/**
* iser_send_command - send command PDU
*/
int iser_send_command(struct iscsi_conn *conn,
struct iscsi_task *task)
{
struct iscsi_iser_conn *iser_conn = conn->dd_data;
struct iscsi_iser_task *iser_task = task->dd_data;
struct iser_dto *send_dto = NULL;
unsigned long edtl;
int err = 0;
struct iser_data_buf *data_buf;
struct iscsi_cmd *hdr = (struct iscsi_cmd *)task->hdr;
struct scsi_cmnd *sc = task->sc;
if (!iser_conn_state_comp(iser_conn->ib_conn, ISER_CONN_UP)) {
iser_err("Failed to send, conn: 0x%p is not up\n", iser_conn->ib_conn);
return -EPERM;
}
if (iser_check_xmit(conn, task))
return -ENOBUFS;
edtl = ntohl(hdr->data_length);
/* build the tx desc regd header and add it to the tx desc dto */
iser_task->desc.type = ISCSI_TX_SCSI_COMMAND;
send_dto = &iser_task->desc.dto;
send_dto->task = iser_task;
iser_create_send_desc(iser_conn, &iser_task->desc);
if (hdr->flags & ISCSI_FLAG_CMD_READ)
data_buf = &iser_task->data[ISER_DIR_IN];
else
data_buf = &iser_task->data[ISER_DIR_OUT];
if (scsi_sg_count(sc)) { /* using a scatter list */
data_buf->buf = scsi_sglist(sc);
data_buf->size = scsi_sg_count(sc);
}
data_buf->data_len = scsi_bufflen(sc);
if (hdr->flags & ISCSI_FLAG_CMD_READ) {
err = iser_prepare_read_cmd(task, edtl);
if (err)
goto send_command_error;
}
if (hdr->flags & ISCSI_FLAG_CMD_WRITE) {
err = iser_prepare_write_cmd(task,
task->imm_count,
task->imm_count +
task->unsol_r2t.data_length,
edtl);
if (err)
goto send_command_error;
}
iser_reg_single(iser_conn->ib_conn->device,
send_dto->regd[0], DMA_TO_DEVICE);
if (iser_post_receive_control(conn) != 0) {
iser_err("post_recv failed!\n");
err = -ENOMEM;
goto send_command_error;
}
iser_task->status = ISER_TASK_STATUS_STARTED;
err = iser_post_send(&iser_task->desc);
if (!err)
return 0;
send_command_error:
iser_dto_buffs_release(send_dto);
iser_err("conn %p failed task->itt %d err %d\n",conn, task->itt, err);
return err;
}
/**
* iser_send_data_out - send data out PDU
*/
int iser_send_data_out(struct iscsi_conn *conn,
struct iscsi_task *task,
struct iscsi_data *hdr)
{
struct iscsi_iser_conn *iser_conn = conn->dd_data;
struct iscsi_iser_task *iser_task = task->dd_data;
struct iser_desc *tx_desc = NULL;
struct iser_dto *send_dto = NULL;
unsigned long buf_offset;
unsigned long data_seg_len;
uint32_t itt;
int err = 0;
if (!iser_conn_state_comp(iser_conn->ib_conn, ISER_CONN_UP)) {
iser_err("Failed to send, conn: 0x%p is not up\n", iser_conn->ib_conn);
return -EPERM;
}
if (iser_check_xmit(conn, task))
return -ENOBUFS;
itt = (__force uint32_t)hdr->itt;
data_seg_len = ntoh24(hdr->dlength);
buf_offset = ntohl(hdr->offset);
iser_dbg("%s itt %d dseg_len %d offset %d\n",
__func__,(int)itt,(int)data_seg_len,(int)buf_offset);
tx_desc = kmem_cache_alloc(ig.desc_cache, GFP_NOIO);
if (tx_desc == NULL) {
iser_err("Failed to alloc desc for post dataout\n");
return -ENOMEM;
}
