hollalinux
/
linux-sg2042
mirror of https://github.com/sophgo/linux-riscv.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

v6.4 merge window RDMA pull request 

Usual wide collection of unrelated items in drivers:
 
 - Driver bug fixes and treewide cleanups in hfi1, siw, qib, mlx5, rxe,
   usnic, usnic, bnxt_re, ocrdma, iser
    * Unnecessary NULL checks
    * kmap obsolescence
    * pci_enable_pcie_error_reporting() obsolescence
    * Unused variables and macros
    * trace event related warnings
    * casting warnings
 
 - Code cleanups for irdm and erdma
 
 - EFA reporting of 128 byte PCIe TLP support
 
 - mlx5 more agressively uses the out of order HW feature
 
 - Big rework of how state machines and tasks work in rxe
 
 - Fix a syzkaller found crash netdev refcount leak in siw
 
 - bnxt_re revises their HW description header
 
 - Congestion control for bnxt_re
 
 - Use mmu_notifiers more safely in hfi1
 
 - mlx5 gets better support for PCIe relaxed ordering inside VMs
 -----BEGIN PGP SIGNATURE-----
 
 iHUEABYIAB0WIQRRRCHOFoQz/8F5bUaFwuHvBreFYQUCZEva5wAKCRCFwuHvBreF
 YZFmAQC9T3b/XQ3bRknYciuzbatC98o9xB0FTqmEFYGj+Y2lVAD9EEVe3HKfHfi3
 t/GxXYB5r22oxg5bgsblZfEdEdTVCg8=
 =akMm
 -----END PGP SIGNATURE-----

Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma

Pull rdma updates from Jason Gunthorpe:
 "Usual wide collection of unrelated items in drivers:

   - Driver bug fixes and treewide cleanups in hfi1, siw, qib, mlx5,
     rxe, usnic, usnic, bnxt_re, ocrdma, iser:
       - remove unnecessary NULL checks
       - kmap obsolescence
       - pci_enable_pcie_error_reporting() obsolescence
       - unused variables and macros
       - trace event related warnings
       - casting warnings

   - Code cleanups for irdm and erdma

   - EFA reporting of 128 byte PCIe TLP support

   - mlx5 more agressively uses the out of order HW feature

   - Big rework of how state machines and tasks work in rxe

   - Fix a syzkaller found crash netdev refcount leak in siw

   - bnxt_re revises their HW description header

   - Congestion control for bnxt_re

   - Use mmu_notifiers more safely in hfi1

   - mlx5 gets better support for PCIe relaxed ordering inside VMs"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (81 commits)
  RDMA/efa: Add rdma write capability to device caps
  RDMA/mlx5: Use correct device num_ports when modify DC
  RDMA/irdma: Drop spurious WQ_UNBOUND from alloc_ordered_workqueue() call
  RDMA/rxe: Fix spinlock recursion deadlock on requester
  RDMA/mlx5: Fix flow counter query via DEVX
  RDMA/rxe: Protect QP state with qp->state_lock
  RDMA/rxe: Move code to check if drained to subroutine
  RDMA/rxe: Remove qp->req.state
  RDMA/rxe: Remove qp->comp.state
  RDMA/rxe: Remove qp->resp.state
  RDMA/mlx5: Allow relaxed ordering read in VFs and VMs
  net/mlx5: Update relaxed ordering read HCA capabilities
  RDMA/mlx5: Check pcie_relaxed_ordering_enabled() in UMR
  RDMA/mlx5: Remove pcie_relaxed_ordering_enabled() check for RO write
  RDMA: Add ib_virt_dma_to_page()
  RDMA/rxe: Fix the error "trying to register non-static key in rxe_cleanup_task"
  RDMA/irdma: Slightly optimize irdma_form_ah_cm_frame()
  RDMA/rxe: Fix incorrect TASKLET_STATE_SCHED check in rxe_task.c
  IB/hfi1: Place struct mmu_rb_handler on cache line start
  IB/hfi1: Fix bugs with non-PAGE_SIZE-end multi-iovec user SDMA requests
  ...
This commit is contained in:
Linus Torvalds 2023-04-29 17:21:24 -07:00
parent 1ae78a1451 531094dc71
commit af3877265d
106 changed files with 7896 additions and 5408 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
drivers/infiniband/core/cm.c
View File

@ -2912,6 +2912,8 @@ static int cm_send_rej_locked(struct cm_id_private *cm_id_priv,
(ari && ari_length > IB_CM_REJ_ARI_LENGTH))
return -EINVAL;
trace_icm_send_rej(&cm_id_priv->id, reason);
switch (state) {
case IB_CM_REQ_SENT:
case IB_CM_MRA_REQ_RCVD:
@ -2942,7 +2944,6 @@ static int cm_send_rej_locked(struct cm_id_private *cm_id_priv,
return -EINVAL;
}
trace_icm_send_rej(&cm_id_priv->id, reason);
ret = ib_post_send_mad(msg, NULL);
if (ret) {
cm_free_msg(msg);

6
drivers/infiniband/core/cma.c
View File

@ -709,8 +709,7 @@ cma_validate_port(struct ib_device *device, u32 port,
}
sgid_attr = rdma_find_gid_by_port(device, gid, gid_type, port, ndev);
if (ndev)
dev_put(ndev);
dev_put(ndev);
return sgid_attr;
}
@ -2429,8 +2428,7 @@ err_unlock:
mutex_unlock(&listen_id->handler_mutex);
net_dev_put:
if (net_dev)
dev_put(net_dev);
dev_put(net_dev);
return ret;
}

23
drivers/infiniband/core/user_mad.c
View File

@ -131,6 +131,11 @@ struct ib_umad_packet {
struct ib_user_mad mad;
};
struct ib_rmpp_mad_hdr {
struct ib_mad_hdr mad_hdr;
struct ib_rmpp_hdr rmpp_hdr;
} __packed;
#define CREATE_TRACE_POINTS
#include <trace/events/ib_umad.h>
@ -494,11 +499,11 @@ static ssize_t ib_umad_write(struct file *filp, const char __user *buf,
size_t count, loff_t *pos)
{
struct ib_umad_file *file = filp->private_data;
struct ib_rmpp_mad_hdr *rmpp_mad_hdr;
struct ib_umad_packet *packet;
struct ib_mad_agent *agent;
struct rdma_ah_attr ah_attr;
struct ib_ah *ah;
struct ib_rmpp_mad *rmpp_mad;
__be64 *tid;
int ret, data_len, hdr_len, copy_offset, rmpp_active;
u8 base_version;
@ -506,7 +511,7 @@ static ssize_t ib_umad_write(struct file *filp, const char __user *buf,
if (count < hdr_size(file) + IB_MGMT_RMPP_HDR)
return -EINVAL;
packet = kzalloc(sizeof *packet + IB_MGMT_RMPP_HDR, GFP_KERNEL);
packet = kzalloc(sizeof(*packet) + IB_MGMT_RMPP_HDR, GFP_KERNEL);
if (!packet)
return -ENOMEM;
@ -560,13 +565,13 @@ static ssize_t ib_umad_write(struct file *filp, const char __user *buf,
goto err_up;
}
rmpp_mad = (struct ib_rmpp_mad *) packet->mad.data;
hdr_len = ib_get_mad_data_offset(rmpp_mad->mad_hdr.mgmt_class);
rmpp_mad_hdr = (struct ib_rmpp_mad_hdr *)packet->mad.data;
hdr_len = ib_get_mad_data_offset(rmpp_mad_hdr->mad_hdr.mgmt_class);
if (ib_is_mad_class_rmpp(rmpp_mad->mad_hdr.mgmt_class)
if (ib_is_mad_class_rmpp(rmpp_mad_hdr->mad_hdr.mgmt_class)
&& ib_mad_kernel_rmpp_agent(agent)) {
copy_offset = IB_MGMT_RMPP_HDR;
rmpp_active = ib_get_rmpp_flags(&rmpp_mad->rmpp_hdr) &
rmpp_active = ib_get_rmpp_flags(&rmpp_mad_hdr->rmpp_hdr) &
IB_MGMT_RMPP_FLAG_ACTIVE;
} else {
copy_offset = IB_MGMT_MAD_HDR;
@ -615,12 +620,12 @@ static ssize_t ib_umad_write(struct file *filp, const char __user *buf,
tid = &((struct ib_mad_hdr *) packet->msg->mad)->tid;
*tid = cpu_to_be64(((u64) agent->hi_tid) << 32 |
(be64_to_cpup(tid) & 0xffffffff));
rmpp_mad->mad_hdr.tid = *tid;
rmpp_mad_hdr->mad_hdr.tid = *tid;
}
if (!ib_mad_kernel_rmpp_agent(agent)
&& ib_is_mad_class_rmpp(rmpp_mad->mad_hdr.mgmt_class)
&& (ib_get_rmpp_flags(&rmpp_mad->rmpp_hdr) & IB_MGMT_RMPP_FLAG_ACTIVE)) {
&& ib_is_mad_class_rmpp(rmpp_mad_hdr->mad_hdr.mgmt_class)
&& (ib_get_rmpp_flags(&rmpp_mad_hdr->rmpp_hdr) & IB_MGMT_RMPP_FLAG_ACTIVE)) {
spin_lock_irq(&file->send_lock);
list_add_tail(&packet->list, &file->send_list);
spin_unlock_irq(&file->send_lock);

109
drivers/infiniband/hw/bnxt_re/ib_verbs.c
View File

@ -2912,6 +2912,106 @@ fail:
return rc;
}
static void bnxt_re_resize_cq_complete(struct bnxt_re_cq *cq)
{
struct bnxt_re_dev *rdev = cq->rdev;
bnxt_qplib_resize_cq_complete(&rdev->qplib_res, &cq->qplib_cq);
cq->qplib_cq.max_wqe = cq->resize_cqe;
if (cq->resize_umem) {
ib_umem_release(cq->umem);
cq->umem = cq->resize_umem;
cq->resize_umem = NULL;
cq->resize_cqe = 0;
}
}
int bnxt_re_resize_cq(struct ib_cq *ibcq, int cqe, struct ib_udata *udata)
{
struct bnxt_qplib_sg_info sg_info = {};
struct bnxt_qplib_dpi *orig_dpi = NULL;
struct bnxt_qplib_dev_attr *dev_attr;
struct bnxt_re_ucontext *uctx = NULL;
struct bnxt_re_resize_cq_req req;
struct bnxt_re_dev *rdev;
struct bnxt_re_cq *cq;
int rc, entries;
cq = container_of(ibcq, struct bnxt_re_cq, ib_cq);
rdev = cq->rdev;
dev_attr = &rdev->dev_attr;
if (!ibcq->uobject) {
ibdev_err(&rdev->ibdev, "Kernel CQ Resize not supported");
return -EOPNOTSUPP;
}
if (cq->resize_umem) {
ibdev_err(&rdev->ibdev, "Resize CQ %#x failed - Busy",
cq->qplib_cq.id);
return -EBUSY;
}
/* Check the requested cq depth out of supported depth */
if (cqe < 1 || cqe > dev_attr->max_cq_wqes) {
ibdev_err(&rdev->ibdev, "Resize CQ %#x failed - out of range cqe %d",
cq->qplib_cq.id, cqe);
return -EINVAL;
}
entries = roundup_pow_of_two(cqe + 1);
if (entries > dev_attr->max_cq_wqes + 1)
entries = dev_attr->max_cq_wqes + 1;
uctx = rdma_udata_to_drv_context(udata, struct bnxt_re_ucontext,
ib_uctx);
/* uverbs consumer */
if (ib_copy_from_udata(&req, udata, sizeof(req))) {
rc = -EFAULT;
goto fail;
}
cq->resize_umem = ib_umem_get(&rdev->ibdev, req.cq_va,
entries * sizeof(struct cq_base),
IB_ACCESS_LOCAL_WRITE);
if (IS_ERR(cq->resize_umem)) {
rc = PTR_ERR(cq->resize_umem);
cq->resize_umem = NULL;
ibdev_err(&rdev->ibdev, "%s: ib_umem_get failed! rc = %d\n",
__func__, rc);
goto fail;
}
cq->resize_cqe = entries;
memcpy(&sg_info, &cq->qplib_cq.sg_info, sizeof(sg_info));
orig_dpi = cq->qplib_cq.dpi;
cq->qplib_cq.sg_info.umem = cq->resize_umem;
cq->qplib_cq.sg_info.pgsize = PAGE_SIZE;
cq->qplib_cq.sg_info.pgshft = PAGE_SHIFT;
cq->qplib_cq.dpi = &uctx->dpi;
rc = bnxt_qplib_resize_cq(&rdev->qplib_res, &cq->qplib_cq, entries);
if (rc) {
ibdev_err(&rdev->ibdev, "Resize HW CQ %#x failed!",
cq->qplib_cq.id);
goto fail;
}
cq->ib_cq.cqe = cq->resize_cqe;
return 0;
fail:
if (cq->resize_umem) {
ib_umem_release(cq->resize_umem);
cq->resize_umem = NULL;
cq->resize_cqe = 0;
memcpy(&cq->qplib_cq.sg_info, &sg_info, sizeof(sg_info));
cq->qplib_cq.dpi = orig_dpi;
}
return rc;
}
static u8 __req_to_ib_wc_status(u8 qstatus)
{
switch (qstatus) {
@ -3425,6 +3525,15 @@ int bnxt_re_poll_cq(struct ib_cq *ib_cq, int num_entries, struct ib_wc *wc)
struct bnxt_re_sqp_entries *sqp_entry = NULL;
unsigned long flags;
/* User CQ; the only processing we do is to
* complete any pending CQ resize operation.
*/
if (cq->umem) {
if (cq->resize_umem)
bnxt_re_resize_cq_complete(cq);
return 0;
}
spin_lock_irqsave(&cq->cq_lock, flags);
budget = min_t(u32, num_entries, cq->max_cql);
num_entries = budget;

3
drivers/infiniband/hw/bnxt_re/ib_verbs.h
View File

@ -104,6 +104,8 @@ struct bnxt_re_cq {
#define MAX_CQL_PER_POLL 1024
u32 max_cql;
struct ib_umem *umem;
struct ib_umem *resize_umem;
int resize_cqe;
};
struct bnxt_re_mr {
@ -191,6 +193,7 @@ int bnxt_re_post_recv(struct ib_qp *qp, const struct ib_recv_wr *recv_wr,
const struct ib_recv_wr **bad_recv_wr);
int bnxt_re_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
struct ib_udata *udata);
int bnxt_re_resize_cq(struct ib_cq *ibcq, int cqe, struct ib_udata *udata);
int bnxt_re_destroy_cq(struct ib_cq *cq, struct ib_udata *udata);
int bnxt_re_poll_cq(struct ib_cq *cq, int num_entries, struct ib_wc *wc);
int bnxt_re_req_notify_cq(struct ib_cq *cq, enum ib_cq_notify_flags flags);

103
drivers/infiniband/hw/bnxt_re/main.c
View File

@ -553,6 +553,7 @@ static const struct ib_device_ops bnxt_re_dev_ops = {
.query_srq = bnxt_re_query_srq,
.reg_user_mr = bnxt_re_reg_user_mr,
.req_notify_cq = bnxt_re_req_notify_cq,
.resize_cq = bnxt_re_resize_cq,
INIT_RDMA_OBJ_SIZE(ib_ah, bnxt_re_ah, ib_ah),
INIT_RDMA_OBJ_SIZE(ib_cq, bnxt_re_cq, ib_cq),
INIT_RDMA_OBJ_SIZE(ib_pd, bnxt_re_pd, ib_pd),
@ -584,6 +585,7 @@ static int bnxt_re_register_ib(struct bnxt_re_dev *rdev)
return ret;
dma_set_max_seg_size(&rdev->en_dev->pdev->dev, UINT_MAX);
ibdev->uverbs_cmd_mask |= BIT_ULL(IB_USER_VERBS_CMD_POLL_CQ);
return ib_register_device(ibdev, "bnxt_re%d", &rdev->en_dev->pdev->dev);
}
@ -919,49 +921,6 @@ static void bnxt_re_dispatch_event(struct ib_device *ibdev, struct ib_qp *qp,
}
}
#define HWRM_QUEUE_PRI2COS_QCFG_INPUT_FLAGS_IVLAN 0x02
static int bnxt_re_query_hwrm_pri2cos(struct bnxt_re_dev *rdev, u8 dir,
u64 *cid_map)
{
struct hwrm_queue_pri2cos_qcfg_input req = {0};
struct hwrm_queue_pri2cos_qcfg_output resp;
struct bnxt_en_dev *en_dev = rdev->en_dev;
struct bnxt_fw_msg fw_msg;
u32 flags = 0;
u8 *qcfgmap, *tmp_map;
int rc = 0, i;
if (!cid_map)
return -EINVAL;
memset(&fw_msg, 0, sizeof(fw_msg));
bnxt_re_init_hwrm_hdr(rdev, (void *)&req,
HWRM_QUEUE_PRI2COS_QCFG, -1, -1);
flags |= (dir & 0x01);
flags |= HWRM_QUEUE_PRI2COS_QCFG_INPUT_FLAGS_IVLAN;
req.flags = cpu_to_le32(flags);
req.port_id = en_dev->pf_port_id;
bnxt_re_fill_fw_msg(&fw_msg, (void *)&req, sizeof(req), (void *)&resp,
sizeof(resp), DFLT_HWRM_CMD_TIMEOUT);
rc = bnxt_send_msg(en_dev, &fw_msg);
if (rc)
return rc;
if (resp.queue_cfg_info) {
ibdev_warn(&rdev->ibdev,
"Asymmetric cos queue configuration detected");
ibdev_warn(&rdev->ibdev,
" on device, QoS may not be fully functional\n");
}
qcfgmap = &resp.pri0_cos_queue_id;
tmp_map = (u8 *)cid_map;
for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
tmp_map[i] = qcfgmap[i];
return rc;
}
static bool bnxt_re_is_qp1_or_shadow_qp(struct bnxt_re_dev *rdev,
struct bnxt_re_qp *qp)
{
@ -1054,26 +1013,9 @@ static u32 bnxt_re_get_priority_mask(struct bnxt_re_dev *rdev)
return prio_map;
}
static void bnxt_re_parse_cid_map(u8 prio_map, u8 *cid_map, u16 *cosq)
{
u16 prio;
u8 id;
for (prio = 0, id = 0; prio < 8; prio++) {
if (prio_map & (1 << prio)) {
cosq[id] = cid_map[prio];
id++;
if (id == 2) /* Max 2 tcs supported */
break;
}
}
}
static int bnxt_re_setup_qos(struct bnxt_re_dev *rdev)
{
u8 prio_map = 0;
u64 cid_map;
int rc;
/* Get priority for roce */
prio_map = bnxt_re_get_priority_mask(rdev);
@ -1081,23 +1023,6 @@ static int bnxt_re_setup_qos(struct bnxt_re_dev *rdev)
if (prio_map == rdev->cur_prio_map)
return 0;
rdev->cur_prio_map = prio_map;
/* Get cosq id for this priority */
rc = bnxt_re_query_hwrm_pri2cos(rdev, 0, &cid_map);
if (rc) {
ibdev_warn(&rdev->ibdev, "no cos for p_mask %x\n", prio_map);
return rc;
}
/* Parse CoS IDs for app priority */
bnxt_re_parse_cid_map(prio_map, (u8 *)&cid_map, rdev->cosq);
/* Config BONO. */
rc = bnxt_qplib_map_tc2cos(&rdev->qplib_res, rdev->cosq);
if (rc) {
ibdev_warn(&rdev->ibdev, "no tc for cos{%x, %x}\n",
rdev->cosq[0], rdev->cosq[1]);
return rc;
}
/* Actual priorities are not programmed as they are already
* done by L2 driver; just enable or disable priority vlan tagging
*/
@ -1407,6 +1332,27 @@ exit:
return rc;
}
static void bnxt_re_setup_cc(struct bnxt_re_dev *rdev, bool enable)
{
struct bnxt_qplib_cc_param cc_param = {};
/* Currently enabling only for GenP5 adapters */
if (!bnxt_qplib_is_chip_gen_p5(rdev->chip_ctx))
return;
if (enable) {
cc_param.enable = 1;
cc_param.cc_mode = CMDQ_MODIFY_ROCE_CC_CC_MODE_PROBABILISTIC_CC_MODE;
}
cc_param.mask = (CMDQ_MODIFY_ROCE_CC_MODIFY_MASK_CC_MODE |
CMDQ_MODIFY_ROCE_CC_MODIFY_MASK_ENABLE_CC |
CMDQ_MODIFY_ROCE_CC_MODIFY_MASK_TOS_ECN);
if (bnxt_qplib_modify_cc(&rdev->qplib_res, &cc_param))
ibdev_err(&rdev->ibdev, "Failed to setup CC enable = %d\n", enable);
}
/*
* "Notifier chain callback can be invoked for the same chain from
* different CPUs at the same time".
@ -1475,7 +1421,7 @@ static void bnxt_re_remove(struct auxiliary_device *adev)
*/
goto skip_remove;
}
bnxt_re_setup_cc(rdev, false);
ib_unregister_device(&rdev->ibdev);
ib_dealloc_device(&rdev->ibdev);
bnxt_re_dev_uninit(rdev);
@ -1507,6 +1453,7 @@ static int bnxt_re_probe(struct auxiliary_device *adev,
goto err;
}
bnxt_re_setup_cc(rdev, true);
mutex_unlock(&bnxt_re_mutex);
return 0;

211
drivers/infiniband/hw/bnxt_re/qplib_fp.c
View File

@ -300,8 +300,6 @@ static void bnxt_qplib_service_nq(struct tasklet_struct *t)
{
struct bnxt_qplib_nq *nq = from_tasklet(nq, t, nq_tasklet);
struct bnxt_qplib_hwq *hwq = &nq->hwq;
int num_srqne_processed = 0;
int num_cqne_processed = 0;
struct bnxt_qplib_cq *cq;
int budget = nq->budget;
u32 sw_cons, raw_cons;
@ -340,9 +338,7 @@ static void bnxt_qplib_service_nq(struct tasklet_struct *t)
DBC_DBC_TYPE_CQ_ARMENA);
spin_lock_bh(&cq->compl_lock);
atomic_set(&cq->arm_state, 0);
if (!nq->cqn_handler(nq, (cq)))
num_cqne_processed++;
else
if (nq->cqn_handler(nq, (cq)))
dev_warn(&nq->pdev->dev,
"cqn - type 0x%x not handled\n", type);
cq->cnq_events++;
@ -361,11 +357,9 @@ static void bnxt_qplib_service_nq(struct tasklet_struct *t)
srq = (struct bnxt_qplib_srq *)q_handle;
bnxt_qplib_armen_db(&srq->dbinfo,
DBC_DBC_TYPE_SRQ_ARMENA);
if (!nq->srqn_handler(nq,
(struct bnxt_qplib_srq *)q_handle,
nqsrqe->event))
num_srqne_processed++;
else
if (nq->srqn_handler(nq,
(struct bnxt_qplib_srq *)q_handle,
nqsrqe->event))
dev_warn(&nq->pdev->dev,
"SRQ event 0x%x not handled\n",
nqsrqe->event);
@ -581,18 +575,20 @@ void bnxt_qplib_destroy_srq(struct bnxt_qplib_res *res,
struct bnxt_qplib_srq *srq)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_destroy_srq req;
struct creq_destroy_srq_resp resp;
u16 cmd_flags = 0;
struct creq_destroy_srq_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_destroy_srq req = {};
int rc;
RCFW_CMD_PREP(req, DESTROY_SRQ, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_DESTROY_SRQ,
sizeof(req));
/* Configure the request */
req.srq_cid = cpu_to_le32(srq->id);
rc = bnxt_qplib_rcfw_send_message(rcfw, (struct cmdq_base *)&req,
(struct creq_base *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req), sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
kfree(srq->swq);
if (rc)
return;
@ -604,10 +600,10 @@ int bnxt_qplib_create_srq(struct bnxt_qplib_res *res,
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct bnxt_qplib_hwq_attr hwq_attr = {};
struct creq_create_srq_resp resp;
struct cmdq_create_srq req;
struct creq_create_srq_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_create_srq req = {};
struct bnxt_qplib_pbl *pbl;
u16 cmd_flags = 0;
u16 pg_sz_lvl;
int rc, idx;
@ -627,7 +623,9 @@ int bnxt_qplib_create_srq(struct bnxt_qplib_res *res,
goto fail;
}
RCFW_CMD_PREP(req, CREATE_SRQ, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_CREATE_SRQ,
sizeof(req));
/* Configure the request */
req.dpi = cpu_to_le32(srq->dpi->dpi);
@ -644,8 +642,8 @@ int bnxt_qplib_create_srq(struct bnxt_qplib_res *res,
req.pd_id = cpu_to_le32(srq->pd->id);
req.eventq_id = cpu_to_le16(srq->eventq_hw_ring_id);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req), sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
goto fail;
@ -700,14 +698,16 @@ int bnxt_qplib_query_srq(struct bnxt_qplib_res *res,
struct bnxt_qplib_srq *srq)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_query_srq req;
struct creq_query_srq_resp resp;
struct creq_query_srq_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct bnxt_qplib_rcfw_sbuf *sbuf;
struct creq_query_srq_resp_sb *sb;
u16 cmd_flags = 0;
struct cmdq_query_srq req = {};
int rc = 0;
RCFW_CMD_PREP(req, QUERY_SRQ, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_QUERY_SRQ,
sizeof(req));
/* Configure the request */
sbuf = bnxt_qplib_rcfw_alloc_sbuf(rcfw, sizeof(*sb));
@ -716,8 +716,9 @@ int bnxt_qplib_query_srq(struct bnxt_qplib_res *res,
req.resp_size = sizeof(*sb) / BNXT_QPLIB_CMDQE_UNITS;
req.srq_cid = cpu_to_le32(srq->id);
sb = sbuf->sb;
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp,
(void *)sbuf, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, sbuf, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
srq->threshold = le16_to_cpu(sb->srq_limit);
bnxt_qplib_rcfw_free_sbuf(rcfw, sbuf);
@ -811,19 +812,20 @@ int bnxt_qplib_create_qp1(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
{
struct bnxt_qplib_hwq_attr hwq_attr = {};
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct creq_create_qp1_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct bnxt_qplib_q *sq = &qp->sq;
struct bnxt_qplib_q *rq = &qp->rq;
struct creq_create_qp1_resp resp;
struct cmdq_create_qp1 req;
struct cmdq_create_qp1 req = {};
struct bnxt_qplib_pbl *pbl;
u16 cmd_flags = 0;
u32 qp_flags = 0;
u8 pg_sz_lvl;
u32 tbl_indx;
int rc;
RCFW_CMD_PREP(req, CREATE_QP1, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_CREATE_QP1,
sizeof(req));
/* General */
req.type = qp->type;
req.dpi = cpu_to_le32(qp->dpi->dpi);
@ -891,8 +893,8 @@ int bnxt_qplib_create_qp1(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
req.qp_flags = cpu_to_le32(qp_flags);
req.pd_id = cpu_to_le32(qp->pd->id);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req), sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
goto fail;
@ -952,20 +954,22 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct bnxt_qplib_hwq_attr hwq_attr = {};
struct bnxt_qplib_sg_info sginfo = {};
struct creq_create_qp_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct bnxt_qplib_q *sq = &qp->sq;
struct bnxt_qplib_q *rq = &qp->rq;
struct creq_create_qp_resp resp;
struct cmdq_create_qp req = {};
int rc, req_size, psn_sz = 0;
struct bnxt_qplib_hwq *xrrq;
struct bnxt_qplib_pbl *pbl;
struct cmdq_create_qp req;
u16 cmd_flags = 0;
u32 qp_flags = 0;
u8 pg_sz_lvl;
u32 tbl_indx;
u16 nsge;
RCFW_CMD_PREP(req, CREATE_QP, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_CREATE_QP,
sizeof(req));
/* General */
req.type = qp->type;
@ -1098,8 +1102,9 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
}
req.pd_id = cpu_to_le32(qp->pd->id);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
goto fail;
@ -1231,14 +1236,16 @@ static void __filter_modify_flags(struct bnxt_qplib_qp *qp)
int bnxt_qplib_modify_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_modify_qp req;
struct creq_modify_qp_resp resp;
u16 cmd_flags = 0;
struct creq_modify_qp_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_modify_qp req = {};
u32 temp32[4];
u32 bmask;
int rc;
RCFW_CMD_PREP(req, MODIFY_QP, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_MODIFY_QP,
sizeof(req));
/* Filter out the qp_attr_mask based on the state->new transition */
__filter_modify_flags(qp);
@ -1286,7 +1293,7 @@ int bnxt_qplib_modify_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
memcpy(req.dest_mac, qp->ah.dmac, 6);
if (bmask & CMDQ_MODIFY_QP_MODIFY_MASK_PATH_MTU)
req.path_mtu = qp->path_mtu;
req.path_mtu_pingpong_push_enable |= qp->path_mtu;
if (bmask & CMDQ_MODIFY_QP_MODIFY_MASK_TIMEOUT)
req.timeout = qp->timeout;
@ -1324,8 +1331,8 @@ int bnxt_qplib_modify_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
req.vlan_pcp_vlan_dei_vlan_id = cpu_to_le16(qp->vlan_id);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req), sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
return rc;
qp->cur_qp_state = qp->state;
@ -1335,15 +1342,17 @@ int bnxt_qplib_modify_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
int bnxt_qplib_query_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_query_qp req;
struct creq_query_qp_resp resp;
struct creq_query_qp_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct bnxt_qplib_rcfw_sbuf *sbuf;
struct creq_query_qp_resp_sb *sb;
u16 cmd_flags = 0;
struct cmdq_query_qp req = {};
u32 temp32[4];
int i, rc = 0;
RCFW_CMD_PREP(req, QUERY_QP, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_QUERY_QP,
sizeof(req));
sbuf = bnxt_qplib_rcfw_alloc_sbuf(rcfw, sizeof(*sb));
if (!sbuf)
@ -1352,8 +1361,9 @@ int bnxt_qplib_query_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
req.qp_cid = cpu_to_le32(qp->id);
req.resp_size = sizeof(*sb) / BNXT_QPLIB_CMDQE_UNITS;
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp,
(void *)sbuf, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, sbuf, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
goto bail;
/* Extract the context from the side buffer */
@ -1460,9 +1470,9 @@ int bnxt_qplib_destroy_qp(struct bnxt_qplib_res *res,
struct bnxt_qplib_qp *qp)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_destroy_qp req;
struct creq_destroy_qp_resp resp;
u16 cmd_flags = 0;
struct creq_destroy_qp_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_destroy_qp req = {};
u32 tbl_indx;
int rc;
@ -1470,11 +1480,14 @@ int bnxt_qplib_destroy_qp(struct bnxt_qplib_res *res,
rcfw->qp_tbl[tbl_indx].qp_id = BNXT_QPLIB_QP_ID_INVALID;
rcfw->qp_tbl[tbl_indx].qp_handle = NULL;
RCFW_CMD_PREP(req, DESTROY_QP, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_DESTROY_QP,
sizeof(req));
req.qp_cid = cpu_to_le32(qp->id);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc) {
rcfw->qp_tbl[tbl_indx].qp_id = qp->id;
rcfw->qp_tbl[tbl_indx].qp_handle = qp;
@ -2036,10 +2049,10 @@ int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct bnxt_qplib_hwq_attr hwq_attr = {};
struct creq_create_cq_resp resp;
struct creq_create_cq_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_create_cq req = {};
struct bnxt_qplib_pbl *pbl;
struct cmdq_create_cq req;
u16 cmd_flags = 0;
u32 pg_sz_lvl;
int rc;
@ -2052,7 +2065,9 @@ int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq)
if (rc)
goto exit;
RCFW_CMD_PREP(req, CREATE_CQ, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_CREATE_CQ,
sizeof(req));
if (!cq->dpi) {
dev_err(&rcfw->pdev->dev,
@ -2071,9 +2086,9 @@ int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq)
req.cq_fco_cnq_id = cpu_to_le32(
(cq->cnq_hw_ring_id & CMDQ_CREATE_CQ_CNQ_ID_MASK) <<
CMDQ_CREATE_CQ_CNQ_ID_SFT);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
goto fail;
@ -2100,20 +2115,70 @@ exit:
return rc;
}
void bnxt_qplib_resize_cq_complete(struct bnxt_qplib_res *res,
struct bnxt_qplib_cq *cq)
{
bnxt_qplib_free_hwq(res, &cq->hwq);
memcpy(&cq->hwq, &cq->resize_hwq, sizeof(cq->hwq));
}
int bnxt_qplib_resize_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq,
int new_cqes)
{
struct bnxt_qplib_hwq_attr hwq_attr = {};
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct creq_resize_cq_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_resize_cq req = {};
struct bnxt_qplib_pbl *pbl;
u32 pg_sz, lvl, new_sz;
int rc;
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_RESIZE_CQ,
sizeof(req));
hwq_attr.sginfo = &cq->sg_info;
hwq_attr.res = res;
hwq_attr.depth = new_cqes;
hwq_attr.stride = sizeof(struct cq_base);
hwq_attr.type = HWQ_TYPE_QUEUE;
rc = bnxt_qplib_alloc_init_hwq(&cq->resize_hwq, &hwq_attr);
if (rc)
return rc;
req.cq_cid = cpu_to_le32(cq->id);
pbl = &cq->resize_hwq.pbl[PBL_LVL_0];
pg_sz = bnxt_qplib_base_pg_size(&cq->resize_hwq);
lvl = (cq->resize_hwq.level << CMDQ_RESIZE_CQ_LVL_SFT) &
CMDQ_RESIZE_CQ_LVL_MASK;
new_sz = (new_cqes << CMDQ_RESIZE_CQ_NEW_CQ_SIZE_SFT) &
CMDQ_RESIZE_CQ_NEW_CQ_SIZE_MASK;
req.new_cq_size_pg_size_lvl = cpu_to_le32(new_sz | pg_sz | lvl);
req.new_pbl = cpu_to_le64(pbl->pg_map_arr[0]);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
return rc;
}
int bnxt_qplib_destroy_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_destroy_cq req;
struct creq_destroy_cq_resp resp;
struct creq_destroy_cq_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_destroy_cq req = {};
u16 total_cnq_events;
u16 cmd_flags = 0;
int rc;
RCFW_CMD_PREP(req, DESTROY_CQ, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_DESTROY_CQ,
sizeof(req));
req.cq_cid = cpu_to_le32(cq->id);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
return rc;
total_cnq_events = le16_to_cpu(resp.total_cnq_events);

5
drivers/infiniband/hw/bnxt_re/qplib_fp.h
View File

@ -400,6 +400,7 @@ struct bnxt_qplib_cq {
u16 count;
u16 period;
struct bnxt_qplib_hwq hwq;
struct bnxt_qplib_hwq resize_hwq;
u32 cnq_hw_ring_id;
struct bnxt_qplib_nq *nq;
bool resize_in_progress;
@ -532,6 +533,10 @@ void bnxt_qplib_post_recv_db(struct bnxt_qplib_qp *qp);
int bnxt_qplib_post_recv(struct bnxt_qplib_qp *qp,
struct bnxt_qplib_swqe *wqe);
int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq);
int bnxt_qplib_resize_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq,
int new_cqes);
void bnxt_qplib_resize_cq_complete(struct bnxt_qplib_res *res,
struct bnxt_qplib_cq *cq);
int bnxt_qplib_destroy_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq);
int bnxt_qplib_poll_cq(struct bnxt_qplib_cq *cq, struct bnxt_qplib_cqe *cqe,
int num, struct bnxt_qplib_qp **qp);

95
drivers/infiniband/hw/bnxt_re/qplib_rcfw.c
View File

@ -49,6 +49,7 @@
#include "qplib_rcfw.h"
#include "qplib_sp.h"
#include "qplib_fp.h"
#include "qplib_tlv.h"
static void bnxt_qplib_service_creq(struct tasklet_struct *t);
@ -85,8 +86,8 @@ done:
return count ? 0 : -ETIMEDOUT;
};
static int __send_message(struct bnxt_qplib_rcfw *rcfw, struct cmdq_base *req,
struct creq_base *resp, void *sb, u8 is_block)
static int __send_message(struct bnxt_qplib_rcfw *rcfw,
struct bnxt_qplib_cmdqmsg *msg)
{
struct bnxt_qplib_cmdq_ctx *cmdq = &rcfw->cmdq;
struct bnxt_qplib_hwq *hwq = &cmdq->hwq;
@ -95,13 +96,13 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, struct cmdq_base *req,
u32 sw_prod, cmdq_prod;
struct pci_dev *pdev;
unsigned long flags;
u32 size, opcode;
u32 bsize, opcode;
u16 cookie, cbit;
u8 *preq;
pdev = rcfw->pdev;
opcode = req->opcode;
opcode = __get_cmdq_base_opcode(msg->req, msg->req_sz);
if (!test_bit(FIRMWARE_INITIALIZED_FLAG, &cmdq->flags) &&
(opcode != CMDQ_BASE_OPCODE_QUERY_FUNC &&
opcode != CMDQ_BASE_OPCODE_INITIALIZE_FW &&
@ -124,7 +125,7 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, struct cmdq_base *req,
* cmdqe
*/
spin_lock_irqsave(&hwq->lock, flags);
if (req->cmd_size >= HWQ_FREE_SLOTS(hwq)) {
if (msg->req->cmd_size >= HWQ_FREE_SLOTS(hwq)) {
dev_err(&pdev->dev, "RCFW: CMDQ is full!\n");
spin_unlock_irqrestore(&hwq->lock, flags);
return -EAGAIN;
@ -133,36 +134,34 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, struct cmdq_base *req,
cookie = cmdq->seq_num & RCFW_MAX_COOKIE_VALUE;
cbit = cookie % rcfw->cmdq_depth;
if (is_block)
if (msg->block)
cookie |= RCFW_CMD_IS_BLOCKING;
set_bit(cbit, cmdq->cmdq_bitmap);
req->cookie = cpu_to_le16(cookie);
__set_cmdq_base_cookie(msg->req, msg->req_sz, cpu_to_le16(cookie));
crsqe = &rcfw->crsqe_tbl[cbit];
if (crsqe->resp) {
spin_unlock_irqrestore(&hwq->lock, flags);
return -EBUSY;
}
size = req->cmd_size;
/* change the cmd_size to the number of 16byte cmdq unit.
* req->cmd_size is modified here
*/
bnxt_qplib_set_cmd_slots(req);
bsize = bnxt_qplib_set_cmd_slots(msg->req);
memset(resp, 0, sizeof(*resp));
crsqe->resp = (struct creq_qp_event *)resp;
crsqe->resp->cookie = req->cookie;
crsqe->req_size = req->cmd_size;
if (req->resp_size && sb) {
struct bnxt_qplib_rcfw_sbuf *sbuf = sb;
req->resp_addr = cpu_to_le64(sbuf->dma_addr);
req->resp_size = (sbuf->size + BNXT_QPLIB_CMDQE_UNITS - 1) /
BNXT_QPLIB_CMDQE_UNITS;
memset(msg->resp, 0, sizeof(*msg->resp));
crsqe->resp = (struct creq_qp_event *)msg->resp;
crsqe->resp->cookie = cpu_to_le16(cookie);
crsqe->req_size = __get_cmdq_base_cmd_size(msg->req, msg->req_sz);
if (__get_cmdq_base_resp_size(msg->req, msg->req_sz) && msg->sb) {
struct bnxt_qplib_rcfw_sbuf *sbuf = msg->sb;
__set_cmdq_base_resp_addr(msg->req, msg->req_sz, cpu_to_le64(sbuf->dma_addr));
__set_cmdq_base_resp_size(msg->req, msg->req_sz,
ALIGN(sbuf->size, BNXT_QPLIB_CMDQE_UNITS));
}
preq = (u8 *)req;
preq = (u8 *)msg->req;
do {
/* Locate the next cmdq slot */
sw_prod = HWQ_CMP(hwq->prod, hwq);
@ -174,11 +173,11 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, struct cmdq_base *req,
}
/* Copy a segment of the req cmd to the cmdq */
memset(cmdqe, 0, sizeof(*cmdqe));
memcpy(cmdqe, preq, min_t(u32, size, sizeof(*cmdqe)));
preq += min_t(u32, size, sizeof(*cmdqe));
size -= min_t(u32, size, sizeof(*cmdqe));
memcpy(cmdqe, preq, min_t(u32, bsize, sizeof(*cmdqe)));
preq += min_t(u32, bsize, sizeof(*cmdqe));
bsize -= min_t(u32, bsize, sizeof(*cmdqe));
hwq->prod++;
} while (size > 0);
} while (bsize > 0);
cmdq->seq_num++;
cmdq_prod = hwq->prod;
@ -191,7 +190,6 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, struct cmdq_base *req,
cmdq_prod |= BIT(FIRMWARE_FIRST_FLAG);
clear_bit(FIRMWARE_FIRST_FLAG, &cmdq->flags);
}
/* ring CMDQ DB */
wmb();
writel(cmdq_prod, cmdq->cmdq_mbox.prod);
@ -203,11 +201,9 @@ done:
}
int bnxt_qplib_rcfw_send_message(struct bnxt_qplib_rcfw *rcfw,
struct cmdq_base *req,
struct creq_base *resp,
void *sb, u8 is_block)
struct bnxt_qplib_cmdqmsg *msg)
{
struct creq_qp_event *evnt = (struct creq_qp_event *)resp;
struct creq_qp_event *evnt = (struct creq_qp_event *)msg->resp;
u16 cookie;
u8 opcode, retry_cnt = 0xFF;
int rc = 0;
@ -217,23 +213,23 @@ int bnxt_qplib_rcfw_send_message(struct bnxt_qplib_rcfw *rcfw,
return 0;
do {
opcode = req->opcode;
rc = __send_message(rcfw, req, resp, sb, is_block);
cookie = le16_to_cpu(req->cookie) & RCFW_MAX_COOKIE_VALUE;
opcode = __get_cmdq_base_opcode(msg->req, msg->req_sz);
rc = __send_message(rcfw, msg);
cookie = le16_to_cpu(__get_cmdq_base_cookie(msg->req, msg->req_sz)) &
RCFW_MAX_COOKIE_VALUE;
if (!rc)
break;
if (!retry_cnt || (rc != -EAGAIN && rc != -EBUSY)) {
/* send failed */
dev_err(&rcfw->pdev->dev, "cmdq[%#x]=%#x send failed\n",
cookie, opcode);
return rc;
}
is_block ? mdelay(1) : usleep_range(500, 1000);
msg->block ? mdelay(1) : usleep_range(500, 1000);
} while (retry_cnt--);
if (is_block)
if (msg->block)
rc = __block_for_resp(rcfw, cookie);
else
rc = __wait_for_resp(rcfw, cookie);
@ -452,14 +448,17 @@ static irqreturn_t bnxt_qplib_creq_irq(int irq, void *dev_instance)
/* RCFW */
int bnxt_qplib_deinit_rcfw(struct bnxt_qplib_rcfw *rcfw)
{
struct cmdq_deinitialize_fw req;
struct creq_deinitialize_fw_resp resp;
u16 cmd_flags = 0;
struct creq_deinitialize_fw_resp resp = {};
struct cmdq_deinitialize_fw req = {};
struct bnxt_qplib_cmdqmsg msg = {};
int rc;
RCFW_CMD_PREP(req, DEINITIALIZE_FW, cmd_flags);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp,
NULL, 0);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_DEINITIALIZE_FW,
sizeof(req));
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL,
sizeof(req), sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
return rc;
@ -470,13 +469,15 @@ int bnxt_qplib_deinit_rcfw(struct bnxt_qplib_rcfw *rcfw)
int bnxt_qplib_init_rcfw(struct bnxt_qplib_rcfw *rcfw,
struct bnxt_qplib_ctx *ctx, int is_virtfn)
{
struct creq_initialize_fw_resp resp;
struct cmdq_initialize_fw req;
u16 cmd_flags = 0;
struct creq_initialize_fw_resp resp = {};
struct cmdq_initialize_fw req = {};
struct bnxt_qplib_cmdqmsg msg = {};
u8 pgsz, lvl;
int rc;
RCFW_CMD_PREP(req, INITIALIZE_FW, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_INITIALIZE_FW,
sizeof(req));
/* Supply (log-base-2-of-host-page-size - base-page-shift)
* to bono to adjust the doorbell page sizes.
*/
@ -545,8 +546,8 @@ config_vf_res:
skip_ctx_setup:
req.stat_ctx_id = cpu_to_le32(ctx->stats.fw_id);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp,
NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req), sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
return rc;
set_bit(FIRMWARE_INITIALIZED_FLAG, &rcfw->cmdq.flags);

