aboutsummaryrefslogtreecommitdiff
path: root/sys/dev/irdma/irdma_kcompat.c
diff options
context:
space:
mode:
Diffstat (limited to 'sys/dev/irdma/irdma_kcompat.c')
-rw-r--r--sys/dev/irdma/irdma_kcompat.c301
1 files changed, 239 insertions, 62 deletions
diff --git a/sys/dev/irdma/irdma_kcompat.c b/sys/dev/irdma/irdma_kcompat.c
index 4261fb45d390..51f44133252c 100644
--- a/sys/dev/irdma/irdma_kcompat.c
+++ b/sys/dev/irdma/irdma_kcompat.c
@@ -1,7 +1,7 @@
/*-
* SPDX-License-Identifier: GPL-2.0 or Linux-OpenIB
*
- * Copyright (c) 2018 - 2023 Intel Corporation
+ * Copyright (c) 2018 - 2026 Intel Corporation
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
@@ -470,25 +470,49 @@ irdma_create_ah_wait(struct irdma_pci_f *rf,
int ret;
if (!sleep) {
- int cnt = rf->sc_dev.hw_attrs.max_cqp_compl_wait_time_ms *
- CQP_TIMEOUT_THRESHOLD;
+ bool timeout = false;
+ u64 start = get_jiffies_64();
+ u64 completed_ops = atomic64_read(&rf->sc_dev.cqp->completed_ops);
struct irdma_cqp_request *cqp_request =
sc_ah->ah_info.cqp_request;
+ const u64 timeout_jiffies =
+ msecs_to_jiffies(rf->sc_dev.hw_attrs.max_cqp_compl_wait_time_ms *
+ CQP_TIMEOUT_THRESHOLD);
+
+ /*
+ * NOTE: irdma_check_cqp_progress is not used here because it relies on a notion of a cycle count, but
+ * we want to avoid unnecessary delays. We are in an atomic context here, so we might as well check in
+ * a tight loop.
+ */
+ while (!READ_ONCE(cqp_request->request_done)) {
+ u64 tmp;
+ u64 curr_jiffies;
- do {
irdma_cqp_ce_handler(rf, &rf->ccq.sc_cq);
- mdelay(1);
- } while (!READ_ONCE(cqp_request->request_done) && --cnt);
- if (cnt && !cqp_request->compl_info.op_ret_val) {
+ curr_jiffies = get_jiffies_64();
+ tmp = atomic64_read(&rf->sc_dev.cqp->completed_ops);
+ if (tmp != completed_ops) {
+ /* CQP is progressing. Reset timer. */
+ completed_ops = tmp;
+ start = curr_jiffies;
+ }
+
+ if ((curr_jiffies - start) > timeout_jiffies) {
+ timeout = true;
+ break;
+ }
+ }
+
+ if (!timeout && !cqp_request->compl_info.op_ret_val) {
irdma_put_cqp_request(&rf->cqp, cqp_request);
sc_ah->ah_info.ah_valid = true;
} else {
- ret = !cnt ? -ETIMEDOUT : -EINVAL;
+ ret = timeout ? -ETIMEDOUT : -EINVAL;
irdma_dev_err(&rf->iwdev->ibdev, "CQP create AH error ret = %d opt_ret_val = %d",
ret, cqp_request->compl_info.op_ret_val);
irdma_put_cqp_request(&rf->cqp, cqp_request);
- if (!cnt && !rf->reset) {
+ if (timeout && !rf->reset) {
rf->reset = true;
rf->gen_ops.request_reset(rf);
}
@@ -501,19 +525,10 @@ irdma_create_ah_wait(struct irdma_pci_f *rf,
#define IRDMA_CREATE_AH_MIN_RESP_LEN offsetofend(struct irdma_create_ah_resp, rsvd)
-/**
- * irdma_create_ah - create address handle
- * @ib_ah: ptr to AH
- * @attr: address handle attributes