tx_desc->type = ISCSI_TX_DATAOUT;
memcpy(&tx_desc->iscsi_header, hdr, sizeof(struct iscsi_hdr));
/* build the tx desc regd header and add it to the tx desc dto */
send_dto = &tx_desc->dto;
send_dto->task = iser_task;
iser_create_send_desc(iser_conn, tx_desc);
iser_reg_single(iser_conn->ib_conn->device,
send_dto->regd[0], DMA_TO_DEVICE);
/* all data was registered for RDMA, we can use the lkey */
iser_dto_add_regd_buff(send_dto,
&iser_task->rdma_regd[ISER_DIR_OUT],
buf_offset,
data_seg_len);
if (buf_offset + data_seg_len > iser_task->data[ISER_DIR_OUT].data_len) {
iser_err("Offset:%ld & DSL:%ld in Data-Out "
"inconsistent with total len:%ld, itt:%d\n",
buf_offset, data_seg_len,
iser_task->data[ISER_DIR_OUT].data_len, itt);
err = -EINVAL;
goto send_data_out_error;
}
iser_dbg("data-out itt: %d, offset: %ld, sz: %ld\n",
itt, buf_offset, data_seg_len);
err = iser_post_send(tx_desc);
if (!err)
return 0;
send_data_out_error:
iser_dto_buffs_release(send_dto);
kmem_cache_free(ig.desc_cache, tx_desc);
iser_err("conn %p failed err %d\n",conn, err);
return err;
}
static void iser_qp_event_callback(struct ib_event *cause, void *context)
{
iser_err("got qp event %d\n",cause->event);
}
/**
* iser_post_receive_control - allocates, initializes and posts receive DTO.
*/
static int iser_post_receive_control(struct iscsi_conn *conn)
{
struct iscsi_iser_conn *iser_conn = conn->dd_data;
struct iser_desc *rx_desc;
struct iser_regd_buf *regd_hdr;
struct iser_regd_buf *regd_data;
struct iser_dto *recv_dto = NULL;
struct iser_device *device = iser_conn->ib_conn->device;
int rx_data_size, err;
int posts, outstanding_unexp_pdus;
/* for the login sequence we must support rx of upto 8K; login is done
* after conn create/bind (connect) and conn stop/bind (reconnect),
* what's common for both schemes is that the connection is not started
*/
if (conn->c_stage != ISCSI_CONN_STARTED)
rx_data_size = ISCSI_DEF_MAX_RECV_SEG_LEN;
else /* FIXME till user space sets conn->max_recv_dlength correctly */
rx_data_size = 128;
outstanding_unexp_pdus =
atomic_xchg(&iser_conn->ib_conn->unexpected_pdu_count, 0);
/*
* in addition to the response buffer, replace those consumed by
* unexpected pdus.
*/
for (posts = 0; posts < 1 + outstanding_unexp_pdus; posts++) {
rx_desc = kmem_cache_alloc(ig.desc_cache, GFP_NOIO);
if (rx_desc == NULL) {
iser_err("Failed to alloc desc for post recv %d\n",
posts);
err = -ENOMEM;
goto post_rx_cache_alloc_failure;
}
rx_desc->type = ISCSI_RX;
rx_desc->data = kmalloc(rx_data_size, GFP_NOIO);
if (rx_desc->data == NULL) {
iser_err("Failed to alloc data buf for post recv %d\n",
posts);
err = -ENOMEM;
goto post_rx_kmalloc_failure;
}
recv_dto = &rx_desc->dto;
recv_dto->ib_conn = iser_conn->ib_conn;
recv_dto->regd_vector_len = 0;
regd_hdr = &rx_desc->hdr_regd_buf;
memset(regd_hdr, 0, sizeof(struct iser_regd_buf));
regd_hdr->device = device;