66
drivers/infiniband/hw/bnxt_re/qplib_rcfw.h
View File

@ -39,6 +39,8 @@
#ifndef __BNXT_QPLIB_RCFW_H__
#define __BNXT_QPLIB_RCFW_H__
#include "qplib_tlv.h"
#define RCFW_CMDQ_TRIG_VAL 1
#define RCFW_COMM_PCI_BAR_REGION 0
#define RCFW_COMM_CONS_PCI_BAR_REGION 2
@ -51,25 +53,25 @@
#define RCFW_DBR_PCI_BAR_REGION 2
#define RCFW_DBR_BASE_PAGE_SHIFT 12
#define RCFW_CMD_PREP(req, CMD, cmd_flags) \
do { \
memset(&(req), 0, sizeof((req))); \
(req).opcode = CMDQ_BASE_OPCODE_##CMD; \
(req).cmd_size = sizeof((req)); \
(req).flags = cpu_to_le16(cmd_flags); \
} while (0)
#define RCFW_CMD_WAIT_TIME_MS 20000 /* 20 Seconds timeout */
/* Cmdq contains a fix number of a 16-Byte slots */
struct bnxt_qplib_cmdqe {
u8 data[16];
};
#define BNXT_QPLIB_CMDQE_UNITS sizeof(struct bnxt_qplib_cmdqe)
static inline void bnxt_qplib_rcfw_cmd_prep(struct cmdq_base *req,
u8 opcode, u8 cmd_size)
{
req->opcode = opcode;
req->cmd_size = cmd_size;
}
#define RCFW_CMD_WAIT_TIME_MS 20000 /* 20 Seconds timeout */
/* CMDQ elements */
#define BNXT_QPLIB_CMDQE_MAX_CNT_256 256
#define BNXT_QPLIB_CMDQE_MAX_CNT_8192 8192
#define BNXT_QPLIB_CMDQE_UNITS sizeof(struct bnxt_qplib_cmdqe)
#define BNXT_QPLIB_CMDQE_BYTES(depth) ((depth) * BNXT_QPLIB_CMDQE_UNITS)
static inline u32 bnxt_qplib_cmdqe_npages(u32 depth)
@ -87,11 +89,21 @@ static inline u32 bnxt_qplib_cmdqe_page_size(u32 depth)
return (bnxt_qplib_cmdqe_npages(depth) * PAGE_SIZE);
}
/* Set the cmd_size to a factor of CMDQE unit */
static inline void bnxt_qplib_set_cmd_slots(struct cmdq_base *req)
static inline u32 bnxt_qplib_set_cmd_slots(struct cmdq_base *req)
{
req->cmd_size = (req->cmd_size + BNXT_QPLIB_CMDQE_UNITS - 1) /
BNXT_QPLIB_CMDQE_UNITS;
u32 cmd_byte = 0;
if (HAS_TLV_HEADER(req)) {
struct roce_tlv *tlv_req = (struct roce_tlv *)req;
cmd_byte = tlv_req->total_size * BNXT_QPLIB_CMDQE_UNITS;
} else {
cmd_byte = req->cmd_size;
req->cmd_size = (req->cmd_size + BNXT_QPLIB_CMDQE_UNITS - 1) /
BNXT_QPLIB_CMDQE_UNITS;
}
return cmd_byte;
}
#define RCFW_MAX_COOKIE_VALUE 0x7FFF
@ -190,6 +202,27 @@ struct bnxt_qplib_rcfw {
u32 cmdq_depth;
};
struct bnxt_qplib_cmdqmsg {
struct cmdq_base *req;
struct creq_base *resp;
void *sb;
u32 req_sz;
u32 res_sz;
u8 block;
};
static inline void bnxt_qplib_fill_cmdqmsg(struct bnxt_qplib_cmdqmsg *msg,
void *req, void *resp, void *sb,
u32 req_sz, u32 res_sz, u8 block)
{
msg->req = req;
msg->resp = resp;
msg->sb = sb;
msg->req_sz = req_sz;
msg->res_sz = res_sz;
msg->block = block;
}
void bnxt_qplib_free_rcfw_channel(struct bnxt_qplib_rcfw *rcfw);
int bnxt_qplib_alloc_rcfw_channel(struct bnxt_qplib_res *res,
struct bnxt_qplib_rcfw *rcfw,
@ -210,8 +243,7 @@ struct bnxt_qplib_rcfw_sbuf *bnxt_qplib_rcfw_alloc_sbuf(
void bnxt_qplib_rcfw_free_sbuf(struct bnxt_qplib_rcfw *rcfw,
struct bnxt_qplib_rcfw_sbuf *sbuf);
int bnxt_qplib_rcfw_send_message(struct bnxt_qplib_rcfw *rcfw,
struct cmdq_base *req, struct creq_base *resp,
void *sbuf, u8 is_block);
struct bnxt_qplib_cmdqmsg *msg);
int bnxt_qplib_deinit_rcfw(struct bnxt_qplib_rcfw *rcfw);
int bnxt_qplib_init_rcfw(struct bnxt_qplib_rcfw *rcfw,

337
drivers/infiniband/hw/bnxt_re/qplib_sp.c
View File

@ -48,6 +48,7 @@
#include "qplib_res.h"
#include "qplib_rcfw.h"
#include "qplib_sp.h"
#include "qplib_tlv.h"
const struct bnxt_qplib_gid bnxt_qplib_gid_zero = {{ 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0 } };
@ -68,15 +69,17 @@ static bool bnxt_qplib_is_atomic_cap(struct bnxt_qplib_rcfw *rcfw)
static void bnxt_qplib_query_version(struct bnxt_qplib_rcfw *rcfw,
char *fw_ver)
{
struct cmdq_query_version req;
struct creq_query_version_resp resp;
u16 cmd_flags = 0;
struct creq_query_version_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_query_version req = {};
int rc = 0;
RCFW_CMD_PREP(req, QUERY_VERSION, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_QUERY_VERSION,
sizeof(req));
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req), sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
return;
fw_ver[0] = resp.fw_maj;
@ -88,16 +91,18 @@ static void bnxt_qplib_query_version(struct bnxt_qplib_rcfw *rcfw,
int bnxt_qplib_get_dev_attr(struct bnxt_qplib_rcfw *rcfw,
struct bnxt_qplib_dev_attr *attr, bool vf)
{
struct cmdq_query_func req;
struct creq_query_func_resp resp;
struct bnxt_qplib_rcfw_sbuf *sbuf;
struct creq_query_func_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct creq_query_func_resp_sb *sb;
u16 cmd_flags = 0;
u32 temp;
struct bnxt_qplib_rcfw_sbuf *sbuf;
struct cmdq_query_func req = {};
u8 *tqm_alloc;
int i, rc = 0;
u32 temp;
RCFW_CMD_PREP(req, QUERY_FUNC, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_QUERY_FUNC,
sizeof(req));
sbuf = bnxt_qplib_rcfw_alloc_sbuf(rcfw, sizeof(*sb));
if (!sbuf) {
@ -108,8 +113,9 @@ int bnxt_qplib_get_dev_attr(struct bnxt_qplib_rcfw *rcfw,
sb = sbuf->sb;
req.resp_size = sizeof(*sb) / BNXT_QPLIB_CMDQE_UNITS;
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp,
(void *)sbuf, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, sbuf, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
goto bail;
@ -174,12 +180,14 @@ int bnxt_qplib_set_func_resources(struct bnxt_qplib_res *res,
struct bnxt_qplib_rcfw *rcfw,
struct bnxt_qplib_ctx *ctx)
{
struct cmdq_set_func_resources req;
struct creq_set_func_resources_resp resp;
u16 cmd_flags = 0;
struct creq_set_func_resources_resp resp = {};
struct cmdq_set_func_resources req = {};
struct bnxt_qplib_cmdqmsg msg = {};
int rc = 0;
RCFW_CMD_PREP(req, SET_FUNC_RESOURCES, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_SET_FUNC_RESOURCES,
sizeof(req));
req.number_of_qp = cpu_to_le32(ctx->qpc_count);
req.number_of_mrw = cpu_to_le32(ctx->mrw_count);
@ -192,9 +200,9 @@ int bnxt_qplib_set_func_resources(struct bnxt_qplib_res *res,
req.max_cq_per_vf = cpu_to_le32(ctx->vf_res.max_cq_per_vf);
req.max_gid_per_vf = cpu_to_le32(ctx->vf_res.max_gid_per_vf);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp,
NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc) {
dev_err(&res->pdev->dev, "Failed to set function resources\n");
}
@ -245,20 +253,23 @@ int bnxt_qplib_del_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl,
}
/* Remove GID from the SGID table */
if (update) {
struct cmdq_delete_gid req;
struct creq_delete_gid_resp resp;
u16 cmd_flags = 0;
struct creq_delete_gid_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_delete_gid req = {};
int rc;
RCFW_CMD_PREP(req, DELETE_GID, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_DELETE_GID,
sizeof(req));
if (sgid_tbl->hw_id[index] == 0xFFFF) {
dev_err(&res->pdev->dev,
"GID entry contains an invalid HW id\n");
return -EINVAL;
}
req.gid_index = cpu_to_le16(sgid_tbl->hw_id[index]);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
return rc;
}
@ -315,12 +326,14 @@ int bnxt_qplib_add_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl,
return -ENOMEM;
}
if (update) {
struct cmdq_add_gid req;
struct creq_add_gid_resp resp;
u16 cmd_flags = 0;
struct creq_add_gid_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_add_gid req = {};
int rc;
RCFW_CMD_PREP(req, ADD_GID, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_ADD_GID,
sizeof(req));
req.gid[0] = cpu_to_be32(((u32 *)gid->data)[3]);
req.gid[1] = cpu_to_be32(((u32 *)gid->data)[2]);
@ -345,8 +358,9 @@ int bnxt_qplib_add_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl,
req.src_mac[1] = cpu_to_be16(((u16 *)smac)[1]);
req.src_mac[2] = cpu_to_be16(((u16 *)smac)[2]);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
return rc;
sgid_tbl->hw_id[free_idx] = le32_to_cpu(resp.xid);
@ -375,12 +389,14 @@ int bnxt_qplib_update_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl,
struct bnxt_qplib_res,
sgid_tbl);
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct creq_modify_gid_resp resp;
struct cmdq_modify_gid req;
struct creq_modify_gid_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_modify_gid req = {};
int rc;
u16 cmd_flags = 0;
RCFW_CMD_PREP(req, MODIFY_GID, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_MODIFY_GID,
sizeof(req));
req.gid[0] = cpu_to_be32(((u32 *)gid->data)[3]);
req.gid[1] = cpu_to_be32(((u32 *)gid->data)[2]);
@ -399,8 +415,9 @@ int bnxt_qplib_update_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl,
req.gid_index = cpu_to_le16(gid_idx);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
return rc;
}
@ -409,14 +426,16 @@ int bnxt_qplib_create_ah(struct bnxt_qplib_res *res, struct bnxt_qplib_ah *ah,
bool block)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_create_ah req;
struct creq_create_ah_resp resp;
u16 cmd_flags = 0;
struct creq_create_ah_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_create_ah req = {};
u32 temp32[4];
u16 temp16[3];
int rc;
RCFW_CMD_PREP(req, CREATE_AH, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_CREATE_AH,
sizeof(req));
memcpy(temp32, ah->dgid.data, sizeof(struct bnxt_qplib_gid));
req.dgid[0] = cpu_to_le32(temp32[0]);
@ -439,8 +458,9 @@ int bnxt_qplib_create_ah(struct bnxt_qplib_res *res, struct bnxt_qplib_ah *ah,
req.dest_mac[1] = cpu_to_le16(temp16[1]);
req.dest_mac[2] = cpu_to_le16(temp16[2]);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp,
NULL, block);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), block);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
return rc;
@ -452,26 +472,29 @@ void bnxt_qplib_destroy_ah(struct bnxt_qplib_res *res, struct bnxt_qplib_ah *ah,
bool block)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_destroy_ah req;
struct creq_destroy_ah_resp resp;
u16 cmd_flags = 0;
struct creq_destroy_ah_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_destroy_ah req = {};
/* Clean up the AH table in the device */
RCFW_CMD_PREP(req, DESTROY_AH, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_DESTROY_AH,
sizeof(req));
req.ah_cid = cpu_to_le32(ah->id);
bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp, NULL,
block);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), block);
bnxt_qplib_rcfw_send_message(rcfw, &msg);
}
/* MRW */
int bnxt_qplib_free_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mrw)
{
struct creq_deallocate_key_resp resp = {};
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_deallocate_key req;
struct creq_deallocate_key_resp resp;
u16 cmd_flags = 0;
struct cmdq_deallocate_key req = {};
struct bnxt_qplib_cmdqmsg msg = {};
int rc;
if (mrw->lkey == 0xFFFFFFFF) {
@ -479,7 +502,9 @@ int bnxt_qplib_free_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mrw)
return 0;
}
RCFW_CMD_PREP(req, DEALLOCATE_KEY, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_DEALLOCATE_KEY,
sizeof(req));
req.mrw_flags = mrw->type;
@ -490,8 +515,9 @@ int bnxt_qplib_free_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mrw)
else
req.key = cpu_to_le32(mrw->lkey);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp,
NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
return rc;
@ -505,13 +531,15 @@ int bnxt_qplib_free_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mrw)
int bnxt_qplib_alloc_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mrw)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_allocate_mrw req;
struct creq_allocate_mrw_resp resp;
u16 cmd_flags = 0;
struct creq_allocate_mrw_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_allocate_mrw req = {};
unsigned long tmp;
int rc;
RCFW_CMD_PREP(req, ALLOCATE_MRW, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_ALLOCATE_MRW,
sizeof(req));
req.pd_id = cpu_to_le32(mrw->pd->id);
req.mrw_flags = mrw->type;
@ -523,8 +551,9 @@ int bnxt_qplib_alloc_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mrw)
tmp = (unsigned long)mrw;
req.mrw_handle = cpu_to_le64(tmp);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
return rc;
@ -541,16 +570,19 @@ int bnxt_qplib_dereg_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mrw,
bool block)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_deregister_mr req;
struct creq_deregister_mr_resp resp;
u16 cmd_flags = 0;
struct creq_deregister_mr_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_deregister_mr req = {};
int rc;
RCFW_CMD_PREP(req, DEREGISTER_MR, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_DEREGISTER_MR,
sizeof(req));
req.lkey = cpu_to_le32(mrw->lkey);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, block);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), block);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
return rc;
@ -570,11 +602,12 @@ int bnxt_qplib_reg_mr(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mr,
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct bnxt_qplib_hwq_attr hwq_attr = {};
struct bnxt_qplib_sg_info sginfo = {};
struct creq_register_mr_resp resp;
struct cmdq_register_mr req;
u16 cmd_flags = 0, level;
struct creq_register_mr_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_register_mr req = {};
int pages, rc;
u32 pg_size;
u16 level;
if (num_pbls) {
pages = roundup_pow_of_two(num_pbls);
@ -602,7 +635,9 @@ int bnxt_qplib_reg_mr(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mr,
}
}
RCFW_CMD_PREP(req, REGISTER_MR, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_REGISTER_MR,
sizeof(req));
/* Configure the request */
if (mr->hwq.level == PBL_LVL_MAX) {
@ -627,8 +662,9 @@ int bnxt_qplib_reg_mr(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mr,
req.key = cpu_to_le32(mr->lkey);
req.mr_size = cpu_to_le64(mr->total_size);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, NULL, false);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
goto fail;
@ -679,32 +715,19 @@ int bnxt_qplib_free_fast_reg_page_list(struct bnxt_qplib_res *res,
return 0;
}
int bnxt_qplib_map_tc2cos(struct bnxt_qplib_res *res, u16 *cids)
{
struct bnxt_qplib_rcfw *rcfw = res->rcfw;
struct cmdq_map_tc_to_cos req;
struct creq_map_tc_to_cos_resp resp;
u16 cmd_flags = 0;
RCFW_CMD_PREP(req, MAP_TC_TO_COS, cmd_flags);
req.cos0 = cpu_to_le16(cids[0]);
req.cos1 = cpu_to_le16(cids[1]);
return bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp,
NULL, 0);
}
int bnxt_qplib_get_roce_stats(struct bnxt_qplib_rcfw *rcfw,
struct bnxt_qplib_roce_stats *stats)
{
struct cmdq_query_roce_stats req;
struct creq_query_roce_stats_resp resp;
struct bnxt_qplib_rcfw_sbuf *sbuf;
struct creq_query_roce_stats_resp resp = {};
struct creq_query_roce_stats_resp_sb *sb;
u16 cmd_flags = 0;
struct cmdq_query_roce_stats req = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct bnxt_qplib_rcfw_sbuf *sbuf;
int rc = 0;
RCFW_CMD_PREP(req, QUERY_ROCE_STATS, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_BASE_OPCODE_QUERY_ROCE_STATS,
sizeof(req));
sbuf = bnxt_qplib_rcfw_alloc_sbuf(rcfw, sizeof(*sb));
if (!sbuf) {
@ -715,8 +738,9 @@ int bnxt_qplib_get_roce_stats(struct bnxt_qplib_rcfw *rcfw,
sb = sbuf->sb;
req.resp_size = sizeof(*sb) / BNXT_QPLIB_CMDQE_UNITS;
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp,
(void *)sbuf, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, sbuf, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
goto bail;
/* Extract the context from the side buffer */
@ -780,8 +804,8 @@ int bnxt_qplib_qext_stat(struct bnxt_qplib_rcfw *rcfw, u32 fid,
struct creq_query_roce_stats_ext_resp resp = {};
struct creq_query_roce_stats_ext_resp_sb *sb;
struct cmdq_query_roce_stats_ext req = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct bnxt_qplib_rcfw_sbuf *sbuf;
u16 cmd_flags = 0;
int rc;
sbuf = bnxt_qplib_rcfw_alloc_sbuf(rcfw, sizeof(*sb));
@ -791,15 +815,18 @@ int bnxt_qplib_qext_stat(struct bnxt_qplib_rcfw *rcfw, u32 fid,
return -ENOMEM;
}
RCFW_CMD_PREP(req, QUERY_ROCE_STATS_EXT, cmd_flags);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)&req,
CMDQ_QUERY_ROCE_STATS_EXT_OPCODE_QUERY_ROCE_STATS,
sizeof(req));
req.resp_size = ALIGN(sizeof(*sb), BNXT_QPLIB_CMDQE_UNITS);
req.resp_addr = cpu_to_le64(sbuf->dma_addr);
req.function_id = cpu_to_le32(fid);
req.flags = cpu_to_le16(CMDQ_QUERY_ROCE_STATS_EXT_FLAGS_FUNCTION_ID);
rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req,
(void *)&resp, (void *)sbuf, 0);
bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, sbuf, sizeof(req),
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(rcfw, &msg);
if (rc)
goto bail;
@ -823,3 +850,111 @@ bail:
bnxt_qplib_rcfw_free_sbuf(rcfw, sbuf);
return rc;
}
static void bnxt_qplib_fill_cc_gen1(struct cmdq_modify_roce_cc_gen1_tlv *ext_req,
struct bnxt_qplib_cc_param_ext *cc_ext)
{
ext_req->modify_mask = cpu_to_le64(cc_ext->ext_mask);
cc_ext->ext_mask = 0;
ext_req->inactivity_th_hi = cpu_to_le16(cc_ext->inact_th_hi);
ext_req->min_time_between_cnps = cpu_to_le16(cc_ext->min_delta_cnp);
ext_req->init_cp = cpu_to_le16(cc_ext->init_cp);
ext_req->tr_update_mode = cc_ext->tr_update_mode;
ext_req->tr_update_cycles = cc_ext->tr_update_cyls;
ext_req->fr_num_rtts = cc_ext->fr_rtt;
ext_req->ai_rate_increase = cc_ext->ai_rate_incr;
ext_req->reduction_relax_rtts_th = cpu_to_le16(cc_ext->rr_rtt_th);
ext_req->additional_relax_cr_th = cpu_to_le16(cc_ext->ar_cr_th);
ext_req->cr_min_th = cpu_to_le16(cc_ext->cr_min_th);
ext_req->bw_avg_weight = cc_ext->bw_avg_weight;
ext_req->actual_cr_factor = cc_ext->cr_factor;
ext_req->max_cp_cr_th = cpu_to_le16(cc_ext->cr_th_max_cp);
ext_req->cp_bias_en = cc_ext->cp_bias_en;
ext_req->cp_bias = cc_ext->cp_bias;
ext_req->cnp_ecn = cc_ext->cnp_ecn;
ext_req->rtt_jitter_en = cc_ext->rtt_jitter_en;
ext_req->link_bytes_per_usec = cpu_to_le16(cc_ext->bytes_per_usec);
ext_req->reset_cc_cr_th = cpu_to_le16(cc_ext->cc_cr_reset_th);
ext_req->cr_width = cc_ext->cr_width;
ext_req->quota_period_min = cc_ext->min_quota;
ext_req->quota_period_max = cc_ext->max_quota;
ext_req->quota_period_abs_max = cc_ext->abs_max_quota;
ext_req->tr_lower_bound = cpu_to_le16(cc_ext->tr_lb);
ext_req->cr_prob_factor = cc_ext->cr_prob_fac;
ext_req->tr_prob_factor = cc_ext->tr_prob_fac;
ext_req->fairness_cr_th = cpu_to_le16(cc_ext->fair_cr_th);
ext_req->red_div = cc_ext->red_div;
ext_req->cnp_ratio_th = cc_ext->cnp_ratio_th;
ext_req->exp_ai_rtts = cpu_to_le16(cc_ext->ai_ext_rtt);
ext_req->exp_ai_cr_cp_ratio = cc_ext->exp_crcp_ratio;
ext_req->use_rate_table = cc_ext->low_rate_en;
ext_req->cp_exp_update_th = cpu_to_le16(cc_ext->cpcr_update_th);
ext_req->high_exp_ai_rtts_th1 = cpu_to_le16(cc_ext->ai_rtt_th1);
ext_req->high_exp_ai_rtts_th2 = cpu_to_le16(cc_ext->ai_rtt_th2);
ext_req->actual_cr_cong_free_rtts_th = cpu_to_le16(cc_ext->cf_rtt_th);
ext_req->severe_cong_cr_th1 = cpu_to_le16(cc_ext->sc_cr_th1);
ext_req->severe_cong_cr_th2 = cpu_to_le16(cc_ext->sc_cr_th2);
ext_req->link64B_per_rtt = cpu_to_le32(cc_ext->l64B_per_rtt);
ext_req->cc_ack_bytes = cc_ext->cc_ack_bytes;
}
int bnxt_qplib_modify_cc(struct bnxt_qplib_res *res,
struct bnxt_qplib_cc_param *cc_param)
{
struct bnxt_qplib_tlv_modify_cc_req tlv_req = {};
struct creq_modify_roce_cc_resp resp = {};
struct bnxt_qplib_cmdqmsg msg = {};
struct cmdq_modify_roce_cc *req;
int req_size;
void *cmd;
int rc;
/* Prepare the older base command */
req = &tlv_req.base_req;
cmd = req;
req_size = sizeof(*req);
bnxt_qplib_rcfw_cmd_prep((struct cmdq_base *)req, CMDQ_BASE_OPCODE_MODIFY_ROCE_CC,
sizeof(*req));
req->modify_mask = cpu_to_le32(cc_param->mask);
req->enable_cc = cc_param->enable;
req->g = cc_param->g;
req->num_phases_per_state = cc_param->nph_per_state;
req->time_per_phase = cc_param->time_pph;
req->pkts_per_phase = cc_param->pkts_pph;
req->init_cr = cpu_to_le16(cc_param->init_cr);
req->init_tr = cpu_to_le16(cc_param->init_tr);
req->tos_dscp_tos_ecn = (cc_param->tos_dscp << CMDQ_MODIFY_ROCE_CC_TOS_DSCP_SFT) |
(cc_param->tos_ecn & CMDQ_MODIFY_ROCE_CC_TOS_ECN_MASK);
req->alt_vlan_pcp = cc_param->alt_vlan_pcp;
req->alt_tos_dscp = cpu_to_le16(cc_param->alt_tos_dscp);
req->rtt = cpu_to_le16(cc_param->rtt);
req->tcp_cp = cpu_to_le16(cc_param->tcp_cp);
req->cc_mode = cc_param->cc_mode;
req->inactivity_th = cpu_to_le16(cc_param->inact_th);
/* For chip gen P5 onwards fill extended cmd and header */
if (bnxt_qplib_is_chip_gen_p5(res->cctx)) {
struct roce_tlv *hdr;
u32 payload;
u32 chunks;
cmd = &tlv_req;
req_size = sizeof(tlv_req);
/* Prepare primary tlv header */
hdr = &tlv_req.tlv_hdr;
chunks = CHUNKS(sizeof(struct bnxt_qplib_tlv_modify_cc_req));
payload = sizeof(struct cmdq_modify_roce_cc);
__roce_1st_tlv_prep(hdr, chunks, payload, true);
/* Prepare secondary tlv header */
hdr = (struct roce_tlv *)&tlv_req.ext_req;
payload = sizeof(struct cmdq_modify_roce_cc_gen1_tlv) -
sizeof(struct roce_tlv);
__roce_ext_tlv_prep(hdr, TLV_TYPE_MODIFY_ROCE_CC_GEN1, payload, false, true);
bnxt_qplib_fill_cc_gen1(&tlv_req.ext_req, &cc_param->cc_ext);
}
bnxt_qplib_fill_cmdqmsg(&msg, cmd, &resp, NULL, req_size,
sizeof(resp), 0);
rc = bnxt_qplib_rcfw_send_message(res->rcfw, &msg);
return rc;
}

68
drivers/infiniband/hw/bnxt_re/qplib_sp.h
View File

@ -244,6 +244,71 @@ struct bnxt_qplib_ext_stat {
u64 rx_ecn_marked;
};
struct bnxt_qplib_cc_param_ext {
u64 ext_mask;
u16 inact_th_hi;
u16 min_delta_cnp;
u16 init_cp;
u8 tr_update_mode;
u8 tr_update_cyls;
u8 fr_rtt;
u8 ai_rate_incr;
u16 rr_rtt_th;
u16 ar_cr_th;
u16 cr_min_th;
u8 bw_avg_weight;
u8 cr_factor;
u16 cr_th_max_cp;
u8 cp_bias_en;
u8 cp_bias;
u8 cnp_ecn;
u8 rtt_jitter_en;
u16 bytes_per_usec;
u16 cc_cr_reset_th;
u8 cr_width;
u8 min_quota;
u8 max_quota;
u8 abs_max_quota;
u16 tr_lb;
u8 cr_prob_fac;
u8 tr_prob_fac;
u16 fair_cr_th;
u8 red_div;
u8 cnp_ratio_th;
u16 ai_ext_rtt;
u8 exp_crcp_ratio;
u8 low_rate_en;
u16 cpcr_update_th;
u16 ai_rtt_th1;
u16 ai_rtt_th2;
u16 cf_rtt_th;
u16 sc_cr_th1; /* severe congestion cr threshold 1 */
u16 sc_cr_th2; /* severe congestion cr threshold 2 */
u32 l64B_per_rtt;
u8 cc_ack_bytes;
u16 reduce_cf_rtt_th;
};
struct bnxt_qplib_cc_param {
u8 alt_vlan_pcp;
u16 alt_tos_dscp;
u8 cc_mode;
u8 enable;
u16 inact_th;
u16 init_cr;
u16 init_tr;
u16 rtt;
u8 g;
u8 nph_per_state;
u8 time_pph;
u8 pkts_pph;
u8 tos_ecn;
u8 tos_dscp;
u16 tcp_cp;
struct bnxt_qplib_cc_param_ext cc_ext;
u32 mask;
};
int bnxt_qplib_get_sgid(struct bnxt_qplib_res *res,
struct bnxt_qplib_sgid_tbl *sgid_tbl, int index,
struct bnxt_qplib_gid *gid);
@ -277,10 +342,11 @@ int bnxt_qplib_alloc_fast_reg_page_list(struct bnxt_qplib_res *res,
struct bnxt_qplib_frpl *frpl, int max);
int bnxt_qplib_free_fast_reg_page_list(struct bnxt_qplib_res *res,
struct bnxt_qplib_frpl *frpl);
int bnxt_qplib_map_tc2cos(struct bnxt_qplib_res *res, u16 *cids);
int bnxt_qplib_get_roce_stats(struct bnxt_qplib_rcfw *rcfw,
struct bnxt_qplib_roce_stats *stats);
int bnxt_qplib_qext_stat(struct bnxt_qplib_rcfw *rcfw, u32 fid,
struct bnxt_qplib_ext_stat *estat);
int bnxt_qplib_modify_cc(struct bnxt_qplib_res *res,
struct bnxt_qplib_cc_param *cc_param);
#endif /* __BNXT_QPLIB_SP_H__*/

162
drivers/infiniband/hw/bnxt_re/qplib_tlv.h Normal file
View File

@ -0,0 +1,162 @@
/* SPDX-License-Identifier: GPL-2.0 or BSD-3-Clause */
#ifndef __QPLIB_TLV_H__
#define __QPLIB_TLV_H__
struct roce_tlv {
struct tlv tlv;
u8 total_size; // in units of 16 byte chunks
u8 unused[7]; // for 16 byte alignment
};
#define CHUNK_SIZE 16
#define CHUNKS(x) (((x) + CHUNK_SIZE - 1) / CHUNK_SIZE)
static inline void __roce_1st_tlv_prep(struct roce_tlv *rtlv, u8 tot_chunks,
u16 content_bytes, u8 flags)
{
rtlv->tlv.cmd_discr = cpu_to_le16(CMD_DISCR_TLV_ENCAP);
rtlv->tlv.tlv_type = cpu_to_le16(TLV_TYPE_ROCE_SP_COMMAND);
rtlv->tlv.length = cpu_to_le16(content_bytes);
rtlv->tlv.flags = TLV_FLAGS_REQUIRED;
rtlv->tlv.flags |= flags ? TLV_FLAGS_MORE : 0;
rtlv->total_size = (tot_chunks);
}
static inline void __roce_ext_tlv_prep(struct roce_tlv *rtlv, u16 tlv_type,
u16 content_bytes, u8 more, u8 flags)
{
rtlv->tlv.cmd_discr = cpu_to_le16(CMD_DISCR_TLV_ENCAP);
rtlv->tlv.tlv_type = cpu_to_le16(tlv_type);
rtlv->tlv.length = cpu_to_le16(content_bytes);
rtlv->tlv.flags |= more ? TLV_FLAGS_MORE : 0;
rtlv->tlv.flags |= flags ? TLV_FLAGS_REQUIRED : 0;
}
/*
* TLV size in units of 16 byte chunks
*/
#define TLV_SIZE ((sizeof(struct roce_tlv) + 15) / 16)
/*
* TLV length in bytes
*/
#define TLV_BYTES (TLV_SIZE * 16)
#define HAS_TLV_HEADER(msg) (le16_to_cpu(((struct tlv *)(msg))->cmd_discr) == CMD_DISCR_TLV_ENCAP)
#define GET_TLV_DATA(tlv) ((void *)&((uint8_t *)(tlv))[TLV_BYTES])
static inline u8 __get_cmdq_base_opcode(struct cmdq_base *req, u32 size)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
return ((struct cmdq_base *)GET_TLV_DATA(req))->opcode;
else
return req->opcode;
}
static inline void __set_cmdq_base_opcode(struct cmdq_base *req,
u32 size, u8 val)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
((struct cmdq_base *)GET_TLV_DATA(req))->opcode = val;
else
req->opcode = val;
}
static inline __le16 __get_cmdq_base_cookie(struct cmdq_base *req, u32 size)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
return ((struct cmdq_base *)GET_TLV_DATA(req))->cookie;
else
return req->cookie;
}
static inline void __set_cmdq_base_cookie(struct cmdq_base *req,
u32 size, __le16 val)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
((struct cmdq_base *)GET_TLV_DATA(req))->cookie = val;
else
req->cookie = val;
}
static inline __le64 __get_cmdq_base_resp_addr(struct cmdq_base *req, u32 size)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
return ((struct cmdq_base *)GET_TLV_DATA(req))->resp_addr;
else
return req->resp_addr;
}
static inline void __set_cmdq_base_resp_addr(struct cmdq_base *req,
u32 size, __le64 val)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
((struct cmdq_base *)GET_TLV_DATA(req))->resp_addr = val;
else
req->resp_addr = val;
}
static inline u8 __get_cmdq_base_resp_size(struct cmdq_base *req, u32 size)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
return ((struct cmdq_base *)GET_TLV_DATA(req))->resp_size;
else
return req->resp_size;
}
static inline void __set_cmdq_base_resp_size(struct cmdq_base *req,
u32 size, u8 val)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
((struct cmdq_base *)GET_TLV_DATA(req))->resp_size = val;
else
req->resp_size = val;
}
static inline u8 __get_cmdq_base_cmd_size(struct cmdq_base *req, u32 size)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
return ((struct roce_tlv *)(req))->total_size;
else
return req->cmd_size;
}
static inline void __set_cmdq_base_cmd_size(struct cmdq_base *req,
u32 size, u8 val)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
((struct cmdq_base *)GET_TLV_DATA(req))->cmd_size = val;
else
req->cmd_size = val;
}
static inline __le16 __get_cmdq_base_flags(struct cmdq_base *req, u32 size)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
return ((struct cmdq_base *)GET_TLV_DATA(req))->flags;
else
return req->flags;
}
static inline void __set_cmdq_base_flags(struct cmdq_base *req,
u32 size, __le16 val)
{
if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
((struct cmdq_base *)GET_TLV_DATA(req))->flags = val;
else
req->flags = val;
}
struct bnxt_qplib_tlv_modify_cc_req {
struct roce_tlv tlv_hdr;
struct cmdq_modify_roce_cc base_req;
__le64 tlvpad;
struct cmdq_modify_roce_cc_gen1_tlv ext_req;
};
struct bnxt_qplib_tlv_query_rcc_sb {
struct roce_tlv tlv_hdr;
struct creq_query_roce_cc_resp_sb base_sb;
struct creq_query_roce_cc_gen1_resp_sb_tlv gen1_sb;
};
#endif /* __QPLIB_TLV_H__ */

7155
drivers/infiniband/hw/bnxt_re/roce_hsi.h
View File

File diff suppressed because it is too large Load Diff

17
drivers/infiniband/hw/efa/efa_admin_cmds_defs.h
View File

@ -1,6 +1,6 @@
/* SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause */
/*
* Copyright 2018-2021 Amazon.com, Inc. or its affiliates. All rights reserved.
* Copyright 2018-2023 Amazon.com, Inc. or its affiliates. All rights reserved.
*/
#ifndef _EFA_ADMIN_CMDS_H_
@ -376,7 +376,9 @@ struct efa_admin_reg_mr_cmd {
* 0 : local_write_enable - Local write permissions:
* must be set for RQ buffers and buffers posted for
* RDMA Read requests
* 1 : reserved1 - MBZ
* 1 : remote_write_enable - Remote write
* permissions: must be set to enable RDMA write to
* the region
* 2 : remote_read_enable - Remote read permissions:
* must be set to enable RDMA read from the region
* 7:3 : reserved2 - MBZ
@ -618,7 +620,11 @@ struct efa_admin_feature_device_attr_desc {
* TX queues
* 1 : rnr_retry - If set, RNR retry is supported on
* modify QP command
* 31:2 : reserved - MBZ
* 2 : data_polling_128 - If set, 128 bytes data
* polling is supported
* 3 : rdma_write - If set, RDMA Write is supported
* on TX queues
* 31:4 : reserved - MBZ
*/
u32 device_caps;
@ -672,7 +678,7 @@ struct efa_admin_feature_queue_attr_desc {
/* The maximum size of LLQ in bytes */
u32 max_llq_size;
/* Maximum number of SGEs for a single RDMA read WQE */
/* Maximum number of SGEs for a single RDMA read/write WQE */
u16 max_wr_rdma_sges;
/*
@ -977,6 +983,7 @@ struct efa_admin_host_info {
#define EFA_ADMIN_REG_MR_CMD_PHYS_PAGE_SIZE_SHIFT_MASK GENMASK(4, 0)
#define EFA_ADMIN_REG_MR_CMD_MEM_ADDR_PHY_MODE_EN_MASK BIT(7)
#define EFA_ADMIN_REG_MR_CMD_LOCAL_WRITE_ENABLE_MASK BIT(0)
#define EFA_ADMIN_REG_MR_CMD_REMOTE_WRITE_ENABLE_MASK BIT(1)
#define EFA_ADMIN_REG_MR_CMD_REMOTE_READ_ENABLE_MASK BIT(2)
/* create_cq_cmd */
@ -991,6 +998,8 @@ struct efa_admin_host_info {
/* feature_device_attr_desc */
#define EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_RDMA_READ_MASK BIT(0)
#define EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_RNR_RETRY_MASK BIT(1)
#define EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_DATA_POLLING_128_MASK BIT(2)
#define EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_RDMA_WRITE_MASK BIT(3)
/* create_eq_cmd */
#define EFA_ADMIN_CREATE_EQ_CMD_ENTRY_SIZE_WORDS_MASK GENMASK(4, 0)

42
drivers/infiniband/hw/efa/efa_io_defs.h
View File

@ -1,6 +1,6 @@
/* SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause */
/*
* Copyright 2018-2022 Amazon.com, Inc. or its affiliates. All rights reserved.
* Copyright 2018-2023 Amazon.com, Inc. or its affiliates. All rights reserved.
*/
#ifndef _EFA_IO_H_
@ -23,6 +23,8 @@ enum efa_io_send_op_type {
EFA_IO_SEND = 0,
/* RDMA read */
EFA_IO_RDMA_READ = 1,
/* RDMA write */
EFA_IO_RDMA_WRITE = 2,
};
enum efa_io_comp_status {
@ -62,8 +64,7 @@ struct efa_io_tx_meta_desc {
/*
* control flags
* 3:0 : op_type - operation type: send/rdma/fast mem
* ops/etc
* 3:0 : op_type - enum efa_io_send_op_type
* 4 : has_imm - immediate_data field carries valid
* data.
* 5 : inline_msg - inline mode - inline message data
@ -219,21 +220,22 @@ struct efa_io_cdesc_common {
* 2:1 : q_type - enum efa_io_queue_type: send/recv
* 3 : has_imm - indicates that immediate data is
* present - for RX completions only
* 7:4 : reserved28 - MBZ
* 6:4 : op_type - enum efa_io_send_op_type
* 7 : reserved31 - MBZ
*/
u8 flags;
/* local QP number */
u16 qp_num;
/* Transferred length */
u16 length;
};
/* Tx completion descriptor */
struct efa_io_tx_cdesc {
/* Common completion info */
struct efa_io_cdesc_common common;
/* MBZ */
u16 reserved16;
};
/* Rx Completion Descriptor */
@ -241,6 +243,9 @@ struct efa_io_rx_cdesc {
/* Common completion info */
struct efa_io_cdesc_common common;
/* Transferred length bits[15:0] */
u16 length;
/* Remote Address Handle FW index, 0xFFFF indicates invalid ah */
u16 ah;
@ -250,16 +255,26 @@ struct efa_io_rx_cdesc {
u32 imm;
};
/* Rx Completion Descriptor RDMA write info */
struct efa_io_rx_cdesc_rdma_write {
/* Transferred length bits[31:16] */
u16 length_hi;
};
/* Extended Rx Completion Descriptor */
struct efa_io_rx_cdesc_ex {
/* Base RX completion info */
struct efa_io_rx_cdesc rx_cdesc_base;
struct efa_io_rx_cdesc base;
/*
* Valid only in case of unknown AH (0xFFFF) and CQ set_src_addr is
* enabled.
*/
u8 src_addr[16];
union {
struct efa_io_rx_cdesc_rdma_write rdma_write;
/*
* Valid only in case of unknown AH (0xFFFF) and CQ
* set_src_addr is enabled.
*/
u8 src_addr[16];
} u;
};
/* tx_meta_desc */
@ -285,5 +300,6 @@ struct efa_io_rx_cdesc_ex {
#define EFA_IO_CDESC_COMMON_PHASE_MASK BIT(0)
#define EFA_IO_CDESC_COMMON_Q_TYPE_MASK GENMASK(2, 1)
#define EFA_IO_CDESC_COMMON_HAS_IMM_MASK BIT(3)
#define EFA_IO_CDESC_COMMON_OP_TYPE_MASK GENMASK(6, 4)
#endif /* _EFA_IO_H_ */

11
drivers/infiniband/hw/efa/efa_verbs.c
View File

@ -1,6 +1,6 @@
// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
/*
* Copyright 2018-2022 Amazon.com, Inc. or its affiliates. All rights reserved.
* Copyright 2018-2023 Amazon.com, Inc. or its affiliates. All rights reserved.
*/
#include <linux/dma-buf.h>
@ -250,6 +250,12 @@ int efa_query_device(struct ib_device *ibdev,
if (EFA_DEV_CAP(dev, RNR_RETRY))
resp.device_caps |= EFA_QUERY_DEVICE_CAPS_RNR_RETRY;
if (EFA_DEV_CAP(dev, DATA_POLLING_128))
resp.device_caps |= EFA_QUERY_DEVICE_CAPS_DATA_POLLING_128;
if (EFA_DEV_CAP(dev, RDMA_WRITE))
resp.device_caps |= EFA_QUERY_DEVICE_CAPS_RDMA_WRITE;
if (dev->neqs)
resp.device_caps |= EFA_QUERY_DEVICE_CAPS_CQ_NOTIFICATIONS;
@ -1569,7 +1575,8 @@ static struct efa_mr *efa_alloc_mr(struct ib_pd *ibpd, int access_flags,
supp_access_flags =
IB_ACCESS_LOCAL_WRITE |
(EFA_DEV_CAP(dev, RDMA_READ) ? IB_ACCESS_REMOTE_READ : 0);
(EFA_DEV_CAP(dev, RDMA_READ) ? IB_ACCESS_REMOTE_READ : 0) |
(EFA_DEV_CAP(dev, RDMA_WRITE) ? IB_ACCESS_REMOTE_WRITE : 0);
access_flags &= ~IB_ACCESS_OPTIONAL;
if (access_flags & ~supp_access_flags) {

2
drivers/infiniband/hw/erdma/erdma.h
View File

@ -32,7 +32,7 @@ struct erdma_eq {
atomic64_t event_num;
atomic64_t notify_num;
u64 __iomem *db_addr;
void __iomem *db;
u64 *db_record;
};

