int mlx4_ib_destroy_cq(struct ib_cq *cq)
{
struct mlx4_ib_dev *dev = to_mdev(cq->device);
struct mlx4_ib_cq *mcq = to_mcq(cq);
mlx4_cq_free(dev->dev, &mcq->mcq);
mlx4_mtt_cleanup(dev->dev, &mcq->buf.mtt);
if (cq->uobject) {
mlx4_ib_db_unmap_user(to_mucontext(cq->uobject->context), &mcq->db);
ib_umem_release(mcq->umem);
} else {
mlx4_ib_free_cq_buf(dev, &mcq->buf, cq->cqe);
mlx4_db_free(dev->dev, &mcq->db);
}
kfree(mcq);
return 0;
}
static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
struct mlx4_ib_qp **cur_qp,
struct ib_wc *wc)
{
struct mlx4_cqe *cqe;
struct mlx4_qp *mqp;
struct mlx4_ib_wq *wq;
struct mlx4_ib_srq *srq;
int is_send;
int is_error;
u32 g_mlpath_rqpn;
u16 wqe_ctr;
repoll:
cqe = next_cqe_sw(cq);
if (!cqe)
return -EAGAIN;
++cq->mcq.cons_index;
/*
* Make sure we read CQ entry contents after we've checked the
* ownership bit.
*/
rmb();
is_send = cqe->owner_sr_opcode & MLX4_CQE_IS_SEND_MASK;
is_error = (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) ==
MLX4_CQE_OPCODE_ERROR;
if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) == MLX4_OPCODE_NOP &&
is_send)) {
printk(KERN_WARNING "Completion for NOP opcode detected!\n");
return -EINVAL;
}
/* Resize CQ in progress */
if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) == MLX4_CQE_OPCODE_RESIZE)) {
if (cq->resize_buf) {
struct mlx4_ib_dev *dev = to_mdev(cq->ibcq.device);
mlx4_ib_free_cq_buf(dev, &cq->buf, cq->ibcq.cqe);
cq->buf = cq->resize_buf->buf;
cq->ibcq.cqe = cq->resize_buf->cqe;
kfree(cq->resize_buf);
cq->resize_buf = NULL;
}
goto repoll;
}
if (!*cur_qp ||
(be32_to_cpu(cqe->vlan_my_qpn) & MLX4_CQE_QPN_MASK) != (*cur_qp)->mqp.qpn) {
/*
* We do not have to take the QP table lock here,
* because CQs will be locked while QPs are removed
* from the table.
*/
mqp = __mlx4_qp_lookup(to_mdev(cq->ibcq.device)->dev,
be32_to_cpu(cqe->vlan_my_qpn));
if (unlikely(!mqp)) {
printk(KERN_WARNING "CQ %06x with entry for unknown QPN %06x\n",
cq->mcq.cqn, be32_to_cpu(cqe->vlan_my_qpn) & MLX4_CQE_QPN_MASK);
return -EINVAL;
}
*cur_qp = to_mibqp(mqp);
}
wc->qp = &(*cur_qp)->ibqp;
if (is_send) {
wq = &(*cur_qp)->sq;
if (!(*cur_qp)->sq_signal_bits) {
wqe_ctr = be16_to_cpu(cqe->wqe_index);
wq->tail += (u16) (wqe_ctr - (u16) wq->tail);
}
wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)];
++wq->tail;
} else if ((*cur_qp)->ibqp.srq) {
srq = to_msrq((*cur_qp)->ibqp.srq);
wqe_ctr = be16_to_cpu(cqe->wqe_index);
wc->wr_id = srq->wrid[wqe_ctr];
mlx4_ib_free_srq_wqe(srq, wqe_ctr);
} else {
wq = &(*cur_qp)->rq;
wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)];
++wq->tail;
}
if (unlikely(is_error)) {
mlx4_ib_handle_error_cqe((struct mlx4_err_cqe *) cqe, wc);
return 0;
}
wc->status = IB_WC_SUCCESS;
if (is_send) {
wc->wc_flags = 0;
switch (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) {