regd_hdr->virt_addr = rx_desc; /* == &rx_desc->iser_header */
regd_hdr->data_size = ISER_TOTAL_HEADERS_LEN;
iser_reg_single(device, regd_hdr, DMA_FROM_DEVICE);
iser_dto_add_regd_buff(recv_dto, regd_hdr, 0, 0);
regd_data = &rx_desc->data_regd_buf;
memset(regd_data, 0, sizeof(struct iser_regd_buf));
regd_data->device = device;
regd_data->virt_addr = rx_desc->data;
regd_data->data_size = rx_data_size;
iser_reg_single(device, regd_data, DMA_FROM_DEVICE);
iser_dto_add_regd_buff(recv_dto, regd_data, 0, 0);
err = iser_post_recv(rx_desc);
if (err) {
iser_err("Failed iser_post_recv for post %d\n", posts);
goto post_rx_post_recv_failure;
}
}
/* all posts successful */
return 0;
post_rx_post_recv_failure:
iser_dto_buffs_release(recv_dto);
kfree(rx_desc->data);
post_rx_kmalloc_failure:
kmem_cache_free(ig.desc_cache, rx_desc);
post_rx_cache_alloc_failure:
if (posts > 0) {
/*
* response buffer posted, but did not replace all unexpected
* pdu recv bufs. Ignore error, retry occurs next send
*/
outstanding_unexp_pdus -= (posts - 1);
err = 0;
}
atomic_add(outstanding_unexp_pdus,
&iser_conn->ib_conn->unexpected_pdu_count);
return err;
}
/**
* iscsi_iser_session_create() - create an iscsi-iser session
* @ep: iscsi end-point handle
* @cmds_max: maximum commands in this session
* @qdepth: session command queue depth
* @initial_cmdsn: initiator command sequnce number
*
* Allocates and adds a scsi host, expose DIF supprot if
* exists, and sets up an iscsi session.
*/
static struct iscsi_cls_session *
iscsi_iser_session_create(struct iscsi_endpoint *ep,
uint16_t cmds_max, uint16_t qdepth,
uint32_t initial_cmdsn)
{
struct iscsi_cls_session *cls_session;
struct iscsi_session *session;
struct Scsi_Host *shost;
struct iser_conn *iser_conn = NULL;
struct ib_conn *ib_conn;
u16 max_cmds;
shost = iscsi_host_alloc(&iscsi_iser_sht, 0, 0);
if (!shost)
return NULL;
shost->transportt = iscsi_iser_scsi_transport;
shost->cmd_per_lun = qdepth;
shost->max_lun = iscsi_max_lun;
shost->max_id = 0;
shost->max_channel = 0;
shost->max_cmd_len = 16;
/*
* older userspace tools (before 2.0-870) did not pass us
* the leading conn's ep so this will be NULL;
*/
if (ep) {
iser_conn = ep->dd_data;
max_cmds = iser_conn->max_cmds;
shost->sg_tablesize = iser_conn->scsi_sg_tablesize;
shost->max_sectors = iser_conn->scsi_max_sectors;
mutex_lock(&iser_conn->state_mutex);
if (iser_conn->state != ISER_CONN_UP) {
iser_err("iser conn %p already started teardown\n",
iser_conn);
mutex_unlock(&iser_conn->state_mutex);
goto free_host;
}
ib_conn = &iser_conn->ib_conn;
if (ib_conn->pi_support) {
u32 sig_caps = ib_conn->device->dev_attr.sig_prot_cap;
scsi_host_set_prot(shost, iser_dif_prot_caps(sig_caps));
scsi_host_set_guard(shost, SHOST_DIX_GUARD_IP |
SHOST_DIX_GUARD_CRC);
}
/*
* Limit the sg_tablesize and max_sectors based on the device
* max fastreg page list length.