10
drivers/infiniband/hw/erdma/erdma_cm.h
View File

@ -33,11 +33,11 @@ struct mpa_rr_params {
* MPA request/response Hdr bits & fields
*/
enum {
MPA_RR_FLAG_MARKERS = __cpu_to_be16(0x8000),
MPA_RR_FLAG_CRC = __cpu_to_be16(0x4000),
MPA_RR_FLAG_REJECT = __cpu_to_be16(0x2000),
MPA_RR_RESERVED = __cpu_to_be16(0x1f00),
MPA_RR_MASK_REVISION = __cpu_to_be16(0x00ff)
MPA_RR_FLAG_MARKERS = cpu_to_be16(0x8000),
MPA_RR_FLAG_CRC = cpu_to_be16(0x4000),
MPA_RR_FLAG_REJECT = cpu_to_be16(0x2000),
MPA_RR_RESERVED = cpu_to_be16(0x1f00),
MPA_RR_MASK_REVISION = cpu_to_be16(0x00ff)
};
/*

42
drivers/infiniband/hw/erdma/erdma_cmdq.c
View File

@ -166,8 +166,7 @@ static int erdma_cmdq_eq_init(struct erdma_dev *dev)
spin_lock_init(&eq->lock);
atomic64_set(&eq->event_num, 0);
eq->db_addr =
(u64 __iomem *)(dev->func_bar + ERDMA_REGS_CEQ_DB_BASE_REG);
eq->db = dev->func_bar + ERDMA_REGS_CEQ_DB_BASE_REG;
eq->db_record = (u64 *)(eq->qbuf + buf_size);
erdma_reg_write32(dev, ERDMA_REGS_CMDQ_EQ_ADDR_H_REG,
@ -183,9 +182,8 @@ static int erdma_cmdq_eq_init(struct erdma_dev *dev)
int erdma_cmdq_init(struct erdma_dev *dev)
{
int err, i;
struct erdma_cmdq *cmdq = &dev->cmdq;
u32 sts, ctrl;
int err;
cmdq->max_outstandings = ERDMA_CMDQ_MAX_OUTSTANDING;
cmdq->use_event = false;
@ -208,34 +206,10 @@ int erdma_cmdq_init(struct erdma_dev *dev)
if (err)
goto err_destroy_cq;
ctrl = FIELD_PREP(ERDMA_REG_DEV_CTRL_INIT_MASK, 1);
erdma_reg_write32(dev, ERDMA_REGS_DEV_CTRL_REG, ctrl);
for (i = 0; i < ERDMA_WAIT_DEV_DONE_CNT; i++) {
sts = erdma_reg_read32_filed(dev, ERDMA_REGS_DEV_ST_REG,
ERDMA_REG_DEV_ST_INIT_DONE_MASK);
if (sts)
break;
msleep(ERDMA_REG_ACCESS_WAIT_MS);
}
if (i == ERDMA_WAIT_DEV_DONE_CNT) {
dev_err(&dev->pdev->dev, "wait init done failed.\n");
err = -ETIMEDOUT;
goto err_destroy_eq;
}
set_bit(ERDMA_CMDQ_STATE_OK_BIT, &cmdq->state);
return 0;
err_destroy_eq:
dma_free_coherent(&dev->pdev->dev,
(cmdq->eq.depth << EQE_SHIFT) +
ERDMA_EXTRA_BUFFER_SIZE,
cmdq->eq.qbuf, cmdq->eq.qbuf_dma_addr);
err_destroy_cq:
dma_free_coherent(&dev->pdev->dev,
(cmdq->cq.depth << CQE_SHIFT) +
@ -283,7 +257,7 @@ static void *get_next_valid_cmdq_cqe(struct erdma_cmdq *cmdq)
__be32 *cqe = get_queue_entry(cmdq->cq.qbuf, cmdq->cq.ci,
cmdq->cq.depth, CQE_SHIFT);
u32 owner = FIELD_GET(ERDMA_CQE_HDR_OWNER_MASK,
__be32_to_cpu(READ_ONCE(*cqe)));
be32_to_cpu(READ_ONCE(*cqe)));
return owner ^ !!(cmdq->cq.ci & cmdq->cq.depth) ? cqe : NULL;
}
@ -319,7 +293,6 @@ static int erdma_poll_single_cmd_completion(struct erdma_cmdq *cmdq)
__be32 *cqe;
u16 ctx_id;
u64 *sqe;
int i;
cqe = get_next_valid_cmdq_cqe(cmdq);
if (!cqe)
@ -328,8 +301,8 @@ static int erdma_poll_single_cmd_completion(struct erdma_cmdq *cmdq)
cmdq->cq.ci++;
dma_rmb();
hdr0 = __be32_to_cpu(*cqe);
sqe_idx = __be32_to_cpu(*(cqe + 1));
hdr0 = be32_to_cpu(*cqe);
sqe_idx = be32_to_cpu(*(cqe + 1));
sqe = get_queue_entry(cmdq->sq.qbuf, sqe_idx, cmdq->sq.depth,
SQEBB_SHIFT);
@ -341,9 +314,8 @@ static int erdma_poll_single_cmd_completion(struct erdma_cmdq *cmdq)
comp_wait->cmd_status = ERDMA_CMD_STATUS_FINISHED;
comp_wait->comp_status = FIELD_GET(ERDMA_CQE_HDR_SYNDROME_MASK, hdr0);
cmdq->sq.ci += cmdq->sq.wqebb_cnt;
for (i = 0; i < 4; i++)
comp_wait->comp_data[i] = __be32_to_cpu(*(cqe + 2 + i));
/* Copy 16B comp data after cqe hdr to outer */
be32_to_cpu_array(comp_wait->comp_data, cqe + 2, 4);
if (cmdq->use_event)
complete(&comp_wait->wait_event);

2
drivers/infiniband/hw/erdma/erdma_cq.c
View File

@ -11,7 +11,7 @@ static void *get_next_valid_cqe(struct erdma_cq *cq)
__be32 *cqe = get_queue_entry(cq->kern_cq.qbuf, cq->kern_cq.ci,
cq->depth, CQE_SHIFT);
u32 owner = FIELD_GET(ERDMA_CQE_HDR_OWNER_MASK,
__be32_to_cpu(READ_ONCE(*cqe)));
be32_to_cpu(READ_ONCE(*cqe)));
return owner ^ !!(cq->kern_cq.ci & cq->depth) ? cqe : NULL;
}

9
drivers/infiniband/hw/erdma/erdma_eq.c
View File

@ -14,7 +14,7 @@ void notify_eq(struct erdma_eq *eq)
FIELD_PREP(ERDMA_EQDB_ARM_MASK, 1);
*eq->db_record = db_data;
writeq(db_data, eq->db_addr);
writeq(db_data, eq->db);
atomic64_inc(&eq->notify_num);
}
@ -98,7 +98,7 @@ int erdma_aeq_init(struct erdma_dev *dev)
atomic64_set(&eq->event_num, 0);
atomic64_set(&eq->notify_num, 0);
eq->db_addr = (u64 __iomem *)(dev->func_bar + ERDMA_REGS_AEQ_DB_REG);
eq->db = dev->func_bar + ERDMA_REGS_AEQ_DB_REG;
eq->db_record = (u64 *)(eq->qbuf + buf_size);
erdma_reg_write32(dev, ERDMA_REGS_AEQ_ADDR_H_REG,
@ -243,9 +243,8 @@ static int erdma_ceq_init_one(struct erdma_dev *dev, u16 ceqn)
atomic64_set(&eq->notify_num, 0);
eq->depth = ERDMA_DEFAULT_EQ_DEPTH;
eq->db_addr =
(u64 __iomem *)(dev->func_bar + ERDMA_REGS_CEQ_DB_BASE_REG +
(ceqn + 1) * ERDMA_DB_SIZE);
eq->db = dev->func_bar + ERDMA_REGS_CEQ_DB_BASE_REG +
(ceqn + 1) * ERDMA_DB_SIZE;
eq->db_record = (u64 *)(eq->qbuf + buf_size);
eq->ci = 0;
dev->ceqs[ceqn].dev = dev;

4
drivers/infiniband/hw/erdma/erdma_hw.h
View File

@ -112,6 +112,10 @@
#define ERDMA_PAGE_SIZE_SUPPORT 0x7FFFF000
/* Hardware page size definition */
#define ERDMA_HW_PAGE_SHIFT 12
#define ERDMA_HW_PAGE_SIZE 4096
/* WQE related. */
#define EQE_SIZE 16
#define EQE_SHIFT 4

39
drivers/infiniband/hw/erdma/erdma_main.c
View File

@ -211,13 +211,36 @@ static int erdma_device_init(struct erdma_dev *dev, struct pci_dev *pdev)
return 0;
}
static void erdma_device_uninit(struct erdma_dev *dev)
static void erdma_hw_reset(struct erdma_dev *dev)
{
u32 ctrl = FIELD_PREP(ERDMA_REG_DEV_CTRL_RESET_MASK, 1);
erdma_reg_write32(dev, ERDMA_REGS_DEV_CTRL_REG, ctrl);
}
static int erdma_wait_hw_init_done(struct erdma_dev *dev)
{
int i;
erdma_reg_write32(dev, ERDMA_REGS_DEV_CTRL_REG,
FIELD_PREP(ERDMA_REG_DEV_CTRL_INIT_MASK, 1));
for (i = 0; i < ERDMA_WAIT_DEV_DONE_CNT; i++) {
if (erdma_reg_read32_filed(dev, ERDMA_REGS_DEV_ST_REG,
ERDMA_REG_DEV_ST_INIT_DONE_MASK))
break;
msleep(ERDMA_REG_ACCESS_WAIT_MS);
}
if (i == ERDMA_WAIT_DEV_DONE_CNT) {
dev_err(&dev->pdev->dev, "wait init done failed.\n");
return -ETIMEDOUT;
}
return 0;
}
static const struct pci_device_id erdma_pci_tbl[] = {
{ PCI_DEVICE(PCI_VENDOR_ID_ALIBABA, 0x107f) },
{}
@ -293,16 +316,22 @@ static int erdma_probe_dev(struct pci_dev *pdev)
if (err)
goto err_uninit_aeq;
err = erdma_ceqs_init(dev);
err = erdma_wait_hw_init_done(dev);
if (err)
goto err_uninit_cmdq;
err = erdma_ceqs_init(dev);
if (err)
goto err_reset_hw;
erdma_finish_cmdq_init(dev);
return 0;
err_reset_hw:
erdma_hw_reset(dev);
err_uninit_cmdq:
erdma_device_uninit(dev);
erdma_cmdq_destroy(dev);
err_uninit_aeq:
@ -334,9 +363,7 @@ static void erdma_remove_dev(struct pci_dev *pdev)
struct erdma_dev *dev = pci_get_drvdata(pdev);
erdma_ceqs_uninit(dev);
erdma_device_uninit(dev);
erdma_hw_reset(dev);
erdma_cmdq_destroy(dev);
erdma_aeq_destroy(dev);
erdma_comm_irq_uninit(dev);

17
drivers/infiniband/hw/erdma/erdma_verbs.c
View File

@ -38,7 +38,7 @@ static int create_qp_cmd(struct erdma_dev *dev, struct erdma_qp *qp)
FIELD_PREP(ERDMA_CMD_CREATE_QP_PD_MASK, pd->pdn);
if (rdma_is_kernel_res(&qp->ibqp.res)) {
u32 pgsz_range = ilog2(SZ_1M) - PAGE_SHIFT;
u32 pgsz_range = ilog2(SZ_1M) - ERDMA_HW_PAGE_SHIFT;
req.sq_cqn_mtt_cfg =
FIELD_PREP(ERDMA_CMD_CREATE_QP_PAGE_SIZE_MASK,
@ -66,13 +66,13 @@ static int create_qp_cmd(struct erdma_dev *dev, struct erdma_qp *qp)
user_qp = &qp->user_qp;
req.sq_cqn_mtt_cfg = FIELD_PREP(
ERDMA_CMD_CREATE_QP_PAGE_SIZE_MASK,
ilog2(user_qp->sq_mtt.page_size) - PAGE_SHIFT);
ilog2(user_qp->sq_mtt.page_size) - ERDMA_HW_PAGE_SHIFT);
req.sq_cqn_mtt_cfg |=
FIELD_PREP(ERDMA_CMD_CREATE_QP_CQN_MASK, qp->scq->cqn);
req.rq_cqn_mtt_cfg = FIELD_PREP(
ERDMA_CMD_CREATE_QP_PAGE_SIZE_MASK,
ilog2(user_qp->rq_mtt.page_size) - PAGE_SHIFT);
ilog2(user_qp->rq_mtt.page_size) - ERDMA_HW_PAGE_SHIFT);
req.rq_cqn_mtt_cfg |=
FIELD_PREP(ERDMA_CMD_CREATE_QP_CQN_MASK, qp->rcq->cqn);
@ -162,7 +162,7 @@ static int create_cq_cmd(struct erdma_dev *dev, struct erdma_cq *cq)
if (rdma_is_kernel_res(&cq->ibcq.res)) {
page_size = SZ_32M;
req.cfg0 |= FIELD_PREP(ERDMA_CMD_CREATE_CQ_PAGESIZE_MASK,
ilog2(page_size) - PAGE_SHIFT);
ilog2(page_size) - ERDMA_HW_PAGE_SHIFT);
req.qbuf_addr_l = lower_32_bits(cq->kern_cq.qbuf_dma_addr);
req.qbuf_addr_h = upper_32_bits(cq->kern_cq.qbuf_dma_addr);
@ -175,8 +175,9 @@ static int create_cq_cmd(struct erdma_dev *dev, struct erdma_cq *cq)
cq->kern_cq.qbuf_dma_addr + (cq->depth << CQE_SHIFT);
} else {
mtt = &cq->user_cq.qbuf_mtt;
req.cfg0 |= FIELD_PREP(ERDMA_CMD_CREATE_CQ_PAGESIZE_MASK,
ilog2(mtt->page_size) - PAGE_SHIFT);
req.cfg0 |=
FIELD_PREP(ERDMA_CMD_CREATE_CQ_PAGESIZE_MASK,
ilog2(mtt->page_size) - ERDMA_HW_PAGE_SHIFT);
if (mtt->mtt_nents == 1) {
req.qbuf_addr_l = lower_32_bits(*(u64 *)mtt->mtt_buf);
req.qbuf_addr_h = upper_32_bits(*(u64 *)mtt->mtt_buf);
@ -636,7 +637,7 @@ static int init_user_qp(struct erdma_qp *qp, struct erdma_ucontext *uctx,
u32 rq_offset;
int ret;
if (len < (PAGE_ALIGN(qp->attrs.sq_size * SQEBB_SIZE) +
if (len < (ALIGN(qp->attrs.sq_size * SQEBB_SIZE, ERDMA_HW_PAGE_SIZE) +
qp->attrs.rq_size * RQE_SIZE))
return -EINVAL;
@ -646,7 +647,7 @@ static int init_user_qp(struct erdma_qp *qp, struct erdma_ucontext *uctx,
if (ret)
return ret;
rq_offset = PAGE_ALIGN(qp->attrs.sq_size << SQEBB_SHIFT);
rq_offset = ALIGN(qp->attrs.sq_size << SQEBB_SHIFT, ERDMA_HW_PAGE_SIZE);
qp->user_qp.rq_offset = rq_offset;
ret = get_mtt_entries(qp->dev, &qp->user_qp.rq_mtt, va + rq_offset,

18
drivers/infiniband/hw/hfi1/chip.c
View File

@ -12135,7 +12135,7 @@ void hfi1_rcvctrl(struct hfi1_devdata *dd, unsigned int op,
set_intr_bits(dd, IS_RCVURGENT_START + rcd->ctxt,
IS_RCVURGENT_START + rcd->ctxt, false);
hfi1_cdbg(RCVCTRL, "ctxt %d rcvctrl 0x%llx\n", ctxt, rcvctrl);
hfi1_cdbg(RCVCTRL, "ctxt %d rcvctrl 0x%llx", ctxt, rcvctrl);
write_kctxt_csr(dd, ctxt, RCV_CTXT_CTRL, rcvctrl);
/* work around sticky RcvCtxtStatus.BlockedRHQFull */
@ -12205,10 +12205,10 @@ u32 hfi1_read_cntrs(struct hfi1_devdata *dd, char **namep, u64 **cntrp)
hfi1_cdbg(CNTR, "reading %s", entry->name);
if (entry->flags & CNTR_DISABLED) {
/* Nothing */
hfi1_cdbg(CNTR, "\tDisabled\n");
hfi1_cdbg(CNTR, "\tDisabled");
} else {
if (entry->flags & CNTR_VL) {
hfi1_cdbg(CNTR, "\tPer VL\n");
hfi1_cdbg(CNTR, "\tPer VL");
for (j = 0; j < C_VL_COUNT; j++) {
val = entry->rw_cntr(entry,
dd, j,
@ -12216,21 +12216,21 @@ u32 hfi1_read_cntrs(struct hfi1_devdata *dd, char **namep, u64 **cntrp)
0);
hfi1_cdbg(
CNTR,
"\t\tRead 0x%llx for %d\n",
"\t\tRead 0x%llx for %d",
val, j);
dd->cntrs[entry->offset + j] =
val;
}
} else if (entry->flags & CNTR_SDMA) {
hfi1_cdbg(CNTR,
"\t Per SDMA Engine\n");
"\t Per SDMA Engine");
for (j = 0; j < chip_sdma_engines(dd);
j++) {
val =
entry->rw_cntr(entry, dd, j,
CNTR_MODE_R, 0);
hfi1_cdbg(CNTR,
"\t\tRead 0x%llx for %d\n",
"\t\tRead 0x%llx for %d",
val, j);
dd->cntrs[entry->offset + j] =
val;
@ -12271,7 +12271,7 @@ u32 hfi1_read_portcntrs(struct hfi1_pportdata *ppd, char **namep, u64 **cntrp)
hfi1_cdbg(CNTR, "reading %s", entry->name);
if (entry->flags & CNTR_DISABLED) {
/* Nothing */
hfi1_cdbg(CNTR, "\tDisabled\n");
hfi1_cdbg(CNTR, "\tDisabled");
continue;
}
@ -12513,7 +12513,7 @@ static void do_update_synth_timer(struct work_struct *work)
hfi1_cdbg(
CNTR,
"[%d] curr tx=0x%llx rx=0x%llx :: last tx=0x%llx rx=0x%llx\n",
"[%d] curr tx=0x%llx rx=0x%llx :: last tx=0x%llx rx=0x%llx",
dd->unit, cur_tx, cur_rx, dd->last_tx, dd->last_rx);
if ((cur_tx < dd->last_tx) || (cur_rx < dd->last_rx)) {
@ -12527,7 +12527,7 @@ static void do_update_synth_timer(struct work_struct *work)
} else {
total_flits = (cur_tx - dd->last_tx) + (cur_rx - dd->last_rx);
hfi1_cdbg(CNTR,
"[%d] total flits 0x%llx limit 0x%llx\n", dd->unit,
"[%d] total flits 0x%llx limit 0x%llx", dd->unit,
total_flits, (u64)CNTR_32BIT_MAX);
if (total_flits >= CNTR_32BIT_MAX) {
hfi1_cdbg(CNTR, "[%d] 32bit limit hit, updating",

2
drivers/infiniband/hw/hfi1/driver.c
View File

@ -1597,7 +1597,7 @@ static int hfi1_setup_bypass_packet(struct hfi1_packet *packet)
return 0;
drop:
hfi1_cdbg(PKT, "%s: packet dropped\n", __func__);
hfi1_cdbg(PKT, "%s: packet dropped", __func__);
ibp->rvp.n_pkt_drops++;
return -EINVAL;
}

2
drivers/infiniband/hw/hfi1/file_ops.c
View File

@ -975,7 +975,7 @@ static int allocate_ctxt(struct hfi1_filedata *fd, struct hfi1_devdata *dd,
ret = -ENOMEM;
goto ctxdata_free;
}
hfi1_cdbg(PROC, "allocated send context %u(%u)\n", uctxt->sc->sw_index,
hfi1_cdbg(PROC, "allocated send context %u(%u)", uctxt->sc->sw_index,
uctxt->sc->hw_context);
ret = sc_enable(uctxt->sc);
if (ret)

12
drivers/infiniband/hw/hfi1/init.c
View File

@ -342,7 +342,7 @@ int hfi1_create_ctxtdata(struct hfi1_pportdata *ppd, int numa,
INIT_LIST_HEAD(&rcd->flow_queue.queue_head);
INIT_LIST_HEAD(&rcd->rarr_queue.queue_head);
hfi1_cdbg(PROC, "setting up context %u\n", rcd->ctxt);
hfi1_cdbg(PROC, "setting up context %u", rcd->ctxt);
/*
* Calculate the context's RcvArray entry starting point.
@ -400,7 +400,7 @@ int hfi1_create_ctxtdata(struct hfi1_pportdata *ppd, int numa,
rcd->egrbufs.count = MAX_EAGER_ENTRIES;
}
hfi1_cdbg(PROC,
"ctxt%u: max Eager buffer RcvArray entries: %u\n",
"ctxt%u: max Eager buffer RcvArray entries: %u",
rcd->ctxt, rcd->egrbufs.count);
/*
@ -432,7 +432,7 @@ int hfi1_create_ctxtdata(struct hfi1_pportdata *ppd, int numa,
if (rcd->egrbufs.size < hfi1_max_mtu) {
rcd->egrbufs.size = __roundup_pow_of_two(hfi1_max_mtu);
hfi1_cdbg(PROC,
"ctxt%u: eager bufs size too small. Adjusting to %u\n",
"ctxt%u: eager bufs size too small. Adjusting to %u",
rcd->ctxt, rcd->egrbufs.size);
}
rcd->egrbufs.rcvtid_size = HFI1_MAX_EAGER_BUFFER_SIZE;
@ -1920,7 +1920,7 @@ int hfi1_setup_eagerbufs(struct hfi1_ctxtdata *rcd)
rcd->egrbufs.size = alloced_bytes;
hfi1_cdbg(PROC,
"ctxt%u: Alloced %u rcv tid entries @ %uKB, total %uKB\n",
"ctxt%u: Alloced %u rcv tid entries @ %uKB, total %uKB",
rcd->ctxt, rcd->egrbufs.alloced,
rcd->egrbufs.rcvtid_size / 1024, rcd->egrbufs.size / 1024);
@ -1943,13 +1943,13 @@ int hfi1_setup_eagerbufs(struct hfi1_ctxtdata *rcd)
rcd->expected_count = MAX_TID_PAIR_ENTRIES * 2;
rcd->expected_base = rcd->eager_base + egrtop;
hfi1_cdbg(PROC, "ctxt%u: eager:%u, exp:%u, egrbase:%u, expbase:%u\n",
hfi1_cdbg(PROC, "ctxt%u: eager:%u, exp:%u, egrbase:%u, expbase:%u",
rcd->ctxt, rcd->egrbufs.alloced, rcd->expected_count,
rcd->eager_base, rcd->expected_base);
if (!hfi1_rcvbuf_validate(rcd->egrbufs.rcvtid_size, PT_EAGER, &order)) {
hfi1_cdbg(PROC,
"ctxt%u: current Eager buffer size is invalid %u\n",
"ctxt%u: current Eager buffer size is invalid %u",
rcd->ctxt, rcd->egrbufs.rcvtid_size);
ret = -EINVAL;
goto bail_rcvegrbuf_phys;

6
drivers/infiniband/hw/hfi1/ipoib_tx.c
View File

@ -215,6 +215,7 @@ static int hfi1_ipoib_build_ulp_payload(struct ipoib_txreq *tx,
const skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
ret = sdma_txadd_page(dd,
NULL,
txreq,
skb_frag_page(frag),
frag->bv_offset,
@ -737,10 +738,13 @@ int hfi1_ipoib_txreq_init(struct hfi1_ipoib_dev_priv *priv)
txq->tx_ring.shift = ilog2(tx_item_size);
txq->tx_ring.avail = hfi1_ipoib_ring_hwat(txq);
tx_ring = &txq->tx_ring;
for (j = 0; j < tx_ring_size; j++)
for (j = 0; j < tx_ring_size; j++) {
hfi1_txreq_from_idx(tx_ring, j)->sdma_hdr =
kzalloc_node(sizeof(*tx->sdma_hdr),
GFP_KERNEL, priv->dd->node);
if (!hfi1_txreq_from_idx(tx_ring, j)->sdma_hdr)
goto free_txqs;
}
netif_napi_add_tx(dev, &txq->napi, hfi1_ipoib_poll_tx_ring);
}

84
drivers/infiniband/hw/hfi1/mmu_rb.c
View File

@ -46,12 +46,14 @@ int hfi1_mmu_rb_register(void *ops_arg,
struct mmu_rb_handler **handler)
{
struct mmu_rb_handler *h;
void *free_ptr;
int ret;
h = kzalloc(sizeof(*h), GFP_KERNEL);
if (!h)
free_ptr = kzalloc(sizeof(*h) + cache_line_size() - 1, GFP_KERNEL);
if (!free_ptr)
return -ENOMEM;
h = PTR_ALIGN(free_ptr, cache_line_size());
h->root = RB_ROOT_CACHED;
h->ops = ops;
h->ops_arg = ops_arg;
@ -62,10 +64,11 @@ int hfi1_mmu_rb_register(void *ops_arg,
INIT_LIST_HEAD(&h->del_list);
INIT_LIST_HEAD(&h->lru_list);
h->wq = wq;
h->free_ptr = free_ptr;
ret = mmu_notifier_register(&h->mn, current->mm);
if (ret) {
kfree(h);
kfree(free_ptr);
return ret;
}
@ -108,7 +111,7 @@ void hfi1_mmu_rb_unregister(struct mmu_rb_handler *handler)
/* Now the mm may be freed. */
mmdrop(handler->mn.mm);
kfree(handler);
kfree(handler->free_ptr);
}
int hfi1_mmu_rb_insert(struct mmu_rb_handler *handler,
@ -126,11 +129,11 @@ int hfi1_mmu_rb_insert(struct mmu_rb_handler *handler,
spin_lock_irqsave(&handler->lock, flags);
node = __mmu_rb_search(handler, mnode->addr, mnode->len);
if (node) {
ret = -EINVAL;
ret = -EEXIST;
goto unlock;
}
__mmu_int_rb_insert(mnode, &handler->root);
list_add(&mnode->list, &handler->lru_list);
list_add_tail(&mnode->list, &handler->lru_list);
ret = handler->ops->insert(handler->ops_arg, mnode);
if (ret) {
@ -143,6 +146,19 @@ unlock:
return ret;
}
/* Caller must hold handler lock */
struct mmu_rb_node *hfi1_mmu_rb_get_first(struct mmu_rb_handler *handler,
unsigned long addr, unsigned long len)
{
struct mmu_rb_node *node;
trace_hfi1_mmu_rb_search(addr, len);
node = __mmu_int_rb_iter_first(&handler->root, addr, (addr + len) - 1);
if (node)
list_move_tail(&node->list, &handler->lru_list);
return node;
}
/* Caller must hold handler lock */
static struct mmu_rb_node *__mmu_rb_search(struct mmu_rb_handler *handler,
unsigned long addr,
@ -167,32 +183,6 @@ static struct mmu_rb_node *__mmu_rb_search(struct mmu_rb_handler *handler,
return node;
}
bool hfi1_mmu_rb_remove_unless_exact(struct mmu_rb_handler *handler,
unsigned long addr, unsigned long len,
struct mmu_rb_node **rb_node)
{
struct mmu_rb_node *node;
unsigned long flags;
bool ret = false;
if (current->mm != handler->mn.mm)
return ret;
spin_lock_irqsave(&handler->lock, flags);
node = __mmu_rb_search(handler, addr, len);
if (node) {
if (node->addr == addr && node->len == len)
goto unlock;
__mmu_int_rb_remove(node, &handler->root);
list_del(&node->list); /* remove from LRU list */
ret = true;
}
unlock:
spin_unlock_irqrestore(&handler->lock, flags);
*rb_node = node;
return ret;
}
void hfi1_mmu_rb_evict(struct mmu_rb_handler *handler, void *evict_arg)
{
struct mmu_rb_node *rbnode, *ptr;
@ -206,8 +196,7 @@ void hfi1_mmu_rb_evict(struct mmu_rb_handler *handler, void *evict_arg)
INIT_LIST_HEAD(&del_list);
spin_lock_irqsave(&handler->lock, flags);
list_for_each_entry_safe_reverse(rbnode, ptr, &handler->lru_list,
list) {
list_for_each_entry_safe(rbnode, ptr, &handler->lru_list, list) {
if (handler->ops->evict(handler->ops_arg, rbnode, evict_arg,
&stop)) {
__mmu_int_rb_remove(rbnode, &handler->root);
@ -219,36 +208,11 @@ void hfi1_mmu_rb_evict(struct mmu_rb_handler *handler, void *evict_arg)
}
spin_unlock_irqrestore(&handler->lock, flags);
while (!list_empty(&del_list)) {
rbnode = list_first_entry(&del_list, struct mmu_rb_node, list);
list_del(&rbnode->list);
list_for_each_entry_safe(rbnode, ptr, &del_list, list) {
handler->ops->remove(handler->ops_arg, rbnode);
}
}
/*
* It is up to the caller to ensure that this function does not race with the
* mmu invalidate notifier which may be calling the users remove callback on
* 'node'.
*/
void hfi1_mmu_rb_remove(struct mmu_rb_handler *handler,
struct mmu_rb_node *node)
{
unsigned long flags;
if (current->mm != handler->mn.mm)
return;
/* Validity of handler and node pointers has been checked by caller. */
trace_hfi1_mmu_rb_remove(node->addr, node->len);
spin_lock_irqsave(&handler->lock, flags);
__mmu_int_rb_remove(node, &handler->root);
list_del(&node->list); /* remove from LRU list */
spin_unlock_irqrestore(&handler->lock, flags);
handler->ops->remove(handler->ops_arg, node);
}
static int mmu_notifier_range_start(struct mmu_notifier *mn,
const struct mmu_notifier_range *range)
{

22
drivers/infiniband/hw/hfi1/mmu_rb.h
View File

@ -33,15 +33,25 @@ struct mmu_rb_ops {
};
struct mmu_rb_handler {
/*
* struct mmu_notifier is 56 bytes, and spinlock_t is 4 bytes, so
* they fit together in one cache line. mn is relatively rarely
* accessed, so co-locating the spinlock with it achieves much of
* the cacheline contention reduction of giving the spinlock its own
* cacheline without the overhead of doing so.
*/
struct mmu_notifier mn;
struct rb_root_cached root;
void *ops_arg;
spinlock_t lock; /* protect the RB tree */
/* Begin on a new cachline boundary here */
struct rb_root_cached root ____cacheline_aligned_in_smp;
void *ops_arg;
struct mmu_rb_ops *ops;
struct list_head lru_list;
struct work_struct del_work;
struct list_head del_list;
struct workqueue_struct *wq;
void *free_ptr;
};
int hfi1_mmu_rb_register(void *ops_arg,
@ -52,10 +62,8 @@ void hfi1_mmu_rb_unregister(struct mmu_rb_handler *handler);
int hfi1_mmu_rb_insert(struct mmu_rb_handler *handler,
struct mmu_rb_node *mnode);
void hfi1_mmu_rb_evict(struct mmu_rb_handler *handler, void *evict_arg);
void hfi1_mmu_rb_remove(struct mmu_rb_handler *handler,
struct mmu_rb_node *mnode);
bool hfi1_mmu_rb_remove_unless_exact(struct mmu_rb_handler *handler,
unsigned long addr, unsigned long len,
struct mmu_rb_node **rb_node);
struct mmu_rb_node *hfi1_mmu_rb_get_first(struct mmu_rb_handler *handler,
unsigned long addr,
unsigned long len);
#endif /* _HFI1_MMU_RB_H */

2
drivers/infiniband/hw/hfi1/pcie.c
View File

@ -7,7 +7,6 @@
#include <linux/io.h>
#include <linux/delay.h>
#include <linux/vmalloc.h>
#include <linux/aer.h>
#include <linux/module.h>
#include "hfi.h"
@ -65,7 +64,6 @@ int hfi1_pcie_init(struct hfi1_devdata *dd)
}
pci_set_master(pdev);
(void)pci_enable_pcie_error_reporting(pdev);
return 0;
bail:

2
drivers/infiniband/hw/hfi1/pio.c
View File

@ -820,7 +820,7 @@ struct send_context *sc_alloc(struct hfi1_devdata *dd, int type,
}
hfi1_cdbg(PIO,
"Send context %u(%u) %s group %u credits %u credit_ctrl 0x%llx threshold %u\n",
"Send context %u(%u) %s group %u credits %u credit_ctrl 0x%llx threshold %u",
sw_index,
hw_context,
sc_type_name(type),

21
drivers/infiniband/hw/hfi1/sdma.c
View File

@ -1593,22 +1593,7 @@ static inline void sdma_unmap_desc(
struct hfi1_devdata *dd,
struct sdma_desc *descp)
{
switch (sdma_mapping_type(descp)) {
case SDMA_MAP_SINGLE:
dma_unmap_single(
&dd->pcidev->dev,
sdma_mapping_addr(descp),
sdma_mapping_len(descp),
DMA_TO_DEVICE);
break;
case SDMA_MAP_PAGE:
dma_unmap_page(
&dd->pcidev->dev,
sdma_mapping_addr(descp),
sdma_mapping_len(descp),
DMA_TO_DEVICE);
break;
}
system_descriptor_complete(dd, descp);
}
/*
@ -3128,7 +3113,7 @@ int ext_coal_sdma_tx_descs(struct hfi1_devdata *dd, struct sdma_txreq *tx,
/* Add descriptor for coalesce buffer */
tx->desc_limit = MAX_DESC;
return _sdma_txadd_daddr(dd, SDMA_MAP_SINGLE, tx,
return _sdma_txadd_daddr(dd, SDMA_MAP_SINGLE, NULL, tx,
addr, tx->tlen);
}
@ -3167,10 +3152,12 @@ int _pad_sdma_tx_descs(struct hfi1_devdata *dd, struct sdma_txreq *tx)
return rval;
}
}
/* finish the one just added */
make_tx_sdma_desc(
tx,
SDMA_MAP_NONE,
NULL,
dd->sdma_pad_phys,
sizeof(u32) - (tx->packet_len & (sizeof(u32) - 1)));
tx->num_desc++;

16
drivers/infiniband/hw/hfi1/sdma.h
View File

@ -594,6 +594,7 @@ static inline dma_addr_t sdma_mapping_addr(struct sdma_desc *d)
static inline void make_tx_sdma_desc(
struct sdma_txreq *tx,
int type,
void *pinning_ctx,
dma_addr_t addr,
size_t len)
{
@ -612,6 +613,7 @@ static inline void make_tx_sdma_desc(
<< SDMA_DESC0_PHY_ADDR_SHIFT) |
(((u64)len & SDMA_DESC0_BYTE_COUNT_MASK)
<< SDMA_DESC0_BYTE_COUNT_SHIFT);
desc->pinning_ctx = pinning_ctx;
}
/* helper to extend txreq */
@ -643,6 +645,7 @@ static inline void _sdma_close_tx(struct hfi1_devdata *dd,
static inline int _sdma_txadd_daddr(
struct hfi1_devdata *dd,
int type,
void *pinning_ctx,
struct sdma_txreq *tx,
dma_addr_t addr,
u16 len)
@ -652,6 +655,7 @@ static inline int _sdma_txadd_daddr(
make_tx_sdma_desc(
tx,
type,
pinning_ctx,
addr, len);
WARN_ON(len > tx->tlen);
tx->num_desc++;
@ -672,6 +676,7 @@ static inline int _sdma_txadd_daddr(
/**
* sdma_txadd_page() - add a page to the sdma_txreq
* @dd: the device to use for mapping
* @pinning_ctx: context to be released at descriptor retirement
* @tx: tx request to which the page is added
* @page: page to map
* @offset: offset within the page
@ -687,6 +692,7 @@ static inline int _sdma_txadd_daddr(
*/
static inline int sdma_txadd_page(
struct hfi1_devdata *dd,
void *pinning_ctx,
struct sdma_txreq *tx,
struct page *page,
unsigned long offset,
@ -714,8 +720,7 @@ static inline int sdma_txadd_page(
return -ENOSPC;
}
return _sdma_txadd_daddr(
dd, SDMA_MAP_PAGE, tx, addr, len);
return _sdma_txadd_daddr(dd, SDMA_MAP_PAGE, pinning_ctx, tx, addr, len);
}
/**
@ -749,7 +754,8 @@ static inline int sdma_txadd_daddr(
return rval;
}
return _sdma_txadd_daddr(dd, SDMA_MAP_NONE, tx, addr, len);
return _sdma_txadd_daddr(dd, SDMA_MAP_NONE, NULL, tx,
addr, len);
}
/**
@ -795,8 +801,7 @@ static inline int sdma_txadd_kvaddr(
return -ENOSPC;
}
return _sdma_txadd_daddr(
dd, SDMA_MAP_SINGLE, tx, addr, len);
return _sdma_txadd_daddr(dd, SDMA_MAP_SINGLE, NULL, tx, addr, len);
}
struct iowait_work;
@ -1030,4 +1035,5 @@ extern uint mod_num_sdma;
void sdma_update_lmc(struct hfi1_devdata *dd, u64 mask, u32 lid);
void system_descriptor_complete(struct hfi1_devdata *dd, struct sdma_desc *descp);
#endif

1
drivers/infiniband/hw/hfi1/sdma_txreq.h
View File

@ -19,6 +19,7 @@
struct sdma_desc {
/* private: don't use directly */
u64 qw[2];
void *pinning_ctx;
};
/**

7
drivers/infiniband/hw/hfi1/trace_dbg.h
View File

@ -22,6 +22,11 @@
#define MAX_MSG_LEN 512
#pragma GCC diagnostic push
#ifndef __clang__
#pragma GCC diagnostic ignored "-Wsuggest-attribute=format"
#endif
DECLARE_EVENT_CLASS(hfi1_trace_template,
TP_PROTO(const char *function, struct va_format *vaf),
TP_ARGS(function, vaf),
@ -36,6 +41,8 @@ DECLARE_EVENT_CLASS(hfi1_trace_template,
__get_str(msg))
);
#pragma GCC diagnostic pop
/*
* It may be nice to macroize the __hfi1_trace but the va_* stuff requires an
* actual function to work and can not be in a macro.

