bytes from QP1 SQ header buf size
+ */
+ if (is_udp && ip_version == 4)
+ sge.size -= 20;
+ /*
+ * Max Header buf size for RoCE V1 is 78.
+ * ETH(14) + VLAN(4) + GRH(40) + BTH(20).
+ * Subtract 8 bytes from QP1 SQ header buf size
+ */
+ if (!is_udp)
+ sge.size -= 8;
+
+ /* Subtract 4 bytes for non vlan packets */
+ if (!is_vlan)
+ sge.size -= 4;
+
+ wqe->sg_list[0].addr = sge.addr;
+ wqe->sg_list[0].lkey = sge.lkey;
+ wqe->sg_list[0].size = sge.size;
+ wqe->num_sge++;
+
+ } else {
+ dev_err(rdev_to_dev(qp->rdev), "QP1 buffer is empty!");
+ rc = -ENOMEM;
+ }
+ return rc;
+}
+
+static int is_ud_qp(struct bnxt_re_qp *qp)
+{
+ return qp->qplib_qp.type == CMDQ_CREATE_QP_TYPE_UD;
+}
+
+static int bnxt_re_build_send_wqe(struct bnxt_re_qp *qp,
+ struct ib_send_wr *wr,
+ struct bnxt_qplib_swqe *wqe)
+{
+ struct bnxt_re_ah *ah = NULL;
+
+ if (is_ud_qp(qp)) {
+ ah = container_of(ud_wr(wr)->ah, struct bnxt_re_ah, ib_ah);
+ wqe->send.q_key = ud_wr(wr)->remote_qkey;
+ wqe->send.dst_qp = ud_wr(wr)->remote_qpn;
+ wqe->send.avid = ah->qplib_ah.id;
+ }
+ switch (wr->opcode) {
+ case IB_WR_SEND:
+ wqe->type = BNXT_QPLIB_SWQE_TYPE_SEND;
+ break;
+ case IB_WR_SEND_WITH_IMM:
+ wqe->type = BNXT_QPLIB_SWQE_TYPE_SEND_WITH_IMM;
+ wqe->send.imm_data = wr->ex.imm_data;
+ break;
+ case IB_WR_SEND_WITH_INV:
+ wqe->type = BNXT_QPLIB_SWQE_TYPE_SEND_WITH_INV;
+ wqe->send.inv_key = wr->ex.invalidate_rkey;
+ break;
+ default:
+ return -EINVAL;
+ }
+ if (wr->send_flags & IB_SEND_SIGNALED)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_SIGNAL_COMP;
+ if (wr->send_flags & IB_SEND_FENCE)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_UC_FENCE;
+ if (wr->send_flags & IB_SEND_SOLICITED)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_SOLICIT_EVENT;
+ if (wr->send_flags & IB_SEND_INLINE)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_INLINE;
+
+ return 0;
+}
+
+static int bnxt_re_build_rdma_wqe(struct ib_send_wr *wr,
+ struct bnxt_qplib_swqe *wqe)
+{
+ switch (wr->opcode) {
+ case IB_WR_RDMA_WRITE:
+ wqe->type = BNXT_QPLIB_SWQE_TYPE_RDMA_WRITE;
+ break;
+ case IB_WR_RDMA_WRITE_WITH_IMM:
+ wqe->type = BNXT_QPLIB_SWQE_TYPE_RDMA_WRITE_WITH_IMM;
+ wqe->rdma.imm_data = wr->ex.imm_data;
+ break;
+ case IB_WR_RDMA_READ:
+ wqe->type = BNXT_QPLIB_SWQE_TYPE_RDMA_READ;
+ wqe->rdma.inv_key = wr->ex.invalidate_rkey;
+ break;
+ default:
+ return -EINVAL;
+ }
+ wqe->rdma.remote_va = rdma_wr(wr)->remote_addr;
+ wqe->rdma.r_key = rdma_wr(wr)->rkey;
+ if (wr->send_flags & IB_SEND_SIGNALED)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_SIGNAL_COMP;
+ if (wr->send_flags & IB_SEND_FENCE)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_UC_FENCE;