- * @flags: AH flags to wait
- * @udata: user data
- *
- * returns 0 on success, error otherwise
- */
-int
-irdma_create_ah(struct ib_ah *ib_ah,
- struct ib_ah_attr *attr, u32 flags,
- struct ib_udata *udata)
+static int
+irdma_create_sleepable_ah(struct ib_ah *ib_ah,
+ struct ib_ah_attr *attr, u32 flags,
+ struct ib_udata *udata)
{
struct irdma_pd *pd = to_iwpd(ib_ah->pd);
struct irdma_ah *ah = container_of(ib_ah, struct irdma_ah, ibah);
@@ -613,6 +628,23 @@ err_gid_l2:
return err;
}
+/**
+ * irdma_create_ah - create address handle
+ * @ib_ah: ptr to AH
+ * @attr: address handle attributes
+ * @flags: AH flags to wait
+ * @udata: user data
+ *
+ * returns 0 on success, error otherwise
+ */
+int
+irdma_create_ah(struct ib_ah *ib_ah,
+ struct ib_ah_attr *attr, u32 flags,
+ struct ib_udata *udata)
+{
+ return irdma_create_sleepable_ah(ib_ah, attr, flags, udata);
+}
+
void
irdma_ether_copy(u8 *dmac, struct ib_ah_attr *attr)
{
@@ -691,6 +723,7 @@ irdma_create_qp(struct ib_pd *ibpd,
struct irdma_uk_attrs *uk_attrs = &dev->hw_attrs.uk_attrs;
struct irdma_qp_init_info init_info = {{0}};
struct irdma_qp_host_ctx_info *ctx_info;
+ u32 next_qp = 0;
unsigned long flags;
err_code = irdma_validate_qp_attrs(init_attr, iwdev);
@@ -743,6 +776,9 @@ irdma_create_qp(struct ib_pd *ibpd,
if (init_attr->qp_type == IB_QPT_GSI)
qp_num = 1;
+ else if (dev->hw_attrs.uk_attrs.hw_rev <= IRDMA_GEN_2)
+ err_code = irdma_alloc_rsrc(rf, rf->allocated_qps, rf->max_qp,
+ &qp_num, &next_qp);
else
err_code = irdma_alloc_rsrc(rf, rf->allocated_qps, rf->max_qp,
&qp_num, &rf->next_qp);
@@ -759,7 +795,7 @@ irdma_create_qp(struct ib_pd *ibpd,
iwqp->host_ctx.size = IRDMA_QP_CTX_SIZE;
init_info.pd = &iwpd->sc_pd;
- init_info.qp_uk_init_info.qp_id = iwqp->ibqp.qp_num;
+ init_info.qp_uk_init_info.qp_id = qp_num;
if (!rdma_protocol_roce(&iwdev->ibdev, 1))
init_info.qp_uk_init_info.first_sq_wq = 1;
iwqp->ctx_info.qp_compl_ctx = (uintptr_t)qp;
@@ -769,10 +805,11 @@ irdma_create_qp(struct ib_pd *ibpd,
spin_lock_init(&iwqp->dwork_flush_lock);
if (udata) {
+ INIT_DELAYED_WORK(&iwqp->dwork_flush, irdma_user_flush_worker);
init_info.qp_uk_init_info.abi_ver = iwpd->sc_pd.abi_ver;
err_code = irdma_setup_umode_qp(udata, iwdev, iwqp, &init_info, init_attr);
} else {
- INIT_DELAYED_WORK(&iwqp->dwork_flush, irdma_flush_worker);
+ INIT_DELAYED_WORK(&iwqp->dwork_flush, irdma_kern_flush_worker);
init_info.qp_uk_init_info.abi_ver = IRDMA_ABI_VER;
err_code = irdma_setup_kmode_qp(iwdev, iwqp, &init_info, init_attr);
}
@@ -946,7 +983,6 @@ irdma_create_cq(struct ib_cq *ibcq,
unsigned long flags;
int err_code;
int entries = attr->cqe;
- bool cqe_64byte_ena;
err_code = cq_validate_flags(attr->flags, dev->hw_attrs.uk_attrs.hw_rev);