4
drivers/infiniband/hw/hfi1/trace_mmu.h
View File

@ -37,10 +37,6 @@ DEFINE_EVENT(hfi1_mmu_rb_template, hfi1_mmu_rb_search,
TP_PROTO(unsigned long addr, unsigned long len),
TP_ARGS(addr, len));
DEFINE_EVENT(hfi1_mmu_rb_template, hfi1_mmu_rb_remove,
TP_PROTO(unsigned long addr, unsigned long len),
TP_ARGS(addr, len));
DEFINE_EVENT(hfi1_mmu_rb_template, hfi1_mmu_mem_invalidate,
TP_PROTO(unsigned long addr, unsigned long len),
TP_ARGS(addr, len));

600
drivers/infiniband/hw/hfi1/user_sdma.c
View File

@ -24,7 +24,6 @@
#include "hfi.h"
#include "sdma.h"
#include "mmu_rb.h"
#include "user_sdma.h"
#include "verbs.h" /* for the headers */
#include "common.h" /* for struct hfi1_tid_info */
@ -39,11 +38,7 @@ static unsigned initial_pkt_count = 8;
static int user_sdma_send_pkts(struct user_sdma_request *req, u16 maxpkts);
static void user_sdma_txreq_cb(struct sdma_txreq *txreq, int status);
static inline void pq_update(struct hfi1_user_sdma_pkt_q *pq);
static void user_sdma_free_request(struct user_sdma_request *req, bool unpin);
static int pin_vector_pages(struct user_sdma_request *req,
struct user_sdma_iovec *iovec);
static void unpin_vector_pages(struct mm_struct *mm, struct page **pages,
unsigned start, unsigned npages);
static void user_sdma_free_request(struct user_sdma_request *req);
static int check_header_template(struct user_sdma_request *req,
struct hfi1_pkt_header *hdr, u32 lrhlen,
u32 datalen);
@ -81,6 +76,11 @@ static struct mmu_rb_ops sdma_rb_ops = {
.invalidate = sdma_rb_invalidate
};
static int add_system_pages_to_sdma_packet(struct user_sdma_request *req,
struct user_sdma_txreq *tx,
struct user_sdma_iovec *iovec,
u32 *pkt_remaining);
static int defer_packet_queue(
struct sdma_engine *sde,
struct iowait_work *wait,
@ -410,6 +410,7 @@ int hfi1_user_sdma_process_request(struct hfi1_filedata *fd,
ret = -EINVAL;
goto free_req;
}
/* Copy the header from the user buffer */
ret = copy_from_user(&req->hdr, iovec[idx].iov_base + sizeof(info),
sizeof(req->hdr));
@ -484,9 +485,8 @@ int hfi1_user_sdma_process_request(struct hfi1_filedata *fd,
memcpy(&req->iovs[i].iov,
iovec + idx++,
sizeof(req->iovs[i].iov));
ret = pin_vector_pages(req, &req->iovs[i]);
if (ret) {
req->data_iovs = i;
if (req->iovs[i].iov.iov_len == 0) {
ret = -EINVAL;
goto free_req;
}
req->data_len += req->iovs[i].iov.iov_len;
@ -584,7 +584,7 @@ free_req:
if (req->seqsubmitted)
wait_event(pq->busy.wait_dma,
(req->seqcomp == req->seqsubmitted - 1));
user_sdma_free_request(req, true);
user_sdma_free_request(req);
pq_update(pq);
set_comp_state(pq, cq, info.comp_idx, ERROR, ret);
}
@ -696,48 +696,6 @@ static int user_sdma_txadd_ahg(struct user_sdma_request *req,
return ret;
}
static int user_sdma_txadd(struct user_sdma_request *req,
struct user_sdma_txreq *tx,
struct user_sdma_iovec *iovec, u32 datalen,
u32 *queued_ptr, u32 *data_sent_ptr,
u64 *iov_offset_ptr)
{
int ret;
unsigned int pageidx, len;
unsigned long base, offset;
u64 iov_offset = *iov_offset_ptr;
u32 queued = *queued_ptr, data_sent = *data_sent_ptr;
struct hfi1_user_sdma_pkt_q *pq = req->pq;
base = (unsigned long)iovec->iov.iov_base;
offset = offset_in_page(base + iovec->offset + iov_offset);
pageidx = (((iovec->offset + iov_offset + base) - (base & PAGE_MASK)) >>
PAGE_SHIFT);
len = offset + req->info.fragsize > PAGE_SIZE ?
PAGE_SIZE - offset : req->info.fragsize;
len = min((datalen - queued), len);
ret = sdma_txadd_page(pq->dd, &tx->txreq, iovec->pages[pageidx],
offset, len);
if (ret) {
SDMA_DBG(req, "SDMA txreq add page failed %d\n", ret);
return ret;
}
iov_offset += len;
queued += len;
data_sent += len;
if (unlikely(queued < datalen && pageidx == iovec->npages &&
req->iov_idx < req->data_iovs - 1)) {
iovec->offset += iov_offset;
iovec = &req->iovs[++req->iov_idx];
iov_offset = 0;
}
*queued_ptr = queued;
*data_sent_ptr = data_sent;
*iov_offset_ptr = iov_offset;
return ret;
}
static int user_sdma_send_pkts(struct user_sdma_request *req, u16 maxpkts)
{
int ret = 0;
@ -769,8 +727,7 @@ static int user_sdma_send_pkts(struct user_sdma_request *req, u16 maxpkts)
maxpkts = req->info.npkts - req->seqnum;
while (npkts < maxpkts) {
u32 datalen = 0, queued = 0, data_sent = 0;
u64 iov_offset = 0;
u32 datalen = 0;
/*
* Check whether any of the completions have come back
@ -863,27 +820,17 @@ static int user_sdma_send_pkts(struct user_sdma_request *req, u16 maxpkts)
goto free_txreq;
}
/*
* If the request contains any data vectors, add up to
* fragsize bytes to the descriptor.
*/
while (queued < datalen &&
(req->sent + data_sent) < req->data_len) {
ret = user_sdma_txadd(req, tx, iovec, datalen,
&queued, &data_sent, &iov_offset);
if (ret)
goto free_txreq;
}
/*
* The txreq was submitted successfully so we can update
* the counters.
*/
req->koffset += datalen;
if (req_opcode(req->info.ctrl) == EXPECTED)
req->tidoffset += datalen;
req->sent += data_sent;
if (req->data_len)
iovec->offset += iov_offset;
req->sent += datalen;
while (datalen) {
ret = add_system_pages_to_sdma_packet(req, tx, iovec,
&datalen);
if (ret)
goto free_txreq;
iovec = &req->iovs[req->iov_idx];
}
list_add_tail(&tx->txreq.list, &req->txps);
/*
* It is important to increment this here as it is used to
@ -920,133 +867,14 @@ free_tx:
static u32 sdma_cache_evict(struct hfi1_user_sdma_pkt_q *pq, u32 npages)
{
struct evict_data evict_data;
struct mmu_rb_handler *handler = pq->handler;
evict_data.cleared = 0;
evict_data.target = npages;
hfi1_mmu_rb_evict(pq->handler, &evict_data);
hfi1_mmu_rb_evict(handler, &evict_data);
return evict_data.cleared;
}
static int pin_sdma_pages(struct user_sdma_request *req,
struct user_sdma_iovec *iovec,
struct sdma_mmu_node *node,
int npages)
{
int pinned, cleared;
struct page **pages;
struct hfi1_user_sdma_pkt_q *pq = req->pq;
pages = kcalloc(npages, sizeof(*pages), GFP_KERNEL);
if (!pages)
return -ENOMEM;
memcpy(pages, node->pages, node->npages * sizeof(*pages));
npages -= node->npages;
retry:
if (!hfi1_can_pin_pages(pq->dd, current->mm,
atomic_read(&pq->n_locked), npages)) {
cleared = sdma_cache_evict(pq, npages);
if (cleared >= npages)
goto retry;
}
pinned = hfi1_acquire_user_pages(current->mm,
((unsigned long)iovec->iov.iov_base +
(node->npages * PAGE_SIZE)), npages, 0,
pages + node->npages);
if (pinned < 0) {
kfree(pages);
return pinned;
}
if (pinned != npages) {
unpin_vector_pages(current->mm, pages, node->npages, pinned);
return -EFAULT;
}
kfree(node->pages);
node->rb.len = iovec->iov.iov_len;
node->pages = pages;
atomic_add(pinned, &pq->n_locked);
return pinned;
}
static void unpin_sdma_pages(struct sdma_mmu_node *node)
{
if (node->npages) {
unpin_vector_pages(mm_from_sdma_node(node), node->pages, 0,
node->npages);
atomic_sub(node->npages, &node->pq->n_locked);
}
}
static int pin_vector_pages(struct user_sdma_request *req,
struct user_sdma_iovec *iovec)
{
int ret = 0, pinned, npages;
struct hfi1_user_sdma_pkt_q *pq = req->pq;
struct sdma_mmu_node *node = NULL;
struct mmu_rb_node *rb_node;
struct iovec *iov;
bool extracted;
extracted =
hfi1_mmu_rb_remove_unless_exact(pq->handler,
(unsigned long)
iovec->iov.iov_base,
iovec->iov.iov_len, &rb_node);
if (rb_node) {
node = container_of(rb_node, struct sdma_mmu_node, rb);
if (!extracted) {
atomic_inc(&node->refcount);
iovec->pages = node->pages;
iovec->npages = node->npages;
iovec->node = node;
return 0;
}
}
if (!node) {
node = kzalloc(sizeof(*node), GFP_KERNEL);
if (!node)
return -ENOMEM;
node->rb.addr = (unsigned long)iovec->iov.iov_base;
node->pq = pq;
atomic_set(&node->refcount, 0);
}
iov = &iovec->iov;
npages = num_user_pages((unsigned long)iov->iov_base, iov->iov_len);
if (node->npages < npages) {
pinned = pin_sdma_pages(req, iovec, node, npages);
if (pinned < 0) {
ret = pinned;
goto bail;
}
node->npages += pinned;
npages = node->npages;
}
iovec->pages = node->pages;
iovec->npages = npages;
iovec->node = node;
ret = hfi1_mmu_rb_insert(req->pq->handler, &node->rb);
if (ret) {
iovec->node = NULL;
goto bail;
}
return 0;
bail:
unpin_sdma_pages(node);
kfree(node);
return ret;
}
static void unpin_vector_pages(struct mm_struct *mm, struct page **pages,
unsigned start, unsigned npages)
{
hfi1_release_user_pages(mm, pages + start, npages, false);
kfree(pages);
}
static int check_header_template(struct user_sdma_request *req,
struct hfi1_pkt_header *hdr, u32 lrhlen,
u32 datalen)
@ -1388,7 +1216,7 @@ static void user_sdma_txreq_cb(struct sdma_txreq *txreq, int status)
if (req->seqcomp != req->info.npkts - 1)
return;
user_sdma_free_request(req, false);
user_sdma_free_request(req);
set_comp_state(pq, cq, req->info.comp_idx, state, status);
pq_update(pq);
}
@ -1399,10 +1227,8 @@ static inline void pq_update(struct hfi1_user_sdma_pkt_q *pq)
wake_up(&pq->wait);
}
static void user_sdma_free_request(struct user_sdma_request *req, bool unpin)
static void user_sdma_free_request(struct user_sdma_request *req)
{
int i;
if (!list_empty(&req->txps)) {
struct sdma_txreq *t, *p;
@ -1415,21 +1241,6 @@ static void user_sdma_free_request(struct user_sdma_request *req, bool unpin)
}
}
for (i = 0; i < req->data_iovs; i++) {
struct sdma_mmu_node *node = req->iovs[i].node;
if (!node)
continue;
req->iovs[i].node = NULL;
if (unpin)
hfi1_mmu_rb_remove(req->pq->handler,
&node->rb);
else
atomic_dec(&node->refcount);
}
kfree(req->tids);
clear_bit(req->info.comp_idx, req->pq->req_in_use);
}
@ -1447,6 +1258,368 @@ static inline void set_comp_state(struct hfi1_user_sdma_pkt_q *pq,
idx, state, ret);
}
static void unpin_vector_pages(struct mm_struct *mm, struct page **pages,
unsigned int start, unsigned int npages)
{
hfi1_release_user_pages(mm, pages + start, npages, false);
kfree(pages);
}
static void free_system_node(struct sdma_mmu_node *node)
{
if (node->npages) {
unpin_vector_pages(mm_from_sdma_node(node), node->pages, 0,
node->npages);
atomic_sub(node->npages, &node->pq->n_locked);
}
kfree(node);
}
static inline void acquire_node(struct sdma_mmu_node *node)
{
atomic_inc(&node->refcount);
WARN_ON(atomic_read(&node->refcount) < 0);
}
static inline void release_node(struct mmu_rb_handler *handler,
struct sdma_mmu_node *node)
{
atomic_dec(&node->refcount);
WARN_ON(atomic_read(&node->refcount) < 0);
}
static struct sdma_mmu_node *find_system_node(struct mmu_rb_handler *handler,
unsigned long start,
unsigned long end)
{
struct mmu_rb_node *rb_node;
struct sdma_mmu_node *node;
unsigned long flags;
spin_lock_irqsave(&handler->lock, flags);
rb_node = hfi1_mmu_rb_get_first(handler, start, (end - start));
if (!rb_node) {
spin_unlock_irqrestore(&handler->lock, flags);
return NULL;
}
node = container_of(rb_node, struct sdma_mmu_node, rb);
acquire_node(node);
spin_unlock_irqrestore(&handler->lock, flags);
return node;
}
static int pin_system_pages(struct user_sdma_request *req,
uintptr_t start_address, size_t length,
struct sdma_mmu_node *node, int npages)
{
struct hfi1_user_sdma_pkt_q *pq = req->pq;
int pinned, cleared;
struct page **pages;
pages = kcalloc(npages, sizeof(*pages), GFP_KERNEL);
if (!pages)
return -ENOMEM;
retry:
if (!hfi1_can_pin_pages(pq->dd, current->mm, atomic_read(&pq->n_locked),
npages)) {
SDMA_DBG(req, "Evicting: nlocked %u npages %u",
atomic_read(&pq->n_locked), npages);
cleared = sdma_cache_evict(pq, npages);
if (cleared >= npages)
goto retry;
}
SDMA_DBG(req, "Acquire user pages start_address %lx node->npages %u npages %u",
start_address, node->npages, npages);
pinned = hfi1_acquire_user_pages(current->mm, start_address, npages, 0,
pages);
if (pinned < 0) {
kfree(pages);
SDMA_DBG(req, "pinned %d", pinned);
return pinned;
}
if (pinned != npages) {
unpin_vector_pages(current->mm, pages, node->npages, pinned);
SDMA_DBG(req, "npages %u pinned %d", npages, pinned);
return -EFAULT;
}
node->rb.addr = start_address;
node->rb.len = length;
node->pages = pages;
node->npages = npages;
atomic_add(pinned, &pq->n_locked);
SDMA_DBG(req, "done. pinned %d", pinned);
return 0;
}
static int add_system_pinning(struct user_sdma_request *req,
struct sdma_mmu_node **node_p,
unsigned long start, unsigned long len)
{
struct hfi1_user_sdma_pkt_q *pq = req->pq;
struct sdma_mmu_node *node;
int ret;
node = kzalloc(sizeof(*node), GFP_KERNEL);
if (!node)
return -ENOMEM;
node->pq = pq;
ret = pin_system_pages(req, start, len, node, PFN_DOWN(len));
if (ret == 0) {
ret = hfi1_mmu_rb_insert(pq->handler, &node->rb);
if (ret)
free_system_node(node);
else
*node_p = node;
return ret;
}
kfree(node);
return ret;
}
static int get_system_cache_entry(struct user_sdma_request *req,
struct sdma_mmu_node **node_p,
size_t req_start, size_t req_len)
{
struct hfi1_user_sdma_pkt_q *pq = req->pq;
u64 start = ALIGN_DOWN(req_start, PAGE_SIZE);
u64 end = PFN_ALIGN(req_start + req_len);
struct mmu_rb_handler *handler = pq->handler;
int ret;
if ((end - start) == 0) {
SDMA_DBG(req,
"Request for empty cache entry req_start %lx req_len %lx start %llx end %llx",
req_start, req_len, start, end);
return -EINVAL;
}
SDMA_DBG(req, "req_start %lx req_len %lu", req_start, req_len);
while (1) {
struct sdma_mmu_node *node =
find_system_node(handler, start, end);
u64 prepend_len = 0;
SDMA_DBG(req, "node %p start %llx end %llu", node, start, end);
if (!node) {
ret = add_system_pinning(req, node_p, start,
end - start);
if (ret == -EEXIST) {
/*
* Another execution context has inserted a
* conficting entry first.
*/
continue;
}
return ret;
}
if (node->rb.addr <= start) {
/*
* This entry covers at least part of the region. If it doesn't extend
* to the end, then this will be called again for the next segment.
*/
*node_p = node;
return 0;
}
SDMA_DBG(req, "prepend: node->rb.addr %lx, node->refcount %d",
node->rb.addr, atomic_read(&node->refcount));
prepend_len = node->rb.addr - start;
/*
* This node will not be returned, instead a new node
* will be. So release the reference.
*/
release_node(handler, node);
/* Prepend a node to cover the beginning of the allocation */
ret = add_system_pinning(req, node_p, start, prepend_len);
if (ret == -EEXIST) {
/* Another execution context has inserted a conficting entry first. */
continue;
}
return ret;
}
}
static int add_mapping_to_sdma_packet(struct user_sdma_request *req,
struct user_sdma_txreq *tx,
struct sdma_mmu_node *cache_entry,
size_t start,
size_t from_this_cache_entry)
{
struct hfi1_user_sdma_pkt_q *pq = req->pq;
unsigned int page_offset;
unsigned int from_this_page;
size_t page_index;
void *ctx;
int ret;
/*
* Because the cache may be more fragmented than the memory that is being accessed,
* it's not strictly necessary to have a descriptor per cache entry.
*/
while (from_this_cache_entry) {
page_index = PFN_DOWN(start - cache_entry->rb.addr);
if (page_index >= cache_entry->npages) {
SDMA_DBG(req,
"Request for page_index %zu >= cache_entry->npages %u",
page_index, cache_entry->npages);
return -EINVAL;
}
page_offset = start - ALIGN_DOWN(start, PAGE_SIZE);
from_this_page = PAGE_SIZE - page_offset;
if (from_this_page < from_this_cache_entry) {
ctx = NULL;
} else {
/*
* In the case they are equal the next line has no practical effect,
* but it's better to do a register to register copy than a conditional
* branch.
*/
from_this_page = from_this_cache_entry;
ctx = cache_entry;
}
ret = sdma_txadd_page(pq->dd, ctx, &tx->txreq,
cache_entry->pages[page_index],
page_offset, from_this_page);
if (ret) {
/*
* When there's a failure, the entire request is freed by
* user_sdma_send_pkts().
*/
SDMA_DBG(req,
"sdma_txadd_page failed %d page_index %lu page_offset %u from_this_page %u",
ret, page_index, page_offset, from_this_page);
return ret;
}
start += from_this_page;
from_this_cache_entry -= from_this_page;
}
return 0;
}
static int add_system_iovec_to_sdma_packet(struct user_sdma_request *req,
struct user_sdma_txreq *tx,
struct user_sdma_iovec *iovec,
size_t from_this_iovec)
{
struct mmu_rb_handler *handler = req->pq->handler;
while (from_this_iovec > 0) {
struct sdma_mmu_node *cache_entry;
size_t from_this_cache_entry;
size_t start;
int ret;
start = (uintptr_t)iovec->iov.iov_base + iovec->offset;
ret = get_system_cache_entry(req, &cache_entry, start,
from_this_iovec);
if (ret) {
SDMA_DBG(req, "pin system segment failed %d", ret);
return ret;
}
from_this_cache_entry = cache_entry->rb.len - (start - cache_entry->rb.addr);
if (from_this_cache_entry > from_this_iovec)
from_this_cache_entry = from_this_iovec;
ret = add_mapping_to_sdma_packet(req, tx, cache_entry, start,
from_this_cache_entry);
if (ret) {
/*
* We're guaranteed that there will be no descriptor
* completion callback that releases this node
* because only the last descriptor referencing it
* has a context attached, and a failure means the
* last descriptor was never added.
*/
release_node(handler, cache_entry);
SDMA_DBG(req, "add system segment failed %d", ret);
return ret;
}
iovec->offset += from_this_cache_entry;
from_this_iovec -= from_this_cache_entry;
}
return 0;
}
static int add_system_pages_to_sdma_packet(struct user_sdma_request *req,
struct user_sdma_txreq *tx,
struct user_sdma_iovec *iovec,
u32 *pkt_data_remaining)
{
size_t remaining_to_add = *pkt_data_remaining;
/*
* Walk through iovec entries, ensure the associated pages
* are pinned and mapped, add data to the packet until no more
* data remains to be added.
*/
while (remaining_to_add > 0) {
struct user_sdma_iovec *cur_iovec;
size_t from_this_iovec;
int ret;
cur_iovec = iovec;
from_this_iovec = iovec->iov.iov_len - iovec->offset;
if (from_this_iovec > remaining_to_add) {
from_this_iovec = remaining_to_add;
} else {
/* The current iovec entry will be consumed by this pass. */
req->iov_idx++;
iovec++;
}
ret = add_system_iovec_to_sdma_packet(req, tx, cur_iovec,
from_this_iovec);
if (ret)
return ret;
remaining_to_add -= from_this_iovec;
}
*pkt_data_remaining = remaining_to_add;
return 0;
}
void system_descriptor_complete(struct hfi1_devdata *dd,
struct sdma_desc *descp)
{
switch (sdma_mapping_type(descp)) {
case SDMA_MAP_SINGLE:
dma_unmap_single(&dd->pcidev->dev, sdma_mapping_addr(descp),
sdma_mapping_len(descp), DMA_TO_DEVICE);
break;
case SDMA_MAP_PAGE:
dma_unmap_page(&dd->pcidev->dev, sdma_mapping_addr(descp),
sdma_mapping_len(descp), DMA_TO_DEVICE);
break;
}
if (descp->pinning_ctx) {
struct sdma_mmu_node *node = descp->pinning_ctx;
release_node(node->rb.handler, node);
}
}
static bool sdma_rb_filter(struct mmu_rb_node *node, unsigned long addr,
unsigned long len)
{
@ -1493,8 +1666,7 @@ static void sdma_rb_remove(void *arg, struct mmu_rb_node *mnode)
struct sdma_mmu_node *node =
container_of(mnode, struct sdma_mmu_node, rb);
unpin_sdma_pages(node);
kfree(node);
free_system_node(node);
}
static int sdma_rb_invalidate(void *arg, struct mmu_rb_node *mnode)

5
drivers/infiniband/hw/hfi1/user_sdma.h
View File

@ -112,16 +112,11 @@ struct sdma_mmu_node {
struct user_sdma_iovec {
struct list_head list;
struct iovec iov;
/* number of pages in this vector */
unsigned int npages;
/* array of pinned pages for this vector */
struct page **pages;
/*
* offset into the virtual address space of the vector at
* which we last left off.
*/
u64 offset;
struct sdma_mmu_node *node;
};
/* evict operation argument */

4
drivers/infiniband/hw/hfi1/verbs.c
View File

@ -778,8 +778,8 @@ static int build_verbs_tx_desc(
/* add icrc, lt byte, and padding to flit */
if (extra_bytes)
ret = sdma_txadd_daddr(sde->dd, &tx->txreq,
sde->dd->sdma_pad_phys, extra_bytes);
ret = sdma_txadd_daddr(sde->dd, &tx->txreq, sde->dd->sdma_pad_phys,
extra_bytes);
bail_txadd:
return ret;

1
drivers/infiniband/hw/hfi1/vnic_sdma.c
View File

@ -64,6 +64,7 @@ static noinline int build_vnic_ulp_payload(struct sdma_engine *sde,
/* combine physically continuous fragments later? */
ret = sdma_txadd_page(sde->dd,
NULL,
&tx->txreq,
skb_frag_page(frag),
skb_frag_off(frag),

187
drivers/infiniband/hw/hns/hns_roce_hw_v2.c
View File

@ -1960,100 +1960,6 @@ static int hns_roce_v2_set_bt(struct hns_roce_dev *hr_dev)
return hns_roce_cmq_send(hr_dev, &desc, 1);
}
/* Use default caps when hns_roce_query_pf_caps() failed or init VF profile */
static void set_default_caps(struct hns_roce_dev *hr_dev)
{
struct hns_roce_caps *caps = &hr_dev->caps;
caps->num_qps = HNS_ROCE_V2_MAX_QP_NUM;
caps->max_wqes = HNS_ROCE_V2_MAX_WQE_NUM;
caps->num_cqs = HNS_ROCE_V2_MAX_CQ_NUM;
caps->num_srqs = HNS_ROCE_V2_MAX_SRQ_NUM;
caps->min_cqes = HNS_ROCE_MIN_CQE_NUM;
caps->max_cqes = HNS_ROCE_V2_MAX_CQE_NUM;
caps->max_sq_sg = HNS_ROCE_V2_MAX_SQ_SGE_NUM;
caps->max_rq_sg = HNS_ROCE_V2_MAX_RQ_SGE_NUM;
caps->num_uars = HNS_ROCE_V2_UAR_NUM;
caps->phy_num_uars = HNS_ROCE_V2_PHY_UAR_NUM;
caps->num_aeq_vectors = HNS_ROCE_V2_AEQE_VEC_NUM;
caps->num_other_vectors = HNS_ROCE_V2_ABNORMAL_VEC_NUM;
caps->num_comp_vectors = 0;
caps->num_mtpts = HNS_ROCE_V2_MAX_MTPT_NUM;
caps->num_pds = HNS_ROCE_V2_MAX_PD_NUM;
caps->qpc_timer_bt_num = HNS_ROCE_V2_MAX_QPC_TIMER_BT_NUM;
caps->cqc_timer_bt_num = HNS_ROCE_V2_MAX_CQC_TIMER_BT_NUM;
caps->max_qp_init_rdma = HNS_ROCE_V2_MAX_QP_INIT_RDMA;
caps->max_qp_dest_rdma = HNS_ROCE_V2_MAX_QP_DEST_RDMA;
caps->max_sq_desc_sz = HNS_ROCE_V2_MAX_SQ_DESC_SZ;
caps->max_rq_desc_sz = HNS_ROCE_V2_MAX_RQ_DESC_SZ;
caps->irrl_entry_sz = HNS_ROCE_V2_IRRL_ENTRY_SZ;
caps->trrl_entry_sz = HNS_ROCE_V2_EXT_ATOMIC_TRRL_ENTRY_SZ;
caps->cqc_entry_sz = HNS_ROCE_V2_CQC_ENTRY_SZ;
caps->srqc_entry_sz = HNS_ROCE_V2_SRQC_ENTRY_SZ;
caps->mtpt_entry_sz = HNS_ROCE_V2_MTPT_ENTRY_SZ;
caps->idx_entry_sz = HNS_ROCE_V2_IDX_ENTRY_SZ;
caps->page_size_cap = HNS_ROCE_V2_PAGE_SIZE_SUPPORTED;
caps->reserved_lkey = 0;
caps->reserved_pds = 0;
caps->reserved_mrws = 1;
caps->reserved_uars = 0;
caps->reserved_cqs = 0;
caps->reserved_srqs = 0;
caps->reserved_qps = HNS_ROCE_V2_RSV_QPS;
caps->qpc_hop_num = HNS_ROCE_CONTEXT_HOP_NUM;
caps->srqc_hop_num = HNS_ROCE_CONTEXT_HOP_NUM;
caps->cqc_hop_num = HNS_ROCE_CONTEXT_HOP_NUM;
caps->mpt_hop_num = HNS_ROCE_CONTEXT_HOP_NUM;
caps->sccc_hop_num = HNS_ROCE_SCCC_HOP_NUM;
caps->mtt_hop_num = HNS_ROCE_MTT_HOP_NUM;
caps->wqe_sq_hop_num = HNS_ROCE_SQWQE_HOP_NUM;
caps->wqe_sge_hop_num = HNS_ROCE_EXT_SGE_HOP_NUM;
caps->wqe_rq_hop_num = HNS_ROCE_RQWQE_HOP_NUM;
caps->cqe_hop_num = HNS_ROCE_CQE_HOP_NUM;
caps->srqwqe_hop_num = HNS_ROCE_SRQWQE_HOP_NUM;
caps->idx_hop_num = HNS_ROCE_IDX_HOP_NUM;
caps->chunk_sz = HNS_ROCE_V2_TABLE_CHUNK_SIZE;
caps->flags = HNS_ROCE_CAP_FLAG_REREG_MR |
HNS_ROCE_CAP_FLAG_ROCE_V1_V2 |
HNS_ROCE_CAP_FLAG_CQ_RECORD_DB |
HNS_ROCE_CAP_FLAG_QP_RECORD_DB;
caps->pkey_table_len[0] = 1;
caps->ceqe_depth = HNS_ROCE_V2_COMP_EQE_NUM;
caps->aeqe_depth = HNS_ROCE_V2_ASYNC_EQE_NUM;
caps->local_ca_ack_delay = 0;
caps->max_mtu = IB_MTU_4096;
caps->max_srq_wrs = HNS_ROCE_V2_MAX_SRQ_WR;
caps->max_srq_sges = HNS_ROCE_V2_MAX_SRQ_SGE;
caps->flags |= HNS_ROCE_CAP_FLAG_ATOMIC | HNS_ROCE_CAP_FLAG_MW |
HNS_ROCE_CAP_FLAG_SRQ | HNS_ROCE_CAP_FLAG_FRMR |
HNS_ROCE_CAP_FLAG_QP_FLOW_CTRL;
caps->gid_table_len[0] = HNS_ROCE_V2_GID_INDEX_NUM;
if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP09) {
caps->flags |= HNS_ROCE_CAP_FLAG_STASH |
HNS_ROCE_CAP_FLAG_DIRECT_WQE |
HNS_ROCE_CAP_FLAG_XRC;
caps->max_sq_inline = HNS_ROCE_V3_MAX_SQ_INLINE;
} else {
caps->max_sq_inline = HNS_ROCE_V2_MAX_SQ_INLINE;
/* The following configuration are only valid for HIP08 */
caps->qpc_sz = HNS_ROCE_V2_QPC_SZ;
caps->sccc_sz = HNS_ROCE_V2_SCCC_SZ;
caps->cqe_sz = HNS_ROCE_V2_CQE_SIZE;
}
}
static void calc_pg_sz(u32 obj_num, u32 obj_size, u32 hop_num, u32 ctx_bt_num,
u32 *buf_page_size, u32 *bt_page_size, u32 hem_type)
{
@ -2239,7 +2145,7 @@ static void apply_func_caps(struct hns_roce_dev *hr_dev)
set_hem_page_size(hr_dev);
}
static int hns_roce_query_pf_caps(struct hns_roce_dev *hr_dev)
static int hns_roce_query_caps(struct hns_roce_dev *hr_dev)
{
struct hns_roce_cmq_desc desc[HNS_ROCE_QUERY_PF_CAPS_CMD_NUM];
struct hns_roce_caps *caps = &hr_dev->caps;
@ -2248,15 +2154,17 @@ static int hns_roce_query_pf_caps(struct hns_roce_dev *hr_dev)
struct hns_roce_query_pf_caps_c *resp_c;
struct hns_roce_query_pf_caps_d *resp_d;
struct hns_roce_query_pf_caps_e *resp_e;
enum hns_roce_opcode_type cmd;
int ctx_hop_num;
int pbl_hop_num;
int ret;
int i;
cmd = hr_dev->is_vf ? HNS_ROCE_OPC_QUERY_VF_CAPS_NUM :
HNS_ROCE_OPC_QUERY_PF_CAPS_NUM;
for (i = 0; i < HNS_ROCE_QUERY_PF_CAPS_CMD_NUM; i++) {
hns_roce_cmq_setup_basic_desc(&desc[i],
HNS_ROCE_OPC_QUERY_PF_CAPS_NUM,
true);
hns_roce_cmq_setup_basic_desc(&desc[i], cmd, true);
if (i < (HNS_ROCE_QUERY_PF_CAPS_CMD_NUM - 1))
desc[i].flag |= cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT);
else
@ -2273,35 +2181,33 @@ static int hns_roce_query_pf_caps(struct hns_roce_dev *hr_dev)
resp_d = (struct hns_roce_query_pf_caps_d *)desc[3].data;
resp_e = (struct hns_roce_query_pf_caps_e *)desc[4].data;
caps->local_ca_ack_delay = resp_a->local_ca_ack_delay;
caps->max_sq_sg = le16_to_cpu(resp_a->max_sq_sg);
caps->max_sq_inline = le16_to_cpu(resp_a->max_sq_inline);
caps->max_rq_sg = le16_to_cpu(resp_a->max_rq_sg);
caps->local_ca_ack_delay = resp_a->local_ca_ack_delay;
caps->max_sq_sg = le16_to_cpu(resp_a->max_sq_sg);
caps->max_sq_inline = le16_to_cpu(resp_a->max_sq_inline);
caps->max_rq_sg = le16_to_cpu(resp_a->max_rq_sg);
caps->max_rq_sg = roundup_pow_of_two(caps->max_rq_sg);
caps->max_srq_sges = le16_to_cpu(resp_a->max_srq_sges);
caps->max_srq_sges = le16_to_cpu(resp_a->max_srq_sges);
caps->max_srq_sges = roundup_pow_of_two(caps->max_srq_sges);
caps->num_aeq_vectors = resp_a->num_aeq_vectors;
caps->num_other_vectors = resp_a->num_other_vectors;
caps->max_sq_desc_sz = resp_a->max_sq_desc_sz;
caps->max_rq_desc_sz = resp_a->max_rq_desc_sz;
caps->cqe_sz = resp_a->cqe_sz;
caps->num_aeq_vectors = resp_a->num_aeq_vectors;
caps->num_other_vectors = resp_a->num_other_vectors;
caps->max_sq_desc_sz = resp_a->max_sq_desc_sz;
caps->max_rq_desc_sz = resp_a->max_rq_desc_sz;
caps->mtpt_entry_sz = resp_b->mtpt_entry_sz;
caps->irrl_entry_sz = resp_b->irrl_entry_sz;
caps->trrl_entry_sz = resp_b->trrl_entry_sz;
caps->cqc_entry_sz = resp_b->cqc_entry_sz;
caps->srqc_entry_sz = resp_b->srqc_entry_sz;
caps->idx_entry_sz = resp_b->idx_entry_sz;
caps->sccc_sz = resp_b->sccc_sz;
caps->max_mtu = resp_b->max_mtu;
caps->qpc_sz = le16_to_cpu(resp_b->qpc_sz);
caps->min_cqes = resp_b->min_cqes;
caps->min_wqes = resp_b->min_wqes;
caps->page_size_cap = le32_to_cpu(resp_b->page_size_cap);
caps->pkey_table_len[0] = resp_b->pkey_table_len;
caps->phy_num_uars = resp_b->phy_num_uars;
ctx_hop_num = resp_b->ctx_hop_num;
pbl_hop_num = resp_b->pbl_hop_num;
caps->mtpt_entry_sz = resp_b->mtpt_entry_sz;
caps->irrl_entry_sz = resp_b->irrl_entry_sz;
caps->trrl_entry_sz = resp_b->trrl_entry_sz;
caps->cqc_entry_sz = resp_b->cqc_entry_sz;
caps->srqc_entry_sz = resp_b->srqc_entry_sz;
caps->idx_entry_sz = resp_b->idx_entry_sz;
caps->sccc_sz = resp_b->sccc_sz;
caps->max_mtu = resp_b->max_mtu;
caps->min_cqes = resp_b->min_cqes;
caps->min_wqes = resp_b->min_wqes;
caps->page_size_cap = le32_to_cpu(resp_b->page_size_cap);
caps->pkey_table_len[0] = resp_b->pkey_table_len;
caps->phy_num_uars = resp_b->phy_num_uars;
ctx_hop_num = resp_b->ctx_hop_num;
pbl_hop_num = resp_b->pbl_hop_num;
caps->num_pds = 1 << hr_reg_read(resp_c, PF_CAPS_C_NUM_PDS);
@ -2324,8 +2230,6 @@ static int hns_roce_query_pf_caps(struct hns_roce_dev *hr_dev)
caps->ceqe_depth = 1 << hr_reg_read(resp_d, PF_CAPS_D_CEQ_DEPTH);
caps->num_comp_vectors = hr_reg_read(resp_d, PF_CAPS_D_NUM_CEQS);
caps->aeqe_depth = 1 << hr_reg_read(resp_d, PF_CAPS_D_AEQ_DEPTH);
caps->default_aeq_arm_st = hr_reg_read(resp_d, PF_CAPS_D_AEQ_ARM_ST);
caps->default_ceq_arm_st = hr_reg_read(resp_d, PF_CAPS_D_CEQ_ARM_ST);
caps->reserved_pds = hr_reg_read(resp_d, PF_CAPS_D_RSV_PDS);
caps->num_uars = 1 << hr_reg_read(resp_d, PF_CAPS_D_NUM_UARS);
caps->reserved_qps = hr_reg_read(resp_d, PF_CAPS_D_RSV_QPS);
@ -2336,10 +2240,6 @@ static int hns_roce_query_pf_caps(struct hns_roce_dev *hr_dev)
caps->reserved_cqs = hr_reg_read(resp_e, PF_CAPS_E_RSV_CQS);
caps->reserved_srqs = hr_reg_read(resp_e, PF_CAPS_E_RSV_SRQS);
caps->reserved_lkey = hr_reg_read(resp_e, PF_CAPS_E_RSV_LKEYS);
caps->default_ceq_max_cnt = le16_to_cpu(resp_e->ceq_max_cnt);
caps->default_ceq_period = le16_to_cpu(resp_e->ceq_period);
caps->default_aeq_max_cnt = le16_to_cpu(resp_e->aeq_max_cnt);
caps->default_aeq_period = le16_to_cpu(resp_e->aeq_period);
caps->qpc_hop_num = ctx_hop_num;
caps->sccc_hop_num = ctx_hop_num;
@ -2357,6 +2257,19 @@ static int hns_roce_query_pf_caps(struct hns_roce_dev *hr_dev)
if (!(caps->page_size_cap & PAGE_SIZE))
caps->page_size_cap = HNS_ROCE_V2_PAGE_SIZE_SUPPORTED;
if (!hr_dev->is_vf) {
caps->cqe_sz = resp_a->cqe_sz;
caps->qpc_sz = le16_to_cpu(resp_b->qpc_sz);
caps->default_aeq_arm_st =
hr_reg_read(resp_d, PF_CAPS_D_AEQ_ARM_ST);
caps->default_ceq_arm_st =
hr_reg_read(resp_d, PF_CAPS_D_CEQ_ARM_ST);
caps->default_ceq_max_cnt = le16_to_cpu(resp_e->ceq_max_cnt);
caps->default_ceq_period = le16_to_cpu(resp_e->ceq_period);
caps->default_aeq_max_cnt = le16_to_cpu(resp_e->aeq_max_cnt);
caps->default_aeq_period = le16_to_cpu(resp_e->aeq_period);
}
return 0;
}
@ -2404,7 +2317,11 @@ static int hns_roce_v2_vf_profile(struct hns_roce_dev *hr_dev)
hr_dev->func_num = 1;
set_default_caps(hr_dev);
ret = hns_roce_query_caps(hr_dev);
if (ret) {
dev_err(dev, "failed to query VF caps, ret = %d.\n", ret);
return ret;
}
ret = hns_roce_query_vf_resource(hr_dev);
if (ret) {
@ -2444,9 +2361,11 @@ static int hns_roce_v2_pf_profile(struct hns_roce_dev *hr_dev)
return ret;
}
ret = hns_roce_query_pf_caps(hr_dev);
if (ret)
set_default_caps(hr_dev);
ret = hns_roce_query_caps(hr_dev);
if (ret) {
dev_err(dev, "failed to query PF caps, ret = %d.\n", ret);
return ret;
}
ret = hns_roce_query_pf_resource(hr_dev);
if (ret) {

31
drivers/infiniband/hw/hns/hns_roce_hw_v2.h
View File

@ -35,43 +35,15 @@
#include <linux/bitops.h>
#define HNS_ROCE_V2_MAX_QP_NUM 0x1000
#define HNS_ROCE_V2_MAX_WQE_NUM 0x8000
#define HNS_ROCE_V2_MAX_SRQ_WR 0x8000
#define HNS_ROCE_V2_MAX_SRQ_SGE 64
#define HNS_ROCE_V2_MAX_CQ_NUM 0x100000
#define HNS_ROCE_V2_MAX_QPC_TIMER_BT_NUM 0x100
#define HNS_ROCE_V2_MAX_CQC_TIMER_BT_NUM 0x100
#define HNS_ROCE_V2_MAX_SRQ_NUM 0x100000
#define HNS_ROCE_V2_MAX_CQE_NUM 0x400000
#define HNS_ROCE_V2_MAX_RQ_SGE_NUM 64
#define HNS_ROCE_V2_MAX_SQ_SGE_NUM 64
#define HNS_ROCE_V2_MAX_SQ_INLINE 0x20
#define HNS_ROCE_V3_MAX_SQ_INLINE 0x400
#define HNS_ROCE_V2_MAX_RC_INL_INN_SZ 32
#define HNS_ROCE_V2_UAR_NUM 256
#define HNS_ROCE_V2_PHY_UAR_NUM 1
#define HNS_ROCE_V2_MTT_ENTRY_SZ 64
#define HNS_ROCE_V2_AEQE_VEC_NUM 1
#define HNS_ROCE_V2_ABNORMAL_VEC_NUM 1
#define HNS_ROCE_V2_MAX_MTPT_NUM 0x100000
#define HNS_ROCE_V2_MAX_SRQWQE_SEGS 0x1000000
#define HNS_ROCE_V2_MAX_IDX_SEGS 0x1000000
#define HNS_ROCE_V2_MAX_PD_NUM 0x1000000
#define HNS_ROCE_V2_MAX_XRCD_NUM 0x1000000
#define HNS_ROCE_V2_RSV_XRCD_NUM 0
#define HNS_ROCE_V2_MAX_QP_INIT_RDMA 128
#define HNS_ROCE_V2_MAX_QP_DEST_RDMA 128
#define HNS_ROCE_V2_MAX_SQ_DESC_SZ 64
#define HNS_ROCE_V2_MAX_RQ_DESC_SZ 16
#define HNS_ROCE_V2_IRRL_ENTRY_SZ 64
#define HNS_ROCE_V2_EXT_ATOMIC_TRRL_ENTRY_SZ 100
#define HNS_ROCE_V2_CQC_ENTRY_SZ 64
#define HNS_ROCE_V2_SRQC_ENTRY_SZ 64
#define HNS_ROCE_V2_MTPT_ENTRY_SZ 64
#define HNS_ROCE_V2_MTT_ENTRY_SZ 64
#define HNS_ROCE_V2_IDX_ENTRY_SZ 4
#define HNS_ROCE_V2_SCCC_SZ 32
#define HNS_ROCE_V3_SCCC_SZ 64
#define HNS_ROCE_V3_GMV_ENTRY_SZ 32
@ -232,6 +204,7 @@ enum hns_roce_opcode_type {
HNS_ROCE_OPC_QUERY_FUNC_INFO = 0x8407,
HNS_ROCE_OPC_QUERY_PF_CAPS_NUM = 0x8408,
HNS_ROCE_OPC_CFG_ENTRY_SIZE = 0x8409,
HNS_ROCE_OPC_QUERY_VF_CAPS_NUM = 0x8410,
HNS_ROCE_OPC_CFG_SGID_TB = 0x8500,
HNS_ROCE_OPC_CFG_SMAC_TB = 0x8501,
HNS_ROCE_OPC_POST_MB = 0x8504,

2
drivers/infiniband/hw/irdma/cm.c
View File

@ -337,7 +337,7 @@ static struct irdma_puda_buf *irdma_form_ah_cm_frame(struct irdma_cm_node *cm_no
pktsize = sizeof(*tcph) + opts_len + hdr_len + pd_len;
memset(buf, 0, pktsize);
memset(buf, 0, sizeof(*tcph));
sqbuf->totallen = pktsize;
sqbuf->tcphlen = sizeof(*tcph) + opts_len;

