This patch adds memory window (mw) deallocation support in
kernel space driver.
Signed-off-by: Yixian Liu <liuyixian@xxxxxxxxxx>
---
drivers/infiniband/hw/hns/hns_roce_device.h | 15 ++++
drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 45 ++++++++++++
drivers/infiniband/hw/hns/hns_roce_hw_v2.h | 10 +++
drivers/infiniband/hw/hns/hns_roce_main.c | 9 +++
drivers/infiniband/hw/hns/hns_roce_mr.c | 109 ++++++++++++++++++++++++++++
5 files changed, 188 insertions(+)
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index cfb88a4..eca294b 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_MW = BIT(7),
};
enum hns_roce_mtt_type {
@@ -293,6 +294,16 @@ struct hns_roce_mtt {
enum hns_roce_mtt_type mtt_type;
};
+struct hns_roce_mw {
+ struct ib_mw ibmw;
+ u32 pdn;
+ u32 rkey;
+ int enabled; /* MW's active status */
+ u32 pbl_hop_num;
+ u32 pbl_ba_pg_sz;
+ u32 pbl_buf_pg_sz;
+};
+
/* Only support 4K page size for mr register */
#define MR_SIZE_4K 0
@@ -765,6 +776,7 @@ struct hns_roce_hw {
struct hns_roce_mr *mr, int flags, u32 pdn,
int mr_access_flags, u64 iova, u64 size,
void *mb_buf);
+ int (*mw_write_mtpt)(void *mb_buf, struct hns_roce_mw *mw);
void (*write_cqc)(struct hns_roce_dev *hr_dev,
struct hns_roce_cq *hr_cq, void *mb_buf, u64 *mtts,
dma_addr_t dma_handle, int nent, u32 vector);
@@ -975,6 +987,9 @@ int hns_roce_hw2sw_mpt(struct hns_roce_dev *hr_dev,
unsigned long mpt_index);
unsigned long key_to_hw_index(u32 key);
+struct ib_mw *hns_roce_alloc_mw(struct ib_pd *pd, enum ib_mw_type,
+ struct ib_udata *udata);
+
void hns_roce_buf_free(struct hns_roce_dev *hr_dev, u32 size,
struct hns_roce_buf *buf);
int hns_roce_buf_alloc(struct hns_roce_dev *hr_dev, u32 size, u32 max_direct,
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index b16ad95..35e2874 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -1255,6 +1255,10 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev)
HNS_ROCE_CAP_FLAG_RQ_INLINE |
HNS_ROCE_CAP_FLAG_RECORD_DB |
HNS_ROCE_CAP_FLAG_SQ_RECORD_DB;
+
+ if (hr_dev->pci_dev->revision == 0x21)
+ caps->flags |= HNS_ROCE_CAP_FLAG_MW;
+
caps->pkey_table_len[0] = 1;
caps->gid_table_len[0] = HNS_ROCE_V2_GID_INDEX_NUM;
caps->ceqe_depth = HNS_ROCE_V2_COMP_EQE_NUM;
@@ -1817,6 +1821,46 @@ static int hns_roce_v2_rereg_write_mtpt(struct hns_roce_dev *hr_dev,
return 0;
}
+static int hns_roce_v2_mw_write_mtpt(void *mb_buf, struct hns_roce_mw *mw)
+{
+ struct hns_roce_v2_mpt_entry *mpt_entry;
+
+ mpt_entry = mb_buf;
+ memset(mpt_entry, 0, sizeof(*mpt_entry));
+
+ roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_MPT_ST_M,
+ V2_MPT_BYTE_4_MPT_ST_S, V2_MPT_ST_FREE);
+ roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PD_M,
+ V2_MPT_BYTE_4_PD_S, mw->pdn);
+ roce_set_field(mpt_entry->byte_4_pd_hop_st,
+ V2_MPT_BYTE_4_PBL_HOP_NUM_M,
+ V2_MPT_BYTE_4_PBL_HOP_NUM_S,
+ mw->pbl_hop_num == HNS_ROCE_HOP_NUM_0 ?