+ if (wr->send_flags & IB_SEND_SOLICITED)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_SOLICIT_EVENT;
+ if (wr->send_flags & IB_SEND_INLINE)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_INLINE;
+
+ return 0;
+}
+
+static int bnxt_re_build_atomic_wqe(struct ib_send_wr *wr,
+ struct bnxt_qplib_swqe *wqe)
+{
+ switch (wr->opcode) {
+ case IB_WR_ATOMIC_CMP_AND_SWP:
+ wqe->type = BNXT_QPLIB_SWQE_TYPE_ATOMIC_CMP_AND_SWP;
+ wqe->atomic.swap_data = atomic_wr(wr)->swap;
+ break;
+ case IB_WR_ATOMIC_FETCH_AND_ADD:
+ wqe->type = BNXT_QPLIB_SWQE_TYPE_ATOMIC_FETCH_AND_ADD;
+ wqe->atomic.cmp_data = atomic_wr(wr)->compare_add;
+ break;
+ default:
+ return -EINVAL;
+ }
+ wqe->atomic.remote_va = atomic_wr(wr)->remote_addr;
+ wqe->atomic.r_key = atomic_wr(wr)->rkey;
+ if (wr->send_flags & IB_SEND_SIGNALED)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_SIGNAL_COMP;
+ if (wr->send_flags & IB_SEND_FENCE)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_UC_FENCE;
+ if (wr->send_flags & IB_SEND_SOLICITED)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_SOLICIT_EVENT;
+ return 0;
+}
+
+static int bnxt_re_build_inv_wqe(struct ib_send_wr *wr,
+ struct bnxt_qplib_swqe *wqe)
+{
+ wqe->type = BNXT_QPLIB_SWQE_TYPE_LOCAL_INV;
+ wqe->local_inv.inv_l_key = wr->ex.invalidate_rkey;
+
+ if (wr->send_flags & IB_SEND_SIGNALED)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_SIGNAL_COMP;
+ if (wr->send_flags & IB_SEND_FENCE)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_UC_FENCE;
+ if (wr->send_flags & IB_SEND_SOLICITED)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_SOLICIT_EVENT;
+
+ return 0;
+}
+
+static int bnxt_re_build_reg_wqe(struct ib_reg_wr *wr,
+ struct bnxt_qplib_swqe *wqe)
+{
+ struct bnxt_re_mr *mr = container_of(wr->mr, struct bnxt_re_mr, ib_mr);
+ struct bnxt_qplib_frpl *qplib_frpl = &mr->qplib_frpl;
+ int access = wr->access;
+
+ wqe->frmr.pbl_ptr = (__le64 *)qplib_frpl->hwq.pbl_ptr[0];
+ wqe->frmr.pbl_dma_ptr = qplib_frpl->hwq.pbl_dma_ptr[0];
+ wqe->frmr.page_list = mr->pages;
+ wqe->frmr.page_list_len = mr->npages;
+ wqe->frmr.levels = qplib_frpl->hwq.level + 1;
+ wqe->type = BNXT_QPLIB_SWQE_TYPE_REG_MR;
+
+ if (wr->wr.send_flags & IB_SEND_FENCE)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_UC_FENCE;
+ if (wr->wr.send_flags & IB_SEND_SIGNALED)
+ wqe->flags |= BNXT_QPLIB_SWQE_FLAGS_SIGNAL_COMP;
+
+ if (access & IB_ACCESS_LOCAL_WRITE)
+ wqe->frmr.access_cntl |= SQ_FR_PMR_ACCESS_CNTL_LOCAL_WRITE;
+ if (access & IB_ACCESS_REMOTE_READ)
+ wqe->frmr.access_cntl |= SQ_FR_PMR_ACCESS_CNTL_REMOTE_READ;
