Add new register and unregister functions to be used with devices that
support FRMRs, provide a local dma lkey, yet do not support DIF/PI.
isert_reg_frmr() only needs to use FRMRs for RDMA READ since RDMA WRITE
can be handled entirely with the local dma lkey. So for RDMA READ,
it calls isert_reg_read_frmr(). Otherwise is uses the lkey map service
isert_map_lkey() for RDMA WRITEs.
isert_reg_read_frmr() will create a linked list of WR triplets of the
form: INV->FRWR->READ. The number of these triplets is dependent on
the devices fast reg page list length limit.
Signed-off-by: Steve Wise <swise@xxxxxxxxxxxxxxxxxxxxx>
---
drivers/infiniband/ulp/isert/ib_isert.c | 224 ++++++++++++++++++++++++++++++-
1 files changed, 218 insertions(+), 6 deletions(-)
diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c
index 47bb790..34b3151 100644
--- a/drivers/infiniband/ulp/isert/ib_isert.c
+++ b/drivers/infiniband/ulp/isert/ib_isert.c
@@ -57,6 +57,11 @@ isert_unreg_frmr_pi(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn);
static int
isert_reg_frmr_pi(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
struct isert_rdma_wr *wr);
+static void
+isert_unreg_frmr(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn);
+static int
+isert_reg_frmr(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
+ struct isert_rdma_wr *wr);
static int
isert_put_response(struct iscsi_conn *conn, struct iscsi_cmd *cmd);
static int
@@ -368,6 +373,7 @@ static int
isert_create_device_ib_res(struct isert_device *device)
{
struct ib_device_attr *dev_attr;
+ int cap_flags;
int ret;
dev_attr = &device->dev_attr;
@@ -376,12 +382,18 @@ isert_create_device_ib_res(struct isert_device *device)
return ret;
/* assign function handlers */
- if (dev_attr->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS &&
- dev_attr->device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY &&
- dev_attr->device_cap_flags & IB_DEVICE_SIGNATURE_HANDOVER) {
- device->use_fastreg = 1;
- device->reg_rdma_mem = isert_reg_frmr_pi;
- device->unreg_rdma_mem = isert_unreg_frmr_pi;
+ cap_flags = dev_attr->device_cap_flags;
+ if (cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS &&
+ cap_flags & IB_DEVICE_LOCAL_DMA_LKEY) {
+ if (cap_flags & IB_DEVICE_SIGNATURE_HANDOVER) {
+ device->use_fastreg = 1;
+ device->reg_rdma_mem = isert_reg_frmr_pi;
+ device->unreg_rdma_mem = isert_unreg_frmr_pi;
+ } else {
+ device->use_fastreg = 1;
+ device->reg_rdma_mem = isert_reg_frmr;
+ device->unreg_rdma_mem = isert_unreg_frmr;
+ }
} else {
device->use_fastreg = 0;
device->reg_rdma_mem = isert_map_lkey;
@@ -1782,6 +1794,50 @@ isert_unreg_frmr_pi(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn)
}
static void
+isert_unreg_frmr(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn)
+{
+ struct isert_rdma_wr *wr = &isert_cmd->rdma_wr;
+
+ if (wr->data.sg) {
+ isert_dbg("Cmd %p unmap_sg op\n", isert_cmd);