case MLX4_OPCODE_RDMA_WRITE_IMM:
wc->wc_flags |= IB_WC_WITH_IMM;
case MLX4_OPCODE_RDMA_WRITE:
wc->opcode = IB_WC_RDMA_WRITE;
break;
case MLX4_OPCODE_SEND_IMM:
wc->wc_flags |= IB_WC_WITH_IMM;
case MLX4_OPCODE_SEND:
case MLX4_OPCODE_SEND_INVAL:
wc->opcode = IB_WC_SEND;
break;
case MLX4_OPCODE_RDMA_READ:
wc->opcode = IB_WC_RDMA_READ;
wc->byte_len = be32_to_cpu(cqe->byte_cnt);
break;
case MLX4_OPCODE_ATOMIC_CS:
wc->opcode = IB_WC_COMP_SWAP;
wc->byte_len = 8;
break;
case MLX4_OPCODE_ATOMIC_FA:
wc->opcode = IB_WC_FETCH_ADD;
wc->byte_len = 8;
break;
case MLX4_OPCODE_MASKED_ATOMIC_CS:
wc->opcode = IB_WC_MASKED_COMP_SWAP;
wc->byte_len = 8;
break;
case MLX4_OPCODE_MASKED_ATOMIC_FA:
wc->opcode = IB_WC_MASKED_FETCH_ADD;
wc->byte_len = 8;
break;
case MLX4_OPCODE_BIND_MW:
wc->opcode = IB_WC_BIND_MW;
break;
case MLX4_OPCODE_LSO:
wc->opcode = IB_WC_LSO;
break;
case MLX4_OPCODE_FMR:
wc->opcode = IB_WC_FAST_REG_MR;
break;
case MLX4_OPCODE_LOCAL_INVAL:
wc->opcode = IB_WC_LOCAL_INV;
break;
}
} else {
wc->byte_len = be32_to_cpu(cqe->byte_cnt);
switch (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) {
case MLX4_RECV_OPCODE_RDMA_WRITE_IMM:
wc->opcode = IB_WC_RECV_RDMA_WITH_IMM;
wc->wc_flags = IB_WC_WITH_IMM;
wc->ex.imm_data = cqe->immed_rss_invalid;
break;
case MLX4_RECV_OPCODE_SEND_INVAL:
wc->opcode = IB_WC_RECV;
wc->wc_flags = IB_WC_WITH_INVALIDATE;
wc->ex.invalidate_rkey = be32_to_cpu(cqe->immed_rss_invalid);
break;
case MLX4_RECV_OPCODE_SEND:
wc->opcode = IB_WC_RECV;
wc->wc_flags = 0;
break;
case MLX4_RECV_OPCODE_SEND_IMM:
wc->opcode = IB_WC_RECV;
wc->wc_flags = IB_WC_WITH_IMM;
wc->ex.imm_data = cqe->immed_rss_invalid;
break;
}
wc->slid = be16_to_cpu(cqe->rlid);
g_mlpath_rqpn = be32_to_cpu(cqe->g_mlpath_rqpn);
wc->src_qp = g_mlpath_rqpn & 0xffffff;
wc->dlid_path_bits = (g_mlpath_rqpn >> 24) & 0x7f;
wc->wc_flags |= g_mlpath_rqpn & 0x80000000 ? IB_WC_GRH : 0;
wc->pkey_index = be32_to_cpu(cqe->immed_rss_invalid) & 0x7f;
wc->csum_ok = mlx4_ib_ipoib_csum_ok(cqe->status, cqe->checksum);
if (rdma_port_get_link_layer(wc->qp->device,
(*cur_qp)->port) == IB_LINK_LAYER_ETHERNET)
wc->sl = be16_to_cpu(cqe->sl_vid) >> 13;
else
wc->sl = be16_to_cpu(cqe->sl_vid) >> 12;
}
struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector,
struct ib_ucontext *context,
struct ib_udata *udata)
{
struct mlx4_ib_dev *dev = to_mdev(ibdev);
struct mlx4_ib_cq *cq;
struct mlx4_uar *uar;
int err;
if (entries < 1 || entries > dev->dev->caps.max_cqes)
return ERR_PTR(-EINVAL);
cq = kmalloc(sizeof *cq, GFP_KERNEL);
if (!cq)
return ERR_PTR(-ENOMEM);
entries = roundup_pow_of_two(entries + 1);
cq->ibcq.cqe = entries - 1;
mutex_init(&cq->resize_mutex);
spin_lock_init(&cq->lock);
cq->resize_buf = NULL;
cq->resize_umem = NULL;
if (context) {
struct mlx4_ib_create_cq ucmd;