+ if (access & IB_ACCESS_REMOTE_WRITE)
+ wqe->frmr.access_cntl |= SQ_FR_PMR_ACCESS_CNTL_REMOTE_WRITE;
+ if (access & IB_ACCESS_REMOTE_ATOMIC)
+ wqe->frmr.access_cntl |= SQ_FR_PMR_ACCESS_CNTL_REMOTE_ATOMIC;
+ if (access & IB_ACCESS_MW_BIND)
+ wqe->frmr.access_cntl |= SQ_FR_PMR_ACCESS_CNTL_WINDOW_BIND;
+
+ wqe->frmr.l_key = wr->key;
+ wqe->frmr.length = wr->mr->length;
+ wqe->frmr.pbl_pg_sz_log = (wr->mr->page_size >> PAGE_SHIFT_4K) - 1;
+ wqe->frmr.va = wr->mr->iova;
+ return 0;
+}
+
+static int bnxt_re_copy_inline_data(struct bnxt_re_dev *rdev,
+ struct ib_send_wr *wr,
+ struct bnxt_qplib_swqe *wqe)
+{
+ /* Copy the inline data to the data field */
+ u8 *in_data;
+ u32 i, sge_len;
+ void *sge_addr;
+
+ in_data = wqe->inline_data;
+ for (i = 0; i < wr->num_sge; i++) {
+ sge_addr = (void *)(unsigned long)
+ wr->sg_list[i].addr;
+ sge_len = wr->sg_list[i].length;
+
+ if ((sge_len + wqe->inline_len) >
+ BNXT_QPLIB_SWQE_MAX_INLINE_LENGTH) {
+ dev_err(rdev_to_dev(rdev),
+ "Inline data size requested > supported value");
+ return -EINVAL;
+ }
+ sge_len = wr->sg_list[i].length;
+
+ memcpy(in_data, sge_addr, sge_len);
+ in_data += wr->sg_list[i].length;
+ wqe->inline_len += wr->sg_list[i].length;
+ }
+ return wqe->inline_len;
+}
+
+static int bnxt_re_copy_wr_payload(struct bnxt_re_dev *rdev,
+ struct ib_send_wr *wr,
+ struct bnxt_qplib_swqe *wqe)
+{
+ int payload_sz = 0;
+
+ if (wr->send_flags & IB_SEND_INLINE)
+ payload_sz = bnxt_re_copy_inline_data(rdev, wr, wqe);
+ else
+ payload_sz = bnxt_re_build_sgl(wr->sg_list, wqe->sg_list,
+ wqe->num_sge);
+
+ return payload_sz;
+}
+
+static int bnxt_re_post_send_shadow_qp(struct bnxt_re_dev *rdev,
+ struct bnxt_re_qp *qp,
+ struct ib_send_wr *wr)
+{
+ struct bnxt_qplib_swqe wqe;
+ int rc = 0, payload_sz = 0;
+ unsigned long flags;
+
+ spin_lock_irqsave(&qp->sq_lock, flags);
+ memset(&wqe, 0, sizeof(wqe));
+ while (wr) {
+ /* House keeping */
+ memset(&wqe, 0, sizeof(wqe));
+
+ /* Common */
+ wqe.num_sge = wr->num_sge;
+ if (wr->num_sge > qp->qplib_qp.sq.max_sge) {
+ dev_err(rdev_to_dev(rdev),
+ "Limit exceeded for Send SGEs");
+ rc = -EINVAL;
+ goto bad;
+ }
+
+ payload_sz = bnxt_re_copy_wr_payload(qp->rdev, wr, &wqe);
+ if (payload_sz < 0) {
+ rc = -EINVAL;
+ goto bad;
+ }
+ wqe.wr_id = wr->wr_id;
+
+ wqe.type = BNXT_QPLIB_SWQE_TYPE_SEND;
+
+ rc = bnxt_re_build_send_wqe(qp, wr, &wqe);
+ if (!rc)
+ rc = bnxt_qplib_post_send(&qp->qplib_qp, &wqe);
+bad:
+ if (rc) {
+ dev_err(rdev_to_dev(rdev),
+ "Post send failed opcode = %#x rc = %d",
+ wr->opcode, rc);
+ break;
+ }
+ wr = wr->next;
+ }
+ bnxt_qplib_post_send_db(&qp->qplib_qp);
+ spin_unlock_irqrestore(&qp->sq_lock, flags);