if (err_code)
@@ -966,10 +1002,9 @@ irdma_create_cq(struct ib_cq *ibcq,
INIT_LIST_HEAD(&iwcq->resize_list);
INIT_LIST_HEAD(&iwcq->cmpl_generated);
info.dev = dev;
- ukinfo->cq_size = max(entries, 4);
+ ukinfo->cq_size = max_t(int, entries, 4);
ukinfo->cq_id = cq_num;
- cqe_64byte_ena = (dev->hw_attrs.uk_attrs.feature_flags & IRDMA_FEATURE_64_BYTE_CQE) ? true : false;
- ukinfo->avoid_mem_cflct = cqe_64byte_ena;
+ iwcq->cq_num = cq_num;
iwcq->ibcq.cqe = info.cq_uk_init_info.cq_size;
atomic_set(&iwcq->armed, 0);
if (attr->comp_vector < rf->ceqs_count)
@@ -1004,8 +1039,6 @@ irdma_create_cq(struct ib_cq *ibcq,
err_code = -EPROTO;
goto cq_free_rsrc;
}
- iwcq->iwpbl = iwpbl;
- iwcq->cq_mem_size = 0;
cqmr = &iwpbl->cq_mr;
if (rf->sc_dev.hw_attrs.uk_attrs.feature_flags &
@@ -1019,7 +1052,6 @@ irdma_create_cq(struct ib_cq *ibcq,
err_code = -EPROTO;
goto cq_free_rsrc;
}
- iwcq->iwpbl_shadow = iwpbl_shadow;
cqmr_shadow = &iwpbl_shadow->cq_mr;
info.shadow_area_pa = cqmr_shadow->cq_pbl.addr;
cqmr->split = true;
@@ -1043,14 +1075,11 @@ irdma_create_cq(struct ib_cq *ibcq,
}
entries++;
- if (!cqe_64byte_ena && dev->hw_attrs.uk_attrs.hw_rev >= IRDMA_GEN_2)
+ if (dev->hw_attrs.uk_attrs.hw_rev == IRDMA_GEN_2)
entries *= 2;
ukinfo->cq_size = entries;
- if (cqe_64byte_ena)
- rsize = info.cq_uk_init_info.cq_size * sizeof(struct irdma_extended_cqe);
- else
- rsize = info.cq_uk_init_info.cq_size * sizeof(struct irdma_cqe);
+ rsize = info.cq_uk_init_info.cq_size * sizeof(struct irdma_cqe);
iwcq->kmem.size = round_up(rsize, IRDMA_HW_PAGE_SIZE);
iwcq->kmem.va = irdma_allocate_dma_mem(dev->hw, &iwcq->kmem,
iwcq->kmem.size, IRDMA_HW_PAGE_SIZE);
@@ -1094,6 +1123,7 @@ irdma_create_cq(struct ib_cq *ibcq,
cqp_info->in.u.cq_create.cq = cq;
cqp_info->in.u.cq_create.check_overflow = true;
cqp_info->in.u.cq_create.scratch = (uintptr_t)cqp_request;
+ cqp_info->create = true;
status = irdma_handle_cqp_op(rf, cqp_request);
irdma_put_cqp_request(&rf->cqp, cqp_request);
if (status) {
@@ -1114,7 +1144,7 @@ irdma_create_cq(struct ib_cq *ibcq,
}
}
- rf->cq_table[cq_num] = iwcq;
+ WRITE_ONCE(rf->cq_table[cq_num], iwcq);
init_completion(&iwcq->free_cq);
return 0;
@@ -1227,6 +1257,86 @@ done:
return 0;
}
+/**
+ * irdma_reg_user_mr - Register a user memory region
+ * @pd: ptr of pd
+ * @start: virtual start address
+ * @len: length of mr
+ * @virt: virtual address
+ * @access: access of mr
+ * @udata: user data
+ */
+struct ib_mr *
+irdma_reg_user_mr(struct ib_pd *pd, u64 start, u64 len,
+ u64 virt, int access,
+ struct ib_udata *udata)
+{
+#define IRDMA_MEM_REG_MIN_REQ_LEN offsetofend(struct irdma_mem_reg_req, sq_pages)
+ struct irdma_device *iwdev = to_iwdev(pd->device);