*/
shost->sg_tablesize = min_t(unsigned short, shost->sg_tablesize,
ib_conn->device->dev_attr.max_fast_reg_page_list_len);
shost->max_sectors = min_t(unsigned int,
1024, (shost->sg_tablesize * PAGE_SIZE) >> 9);
if (iscsi_host_add(shost,
ib_conn->device->ib_device->dma_device)) {
mutex_unlock(&iser_conn->state_mutex);
goto free_host;
}
mutex_unlock(&iser_conn->state_mutex);
} else {
/**
* iser_create_ib_conn_res - Creates FMR pool and Queue-Pair (QP)
*
* returns 0 on success, -1 on failure
*/
static int iser_create_ib_conn_res(struct iser_conn *ib_conn)
{
struct iser_device *device;
struct ib_qp_init_attr init_attr;
int ret;
struct ib_fmr_pool_param params;
BUG_ON(ib_conn->device == NULL);
device = ib_conn->device;
ib_conn->page_vec = kmalloc(sizeof(struct iser_page_vec) +
(sizeof(u64) * (ISCSI_ISER_SG_TABLESIZE +1)),
GFP_KERNEL);
if (!ib_conn->page_vec) {
ret = -ENOMEM;
goto alloc_err;
}
ib_conn->page_vec->pages = (u64 *) (ib_conn->page_vec + 1);
params.page_shift = SHIFT_4K;
/* when the first/last SG element are not start/end *
* page aligned, the map whould be of N+1 pages */
params.max_pages_per_fmr = ISCSI_ISER_SG_TABLESIZE + 1;
/* make the pool size twice the max number of SCSI commands *
* the ML is expected to queue, watermark for unmap at 50% */
params.pool_size = ISCSI_DEF_XMIT_CMDS_MAX * 2;
params.dirty_watermark = ISCSI_DEF_XMIT_CMDS_MAX;
params.cache = 0;
params.flush_function = NULL;
params.access = (IB_ACCESS_LOCAL_WRITE |
IB_ACCESS_REMOTE_WRITE |
IB_ACCESS_REMOTE_READ);
ib_conn->fmr_pool = ib_create_fmr_pool(device->pd, ¶ms);
if (IS_ERR(ib_conn->fmr_pool)) {
ret = PTR_ERR(ib_conn->fmr_pool);
goto fmr_pool_err;
}
memset(&init_attr, 0, sizeof init_attr);
init_attr.event_handler = iser_qp_event_callback;
init_attr.qp_context = (void *)ib_conn;
init_attr.send_cq = device->cq;
init_attr.recv_cq = device->cq;
init_attr.cap.max_send_wr = ISER_QP_MAX_REQ_DTOS;
init_attr.cap.max_recv_wr = ISER_QP_MAX_RECV_DTOS;
init_attr.cap.max_send_sge = MAX_REGD_BUF_VECTOR_LEN;
init_attr.cap.max_recv_sge = 2;
init_attr.sq_sig_type = IB_SIGNAL_REQ_WR;
init_attr.qp_type = IB_QPT_RC;
ret = rdma_create_qp(ib_conn->cma_id, device->pd, &init_attr);
if (ret)
goto qp_err;
ib_conn->qp = ib_conn->cma_id->qp;
iser_err("setting conn %p cma_id %p: fmr_pool %p qp %p\n",
ib_conn, ib_conn->cma_id,
ib_conn->fmr_pool, ib_conn->cma_id->qp);
return ret;
qp_err:
(void)ib_destroy_fmr_pool(ib_conn->fmr_pool);
fmr_pool_err:
kfree(ib_conn->page_vec);
alloc_err:
iser_err("unable to alloc mem or create resource, err %d\n", ret);
return ret;
}
/**
* iser_reg_rdma_mem - Registers memory intended for RDMA,
* obtaining rkey and va
*
* returns 0 on success, errno code on failure
*/
int iser_reg_rdma_mem(struct iscsi_iser_cmd_task *iser_ctask,
enum iser_data_dir cmd_dir)
{
struct iser_conn *ib_conn = iser_ctask->iser_conn->ib_conn;
struct iser_device *device = ib_conn->device;
struct ib_device *ibdev = device->ib_device;
struct iser_data_buf *mem = &iser_ctask->data[cmd_dir];
struct iser_regd_buf *regd_buf;