+ return rc;
+}
+
+int bnxt_re_post_send(struct ib_qp *ib_qp, struct ib_send_wr *wr,
+ struct ib_send_wr **bad_wr)
+{
+ struct bnxt_re_qp *qp = container_of(ib_qp, struct bnxt_re_qp, ib_qp);
+ struct bnxt_qplib_swqe wqe;
+ int rc = 0, payload_sz = 0;
+ unsigned long flags;
+
+ spin_lock_irqsave(&qp->sq_lock, flags);
+ while (wr) {
+ /* House keeping */
+ memset(&wqe, 0, sizeof(wqe));
+
+ /* Common */
+ wqe.num_sge = wr->num_sge;
+ if (wr->num_sge > qp->qplib_qp.sq.max_sge) {
+ dev_err(rdev_to_dev(qp->rdev),
+ "Limit exceeded for Send SGEs");
+ rc = -EINVAL;
+ goto bad;
+ }
+
+ payload_sz = bnxt_re_copy_wr_payload(qp->rdev, wr, &wqe);
+ if (payload_sz < 0) {
+ rc = -EINVAL;
+ goto bad;
+ }
+ wqe.wr_id = wr->wr_id;
+
+ switch (wr->opcode) {
+ case IB_WR_SEND:
+ case IB_WR_SEND_WITH_IMM:
+ if (ib_qp->qp_type == IB_QPT_GSI) {
+ rc = bnxt_re_build_qp1_send_v2(qp, wr, &wqe,
+ payload_sz);
+ if (rc)
+ goto bad;
+ wqe.rawqp1.lflags |=
+ SQ_SEND_RAWETH_QP1_LFLAGS_ROCE_CRC;
+ }
+ switch (wr->send_flags) {
+ case IB_SEND_IP_CSUM:
+ wqe.rawqp1.lflags |=
+ SQ_SEND_RAWETH_QP1_LFLAGS_IP_CHKSUM;
+ break;
+ default:
+ break;
+ }
+ /* Fall thru to build the wqe */
+ case IB_WR_SEND_WITH_INV:
+ rc = bnxt_re_build_send_wqe(qp, wr, &wqe);
+ break;
+ case IB_WR_RDMA_WRITE:
+ case IB_WR_RDMA_WRITE_WITH_IMM:
+ case IB_WR_RDMA_READ:
+ rc = bnxt_re_build_rdma_wqe(wr, &wqe);
+ break;
+ case IB_WR_ATOMIC_CMP_AND_SWP:
+ case IB_WR_ATOMIC_FETCH_AND_ADD:
+ rc = bnxt_re_build_atomic_wqe(wr, &wqe);
+ break;
+ case IB_WR_RDMA_READ_WITH_INV:
+ dev_err(rdev_to_dev(qp->rdev),
+ "RDMA Read with Invalidate is not supported");
+ rc = -EINVAL;
+ goto bad;
+ case IB_WR_LOCAL_INV:
+ rc = bnxt_re_build_inv_wqe(wr, &wqe);
+ break;
+ case IB_WR_REG_MR:
+ rc = bnxt_re_build_reg_wqe(reg_wr(wr), &wqe);
+ break;
+ default:
+ /* Unsupported WRs */
+ dev_err(rdev_to_dev(qp->rdev),
+ "WR (%#x) is not supported", wr->opcode);
+ rc = -EINVAL;
+ goto bad;
+ }
+ if (!rc)
+ rc = bnxt_qplib_post_send(&qp->qplib_qp, &wqe);
+bad:
+ if (rc) {
+ dev_err(rdev_to_dev(qp->rdev),
+ "post_send failed op:%#x qps = %#x rc = %d\n",
+ wr->opcode, qp->qplib_qp.state, rc);
+ *bad_wr = wr;
+ break;
+ }
+ wr = wr->next;
+ }
+ bnxt_qplib_post_send_db(&qp->qplib_qp);
+ spin_unlock_irqrestore(&qp->sq_lock, flags);
+
+ return rc;
+}
+
/* Completion Queues */
int bnxt_re_destroy_cq(struct ib_cq *ib_cq)
{
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.h b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
index a3c7ce3..2bb36b9 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.h
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
