RDMA/hns: Fix bug when wqe num is larger than 16K

hip08 can support up to 32768 wqes in one qp. currently if the wqe num
is larger than 16384, the driver will lead a calltrace as follows.

[21361.393725] Call trace:
[21361.398605]  hns_roce_v2_modify_qp+0xbcc/0x1360 [hns_roce_hw_v2]
[21361.410627]  hns_roce_modify_qp+0x1d8/0x2f8 [hns_roce]
[21361.420906]  _ib_modify_qp+0x70/0x118
[21361.428222]  ib_modify_qp+0x14/0x1c
[21361.435193]  rt_ktest_modify_qp+0xb8/0x650 [rdma_test]
[21361.445472]  exec_modify_qp_cmd+0x110/0x4d8 [rdma_test]
[21361.455924]  rt_ktest_dispatch_cmd_3+0xa94/0x2edc [rdma_test]
[21361.467422]  rt_ktest_dispatch_cmd_2+0x9c/0x108 [rdma_test]
[21361.478570]  rt_ktest_dispatch_cmd+0x138/0x904 [rdma_test]
[21361.489545]  rt_ktest_dev_write+0x328/0x4b0 [rdma_test]
[21361.499998]  __vfs_write+0x38/0x15c
[21361.506966]  vfs_write+0xa8/0x1a0
[21361.513586]  ksys_write+0x50/0xb0
[21361.520206]  sys_write+0xc/0x14
[21361.526479]  el0_svc_naked+0x30/0x34
[21361.533622] Code: 1ac10841 d37d7c22 0b000021 d37df021 (f86268c0)
[21361.545815] ---[ end trace e2a1feb2c3d7f13c ]---

When the wqe num is larger than 16384, hns_roce_table_find will return an
invalid mtt, this will lead an kernel paging requet error if the driver try
to access it. It's the mtt design defect which can't support up to the max
wqe num of hip08.

This patch fixs it by replacing mtt with mtr for wqe.

Fixes: 926a01d ("RDMA/hns: Add QP operations support for hip08 SoC")
Signed-off-by: Xi Wang <[email protected]>
Signed-off-by: Lijun Ou <[email protected]>
Signed-off-by: Doug Ledford <[email protected]>

Author: oulijun

drivers/infiniband/hw/hns/hns_roce_device.h

@@ -660,6 +660,14 @@ struct hns_roce_qp {
 
 	struct ib_umem		*umem;
 	struct hns_roce_mtt	mtt;
+	struct hns_roce_mtr	mtr;
+
+	/* this define must less than HNS_ROCE_MAX_BT_REGION */
+#define HNS_ROCE_WQE_REGION_MAX	 3
+	struct hns_roce_buf_region regions[HNS_ROCE_WQE_REGION_MAX];
+	int			region_cnt;
+	int                     wqe_bt_pg_shift;
+
 	u32			buff_size;
 	struct mutex		mutex;
 	u8			port;
@@ -870,6 +878,9 @@ struct hns_roce_caps {
 	u32		mtt_ba_pg_sz;
 	u32		mtt_buf_pg_sz;
 	u32		mtt_hop_num;
+	u32		wqe_sq_hop_num;
+	u32		wqe_sge_hop_num;
+	u32		wqe_rq_hop_num;
 	u32		sccc_ba_pg_sz;
 	u32		sccc_buf_pg_sz;
 	u32		sccc_hop_num;

drivers/infiniband/hw/hns/hns_roce_hw_v2.c

@@ -1574,6 +1574,9 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev)
 	caps->mtt_ba_pg_sz	= 0;
 	caps->mtt_buf_pg_sz	= 0;
 	caps->mtt_hop_num	= HNS_ROCE_MTT_HOP_NUM;
+	caps->wqe_sq_hop_num	= 2;
+	caps->wqe_sge_hop_num	= 1;
+	caps->wqe_rq_hop_num	= 2;
 	caps->cqe_ba_pg_sz	= 0;
 	caps->cqe_buf_pg_sz	= 0;
 	caps->cqe_hop_num	= HNS_ROCE_CQE_HOP_NUM;
@@ -3021,7 +3024,6 @@ static int hns_roce_v2_clear_hem(struct hns_roce_dev *hr_dev,
 }
 