318
drivers/infiniband/hw/irdma/ctrl.c
View File

@ -1867,8 +1867,6 @@ void irdma_sc_vsi_init(struct irdma_sc_vsi *vsi,
vsi->mtu = info->params->mtu;
vsi->exception_lan_q = info->exception_lan_q;
vsi->vsi_idx = info->pf_data_vsi_num;
if (vsi->dev->hw_attrs.uk_attrs.hw_rev == IRDMA_GEN_1)
vsi->fcn_id = info->dev->hmc_fn_id;
irdma_set_qos_info(vsi, info->params);
for (i = 0; i < IRDMA_MAX_USER_PRIORITY; i++) {
@ -1887,32 +1885,56 @@ void irdma_sc_vsi_init(struct irdma_sc_vsi *vsi,
}
/**
* irdma_get_fcn_id - Return the function id
* irdma_get_stats_idx - Return stats index
* @vsi: pointer to the vsi
*/
static u8 irdma_get_fcn_id(struct irdma_sc_vsi *vsi)
static u8 irdma_get_stats_idx(struct irdma_sc_vsi *vsi)
{
struct irdma_stats_inst_info stats_info = {};
struct irdma_sc_dev *dev = vsi->dev;
u8 fcn_id = IRDMA_INVALID_FCN_ID;
u8 start_idx, max_stats, i;
u8 i;
if (dev->hw_attrs.uk_attrs.hw_rev != IRDMA_GEN_1) {
if (dev->hw_attrs.uk_attrs.hw_rev >= IRDMA_GEN_2) {
if (!irdma_cqp_stats_inst_cmd(vsi, IRDMA_OP_STATS_ALLOCATE,
&stats_info))
return stats_info.stats_idx;
}
start_idx = 1;
max_stats = 16;
for (i = start_idx; i < max_stats; i++)
if (!dev->fcn_id_array[i]) {
fcn_id = i;
dev->fcn_id_array[i] = true;
break;
for (i = 0; i < IRDMA_MAX_STATS_COUNT_GEN_1; i++) {
if (!dev->stats_idx_array[i]) {
dev->stats_idx_array[i] = true;
return i;
}
}
return fcn_id;
return IRDMA_INVALID_STATS_IDX;
}
/**
* irdma_hw_stats_init_gen1 - Initialize stat reg table used for gen1
* @vsi: vsi structure where hw_regs are set
*
* Populate the HW stats table
*/
static void irdma_hw_stats_init_gen1(struct irdma_sc_vsi *vsi)
{
struct irdma_sc_dev *dev = vsi->dev;
const struct irdma_hw_stat_map *map;
u64 *stat_reg = vsi->hw_stats_regs;
u64 *regs = dev->hw_stats_regs;
u16 i, stats_reg_set = vsi->stats_idx;
map = dev->hw_stats_map;
/* First 4 stat instances are reserved for port level statistics. */
stats_reg_set += vsi->stats_inst_alloc ? IRDMA_FIRST_NON_PF_STAT : 0;
for (i = 0; i < dev->hw_attrs.max_stat_idx; i++) {
if (map[i].bitmask <= IRDMA_MAX_STATS_32)
stat_reg[i] = regs[i] + stats_reg_set * sizeof(u32);
else
stat_reg[i] = regs[i] + stats_reg_set * sizeof(u64);
}
}
/**
@ -1923,7 +1945,6 @@ static u8 irdma_get_fcn_id(struct irdma_sc_vsi *vsi)
int irdma_vsi_stats_init(struct irdma_sc_vsi *vsi,
struct irdma_vsi_stats_info *info)
{
u8 fcn_id = info->fcn_id;
struct irdma_dma_mem *stats_buff_mem;
vsi->pestat = info->pestat;
@ -1944,26 +1965,24 @@ int irdma_vsi_stats_init(struct irdma_sc_vsi *vsi,
IRDMA_GATHER_STATS_BUF_SIZE);
irdma_hw_stats_start_timer(vsi);
if (info->alloc_fcn_id)
fcn_id = irdma_get_fcn_id(vsi);
if (fcn_id == IRDMA_INVALID_FCN_ID)
goto stats_error;
vsi->stats_fcn_id_alloc = info->alloc_fcn_id;
vsi->fcn_id = fcn_id;
if (info->alloc_fcn_id) {
vsi->pestat->gather_info.use_stats_inst = true;
vsi->pestat->gather_info.stats_inst_index = fcn_id;
/* when stat allocation is not required default to fcn_id. */
vsi->stats_idx = info->fcn_id;
if (info->alloc_stats_inst) {
u8 stats_idx = irdma_get_stats_idx(vsi);
if (stats_idx != IRDMA_INVALID_STATS_IDX) {
vsi->stats_inst_alloc = true;
vsi->stats_idx = stats_idx;
vsi->pestat->gather_info.use_stats_inst = true;
vsi->pestat->gather_info.stats_inst_index = stats_idx;
}
}
if (vsi->dev->hw_attrs.uk_attrs.hw_rev == IRDMA_GEN_1)
irdma_hw_stats_init_gen1(vsi);
return 0;
stats_error:
dma_free_coherent(vsi->pestat->hw->device, stats_buff_mem->size,
stats_buff_mem->va, stats_buff_mem->pa);
stats_buff_mem->va = NULL;
return -EIO;
}
/**
@ -1973,19 +1992,19 @@ stats_error:
void irdma_vsi_stats_free(struct irdma_sc_vsi *vsi)
{
struct irdma_stats_inst_info stats_info = {};
u8 fcn_id = vsi->fcn_id;
struct irdma_sc_dev *dev = vsi->dev;
u8 stats_idx = vsi->stats_idx;
if (dev->hw_attrs.uk_attrs.hw_rev != IRDMA_GEN_1) {
if (vsi->stats_fcn_id_alloc) {
stats_info.stats_idx = vsi->fcn_id;
if (dev->hw_attrs.uk_attrs.hw_rev >= IRDMA_GEN_2) {
if (vsi->stats_inst_alloc) {
stats_info.stats_idx = vsi->stats_idx;
irdma_cqp_stats_inst_cmd(vsi, IRDMA_OP_STATS_FREE,
&stats_info);
}
} else {
if (vsi->stats_fcn_id_alloc &&
fcn_id < vsi->dev->hw_attrs.max_stat_inst)
vsi->dev->fcn_id_array[fcn_id] = false;
if (vsi->stats_inst_alloc &&
stats_idx < vsi->dev->hw_attrs.max_stat_inst)
vsi->dev->stats_idx_array[stats_idx] = false;
}
if (!vsi->pestat)
@ -5297,7 +5316,8 @@ void sc_vsi_update_stats(struct irdma_sc_vsi *vsi)
gather_stats = vsi->pestat->gather_info.gather_stats_va;
last_gather_stats = vsi->pestat->gather_info.last_gather_stats_va;
irdma_update_stats(&vsi->pestat->hw_stats, gather_stats,
last_gather_stats);
last_gather_stats, vsi->dev->hw_stats_map,
vsi->dev->hw_attrs.max_stat_idx);
}
/**
@ -5404,186 +5424,62 @@ int irdma_sc_dev_init(enum irdma_vers ver, struct irdma_sc_dev *dev,
return ret_code;
}
/**
* irdma_stat_val - Extract HW counter value from statistics buffer
* @stats_val: pointer to statistics buffer
* @byteoff: byte offset of counter value in the buffer (8B-aligned)
* @bitoff: bit offset of counter value within 8B entry
* @bitmask: maximum counter value (e.g. 0xffffff for 24-bit counter)
*/
static inline u64 irdma_stat_val(const u64 *stats_val, u16 byteoff, u8 bitoff,
u64 bitmask)
{
u16 idx = byteoff / sizeof(*stats_val);
return (stats_val[idx] >> bitoff) & bitmask;
}
/**
* irdma_stat_delta - Calculate counter delta
* @new_val: updated counter value
* @old_val: last counter value
* @max_val: maximum counter value (e.g. 0xffffff for 24-bit counter)
*/
static inline u64 irdma_stat_delta(u64 new_val, u64 old_val, u64 max_val)
{
if (new_val >= old_val)
return new_val - old_val;
/* roll-over case */
return max_val - old_val + new_val + 1;
}
/**
* irdma_update_stats - Update statistics
* @hw_stats: hw_stats instance to update
* @gather_stats: updated stat counters
* @last_gather_stats: last stat counters
* @map: HW stat map (hw_stats => gather_stats)
* @max_stat_idx: number of HW stats
*/
void irdma_update_stats(struct irdma_dev_hw_stats *hw_stats,
struct irdma_gather_stats *gather_stats,
struct irdma_gather_stats *last_gather_stats)
struct irdma_gather_stats *last_gather_stats,
const struct irdma_hw_stat_map *map, u16 max_stat_idx)
{
u64 *stats_val = hw_stats->stats_val_32;
u64 *stats_val = hw_stats->stats_val;
u16 i;
for (i = 0; i < max_stat_idx; i++) {
u64 new_val = irdma_stat_val(gather_stats->val, map[i].byteoff,
map[i].bitoff, map[i].bitmask);
u64 last_val = irdma_stat_val(last_gather_stats->val,
map[i].byteoff, map[i].bitoff,
map[i].bitmask);
stats_val[i] +=
irdma_stat_delta(new_val, last_val, map[i].bitmask);
}
stats_val[IRDMA_HW_STAT_INDEX_RXVLANERR] +=
IRDMA_STATS_DELTA(gather_stats->rxvlanerr,
last_gather_stats->rxvlanerr,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_IP4RXDISCARD] +=
IRDMA_STATS_DELTA(gather_stats->ip4rxdiscard,
last_gather_stats->ip4rxdiscard,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_IP4RXTRUNC] +=
IRDMA_STATS_DELTA(gather_stats->ip4rxtrunc,
last_gather_stats->ip4rxtrunc,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_IP4TXNOROUTE] +=
IRDMA_STATS_DELTA(gather_stats->ip4txnoroute,
last_gather_stats->ip4txnoroute,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_IP6RXDISCARD] +=
IRDMA_STATS_DELTA(gather_stats->ip6rxdiscard,
last_gather_stats->ip6rxdiscard,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_IP6RXTRUNC] +=
IRDMA_STATS_DELTA(gather_stats->ip6rxtrunc,
last_gather_stats->ip6rxtrunc,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_IP6TXNOROUTE] +=
IRDMA_STATS_DELTA(gather_stats->ip6txnoroute,
last_gather_stats->ip6txnoroute,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_TCPRTXSEG] +=
IRDMA_STATS_DELTA(gather_stats->tcprtxseg,
last_gather_stats->tcprtxseg,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_TCPRXOPTERR] +=
IRDMA_STATS_DELTA(gather_stats->tcprxopterr,
last_gather_stats->tcprxopterr,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_TCPRXPROTOERR] +=
IRDMA_STATS_DELTA(gather_stats->tcprxprotoerr,
last_gather_stats->tcprxprotoerr,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_RXRPCNPHANDLED] +=
IRDMA_STATS_DELTA(gather_stats->rxrpcnphandled,
last_gather_stats->rxrpcnphandled,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_RXRPCNPIGNORED] +=
IRDMA_STATS_DELTA(gather_stats->rxrpcnpignored,
last_gather_stats->rxrpcnpignored,
IRDMA_MAX_STATS_32);
stats_val[IRDMA_HW_STAT_INDEX_TXNPCNPSENT] +=
IRDMA_STATS_DELTA(gather_stats->txnpcnpsent,
last_gather_stats->txnpcnpsent,
IRDMA_MAX_STATS_32);
stats_val = hw_stats->stats_val_64;
stats_val[IRDMA_HW_STAT_INDEX_IP4RXOCTS] +=
IRDMA_STATS_DELTA(gather_stats->ip4rxocts,
last_gather_stats->ip4rxocts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP4RXPKTS] +=
IRDMA_STATS_DELTA(gather_stats->ip4rxpkts,
last_gather_stats->ip4rxpkts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP4RXFRAGS] +=
IRDMA_STATS_DELTA(gather_stats->ip4txfrag,
last_gather_stats->ip4txfrag,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP4RXMCPKTS] +=
IRDMA_STATS_DELTA(gather_stats->ip4rxmcpkts,
last_gather_stats->ip4rxmcpkts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP4TXOCTS] +=
IRDMA_STATS_DELTA(gather_stats->ip4txocts,
last_gather_stats->ip4txocts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP4TXPKTS] +=
IRDMA_STATS_DELTA(gather_stats->ip4txpkts,
last_gather_stats->ip4txpkts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP4TXFRAGS] +=
IRDMA_STATS_DELTA(gather_stats->ip4txfrag,
last_gather_stats->ip4txfrag,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP4TXMCPKTS] +=
IRDMA_STATS_DELTA(gather_stats->ip4txmcpkts,
last_gather_stats->ip4txmcpkts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP6RXOCTS] +=
IRDMA_STATS_DELTA(gather_stats->ip6rxocts,
last_gather_stats->ip6rxocts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP6RXPKTS] +=
IRDMA_STATS_DELTA(gather_stats->ip6rxpkts,
last_gather_stats->ip6rxpkts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP6RXFRAGS] +=
IRDMA_STATS_DELTA(gather_stats->ip6txfrags,
last_gather_stats->ip6txfrags,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP6RXMCPKTS] +=
IRDMA_STATS_DELTA(gather_stats->ip6rxmcpkts,
last_gather_stats->ip6rxmcpkts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP6TXOCTS] +=
IRDMA_STATS_DELTA(gather_stats->ip6txocts,
last_gather_stats->ip6txocts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP6TXPKTS] +=
IRDMA_STATS_DELTA(gather_stats->ip6txpkts,
last_gather_stats->ip6txpkts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP6TXFRAGS] +=
IRDMA_STATS_DELTA(gather_stats->ip6txfrags,
last_gather_stats->ip6txfrags,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_IP6TXMCPKTS] +=
IRDMA_STATS_DELTA(gather_stats->ip6txmcpkts,
last_gather_stats->ip6txmcpkts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_TCPRXSEGS] +=
IRDMA_STATS_DELTA(gather_stats->tcprxsegs,
last_gather_stats->tcprxsegs,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_TCPTXSEG] +=
IRDMA_STATS_DELTA(gather_stats->tcptxsegs,
last_gather_stats->tcptxsegs,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_RDMARXRDS] +=
IRDMA_STATS_DELTA(gather_stats->rdmarxrds,
last_gather_stats->rdmarxrds,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_RDMARXSNDS] +=
IRDMA_STATS_DELTA(gather_stats->rdmarxsnds,
last_gather_stats->rdmarxsnds,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_RDMARXWRS] +=
IRDMA_STATS_DELTA(gather_stats->rdmarxwrs,
last_gather_stats->rdmarxwrs,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_RDMATXRDS] +=
IRDMA_STATS_DELTA(gather_stats->rdmatxrds,
last_gather_stats->rdmatxrds,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_RDMATXSNDS] +=
IRDMA_STATS_DELTA(gather_stats->rdmatxsnds,
last_gather_stats->rdmatxsnds,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_RDMATXWRS] +=
IRDMA_STATS_DELTA(gather_stats->rdmatxwrs,
last_gather_stats->rdmatxwrs,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_RDMAVBND] +=
IRDMA_STATS_DELTA(gather_stats->rdmavbn,
last_gather_stats->rdmavbn,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_RDMAVINV] +=
IRDMA_STATS_DELTA(gather_stats->rdmavinv,
last_gather_stats->rdmavinv,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_UDPRXPKTS] +=
IRDMA_STATS_DELTA(gather_stats->udprxpkts,
last_gather_stats->udprxpkts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_UDPTXPKTS] +=
IRDMA_STATS_DELTA(gather_stats->udptxpkts,
last_gather_stats->udptxpkts,
IRDMA_MAX_STATS_48);
stats_val[IRDMA_HW_STAT_INDEX_RXNPECNMARKEDPKTS] +=
IRDMA_STATS_DELTA(gather_stats->rxnpecnmrkpkts,
last_gather_stats->rxnpecnmrkpkts,
IRDMA_MAX_STATS_48);
memcpy(last_gather_stats, gather_stats, sizeof(*last_gather_stats));
}

9
drivers/infiniband/hw/irdma/defs.h
View File

@ -36,6 +36,7 @@ enum irdma_protocol_used {
#define IRDMA_QP_STATE_ERROR 6
#define IRDMA_MAX_TRAFFIC_CLASS 8
#define IRDMA_MAX_STATS_COUNT_GEN_1 12
#define IRDMA_MAX_USER_PRIORITY 8
#define IRDMA_MAX_APPS 8
#define IRDMA_MAX_STATS_COUNT 128
@ -365,9 +366,11 @@ enum irdma_cqp_op_type {
#define FLD_RS_32(dev, val, field) \
((u64)((val) & (dev)->hw_masks[field ## _M]) >> (dev)->hw_shifts[field ## _S])
#define IRDMA_STATS_DELTA(a, b, c) ((a) >= (b) ? (a) - (b) : (a) + (c) - (b))
#define IRDMA_MAX_STATS_32 0xFFFFFFFFULL
#define IRDMA_MAX_STATS_48 0xFFFFFFFFFFFFULL
#define IRDMA_MAX_STATS_24 0xffffffULL
#define IRDMA_MAX_STATS_32 0xffffffffULL
#define IRDMA_MAX_STATS_48 0xffffffffffffULL
#define IRDMA_MAX_STATS_56 0xffffffffffffffULL
#define IRDMA_MAX_STATS_64 0xffffffffffffffffULL
#define IRDMA_MAX_CQ_READ_THRESH 0x3FFFF
#define IRDMA_CQPSQ_QHASH_VLANID GENMASK_ULL(43, 32)

17
drivers/infiniband/hw/irdma/hw.c
View File

@ -1092,14 +1092,19 @@ static int irdma_cfg_ceq_vector(struct irdma_pci_f *rf, struct irdma_ceq *iwceq,
int status;
if (rf->msix_shared && !ceq_id) {
snprintf(msix_vec->name, sizeof(msix_vec->name) - 1,
"irdma-%s-AEQCEQ-0", dev_name(&rf->pcidev->dev));
tasklet_setup(&rf->dpc_tasklet, irdma_dpc);
status = request_irq(msix_vec->irq, irdma_irq_handler, 0,
"AEQCEQ", rf);
msix_vec->name, rf);
} else {
snprintf(msix_vec->name, sizeof(msix_vec->name) - 1,
"irdma-%s-CEQ-%d",
dev_name(&rf->pcidev->dev), ceq_id);
tasklet_setup(&iwceq->dpc_tasklet, irdma_ceq_dpc);
status = request_irq(msix_vec->irq, irdma_ceq_handler, 0,
"CEQ", iwceq);
msix_vec->name, iwceq);
}
cpumask_clear(&msix_vec->mask);
cpumask_set_cpu(msix_vec->cpu_affinity, &msix_vec->mask);
@ -1128,9 +1133,11 @@ static int irdma_cfg_aeq_vector(struct irdma_pci_f *rf)
u32 ret = 0;
if (!rf->msix_shared) {
snprintf(msix_vec->name, sizeof(msix_vec->name) - 1,
"irdma-%s-AEQ", dev_name(&rf->pcidev->dev));
tasklet_setup(&rf->dpc_tasklet, irdma_dpc);
ret = request_irq(msix_vec->irq, irdma_irq_handler, 0,
"irdma", rf);
msix_vec->name, rf);
}
if (ret) {
ibdev_dbg(&rf->iwdev->ibdev, "ERR: aeq irq config fail\n");
@ -1904,8 +1911,8 @@ int irdma_ctrl_init_hw(struct irdma_pci_f *rf)
break;
rf->init_state = CEQ0_CREATED;
/* Handles processing of CQP completions */
rf->cqp_cmpl_wq = alloc_ordered_workqueue("cqp_cmpl_wq",
WQ_HIGHPRI | WQ_UNBOUND);
rf->cqp_cmpl_wq =
alloc_ordered_workqueue("cqp_cmpl_wq", WQ_HIGHPRI);
if (!rf->cqp_cmpl_wq) {
status = -ENOMEM;
break;

60
drivers/infiniband/hw/irdma/i40iw_hw.c
View File

@ -32,7 +32,7 @@ static u32 i40iw_regs[IRDMA_MAX_REGS] = {
0xffffffff /* PFINT_RATEN not used in FPK */
};
static u32 i40iw_stat_offsets_32[IRDMA_HW_STAT_INDEX_MAX_32] = {
static u32 i40iw_stat_offsets[] = {
I40E_GLPES_PFIP4RXDISCARD(0),
I40E_GLPES_PFIP4RXTRUNC(0),
I40E_GLPES_PFIP4TXNOROUTE(0),
@ -42,10 +42,8 @@ static u32 i40iw_stat_offsets_32[IRDMA_HW_STAT_INDEX_MAX_32] = {
I40E_GLPES_PFTCPRTXSEG(0),
I40E_GLPES_PFTCPRXOPTERR(0),
I40E_GLPES_PFTCPRXPROTOERR(0),
I40E_GLPES_PFRXVLANERR(0)
};
I40E_GLPES_PFRXVLANERR(0),
static u32 i40iw_stat_offsets_64[IRDMA_HW_STAT_INDEX_MAX_64] = {
I40E_GLPES_PFIP4RXOCTSLO(0),
I40E_GLPES_PFIP4RXPKTSLO(0),
I40E_GLPES_PFIP4RXFRAGSLO(0),
@ -158,6 +156,51 @@ static const struct irdma_irq_ops i40iw_irq_ops = {
.irdma_en_irq = i40iw_ena_irq,
};
static const struct irdma_hw_stat_map i40iw_hw_stat_map[] = {
[IRDMA_HW_STAT_INDEX_RXVLANERR] = { 0, 0, IRDMA_MAX_STATS_24 },
[IRDMA_HW_STAT_INDEX_IP4RXOCTS] = { 8, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4RXPKTS] = { 16, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4RXDISCARD] = { 24, 0, IRDMA_MAX_STATS_32 },
[IRDMA_HW_STAT_INDEX_IP4RXTRUNC] = { 32, 0, IRDMA_MAX_STATS_32 },
[IRDMA_HW_STAT_INDEX_IP4RXFRAGS] = { 40, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4RXMCPKTS] = { 48, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6RXOCTS] = { 56, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6RXPKTS] = { 64, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6RXDISCARD] = { 72, 0, IRDMA_MAX_STATS_32 },
[IRDMA_HW_STAT_INDEX_IP6RXTRUNC] = { 80, 0, IRDMA_MAX_STATS_32 },
[IRDMA_HW_STAT_INDEX_IP6RXFRAGS] = { 88, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6RXMCPKTS] = { 96, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXOCTS] = { 104, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXPKTS] = { 112, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXFRAGS] = { 120, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXMCPKTS] = { 128, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6TXOCTS] = { 136, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6TXPKTS] = { 144, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6TXFRAGS] = { 152, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6TXMCPKTS] = { 160, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXNOROUTE] = { 168, 0, IRDMA_MAX_STATS_24 },
[IRDMA_HW_STAT_INDEX_IP6TXNOROUTE] = { 176, 0, IRDMA_MAX_STATS_24 },
[IRDMA_HW_STAT_INDEX_TCPRXSEGS] = { 184, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_TCPRXOPTERR] = { 192, 0, IRDMA_MAX_STATS_24 },
[IRDMA_HW_STAT_INDEX_TCPRXPROTOERR] = { 200, 0, IRDMA_MAX_STATS_24 },
[IRDMA_HW_STAT_INDEX_TCPTXSEG] = { 208, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_TCPRTXSEG] = { 216, 0, IRDMA_MAX_STATS_32 },
[IRDMA_HW_STAT_INDEX_RDMARXWRS] = { 224, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMARXRDS] = { 232, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMARXSNDS] = { 240, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMATXWRS] = { 248, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMATXRDS] = { 256, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMATXSNDS] = { 264, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMAVBND] = { 272, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMAVINV] = { 280, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4RXMCOCTS] = { 288, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXMCOCTS] = { 296, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6RXMCOCTS] = { 304, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6TXMCOCTS] = { 312, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_UDPRXPKTS] = { 320, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_UDPTXPKTS] = { 328, 0, IRDMA_MAX_STATS_48 },
};
void i40iw_init_hw(struct irdma_sc_dev *dev)
{
int i;
@ -172,11 +215,8 @@ void i40iw_init_hw(struct irdma_sc_dev *dev)
dev->hw_regs[i] = (u32 __iomem *)(i40iw_regs[i] + hw_addr);
}
for (i = 0; i < IRDMA_HW_STAT_INDEX_MAX_32; ++i)
dev->hw_stats_regs_32[i] = i40iw_stat_offsets_32[i];
for (i = 0; i < IRDMA_HW_STAT_INDEX_MAX_64; ++i)
dev->hw_stats_regs_64[i] = i40iw_stat_offsets_64[i];
for (i = 0; i < IRDMA_HW_STAT_INDEX_MAX_GEN_1; ++i)
dev->hw_stats_regs[i] = i40iw_stat_offsets[i];
dev->hw_attrs.first_hw_vf_fpm_id = I40IW_FIRST_VF_FPM_ID;
dev->hw_attrs.max_hw_vf_fpm_id = IRDMA_MAX_VF_FPM_ID;
@ -195,6 +235,7 @@ void i40iw_init_hw(struct irdma_sc_dev *dev)
dev->ceq_itr_mask_db = NULL;
dev->aeq_itr_mask_db = NULL;
dev->irq_ops = &i40iw_irq_ops;
dev->hw_stats_map = i40iw_hw_stat_map;
/* Setup the hardware limits, hmc may limit further */
dev->hw_attrs.uk_attrs.max_hw_wq_frags = I40IW_MAX_WQ_FRAGMENT_COUNT;
@ -210,6 +251,7 @@ void i40iw_init_hw(struct irdma_sc_dev *dev)
dev->hw_attrs.uk_attrs.max_hw_sq_chunk = I40IW_MAX_QUANTA_PER_WR;
dev->hw_attrs.max_hw_pds = I40IW_MAX_PDS;
dev->hw_attrs.max_stat_inst = I40IW_MAX_STATS_COUNT;
dev->hw_attrs.max_stat_idx = IRDMA_HW_STAT_INDEX_MAX_GEN_1;
dev->hw_attrs.max_hw_outbound_msg_size = I40IW_MAX_OUTBOUND_MSG_SIZE;
dev->hw_attrs.max_hw_inbound_msg_size = I40IW_MAX_INBOUND_MSG_SIZE;
dev->hw_attrs.max_qp_wr = I40IW_MAX_QP_WRS;

51
drivers/infiniband/hw/irdma/icrdma_hw.c
View File

@ -111,6 +111,55 @@ static const struct irdma_irq_ops icrdma_irq_ops = {
.irdma_en_irq = icrdma_ena_irq,
};
static const struct irdma_hw_stat_map icrdma_hw_stat_map[] = {
[IRDMA_HW_STAT_INDEX_RXVLANERR] = { 0, 32, IRDMA_MAX_STATS_24 },
[IRDMA_HW_STAT_INDEX_IP4RXOCTS] = { 8, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4RXPKTS] = { 16, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4RXDISCARD] = { 24, 32, IRDMA_MAX_STATS_32 },
[IRDMA_HW_STAT_INDEX_IP4RXTRUNC] = { 24, 0, IRDMA_MAX_STATS_32 },
[IRDMA_HW_STAT_INDEX_IP4RXFRAGS] = { 32, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4RXMCOCTS] = { 40, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4RXMCPKTS] = { 48, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6RXOCTS] = { 56, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6RXPKTS] = { 64, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6RXDISCARD] = { 72, 32, IRDMA_MAX_STATS_32 },
[IRDMA_HW_STAT_INDEX_IP6RXTRUNC] = { 72, 0, IRDMA_MAX_STATS_32 },
[IRDMA_HW_STAT_INDEX_IP6RXFRAGS] = { 80, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6RXMCOCTS] = { 88, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6RXMCPKTS] = { 96, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXOCTS] = { 104, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXPKTS] = { 112, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXFRAGS] = { 120, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXMCOCTS] = { 128, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXMCPKTS] = { 136, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6TXOCTS] = { 144, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6TXPKTS] = { 152, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6TXFRAGS] = { 160, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6TXMCOCTS] = { 168, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP6TXMCPKTS] = { 176, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_IP4TXNOROUTE] = { 184, 32, IRDMA_MAX_STATS_24 },
[IRDMA_HW_STAT_INDEX_IP6TXNOROUTE] = { 184, 0, IRDMA_MAX_STATS_24 },
[IRDMA_HW_STAT_INDEX_TCPRXSEGS] = { 192, 32, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_TCPRXOPTERR] = { 200, 32, IRDMA_MAX_STATS_24 },
[IRDMA_HW_STAT_INDEX_TCPRXPROTOERR] = { 200, 0, IRDMA_MAX_STATS_24 },
[IRDMA_HW_STAT_INDEX_TCPTXSEG] = { 208, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_TCPRTXSEG] = { 216, 32, IRDMA_MAX_STATS_32 },
[IRDMA_HW_STAT_INDEX_UDPRXPKTS] = { 224, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_UDPTXPKTS] = { 232, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMARXWRS] = { 240, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMARXRDS] = { 248, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMARXSNDS] = { 256, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMATXWRS] = { 264, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMATXRDS] = { 272, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMATXSNDS] = { 280, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMAVBND] = { 288, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RDMAVINV] = { 296, 0, IRDMA_MAX_STATS_48 },
[IRDMA_HW_STAT_INDEX_RXNPECNMARKEDPKTS] = { 304, 0, IRDMA_MAX_STATS_56 },
[IRDMA_HW_STAT_INDEX_RXRPCNPIGNORED] = { 312, 32, IRDMA_MAX_STATS_24 },
[IRDMA_HW_STAT_INDEX_RXRPCNPHANDLED] = { 312, 0, IRDMA_MAX_STATS_32 },
[IRDMA_HW_STAT_INDEX_TXNPCNPSENT] = { 320, 0, IRDMA_MAX_STATS_32 },
};
void icrdma_init_hw(struct irdma_sc_dev *dev)
{
int i;
@ -140,9 +189,11 @@ void icrdma_init_hw(struct irdma_sc_dev *dev)
dev->cq_ack_db = dev->hw_regs[IRDMA_CQACK];
dev->irq_ops = &icrdma_irq_ops;
dev->hw_attrs.page_size_cap = SZ_4K | SZ_2M | SZ_1G;
dev->hw_stats_map = icrdma_hw_stat_map;
dev->hw_attrs.max_hw_ird = ICRDMA_MAX_IRD_SIZE;
dev->hw_attrs.max_hw_ord = ICRDMA_MAX_ORD_SIZE;
dev->hw_attrs.max_stat_inst = ICRDMA_MAX_STATS_COUNT;
dev->hw_attrs.max_stat_idx = IRDMA_HW_STAT_INDEX_MAX_GEN_2;
dev->hw_attrs.uk_attrs.max_hw_sq_chunk = IRDMA_MAX_QUANTA_PER_WR;
dev->hw_attrs.uk_attrs.feature_flags |= IRDMA_FEATURE_RTS_AE |

1
drivers/infiniband/hw/irdma/irdma.h
View File

@ -147,6 +147,7 @@ struct irdma_hw_attrs {
u32 max_sleep_count;
u32 max_cqp_compl_wait_time_ms;
u16 max_stat_inst;
u16 max_stat_idx;
};
void i40iw_init_hw(struct irdma_sc_dev *dev);

3
drivers/infiniband/hw/irdma/main.h
View File

@ -115,6 +115,8 @@ extern struct auxiliary_driver i40iw_auxiliary_drv;
#define IRDMA_REFLUSH BIT(2)
#define IRDMA_FLUSH_WAIT BIT(3)
#define IRDMA_IRQ_NAME_STR_LEN (64)
enum init_completion_state {
INVALID_STATE = 0,
INITIAL_STATE,
@ -212,6 +214,7 @@ struct irdma_msix_vector {
u32 cpu_affinity;
u32 ceq_id;
cpumask_t mask;
char name[IRDMA_IRQ_NAME_STR_LEN];
};
struct irdma_mc_table_info {

16
drivers/infiniband/hw/irdma/pble.c
View File

@ -423,15 +423,15 @@ static int get_lvl1_pble(struct irdma_hmc_pble_rsrc *pble_rsrc,
* get_lvl1_lvl2_pble - calls get_lvl1 and get_lvl2 pble routine
* @pble_rsrc: pble resources
* @palloc: contains all inforamtion regarding pble (idx + pble addr)
* @level1_only: flag for a level 1 PBLE
* @lvl: Bitmask for requested pble level
*/
static int get_lvl1_lvl2_pble(struct irdma_hmc_pble_rsrc *pble_rsrc,
struct irdma_pble_alloc *palloc, bool level1_only)
struct irdma_pble_alloc *palloc, u8 lvl)
{
int status = 0;
status = get_lvl1_pble(pble_rsrc, palloc);
if (!status || level1_only || palloc->total_cnt <= PBLE_PER_PAGE)
if (!status || lvl == PBLE_LEVEL_1 || palloc->total_cnt <= PBLE_PER_PAGE)
return status;
status = get_lvl2_pble(pble_rsrc, palloc);
@ -444,11 +444,11 @@ static int get_lvl1_lvl2_pble(struct irdma_hmc_pble_rsrc *pble_rsrc,
* @pble_rsrc: pble resources
* @palloc: contains all inforamtion regarding pble (idx + pble addr)
* @pble_cnt: #of pbles requested
* @level1_only: true if only pble level 1 to acquire
* @lvl: requested pble level mask
*/
int irdma_get_pble(struct irdma_hmc_pble_rsrc *pble_rsrc,
struct irdma_pble_alloc *palloc, u32 pble_cnt,
bool level1_only)
u8 lvl)
{
int status = 0;
int max_sds = 0;
@ -462,7 +462,7 @@ int irdma_get_pble(struct irdma_hmc_pble_rsrc *pble_rsrc,
/*check first to see if we can get pble's without acquiring
* additional sd's
*/
status = get_lvl1_lvl2_pble(pble_rsrc, palloc, level1_only);
status = get_lvl1_lvl2_pble(pble_rsrc, palloc, lvl);
if (!status)
goto exit;
@ -472,9 +472,9 @@ int irdma_get_pble(struct irdma_hmc_pble_rsrc *pble_rsrc,
if (status)
break;
status = get_lvl1_lvl2_pble(pble_rsrc, palloc, level1_only);
status = get_lvl1_lvl2_pble(pble_rsrc, palloc, lvl);
/* if level1_only, only go through it once */
if (!status || level1_only)
if (!status || lvl)
break;
}

2
drivers/infiniband/hw/irdma/pble.h
View File

@ -114,7 +114,7 @@ void irdma_free_pble(struct irdma_hmc_pble_rsrc *pble_rsrc,
struct irdma_pble_alloc *palloc);
int irdma_get_pble(struct irdma_hmc_pble_rsrc *pble_rsrc,
struct irdma_pble_alloc *palloc, u32 pble_cnt,
bool level1_only);
u8 lvl);
int irdma_prm_add_pble_mem(struct irdma_pble_prm *pprm,
struct irdma_chunk *pchunk);
int irdma_prm_get_pbles(struct irdma_pble_prm *pprm,

8
drivers/infiniband/hw/irdma/protos.h
View File

@ -28,9 +28,7 @@ int irdma_cqp_gather_stats_cmd(struct irdma_sc_dev *dev,
void irdma_cqp_gather_stats_gen1(struct irdma_sc_dev *dev,
struct irdma_vsi_pestat *pestat);
void irdma_hw_stats_read_all(struct irdma_vsi_pestat *stats,
struct irdma_dev_hw_stats *stats_values,
u64 *hw_stats_regs_32, u64 *hw_stats_regs_64,
u8 hw_rev);
const u64 *hw_stats_regs);
int irdma_cqp_ws_node_cmd(struct irdma_sc_dev *dev, u8 cmd,
struct irdma_ws_node_info *node_info);
int irdma_cqp_ceq_cmd(struct irdma_sc_dev *dev, struct irdma_sc_ceq *sc_ceq,
@ -43,7 +41,9 @@ u16 irdma_alloc_ws_node_id(struct irdma_sc_dev *dev);
void irdma_free_ws_node_id(struct irdma_sc_dev *dev, u16 node_id);
void irdma_update_stats(struct irdma_dev_hw_stats *hw_stats,
struct irdma_gather_stats *gather_stats,
struct irdma_gather_stats *last_gather_stats);
struct irdma_gather_stats *last_gather_stats,
const struct irdma_hw_stat_map *map, u16 max_stat_idx);
/* vsi functions */
int irdma_vsi_stats_init(struct irdma_sc_vsi *vsi,
struct irdma_vsi_stats_info *info);

166
drivers/infiniband/hw/irdma/type.h
View File

@ -101,7 +101,8 @@ enum irdma_qp_event_type {
IRDMA_QP_EVENT_REQ_ERR,
};
enum irdma_hw_stats_index_32b {
enum irdma_hw_stats_index {
/* gen1 - 32-bit */
IRDMA_HW_STAT_INDEX_IP4RXDISCARD = 0,
IRDMA_HW_STAT_INDEX_IP4RXTRUNC = 1,
IRDMA_HW_STAT_INDEX_IP4TXNOROUTE = 2,
@ -111,50 +112,48 @@ enum irdma_hw_stats_index_32b {
IRDMA_HW_STAT_INDEX_TCPRTXSEG = 6,
IRDMA_HW_STAT_INDEX_TCPRXOPTERR = 7,
IRDMA_HW_STAT_INDEX_TCPRXPROTOERR = 8,
IRDMA_HW_STAT_INDEX_MAX_32_GEN_1 = 9, /* Must be same value as next entry */
IRDMA_HW_STAT_INDEX_RXVLANERR = 9,
IRDMA_HW_STAT_INDEX_RXRPCNPHANDLED = 10,
IRDMA_HW_STAT_INDEX_RXRPCNPIGNORED = 11,
IRDMA_HW_STAT_INDEX_TXNPCNPSENT = 12,
IRDMA_HW_STAT_INDEX_MAX_32, /* Must be last entry */
};
enum irdma_hw_stats_index_64b {
IRDMA_HW_STAT_INDEX_IP4RXOCTS = 0,
IRDMA_HW_STAT_INDEX_IP4RXPKTS = 1,
IRDMA_HW_STAT_INDEX_IP4RXFRAGS = 2,
IRDMA_HW_STAT_INDEX_IP4RXMCPKTS = 3,
IRDMA_HW_STAT_INDEX_IP4TXOCTS = 4,
IRDMA_HW_STAT_INDEX_IP4TXPKTS = 5,
IRDMA_HW_STAT_INDEX_IP4TXFRAGS = 6,
IRDMA_HW_STAT_INDEX_IP4TXMCPKTS = 7,
IRDMA_HW_STAT_INDEX_IP6RXOCTS = 8,
IRDMA_HW_STAT_INDEX_IP6RXPKTS = 9,
IRDMA_HW_STAT_INDEX_IP6RXFRAGS = 10,
IRDMA_HW_STAT_INDEX_IP6RXMCPKTS = 11,
IRDMA_HW_STAT_INDEX_IP6TXOCTS = 12,
IRDMA_HW_STAT_INDEX_IP6TXPKTS = 13,
IRDMA_HW_STAT_INDEX_IP6TXFRAGS = 14,
IRDMA_HW_STAT_INDEX_IP6TXMCPKTS = 15,
IRDMA_HW_STAT_INDEX_TCPRXSEGS = 16,
IRDMA_HW_STAT_INDEX_TCPTXSEG = 17,
IRDMA_HW_STAT_INDEX_RDMARXRDS = 18,
IRDMA_HW_STAT_INDEX_RDMARXSNDS = 19,
IRDMA_HW_STAT_INDEX_RDMARXWRS = 20,
IRDMA_HW_STAT_INDEX_RDMATXRDS = 21,
IRDMA_HW_STAT_INDEX_RDMATXSNDS = 22,
IRDMA_HW_STAT_INDEX_RDMATXWRS = 23,
IRDMA_HW_STAT_INDEX_RDMAVBND = 24,
IRDMA_HW_STAT_INDEX_RDMAVINV = 25,
IRDMA_HW_STAT_INDEX_MAX_64_GEN_1 = 26, /* Must be same value as next entry */
IRDMA_HW_STAT_INDEX_IP4RXMCOCTS = 26,
IRDMA_HW_STAT_INDEX_IP4TXMCOCTS = 27,
IRDMA_HW_STAT_INDEX_IP6RXMCOCTS = 28,
IRDMA_HW_STAT_INDEX_IP6TXMCOCTS = 29,
IRDMA_HW_STAT_INDEX_UDPRXPKTS = 30,
IRDMA_HW_STAT_INDEX_UDPTXPKTS = 31,
IRDMA_HW_STAT_INDEX_RXNPECNMARKEDPKTS = 32,
IRDMA_HW_STAT_INDEX_MAX_64, /* Must be last entry */
/* gen1 - 64-bit */
IRDMA_HW_STAT_INDEX_IP4RXOCTS = 10,
IRDMA_HW_STAT_INDEX_IP4RXPKTS = 11,
IRDMA_HW_STAT_INDEX_IP4RXFRAGS = 12,
IRDMA_HW_STAT_INDEX_IP4RXMCPKTS = 13,
IRDMA_HW_STAT_INDEX_IP4TXOCTS = 14,
IRDMA_HW_STAT_INDEX_IP4TXPKTS = 15,
IRDMA_HW_STAT_INDEX_IP4TXFRAGS = 16,
IRDMA_HW_STAT_INDEX_IP4TXMCPKTS = 17,
IRDMA_HW_STAT_INDEX_IP6RXOCTS = 18,
IRDMA_HW_STAT_INDEX_IP6RXPKTS = 19,
IRDMA_HW_STAT_INDEX_IP6RXFRAGS = 20,
IRDMA_HW_STAT_INDEX_IP6RXMCPKTS = 21,
IRDMA_HW_STAT_INDEX_IP6TXOCTS = 22,
IRDMA_HW_STAT_INDEX_IP6TXPKTS = 23,
IRDMA_HW_STAT_INDEX_IP6TXFRAGS = 24,
IRDMA_HW_STAT_INDEX_IP6TXMCPKTS = 25,
IRDMA_HW_STAT_INDEX_TCPRXSEGS = 26,
IRDMA_HW_STAT_INDEX_TCPTXSEG = 27,
IRDMA_HW_STAT_INDEX_RDMARXRDS = 28,
IRDMA_HW_STAT_INDEX_RDMARXSNDS = 29,
IRDMA_HW_STAT_INDEX_RDMARXWRS = 30,
IRDMA_HW_STAT_INDEX_RDMATXRDS = 31,
IRDMA_HW_STAT_INDEX_RDMATXSNDS = 32,
IRDMA_HW_STAT_INDEX_RDMATXWRS = 33,
IRDMA_HW_STAT_INDEX_RDMAVBND = 34,
IRDMA_HW_STAT_INDEX_RDMAVINV = 35,
IRDMA_HW_STAT_INDEX_IP4RXMCOCTS = 36,
IRDMA_HW_STAT_INDEX_IP4TXMCOCTS = 37,
IRDMA_HW_STAT_INDEX_IP6RXMCOCTS = 38,
IRDMA_HW_STAT_INDEX_IP6TXMCOCTS = 39,
IRDMA_HW_STAT_INDEX_UDPRXPKTS = 40,
IRDMA_HW_STAT_INDEX_UDPTXPKTS = 41,
IRDMA_HW_STAT_INDEX_MAX_GEN_1 = 42, /* Must be same value as next entry */
/* gen2 - 64-bit */
IRDMA_HW_STAT_INDEX_RXNPECNMARKEDPKTS = 42,
/* gen2 - 32-bit */
IRDMA_HW_STAT_INDEX_RXRPCNPHANDLED = 43,
IRDMA_HW_STAT_INDEX_RXRPCNPIGNORED = 44,
IRDMA_HW_STAT_INDEX_TXNPCNPSENT = 45,
IRDMA_HW_STAT_INDEX_MAX_GEN_2 = 46,
};
enum irdma_feature_type {
@ -274,65 +273,21 @@ struct irdma_cq_shadow_area {
};
struct irdma_dev_hw_stats_offsets {
u32 stats_offset_32[IRDMA_HW_STAT_INDEX_MAX_32];
u32 stats_offset_64[IRDMA_HW_STAT_INDEX_MAX_64];
u32 stats_offset[IRDMA_HW_STAT_INDEX_MAX_GEN_1];
};
struct irdma_dev_hw_stats {
u64 stats_val_32[IRDMA_HW_STAT_INDEX_MAX_32];
u64 stats_val_64[IRDMA_HW_STAT_INDEX_MAX_64];
u64 stats_val[IRDMA_GATHER_STATS_BUF_SIZE / sizeof(u64)];
};
struct irdma_gather_stats {
u32 rsvd1;
u32 rxvlanerr;
u64 ip4rxocts;
u64 ip4rxpkts;
u32 ip4rxtrunc;
u32 ip4rxdiscard;
u64 ip4rxfrags;
u64 ip4rxmcocts;
u64 ip4rxmcpkts;
u64 ip6rxocts;
u64 ip6rxpkts;
u32 ip6rxtrunc;
u32 ip6rxdiscard;
u64 ip6rxfrags;
u64 ip6rxmcocts;
u64 ip6rxmcpkts;
u64 ip4txocts;
u64 ip4txpkts;
u64 ip4txfrag;
u64 ip4txmcocts;
u64 ip4txmcpkts;
u64 ip6txocts;
u64 ip6txpkts;
u64 ip6txfrags;
u64 ip6txmcocts;
u64 ip6txmcpkts;
u32 ip6txnoroute;
u32 ip4txnoroute;
u64 tcprxsegs;
u32 tcprxprotoerr;
u32 tcprxopterr;
u64 tcptxsegs;
u32 rsvd2;
u32 tcprtxseg;
u64 udprxpkts;
u64 udptxpkts;
u64 rdmarxwrs;
u64 rdmarxrds;
u64 rdmarxsnds;
u64 rdmatxwrs;
u64 rdmatxrds;
u64 rdmatxsnds;
u64 rdmavbn;
u64 rdmavinv;
u64 rxnpecnmrkpkts;
u32 rxrpcnphandled;
u32 rxrpcnpignored;
u32 txnpcnpsent;
u32 rsvd3[88];
u64 val[IRDMA_GATHER_STATS_BUF_SIZE / sizeof(u64)];
};
struct irdma_hw_stat_map {
u16 byteoff;
u8 bitoff;
u64 bitmask;
};
struct irdma_stats_gather_info {
@ -584,7 +539,7 @@ struct irdma_qos {
bool valid;
};
#define IRDMA_INVALID_FCN_ID 0xff
#define IRDMA_INVALID_STATS_IDX 0xff
struct irdma_sc_vsi {
u16 vsi_idx;
struct irdma_sc_dev *dev;
@ -598,11 +553,9 @@ struct irdma_sc_vsi {
u32 exception_lan_q;
u16 mtu;
u16 vm_id;
u8 fcn_id;
enum irdma_vm_vf_type vm_vf_type;
bool stats_fcn_id_alloc:1;
bool stats_inst_alloc:1;
bool tc_change_pending:1;
struct irdma_qos qos[IRDMA_MAX_USER_PRIORITY];
struct irdma_vsi_pestat *pestat;
atomic_t qp_suspend_reqs;
int (*register_qset)(struct irdma_sc_vsi *vsi,
@ -611,14 +564,17 @@ struct irdma_sc_vsi {
struct irdma_ws_node *tc_node);
u8 qos_rel_bw;
u8 qos_prio_type;
u8 stats_idx;
u8 dscp_map[IIDC_MAX_DSCP_MAPPING];
struct irdma_qos qos[IRDMA_MAX_USER_PRIORITY];
u64 hw_stats_regs[IRDMA_HW_STAT_INDEX_MAX_GEN_1];
bool dscp_mode:1;
};
struct irdma_sc_dev {
struct list_head cqp_cmd_head; /* head of the CQP command list */
spinlock_t cqp_lock; /* protect CQP list access */
bool fcn_id_array[IRDMA_MAX_STATS_COUNT];
bool stats_idx_array[IRDMA_MAX_STATS_COUNT_GEN_1];
struct irdma_dma_mem vf_fpm_query_buf[IRDMA_MAX_PE_ENA_VF_COUNT];
u64 fpm_query_buf_pa;
u64 fpm_commit_buf_pa;
@ -637,8 +593,8 @@ struct irdma_sc_dev {
u32 ceq_itr; /* Interrupt throttle, usecs between interrupts: 0 disabled. 2 - 8160 */
u64 hw_masks[IRDMA_MAX_MASKS];
u64 hw_shifts[IRDMA_MAX_SHIFTS];
u64 hw_stats_regs_32[IRDMA_HW_STAT_INDEX_MAX_32];
u64 hw_stats_regs_64[IRDMA_HW_STAT_INDEX_MAX_64];
const struct irdma_hw_stat_map *hw_stats_map;
u64 hw_stats_regs[IRDMA_HW_STAT_INDEX_MAX_GEN_1];
u64 feature_info[IRDMA_MAX_FEATURES];
u64 cqp_cmd_stats[IRDMA_MAX_CQP_OPS];
struct irdma_hw_attrs hw_attrs;
@ -763,7 +719,7 @@ struct irdma_vsi_init_info {
struct irdma_vsi_stats_info {
struct irdma_vsi_pestat *pestat;
u8 fcn_id;
bool alloc_fcn_id;
bool alloc_stats_inst;
};
struct irdma_device_init_info {