+ struct irdma_mem_reg_req req = {};
+ struct ib_umem *region;
+ struct irdma_mr *iwmr;
+ int err;
+
+ if (len > iwdev->rf->sc_dev.hw_attrs.max_mr_size)
+ return ERR_PTR(-EINVAL);
+
+ if (udata->inlen < IRDMA_MEM_REG_MIN_REQ_LEN)
+ return ERR_PTR(-EINVAL);
+
+ region = ib_umem_get(pd->uobject->context, start, len, access, 0);
+
+ if (IS_ERR(region)) {
+ irdma_debug(&iwdev->rf->sc_dev, IRDMA_DEBUG_VERBS,
+ "Failed to create ib_umem region err=%ld\n",
+ PTR_ERR(region));
+ return (struct ib_mr *)region;
+ }
+
+ if (ib_copy_from_udata(&req, udata, min(sizeof(req), udata->inlen))) {
+ ib_umem_release(region);
+ return ERR_PTR(-EFAULT);
+ }
+
+ iwmr = irdma_alloc_iwmr(region, pd, virt, req.reg_type);
+ if (IS_ERR(iwmr)) {
+ ib_umem_release(region);
+ return (struct ib_mr *)iwmr;
+ }
+
+ switch (req.reg_type) {
+ case IRDMA_MEMREG_TYPE_QP:
+ err = irdma_reg_user_mr_type_qp(req, udata, iwmr);
+ if (err)
+ goto error;
+
+ break;
+ case IRDMA_MEMREG_TYPE_CQ:
+ err = irdma_reg_user_mr_type_cq(req, udata, iwmr);
+ if (err)
+ goto error;
+
+ break;
+ case IRDMA_MEMREG_TYPE_MEM:
+ err = irdma_reg_user_mr_type_mem(iwmr, access, true);
+ if (err)
+ goto error;
+
+ break;
+ default:
+ err = -EINVAL;
+ goto error;
+ }
+
+ return &iwmr->ibmr;
+
+error:
+ ib_umem_release(region);
+ irdma_free_iwmr(iwmr);
+
+ return ERR_PTR(err);
+}
+
/*
* irdma_rereg_user_mr - Re-Register a user memory region @ibmr: ib mem to access iwarp mr pointer @flags: bit mask to
* indicate which of the attr's of MR modified @start: virtual start address @len: length of mr @virt: virtual address
@@ -1512,20 +1622,19 @@ irdma_query_port(struct ib_device *ibdev, u8 port,
/* no need to zero out pros here. done by caller */
props->max_mtu = IB_MTU_4096;
- props->active_mtu = ib_mtu_int_to_enum(if_getmtu(netdev));
+ props->active_mtu = min(props->max_mtu, iboe_get_mtu(if_getmtu(netdev)));
props->lid = 1;
props->lmc = 0;
props->sm_lid = 0;
props->sm_sl = 0;
- if ((if_getlinkstate(netdev) == LINK_STATE_UP) &&
- (if_getdrvflags(netdev) & IFF_DRV_RUNNING)) {
+ if ((if_getlinkstate(netdev) == LINK_STATE_UP) && (if_getdrvflags(netdev) & IFF_DRV_RUNNING)) {
props->state = IB_PORT_ACTIVE;
props->phys_state = IB_PORT_PHYS_STATE_LINK_UP;
} else {
props->state = IB_PORT_DOWN;
props->phys_state = IB_PORT_PHYS_STATE_DISABLED;
}
- ib_get_eth_speed(ibdev, port, &props->active_speed, &props->active_width);
+ ib_get_eth_speed(ibdev, port, (u16 *)& props->active_speed, &props->active_width);
if (rdma_protocol_roce(ibdev, 1)) {
props->gid_tbl_len = 32;
@@ -1720,36 +1829,104 @@ kc_set_rdma_uverbs_cmd_mask(struct irdma_device *iwdev)
iwdev->ibdev.uverbs_ex_cmd_mask |= BIT_ULL(IB_USER_VERBS_EX_CMD_CREATE_CQ);
}
-int