int aligned_len;
int err;
int i;
struct scatterlist *sg;
regd_buf = &iser_ctask->rdma_regd[cmd_dir];
aligned_len = iser_data_buf_aligned_len(mem, ibdev);
if (aligned_len != mem->dma_nents) {
iser_err("rdma alignment violation %d/%d aligned\n",
aligned_len, mem->size);
iser_data_buf_dump(mem, ibdev);
/* unmap the command data before accessing it */
iser_dma_unmap_task_data(iser_ctask);
/* allocate copy buf, if we are writing, copy the */
/* unaligned scatterlist, dma map the copy */
if (iser_start_rdma_unaligned_sg(iser_ctask, cmd_dir) != 0)
return -ENOMEM;
mem = &iser_ctask->data_copy[cmd_dir];
}
/* if there a single dma entry, FMR is not needed */
if (mem->dma_nents == 1) {
sg = (struct scatterlist *)mem->buf;
regd_buf->reg.lkey = device->mr->lkey;
regd_buf->reg.rkey = device->mr->rkey;
regd_buf->reg.len = ib_sg_dma_len(ibdev, &sg[0]);
regd_buf->reg.va = ib_sg_dma_address(ibdev, &sg[0]);
regd_buf->reg.is_fmr = 0;
iser_dbg("PHYSICAL Mem.register: lkey: 0x%08X rkey: 0x%08X "
"va: 0x%08lX sz: %ld]\n",
(unsigned int)regd_buf->reg.lkey,
(unsigned int)regd_buf->reg.rkey,
(unsigned long)regd_buf->reg.va,
(unsigned long)regd_buf->reg.len);
} else { /* use FMR for multiple dma entries */
iser_page_vec_build(mem, ib_conn->page_vec, ibdev);
err = iser_reg_page_vec(ib_conn, ib_conn->page_vec, ®d_buf->reg);
if (err) {
iser_data_buf_dump(mem, ibdev);
iser_err("mem->dma_nents = %d (dlength = 0x%x)\n", mem->dma_nents,
ntoh24(iser_ctask->desc.iscsi_header.dlength));
iser_err("page_vec: data_size = 0x%x, length = %d, offset = 0x%x\n",
ib_conn->page_vec->data_size, ib_conn->page_vec->length,
ib_conn->page_vec->offset);
for (i=0 ; i<ib_conn->page_vec->length ; i++)
iser_err("page_vec[%d] = 0x%llx\n", i,
(unsigned long long) ib_conn->page_vec->pages[i]);
return err;
}
}
/* take a reference on this regd buf such that it will not be released *
* (eg in send dto completion) before we get the scsi response */
atomic_inc(®d_buf->ref_count);
return 0;
}
/**
* iser_post_receive_control - allocates, initializes and posts receive DTO.
*/
static int iser_post_receive_control(struct iscsi_conn *conn)
{
struct iscsi_iser_conn *iser_conn = conn->dd_data;
struct iser_desc *rx_desc;
struct iser_regd_buf *regd_hdr;
struct iser_regd_buf *regd_data;
struct iser_dto *recv_dto = NULL;
struct iser_device *device = iser_conn->ib_conn->device;
int rx_data_size, err = 0;
rx_desc = kmem_cache_alloc(ig.desc_cache, GFP_NOIO);
if (rx_desc == NULL) {
iser_err("Failed to alloc desc for post recv\n");
return -ENOMEM;
}
rx_desc->type = ISCSI_RX;
/* for the login sequence we must support rx of upto 8K; login is done
* after conn create/bind (connect) and conn stop/bind (reconnect),
* what's common for both schemes is that the connection is not started
*/
if (conn->c_stage != ISCSI_CONN_STARTED)
rx_data_size = ISCSI_DEF_MAX_RECV_SEG_LEN;
else /* FIXME till user space sets conn->max_recv_dlength correctly */
rx_data_size = 128;
rx_desc->data = kmalloc(rx_data_size, GFP_NOIO);
if (rx_desc->data == NULL) {