 static int hns_roce_v2_qp_modify(struct hns_roce_dev *hr_dev,
-				 struct hns_roce_mtt *mtt,
 				 enum ib_qp_state cur_state,
 				 enum ib_qp_state new_state,
 				 struct hns_roce_v2_qp_context *context,
@@ -3517,6 +3519,31 @@ static void modify_qp_init_to_init(struct ib_qp *ibqp,
 	}
 }
 
+static bool check_wqe_rq_mtt_count(struct hns_roce_dev *hr_dev,
+				   struct hns_roce_qp *hr_qp, int mtt_cnt,
+				   u32 page_size)
+{
+	struct device *dev = hr_dev->dev;
+
+	if (hr_qp->rq.wqe_cnt < 1)
+		return true;
+
+	if (mtt_cnt < 1) {
+		dev_err(dev, "qp(0x%lx) rqwqe buf ba find failed\n",
+			hr_qp->qpn);
+		return false;
+	}
+
+	if (mtt_cnt < MTT_MIN_COUNT &&
+		(hr_qp->rq.offset + page_size) < hr_qp->buff_size) {
+		dev_err(dev, "qp(0x%lx) next rqwqe buf ba find failed\n",
+			hr_qp->qpn);
+		return false;
+	}
+
+	return true;
+}
+
 static int modify_qp_init_to_rtr(struct ib_qp *ibqp,
 				 const struct ib_qp_attr *attr, int attr_mask,
 				 struct hns_roce_v2_qp_context *context,
@@ -3526,25 +3553,27 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp,
 	struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device);
 	struct hns_roce_qp *hr_qp = to_hr_qp(ibqp);
 	struct device *dev = hr_dev->dev;
+	u64 mtts[MTT_MIN_COUNT] = { 0 };
 	dma_addr_t dma_handle_3;
 	dma_addr_t dma_handle_2;
-	dma_addr_t dma_handle;
+	u64 wqe_sge_ba;
 	u32 page_size;
 	u8 port_num;
 	u64 *mtts_3;
 	u64 *mtts_2;
-	u64 *mtts;
+	int count;
 	u8 *dmac;
 	u8 *smac;
 	int port;
 
 	/* Search qp buf's mtts */
-	mtts = hns_roce_table_find(hr_dev, &hr_dev->mr_table.mtt_table,
-				   hr_qp->mtt.first_seg, &dma_handle);
-	if (!mtts) {
-		dev_err(dev, "qp buf pa find failed\n");
-		return -EINVAL;
-	}
+	page_size = 1 << (hr_dev->caps.mtt_buf_pg_sz + PAGE_SHIFT);
+	count = hns_roce_mtr_find(hr_dev, &hr_qp->mtr,
+				  hr_qp->rq.offset / page_size, mtts,
+				  MTT_MIN_COUNT, &wqe_sge_ba);
+	if (!ibqp->srq)
+		if (!check_wqe_rq_mtt_count(hr_dev, hr_qp, count, page_size))
+			return -EINVAL;
 
 	/* Search IRRL's mtts */
 	mtts_2 = hns_roce_table_find(hr_dev, &hr_dev->qp_table.irrl_table,
@@ -3568,7 +3597,7 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp,
 	}
 
 	dmac = (u8 *)attr->ah_attr.roce.dmac;
-	context->wqe_sge_ba = (u32)(dma_handle >> 3);
+	context->wqe_sge_ba = (u32)(wqe_sge_ba >> 3);
 	qpc_mask->wqe_sge_ba = 0;
 