+ isert_unmap_data_buf(isert_conn, &wr->data);
+ }
+
+ if (wr->send_wr) {
+ int i;
+
+ isert_dbg("Cmd %p free send_wr wr_num %d\n", isert_cmd,
+ wr->send_wr_num);
+ if (wr->iser_ib_op == ISER_IB_RDMA_READ) {
+ spin_lock_bh(&isert_conn->pool_lock);
+ for (i = 0; i < wr->send_wr_num; i += 3) {
+ struct ib_send_wr *fr_wr;
+ struct fast_reg_descriptor *fr_desc;
+
+ fr_wr = &wr->send_wr[i + 1];
+ fr_desc = (struct fast_reg_descriptor *)
+ (uintptr_t) fr_wr->wr_id;
+ isert_dbg("Cmd %p free fr_desc %p\n", isert_cmd,
+ fr_desc);
+ list_add_tail(&fr_desc->list,
+ &isert_conn->fr_pool);
+ }
+ spin_unlock_bh(&isert_conn->pool_lock);
+ }
+ isert_dbg("Cmd %p free wr->send_wr\n", isert_cmd);
+ kfree(wr->send_wr);
+ wr->send_wr = NULL;
+ wr->send_wr_num = 0;
+ }
+
+ if (wr->ib_sge) {
+ isert_dbg("Cmd %p free ib_sge\n", isert_cmd);
+ kfree(wr->ib_sge);
+ wr->ib_sge = NULL;
+ }
+}
+
+static void
isert_put_cmd(struct isert_cmd *isert_cmd, bool comp_err)
{
struct iscsi_cmd *cmd = isert_cmd->iscsi_cmd;
@@ -2958,6 +3014,162 @@ unmap_cmd:
return ret;
}
+static void
+isert_build_inv_fr_wrs(struct isert_conn *isert_conn,
+ struct isert_cmd *isert_cmd, struct ib_sge *ib_sge,
+ struct ib_send_wr *inv_wr, u32 data_len, u32 offset)
+{
+ struct iscsi_cmd *cmd = isert_cmd->iscsi_cmd;
+ struct scatterlist *sg_start;
+ struct isert_device *device = isert_conn->device;
+ struct ib_device *ib_dev = device->ib_device;
+ struct ib_fast_reg_page_list *frpl;
+ struct fast_reg_descriptor *fr_desc;
+ struct ib_send_wr *fr_wr;
+ u32 sg_off, page_off;
+ int sg_nents;
+ int frpl_len;
+ unsigned long flags;
+
+ sg_off = offset / PAGE_SIZE;
+ sg_start = &cmd->se_cmd.t_data_sg[sg_off];
+ sg_nents = min(cmd->se_cmd.t_data_nents - sg_off,
+ isert_conn->max_frpl_len);
+ page_off = offset % PAGE_SIZE;
+
+ spin_lock_irqsave(&isert_conn->pool_lock, flags);
+ fr_desc = list_first_entry(&isert_conn->fr_pool,
+ struct fast_reg_descriptor, list);
+ list_del(&fr_desc->list);
+ spin_unlock_irqrestore(&isert_conn->pool_lock, flags);
+ fr_desc->ind &= ~ISERT_DATA_KEY_VALID;
+
+ /* Build the INV WR */
+ isert_inv_rkey(inv_wr, fr_desc->data_mr);
+
+ /* Build the FR WR */
+ frpl = fr_desc->data_frpl;
+ frpl_len = isert_map_fr_pagelist(ib_dev, sg_start, sg_nents,
+ &frpl->page_list[0]);
+ fr_wr = inv_wr + 1;
+ memset(fr_wr, 0, sizeof(fr_wr));
+ fr_wr->wr_id = (uintptr_t)fr_desc;
+ fr_wr->opcode = IB_WR_FAST_REG_MR;
+ fr_wr->wr.fast_reg.iova_start = frpl->page_list[0] + page_off;
+ fr_wr->wr.fast_reg.page_list = frpl;
+ fr_wr->wr.fast_reg.page_list_len = frpl_len;
+ fr_wr->wr.fast_reg.page_shift = PAGE_SHIFT;
+ fr_wr->wr.fast_reg.length = data_len;
+ fr_wr->wr.fast_reg.rkey = fr_desc->data_mr->rkey;
+ fr_wr->wr.fast_reg.access_flags = IB_ACCESS_REMOTE_WRITE |
+ IB_ACCESS_LOCAL_WRITE;
+
+ ib_sge->addr = frpl->page_list[0] + page_off;
+ ib_sge->length = data_len;
+ ib_sge->lkey = fr_desc->data_mr->rkey;
+
