This patch adds atomic operations for hip08, includes
fetchadd and cmpswap operation.
In order to enable atomic, the driver needs to do the
following step:
1. Enable the atomic caps for RoCE device
2. Post the wqe context of atomic type
3. Configure the atomic type of mtpt
Signed-off-by: Lijun Ou <oulijun@xxxxxxxxxx>
---
drivers/infiniband/hw/hns/hns_roce_device.h | 1 +
drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 39 +++++++++++++++++++++++++++--
drivers/infiniband/hw/hns/hns_roce_hw_v2.h | 5 ++++
drivers/infiniband/hw/hns/hns_roce_main.c | 3 ++-
4 files changed, 45 insertions(+), 3 deletions(-)
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index cfb88a4..cdd0bb0 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -193,6 +193,7 @@ enum {
HNS_ROCE_CAP_FLAG_RQ_INLINE = BIT(2),
HNS_ROCE_CAP_FLAG_RECORD_DB = BIT(3),
HNS_ROCE_CAP_FLAG_SQ_RECORD_DB = BIT(4),
+ HNS_ROCE_CAP_FLAG_ATOMIC = BIT(10),
};
enum hns_roce_mtt_type {
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 903ed66..97ff5da 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -54,6 +54,18 @@ static void set_data_seg_v2(struct hns_roce_v2_wqe_data_seg *dseg,
dseg->len = cpu_to_le32(sg->length);
}
+static void set_atomic_seg(struct hns_roce_wqe_atomic_seg *aseg,
+ const struct ib_atomic_wr *wr)
+{
+ if (wr->wr.opcode == IB_WR_ATOMIC_CMP_AND_SWP) {
+ aseg->fetchadd_swap_data = cpu_to_le64(wr->swap);
+ aseg->cmp_data = cpu_to_le64(wr->compare_add);
+ } else {
+ aseg->fetchadd_swap_data = cpu_to_le64(wr->compare_add);
+ aseg->cmp_data = 0;
+ }
+}
+
static void set_extend_sge(struct hns_roce_qp *qp, const struct ib_send_wr *wr,
unsigned int *sge_ind)
{
@@ -179,6 +191,7 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
struct hns_roce_v2_ud_send_wqe *ud_sq_wqe;
struct hns_roce_v2_rc_send_wqe *rc_sq_wqe;
struct hns_roce_qp *qp = to_hr_qp(ibqp);
+ struct hns_roce_v2_wqe_data_seg *dseg;
struct device *dev = hr_dev->dev;
struct hns_roce_v2_db sq_db;
struct ib_qp_attr attr;
@@ -407,6 +420,7 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
roce_set_bit(rc_sq_wqe->byte_4,
V2_RC_SEND_WQE_BYTE_4_OWNER_S, owner_bit);
+ wqe += sizeof(struct hns_roce_v2_rc_send_wqe);
switch (wr->opcode) {
case IB_WR_RDMA_READ:
hr_op = HNS_ROCE_V2_WQE_OP_RDMA_READ;
@@ -443,9 +457,21 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
break;
case IB_WR_ATOMIC_CMP_AND_SWP:
hr_op = HNS_ROCE_V2_WQE_OP_ATOM_CMP_AND_SWAP;
+ rc_sq_wqe->rkey =
+ cpu_to_le32(atomic_wr(wr)->rkey);
+ rc_sq_wqe->va =
+ cpu_to_le32(atomic_wr(wr)->remote_addr);
+ wqe += sizeof(struct hns_roce_v2_wqe_data_seg);
+ set_atomic_seg(wqe, atomic_wr(wr));
break;
case IB_WR_ATOMIC_FETCH_AND_ADD:
hr_op = HNS_ROCE_V2_WQE_OP_ATOM_FETCH_AND_ADD;
+ rc_sq_wqe->rkey =
+ cpu_to_le32(atomic_wr(wr)->rkey);
+ rc_sq_wqe->va =
+ cpu_to_le32(atomic_wr(wr)->remote_addr);
+ wqe += sizeof(struct hns_roce_v2_wqe_data_seg);
+ set_atomic_seg(wqe, atomic_wr(wr));
break;
case IB_WR_MASKED_ATOMIC_CMP_AND_SWP:
hr_op =
@@ -463,7 +489,12 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
roce_set_field(rc_sq_wqe->byte_4,
V2_RC_SEND_WQE_BYTE_4_OPCODE_M,
V2_RC_SEND_WQE_BYTE_4_OPCODE_S, hr_op);
- wqe += sizeof(struct hns_roce_v2_rc_send_wqe);
+ if (wr->opcode == IB_WR_ATOMIC_CMP_AND_SWP ||
+ wr->opcode == IB_WR_ATOMIC_FETCH_AND_ADD)
+ dseg =
+ wqe - sizeof(struct hns_roce_v2_wqe_data_seg);
+ else
+ dseg = wqe;
ret = set_rwqe_data_seg(ibqp, wr, rc_sq_wqe, wqe,
&sge_ind, bad_wr);
@@ -1232,6 +1263,9 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev)
caps->local_ca_ack_delay = 0;
caps->max_mtu = IB_MTU_4096;
+ if (hr_dev->pci_dev->revision == 0x21)
+ caps->flags |= HNS_ROCE_CAP_FLAG_ATOMIC;
+
ret = hns_roce_v2_set_bt(hr_dev);
if (ret)
dev_err(hr_dev->dev, "Configure bt attribute fail, ret = %d.\n",
@@ -1663,7 +1697,8 @@ static int hns_roce_v2_write_mtpt(void *mb_buf, struct hns_roce_mr *mr,
roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_L_INV_EN_S, 0);
roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_BIND_EN_S,
(mr->access & IB_ACCESS_MW_BIND ? 1 : 0));
- roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_ATOMIC_EN_S, 0);
+ roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_ATOMIC_EN_S,
+ mr->access & IB_ACCESS_REMOTE_ATOMIC ? 1 : 0);
roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_RR_EN_S,
(mr->access & IB_ACCESS_REMOTE_READ ? 1 : 0));
roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_RW_EN_S,
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
index 14aa308..6f92722 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
@@ -1564,4 +1564,9 @@ struct hns_roce_eq_context {
#define HNS_ROCE_V2_AEQE_EVENT_QUEUE_NUM_S 0
#define HNS_ROCE_V2_AEQE_EVENT_QUEUE_NUM_M GENMASK(23, 0)
+struct hns_roce_wqe_atomic_seg {
+ __le64 fetchadd_swap_data;
+ __le64 cmp_data;
+};
+
#endif
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index c5cae9a..1306d79 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -215,7 +215,8 @@ static int hns_roce_query_device(struct ib_device *ib_dev,
props->max_pd = hr_dev->caps.num_pds;
props->max_qp_rd_atom = hr_dev->caps.max_qp_dest_rdma;
props->max_qp_init_rd_atom = hr_dev->caps.max_qp_init_rdma;
- props->atomic_cap = IB_ATOMIC_NONE;
+ props->atomic_cap = hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_ATOMIC ?
+ IB_ATOMIC_HCA : IB_ATOMIC_NONE;
props->max_pkeys = 1;
props->local_ca_ack_delay = hr_dev->caps.local_ca_ack_delay;
--
1.9.1