iser_err("Failed to alloc data buf for post recv\n");
err = -ENOMEM;
goto post_rx_kmalloc_failure;
}
recv_dto = &rx_desc->dto;
recv_dto->ib_conn = iser_conn->ib_conn;
recv_dto->regd_vector_len = 0;
regd_hdr = &rx_desc->hdr_regd_buf;
memset(regd_hdr, 0, sizeof(struct iser_regd_buf));
regd_hdr->device = device;
regd_hdr->virt_addr = rx_desc; /* == &rx_desc->iser_header */
regd_hdr->data_size = ISER_TOTAL_HEADERS_LEN;
iser_reg_single(device, regd_hdr, DMA_FROM_DEVICE);
iser_dto_add_regd_buff(recv_dto, regd_hdr, 0, 0);
regd_data = &rx_desc->data_regd_buf;
memset(regd_data, 0, sizeof(struct iser_regd_buf));
regd_data->device = device;
regd_data->virt_addr = rx_desc->data;
regd_data->data_size = rx_data_size;
iser_reg_single(device, regd_data, DMA_FROM_DEVICE);
iser_dto_add_regd_buff(recv_dto, regd_data, 0, 0);
err = iser_post_recv(rx_desc);
if (!err)
return 0;
/* iser_post_recv failed */
iser_dto_buffs_release(recv_dto);
kfree(rx_desc->data);
post_rx_kmalloc_failure:
kmem_cache_free(ig.desc_cache, rx_desc);
return err;
}
static void iser_event_handler(struct ib_event_handler *handler,
struct ib_event *event)
{
iser_err("async event %d on device %s port %d\n", event->event,
event->device->name, event->element.port_num);
}
/* Register user buffer memory and initialize passive rdma
* dto descriptor. Data size is stored in
* task->data[ISER_DIR_OUT].data_len, Protection size
* is stored at task->prot[ISER_DIR_OUT].data_len
*/
static int
iser_prepare_write_cmd(struct iscsi_task *task,
unsigned int imm_sz,
unsigned int unsol_sz,
unsigned int edtl)
{
struct iscsi_iser_task *iser_task = task->dd_data;
struct iser_device *device = iser_task->iser_conn->ib_conn.device;
struct iser_regd_buf *regd_buf;
int err;
struct iser_hdr *hdr = &iser_task->desc.iser_header;
struct iser_data_buf *buf_out = &iser_task->data[ISER_DIR_OUT];
struct ib_sge *tx_dsg = &iser_task->desc.tx_sg[1];
err = iser_dma_map_task_data(iser_task,
buf_out,
ISER_DIR_OUT,
DMA_TO_DEVICE);
if (err)
return err;
if (scsi_prot_sg_count(iser_task->sc)) {
struct iser_data_buf *pbuf_out = &iser_task->prot[ISER_DIR_OUT];
err = iser_dma_map_task_data(iser_task,
pbuf_out,
ISER_DIR_OUT,
DMA_TO_DEVICE);
if (err)
return err;
}
err = device->iser_reg_rdma_mem(iser_task, ISER_DIR_OUT);
if (err != 0) {
iser_err("Failed to register write cmd RDMA mem\n");
return err;
}
regd_buf = &iser_task->rdma_regd[ISER_DIR_OUT];
if (unsol_sz < edtl) {
hdr->flags |= ISER_WSV;
hdr->write_stag = cpu_to_be32(regd_buf->reg.rkey);
hdr->write_va = cpu_to_be64(regd_buf->reg.va + unsol_sz);
iser_dbg("Cmd itt:%d, WRITE tags, RKEY:%#.4X "
"VA:%#llX + unsol:%d\n",
task->itt, regd_buf->reg.rkey,
(unsigned long long)regd_buf->reg.va, unsol_sz);
}
if (imm_sz > 0) {
iser_dbg("Cmd itt:%d, WRITE, adding imm.data sz: %d\n",
task->itt, imm_sz);
tx_dsg->addr = regd_buf->reg.va;
tx_dsg->length = imm_sz;
tx_dsg->lkey = regd_buf->reg.lkey;
iser_task->desc.num_sge = 2;
}
return 0;
}
/**
* iser_send_data_out - send data out PDU
*/
int iser_send_data_out(struct iscsi_conn *conn,