@@ -163,6 +163,8 @@ int bnxt_re_modify_qp(struct ib_qp *qp, struct ib_qp_attr *qp_attr,
int bnxt_re_query_qp(struct ib_qp *qp, struct ib_qp_attr *qp_attr,
int qp_attr_mask, struct ib_qp_init_attr *qp_init_attr);
int bnxt_re_destroy_qp(struct ib_qp *qp);
+int bnxt_re_post_send(struct ib_qp *qp, struct ib_send_wr *send_wr,
+ struct ib_send_wr **bad_send_wr);
struct ib_cq *bnxt_re_create_cq(struct ib_device *ibdev,
const struct ib_cq_init_attr *attr,
struct ib_ucontext *context,
diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c
index ef99321..cb909c4 100644
--- a/drivers/infiniband/hw/bnxt_re/main.c
+++ b/drivers/infiniband/hw/bnxt_re/main.c
@@ -458,6 +458,7 @@ static int bnxt_re_register_ib(struct bnxt_re_dev *rdev)
ibdev->query_qp = bnxt_re_query_qp;
ibdev->destroy_qp = bnxt_re_destroy_qp;
+ ibdev->post_send = bnxt_re_post_send;
ibdev->create_cq = bnxt_re_create_cq;
ibdev->destroy_cq = bnxt_re_destroy_cq;
ibdev->req_notify_cq = bnxt_re_req_notify_cq;
diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.c b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
index b1a5a5e..b2e9afe 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_fp.c
+++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
@@ -1078,6 +1078,274 @@ int bnxt_qplib_destroy_qp(struct bnxt_qplib_res *res,
return 0;
}
+void *bnxt_qplib_get_qp1_sq_buf(struct bnxt_qplib_qp *qp,
+ struct bnxt_qplib_sge *sge)
+{
+ struct bnxt_qplib_q *sq = &qp->sq;
+ u32 sw_prod;
+
+ memset(sge, 0, sizeof(*sge));
+
+ if (qp->sq_hdr_buf) {
+ sw_prod = HWQ_CMP(sq->hwq.prod, &sq->hwq);
+ sge->addr = (dma_addr_t)(qp->sq_hdr_buf_map +
+ sw_prod * qp->sq_hdr_buf_size);
+ sge->lkey = 0xFFFFFFFF;
+ sge->size = qp->sq_hdr_buf_size;
+ return qp->sq_hdr_buf + sw_prod * sge->size;
+ }
+ return NULL;
+}
+
+void bnxt_qplib_post_send_db(struct bnxt_qplib_qp *qp)
+{
+ struct bnxt_qplib_q *sq = &qp->sq;
+ struct dbr_dbr db_msg = { 0 };
+ u32 sw_prod;
+
+ sw_prod = HWQ_CMP(sq->hwq.prod, &sq->hwq);
+
+ db_msg.index = cpu_to_le32((sw_prod << DBR_DBR_INDEX_SFT) &
+ DBR_DBR_INDEX_MASK);
+ db_msg.type_xid =
+ cpu_to_le32(((qp->id << DBR_DBR_XID_SFT) & DBR_DBR_XID_MASK) |
+ DBR_DBR_TYPE_SQ);
+ /* Flush all the WQE writes to HW */
+ wmb();
+ __iowrite64_copy(qp->dpi->dbr, &db_msg, sizeof(db_msg) / sizeof(u64));
+}
+
+int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp,
+ struct bnxt_qplib_swqe *wqe)
+{
+ struct bnxt_qplib_q *sq = &qp->sq;
+ struct bnxt_qplib_swq *swq;
+ struct sq_send *hw_sq_send_hdr, **hw_sq_send_ptr;
+ struct sq_sge *hw_sge;
+ u32 sw_prod;
+ u8 wqe_size16;
+ int i, rc = 0, data_len = 0, pkt_num = 0;
+ __le32 temp32;
+