 	/*
@@ -3578,39 +3607,40 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp,
 	 * 0 at the same time, else set them to 0x1.
 	 */
 	roce_set_field(context->byte_12_sq_hop, V2_QPC_BYTE_12_WQE_SGE_BA_M,
-		       V2_QPC_BYTE_12_WQE_SGE_BA_S, dma_handle >> (32 + 3));
+		       V2_QPC_BYTE_12_WQE_SGE_BA_S, wqe_sge_ba >> (32 + 3));
 	roce_set_field(qpc_mask->byte_12_sq_hop, V2_QPC_BYTE_12_WQE_SGE_BA_M,
 		       V2_QPC_BYTE_12_WQE_SGE_BA_S, 0);
 
 	roce_set_field(context->byte_12_sq_hop, V2_QPC_BYTE_12_SQ_HOP_NUM_M,
 		       V2_QPC_BYTE_12_SQ_HOP_NUM_S,
-		       hr_dev->caps.mtt_hop_num == HNS_ROCE_HOP_NUM_0 ?
-		       0 : hr_dev->caps.mtt_hop_num);
+		       hr_dev->caps.wqe_sq_hop_num == HNS_ROCE_HOP_NUM_0 ?
+		       0 : hr_dev->caps.wqe_sq_hop_num);
 	roce_set_field(qpc_mask->byte_12_sq_hop, V2_QPC_BYTE_12_SQ_HOP_NUM_M,
 		       V2_QPC_BYTE_12_SQ_HOP_NUM_S, 0);
 
 	roce_set_field(context->byte_20_smac_sgid_idx,
 		       V2_QPC_BYTE_20_SGE_HOP_NUM_M,
 		       V2_QPC_BYTE_20_SGE_HOP_NUM_S,
-		       ((ibqp->qp_type == IB_QPT_GSI) || hr_qp->sq.max_gs > 2) ?
-		       hr_dev->caps.mtt_hop_num : 0);
+		       ((ibqp->qp_type == IB_QPT_GSI) ||
+		       hr_qp->sq.max_gs > HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE) ?
+		       hr_dev->caps.wqe_sge_hop_num : 0);
 	roce_set_field(qpc_mask->byte_20_smac_sgid_idx,
 		       V2_QPC_BYTE_20_SGE_HOP_NUM_M,
 		       V2_QPC_BYTE_20_SGE_HOP_NUM_S, 0);
 
 	roce_set_field(context->byte_20_smac_sgid_idx,
 		       V2_QPC_BYTE_20_RQ_HOP_NUM_M,
 		       V2_QPC_BYTE_20_RQ_HOP_NUM_S,
-		       hr_dev->caps.mtt_hop_num == HNS_ROCE_HOP_NUM_0 ?
-		       0 : hr_dev->caps.mtt_hop_num);
+		       hr_dev->caps.wqe_rq_hop_num == HNS_ROCE_HOP_NUM_0 ?
+		       0 : hr_dev->caps.wqe_rq_hop_num);
 	roce_set_field(qpc_mask->byte_20_smac_sgid_idx,
 		       V2_QPC_BYTE_20_RQ_HOP_NUM_M,
 		       V2_QPC_BYTE_20_RQ_HOP_NUM_S, 0);
 
 	roce_set_field(context->byte_16_buf_ba_pg_sz,
 		       V2_QPC_BYTE_16_WQE_SGE_BA_PG_SZ_M,
 		       V2_QPC_BYTE_16_WQE_SGE_BA_PG_SZ_S,
-		       hr_dev->caps.mtt_ba_pg_sz + PG_SHIFT_OFFSET);
+		       hr_qp->wqe_bt_pg_shift + PG_SHIFT_OFFSET);
 	roce_set_field(qpc_mask->byte_16_buf_ba_pg_sz,
 		       V2_QPC_BYTE_16_WQE_SGE_BA_PG_SZ_M,
 		       V2_QPC_BYTE_16_WQE_SGE_BA_PG_SZ_S, 0);
@@ -3623,29 +3653,24 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp,
 		       V2_QPC_BYTE_16_WQE_SGE_BUF_PG_SZ_M,
 		       V2_QPC_BYTE_16_WQE_SGE_BUF_PG_SZ_S, 0);
 