struct iscsi_task *task,
struct iscsi_data *hdr)
{
struct iser_conn *iser_conn = conn->dd_data;
struct iscsi_iser_task *iser_task = task->dd_data;
struct iser_tx_desc *tx_desc = NULL;
struct iser_regd_buf *regd_buf;
unsigned long buf_offset;
unsigned long data_seg_len;
uint32_t itt;
int err = 0;
struct ib_sge *tx_dsg;
itt = (__force uint32_t)hdr->itt;
data_seg_len = ntoh24(hdr->dlength);
buf_offset = ntohl(hdr->offset);
iser_dbg("%s itt %d dseg_len %d offset %d\n",
__func__,(int)itt,(int)data_seg_len,(int)buf_offset);
tx_desc = kmem_cache_zalloc(ig.desc_cache, GFP_ATOMIC);
if (tx_desc == NULL) {
iser_err("Failed to alloc desc for post dataout\n");
return -ENOMEM;
}
tx_desc->type = ISCSI_TX_DATAOUT;
tx_desc->iser_header.flags = ISER_VER;
memcpy(&tx_desc->iscsi_header, hdr, sizeof(struct iscsi_hdr));
/* build the tx desc */
iser_initialize_task_headers(task, tx_desc);
regd_buf = &iser_task->rdma_regd[ISER_DIR_OUT];
tx_dsg = &tx_desc->tx_sg[1];
tx_dsg->addr = regd_buf->reg.va + buf_offset;
tx_dsg->length = data_seg_len;
tx_dsg->lkey = regd_buf->reg.lkey;
tx_desc->num_sge = 2;
if (buf_offset + data_seg_len > iser_task->data[ISER_DIR_OUT].data_len) {
iser_err("Offset:%ld & DSL:%ld in Data-Out "
"inconsistent with total len:%ld, itt:%d\n",
buf_offset, data_seg_len,
iser_task->data[ISER_DIR_OUT].data_len, itt);
err = -EINVAL;
goto send_data_out_error;
}
iser_dbg("data-out itt: %d, offset: %ld, sz: %ld\n",
itt, buf_offset, data_seg_len);
err = iser_post_send(&iser_conn->ib_conn, tx_desc, true);
if (!err)
return 0;
send_data_out_error:
kmem_cache_free(ig.desc_cache, tx_desc);
iser_err("conn %p failed err %d\n",conn, err);
return err;
}
/**
* iser_send_command - send command PDU
*/
int iser_send_command(struct iscsi_conn *conn,
struct iscsi_task *task)
{
struct iser_conn *iser_conn = conn->dd_data;
struct iscsi_iser_task *iser_task = task->dd_data;
unsigned long edtl;
int err;
struct iser_data_buf *data_buf, *prot_buf;
struct iscsi_scsi_req *hdr = (struct iscsi_scsi_req *)task->hdr;
struct scsi_cmnd *sc = task->sc;
struct iser_tx_desc *tx_desc = &iser_task->desc;
u8 sig_count = ++iser_conn->ib_conn.sig_count;
edtl = ntohl(hdr->data_length);
/* build the tx desc regd header and add it to the tx desc dto */
tx_desc->type = ISCSI_TX_SCSI_COMMAND;
iser_create_send_desc(iser_conn, tx_desc);
if (hdr->flags & ISCSI_FLAG_CMD_READ) {
data_buf = &iser_task->data[ISER_DIR_IN];
prot_buf = &iser_task->prot[ISER_DIR_IN];
} else {
data_buf = &iser_task->data[ISER_DIR_OUT];
prot_buf = &iser_task->prot[ISER_DIR_OUT];
}
if (scsi_sg_count(sc)) { /* using a scatter list */
data_buf->buf = scsi_sglist(sc);
data_buf->size = scsi_sg_count(sc);
}
data_buf->data_len = scsi_bufflen(sc);
if (scsi_prot_sg_count(sc)) {
prot_buf->buf = scsi_prot_sglist(sc);
prot_buf->size = scsi_prot_sg_count(sc);
prot_buf->data_len = (data_buf->data_len >>
ilog2(sc->device->sector_size)) * 8;
}
if (hdr->flags & ISCSI_FLAG_CMD_READ) {
err = iser_prepare_read_cmd(task);
if (err)
goto send_command_error;
}