+ if (qp->state != CMDQ_MODIFY_QP_NEW_STATE_RTS) {
+ rc = -EINVAL;
+ goto done;
+ }
+ if (HWQ_CMP((sq->hwq.prod + 1), &sq->hwq) ==
+ HWQ_CMP(sq->hwq.cons, &sq->hwq)) {
+ rc = -ENOMEM;
+ goto done;
+ }
+ sw_prod = HWQ_CMP(sq->hwq.prod, &sq->hwq);
+ swq = &sq->swq[sw_prod];
+ swq->wr_id = wqe->wr_id;
+ swq->type = wqe->type;
+ swq->flags = wqe->flags;
+ if (qp->sig_type)
+ swq->flags |= SQ_SEND_FLAGS_SIGNAL_COMP;
+ swq->start_psn = sq->psn & BTH_PSN_MASK;
+
+ hw_sq_send_ptr = (struct sq_send **)sq->hwq.pbl_ptr;
+ hw_sq_send_hdr = &hw_sq_send_ptr[get_sqe_pg(sw_prod)]
+ [get_sqe_idx(sw_prod)];
+
+ memset(hw_sq_send_hdr, 0, BNXT_QPLIB_MAX_SQE_ENTRY_SIZE);
+
+ if (wqe->flags & BNXT_QPLIB_SWQE_FLAGS_INLINE) {
+ /* Copy the inline data */
+ if (wqe->inline_len > BNXT_QPLIB_SWQE_MAX_INLINE_LENGTH) {
+ dev_warn(&sq->hwq.pdev->dev,
+ "QPLIB: Inline data length > 96 detected");
+ data_len = BNXT_QPLIB_SWQE_MAX_INLINE_LENGTH;
+ } else {
+ data_len = wqe->inline_len;
+ }
+ memcpy(hw_sq_send_hdr->data, wqe->inline_data, data_len);
+ wqe_size16 = (data_len + 15) >> 4;
+ } else {
+ for (i = 0, hw_sge = (struct sq_sge *)hw_sq_send_hdr->data;
+ i < wqe->num_sge; i++, hw_sge++) {
+ hw_sge->va_or_pa = cpu_to_le64(wqe->sg_list[i].addr);
+ hw_sge->l_key = cpu_to_le32(wqe->sg_list[i].lkey);
+ hw_sge->size = cpu_to_le32(wqe->sg_list[i].size);
+ data_len += wqe->sg_list[i].size;
+ }
+ /* Each SGE entry = 1 WQE size16 */
+ wqe_size16 = wqe->num_sge;
+ }
+
+ /* Specifics */
+ switch (wqe->type) {
+ case BNXT_QPLIB_SWQE_TYPE_SEND:
+ if (qp->type == CMDQ_CREATE_QP1_TYPE_GSI) {
+ /* Assemble info for Raw Ethertype QPs */
+ struct sq_send_raweth_qp1 *sqe =
+ (struct sq_send_raweth_qp1 *)hw_sq_send_hdr;
+
+ sqe->wqe_type = wqe->type;
+ sqe->flags = wqe->flags;
+ sqe->wqe_size = wqe_size16 +
+ ((offsetof(typeof(*sqe), data) + 15) >> 4);
+ sqe->cfa_action = cpu_to_le16(wqe->rawqp1.cfa_action);
+ sqe->lflags = cpu_to_le16(wqe->rawqp1.lflags);
+ sqe->length = cpu_to_le32(data_len);
+ sqe->cfa_meta = cpu_to_le32((wqe->rawqp1.cfa_meta &
+ SQ_SEND_RAWETH_QP1_CFA_META_VLAN_VID_MASK) <<
+ SQ_SEND_RAWETH_QP1_CFA_META_VLAN_VID_SFT);
+
+ break;
+ }
+ /* else, just fall thru */
+ case BNXT_QPLIB_SWQE_TYPE_SEND_WITH_IMM:
+ case BNXT_QPLIB_SWQE_TYPE_SEND_WITH_INV:
+ {
+ struct sq_send *sqe = (struct sq_send *)hw_sq_send_hdr;
+
+ sqe->wqe_type = wqe->type;
+ sqe->flags = wqe->flags;
+ sqe->wqe_size = wqe_size16 +
+ ((offsetof(typeof(*sqe), data) + 15) >> 4);
+ sqe->inv_key_or_imm_data = cpu_to_le32(
+ wqe->send.inv_key);
+ if (qp->type == CMDQ_CREATE_QP_TYPE_UD) {
+ sqe->q_key = cpu_to_le32(wqe->send.q_key);
+ sqe->dst_qp = cpu_to_le32(