+ 0 : mw->pbl_hop_num);
+ roce_set_field(mpt_entry->byte_4_pd_hop_st,
+ V2_MPT_BYTE_4_PBL_BA_PG_SZ_M,
+ V2_MPT_BYTE_4_PBL_BA_PG_SZ_S,
+ mw->pbl_ba_pg_sz + PG_SHIFT_OFFSET);
+
+ roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_R_INV_EN_S, 1);
+ roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_L_INV_EN_S, 1);
+
+ roce_set_bit(mpt_entry->byte_12_mw_pa, V2_MPT_BYTE_12_PA_S, 0);
+ roce_set_bit(mpt_entry->byte_12_mw_pa, V2_MPT_BYTE_12_MR_MW_S, 1);
+ roce_set_bit(mpt_entry->byte_12_mw_pa, V2_MPT_BYTE_12_BPD_S, 1);
+ roce_set_bit(mpt_entry->byte_12_mw_pa, V2_MPT_BYTE_12_BQP_S,
+ mw->ibmw.type == IB_MW_TYPE_1 ? 0 : 1);
+
+ roce_set_field(mpt_entry->byte_64_buf_pa1,
+ V2_MPT_BYTE_64_PBL_BUF_PG_SZ_M,
+ V2_MPT_BYTE_64_PBL_BUF_PG_SZ_S,
+ mw->pbl_buf_pg_sz + PG_SHIFT_OFFSET);
+
+ mpt_entry->lkey = cpu_to_le32(mw->rkey);
+
+ return 0;
+}
+
static void *get_cqe_v2(struct hns_roce_cq *hr_cq, int n)
{
return hns_roce_buf_offset(&hr_cq->hr_buf.hr_buf,
@@ -5159,6 +5203,7 @@ static const struct hns_roce_hw hns_roce_hw_v2 = {
.set_mac = hns_roce_v2_set_mac,
.write_mtpt = hns_roce_v2_write_mtpt,
.rereg_write_mtpt = hns_roce_v2_rereg_write_mtpt,
+ .mw_write_mtpt = hns_roce_v2_mw_write_mtpt,
.write_cqc = hns_roce_v2_write_cqc,
.set_hem = hns_roce_v2_set_hem,
.clear_hem = hns_roce_v2_clear_hem,
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
index 14aa308..746fe80 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
@@ -324,6 +324,7 @@ struct hns_roce_v2_cq_context {
enum{
V2_MPT_ST_VALID = 0x1,
+ V2_MPT_ST_FREE = 0x2,
};
enum hns_roce_v2_qp_state {
@@ -878,8 +879,17 @@ struct hns_roce_v2_mpt_entry {
#define V2_MPT_BYTE_8_LW_EN_S 7
+#define V2_MPT_BYTE_8_MW_CNT_S 8
+#define V2_MPT_BYTE_8_MW_CNT_M GENMASK(31, 8)
+
#define V2_MPT_BYTE_12_PA_S 1
+#define V2_MPT_BYTE_12_MR_MW_S 4
+
+#define V2_MPT_BYTE_12_BPD_S 5
+
+#define V2_MPT_BYTE_12_BQP_S 6
+
#define V2_MPT_BYTE_12_INNER_PA_VLD_S 7
#define V2_MPT_BYTE_12_MW_BIND_QPN_S 8
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index c5cae9a..80ef384 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -205,6 +205,8 @@ static int hns_roce_query_device(struct ib_device *ib_dev,
props->max_qp = hr_dev->caps.num_qps;
props->max_qp_wr = hr_dev->caps.max_wqes;
props->device_cap_flags = IB_DEVICE_PORT_ACTIVE_EVENT |
+ IB_DEVICE_MEM_WINDOW |
+ IB_DEVICE_MEM_WINDOW_TYPE_2B |
IB_DEVICE_RC_RNR_NAK_GEN;
props->max_send_sge = hr_dev->caps.max_sq_sg;
props->max_recv_sge = hr_dev->caps.max_rq_sg;
@@ -212,6 +214,7 @@ static int hns_roce_query_device(struct ib_device *ib_dev,
props->max_cq = hr_dev->caps.num_cqs;
props->max_cqe = hr_dev->caps.max_cqes;
props->max_mr = hr_dev->caps.num_mtpts;
+ props->max_mw = hr_dev->caps.num_mtpts;
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;
@@ -584,6 +587,12 @@ static int hns_roce_register_device(struct hns_roce_dev *hr_dev)
ib_dev->uverbs_cmd_mask |= (1ULL << IB_USER_VERBS_CMD_REREG_MR);
}
+ /* MW */
+ if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_MW) {
+ ib_dev->alloc_mw = hns_roce_alloc_mw;