+ /* Link up the wrs: inv->fr->read */
+ inv_wr->next = fr_wr;
+ fr_wr->next = fr_wr + 1;
+}
+
+static int
+isert_reg_read_frmr(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
+ struct isert_rdma_wr *wr)
+{
+ struct se_cmd *se_cmd = &cmd->se_cmd;
+ struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd);
+ struct isert_conn *isert_conn = conn->context;
+ struct isert_data_buf *data = &wr->data;
+ struct ib_send_wr *send_wr = NULL;
+ struct ib_sge *ib_sge;
+ u32 offset, data_len, data_left, rdma_max_len, va_offset = 0;
+ int ret = 0, i;
+ u32 num_frwrs;
+
+ isert_cmd->tx_desc.isert_cmd = isert_cmd;
+
+ offset = cmd->write_data_done;
+ ret = isert_map_data_buf(isert_conn, isert_cmd, se_cmd->t_data_sg,
+ se_cmd->t_data_nents, se_cmd->data_length,
+ offset, wr->iser_ib_op, &wr->data);
+ if (ret)
+ return ret;
+
+ data_left = data->len;
+ offset = data->offset;
+ num_frwrs = DIV_ROUND_UP(data->nents, isert_conn->max_frpl_len);
+
+ ib_sge = kcalloc(num_frwrs, sizeof(struct ib_sge), GFP_KERNEL);
+ if (!ib_sge) {
+ ret = -ENOMEM;
+ goto unmap_cmd;
+ }
+ wr->ib_sge = ib_sge;
+
+ /* always do INV + FR + READ for now */
+ wr->send_wr_num = 3 * num_frwrs;
+ wr->send_wr = kcalloc(wr->send_wr_num, sizeof(struct ib_send_wr),
+ GFP_KERNEL);
+ if (!wr->send_wr) {
+ ret = -ENOMEM;
+ goto unmap_cmd;
+ }
+ isert_info("num_frwrs %d send_wr_num %d\n", num_frwrs, wr->send_wr_num);
+
+ wr->isert_cmd = isert_cmd;
+ rdma_max_len = isert_conn->max_frpl_len * PAGE_SIZE;
+
+ for (i = 0; i < wr->send_wr_num; i += 3) {
+
+ /*
+ * i = INV, i+1 = FR, i+2 = READ
+ */
+ send_wr = &isert_cmd->rdma_wr.send_wr[i+2];
+ data_len = min(data_left, rdma_max_len);
+ isert_build_inv_fr_wrs(isert_conn, isert_cmd, ib_sge,
+ send_wr - 2, data_len, offset);
+ send_wr->opcode = IB_WR_RDMA_READ;
+ send_wr->wr.rdma.remote_addr = isert_cmd->write_va + va_offset;
+ send_wr->wr.rdma.rkey = isert_cmd->write_stag;
+ send_wr->sg_list = ib_sge;
+ send_wr->num_sge = 1;
+ if (i + 3 == wr->send_wr_num) {
+ send_wr->wr_id = (uintptr_t)&isert_cmd->tx_desc;
+ send_wr->send_flags = IB_SEND_SIGNALED;
+ } else {
+ send_wr->next = send_wr + 1;
+ }
+
+ ib_sge++;
+ offset += data_len;
+ va_offset += data_len;
+ data_left -= data_len;
+ }
+ isert_info("send_wr->send_flags 0x%x\n", send_wr->send_flags);
+
+ return 0;
+unmap_cmd:
+ isert_unmap_data_buf(isert_conn, data);
+
+ return ret;
+}
+
+static int
+isert_reg_frmr(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
+ struct isert_rdma_wr *wr)
+{
+
+ /*
+ * Use the local dma lkey for Writes since it only requires local
+ * access flags. For reads, build up inv+fr+read
+ * wr chains as needed.
+ */
+ if (wr->iser_ib_op == ISER_IB_RDMA_WRITE)
+ return isert_map_lkey(conn, cmd, wr);
+ return isert_reg_read_frmr(conn, cmd, wr);
+}
+
static int
isert_put_datain(struct iscsi_conn *conn, struct iscsi_cmd *cmd)
{
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