if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd)) {
err = -EFAULT;
goto err_cq;
}
err = mlx4_ib_get_cq_umem(dev, context, &cq->buf, &cq->umem,
ucmd.buf_addr, entries);
if (err)
goto err_cq;
err = mlx4_ib_db_map_user(to_mucontext(context), ucmd.db_addr,
&cq->db);
if (err)
goto err_mtt;
uar = &to_mucontext(context)->uar;
} else {
err = mlx4_db_alloc(dev->dev, &cq->db, 1);
if (err)
goto err_cq;
cq->mcq.set_ci_db = cq->db.db;
cq->mcq.arm_db = cq->db.db + 1;
*cq->mcq.set_ci_db = 0;
*cq->mcq.arm_db = 0;
err = mlx4_ib_alloc_cq_buf(dev, &cq->buf, entries);
if (err)
goto err_db;
uar = &dev->priv_uar;
}
err = mlx4_cq_alloc(dev->dev, entries, &cq->buf.mtt, uar,
cq->db.dma, &cq->mcq, vector, 0);
if (err)
goto err_dbmap;
cq->mcq.comp = mlx4_ib_cq_comp;
cq->mcq.event = mlx4_ib_cq_event;
if (context)
if (ib_copy_to_udata(udata, &cq->mcq.cqn, sizeof (__u32))) {
err = -EFAULT;
goto err_dbmap;
}
return &cq->ibcq;
err_dbmap:
if (context)
mlx4_ib_db_unmap_user(to_mucontext(context), &cq->db);
err_mtt:
mlx4_mtt_cleanup(dev->dev, &cq->buf.mtt);
if (context)
ib_umem_release(cq->umem);
else
mlx4_ib_free_cq_buf(dev, &cq->buf, cq->ibcq.cqe);
err_db:
if (!context)
mlx4_db_free(dev->dev, &cq->db);
err_cq:
kfree(cq);
return ERR_PTR(err);
}
int mlx4_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata)
{
struct mlx4_ib_dev *dev = to_mdev(ibcq->device);
struct mlx4_ib_cq *cq = to_mcq(ibcq);
struct mlx4_mtt mtt;
int outst_cqe;
int err;
mutex_lock(&cq->resize_mutex);
if (entries < 1 || entries > dev->dev->caps.max_cqes) {
err = -EINVAL;
goto out;
}
entries = roundup_pow_of_two(entries + 1);
if (entries == ibcq->cqe + 1) {
err = 0;
goto out;
}
if (ibcq->uobject) {
err = mlx4_alloc_resize_umem(dev, cq, entries, udata);
if (err)
goto out;
} else {
/* Can't be smaller than the number of outstanding CQEs */
outst_cqe = mlx4_ib_get_outstanding_cqes(cq);
if (entries < outst_cqe + 1) {
err = 0;
goto out;
}
err = mlx4_alloc_resize_buf(dev, cq, entries);
if (err)
goto out;
}
mtt = cq->buf.mtt;
err = mlx4_cq_resize(dev->dev, &cq->mcq, entries, &cq->resize_buf->buf.mtt);
if (err)
goto err_buf;
mlx4_mtt_cleanup(dev->dev, &mtt);
if (ibcq->uobject) {
cq->buf = cq->resize_buf->buf;
cq->ibcq.cqe = cq->resize_buf->cqe;
ib_umem_release(cq->umem);
cq->umem = cq->resize_umem;
kfree(cq->resize_buf);
cq->resize_buf = NULL;
cq->resize_umem = NULL;
} else {
struct mlx4_ib_cq_buf tmp_buf;
int tmp_cqe = 0;
spin_lock_irq(&cq->lock);
if (cq->resize_buf) {
mlx4_ib_cq_resize_copy_cqes(cq);
tmp_buf = cq->buf;
tmp_cqe = cq->ibcq.cqe;
cq->buf = cq->resize_buf->buf;
cq->ibcq.cqe = cq->resize_buf->cqe;
kfree(cq->resize_buf);
cq->resize_buf = NULL;
}
spin_unlock_irq(&cq->lock);
if (tmp_cqe)
mlx4_ib_free_cq_buf(dev, &tmp_buf, tmp_cqe);
}
goto out;
err_buf:
mlx4_mtt_cleanup(dev->dev, &cq->resize_buf->buf.mtt);
if (!ibcq->uobject)