-ib_get_eth_speed(struct ib_device *ibdev, u32 port_num, u8 *speed, u8 *width)
+static void
+ib_get_width_and_speed(u32 netdev_speed, u32 lanes,
+ u16 *speed, u8 *width)
{
- if_t netdev = ibdev->get_netdev(ibdev, port_num);
- u32 netdev_speed;
+ if (!lanes) {
+ if (netdev_speed <= SPEED_1000) {
+ *width = IB_WIDTH_1X;
+ *speed = IB_SPEED_SDR;
+ } else if (netdev_speed <= SPEED_10000) {
+ *width = IB_WIDTH_1X;
+ *speed = IB_SPEED_FDR10;
+ } else if (netdev_speed <= SPEED_20000) {
+ *width = IB_WIDTH_4X;
+ *speed = IB_SPEED_DDR;
+ } else if (netdev_speed <= SPEED_25000) {
+ *width = IB_WIDTH_1X;
+ *speed = IB_SPEED_EDR;
+ } else if (netdev_speed <= SPEED_40000) {
+ *width = IB_WIDTH_4X;
+ *speed = IB_SPEED_FDR10;
+ } else if (netdev_speed <= SPEED_50000) {
+ *width = IB_WIDTH_2X;
+ *speed = IB_SPEED_EDR;
+ } else if (netdev_speed <= SPEED_100000) {
+ *width = IB_WIDTH_4X;
+ *speed = IB_SPEED_EDR;
+ } else if (netdev_speed <= SPEED_200000) {
+ *width = IB_WIDTH_4X;
+ *speed = IB_SPEED_HDR;
+ } else {
+ *width = IB_WIDTH_4X;
+ *speed = IB_SPEED_NDR;
+ }
- if (!netdev)
- return -ENODEV;
+ return;
+ }
- netdev_speed = if_getbaudrate(netdev);
- dev_put(netdev);
- if (netdev_speed <= SPEED_1000) {
+ switch (lanes) {
+ case 1:
*width = IB_WIDTH_1X;
- *speed = IB_SPEED_SDR;
- } else if (netdev_speed <= SPEED_10000) {
- *width = IB_WIDTH_1X;
- *speed = IB_SPEED_FDR10;
- } else if (netdev_speed <= SPEED_20000) {
+ break;
+ case 2:
+ *width = IB_WIDTH_2X;
+ break;
+ case 4:
*width = IB_WIDTH_4X;
- *speed = IB_SPEED_DDR;
- } else if (netdev_speed <= SPEED_25000) {
+ break;
+ case 8:
+ *width = IB_WIDTH_8X;
+ break;
+ case 12:
+ *width = IB_WIDTH_12X;
+ break;
+ default:
*width = IB_WIDTH_1X;
- *speed = IB_SPEED_EDR;
- } else if (netdev_speed <= SPEED_40000) {
- *width = IB_WIDTH_4X;
+ }
+
+ switch (netdev_speed / lanes) {
+ case SPEED_2500:
+ *speed = IB_SPEED_SDR;
+ break;
+ case SPEED_5000:
+ *speed = IB_SPEED_DDR;
+ break;
+ case SPEED_10000:
*speed = IB_SPEED_FDR10;
- } else {
- *width = IB_WIDTH_4X;
+ break;
+ case SPEED_14000:
+ *speed = IB_SPEED_FDR;
+ break;
+ case SPEED_25000:
*speed = IB_SPEED_EDR;
+ break;
+ case SPEED_50000:
+ *speed = IB_SPEED_HDR;
+ break;
+ case SPEED_100000:
+ *speed = IB_SPEED_NDR;
+ break;
+ default:
+ *speed = IB_SPEED_SDR;
}
+}
+
+int
+ib_get_eth_speed(struct ib_device *ibdev, u32 port_num, u16 *speed, u8 *width)
+{
+ if_t netdev = ibdev->get_netdev(ibdev, port_num);
+ u32 netdev_speed, lanes;
+
+ if (!netdev)
+ return -ENODEV;
+
+ netdev_speed = (u32)if_getbaudrate(netdev);
+ dev_put(netdev);
+ lanes = 0;
+
+ ib_get_width_and_speed(netdev_speed, lanes, speed, width);
return 0;
}
generated by cgit v1.2.3 (git 2.25.1) at 2026-04-27 10:08:58 +0000