-	page_size = 1 << (hr_dev->caps.mtt_buf_pg_sz + PAGE_SHIFT);
-	context->rq_cur_blk_addr = (u32)(mtts[hr_qp->rq.offset / page_size]
-				    >> PAGE_ADDR_SHIFT);
+	context->rq_cur_blk_addr = (u32)(mtts[0] >> PAGE_ADDR_SHIFT);
 	qpc_mask->rq_cur_blk_addr = 0;
 
 	roce_set_field(context->byte_92_srq_info,
 		       V2_QPC_BYTE_92_RQ_CUR_BLK_ADDR_M,
 		       V2_QPC_BYTE_92_RQ_CUR_BLK_ADDR_S,
-		       mtts[hr_qp->rq.offset / page_size]
-		       >> (32 + PAGE_ADDR_SHIFT));
+		       mtts[0] >> (32 + PAGE_ADDR_SHIFT));
 	roce_set_field(qpc_mask->byte_92_srq_info,
 		       V2_QPC_BYTE_92_RQ_CUR_BLK_ADDR_M,
 		       V2_QPC_BYTE_92_RQ_CUR_BLK_ADDR_S, 0);
 
-	context->rq_nxt_blk_addr = (u32)(mtts[hr_qp->rq.offset / page_size + 1]
-				    >> PAGE_ADDR_SHIFT);
+	context->rq_nxt_blk_addr = (u32)(mtts[1] >> PAGE_ADDR_SHIFT);
 	qpc_mask->rq_nxt_blk_addr = 0;
 
 	roce_set_field(context->byte_104_rq_sge,
 		       V2_QPC_BYTE_104_RQ_NXT_BLK_ADDR_M,
 		       V2_QPC_BYTE_104_RQ_NXT_BLK_ADDR_S,
-		       mtts[hr_qp->rq.offset / page_size + 1]
-		       >> (32 + PAGE_ADDR_SHIFT));
+		       mtts[1] >> (32 + PAGE_ADDR_SHIFT));
 	roce_set_field(qpc_mask->byte_104_rq_sge,
 		       V2_QPC_BYTE_104_RQ_NXT_BLK_ADDR_M,
 		       V2_QPC_BYTE_104_RQ_NXT_BLK_ADDR_S, 0);
@@ -3773,18 +3798,30 @@ static int modify_qp_rtr_to_rts(struct ib_qp *ibqp,
 	struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device);
 	struct hns_roce_qp *hr_qp = to_hr_qp(ibqp);
 	struct device *dev = hr_dev->dev;
-	dma_addr_t dma_handle;
+	u64 sge_cur_blk = 0;
+	u64 sq_cur_blk = 0;
 	u32 page_size;
-	u64 *mtts;
+	int count;
 
 	/* Search qp buf's mtts */
-	mtts = hns_roce_table_find(hr_dev, &hr_dev->mr_table.mtt_table,
-				   hr_qp->mtt.first_seg, &dma_handle);
-	if (!mtts) {
-		dev_err(dev, "qp buf pa find failed\n");
+	count = hns_roce_mtr_find(hr_dev, &hr_qp->mtr, 0, &sq_cur_blk, 1, NULL);
+	if (count < 1) {
+		dev_err(dev, "qp(0x%lx) buf pa find failed\n", hr_qp->qpn);
 		return -EINVAL;
 	}
 