mlx4_ib_free_cq_buf(dev, &cq->resize_buf->buf,
cq->resize_buf->cqe);
kfree(cq->resize_buf);
cq->resize_buf = NULL;
if (cq->resize_umem) {
ib_umem_release(cq->resize_umem);
cq->resize_umem = NULL;
}
out:
mutex_unlock(&cq->resize_mutex);
return err;
}
struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev,
const struct ib_cq_init_attr *attr,
struct ib_udata *udata)
{
int entries = attr->cqe;
int vector = attr->comp_vector;
struct mlx4_ib_dev *dev = to_mdev(ibdev);
struct mlx4_ib_cq *cq;
struct mlx4_uar *uar;
void *buf_addr;
int err;
struct mlx4_ib_ucontext *context = rdma_udata_to_drv_context(
udata, struct mlx4_ib_ucontext, ibucontext);
if (entries < 1 || entries > dev->dev->caps.max_cqes)
return ERR_PTR(-EINVAL);
if (attr->flags & ~CQ_CREATE_FLAGS_SUPPORTED)
return ERR_PTR(-EINVAL);
cq = kzalloc(sizeof(*cq), GFP_KERNEL);
if (!cq)
return ERR_PTR(-ENOMEM);
entries = roundup_pow_of_two(entries + 1);
cq->ibcq.cqe = entries - 1;
mutex_init(&cq->resize_mutex);
spin_lock_init(&cq->lock);
cq->resize_buf = NULL;
cq->resize_umem = NULL;
cq->create_flags = attr->flags;
INIT_LIST_HEAD(&cq->send_qp_list);
INIT_LIST_HEAD(&cq->recv_qp_list);
if (udata) {
struct mlx4_ib_create_cq ucmd;
if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd)) {
err = -EFAULT;
goto err_cq;
}
buf_addr = (void *)(unsigned long)ucmd.buf_addr;
err = mlx4_ib_get_cq_umem(dev, udata, &cq->buf, &cq->umem,
ucmd.buf_addr, entries);
if (err)
goto err_cq;
err = mlx4_ib_db_map_user(udata, ucmd.db_addr, &cq->db);
if (err)
goto err_mtt;
uar = &context->uar;
cq->mcq.usage = MLX4_RES_USAGE_USER_VERBS;
} else {
err = mlx4_db_alloc(dev->dev, &cq->db, 1);
if (err)
goto err_cq;
cq->mcq.set_ci_db = cq->db.db;
cq->mcq.arm_db = cq->db.db + 1;
*cq->mcq.set_ci_db = 0;
*cq->mcq.arm_db = 0;
err = mlx4_ib_alloc_cq_buf(dev, &cq->buf, entries);
if (err)
goto err_db;
buf_addr = &cq->buf.buf;
uar = &dev->priv_uar;
cq->mcq.usage = MLX4_RES_USAGE_DRIVER;
}
if (dev->eq_table)
vector = dev->eq_table[vector % ibdev->num_comp_vectors];
err = mlx4_cq_alloc(dev->dev, entries, &cq->buf.mtt, uar, cq->db.dma,
&cq->mcq, vector, 0,
!!(cq->create_flags &
IB_UVERBS_CQ_FLAGS_TIMESTAMP_COMPLETION),
buf_addr, !!udata);
if (err)
goto err_dbmap;
if (udata)
cq->mcq.tasklet_ctx.comp = mlx4_ib_cq_comp;
else
cq->mcq.comp = mlx4_ib_cq_comp;
cq->mcq.event = mlx4_ib_cq_event;
if (udata)
if (ib_copy_to_udata(udata, &cq->mcq.cqn, sizeof (__u32))) {
err = -EFAULT;
goto err_cq_free;
}
return &cq->ibcq;
err_cq_free:
mlx4_cq_free(dev->dev, &cq->mcq);
err_dbmap:
if (udata)
mlx4_ib_db_unmap_user(context, &cq->db);
err_mtt:
mlx4_mtt_cleanup(dev->dev, &cq->buf.mtt);
if (udata)
ib_umem_release(cq->umem);
else
mlx4_ib_free_cq_buf(dev, &cq->buf, cq->ibcq.cqe);
err_db:
if (!udata)
mlx4_db_free(dev->dev, &cq->db);
err_cq:
kfree(cq);
return ERR_PTR(err);
}