172
drivers/infiniband/hw/irdma/utils.c
View File

@ -1634,10 +1634,10 @@ static void irdma_hw_stats_timeout(struct timer_list *t)
from_timer(pf_devstat, t, stats_timer);
struct irdma_sc_vsi *sc_vsi = pf_devstat->vsi;
if (sc_vsi->dev->hw_attrs.uk_attrs.hw_rev == IRDMA_GEN_1)
irdma_cqp_gather_stats_gen1(sc_vsi->dev, sc_vsi->pestat);
else
if (sc_vsi->dev->hw_attrs.uk_attrs.hw_rev >= IRDMA_GEN_2)
irdma_cqp_gather_stats_cmd(sc_vsi->dev, sc_vsi->pestat, false);
else
irdma_cqp_gather_stats_gen1(sc_vsi->dev, sc_vsi->pestat);
mod_timer(&pf_devstat->stats_timer,
jiffies + msecs_to_jiffies(STATS_TIMER_DELAY));
@ -1686,164 +1686,28 @@ void irdma_cqp_gather_stats_gen1(struct irdma_sc_dev *dev,
{
struct irdma_gather_stats *gather_stats =
pestat->gather_info.gather_stats_va;
const struct irdma_hw_stat_map *map = dev->hw_stats_map;
u16 max_stats_idx = dev->hw_attrs.max_stat_idx;
u32 stats_inst_offset_32;
u32 stats_inst_offset_64;
u64 new_val;
u16 i;
stats_inst_offset_32 = (pestat->gather_info.use_stats_inst) ?
pestat->gather_info.stats_inst_index :
pestat->hw->hmc.hmc_fn_id;
pestat->gather_info.stats_inst_index :
pestat->hw->hmc.hmc_fn_id;
stats_inst_offset_32 *= 4;
stats_inst_offset_64 = stats_inst_offset_32 * 2;
gather_stats->rxvlanerr =
rd32(dev->hw,
dev->hw_stats_regs_32[IRDMA_HW_STAT_INDEX_RXVLANERR]
+ stats_inst_offset_32);
gather_stats->ip4rxdiscard =
rd32(dev->hw,
dev->hw_stats_regs_32[IRDMA_HW_STAT_INDEX_IP4RXDISCARD]
+ stats_inst_offset_32);
gather_stats->ip4rxtrunc =
rd32(dev->hw,
dev->hw_stats_regs_32[IRDMA_HW_STAT_INDEX_IP4RXTRUNC]
+ stats_inst_offset_32);
gather_stats->ip4txnoroute =
rd32(dev->hw,
dev->hw_stats_regs_32[IRDMA_HW_STAT_INDEX_IP4TXNOROUTE]
+ stats_inst_offset_32);
gather_stats->ip6rxdiscard =
rd32(dev->hw,
dev->hw_stats_regs_32[IRDMA_HW_STAT_INDEX_IP6RXDISCARD]
+ stats_inst_offset_32);
gather_stats->ip6rxtrunc =
rd32(dev->hw,
dev->hw_stats_regs_32[IRDMA_HW_STAT_INDEX_IP6RXTRUNC]
+ stats_inst_offset_32);
gather_stats->ip6txnoroute =
rd32(dev->hw,
dev->hw_stats_regs_32[IRDMA_HW_STAT_INDEX_IP6TXNOROUTE]
+ stats_inst_offset_32);
gather_stats->tcprtxseg =
rd32(dev->hw,
dev->hw_stats_regs_32[IRDMA_HW_STAT_INDEX_TCPRTXSEG]
+ stats_inst_offset_32);
gather_stats->tcprxopterr =
rd32(dev->hw,
dev->hw_stats_regs_32[IRDMA_HW_STAT_INDEX_TCPRXOPTERR]
+ stats_inst_offset_32);
gather_stats->ip4rxocts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP4RXOCTS]
+ stats_inst_offset_64);
gather_stats->ip4rxpkts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP4RXPKTS]
+ stats_inst_offset_64);
gather_stats->ip4txfrag =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP4RXFRAGS]
+ stats_inst_offset_64);
gather_stats->ip4rxmcpkts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP4RXMCPKTS]
+ stats_inst_offset_64);
gather_stats->ip4txocts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP4TXOCTS]
+ stats_inst_offset_64);
gather_stats->ip4txpkts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP4TXPKTS]
+ stats_inst_offset_64);
gather_stats->ip4txfrag =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP4TXFRAGS]
+ stats_inst_offset_64);
gather_stats->ip4txmcpkts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP4TXMCPKTS]
+ stats_inst_offset_64);
gather_stats->ip6rxocts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP6RXOCTS]
+ stats_inst_offset_64);
gather_stats->ip6rxpkts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP6RXPKTS]
+ stats_inst_offset_64);
gather_stats->ip6txfrags =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP6RXFRAGS]
+ stats_inst_offset_64);
gather_stats->ip6rxmcpkts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP6RXMCPKTS]
+ stats_inst_offset_64);
gather_stats->ip6txocts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP6TXOCTS]
+ stats_inst_offset_64);
gather_stats->ip6txpkts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP6TXPKTS]
+ stats_inst_offset_64);
gather_stats->ip6txfrags =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP6TXFRAGS]
+ stats_inst_offset_64);
gather_stats->ip6txmcpkts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_IP6TXMCPKTS]
+ stats_inst_offset_64);
gather_stats->tcprxsegs =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_TCPRXSEGS]
+ stats_inst_offset_64);
gather_stats->tcptxsegs =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_TCPTXSEG]
+ stats_inst_offset_64);
gather_stats->rdmarxrds =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_RDMARXRDS]
+ stats_inst_offset_64);
gather_stats->rdmarxsnds =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_RDMARXSNDS]
+ stats_inst_offset_64);
gather_stats->rdmarxwrs =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_RDMARXWRS]
+ stats_inst_offset_64);
gather_stats->rdmatxrds =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_RDMATXRDS]
+ stats_inst_offset_64);
gather_stats->rdmatxsnds =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_RDMATXSNDS]
+ stats_inst_offset_64);
gather_stats->rdmatxwrs =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_RDMATXWRS]
+ stats_inst_offset_64);
gather_stats->rdmavbn =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_RDMAVBND]
+ stats_inst_offset_64);
gather_stats->rdmavinv =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_RDMAVINV]
+ stats_inst_offset_64);
gather_stats->udprxpkts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_UDPRXPKTS]
+ stats_inst_offset_64);
gather_stats->udptxpkts =
rd64(dev->hw,
dev->hw_stats_regs_64[IRDMA_HW_STAT_INDEX_UDPTXPKTS]
+ stats_inst_offset_64);
for (i = 0; i < max_stats_idx; i++) {
if (map[i].bitmask <= IRDMA_MAX_STATS_32)
new_val = rd32(dev->hw,
dev->hw_stats_regs[i] + stats_inst_offset_32);
else
new_val = rd64(dev->hw,
dev->hw_stats_regs[i] + stats_inst_offset_64);
gather_stats->val[map[i].byteoff / sizeof(u64)] = new_val;
}
irdma_process_stats(pestat);
}

198
drivers/infiniband/hw/irdma/verbs.c
View File

@ -1226,10 +1226,6 @@ int irdma_modify_qp_roce(struct ib_qp *ibqp, struct ib_qp_attr *attr,
udp_info->ipv4 = false;
irdma_copy_ip_ntohl(local_ip, daddr);
udp_info->arp_idx = irdma_arp_table(iwdev->rf,
&local_ip[0],
false, NULL,
IRDMA_ARP_RESOLVE);
} else if (av->net_type == RDMA_NETWORK_IPV4) {
__be32 saddr = av->sgid_addr.saddr_in.sin_addr.s_addr;
__be32 daddr = av->dgid_addr.saddr_in.sin_addr.s_addr;
@ -2329,11 +2325,10 @@ static bool irdma_check_mr_contiguous(struct irdma_pble_alloc *palloc,
* irdma_setup_pbles - copy user pg address to pble's
* @rf: RDMA PCI function
* @iwmr: mr pointer for this memory registration
* @use_pbles: flag if to use pble's
* @lvl_1_only: request only level 1 pble if true
* @lvl: requested pble levels
*/
static int irdma_setup_pbles(struct irdma_pci_f *rf, struct irdma_mr *iwmr,
bool use_pbles, bool lvl_1_only)
u8 lvl)
{
struct irdma_pbl *iwpbl = &iwmr->iwpbl;
struct irdma_pble_alloc *palloc = &iwpbl->pble_alloc;
@ -2342,9 +2337,9 @@ static int irdma_setup_pbles(struct irdma_pci_f *rf, struct irdma_mr *iwmr,
int status;
enum irdma_pble_level level = PBLE_LEVEL_1;
if (use_pbles) {
if (lvl) {
status = irdma_get_pble(rf->pble_rsrc, palloc, iwmr->page_cnt,
lvl_1_only);
lvl);
if (status)
return status;
@ -2359,7 +2354,7 @@ static int irdma_setup_pbles(struct irdma_pci_f *rf, struct irdma_mr *iwmr,
irdma_copy_user_pgaddrs(iwmr, pbl, level);
if (use_pbles)
if (lvl)
iwmr->pgaddrmem[0] = *pbl;
return 0;
@ -2370,11 +2365,11 @@ static int irdma_setup_pbles(struct irdma_pci_f *rf, struct irdma_mr *iwmr,
* @iwdev: irdma device
* @req: information for q memory management
* @iwpbl: pble struct
* @use_pbles: flag to use pble
* @lvl: pble level mask
*/
static int irdma_handle_q_mem(struct irdma_device *iwdev,
struct irdma_mem_reg_req *req,
struct irdma_pbl *iwpbl, bool use_pbles)
struct irdma_pbl *iwpbl, u8 lvl)
{
struct irdma_pble_alloc *palloc = &iwpbl->pble_alloc;
struct irdma_mr *iwmr = iwpbl->iwmr;
@ -2387,11 +2382,11 @@ static int irdma_handle_q_mem(struct irdma_device *iwdev,
bool ret = true;
pg_size = iwmr->page_size;
err = irdma_setup_pbles(iwdev->rf, iwmr, use_pbles, true);
err = irdma_setup_pbles(iwdev->rf, iwmr, lvl);
if (err)
return err;
if (use_pbles)
if (lvl)
arr = palloc->level1.addr;
switch (iwmr->type) {
@ -2400,7 +2395,7 @@ static int irdma_handle_q_mem(struct irdma_device *iwdev,
hmc_p = &qpmr->sq_pbl;
qpmr->shadow = (dma_addr_t)arr[total];
if (use_pbles) {
if (lvl) {
ret = irdma_check_mem_contiguous(arr, req->sq_pages,
pg_size);
if (ret)
@ -2425,7 +2420,7 @@ static int irdma_handle_q_mem(struct irdma_device *iwdev,
if (!cqmr->split)
cqmr->shadow = (dma_addr_t)arr[req->cq_pages];
if (use_pbles)
if (lvl)
ret = irdma_check_mem_contiguous(arr, req->cq_pages,
pg_size);
@ -2439,7 +2434,7 @@ static int irdma_handle_q_mem(struct irdma_device *iwdev,
err = -EINVAL;
}
if (use_pbles && ret) {
if (lvl && ret) {
irdma_free_pble(iwdev->rf->pble_rsrc, palloc);
iwpbl->pbl_allocated = false;
}
@ -2749,17 +2744,17 @@ static int irdma_reg_user_mr_type_mem(struct irdma_mr *iwmr, int access)
{
struct irdma_device *iwdev = to_iwdev(iwmr->ibmr.device);
struct irdma_pbl *iwpbl = &iwmr->iwpbl;
bool use_pbles;
u32 stag;
u8 lvl;
int err;
use_pbles = iwmr->page_cnt != 1;
lvl = iwmr->page_cnt != 1 ? PBLE_LEVEL_1 | PBLE_LEVEL_2 : PBLE_LEVEL_0;
err = irdma_setup_pbles(iwdev->rf, iwmr, use_pbles, false);
err = irdma_setup_pbles(iwdev->rf, iwmr, lvl);
if (err)
return err;
if (use_pbles) {
if (lvl) {
err = irdma_check_mr_contiguous(&iwpbl->pble_alloc,
iwmr->page_size);
if (err) {
@ -2843,17 +2838,17 @@ static int irdma_reg_user_mr_type_qp(struct irdma_mem_reg_req req,
struct irdma_pbl *iwpbl = &iwmr->iwpbl;
struct irdma_ucontext *ucontext = NULL;
unsigned long flags;
bool use_pbles;
u32 total;
int err;
u8 lvl;
total = req.sq_pages + req.rq_pages + 1;
if (total > iwmr->page_cnt)
return -EINVAL;
total = req.sq_pages + req.rq_pages;
use_pbles = total > 2;
err = irdma_handle_q_mem(iwdev, &req, iwpbl, use_pbles);
lvl = total > 2 ? PBLE_LEVEL_1 : PBLE_LEVEL_0;
err = irdma_handle_q_mem(iwdev, &req, iwpbl, lvl);
if (err)
return err;
@ -2876,9 +2871,9 @@ static int irdma_reg_user_mr_type_cq(struct irdma_mem_reg_req req,
struct irdma_ucontext *ucontext = NULL;
u8 shadow_pgcnt = 1;
unsigned long flags;
bool use_pbles;
u32 total;
int err;
u8 lvl;
if (iwdev->rf->sc_dev.hw_attrs.uk_attrs.feature_flags & IRDMA_FEATURE_CQ_RESIZE)
shadow_pgcnt = 0;
@ -2886,8 +2881,8 @@ static int irdma_reg_user_mr_type_cq(struct irdma_mem_reg_req req,
if (total > iwmr->page_cnt)
return -EINVAL;
use_pbles = req.cq_pages > 1;
err = irdma_handle_q_mem(iwdev, &req, iwpbl, use_pbles);
lvl = req.cq_pages > 1 ? PBLE_LEVEL_1 : PBLE_LEVEL_0;
err = irdma_handle_q_mem(iwdev, &req, iwpbl, lvl);
if (err)
return err;
@ -3708,89 +3703,59 @@ static int irdma_iw_port_immutable(struct ib_device *ibdev, u32 port_num,
return 0;
}
static const struct rdma_stat_desc irdma_hw_stat_descs[] = {
/* 32bit names */
[IRDMA_HW_STAT_INDEX_RXVLANERR].name = "rxVlanErrors",
[IRDMA_HW_STAT_INDEX_IP4RXDISCARD].name = "ip4InDiscards",
[IRDMA_HW_STAT_INDEX_IP4RXTRUNC].name = "ip4InTruncatedPkts",
[IRDMA_HW_STAT_INDEX_IP4TXNOROUTE].name = "ip4OutNoRoutes",
[IRDMA_HW_STAT_INDEX_IP6RXDISCARD].name = "ip6InDiscards",
[IRDMA_HW_STAT_INDEX_IP6RXTRUNC].name = "ip6InTruncatedPkts",
[IRDMA_HW_STAT_INDEX_IP6TXNOROUTE].name = "ip6OutNoRoutes",
[IRDMA_HW_STAT_INDEX_TCPRTXSEG].name = "tcpRetransSegs",
[IRDMA_HW_STAT_INDEX_TCPRXOPTERR].name = "tcpInOptErrors",
[IRDMA_HW_STAT_INDEX_TCPRXPROTOERR].name = "tcpInProtoErrors",
[IRDMA_HW_STAT_INDEX_RXRPCNPHANDLED].name = "cnpHandled",
[IRDMA_HW_STAT_INDEX_RXRPCNPIGNORED].name = "cnpIgnored",
[IRDMA_HW_STAT_INDEX_TXNPCNPSENT].name = "cnpSent",
static const struct rdma_stat_desc irdma_hw_stat_names[] = {
/* gen1 - 32-bit */
[IRDMA_HW_STAT_INDEX_IP4RXDISCARD].name = "ip4InDiscards",
[IRDMA_HW_STAT_INDEX_IP4RXTRUNC].name = "ip4InTruncatedPkts",
[IRDMA_HW_STAT_INDEX_IP4TXNOROUTE].name = "ip4OutNoRoutes",
[IRDMA_HW_STAT_INDEX_IP6RXDISCARD].name = "ip6InDiscards",
[IRDMA_HW_STAT_INDEX_IP6RXTRUNC].name = "ip6InTruncatedPkts",
[IRDMA_HW_STAT_INDEX_IP6TXNOROUTE].name = "ip6OutNoRoutes",
[IRDMA_HW_STAT_INDEX_TCPRTXSEG].name = "tcpRetransSegs",
[IRDMA_HW_STAT_INDEX_TCPRXOPTERR].name = "tcpInOptErrors",
[IRDMA_HW_STAT_INDEX_TCPRXPROTOERR].name = "tcpInProtoErrors",
[IRDMA_HW_STAT_INDEX_RXVLANERR].name = "rxVlanErrors",
/* gen1 - 64-bit */
[IRDMA_HW_STAT_INDEX_IP4RXOCTS].name = "ip4InOctets",
[IRDMA_HW_STAT_INDEX_IP4RXPKTS].name = "ip4InPkts",
[IRDMA_HW_STAT_INDEX_IP4RXFRAGS].name = "ip4InReasmRqd",
[IRDMA_HW_STAT_INDEX_IP4RXMCPKTS].name = "ip4InMcastPkts",
[IRDMA_HW_STAT_INDEX_IP4TXOCTS].name = "ip4OutOctets",
[IRDMA_HW_STAT_INDEX_IP4TXPKTS].name = "ip4OutPkts",
[IRDMA_HW_STAT_INDEX_IP4TXFRAGS].name = "ip4OutSegRqd",
[IRDMA_HW_STAT_INDEX_IP4TXMCPKTS].name = "ip4OutMcastPkts",
[IRDMA_HW_STAT_INDEX_IP6RXOCTS].name = "ip6InOctets",
[IRDMA_HW_STAT_INDEX_IP6RXPKTS].name = "ip6InPkts",
[IRDMA_HW_STAT_INDEX_IP6RXFRAGS].name = "ip6InReasmRqd",
[IRDMA_HW_STAT_INDEX_IP6RXMCPKTS].name = "ip6InMcastPkts",
[IRDMA_HW_STAT_INDEX_IP6TXOCTS].name = "ip6OutOctets",
[IRDMA_HW_STAT_INDEX_IP6TXPKTS].name = "ip6OutPkts",
[IRDMA_HW_STAT_INDEX_IP6TXFRAGS].name = "ip6OutSegRqd",
[IRDMA_HW_STAT_INDEX_IP6TXMCPKTS].name = "ip6OutMcastPkts",
[IRDMA_HW_STAT_INDEX_TCPRXSEGS].name = "tcpInSegs",
[IRDMA_HW_STAT_INDEX_TCPTXSEG].name = "tcpOutSegs",
[IRDMA_HW_STAT_INDEX_RDMARXRDS].name = "iwInRdmaReads",
[IRDMA_HW_STAT_INDEX_RDMARXSNDS].name = "iwInRdmaSends",
[IRDMA_HW_STAT_INDEX_RDMARXWRS].name = "iwInRdmaWrites",
[IRDMA_HW_STAT_INDEX_RDMATXRDS].name = "iwOutRdmaReads",
[IRDMA_HW_STAT_INDEX_RDMATXSNDS].name = "iwOutRdmaSends",
[IRDMA_HW_STAT_INDEX_RDMATXWRS].name = "iwOutRdmaWrites",
[IRDMA_HW_STAT_INDEX_RDMAVBND].name = "iwRdmaBnd",
[IRDMA_HW_STAT_INDEX_RDMAVINV].name = "iwRdmaInv",
/* gen2 - 32-bit */
[IRDMA_HW_STAT_INDEX_RXRPCNPHANDLED].name = "cnpHandled",
[IRDMA_HW_STAT_INDEX_RXRPCNPIGNORED].name = "cnpIgnored",
[IRDMA_HW_STAT_INDEX_TXNPCNPSENT].name = "cnpSent",
/* gen2 - 64-bit */
[IRDMA_HW_STAT_INDEX_IP4RXMCOCTS].name = "ip4InMcastOctets",
[IRDMA_HW_STAT_INDEX_IP4TXMCOCTS].name = "ip4OutMcastOctets",
[IRDMA_HW_STAT_INDEX_IP6RXMCOCTS].name = "ip6InMcastOctets",
[IRDMA_HW_STAT_INDEX_IP6TXMCOCTS].name = "ip6OutMcastOctets",
[IRDMA_HW_STAT_INDEX_UDPRXPKTS].name = "RxUDP",
[IRDMA_HW_STAT_INDEX_UDPTXPKTS].name = "TxUDP",
[IRDMA_HW_STAT_INDEX_RXNPECNMARKEDPKTS].name = "RxECNMrkd",
/* 64bit names */
[IRDMA_HW_STAT_INDEX_IP4RXOCTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip4InOctets",
[IRDMA_HW_STAT_INDEX_IP4RXPKTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip4InPkts",
[IRDMA_HW_STAT_INDEX_IP4RXFRAGS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip4InReasmRqd",
[IRDMA_HW_STAT_INDEX_IP4RXMCOCTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip4InMcastOctets",
[IRDMA_HW_STAT_INDEX_IP4RXMCPKTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip4InMcastPkts",
[IRDMA_HW_STAT_INDEX_IP4TXOCTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip4OutOctets",
[IRDMA_HW_STAT_INDEX_IP4TXPKTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip4OutPkts",
[IRDMA_HW_STAT_INDEX_IP4TXFRAGS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip4OutSegRqd",
[IRDMA_HW_STAT_INDEX_IP4TXMCOCTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip4OutMcastOctets",
[IRDMA_HW_STAT_INDEX_IP4TXMCPKTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip4OutMcastPkts",
[IRDMA_HW_STAT_INDEX_IP6RXOCTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip6InOctets",
[IRDMA_HW_STAT_INDEX_IP6RXPKTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip6InPkts",
[IRDMA_HW_STAT_INDEX_IP6RXFRAGS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip6InReasmRqd",
[IRDMA_HW_STAT_INDEX_IP6RXMCOCTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip6InMcastOctets",
[IRDMA_HW_STAT_INDEX_IP6RXMCPKTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip6InMcastPkts",
[IRDMA_HW_STAT_INDEX_IP6TXOCTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip6OutOctets",
[IRDMA_HW_STAT_INDEX_IP6TXPKTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip6OutPkts",
[IRDMA_HW_STAT_INDEX_IP6TXFRAGS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip6OutSegRqd",
[IRDMA_HW_STAT_INDEX_IP6TXMCOCTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip6OutMcastOctets",
[IRDMA_HW_STAT_INDEX_IP6TXMCPKTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"ip6OutMcastPkts",
[IRDMA_HW_STAT_INDEX_TCPRXSEGS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"tcpInSegs",
[IRDMA_HW_STAT_INDEX_TCPTXSEG + IRDMA_HW_STAT_INDEX_MAX_32].name =
"tcpOutSegs",
[IRDMA_HW_STAT_INDEX_RDMARXRDS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"iwInRdmaReads",
[IRDMA_HW_STAT_INDEX_RDMARXSNDS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"iwInRdmaSends",
[IRDMA_HW_STAT_INDEX_RDMARXWRS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"iwInRdmaWrites",
[IRDMA_HW_STAT_INDEX_RDMATXRDS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"iwOutRdmaReads",
[IRDMA_HW_STAT_INDEX_RDMATXSNDS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"iwOutRdmaSends",
[IRDMA_HW_STAT_INDEX_RDMATXWRS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"iwOutRdmaWrites",
[IRDMA_HW_STAT_INDEX_RDMAVBND + IRDMA_HW_STAT_INDEX_MAX_32].name =
"iwRdmaBnd",
[IRDMA_HW_STAT_INDEX_RDMAVINV + IRDMA_HW_STAT_INDEX_MAX_32].name =
"iwRdmaInv",
[IRDMA_HW_STAT_INDEX_UDPRXPKTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"RxUDP",
[IRDMA_HW_STAT_INDEX_UDPTXPKTS + IRDMA_HW_STAT_INDEX_MAX_32].name =
"TxUDP",
[IRDMA_HW_STAT_INDEX_RXNPECNMARKEDPKTS + IRDMA_HW_STAT_INDEX_MAX_32]
.name = "RxECNMrkd",
};
static void irdma_get_dev_fw_str(struct ib_device *dev, char *str)
@ -3810,14 +3775,13 @@ static void irdma_get_dev_fw_str(struct ib_device *dev, char *str)
static struct rdma_hw_stats *irdma_alloc_hw_port_stats(struct ib_device *ibdev,
u32 port_num)
{
int num_counters = IRDMA_HW_STAT_INDEX_MAX_32 +
IRDMA_HW_STAT_INDEX_MAX_64;
struct irdma_device *iwdev = to_iwdev(ibdev);
struct irdma_sc_dev *dev = &iwdev->rf->sc_dev;
int num_counters = dev->hw_attrs.max_stat_idx;
unsigned long lifespan = RDMA_HW_STATS_DEFAULT_LIFESPAN;
BUILD_BUG_ON(ARRAY_SIZE(irdma_hw_stat_descs) !=
(IRDMA_HW_STAT_INDEX_MAX_32 + IRDMA_HW_STAT_INDEX_MAX_64));
return rdma_alloc_hw_stats_struct(irdma_hw_stat_descs, num_counters,
return rdma_alloc_hw_stats_struct(irdma_hw_stat_names, num_counters,
lifespan);
}
@ -3840,7 +3804,7 @@ static int irdma_get_hw_stats(struct ib_device *ibdev,
else
irdma_cqp_gather_stats_gen1(&iwdev->rf->sc_dev, iwdev->vsi.pestat);
memcpy(&stats->value[0], hw_stats, sizeof(*hw_stats));
memcpy(&stats->value[0], hw_stats, sizeof(u64) * stats->num_counters);
return stats->num_counters;
}
@ -4054,7 +4018,7 @@ static int irdma_attach_mcast(struct ib_qp *ibqp, union ib_gid *ibgid, u16 lid)
mc_qht_elem->mc_grp_ctx.vlan_id = vlan_id;
if (vlan_id < VLAN_N_VID)
mc_qht_elem->mc_grp_ctx.vlan_valid = true;
mc_qht_elem->mc_grp_ctx.hmc_fcn_id = iwdev->vsi.fcn_id;
mc_qht_elem->mc_grp_ctx.hmc_fcn_id = iwdev->rf->sc_dev.hmc_fn_id;
mc_qht_elem->mc_grp_ctx.qs_handle =
iwqp->sc_qp.vsi->qos[iwqp->sc_qp.user_pri].qs_handle;
ether_addr_copy(mc_qht_elem->mc_grp_ctx.dest_mac_addr, dmac);

8
drivers/infiniband/hw/mlx4/qp.c
View File

@ -447,9 +447,13 @@ static int set_user_sq_size(struct mlx4_ib_dev *dev,
struct mlx4_ib_qp *qp,
struct mlx4_ib_create_qp *ucmd)
{
u32 cnt;
/* Sanity check SQ size before proceeding */
if ((1 << ucmd->log_sq_bb_count) > dev->dev->caps.max_wqes ||
ucmd->log_sq_stride >
if (check_shl_overflow(1, ucmd->log_sq_bb_count, &cnt) ||
cnt > dev->dev->caps.max_wqes)
return -EINVAL;
if (ucmd->log_sq_stride >
ilog2(roundup_pow_of_two(dev->dev->caps.max_sq_desc_sz)) ||
ucmd->log_sq_stride < MLX4_IB_MIN_SQ_STRIDE)
return -EINVAL;

171
drivers/infiniband/hw/mlx5/counters.c
View File

@ -5,6 +5,7 @@
#include "mlx5_ib.h"
#include <linux/mlx5/eswitch.h>
#include <linux/mlx5/vport.h>
#include "counters.h"
#include "ib_rep.h"
#include "qp.h"
@ -18,6 +19,10 @@ struct mlx5_ib_counter {
#define INIT_Q_COUNTER(_name) \
{ .name = #_name, .offset = MLX5_BYTE_OFF(query_q_counter_out, _name)}
#define INIT_VPORT_Q_COUNTER(_name) \
{ .name = "vport_" #_name, .offset = \
MLX5_BYTE_OFF(query_q_counter_out, _name)}
static const struct mlx5_ib_counter basic_q_cnts[] = {
INIT_Q_COUNTER(rx_write_requests),
INIT_Q_COUNTER(rx_read_requests),
@ -37,6 +42,25 @@ static const struct mlx5_ib_counter retrans_q_cnts[] = {
INIT_Q_COUNTER(local_ack_timeout_err),
};
static const struct mlx5_ib_counter vport_basic_q_cnts[] = {
INIT_VPORT_Q_COUNTER(rx_write_requests),
INIT_VPORT_Q_COUNTER(rx_read_requests),
INIT_VPORT_Q_COUNTER(rx_atomic_requests),
INIT_VPORT_Q_COUNTER(out_of_buffer),
};
static const struct mlx5_ib_counter vport_out_of_seq_q_cnts[] = {
INIT_VPORT_Q_COUNTER(out_of_sequence),
};
static const struct mlx5_ib_counter vport_retrans_q_cnts[] = {
INIT_VPORT_Q_COUNTER(duplicate_request),
INIT_VPORT_Q_COUNTER(rnr_nak_retry_err),
INIT_VPORT_Q_COUNTER(packet_seq_err),
INIT_VPORT_Q_COUNTER(implied_nak_seq_err),
INIT_VPORT_Q_COUNTER(local_ack_timeout_err),
};
#define INIT_CONG_COUNTER(_name) \
{ .name = #_name, .offset = \
MLX5_BYTE_OFF(query_cong_statistics_out, _name ## _high)}
@ -67,6 +91,25 @@ static const struct mlx5_ib_counter roce_accl_cnts[] = {
INIT_Q_COUNTER(roce_slow_restart_trans),
};
static const struct mlx5_ib_counter vport_extended_err_cnts[] = {
INIT_VPORT_Q_COUNTER(resp_local_length_error),
INIT_VPORT_Q_COUNTER(resp_cqe_error),
INIT_VPORT_Q_COUNTER(req_cqe_error),
INIT_VPORT_Q_COUNTER(req_remote_invalid_request),
INIT_VPORT_Q_COUNTER(req_remote_access_errors),
INIT_VPORT_Q_COUNTER(resp_remote_access_errors),
INIT_VPORT_Q_COUNTER(resp_cqe_flush_error),
INIT_VPORT_Q_COUNTER(req_cqe_flush_error),
};
static const struct mlx5_ib_counter vport_roce_accl_cnts[] = {
INIT_VPORT_Q_COUNTER(roce_adp_retrans),
INIT_VPORT_Q_COUNTER(roce_adp_retrans_to),
INIT_VPORT_Q_COUNTER(roce_slow_restart),
INIT_VPORT_Q_COUNTER(roce_slow_restart_cnps),
INIT_VPORT_Q_COUNTER(roce_slow_restart_trans),
};
#define INIT_EXT_PPCNT_COUNTER(_name) \
{ .name = #_name, .offset = \
MLX5_BYTE_OFF(ppcnt_reg, \
@ -153,12 +196,20 @@ static int mlx5_ib_create_counters(struct ib_counters *counters,
return 0;
}
static bool vport_qcounters_supported(struct mlx5_ib_dev *dev)
{
return MLX5_CAP_GEN(dev->mdev, q_counter_other_vport) &&
MLX5_CAP_GEN(dev->mdev, q_counter_aggregation);
}
static const struct mlx5_ib_counters *get_counters(struct mlx5_ib_dev *dev,
u32 port_num)
{
return is_mdev_switchdev_mode(dev->mdev) ? &dev->port[0].cnts :
&dev->port[port_num].cnts;
if ((is_mdev_switchdev_mode(dev->mdev) &&
!vport_qcounters_supported(dev)) || !port_num)
return &dev->port[0].cnts;
return &dev->port[port_num - 1].cnts;
}
/**
@ -172,7 +223,7 @@ static const struct mlx5_ib_counters *get_counters(struct mlx5_ib_dev *dev,
*/
u16 mlx5_ib_get_counters_id(struct mlx5_ib_dev *dev, u32 port_num)
{
const struct mlx5_ib_counters *cnts = get_counters(dev, port_num);
const struct mlx5_ib_counters *cnts = get_counters(dev, port_num + 1);
return cnts->set_id;
}
@ -270,12 +321,44 @@ free:
return ret;
}
static int mlx5_ib_query_q_counters_vport(struct mlx5_ib_dev *dev,
u32 port_num,
const struct mlx5_ib_counters *cnts,
struct rdma_hw_stats *stats)
{
u32 out[MLX5_ST_SZ_DW(query_q_counter_out)] = {};
u32 in[MLX5_ST_SZ_DW(query_q_counter_in)] = {};
__be32 val;
int ret, i;
if (!dev->port[port_num].rep ||
dev->port[port_num].rep->vport == MLX5_VPORT_UPLINK)
return 0;
MLX5_SET(query_q_counter_in, in, opcode, MLX5_CMD_OP_QUERY_Q_COUNTER);
MLX5_SET(query_q_counter_in, in, other_vport, 1);
MLX5_SET(query_q_counter_in, in, vport_number,
dev->port[port_num].rep->vport);
MLX5_SET(query_q_counter_in, in, aggregate, 1);
ret = mlx5_cmd_exec_inout(dev->mdev, query_q_counter, in, out);
if (ret)
return ret;
for (i = 0; i < cnts->num_q_counters; i++) {
val = *(__be32 *)((void *)out + cnts->offsets[i]);
stats->value[i] = (u64)be32_to_cpu(val);
}
return 0;
}
static int do_get_hw_stats(struct ib_device *ibdev,
struct rdma_hw_stats *stats,
u32 port_num, int index)
{
struct mlx5_ib_dev *dev = to_mdev(ibdev);
const struct mlx5_ib_counters *cnts = get_counters(dev, port_num - 1);
const struct mlx5_ib_counters *cnts = get_counters(dev, port_num);
struct mlx5_core_dev *mdev;
int ret, num_counters;
@ -286,11 +369,19 @@ static int do_get_hw_stats(struct ib_device *ibdev,
cnts->num_cong_counters +
cnts->num_ext_ppcnt_counters;
/* q_counters are per IB device, query the master mdev */
ret = mlx5_ib_query_q_counters(dev->mdev, cnts, stats, cnts->set_id);
if (is_mdev_switchdev_mode(dev->mdev) && dev->is_rep && port_num != 0)
ret = mlx5_ib_query_q_counters_vport(dev, port_num - 1, cnts,
stats);
else
ret = mlx5_ib_query_q_counters(dev->mdev, cnts, stats,
cnts->set_id);
if (ret)
return ret;
/* We don't expose device counters over Vports */
if (is_mdev_switchdev_mode(dev->mdev) && port_num != 0)
goto done;
if (MLX5_CAP_PCAM_FEATURE(dev->mdev, rx_icrc_encapsulated_counter)) {
ret = mlx5_ib_query_ext_ppcnt_counters(dev, cnts, stats);
if (ret)
@ -335,7 +426,8 @@ static int do_get_op_stat(struct ib_device *ibdev,
u32 type;
int ret;
cnts = get_counters(dev, port_num - 1);
cnts = get_counters(dev, port_num);
opfcs = cnts->opfcs;
type = *(u32 *)cnts->descs[index].priv;
if (type >= MLX5_IB_OPCOUNTER_MAX)
@ -362,7 +454,7 @@ static int do_get_op_stats(struct ib_device *ibdev,
const struct mlx5_ib_counters *cnts;
int index, ret, num_hw_counters;
cnts = get_counters(dev, port_num - 1);
cnts = get_counters(dev, port_num);
num_hw_counters = cnts->num_q_counters + cnts->num_cong_counters +
cnts->num_ext_ppcnt_counters;
for (index = num_hw_counters;
@ -383,7 +475,7 @@ static int mlx5_ib_get_hw_stats(struct ib_device *ibdev,
struct mlx5_ib_dev *dev = to_mdev(ibdev);
const struct mlx5_ib_counters *cnts;
cnts = get_counters(dev, port_num - 1);
cnts = get_counters(dev, port_num);
num_hw_counters = cnts->num_q_counters + cnts->num_cong_counters +
cnts->num_ext_ppcnt_counters;
num_counters = num_hw_counters + cnts->num_op_counters;
@ -410,8 +502,7 @@ static struct rdma_hw_stats *
mlx5_ib_counter_alloc_stats(struct rdma_counter *counter)
{
struct mlx5_ib_dev *dev = to_mdev(counter->device);
const struct mlx5_ib_counters *cnts =
get_counters(dev, counter->port - 1);
const struct mlx5_ib_counters *cnts = get_counters(dev, counter->port);
return do_alloc_stats(cnts);
}
@ -419,8 +510,7 @@ mlx5_ib_counter_alloc_stats(struct rdma_counter *counter)
static int mlx5_ib_counter_update_stats(struct rdma_counter *counter)
{
struct mlx5_ib_dev *dev = to_mdev(counter->device);
const struct mlx5_ib_counters *cnts =
get_counters(dev, counter->port - 1);
const struct mlx5_ib_counters *cnts = get_counters(dev, counter->port);
return mlx5_ib_query_q_counters(dev->mdev, cnts,
counter->stats, counter->id);
@ -479,44 +569,55 @@ static int mlx5_ib_counter_unbind_qp(struct ib_qp *qp)
}
static void mlx5_ib_fill_counters(struct mlx5_ib_dev *dev,
struct rdma_stat_desc *descs, size_t *offsets)
struct rdma_stat_desc *descs, size_t *offsets,
u32 port_num)
{
int i;
int j = 0;
bool is_vport = is_mdev_switchdev_mode(dev->mdev) &&
port_num != MLX5_VPORT_PF;
const struct mlx5_ib_counter *names;
int j = 0, i;
names = is_vport ? vport_basic_q_cnts : basic_q_cnts;
for (i = 0; i < ARRAY_SIZE(basic_q_cnts); i++, j++) {
descs[j].name = basic_q_cnts[i].name;
descs[j].name = names[i].name;
offsets[j] = basic_q_cnts[i].offset;
}
names = is_vport ? vport_out_of_seq_q_cnts : out_of_seq_q_cnts;
if (MLX5_CAP_GEN(dev->mdev, out_of_seq_cnt)) {
for (i = 0; i < ARRAY_SIZE(out_of_seq_q_cnts); i++, j++) {
descs[j].name = out_of_seq_q_cnts[i].name;
descs[j].name = names[i].name;
offsets[j] = out_of_seq_q_cnts[i].offset;
}
}
names = is_vport ? vport_retrans_q_cnts : retrans_q_cnts;
if (MLX5_CAP_GEN(dev->mdev, retransmission_q_counters)) {
for (i = 0; i < ARRAY_SIZE(retrans_q_cnts); i++, j++) {
descs[j].name = retrans_q_cnts[i].name;
descs[j].name = names[i].name;
offsets[j] = retrans_q_cnts[i].offset;
}
}
names = is_vport ? vport_extended_err_cnts : extended_err_cnts;
if (MLX5_CAP_GEN(dev->mdev, enhanced_error_q_counters)) {
for (i = 0; i < ARRAY_SIZE(extended_err_cnts); i++, j++) {
descs[j].name = extended_err_cnts[i].name;
descs[j].name = names[i].name;
offsets[j] = extended_err_cnts[i].offset;
}
}
names = is_vport ? vport_roce_accl_cnts : roce_accl_cnts;
if (MLX5_CAP_GEN(dev->mdev, roce_accl)) {
for (i = 0; i < ARRAY_SIZE(roce_accl_cnts); i++, j++) {
descs[j].name = roce_accl_cnts[i].name;
descs[j].name = names[i].name;
offsets[j] = roce_accl_cnts[i].offset;
}
}
if (is_vport)
return;
if (MLX5_CAP_GEN(dev->mdev, cc_query_allowed)) {
for (i = 0; i < ARRAY_SIZE(cong_cnts); i++, j++) {
descs[j].name = cong_cnts[i].name;
@ -558,9 +659,9 @@ static void mlx5_ib_fill_counters(struct mlx5_ib_dev *dev,
static int __mlx5_ib_alloc_counters(struct mlx5_ib_dev *dev,
struct mlx5_ib_counters *cnts)
struct mlx5_ib_counters *cnts, u32 port_num)
{
u32 num_counters, num_op_counters;
u32 num_counters, num_op_counters = 0;
num_counters = ARRAY_SIZE(basic_q_cnts);
@ -578,6 +679,9 @@ static int __mlx5_ib_alloc_counters(struct mlx5_ib_dev *dev,
cnts->num_q_counters = num_counters;
if (is_mdev_switchdev_mode(dev->mdev) && port_num != MLX5_VPORT_PF)
goto skip_non_qcounters;
if (MLX5_CAP_GEN(dev->mdev, cc_query_allowed)) {
cnts->num_cong_counters = ARRAY_SIZE(cong_cnts);
num_counters += ARRAY_SIZE(cong_cnts);
@ -597,6 +701,7 @@ static int __mlx5_ib_alloc_counters(struct mlx5_ib_dev *dev,
ft_field_support_2_nic_transmit_rdma.bth_opcode))
num_op_counters += ARRAY_SIZE(rdmatx_cnp_op_cnts);
skip_non_qcounters:
cnts->num_op_counters = num_op_counters;
num_counters += num_op_counters;
cnts->descs = kcalloc(num_counters,
@ -623,7 +728,8 @@ static void mlx5_ib_dealloc_counters(struct mlx5_ib_dev *dev)
int num_cnt_ports;
int i, j;
num_cnt_ports = is_mdev_switchdev_mode(dev->mdev) ? 1 : dev->num_ports;
num_cnt_ports = (!is_mdev_switchdev_mode(dev->mdev) ||
vport_qcounters_supported(dev)) ? dev->num_ports : 1;
MLX5_SET(dealloc_q_counter_in, in, opcode,
MLX5_CMD_OP_DEALLOC_Q_COUNTER);
@ -662,15 +768,16 @@ static int mlx5_ib_alloc_counters(struct mlx5_ib_dev *dev)
MLX5_SET(alloc_q_counter_in, in, opcode, MLX5_CMD_OP_ALLOC_Q_COUNTER);
is_shared = MLX5_CAP_GEN(dev->mdev, log_max_uctx) != 0;
num_cnt_ports = is_mdev_switchdev_mode(dev->mdev) ? 1 : dev->num_ports;
num_cnt_ports = (!is_mdev_switchdev_mode(dev->mdev) ||
vport_qcounters_supported(dev)) ? dev->num_ports : 1;
for (i = 0; i < num_cnt_ports; i++) {
err = __mlx5_ib_alloc_counters(dev, &dev->port[i].cnts);
err = __mlx5_ib_alloc_counters(dev, &dev->port[i].cnts, i);
if (err)
goto err_alloc;
mlx5_ib_fill_counters(dev, dev->port[i].cnts.descs,
dev->port[i].cnts.offsets);
dev->port[i].cnts.offsets, i);
MLX5_SET(alloc_q_counter_in, in, uid,
is_shared ? MLX5_SHARED_RESOURCE_UID : 0);
@ -889,6 +996,10 @@ static const struct ib_device_ops hw_stats_ops = {
mlx5_ib_modify_stat : NULL,
};
static const struct ib_device_ops hw_switchdev_vport_op = {
.alloc_hw_port_stats = mlx5_ib_alloc_hw_port_stats,
};
static const struct ib_device_ops hw_switchdev_stats_ops = {
.alloc_hw_device_stats = mlx5_ib_alloc_hw_device_stats,
.get_hw_stats = mlx5_ib_get_hw_stats,
@ -914,9 +1025,11 @@ int mlx5_ib_counters_init(struct mlx5_ib_dev *dev)
if (!MLX5_CAP_GEN(dev->mdev, max_qp_cnt))
return 0;
if (is_mdev_switchdev_mode(dev->mdev))
if (is_mdev_switchdev_mode(dev->mdev)) {
ib_set_device_ops(&dev->ib_dev, &hw_switchdev_stats_ops);
else
if (vport_qcounters_supported(dev))
ib_set_device_ops(&dev->ib_dev, &hw_switchdev_vport_op);
} else
ib_set_device_ops(&dev->ib_dev, &hw_stats_ops);
return mlx5_ib_alloc_counters(dev);
}