+ ib_dev->uverbs_cmd_mask |= (1ULL << IB_USER_VERBS_CMD_ALLOC_MW);
+ }
+
/* OTHERS */
ib_dev->get_port_immutable = hns_roce_port_immutable;
ib_dev->disassociate_ucontext = hns_roce_disassociate_ucontext;
diff --git a/drivers/infiniband/hw/hns/hns_roce_mr.c b/drivers/infiniband/hw/hns/hns_roce_mr.c
index eb26a5f..d45d46b 100644
--- a/drivers/infiniband/hw/hns/hns_roce_mr.c
+++ b/drivers/infiniband/hw/hns/hns_roce_mr.c
@@ -1201,3 +1201,112 @@ int hns_roce_dereg_mr(struct ib_mr *ibmr)
return ret;
}
+
+static void hns_roce_mw_free(struct hns_roce_dev *hr_dev,
+ struct hns_roce_mw *mw)
+{
+ struct device *dev = hr_dev->dev;
+ int ret;
+
+ if (mw->enabled) {
+ ret = hns_roce_hw2sw_mpt(hr_dev, NULL, key_to_hw_index(mw->rkey)
+ & (hr_dev->caps.num_mtpts - 1));
+ if (ret)
+ dev_warn(dev, "MW HW2SW_MPT failed (%d)\n", ret);
+
+ hns_roce_table_put(hr_dev, &hr_dev->mr_table.mtpt_table,
+ key_to_hw_index(mw->rkey));
+ }
+
+ hns_roce_bitmap_free(&hr_dev->mr_table.mtpt_bitmap,
+ key_to_hw_index(mw->rkey), BITMAP_NO_RR);
+}
+
+static int hns_roce_mw_enable(struct hns_roce_dev *hr_dev,
+ struct hns_roce_mw *mw)
+{
+ unsigned long mtpt_idx = key_to_hw_index(mw->rkey);
+ struct device *dev = hr_dev->dev;
+ struct hns_roce_cmd_mailbox *mailbox;
+ struct hns_roce_mr_table *mr_table = &hr_dev->mr_table;
+ int ret;
+
+ /* prepare HEM entry memory */
+ ret = hns_roce_table_get(hr_dev, &mr_table->mtpt_table, mtpt_idx);
+ if (ret)
+ return ret;
+
+ mailbox = hns_roce_alloc_cmd_mailbox(hr_dev);
+ if (IS_ERR(mailbox)) {
+ ret = PTR_ERR(mailbox);
+ goto err_table;
+ }
+
+ ret = hr_dev->hw->mw_write_mtpt(mailbox->buf, mw);
+ if (ret) {
+ dev_err(dev, "MW write mtpt fail!\n");
+ goto err_page;
+ }
+
+ ret = hns_roce_sw2hw_mpt(hr_dev, mailbox,
+ mtpt_idx & (hr_dev->caps.num_mtpts - 1));
+ if (ret) {
+ dev_err(dev, "MW sw2hw_mpt failed (%d)\n", ret);
+ goto err_page;
+ }
+
+ mw->enabled = 1;
+
+ hns_roce_free_cmd_mailbox(hr_dev, mailbox);
+
+ return 0;
+
+err_page:
+ hns_roce_free_cmd_mailbox(hr_dev, mailbox);
+
+err_table:
+ hns_roce_table_put(hr_dev, &mr_table->mtpt_table, mtpt_idx);
+
+ return ret;
+}
+
+struct ib_mw *hns_roce_alloc_mw(struct ib_pd *ib_pd, enum ib_mw_type type,
+ struct ib_udata *udata)
+{
+ struct hns_roce_dev *hr_dev = to_hr_dev(ib_pd->device);
+ struct hns_roce_mw *mw;
+ unsigned long index = 0;
+ int ret;
+
+ mw = kmalloc(sizeof(*mw), GFP_KERNEL);
+ if (!mw)
+ return ERR_PTR(-ENOMEM);
+
+ /* Allocate a key for mw from bitmap */
+ ret = hns_roce_bitmap_alloc(&hr_dev->mr_table.mtpt_bitmap, &index);
+ if (ret)
+ goto err_bitmap;
+
+ mw->rkey = hw_index_to_key(index);
+
+ mw->ibmw.rkey = mw->rkey;
+ mw->ibmw.type = type;
+ mw->pdn = to_hr_pd(ib_pd)->pdn;
+ mw->pbl_hop_num = hr_dev->caps.pbl_hop_num;
+ mw->pbl_ba_pg_sz = hr_dev->caps.pbl_ba_pg_sz;
+ mw->pbl_buf_pg_sz = hr_dev->caps.pbl_buf_pg_sz;
+
+ ret = hns_roce_mw_enable(hr_dev, mw);
+ if (ret)
+ goto err_mw;
+
+ return &mw->ibmw;
+
+err_mw:
+ hns_roce_mw_free(hr_dev, mw);
+
+err_bitmap:
+ kfree(mw);
+
+ return ERR_PTR(ret);
+}
--
2.7.4