if (hdr->flags & ISCSI_FLAG_CMD_WRITE) {
err = iser_prepare_write_cmd(task,
task->imm_count,
task->imm_count +
task->unsol_r2t.data_length,
edtl);
if (err)
goto send_command_error;
}
iser_task->status = ISER_TASK_STATUS_STARTED;
err = iser_post_send(&iser_conn->ib_conn, tx_desc,
iser_signal_comp(sig_count));
if (!err)
return 0;
send_command_error:
iser_err("conn %p failed task->itt %d err %d\n",conn, task->itt, err);
return err;
}
static int iser_create_ib_conn_res(struct iser_conn *ib_conn)
{
struct iser_device *device;
struct ib_qp_init_attr init_attr;
int req_err, resp_err, ret = -ENOMEM;
struct ib_fmr_pool_param params;
BUG_ON(ib_conn->device == NULL);
device = ib_conn->device;
ib_conn->login_buf = kmalloc(ISCSI_DEF_MAX_RECV_SEG_LEN +
ISER_RX_LOGIN_SIZE, GFP_KERNEL);
if (!ib_conn->login_buf)
goto out_err;
ib_conn->login_req_buf = ib_conn->login_buf;
ib_conn->login_resp_buf = ib_conn->login_buf + ISCSI_DEF_MAX_RECV_SEG_LEN;
ib_conn->login_req_dma = ib_dma_map_single(ib_conn->device->ib_device,
(void *)ib_conn->login_req_buf,
ISCSI_DEF_MAX_RECV_SEG_LEN, DMA_TO_DEVICE);
ib_conn->login_resp_dma = ib_dma_map_single(ib_conn->device->ib_device,
(void *)ib_conn->login_resp_buf,
ISER_RX_LOGIN_SIZE, DMA_FROM_DEVICE);
req_err = ib_dma_mapping_error(device->ib_device, ib_conn->login_req_dma);
resp_err = ib_dma_mapping_error(device->ib_device, ib_conn->login_resp_dma);
if (req_err || resp_err) {
if (req_err)
ib_conn->login_req_dma = 0;
if (resp_err)
ib_conn->login_resp_dma = 0;
goto out_err;
}
ib_conn->page_vec = kmalloc(sizeof(struct iser_page_vec) +
(sizeof(u64) * (ISCSI_ISER_SG_TABLESIZE +1)),
GFP_KERNEL);
if (!ib_conn->page_vec)
goto out_err;
ib_conn->page_vec->pages = (u64 *) (ib_conn->page_vec + 1);
params.page_shift = SHIFT_4K;
/*
*/
params.max_pages_per_fmr = ISCSI_ISER_SG_TABLESIZE + 1;
/*
*/
params.pool_size = ISCSI_DEF_XMIT_CMDS_MAX * 2;
params.dirty_watermark = ISCSI_DEF_XMIT_CMDS_MAX;
params.cache = 0;
params.flush_function = NULL;
params.access = (IB_ACCESS_LOCAL_WRITE |
IB_ACCESS_REMOTE_WRITE |
IB_ACCESS_REMOTE_READ);
ib_conn->fmr_pool = ib_create_fmr_pool(device->pd, ¶ms);
if (IS_ERR(ib_conn->fmr_pool)) {
ret = PTR_ERR(ib_conn->fmr_pool);
ib_conn->fmr_pool = NULL;
goto out_err;
}
memset(&init_attr, 0, sizeof init_attr);
init_attr.event_handler = iser_qp_event_callback;
init_attr.qp_context = (void *)ib_conn;
init_attr.send_cq = device->tx_cq;
init_attr.recv_cq = device->rx_cq;
init_attr.cap.max_send_wr = ISER_QP_MAX_REQ_DTOS;
init_attr.cap.max_recv_wr = ISER_QP_MAX_RECV_DTOS;
init_attr.cap.max_send_sge = 2;
init_attr.cap.max_recv_sge = 1;
init_attr.sq_sig_type = IB_SIGNAL_REQ_WR;
init_attr.qp_type = IB_QPT_RC;
ret = rdma_create_qp(ib_conn->cma_id, device->pd, &init_attr);
if (ret)
goto out_err;
ib_conn->qp = ib_conn->cma_id->qp;
iser_err("setting conn %p cma_id %p: fmr_pool %p qp %p\n",
ib_conn, ib_conn->cma_id,
ib_conn->fmr_pool, ib_conn->cma_id->qp);
return ret;
out_err:
iser_err("unable to alloc mem or create resource, err %d\n", ret);
return ret;
}