+	if (hr_qp->sge.offset) {
+		page_size = 1 << (hr_dev->caps.mtt_buf_pg_sz + PAGE_SHIFT);
+		count = hns_roce_mtr_find(hr_dev, &hr_qp->mtr,
+					  hr_qp->sge.offset / page_size,
+					  &sge_cur_blk, 1, NULL);
+		if (count < 1) {
+			dev_err(dev, "qp(0x%lx) sge pa find failed\n",
+				hr_qp->qpn);
+			return -EINVAL;
+		}
+	}
+
 	/* Not support alternate path and path migration */
 	if ((attr_mask & IB_QP_ALT_PATH) ||
 	    (attr_mask & IB_QP_PATH_MIG_STATE)) {
@@ -3798,38 +3835,37 @@ static int modify_qp_rtr_to_rts(struct ib_qp *ibqp,
 	 * we should set all bits of the relevant fields in context mask to
 	 * 0 at the same time, else set them to 0x1.
 	 */
-	context->sq_cur_blk_addr = (u32)(mtts[0] >> PAGE_ADDR_SHIFT);
+	context->sq_cur_blk_addr = (u32)(sq_cur_blk >> PAGE_ADDR_SHIFT);
 	roce_set_field(context->byte_168_irrl_idx,
 		       V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_M,
 		       V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_S,
-		       mtts[0] >> (32 + PAGE_ADDR_SHIFT));
+		       sq_cur_blk >> (32 + PAGE_ADDR_SHIFT));
 	qpc_mask->sq_cur_blk_addr = 0;
 	roce_set_field(qpc_mask->byte_168_irrl_idx,
 		       V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_M,
 		       V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_S, 0);
 
-	page_size = 1 << (hr_dev->caps.mtt_buf_pg_sz + PAGE_SHIFT);
 	context->sq_cur_sge_blk_addr = ((ibqp->qp_type == IB_QPT_GSI) ||
 		       hr_qp->sq.max_gs > HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE) ?
-		       ((u32)(mtts[hr_qp->sge.offset / page_size] >>
+		       ((u32)(sge_cur_blk >>
 		       PAGE_ADDR_SHIFT)) : 0;
 	roce_set_field(context->byte_184_irrl_idx,
 		       V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_M,
 		       V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_S,
 		       ((ibqp->qp_type == IB_QPT_GSI) || hr_qp->sq.max_gs >
 		       HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE) ?
-		       (mtts[hr_qp->sge.offset / page_size] >>
+		       (sge_cur_blk >>
 		       (32 + PAGE_ADDR_SHIFT)) : 0);
 	qpc_mask->sq_cur_sge_blk_addr = 0;
 	roce_set_field(qpc_mask->byte_184_irrl_idx,
 		       V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_M,
 		       V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_S, 0);
 
-	context->rx_sq_cur_blk_addr = (u32)(mtts[0] >> PAGE_ADDR_SHIFT);
+	context->rx_sq_cur_blk_addr = (u32)(sq_cur_blk >> PAGE_ADDR_SHIFT);
 	roce_set_field(context->byte_232_irrl_sge,
 		       V2_QPC_BYTE_232_RX_SQ_CUR_BLK_ADDR_M,
 		       V2_QPC_BYTE_232_RX_SQ_CUR_BLK_ADDR_S,
-		       mtts[0] >> (32 + PAGE_ADDR_SHIFT));
+		       sq_cur_blk >> (32 + PAGE_ADDR_SHIFT));
 	qpc_mask->rx_sq_cur_blk_addr = 0;
 	roce_set_field(qpc_mask->byte_232_irrl_sge,
 		       V2_QPC_BYTE_232_RX_SQ_CUR_BLK_ADDR_M,
@@ -4230,7 +4266,7 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp,
 		       V2_QPC_BYTE_60_QP_ST_S, 0);
 
 	/* SW pass context to HW */
-	ret = hns_roce_v2_qp_modify(hr_dev, &hr_qp->mtt, cur_state, new_state,
+	ret = hns_roce_v2_qp_modify(hr_dev, cur_state, new_state,
 				    context, hr_qp);
 	if (ret) {
 		dev_err(dev, "hns_roce_qp_modify failed(%d)\n", ret);