+ wqe->send.dst_qp & SQ_SEND_DST_QP_MASK);
+ sqe->length = cpu_to_le32(data_len);
+ sqe->avid = cpu_to_le32(wqe->send.avid &
+ SQ_SEND_AVID_MASK);
+ sq->psn = (sq->psn + 1) & BTH_PSN_MASK;
+ } else {
+ sqe->length = cpu_to_le32(data_len);
+ sqe->dst_qp = 0;
+ sqe->avid = 0;
+ if (qp->mtu)
+ pkt_num = (data_len + qp->mtu - 1) / qp->mtu;
+ if (!pkt_num)
+ pkt_num = 1;
+ sq->psn = (sq->psn + pkt_num) & BTH_PSN_MASK;
+ }
+ break;
+ }
+ case BNXT_QPLIB_SWQE_TYPE_RDMA_WRITE:
+ case BNXT_QPLIB_SWQE_TYPE_RDMA_WRITE_WITH_IMM:
+ case BNXT_QPLIB_SWQE_TYPE_RDMA_READ:
+ {
+ struct sq_rdma *sqe = (struct sq_rdma *)hw_sq_send_hdr;
+
+ sqe->wqe_type = wqe->type;
+ sqe->flags = wqe->flags;
+ sqe->wqe_size = wqe_size16 +
+ ((offsetof(typeof(*sqe), data) + 15) >> 4);
+ sqe->imm_data = cpu_to_le32(wqe->rdma.inv_key);
+ sqe->length = cpu_to_le32((u32)data_len);
+ sqe->remote_va = cpu_to_le64(wqe->rdma.remote_va);
+ sqe->remote_key = cpu_to_le32(wqe->rdma.r_key);
+ if (qp->mtu)
+ pkt_num = (data_len + qp->mtu - 1) / qp->mtu;
+ if (!pkt_num)
+ pkt_num = 1;
+ sq->psn = (sq->psn + pkt_num) & BTH_PSN_MASK;
+ break;
+ }
+ case BNXT_QPLIB_SWQE_TYPE_ATOMIC_CMP_AND_SWP:
+ case BNXT_QPLIB_SWQE_TYPE_ATOMIC_FETCH_AND_ADD:
+ {
+ struct sq_atomic *sqe = (struct sq_atomic *)hw_sq_send_hdr;
+
+ sqe->wqe_type = wqe->type;
+ sqe->flags = wqe->flags;
+ sqe->remote_key = cpu_to_le32(wqe->atomic.r_key);
+ sqe->remote_va = cpu_to_le64(wqe->atomic.remote_va);
+ sqe->swap_data = cpu_to_le64(wqe->atomic.swap_data);
+ sqe->cmp_data = cpu_to_le64(wqe->atomic.cmp_data);
+ if (qp->mtu)
+ pkt_num = (data_len + qp->mtu - 1) / qp->mtu;
+ if (!pkt_num)
+ pkt_num = 1;
+ sq->psn = (sq->psn + pkt_num) & BTH_PSN_MASK;
+ break;
+ }
+ case BNXT_QPLIB_SWQE_TYPE_LOCAL_INV:
+ {
+ struct sq_localinvalidate *sqe =
+ (struct sq_localinvalidate *)hw_sq_send_hdr;
+
+ sqe->wqe_type = wqe->type;
+ sqe->flags = wqe->flags;
+ sqe->inv_l_key = cpu_to_le32(wqe->local_inv.inv_l_key);
+
+ break;
+ }
+ case BNXT_QPLIB_SWQE_TYPE_FAST_REG_MR:
+ {
+ struct sq_fr_pmr *sqe = (struct sq_fr_pmr *)hw_sq_send_hdr;
+
+ sqe->wqe_type = wqe->type;
+ sqe->flags = wqe->flags;
+ sqe->access_cntl = wqe->frmr.access_cntl |
+ SQ_FR_PMR_ACCESS_CNTL_LOCAL_WRITE;
+ sqe->zero_based_page_size_log =
+ (wqe->frmr.pg_sz_log & SQ_FR_PMR_PAGE_SIZE_LOG_MASK) <<
+ SQ_FR_PMR_PAGE_SIZE_LOG_SFT |
+ (wqe->frmr.zero_based ? SQ_FR_PMR_ZERO_BASED : 0);
+ sqe->l_key = cpu_to_le32(wqe->frmr.l_key);
+ temp32 = cpu_to_le32(wqe->frmr.length);
+ memcpy(sqe->length, &temp32, sizeof(wqe->frmr.length));
+ sqe->numlevels_pbl_page_size_log =
+ ((wqe->frmr.pbl_pg_sz_log <<