33
drivers/infiniband/hw/mlx5/devx.c
View File

@ -666,7 +666,21 @@ static bool devx_is_valid_obj_id(struct uverbs_attr_bundle *attrs,
obj_id;
case MLX5_IB_OBJECT_DEVX_OBJ:
return ((struct devx_obj *)uobj->object)->obj_id == obj_id;
{
u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, in, opcode);
struct devx_obj *devx_uobj = uobj->object;
if (opcode == MLX5_CMD_OP_QUERY_FLOW_COUNTER &&
devx_uobj->flow_counter_bulk_size) {
u64 end;
end = devx_uobj->obj_id +
devx_uobj->flow_counter_bulk_size;
return devx_uobj->obj_id <= obj_id && end > obj_id;
}
return devx_uobj->obj_id == obj_id;
}
default:
return false;
@ -1517,10 +1531,17 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_CREATE)(
goto obj_free;
if (opcode == MLX5_CMD_OP_ALLOC_FLOW_COUNTER) {
u8 bulk = MLX5_GET(alloc_flow_counter_in,
cmd_in,
flow_counter_bulk);
obj->flow_counter_bulk_size = 128UL * bulk;
u32 bulk = MLX5_GET(alloc_flow_counter_in,
cmd_in,
flow_counter_bulk_log_size);
if (bulk)
bulk = 1 << bulk;
else
bulk = 128UL * MLX5_GET(alloc_flow_counter_in,
cmd_in,
flow_counter_bulk);
obj->flow_counter_bulk_size = bulk;
}
uobj->object = obj;
@ -1993,7 +2014,6 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_SUBSCRIBE_EVENT)(
int redirect_fd;
bool use_eventfd = false;
int num_events;
int num_alloc_xa_entries = 0;
u16 obj_type = 0;
u64 cookie = 0;
u32 obj_id = 0;
@ -2075,7 +2095,6 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_SUBSCRIBE_EVENT)(
if (err)
goto err;
num_alloc_xa_entries++;
event_sub = kzalloc(sizeof(*event_sub), GFP_KERNEL);
if (!event_sub) {
err = -ENOMEM;

12
drivers/infiniband/hw/mlx5/mr.c
View File

@ -67,11 +67,14 @@ static void set_mkc_access_pd_addr_fields(void *mkc, int acc, u64 start_addr,
MLX5_SET(mkc, mkc, lw, !!(acc & IB_ACCESS_LOCAL_WRITE));
MLX5_SET(mkc, mkc, lr, 1);
if ((acc & IB_ACCESS_RELAXED_ORDERING) &&
pcie_relaxed_ordering_enabled(dev->mdev->pdev)) {
if (acc & IB_ACCESS_RELAXED_ORDERING) {
if (MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write))
MLX5_SET(mkc, mkc, relaxed_ordering_write, 1);
if (MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read))
if (MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read) ||
(MLX5_CAP_GEN(dev->mdev,
relaxed_ordering_read_pci_enabled) &&
pcie_relaxed_ordering_enabled(dev->mdev->pdev)))
MLX5_SET(mkc, mkc, relaxed_ordering_read, 1);
}
@ -791,7 +794,8 @@ static int get_unchangeable_access_flags(struct mlx5_ib_dev *dev,
ret |= IB_ACCESS_RELAXED_ORDERING;
if ((access_flags & IB_ACCESS_RELAXED_ORDERING) &&
MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read) &&
(MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read) ||
MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read_pci_enabled)) &&
!MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read_umr))
ret |= IB_ACCESS_RELAXED_ORDERING;

19
drivers/infiniband/hw/mlx5/qp.c
View File

@ -60,6 +60,10 @@ enum raw_qp_set_mask_map {
MLX5_RAW_QP_RATE_LIMIT = 1UL << 1,
};
enum {
MLX5_QP_RM_GO_BACK_N = 0x1,
};
struct mlx5_modify_raw_qp_param {
u16 operation;
@ -2519,6 +2523,10 @@ static int create_kernel_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd,
if (qp->flags & IB_QP_CREATE_IPOIB_UD_LSO)
MLX5_SET(qpc, qpc, ulp_stateless_offload_mode, 1);
if (qp->flags & IB_QP_CREATE_INTEGRITY_EN &&
MLX5_CAP_GEN(mdev, go_back_n))
MLX5_SET(qpc, qpc, retry_mode, MLX5_QP_RM_GO_BACK_N);
err = mlx5_qpc_create_qp(dev, &base->mqp, in, inlen, out);
kvfree(in);
if (err)
@ -2846,9 +2854,9 @@ static void process_vendor_flag(struct mlx5_ib_dev *dev, int *flags, int flag,
case MLX5_QP_FLAG_SCATTER_CQE:
case MLX5_QP_FLAG_ALLOW_SCATTER_CQE:
/*
* We don't return error if these flags were provided,
* and mlx5 doesn't have right capability.
*/
* We don't return error if these flags were provided,
* and mlx5 doesn't have right capability.
*/
*flags &= ~(MLX5_QP_FLAG_SCATTER_CQE |
MLX5_QP_FLAG_ALLOW_SCATTER_CQE);
return;
@ -4485,7 +4493,7 @@ static int mlx5_ib_modify_dct(struct ib_qp *ibqp, struct ib_qp_attr *attr,
return -EINVAL;
if (attr->port_num == 0 ||
attr->port_num > MLX5_CAP_GEN(dev->mdev, num_ports)) {
attr->port_num > dev->num_ports) {
mlx5_ib_dbg(dev, "invalid port number %d. number of ports is %d\n",
attr->port_num, dev->num_ports);
return -EINVAL;
@ -5592,8 +5600,7 @@ int mlx5_ib_modify_wq(struct ib_wq *wq, struct ib_wq_attr *wq_attr,
if (wq_attr->flags_mask & IB_WQ_FLAGS_CVLAN_STRIPPING) {
if (!(MLX5_CAP_GEN(dev->mdev, eth_net_offloads) &&
MLX5_CAP_ETH(dev->mdev, vlan_cap))) {
mlx5_ib_dbg(dev, "VLAN offloads are not "
"supported\n");
mlx5_ib_dbg(dev, "VLAN offloads are not supported\n");
err = -EOPNOTSUPP;
goto out;
}

7
drivers/infiniband/hw/mlx5/umr.c
View File

@ -380,6 +380,10 @@ static void mlx5r_umr_set_access_flags(struct mlx5_ib_dev *dev,
struct mlx5_mkey_seg *seg,
unsigned int access_flags)
{
bool ro_read = (access_flags & IB_ACCESS_RELAXED_ORDERING) &&
(MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read) ||
pcie_relaxed_ordering_enabled(dev->mdev->pdev));
MLX5_SET(mkc, seg, a, !!(access_flags & IB_ACCESS_REMOTE_ATOMIC));
MLX5_SET(mkc, seg, rw, !!(access_flags & IB_ACCESS_REMOTE_WRITE));
MLX5_SET(mkc, seg, rr, !!(access_flags & IB_ACCESS_REMOTE_READ));
@ -387,8 +391,7 @@ static void mlx5r_umr_set_access_flags(struct mlx5_ib_dev *dev,
MLX5_SET(mkc, seg, lr, 1);
MLX5_SET(mkc, seg, relaxed_ordering_write,
!!(access_flags & IB_ACCESS_RELAXED_ORDERING));
MLX5_SET(mkc, seg, relaxed_ordering_read,
!!(access_flags & IB_ACCESS_RELAXED_ORDERING));
MLX5_SET(mkc, seg, relaxed_ordering_read, ro_read);
}
int mlx5r_umr_rereg_pd_access(struct mlx5_ib_mr *mr, struct ib_pd *pd,

3
drivers/infiniband/hw/mlx5/umr.h
View File

@ -62,7 +62,8 @@ static inline bool mlx5r_umr_can_reconfig(struct mlx5_ib_dev *dev,
return false;
if ((diffs & IB_ACCESS_RELAXED_ORDERING) &&
MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read) &&
(MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read) ||
MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read_pci_enabled)) &&
!MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read_umr))
return false;

2
drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
View File

@ -1589,7 +1589,6 @@ static void ocrdma_discard_cqes(struct ocrdma_qp *qp, struct ocrdma_cq *cq)
{
unsigned long cq_flags;
unsigned long flags;
int discard_cnt = 0;
u32 cur_getp, stop_getp;
struct ocrdma_cqe *cqe;
u32 qpn = 0, wqe_idx = 0;
@ -1641,7 +1640,6 @@ static void ocrdma_discard_cqes(struct ocrdma_qp *qp, struct ocrdma_cq *cq)
/* mark cqe discarded so that it is not picked up later
* in the poll_cq().
*/
discard_cnt += 1;
cqe->cmn.qpn = 0;
skip_cqe:
cur_getp = (cur_getp + 1) % cq->max_hw_cqe;

9
drivers/infiniband/hw/qib/qib_file_ops.c
View File

@ -484,7 +484,7 @@ static int qib_tid_free(struct qib_ctxtdata *rcd, unsigned subctxt,
const struct qib_tid_info *ti)
{
int ret = 0;
u32 tid, ctxttid, cnt, limit, tidcnt;
u32 tid, ctxttid, limit, tidcnt;
struct qib_devdata *dd = rcd->dd;
u64 __iomem *tidbase;
unsigned long tidmap[8];
@ -520,7 +520,7 @@ static int qib_tid_free(struct qib_ctxtdata *rcd, unsigned subctxt,
/* just in case size changes in future */
limit = tidcnt;
tid = find_first_bit(tidmap, limit);
for (cnt = 0; tid < limit; tid++) {
for (; tid < limit; tid++) {
/*
* small optimization; if we detect a run of 3 or so without
* any set, use find_first_bit again. That's mainly to
@ -530,7 +530,7 @@ static int qib_tid_free(struct qib_ctxtdata *rcd, unsigned subctxt,
*/
if (!test_bit(tid, tidmap))
continue;
cnt++;
if (dd->pageshadow[ctxttid + tid]) {
struct page *p;
dma_addr_t phys;
@ -1768,7 +1768,7 @@ static void unlock_expected_tids(struct qib_ctxtdata *rcd)
{
struct qib_devdata *dd = rcd->dd;
int ctxt_tidbase = rcd->ctxt * dd->rcvtidcnt;
int i, cnt = 0, maxtid = ctxt_tidbase + dd->rcvtidcnt;
int i, maxtid = ctxt_tidbase + dd->rcvtidcnt;
for (i = ctxt_tidbase; i < maxtid; i++) {
struct page *p = dd->pageshadow[i];
@ -1783,7 +1783,6 @@ static void unlock_expected_tids(struct qib_ctxtdata *rcd)
dma_unmap_page(&dd->pcidev->dev, phys, PAGE_SIZE,
DMA_FROM_DEVICE);
qib_release_user_pages(&p, 1);
cnt++;
}
}

8
drivers/infiniband/hw/qib/qib_pcie.c
View File

@ -35,7 +35,6 @@
#include <linux/io.h>
#include <linux/delay.h>
#include <linux/vmalloc.h>
#include <linux/aer.h>
#include <linux/module.h>
#include "qib.h"
@ -105,13 +104,6 @@ int qib_pcie_init(struct pci_dev *pdev, const struct pci_device_id *ent)
}
pci_set_master(pdev);
ret = pci_enable_pcie_error_reporting(pdev);
if (ret) {
qib_early_err(&pdev->dev,
"Unable to enable pcie error reporting: %d\n",
ret);
ret = 0;
}
goto done;
bail:

11
drivers/infiniband/hw/qib/qib_user_sdma.c
View File

@ -320,7 +320,6 @@ static int qib_user_sdma_page_to_frags(const struct qib_devdata *dd,
unpin_user_page(page);
} else {
/* coalesce case */
kunmap(page);
__free_page(page);
}
ret = -ENOMEM;
@ -572,7 +571,7 @@ static int qib_user_sdma_coalesce(const struct qib_devdata *dd,
goto done;
}
mpage = kmap(page);
mpage = page_address(page);
mpage_save = mpage;
for (i = 0; i < niov; i++) {
int cfur;
@ -581,7 +580,7 @@ static int qib_user_sdma_coalesce(const struct qib_devdata *dd,
iov[i].iov_base, iov[i].iov_len);
if (cfur) {
ret = -EFAULT;
goto free_unmap;
goto page_free;
}
mpage += iov[i].iov_len;
@ -592,8 +591,7 @@ static int qib_user_sdma_coalesce(const struct qib_devdata *dd,
page, 0, 0, len, mpage_save);
goto done;
free_unmap:
kunmap(page);
page_free:
__free_page(page);
done:
return ret;
@ -627,9 +625,6 @@ static void qib_user_sdma_free_pkt_frag(struct device *dev,
pkt->addr[i].dma_length,
DMA_TO_DEVICE);
if (pkt->addr[i].kvaddr)
kunmap(pkt->addr[i].page);
if (pkt->addr[i].put_page)
unpin_user_page(pkt->addr[i].page);
else

2
drivers/infiniband/hw/usnic/usnic_ib_main.c
View File

@ -602,7 +602,6 @@ out_clean_vnic:
usnic_vnic_free(vf->vnic);
out_release_regions:
pci_set_drvdata(pdev, NULL);
pci_clear_master(pdev);
pci_release_regions(pdev);
out_disable_device:
pci_disable_device(pdev);
@ -623,7 +622,6 @@ static void usnic_ib_pci_remove(struct pci_dev *pdev)
kref_put(&pf->vf_cnt, usnic_ib_undiscover_pf);
usnic_vnic_free(vf->vnic);
pci_set_drvdata(pdev, NULL);
pci_clear_master(pdev);
pci_release_regions(pdev);
pci_disable_device(pdev);
kfree(vf);

4
drivers/infiniband/sw/rdmavt/qp.c
View File

@ -464,8 +464,6 @@ void rvt_qp_exit(struct rvt_dev_info *rdi)
if (qps_inuse)
rvt_pr_err(rdi, "QP memory leak! %u still in use\n",
qps_inuse);
if (!rdi->qp_dev)
return;
kfree(rdi->qp_dev->qp_table);
free_qpn_table(&rdi->qp_dev->qpn_table);
@ -2040,7 +2038,7 @@ static int rvt_post_one_wr(struct rvt_qp *qp,
wqe = rvt_get_swqe_ptr(qp, qp->s_head);
/* cplen has length from above */
memcpy(&wqe->wr, wr, cplen);
memcpy(&wqe->ud_wr, wr, cplen);
wqe->length = 0;
j = 0;

16
drivers/infiniband/sw/rxe/rxe.c
View File

@ -160,6 +160,8 @@ void rxe_set_mtu(struct rxe_dev *rxe, unsigned int ndev_mtu)
port->attr.active_mtu = mtu;
port->mtu_cap = ib_mtu_enum_to_int(mtu);
rxe_info_dev(rxe, "Set mtu to %d", port->mtu_cap);
}
/* called by ifc layer to create new rxe device.
@ -175,26 +177,26 @@ int rxe_add(struct rxe_dev *rxe, unsigned int mtu, const char *ibdev_name)
static int rxe_newlink(const char *ibdev_name, struct net_device *ndev)
{
struct rxe_dev *exists;
struct rxe_dev *rxe;
int err = 0;
if (is_vlan_dev(ndev)) {
pr_err("rxe creation allowed on top of a real device only\n");
rxe_err("rxe creation allowed on top of a real device only");
err = -EPERM;
goto err;
}
exists = rxe_get_dev_from_net(ndev);
if (exists) {
ib_device_put(&exists->ib_dev);
rxe_dbg(exists, "already configured on %s\n", ndev->name);
rxe = rxe_get_dev_from_net(ndev);
if (rxe) {
ib_device_put(&rxe->ib_dev);
rxe_err_dev(rxe, "already configured on %s", ndev->name);
err = -EEXIST;
goto err;
}
err = rxe_net_add(ibdev_name, ndev);
if (err) {
rxe_dbg(exists, "failed to add %s\n", ndev->name);
rxe_err("failed to add %s\n", ndev->name);
goto err;
}
err:

46
drivers/infiniband/sw/rxe/rxe.h
View File

@ -38,7 +38,8 @@
#define RXE_ROCE_V2_SPORT (0xc000)
#define rxe_dbg(rxe, fmt, ...) ibdev_dbg(&(rxe)->ib_dev, \
#define rxe_dbg(fmt, ...) pr_debug("%s: " fmt "\n", __func__, ##__VA_ARGS__)
#define rxe_dbg_dev(rxe, fmt, ...) ibdev_dbg(&(rxe)->ib_dev, \
"%s: " fmt, __func__, ##__VA_ARGS__)
#define rxe_dbg_uc(uc, fmt, ...) ibdev_dbg((uc)->ibuc.device, \
"uc#%d %s: " fmt, (uc)->elem.index, __func__, ##__VA_ARGS__)
@ -57,6 +58,48 @@
#define rxe_dbg_mw(mw, fmt, ...) ibdev_dbg((mw)->ibmw.device, \
"mw#%d %s: " fmt, (mw)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_err(fmt, ...) pr_err_ratelimited("%s: " fmt "\n", __func__, \
##__VA_ARGS__)
#define rxe_err_dev(rxe, fmt, ...) ibdev_err_ratelimited(&(rxe)->ib_dev, \
"%s: " fmt, __func__, ##__VA_ARGS__)
#define rxe_err_uc(uc, fmt, ...) ibdev_err_ratelimited((uc)->ibuc.device, \
"uc#%d %s: " fmt, (uc)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_err_pd(pd, fmt, ...) ibdev_err_ratelimited((pd)->ibpd.device, \
"pd#%d %s: " fmt, (pd)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_err_ah(ah, fmt, ...) ibdev_err_ratelimited((ah)->ibah.device, \
"ah#%d %s: " fmt, (ah)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_err_srq(srq, fmt, ...) ibdev_err_ratelimited((srq)->ibsrq.device, \
"srq#%d %s: " fmt, (srq)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_err_qp(qp, fmt, ...) ibdev_err_ratelimited((qp)->ibqp.device, \
"qp#%d %s: " fmt, (qp)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_err_cq(cq, fmt, ...) ibdev_err_ratelimited((cq)->ibcq.device, \
"cq#%d %s: " fmt, (cq)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_err_mr(mr, fmt, ...) ibdev_err_ratelimited((mr)->ibmr.device, \
"mr#%d %s: " fmt, (mr)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_err_mw(mw, fmt, ...) ibdev_err_ratelimited((mw)->ibmw.device, \
"mw#%d %s: " fmt, (mw)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_info(fmt, ...) pr_info_ratelimited("%s: " fmt "\n", __func__, \
##__VA_ARGS__)
#define rxe_info_dev(rxe, fmt, ...) ibdev_info_ratelimited(&(rxe)->ib_dev, \
"%s: " fmt, __func__, ##__VA_ARGS__)
#define rxe_info_uc(uc, fmt, ...) ibdev_info_ratelimited((uc)->ibuc.device, \
"uc#%d %s: " fmt, (uc)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_info_pd(pd, fmt, ...) ibdev_info_ratelimited((pd)->ibpd.device, \
"pd#%d %s: " fmt, (pd)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_info_ah(ah, fmt, ...) ibdev_info_ratelimited((ah)->ibah.device, \
"ah#%d %s: " fmt, (ah)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_info_srq(srq, fmt, ...) ibdev_info_ratelimited((srq)->ibsrq.device, \
"srq#%d %s: " fmt, (srq)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_info_qp(qp, fmt, ...) ibdev_info_ratelimited((qp)->ibqp.device, \
"qp#%d %s: " fmt, (qp)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_info_cq(cq, fmt, ...) ibdev_info_ratelimited((cq)->ibcq.device, \
"cq#%d %s: " fmt, (cq)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_info_mr(mr, fmt, ...) ibdev_info_ratelimited((mr)->ibmr.device, \
"mr#%d %s: " fmt, (mr)->elem.index, __func__, ##__VA_ARGS__)
#define rxe_info_mw(mw, fmt, ...) ibdev_info_ratelimited((mw)->ibmw.device, \
"mw#%d %s: " fmt, (mw)->elem.index, __func__, ##__VA_ARGS__)
/* responder states */
enum resp_states {
RESPST_NONE,
@ -90,7 +133,6 @@ enum resp_states {
RESPST_ERR_LENGTH,
RESPST_ERR_CQ_OVERFLOW,
RESPST_ERROR,
RESPST_RESET,
RESPST_DONE,
RESPST_EXIT,
};

163
drivers/infiniband/sw/rxe/rxe_comp.c
View File

@ -118,10 +118,12 @@ void retransmit_timer(struct timer_list *t)
rxe_dbg_qp(qp, "retransmit timer fired\n");
spin_lock_bh(&qp->state_lock);
if (qp->valid) {
qp->comp.timeout = 1;
rxe_sched_task(&qp->comp.task);
}
spin_unlock_bh(&qp->state_lock);
}
void rxe_comp_queue_pkt(struct rxe_qp *qp, struct sk_buff *skb)
@ -322,7 +324,7 @@ static inline enum comp_state check_ack(struct rxe_qp *qp,
qp->comp.psn = pkt->psn;
if (qp->req.wait_psn) {
qp->req.wait_psn = 0;
rxe_run_task(&qp->req.task);
rxe_sched_task(&qp->req.task);
}
}
return COMPST_ERROR_RETRY;
@ -428,6 +430,10 @@ static void make_send_cqe(struct rxe_qp *qp, struct rxe_send_wqe *wqe,
uwc->wc_flags = IB_WC_WITH_IMM;
uwc->byte_len = wqe->dma.length;
}
} else {
if (wqe->status != IB_WC_WR_FLUSH_ERR)
rxe_err_qp(qp, "non-flush error status = %d",
wqe->status);
}
}
@ -469,10 +475,33 @@ static void do_complete(struct rxe_qp *qp, struct rxe_send_wqe *wqe)
*/
if (qp->req.wait_fence) {
qp->req.wait_fence = 0;
rxe_run_task(&qp->req.task);
rxe_sched_task(&qp->req.task);
}
}
static void comp_check_sq_drain_done(struct rxe_qp *qp)
{
spin_lock_bh(&qp->state_lock);
if (unlikely(qp_state(qp) == IB_QPS_SQD)) {
if (qp->attr.sq_draining && qp->comp.psn == qp->req.psn) {
qp->attr.sq_draining = 0;
spin_unlock_bh(&qp->state_lock);
if (qp->ibqp.event_handler) {
struct ib_event ev;
ev.device = qp->ibqp.device;
ev.element.qp = &qp->ibqp;
ev.event = IB_EVENT_SQ_DRAINED;
qp->ibqp.event_handler(&ev,
qp->ibqp.qp_context);
}
return;
}
}
spin_unlock_bh(&qp->state_lock);
}
static inline enum comp_state complete_ack(struct rxe_qp *qp,
struct rxe_pkt_info *pkt,
struct rxe_send_wqe *wqe)
@ -483,31 +512,11 @@ static inline enum comp_state complete_ack(struct rxe_qp *qp,
if (qp->req.need_rd_atomic) {
qp->comp.timeout_retry = 0;
qp->req.need_rd_atomic = 0;
rxe_run_task(&qp->req.task);
rxe_sched_task(&qp->req.task);
}
}
if (unlikely(qp->req.state == QP_STATE_DRAIN)) {
/* state_lock used by requester & completer */
spin_lock_bh(&qp->state_lock);
if ((qp->req.state == QP_STATE_DRAIN) &&
(qp->comp.psn == qp->req.psn)) {
qp->req.state = QP_STATE_DRAINED;
spin_unlock_bh(&qp->state_lock);
if (qp->ibqp.event_handler) {
struct ib_event ev;
ev.device = qp->ibqp.device;
ev.element.qp = &qp->ibqp;
ev.event = IB_EVENT_SQ_DRAINED;
qp->ibqp.event_handler(&ev,
qp->ibqp.qp_context);
}
} else {
spin_unlock_bh(&qp->state_lock);
}
}
comp_check_sq_drain_done(qp);
do_complete(qp, wqe);
@ -538,25 +547,60 @@ static inline enum comp_state complete_wqe(struct rxe_qp *qp,
return COMPST_GET_WQE;
}
static void rxe_drain_resp_pkts(struct rxe_qp *qp, bool notify)
/* drain incoming response packet queue */
static void drain_resp_pkts(struct rxe_qp *qp)
{
struct sk_buff *skb;
struct rxe_send_wqe *wqe;
struct rxe_queue *q = qp->sq.queue;
while ((skb = skb_dequeue(&qp->resp_pkts))) {
rxe_put(qp);
kfree_skb(skb);
ib_device_put(qp->ibqp.device);
}
}
/* complete send wqe with flush error */
static int flush_send_wqe(struct rxe_qp *qp, struct rxe_send_wqe *wqe)
{
struct rxe_cqe cqe = {};
struct ib_wc *wc = &cqe.ibwc;
struct ib_uverbs_wc *uwc = &cqe.uibwc;
int err;
if (qp->is_user) {
uwc->wr_id = wqe->wr.wr_id;
uwc->status = IB_WC_WR_FLUSH_ERR;
uwc->qp_num = qp->ibqp.qp_num;
} else {
wc->wr_id = wqe->wr.wr_id;
wc->status = IB_WC_WR_FLUSH_ERR;
wc->qp = &qp->ibqp;
}
err = rxe_cq_post(qp->scq, &cqe, 0);
if (err)
rxe_dbg_cq(qp->scq, "post cq failed, err = %d", err);
return err;
}
/* drain and optionally complete the send queue
* if unable to complete a wqe, i.e. cq is full, stop
* completing and flush the remaining wqes
*/
static void flush_send_queue(struct rxe_qp *qp, bool notify)
{
struct rxe_send_wqe *wqe;
struct rxe_queue *q = qp->sq.queue;
int err;
while ((wqe = queue_head(q, q->type))) {
if (notify) {
wqe->status = IB_WC_WR_FLUSH_ERR;
do_complete(qp, wqe);
} else {
queue_advance_consumer(q, q->type);
err = flush_send_wqe(qp, wqe);
if (err)
notify = 0;
}
queue_advance_consumer(q, q->type);
}
}
@ -571,9 +615,28 @@ static void free_pkt(struct rxe_pkt_info *pkt)
ib_device_put(dev);
}
int rxe_completer(void *arg)
/* reset the retry timer if
* - QP is type RC
* - there is a packet sent by the requester that
* might be acked (we still might get spurious
* timeouts but try to keep them as few as possible)
* - the timeout parameter is set
* - the QP is alive
*/
static void reset_retry_timer(struct rxe_qp *qp)
{
if (qp_type(qp) == IB_QPT_RC && qp->qp_timeout_jiffies) {
spin_lock_bh(&qp->state_lock);
if (qp_state(qp) >= IB_QPS_RTS &&
psn_compare(qp->req.psn, qp->comp.psn) > 0)
mod_timer(&qp->retrans_timer,
jiffies + qp->qp_timeout_jiffies);
spin_unlock_bh(&qp->state_lock);
}
}
int rxe_completer(struct rxe_qp *qp)
{
struct rxe_qp *qp = (struct rxe_qp *)arg;
struct rxe_dev *rxe = to_rdev(qp->ibqp.device);
struct rxe_send_wqe *wqe = NULL;
struct sk_buff *skb = NULL;
@ -581,15 +644,17 @@ int rxe_completer(void *arg)
enum comp_state state;
int ret;
if (!rxe_get(qp))
return -EAGAIN;
spin_lock_bh(&qp->state_lock);
if (!qp->valid || qp_state(qp) == IB_QPS_ERR ||
qp_state(qp) == IB_QPS_RESET) {
bool notify = qp->valid && (qp_state(qp) == IB_QPS_ERR);
if (!qp->valid || qp->comp.state == QP_STATE_ERROR ||
qp->comp.state == QP_STATE_RESET) {
rxe_drain_resp_pkts(qp, qp->valid &&
qp->comp.state == QP_STATE_ERROR);
drain_resp_pkts(qp);
flush_send_queue(qp, notify);
spin_unlock_bh(&qp->state_lock);
goto exit;
}
spin_unlock_bh(&qp->state_lock);
if (qp->comp.timeout) {
qp->comp.timeout_retry = 1;
@ -677,20 +742,7 @@ int rxe_completer(void *arg)
break;
}
/* re reset the timeout counter if
* (1) QP is type RC
* (2) the QP is alive
* (3) there is a packet sent by the requester that
* might be acked (we still might get spurious
* timeouts but try to keep them as few as possible)
* (4) the timeout parameter is set
*/
if ((qp_type(qp) == IB_QPT_RC) &&
(qp->req.state == QP_STATE_READY) &&
(psn_compare(qp->req.psn, qp->comp.psn) > 0) &&
qp->qp_timeout_jiffies)
mod_timer(&qp->retrans_timer,
jiffies + qp->qp_timeout_jiffies);
reset_retry_timer(qp);
goto exit;
case COMPST_ERROR_RETRY:
@ -730,7 +782,7 @@ int rxe_completer(void *arg)
RXE_CNT_COMP_RETRY);
qp->req.need_retry = 1;
qp->comp.started_retry = 1;
rxe_run_task(&qp->req.task);
rxe_sched_task(&qp->req.task);
}
goto done;
@ -752,6 +804,7 @@ int rxe_completer(void *arg)
*/
qp->req.wait_for_rnr_timer = 1;
rxe_dbg_qp(qp, "set rnr nak timer\n");
// TODO who protects from destroy_qp??
mod_timer(&qp->rnr_nak_timer,
jiffies + rnrnak_jiffies(aeth_syn(pkt)
& ~AETH_TYPE_MASK));
@ -784,7 +837,5 @@ exit:
out:
if (pkt)
free_pkt(pkt);
rxe_put(qp);
return ret;
}

39
drivers/infiniband/sw/rxe/rxe_cq.c
View File

@ -14,12 +14,12 @@ int rxe_cq_chk_attr(struct rxe_dev *rxe, struct rxe_cq *cq,
int count;
if (cqe <= 0) {
rxe_dbg(rxe, "cqe(%d) <= 0\n", cqe);
rxe_dbg_dev(rxe, "cqe(%d) <= 0\n", cqe);
goto err1;
}
if (cqe > rxe->attr.max_cqe) {
rxe_dbg(rxe, "cqe(%d) > max_cqe(%d)\n",
rxe_dbg_dev(rxe, "cqe(%d) > max_cqe(%d)\n",
cqe, rxe->attr.max_cqe);
goto err1;
}
@ -39,21 +39,6 @@ err1:
return -EINVAL;
}
static void rxe_send_complete(struct tasklet_struct *t)
{
struct rxe_cq *cq = from_tasklet(cq, t, comp_task);
unsigned long flags;
spin_lock_irqsave(&cq->cq_lock, flags);
if (cq->is_dying) {
spin_unlock_irqrestore(&cq->cq_lock, flags);
return;
}
spin_unlock_irqrestore(&cq->cq_lock, flags);
cq->ibcq.comp_handler(&cq->ibcq, cq->ibcq.cq_context);
}
int rxe_cq_from_init(struct rxe_dev *rxe, struct rxe_cq *cq, int cqe,
int comp_vector, struct ib_udata *udata,
struct rxe_create_cq_resp __user *uresp)
@ -65,7 +50,7 @@ int rxe_cq_from_init(struct rxe_dev *rxe, struct rxe_cq *cq, int cqe,
cq->queue = rxe_queue_init(rxe, &cqe,
sizeof(struct rxe_cqe), type);
if (!cq->queue) {
rxe_dbg(rxe, "unable to create cq\n");
rxe_dbg_dev(rxe, "unable to create cq\n");
return -ENOMEM;
}
@ -79,10 +64,6 @@ int rxe_cq_from_init(struct rxe_dev *rxe, struct rxe_cq *cq, int cqe,
cq->is_user = uresp;
cq->is_dying = false;
tasklet_setup(&cq->comp_task, rxe_send_complete);
spin_lock_init(&cq->cq_lock);
cq->ibcq.cqe = cqe;
return 0;
@ -103,6 +84,7 @@ int rxe_cq_resize_queue(struct rxe_cq *cq, int cqe,
return err;
}
/* caller holds reference to cq */
int rxe_cq_post(struct rxe_cq *cq, struct rxe_cqe *cqe, int solicited)
{
struct ib_event ev;
@ -114,6 +96,7 @@ int rxe_cq_post(struct rxe_cq *cq, struct rxe_cqe *cqe, int solicited)
full = queue_full(cq->queue, QUEUE_TYPE_TO_CLIENT);
if (unlikely(full)) {
rxe_err_cq(cq, "queue full");
spin_unlock_irqrestore(&cq->cq_lock, flags);
if (cq->ibcq.event_handler) {
ev.device = cq->ibcq.device;
@ -135,21 +118,13 @@ int rxe_cq_post(struct rxe_cq *cq, struct rxe_cqe *cqe, int solicited)
if ((cq->notify == IB_CQ_NEXT_COMP) ||
(cq->notify == IB_CQ_SOLICITED && solicited)) {
cq->notify = 0;
tasklet_schedule(&cq->comp_task);
cq->ibcq.comp_handler(&cq->ibcq, cq->ibcq.cq_context);
}
return 0;
}
void rxe_cq_disable(struct rxe_cq *cq)
{
unsigned long flags;
spin_lock_irqsave(&cq->cq_lock, flags);
cq->is_dying = true;
spin_unlock_irqrestore(&cq->cq_lock, flags);
}
void rxe_cq_cleanup(struct rxe_pool_elem *elem)
{
struct rxe_cq *cq = container_of(elem, typeof(*cq), elem);

4
drivers/infiniband/sw/rxe/rxe_icrc.c
View File

@ -21,7 +21,7 @@ int rxe_icrc_init(struct rxe_dev *rxe)
tfm = crypto_alloc_shash("crc32", 0, 0);
if (IS_ERR(tfm)) {
rxe_dbg(rxe, "failed to init crc32 algorithm err: %ld\n",
rxe_dbg_dev(rxe, "failed to init crc32 algorithm err: %ld\n",
PTR_ERR(tfm));
return PTR_ERR(tfm);
}
@ -51,7 +51,7 @@ static __be32 rxe_crc32(struct rxe_dev *rxe, __be32 crc, void *next, size_t len)
*(__be32 *)shash_desc_ctx(shash) = crc;
err = crypto_shash_update(shash, next, len);
if (unlikely(err)) {
rxe_dbg(rxe, "failed crc calculation, err: %d\n", err);
rxe_dbg_dev(rxe, "failed crc calculation, err: %d\n", err);
return (__force __be32)crc32_le((__force u32)crc, next, len);
}

7
drivers/infiniband/sw/rxe/rxe_loc.h
View File

@ -80,7 +80,6 @@ int mr_check_range(struct rxe_mr *mr, u64 iova, size_t length);
int advance_dma_data(struct rxe_dma_info *dma, unsigned int length);
int rxe_invalidate_mr(struct rxe_qp *qp, u32 key);
int rxe_reg_fast_mr(struct rxe_qp *qp, struct rxe_send_wqe *wqe);
int rxe_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata);
void rxe_mr_cleanup(struct rxe_pool_elem *elem);
/* rxe_mw.c */
@ -171,9 +170,9 @@ void rxe_srq_cleanup(struct rxe_pool_elem *elem);
void rxe_dealloc(struct ib_device *ib_dev);
int rxe_completer(void *arg);
int rxe_requester(void *arg);
int rxe_responder(void *arg);
int rxe_completer(struct rxe_qp *qp);
int rxe_requester(struct rxe_qp *qp);
int rxe_responder(struct rxe_qp *qp);
/* rxe_icrc.c */
int rxe_icrc_init(struct rxe_dev *rxe);

6
drivers/infiniband/sw/rxe/rxe_mmap.c
View File

@ -79,7 +79,7 @@ int rxe_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
/* Don't allow a mmap larger than the object. */
if (size > ip->info.size) {
rxe_dbg(rxe, "mmap region is larger than the object!\n");
rxe_dbg_dev(rxe, "mmap region is larger than the object!\n");
spin_unlock_bh(&rxe->pending_lock);
ret = -EINVAL;
goto done;
@ -87,7 +87,7 @@ int rxe_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
goto found_it;
}
rxe_dbg(rxe, "unable to find pending mmap info\n");
rxe_dbg_dev(rxe, "unable to find pending mmap info\n");
spin_unlock_bh(&rxe->pending_lock);
ret = -EINVAL;
goto done;
@ -98,7 +98,7 @@ found_it:
ret = remap_vmalloc_range(vma, ip->obj, 0);
if (ret) {
rxe_dbg(rxe, "err %d from remap_vmalloc_range\n", ret);
rxe_dbg_dev(rxe, "err %d from remap_vmalloc_range\n", ret);
goto done;
}

29
drivers/infiniband/sw/rxe/rxe_mr.c
View File

@ -210,10 +210,10 @@ err1:
return err;
}
static int rxe_set_page(struct ib_mr *ibmr, u64 iova)
static int rxe_set_page(struct ib_mr *ibmr, u64 dma_addr)
{
struct rxe_mr *mr = to_rmr(ibmr);
struct page *page = virt_to_page(iova & mr->page_mask);
struct page *page = ib_virt_dma_to_page(dma_addr);
bool persistent = !!(mr->access & IB_ACCESS_FLUSH_PERSISTENT);
int err;
@ -279,16 +279,16 @@ static int rxe_mr_copy_xarray(struct rxe_mr *mr, u64 iova, void *addr,
return 0;
}
static void rxe_mr_copy_dma(struct rxe_mr *mr, u64 iova, void *addr,
static void rxe_mr_copy_dma(struct rxe_mr *mr, u64 dma_addr, void *addr,
unsigned int length, enum rxe_mr_copy_dir dir)
{
unsigned int page_offset = iova & (PAGE_SIZE - 1);
unsigned int page_offset = dma_addr & (PAGE_SIZE - 1);
unsigned int bytes;
struct page *page;
u8 *va;
while (length) {
page = virt_to_page(iova & mr->page_mask);
page = ib_virt_dma_to_page(dma_addr);
bytes = min_t(unsigned int, length,
PAGE_SIZE - page_offset);
va = kmap_local_page(page);
@ -300,7 +300,7 @@ static void rxe_mr_copy_dma(struct rxe_mr *mr, u64 iova, void *addr,
kunmap_local(va);
page_offset = 0;
iova += bytes;
dma_addr += bytes;
addr += bytes;
length -= bytes;
}
@ -488,7 +488,7 @@ int rxe_mr_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
if (mr->ibmr.type == IB_MR_TYPE_DMA) {
page_offset = iova & (PAGE_SIZE - 1);
page = virt_to_page(iova & PAGE_MASK);
page = ib_virt_dma_to_page(iova);
} else {
unsigned long index;
int err;
@ -545,7 +545,7 @@ int rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
if (mr->ibmr.type == IB_MR_TYPE_DMA) {
page_offset = iova & (PAGE_SIZE - 1);
page = virt_to_page(iova & PAGE_MASK);
page = ib_virt_dma_to_page(iova);
} else {
unsigned long index;
int err;
@ -722,19 +722,6 @@ int rxe_reg_fast_mr(struct rxe_qp *qp, struct rxe_send_wqe *wqe)
return 0;
}
int rxe_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata)
{
struct rxe_mr *mr = to_rmr(ibmr);
/* See IBA 10.6.7.2.6 */
if (atomic_read(&mr->num_mw) > 0)
return -EINVAL;
rxe_cleanup(mr);
kfree_rcu(mr);
return 0;
}
void rxe_mr_cleanup(struct rxe_pool_elem *elem)
{
struct rxe_mr *mr = container_of(elem, typeof(*mr), elem);

11
drivers/infiniband/sw/rxe/rxe_net.c
View File

@ -413,11 +413,14 @@ int rxe_xmit_packet(struct rxe_qp *qp, struct rxe_pkt_info *pkt,
int is_request = pkt->mask & RXE_REQ_MASK;
struct rxe_dev *rxe = to_rdev(qp->ibqp.device);
if ((is_request && (qp->req.state != QP_STATE_READY)) ||
(!is_request && (qp->resp.state != QP_STATE_READY))) {
spin_lock_bh(&qp->state_lock);
if ((is_request && (qp_state(qp) < IB_QPS_RTS)) ||
(!is_request && (qp_state(qp) < IB_QPS_RTR))) {
spin_unlock_bh(&qp->state_lock);
rxe_dbg_qp(qp, "Packet dropped. QP is not in ready state\n");
goto drop;
}
spin_unlock_bh(&qp->state_lock);
rxe_icrc_generate(skb, pkt);
@ -596,7 +599,7 @@ static int rxe_notify(struct notifier_block *not_blk,
rxe_port_down(rxe);
break;
case NETDEV_CHANGEMTU:
rxe_dbg(rxe, "%s changed mtu to %d\n", ndev->name, ndev->mtu);
rxe_dbg_dev(rxe, "%s changed mtu to %d\n", ndev->name, ndev->mtu);
rxe_set_mtu(rxe, ndev->mtu);
break;
case NETDEV_CHANGE:
@ -608,7 +611,7 @@ static int rxe_notify(struct notifier_block *not_blk,
case NETDEV_CHANGENAME:
case NETDEV_FEAT_CHANGE:
default:
rxe_dbg(rxe, "ignoring netdev event = %ld for %s\n",
rxe_dbg_dev(rxe, "ignoring netdev event = %ld for %s\n",
event, ndev->name);
break;
}