+ SQ_FR_PMR_PBL_PAGE_SIZE_LOG_SFT) &
+ SQ_FR_PMR_PBL_PAGE_SIZE_LOG_MASK) |
+ ((wqe->frmr.levels << SQ_FR_PMR_NUMLEVELS_SFT) &
+ SQ_FR_PMR_NUMLEVELS_MASK);
+
+ for (i = 0; i < wqe->frmr.page_list_len; i++)
+ wqe->frmr.pbl_ptr[i] = cpu_to_le64(
+ wqe->frmr.page_list[i] |
+ PTU_PTE_VALID);
+ sqe->pblptr = cpu_to_le64(wqe->frmr.pbl_dma_ptr);
+ sqe->va = cpu_to_le64(wqe->frmr.va);
+
+ break;
+ }
+ case BNXT_QPLIB_SWQE_TYPE_BIND_MW:
+ {
+ struct sq_bind *sqe = (struct sq_bind *)hw_sq_send_hdr;
+
+ sqe->wqe_type = wqe->type;
+ sqe->flags = wqe->flags;
+ sqe->access_cntl = wqe->bind.access_cntl;
+ sqe->mw_type_zero_based = wqe->bind.mw_type |
+ (wqe->bind.zero_based ? SQ_BIND_ZERO_BASED : 0);
+ sqe->parent_l_key = cpu_to_le32(wqe->bind.parent_l_key);
+ sqe->l_key = cpu_to_le32(wqe->bind.r_key);
+ sqe->va = cpu_to_le64(wqe->bind.va);
+ temp32 = cpu_to_le32(wqe->bind.length);
+ memcpy(&sqe->length, &temp32, sizeof(wqe->bind.length));
+ break;
+ }
+ default:
+ /* Bad wqe, return error */
+ rc = -EINVAL;
+ goto done;
+ }
+ swq->next_psn = sq->psn & BTH_PSN_MASK;
+ if (swq->psn_search) {
+ swq->psn_search->opcode_start_psn = cpu_to_le32(
+ ((swq->start_psn << SQ_PSN_SEARCH_START_PSN_SFT) &
+ SQ_PSN_SEARCH_START_PSN_MASK) |
+ ((wqe->type << SQ_PSN_SEARCH_OPCODE_SFT) &
+ SQ_PSN_SEARCH_OPCODE_MASK));
+ swq->psn_search->flags_next_psn = cpu_to_le32(
+ ((swq->next_psn << SQ_PSN_SEARCH_NEXT_PSN_SFT) &
+ SQ_PSN_SEARCH_NEXT_PSN_MASK));
+ }
+
+ sq->hwq.prod++;
+done:
+ return rc;
+}
+
/* CQ */
/* Spinlock must be held */
diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.h b/drivers/infiniband/hw/bnxt_re/qplib_fp.h
index 32e11c9..3db8fb9 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_fp.h
+++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.h
@@ -38,6 +38,7 @@
#ifndef __BNXT_QPLIB_FP_H__
#define __BNXT_QPLIB_FP_H__
+
struct bnxt_qplib_sge {
u64 addr;
u32 lkey;
@@ -415,6 +416,11 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp);
int bnxt_qplib_modify_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp);
int bnxt_qplib_query_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp);
int bnxt_qplib_destroy_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp);
+void *bnxt_qplib_get_qp1_sq_buf(struct bnxt_qplib_qp *qp,
+ struct bnxt_qplib_sge *sge);
+void bnxt_qplib_post_send_db(struct bnxt_qplib_qp *qp);
+int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp,
+ struct bnxt_qplib_swqe *wqe);
int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq);
int bnxt_qplib_destroy_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq);
--
2.5.5
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