267
drivers/infiniband/sw/rxe/rxe_qp.c
View File

@ -19,33 +19,33 @@ static int rxe_qp_chk_cap(struct rxe_dev *rxe, struct ib_qp_cap *cap,
int has_srq)
{
if (cap->max_send_wr > rxe->attr.max_qp_wr) {
rxe_dbg(rxe, "invalid send wr = %u > %d\n",
rxe_dbg_dev(rxe, "invalid send wr = %u > %d\n",
cap->max_send_wr, rxe->attr.max_qp_wr);
goto err1;
}
if (cap->max_send_sge > rxe->attr.max_send_sge) {
rxe_dbg(rxe, "invalid send sge = %u > %d\n",
rxe_dbg_dev(rxe, "invalid send sge = %u > %d\n",
cap->max_send_sge, rxe->attr.max_send_sge);
goto err1;
}
if (!has_srq) {
if (cap->max_recv_wr > rxe->attr.max_qp_wr) {
rxe_dbg(rxe, "invalid recv wr = %u > %d\n",
rxe_dbg_dev(rxe, "invalid recv wr = %u > %d\n",
cap->max_recv_wr, rxe->attr.max_qp_wr);
goto err1;
}
if (cap->max_recv_sge > rxe->attr.max_recv_sge) {
rxe_dbg(rxe, "invalid recv sge = %u > %d\n",
rxe_dbg_dev(rxe, "invalid recv sge = %u > %d\n",
cap->max_recv_sge, rxe->attr.max_recv_sge);
goto err1;
}
}
if (cap->max_inline_data > rxe->max_inline_data) {
rxe_dbg(rxe, "invalid max inline data = %u > %d\n",
rxe_dbg_dev(rxe, "invalid max inline data = %u > %d\n",
cap->max_inline_data, rxe->max_inline_data);
goto err1;
}
@ -73,7 +73,7 @@ int rxe_qp_chk_init(struct rxe_dev *rxe, struct ib_qp_init_attr *init)
}
if (!init->recv_cq || !init->send_cq) {
rxe_dbg(rxe, "missing cq\n");
rxe_dbg_dev(rxe, "missing cq\n");
goto err1;
}
@ -82,14 +82,14 @@ int rxe_qp_chk_init(struct rxe_dev *rxe, struct ib_qp_init_attr *init)
if (init->qp_type == IB_QPT_GSI) {
if (!rdma_is_port_valid(&rxe->ib_dev, port_num)) {
rxe_dbg(rxe, "invalid port = %d\n", port_num);
rxe_dbg_dev(rxe, "invalid port = %d\n", port_num);
goto err1;
}
port = &rxe->port;
if (init->qp_type == IB_QPT_GSI && port->qp_gsi_index) {
rxe_dbg(rxe, "GSI QP exists for port %d\n", port_num);
rxe_dbg_dev(rxe, "GSI QP exists for port %d\n", port_num);
goto err1;
}
}
@ -231,8 +231,6 @@ static int rxe_qp_init_req(struct rxe_dev *rxe, struct rxe_qp *qp,
qp->req.wqe_index = queue_get_producer(qp->sq.queue,
QUEUE_TYPE_FROM_CLIENT);
qp->req.state = QP_STATE_RESET;
qp->comp.state = QP_STATE_RESET;
qp->req.opcode = -1;
qp->comp.opcode = -1;
@ -287,7 +285,6 @@ static int rxe_qp_init_resp(struct rxe_dev *rxe, struct rxe_qp *qp,
qp->resp.opcode = OPCODE_NONE;
qp->resp.msn = 0;
qp->resp.state = QP_STATE_RESET;
return 0;
}
@ -328,8 +325,10 @@ int rxe_qp_from_init(struct rxe_dev *rxe, struct rxe_qp *qp, struct rxe_pd *pd,
if (err)
goto err2;
spin_lock_bh(&qp->state_lock);
qp->attr.qp_state = IB_QPS_RESET;
qp->valid = 1;
spin_unlock_bh(&qp->state_lock);
return 0;
@ -380,30 +379,9 @@ int rxe_qp_to_init(struct rxe_qp *qp, struct ib_qp_init_attr *init)
return 0;
}
/* called by the modify qp verb, this routine checks all the parameters before
* making any changes
*/
int rxe_qp_chk_attr(struct rxe_dev *rxe, struct rxe_qp *qp,
struct ib_qp_attr *attr, int mask)
{
enum ib_qp_state cur_state = (mask & IB_QP_CUR_STATE) ?
attr->cur_qp_state : qp->attr.qp_state;
enum ib_qp_state new_state = (mask & IB_QP_STATE) ?
attr->qp_state : cur_state;
if (!ib_modify_qp_is_ok(cur_state, new_state, qp_type(qp), mask)) {
rxe_dbg_qp(qp, "invalid mask or state\n");
goto err1;
}
if (mask & IB_QP_STATE) {
if (cur_state == IB_QPS_SQD) {
if (qp->req.state == QP_STATE_DRAIN &&
new_state != IB_QPS_ERR)
goto err1;
}
}
if (mask & IB_QP_PORT) {
if (!rdma_is_port_valid(&rxe->ib_dev, attr->port_num)) {
rxe_dbg_qp(qp, "invalid port %d\n", attr->port_num);
@ -473,29 +451,18 @@ static void rxe_qp_reset(struct rxe_qp *qp)
{
/* stop tasks from running */
rxe_disable_task(&qp->resp.task);
rxe_disable_task(&qp->comp.task);
rxe_disable_task(&qp->req.task);
/* stop request/comp */
if (qp->sq.queue) {
if (qp_type(qp) == IB_QPT_RC)
rxe_disable_task(&qp->comp.task);
rxe_disable_task(&qp->req.task);
}
/* drain work and packet queuesc */
rxe_requester(qp);
rxe_completer(qp);
rxe_responder(qp);
/* move qp to the reset state */
qp->req.state = QP_STATE_RESET;
qp->comp.state = QP_STATE_RESET;
qp->resp.state = QP_STATE_RESET;
/* let state machines reset themselves drain work and packet queues
* etc.
*/
__rxe_do_task(&qp->resp.task);
if (qp->sq.queue) {
__rxe_do_task(&qp->comp.task);
__rxe_do_task(&qp->req.task);
if (qp->rq.queue)
rxe_queue_reset(qp->rq.queue);
if (qp->sq.queue)
rxe_queue_reset(qp->sq.queue);
}
/* cleanup attributes */
atomic_set(&qp->ssn, 0);
@ -518,54 +485,103 @@ static void rxe_qp_reset(struct rxe_qp *qp)
/* reenable tasks */
rxe_enable_task(&qp->resp.task);
if (qp->sq.queue) {
if (qp_type(qp) == IB_QPT_RC)
rxe_enable_task(&qp->comp.task);
rxe_enable_task(&qp->req.task);
}
}
/* drain the send queue */
static void rxe_qp_drain(struct rxe_qp *qp)
{
if (qp->sq.queue) {
if (qp->req.state != QP_STATE_DRAINED) {
qp->req.state = QP_STATE_DRAIN;
if (qp_type(qp) == IB_QPT_RC)
rxe_sched_task(&qp->comp.task);
else
__rxe_do_task(&qp->comp.task);
rxe_sched_task(&qp->req.task);
}
}
rxe_enable_task(&qp->comp.task);
rxe_enable_task(&qp->req.task);
}
/* move the qp to the error state */
void rxe_qp_error(struct rxe_qp *qp)
{
qp->req.state = QP_STATE_ERROR;
qp->resp.state = QP_STATE_ERROR;
qp->comp.state = QP_STATE_ERROR;
spin_lock_bh(&qp->state_lock);
qp->attr.qp_state = IB_QPS_ERR;
/* drain work and packet queues */
rxe_sched_task(&qp->resp.task);
if (qp_type(qp) == IB_QPT_RC)
rxe_sched_task(&qp->comp.task);
else
__rxe_do_task(&qp->comp.task);
rxe_sched_task(&qp->comp.task);
rxe_sched_task(&qp->req.task);
spin_unlock_bh(&qp->state_lock);
}
static void rxe_qp_sqd(struct rxe_qp *qp, struct ib_qp_attr *attr,
int mask)
{
spin_lock_bh(&qp->state_lock);
qp->attr.sq_draining = 1;
rxe_sched_task(&qp->comp.task);
rxe_sched_task(&qp->req.task);
spin_unlock_bh(&qp->state_lock);
}
/* caller should hold qp->state_lock */
static int __qp_chk_state(struct rxe_qp *qp, struct ib_qp_attr *attr,
int mask)
{
enum ib_qp_state cur_state;
enum ib_qp_state new_state;
cur_state = (mask & IB_QP_CUR_STATE) ?
attr->cur_qp_state : qp->attr.qp_state;
new_state = (mask & IB_QP_STATE) ?
attr->qp_state : cur_state;
if (!ib_modify_qp_is_ok(cur_state, new_state, qp_type(qp), mask))
return -EINVAL;
if (mask & IB_QP_STATE && cur_state == IB_QPS_SQD) {
if (qp->attr.sq_draining && new_state != IB_QPS_ERR)
return -EINVAL;
}
return 0;
}
static const char *const qps2str[] = {
[IB_QPS_RESET] = "RESET",
[IB_QPS_INIT] = "INIT",
[IB_QPS_RTR] = "RTR",
[IB_QPS_RTS] = "RTS",
[IB_QPS_SQD] = "SQD",
[IB_QPS_SQE] = "SQE",
[IB_QPS_ERR] = "ERR",
};
/* called by the modify qp verb */
int rxe_qp_from_attr(struct rxe_qp *qp, struct ib_qp_attr *attr, int mask,
struct ib_udata *udata)
{
int err;
if (mask & IB_QP_CUR_STATE)
qp->attr.cur_qp_state = attr->qp_state;
if (mask & IB_QP_STATE) {
spin_lock_bh(&qp->state_lock);
err = __qp_chk_state(qp, attr, mask);
if (!err) {
qp->attr.qp_state = attr->qp_state;
rxe_dbg_qp(qp, "state -> %s\n",
qps2str[attr->qp_state]);
}
spin_unlock_bh(&qp->state_lock);
if (err)
return err;
switch (attr->qp_state) {
case IB_QPS_RESET:
rxe_qp_reset(qp);
break;
case IB_QPS_SQD:
rxe_qp_sqd(qp, attr, mask);
break;
case IB_QPS_ERR:
rxe_qp_error(qp);
break;
default:
break;
}
}
if (mask & IB_QP_MAX_QP_RD_ATOMIC) {
int max_rd_atomic = attr->max_rd_atomic ?
roundup_pow_of_two(attr->max_rd_atomic) : 0;
@ -587,9 +603,6 @@ int rxe_qp_from_attr(struct rxe_qp *qp, struct ib_qp_attr *attr, int mask,
return err;
}
if (mask & IB_QP_CUR_STATE)
qp->attr.cur_qp_state = attr->qp_state;
if (mask & IB_QP_EN_SQD_ASYNC_NOTIFY)
qp->attr.en_sqd_async_notify = attr->en_sqd_async_notify;
@ -669,50 +682,6 @@ int rxe_qp_from_attr(struct rxe_qp *qp, struct ib_qp_attr *attr, int mask,
if (mask & IB_QP_DEST_QPN)
qp->attr.dest_qp_num = attr->dest_qp_num;
if (mask & IB_QP_STATE) {
qp->attr.qp_state = attr->qp_state;
switch (attr->qp_state) {
case IB_QPS_RESET:
rxe_dbg_qp(qp, "state -> RESET\n");
rxe_qp_reset(qp);
break;
case IB_QPS_INIT:
rxe_dbg_qp(qp, "state -> INIT\n");
qp->req.state = QP_STATE_INIT;
qp->resp.state = QP_STATE_INIT;
qp->comp.state = QP_STATE_INIT;
break;
case IB_QPS_RTR:
rxe_dbg_qp(qp, "state -> RTR\n");
qp->resp.state = QP_STATE_READY;
break;
case IB_QPS_RTS:
rxe_dbg_qp(qp, "state -> RTS\n");
qp->req.state = QP_STATE_READY;
qp->comp.state = QP_STATE_READY;
break;
case IB_QPS_SQD:
rxe_dbg_qp(qp, "state -> SQD\n");
rxe_qp_drain(qp);
break;
case IB_QPS_SQE:
rxe_dbg_qp(qp, "state -> SQE !!?\n");
/* Not possible from modify_qp. */
break;
case IB_QPS_ERR:
rxe_dbg_qp(qp, "state -> ERR\n");
rxe_qp_error(qp);
break;
}
}
return 0;
}
@ -736,18 +705,15 @@ int rxe_qp_to_attr(struct rxe_qp *qp, struct ib_qp_attr *attr, int mask)
rxe_av_to_attr(&qp->pri_av, &attr->ah_attr);
rxe_av_to_attr(&qp->alt_av, &attr->alt_ah_attr);
if (qp->req.state == QP_STATE_DRAIN) {
attr->sq_draining = 1;
/* applications that get this state
* typically spin on it. yield the
* processor
*/
/* Applications that get this state typically spin on it.
* Yield the processor
*/
spin_lock_bh(&qp->state_lock);
if (qp->attr.sq_draining) {
spin_unlock_bh(&qp->state_lock);
cond_resched();
} else {
attr->sq_draining = 0;
}
rxe_dbg_qp(qp, "attr->sq_draining = %d\n", attr->sq_draining);
spin_unlock_bh(&qp->state_lock);
return 0;
}
@ -771,26 +737,29 @@ static void rxe_qp_do_cleanup(struct work_struct *work)
{
struct rxe_qp *qp = container_of(work, typeof(*qp), cleanup_work.work);
spin_lock_bh(&qp->state_lock);
qp->valid = 0;
spin_unlock_bh(&qp->state_lock);
qp->qp_timeout_jiffies = 0;
rxe_cleanup_task(&qp->resp.task);
if (qp_type(qp) == IB_QPT_RC) {
del_timer_sync(&qp->retrans_timer);
del_timer_sync(&qp->rnr_nak_timer);
}
rxe_cleanup_task(&qp->req.task);
rxe_cleanup_task(&qp->comp.task);
if (qp->resp.task.func)
rxe_cleanup_task(&qp->resp.task);
if (qp->req.task.func)
rxe_cleanup_task(&qp->req.task);
if (qp->comp.task.func)
rxe_cleanup_task(&qp->comp.task);
/* flush out any receive wr's or pending requests */
if (qp->req.task.func)
__rxe_do_task(&qp->req.task);
if (qp->sq.queue) {
__rxe_do_task(&qp->comp.task);
__rxe_do_task(&qp->req.task);
}
rxe_requester(qp);
rxe_completer(qp);
rxe_responder(qp);
if (qp->sq.queue)
rxe_queue_cleanup(qp->sq.queue);

5
drivers/infiniband/sw/rxe/rxe_queue.c
View File

@ -61,11 +61,11 @@ struct rxe_queue *rxe_queue_init(struct rxe_dev *rxe, int *num_elem,
/* num_elem == 0 is allowed, but uninteresting */
if (*num_elem < 0)
goto err1;
return NULL;
q = kzalloc(sizeof(*q), GFP_KERNEL);
if (!q)
goto err1;
return NULL;
q->rxe = rxe;
q->type = type;
@ -100,7 +100,6 @@ struct rxe_queue *rxe_queue_init(struct rxe_dev *rxe, int *num_elem,
err2:
kfree(q);
err1:
return NULL;
}

15
drivers/infiniband/sw/rxe/rxe_recv.c
View File

@ -38,12 +38,19 @@ static int check_type_state(struct rxe_dev *rxe, struct rxe_pkt_info *pkt,
return -EINVAL;
}
spin_lock_bh(&qp->state_lock);
if (pkt->mask & RXE_REQ_MASK) {
if (unlikely(qp->resp.state != QP_STATE_READY))
if (unlikely(qp_state(qp) < IB_QPS_RTR)) {
spin_unlock_bh(&qp->state_lock);
return -EINVAL;
} else if (unlikely(qp->req.state < QP_STATE_READY ||
qp->req.state > QP_STATE_DRAINED))
return -EINVAL;
}
} else {
if (unlikely(qp_state(qp) < IB_QPS_RTS)) {
spin_unlock_bh(&qp->state_lock);
return -EINVAL;
}
}
spin_unlock_bh(&qp->state_lock);
return 0;
}

104
drivers/infiniband/sw/rxe/rxe_req.c
View File

@ -102,44 +102,44 @@ void rnr_nak_timer(struct timer_list *t)
rxe_dbg_qp(qp, "nak timer fired\n");
/* request a send queue retry */
qp->req.need_retry = 1;
qp->req.wait_for_rnr_timer = 0;
rxe_sched_task(&qp->req.task);
spin_lock_bh(&qp->state_lock);
if (qp->valid) {
/* request a send queue retry */
qp->req.need_retry = 1;
qp->req.wait_for_rnr_timer = 0;
rxe_sched_task(&qp->req.task);
}
spin_unlock_bh(&qp->state_lock);
}
static struct rxe_send_wqe *req_next_wqe(struct rxe_qp *qp)
static void req_check_sq_drain_done(struct rxe_qp *qp)
{
struct rxe_send_wqe *wqe;
struct rxe_queue *q = qp->sq.queue;
unsigned int index = qp->req.wqe_index;
struct rxe_queue *q;
unsigned int index;
unsigned int cons;
unsigned int prod;
struct rxe_send_wqe *wqe;
wqe = queue_head(q, QUEUE_TYPE_FROM_CLIENT);
cons = queue_get_consumer(q, QUEUE_TYPE_FROM_CLIENT);
prod = queue_get_producer(q, QUEUE_TYPE_FROM_CLIENT);
spin_lock_bh(&qp->state_lock);
if (qp_state(qp) == IB_QPS_SQD) {
q = qp->sq.queue;
index = qp->req.wqe_index;
cons = queue_get_consumer(q, QUEUE_TYPE_FROM_CLIENT);
wqe = queue_addr_from_index(q, cons);
if (unlikely(qp->req.state == QP_STATE_DRAIN)) {
/* check to see if we are drained;
* state_lock used by requester and completer
*/
spin_lock_bh(&qp->state_lock);
do {
if (qp->req.state != QP_STATE_DRAIN) {
if (!qp->attr.sq_draining)
/* comp just finished */
spin_unlock_bh(&qp->state_lock);
break;
}
if (wqe && ((index != cons) ||
(wqe->state != wqe_state_posted))) {
(wqe->state != wqe_state_posted)))
/* comp not done yet */
spin_unlock_bh(&qp->state_lock);
break;
}
qp->req.state = QP_STATE_DRAINED;
qp->attr.sq_draining = 0;
spin_unlock_bh(&qp->state_lock);
if (qp->ibqp.event_handler) {
@ -151,19 +151,43 @@ static struct rxe_send_wqe *req_next_wqe(struct rxe_qp *qp)
qp->ibqp.event_handler(&ev,
qp->ibqp.qp_context);
}
return;
} while (0);
}
spin_unlock_bh(&qp->state_lock);
}
static struct rxe_send_wqe *__req_next_wqe(struct rxe_qp *qp)
{
struct rxe_queue *q = qp->sq.queue;
unsigned int index = qp->req.wqe_index;
unsigned int prod;
prod = queue_get_producer(q, QUEUE_TYPE_FROM_CLIENT);
if (index == prod)
return NULL;
else
return queue_addr_from_index(q, index);
}
wqe = queue_addr_from_index(q, index);
static struct rxe_send_wqe *req_next_wqe(struct rxe_qp *qp)
{
struct rxe_send_wqe *wqe;
if (unlikely((qp->req.state == QP_STATE_DRAIN ||
qp->req.state == QP_STATE_DRAINED) &&
(wqe->state != wqe_state_processing)))
req_check_sq_drain_done(qp);
wqe = __req_next_wqe(qp);
if (wqe == NULL)
return NULL;
spin_lock_bh(&qp->state_lock);
if (unlikely((qp_state(qp) == IB_QPS_SQD) &&
(wqe->state != wqe_state_processing))) {
spin_unlock_bh(&qp->state_lock);
return NULL;
}
spin_unlock_bh(&qp->state_lock);
wqe->mask = wr_opcode_mask(wqe->wr.opcode, qp);
return wqe;
}
@ -635,9 +659,8 @@ static int rxe_do_local_ops(struct rxe_qp *qp, struct rxe_send_wqe *wqe)
return 0;
}
int rxe_requester(void *arg)
int rxe_requester(struct rxe_qp *qp)
{
struct rxe_qp *qp = (struct rxe_qp *)arg;
struct rxe_dev *rxe = to_rdev(qp->ibqp.device);
struct rxe_pkt_info pkt;
struct sk_buff *skb;
@ -654,24 +677,22 @@ int rxe_requester(void *arg)
struct rxe_ah *ah;
struct rxe_av *av;
if (!rxe_get(qp))
return -EAGAIN;
if (unlikely(!qp->valid))
spin_lock_bh(&qp->state_lock);
if (unlikely(!qp->valid)) {
spin_unlock_bh(&qp->state_lock);
goto exit;
}
if (unlikely(qp->req.state == QP_STATE_ERROR)) {
wqe = req_next_wqe(qp);
if (unlikely(qp_state(qp) == IB_QPS_ERR)) {
wqe = __req_next_wqe(qp);
spin_unlock_bh(&qp->state_lock);
if (wqe)
/*
* Generate an error completion for error qp state
*/
goto err;
else
goto exit;
}
if (unlikely(qp->req.state == QP_STATE_RESET)) {
if (unlikely(qp_state(qp) == IB_QPS_RESET)) {
qp->req.wqe_index = queue_get_consumer(q,
QUEUE_TYPE_FROM_CLIENT);
qp->req.opcode = -1;
@ -679,8 +700,10 @@ int rxe_requester(void *arg)
qp->req.wait_psn = 0;
qp->req.need_retry = 0;
qp->req.wait_for_rnr_timer = 0;
spin_unlock_bh(&qp->state_lock);
goto exit;
}
spin_unlock_bh(&qp->state_lock);
/* we come here if the retransmit timer has fired
* or if the rnr timer has fired. If the retransmit
@ -757,7 +780,7 @@ int rxe_requester(void *arg)
qp->req.wqe_index);
wqe->state = wqe_state_done;
wqe->status = IB_WC_SUCCESS;
rxe_run_task(&qp->comp.task);
rxe_sched_task(&qp->comp.task);
goto done;
}
payload = mtu;
@ -840,12 +863,9 @@ err:
/* update wqe_index for each wqe completion */
qp->req.wqe_index = queue_next_index(qp->sq.queue, qp->req.wqe_index);
wqe->state = wqe_state_error;
qp->req.state = QP_STATE_ERROR;
rxe_run_task(&qp->comp.task);
rxe_qp_error(qp);
exit:
ret = -EAGAIN;
out:
rxe_put(qp);
return ret;
}

136
drivers/infiniband/sw/rxe/rxe_resp.c
View File

@ -42,7 +42,6 @@ static char *resp_state_name[] = {
[RESPST_ERR_LENGTH] = "ERR_LENGTH",
[RESPST_ERR_CQ_OVERFLOW] = "ERR_CQ_OVERFLOW",
[RESPST_ERROR] = "ERROR",
[RESPST_RESET] = "RESET",
[RESPST_DONE] = "DONE",
[RESPST_EXIT] = "EXIT",
};
@ -69,17 +68,6 @@ static inline enum resp_states get_req(struct rxe_qp *qp,
{
struct sk_buff *skb;
if (qp->resp.state == QP_STATE_ERROR) {
while ((skb = skb_dequeue(&qp->req_pkts))) {
rxe_put(qp);
kfree_skb(skb);
ib_device_put(qp->ibqp.device);
}
/* go drain recv wr queue */
return RESPST_CHK_RESOURCE;
}
skb = skb_peek(&qp->req_pkts);
if (!skb)
return RESPST_EXIT;
@ -334,24 +322,6 @@ static enum resp_states check_resource(struct rxe_qp *qp,
{
struct rxe_srq *srq = qp->srq;
if (qp->resp.state == QP_STATE_ERROR) {
if (qp->resp.wqe) {
qp->resp.status = IB_WC_WR_FLUSH_ERR;
return RESPST_COMPLETE;
} else if (!srq) {
qp->resp.wqe = queue_head(qp->rq.queue,
QUEUE_TYPE_FROM_CLIENT);
if (qp->resp.wqe) {
qp->resp.status = IB_WC_WR_FLUSH_ERR;
return RESPST_COMPLETE;
} else {
return RESPST_EXIT;
}
} else {
return RESPST_EXIT;
}
}
if (pkt->mask & (RXE_READ_OR_ATOMIC_MASK | RXE_ATOMIC_WRITE_MASK)) {
/* it is the requesters job to not send
* too many read/atomic ops, we just
@ -1151,6 +1121,10 @@ static enum resp_states do_complete(struct rxe_qp *qp,
wc->port_num = qp->attr.port_num;
}
} else {
if (wc->status != IB_WC_WR_FLUSH_ERR)
rxe_err_qp(qp, "non-flush error status = %d",
wc->status);
}
/* have copy for srq and reference for !srq */
@ -1163,8 +1137,13 @@ static enum resp_states do_complete(struct rxe_qp *qp,
return RESPST_ERR_CQ_OVERFLOW;
finish:
if (unlikely(qp->resp.state == QP_STATE_ERROR))
spin_lock_bh(&qp->state_lock);
if (unlikely(qp_state(qp) == IB_QPS_ERR)) {
spin_unlock_bh(&qp->state_lock);
return RESPST_CHK_RESOURCE;
}
spin_unlock_bh(&qp->state_lock);
if (unlikely(!pkt))
return RESPST_DONE;
if (qp_type(qp) == IB_QPT_RC)
@ -1421,49 +1400,90 @@ static enum resp_states do_class_d1e_error(struct rxe_qp *qp)
}
}
static void rxe_drain_req_pkts(struct rxe_qp *qp, bool notify)
/* drain incoming request packet queue */
static void drain_req_pkts(struct rxe_qp *qp)
{
struct sk_buff *skb;
struct rxe_queue *q = qp->rq.queue;
while ((skb = skb_dequeue(&qp->req_pkts))) {
rxe_put(qp);
kfree_skb(skb);
ib_device_put(qp->ibqp.device);
}
if (notify)
return;
while (!qp->srq && q && queue_head(q, q->type))
queue_advance_consumer(q, q->type);
}
int rxe_responder(void *arg)
/* complete receive wqe with flush error */
static int flush_recv_wqe(struct rxe_qp *qp, struct rxe_recv_wqe *wqe)
{
struct rxe_cqe cqe = {};
struct ib_wc *wc = &cqe.ibwc;
struct ib_uverbs_wc *uwc = &cqe.uibwc;
int err;
if (qp->rcq->is_user) {
uwc->wr_id = wqe->wr_id;
uwc->status = IB_WC_WR_FLUSH_ERR;
uwc->qp_num = qp_num(qp);
} else {
wc->wr_id = wqe->wr_id;
wc->status = IB_WC_WR_FLUSH_ERR;
wc->qp = &qp->ibqp;
}
err = rxe_cq_post(qp->rcq, &cqe, 0);
if (err)
rxe_dbg_cq(qp->rcq, "post cq failed err = %d", err);
return err;
}
/* drain and optionally complete the recive queue
* if unable to complete a wqe stop completing and
* just flush the remaining wqes
*/
static void flush_recv_queue(struct rxe_qp *qp, bool notify)
{
struct rxe_queue *q = qp->rq.queue;
struct rxe_recv_wqe *wqe;
int err;
if (qp->srq)
return;
while ((wqe = queue_head(q, q->type))) {
if (notify) {
err = flush_recv_wqe(qp, wqe);
if (err)
notify = 0;
}
queue_advance_consumer(q, q->type);
}
qp->resp.wqe = NULL;
}
int rxe_responder(struct rxe_qp *qp)
{
struct rxe_qp *qp = (struct rxe_qp *)arg;
struct rxe_dev *rxe = to_rdev(qp->ibqp.device);
enum resp_states state;
struct rxe_pkt_info *pkt = NULL;
int ret;
if (!rxe_get(qp))
return -EAGAIN;
spin_lock_bh(&qp->state_lock);
if (!qp->valid || qp_state(qp) == IB_QPS_ERR ||
qp_state(qp) == IB_QPS_RESET) {
bool notify = qp->valid && (qp_state(qp) == IB_QPS_ERR);
drain_req_pkts(qp);
flush_recv_queue(qp, notify);
spin_unlock_bh(&qp->state_lock);
goto exit;
}
spin_unlock_bh(&qp->state_lock);
qp->resp.aeth_syndrome = AETH_ACK_UNLIMITED;
if (!qp->valid)
goto exit;
switch (qp->resp.state) {
case QP_STATE_RESET:
state = RESPST_RESET;
break;
default:
state = RESPST_GET_REQ;
break;
}
state = RESPST_GET_REQ;
while (1) {
rxe_dbg_qp(qp, "state = %s\n", resp_state_name[state]);
@ -1622,11 +1642,6 @@ int rxe_responder(void *arg)
goto exit;
case RESPST_RESET:
rxe_drain_req_pkts(qp, false);
qp->resp.wqe = NULL;
goto exit;
case RESPST_ERROR:
qp->resp.goto_error = 0;
rxe_dbg_qp(qp, "moved to error state\n");
@ -1648,6 +1663,5 @@ done:
exit:
ret = -EAGAIN;
out:
rxe_put(qp);
return ret;
}

6
drivers/infiniband/sw/rxe/rxe_srq.c
View File

@ -13,13 +13,13 @@ int rxe_srq_chk_init(struct rxe_dev *rxe, struct ib_srq_init_attr *init)
struct ib_srq_attr *attr = &init->attr;
if (attr->max_wr > rxe->attr.max_srq_wr) {
rxe_dbg(rxe, "max_wr(%d) > max_srq_wr(%d)\n",
rxe_dbg_dev(rxe, "max_wr(%d) > max_srq_wr(%d)\n",
attr->max_wr, rxe->attr.max_srq_wr);
goto err1;
}
if (attr->max_wr <= 0) {
rxe_dbg(rxe, "max_wr(%d) <= 0\n", attr->max_wr);
rxe_dbg_dev(rxe, "max_wr(%d) <= 0\n", attr->max_wr);
goto err1;
}
@ -27,7 +27,7 @@ int rxe_srq_chk_init(struct rxe_dev *rxe, struct ib_srq_init_attr *init)
attr->max_wr = RXE_MIN_SRQ_WR;
if (attr->max_sge > rxe->attr.max_srq_sge) {
rxe_dbg(rxe, "max_sge(%d) > max_srq_sge(%d)\n",
rxe_dbg_dev(rxe, "max_sge(%d) > max_srq_sge(%d)\n",
attr->max_sge, rxe->attr.max_srq_sge);
goto err1;
}

270
drivers/infiniband/sw/rxe/rxe_task.c
View File

@ -6,69 +6,128 @@
#include "rxe.h"
int __rxe_do_task(struct rxe_task *task)
/* Check if task is idle i.e. not running, not scheduled in
* tasklet queue and not draining. If so move to busy to
* reserve a slot in do_task() by setting to busy and taking
* a qp reference to cover the gap from now until the task finishes.
* state will move out of busy if task returns a non zero value
* in do_task(). If state is already busy it is raised to armed
* to indicate to do_task that additional pass should be made
* over the task.
* Context: caller should hold task->lock.
* Returns: true if state transitioned from idle to busy else false.
*/
static bool __reserve_if_idle(struct rxe_task *task)
{
int ret;
WARN_ON(rxe_read(task->qp) <= 0);
while ((ret = task->func(task->arg)) == 0)
;
if (task->tasklet.state & BIT(TASKLET_STATE_SCHED))
return false;
task->ret = ret;
if (task->state == TASK_STATE_IDLE) {
rxe_get(task->qp);
task->state = TASK_STATE_BUSY;
task->num_sched++;
return true;
}
return ret;
if (task->state == TASK_STATE_BUSY)
task->state = TASK_STATE_ARMED;
return false;
}
/*
* this locking is due to a potential race where
* a second caller finds the task already running
* but looks just after the last call to func
/* check if task is idle or drained and not currently
* scheduled in the tasklet queue. This routine is
* called by rxe_cleanup_task or rxe_disable_task to
* see if the queue is empty.
* Context: caller should hold task->lock.
* Returns true if done else false.
*/
static bool __is_done(struct rxe_task *task)
{
if (task->tasklet.state & BIT(TASKLET_STATE_SCHED))
return false;
if (task->state == TASK_STATE_IDLE ||
task->state == TASK_STATE_DRAINED) {
return true;
}
return false;
}
/* a locked version of __is_done */
static bool is_done(struct rxe_task *task)
{
unsigned long flags;
int done;
spin_lock_irqsave(&task->lock, flags);
done = __is_done(task);
spin_unlock_irqrestore(&task->lock, flags);
return done;
}
/* do_task is a wrapper for the three tasks (requester,
* completer, responder) and calls them in a loop until
* they return a non-zero value. It is called either
* directly by rxe_run_task or indirectly if rxe_sched_task
* schedules the task. They must call __reserve_if_idle to
* move the task to busy before calling or scheduling.
* The task can also be moved to drained or invalid
* by calls to rxe-cleanup_task or rxe_disable_task.
* In that case tasks which get here are not executed but
* just flushed. The tasks are designed to look to see if
* there is work to do and do part of it before returning
* here with a return value of zero until all the work
* has been consumed then it retuens a non-zero value.
* The number of times the task can be run is limited by
* max iterations so one task cannot hold the cpu forever.
*/
static void do_task(struct tasklet_struct *t)
{
int cont;
int ret;
struct rxe_task *task = from_tasklet(task, t, tasklet);
struct rxe_qp *qp = (struct rxe_qp *)task->arg;
unsigned int iterations = RXE_MAX_ITERATIONS;
unsigned int iterations;
unsigned long flags;
int resched = 0;
spin_lock_bh(&task->lock);
switch (task->state) {
case TASK_STATE_START:
task->state = TASK_STATE_BUSY;
spin_unlock_bh(&task->lock);
break;
WARN_ON(rxe_read(task->qp) <= 0);
case TASK_STATE_BUSY:
task->state = TASK_STATE_ARMED;
fallthrough;
case TASK_STATE_ARMED:
spin_unlock_bh(&task->lock);
return;
default:
spin_unlock_bh(&task->lock);
rxe_dbg_qp(qp, "failed with bad state %d\n", task->state);
spin_lock_irqsave(&task->lock, flags);
if (task->state >= TASK_STATE_DRAINED) {
rxe_put(task->qp);
task->num_done++;
spin_unlock_irqrestore(&task->lock, flags);
return;
}
spin_unlock_irqrestore(&task->lock, flags);
do {
iterations = RXE_MAX_ITERATIONS;
cont = 0;
ret = task->func(task->arg);
spin_lock_bh(&task->lock);
do {
ret = task->func(task->qp);
} while (ret == 0 && iterations-- > 0);
spin_lock_irqsave(&task->lock, flags);
switch (task->state) {
case TASK_STATE_BUSY:
if (ret) {
task->state = TASK_STATE_START;
} else if (iterations--) {
cont = 1;
task->state = TASK_STATE_IDLE;
} else {
/* reschedule the tasklet and exit
/* This can happen if the client
* can add work faster than the
* tasklet can finish it.
* Reschedule the tasklet and exit
* the loop to give up the cpu
*/
tasklet_schedule(&task->tasklet);
task->state = TASK_STATE_START;
task->state = TASK_STATE_IDLE;
resched = 1;
}
break;
@ -81,71 +140,158 @@ static void do_task(struct tasklet_struct *t)
cont = 1;
break;
case TASK_STATE_DRAINING:
if (ret)
task->state = TASK_STATE_DRAINED;
else
cont = 1;
break;
default:
rxe_dbg_qp(qp, "failed with bad state %d\n",
task->state);
WARN_ON(1);
rxe_info_qp(task->qp, "unexpected task state = %d", task->state);
}
spin_unlock_bh(&task->lock);
if (!cont) {
task->num_done++;
if (WARN_ON(task->num_done != task->num_sched))
rxe_err_qp(task->qp, "%ld tasks scheduled, %ld tasks done",
task->num_sched, task->num_done);
}
spin_unlock_irqrestore(&task->lock, flags);
} while (cont);
task->ret = ret;
if (resched)
rxe_sched_task(task);
rxe_put(task->qp);
}
int rxe_init_task(struct rxe_task *task, void *arg, int (*func)(void *))
int rxe_init_task(struct rxe_task *task, struct rxe_qp *qp,
int (*func)(struct rxe_qp *))
{
task->arg = arg;
task->func = func;
task->destroyed = false;
WARN_ON(rxe_read(qp) <= 0);
task->qp = qp;
task->func = func;
tasklet_setup(&task->tasklet, do_task);
task->state = TASK_STATE_START;
task->state = TASK_STATE_IDLE;
spin_lock_init(&task->lock);
return 0;
}
/* rxe_cleanup_task is only called from rxe_do_qp_cleanup in
* process context. The qp is already completed with no
* remaining references. Once the queue is drained the
* task is moved to invalid and returns. The qp cleanup
* code then calls the task functions directly without
* using the task struct to drain any late arriving packets
* or work requests.
*/
void rxe_cleanup_task(struct rxe_task *task)
{
bool idle;
unsigned long flags;
/*
* Mark the task, then wait for it to finish. It might be
* running in a non-tasklet (direct call) context.
spin_lock_irqsave(&task->lock, flags);
if (!__is_done(task) && task->state < TASK_STATE_DRAINED) {
task->state = TASK_STATE_DRAINING;
} else {
task->state = TASK_STATE_INVALID;
spin_unlock_irqrestore(&task->lock, flags);
return;
}
spin_unlock_irqrestore(&task->lock, flags);
/* now the task cannot be scheduled or run just wait
* for the previously scheduled tasks to finish.
*/
task->destroyed = true;
do {
spin_lock_bh(&task->lock);
idle = (task->state == TASK_STATE_START);
spin_unlock_bh(&task->lock);
} while (!idle);
while (!is_done(task))
cond_resched();
tasklet_kill(&task->tasklet);
spin_lock_irqsave(&task->lock, flags);
task->state = TASK_STATE_INVALID;
spin_unlock_irqrestore(&task->lock, flags);
}
/* run the task inline if it is currently idle
* cannot call do_task holding the lock
*/
void rxe_run_task(struct rxe_task *task)
{
if (task->destroyed)
return;
unsigned long flags;
int run;
do_task(&task->tasklet);
WARN_ON(rxe_read(task->qp) <= 0);
spin_lock_irqsave(&task->lock, flags);
run = __reserve_if_idle(task);
spin_unlock_irqrestore(&task->lock, flags);
if (run)
do_task(&task->tasklet);
}
/* schedule the task to run later as a tasklet.
* the tasklet)schedule call can be called holding
* the lock.
*/
void rxe_sched_task(struct rxe_task *task)
{
if (task->destroyed)
return;
unsigned long flags;
tasklet_schedule(&task->tasklet);
WARN_ON(rxe_read(task->qp) <= 0);
spin_lock_irqsave(&task->lock, flags);
if (__reserve_if_idle(task))
tasklet_schedule(&task->tasklet);
spin_unlock_irqrestore(&task->lock, flags);
}
/* rxe_disable/enable_task are only called from
* rxe_modify_qp in process context. Task is moved
* to the drained state by do_task.
*/
void rxe_disable_task(struct rxe_task *task)
{
unsigned long flags;
WARN_ON(rxe_read(task->qp) <= 0);
spin_lock_irqsave(&task->lock, flags);
if (!__is_done(task) && task->state < TASK_STATE_DRAINED) {
task->state = TASK_STATE_DRAINING;
} else {
task->state = TASK_STATE_DRAINED;
spin_unlock_irqrestore(&task->lock, flags);
return;
}
spin_unlock_irqrestore(&task->lock, flags);
while (!is_done(task))
cond_resched();
tasklet_disable(&task->tasklet);
}
void rxe_enable_task(struct rxe_task *task)
{
unsigned long flags;
WARN_ON(rxe_read(task->qp) <= 0);
spin_lock_irqsave(&task->lock, flags);
if (task->state == TASK_STATE_INVALID) {
spin_unlock_irqrestore(&task->lock, flags);
return;
}
task->state = TASK_STATE_IDLE;
tasklet_enable(&task->tasklet);
spin_unlock_irqrestore(&task->lock, flags);
}

23
drivers/infiniband/sw/rxe/rxe_task.h
View File

@ -8,9 +8,12 @@
#define RXE_TASK_H
enum {
TASK_STATE_START = 0,
TASK_STATE_IDLE = 0,
TASK_STATE_BUSY = 1,
TASK_STATE_ARMED = 2,
TASK_STATE_DRAINING = 3,
TASK_STATE_DRAINED = 4,
TASK_STATE_INVALID = 5,
};
/*
@ -22,28 +25,24 @@ struct rxe_task {
struct tasklet_struct tasklet;
int state;
spinlock_t lock;
void *arg;
int (*func)(void *arg);
struct rxe_qp *qp;
int (*func)(struct rxe_qp *qp);
int ret;
bool destroyed;
long num_sched;
long num_done;
};
/*
* init rxe_task structure
* arg => parameter to pass to fcn
* qp => parameter to pass to func
* func => function to call until it returns != 0
*/
int rxe_init_task(struct rxe_task *task, void *arg, int (*func)(void *));
int rxe_init_task(struct rxe_task *task, struct rxe_qp *qp,
int (*func)(struct rxe_qp *));
/* cleanup task */
void rxe_cleanup_task(struct rxe_task *task);
/*
* raw call to func in loop without any checking
* can call when tasklets are disabled
*/
int __rxe_do_task(struct rxe_task *task);
void rxe_run_task(struct rxe_task *task);
void rxe_sched_task(struct rxe_task *task);

1023
drivers/infiniband/sw/rxe/rxe_verbs.c
View File

File diff suppressed because it is too large Load Diff

14
drivers/infiniband/sw/rxe/rxe_verbs.h
View File

@ -63,9 +63,7 @@ struct rxe_cq {
struct rxe_queue *queue;
spinlock_t cq_lock;
u8 notify;
bool is_dying;
bool is_user;
struct tasklet_struct comp_task;
atomic_t num_wq;
};
@ -104,17 +102,7 @@ struct rxe_srq {
int error;
};
enum rxe_qp_state {
QP_STATE_RESET,
QP_STATE_INIT,
QP_STATE_READY,
QP_STATE_DRAIN, /* req only */
QP_STATE_DRAINED, /* req only */
QP_STATE_ERROR
};
struct rxe_req_info {
enum rxe_qp_state state;
int wqe_index;
u32 psn;
int opcode;
@ -129,7 +117,6 @@ struct rxe_req_info {
};
struct rxe_comp_info {
enum rxe_qp_state state;
u32 psn;
int opcode;
int timeout;
@ -175,7 +162,6 @@ struct resp_res {
};
struct rxe_resp_info {
enum rxe_qp_state state;
u32 msn;
u32 psn;
u32 ack_psn;

3
drivers/infiniband/sw/siw/siw_main.c
View File

@ -437,9 +437,6 @@ static int siw_netdev_event(struct notifier_block *nb, unsigned long event,
dev_dbg(&netdev->dev, "siw: event %lu\n", event);
if (dev_net(netdev) != &init_net)
return NOTIFY_OK;
base_dev = ib_device_get_by_netdev(netdev, RDMA_DRIVER_SIW);
if (!base_dev)
return NOTIFY_OK;

6
drivers/infiniband/sw/siw/siw_qp_rx.c
View File

@ -139,7 +139,7 @@ static int siw_rx_pbl(struct siw_rx_stream *srx, int *pbl_idx,
break;
bytes = min(bytes, len);
if (siw_rx_kva(srx, (void *)(uintptr_t)buf_addr, bytes) ==
if (siw_rx_kva(srx, ib_virt_dma_to_ptr(buf_addr), bytes) ==
bytes) {
copied += bytes;
offset += bytes;
@ -487,7 +487,7 @@ int siw_proc_send(struct siw_qp *qp)
mem_p = *mem;
if (mem_p->mem_obj == NULL)
rv = siw_rx_kva(srx,
(void *)(uintptr_t)(sge->laddr + frx->sge_off),
ib_virt_dma_to_ptr(sge->laddr + frx->sge_off),
sge_bytes);
else if (!mem_p->is_pbl)
rv = siw_rx_umem(srx, mem_p->umem,
@ -852,7 +852,7 @@ int siw_proc_rresp(struct siw_qp *qp)
if (mem_p->mem_obj == NULL)
rv = siw_rx_kva(srx,
(void *)(uintptr_t)(sge->laddr + wqe->processed),
ib_virt_dma_to_ptr(sge->laddr + wqe->processed),
bytes);
else if (!mem_p->is_pbl)
rv = siw_rx_umem(srx, mem_p->umem, sge->laddr + wqe->processed,

21
drivers/infiniband/sw/siw/siw_qp_tx.c
View File

@ -29,7 +29,7 @@ static struct page *siw_get_pblpage(struct siw_mem *mem, u64 addr, int *idx)
dma_addr_t paddr = siw_pbl_get_buffer(pbl, offset, NULL, idx);
if (paddr)
return virt_to_page((void *)(uintptr_t)paddr);
return ib_virt_dma_to_page(paddr);
return NULL;
}
@ -56,8 +56,7 @@ static int siw_try_1seg(struct siw_iwarp_tx *c_tx, void *paddr)
if (!mem->mem_obj) {
/* Kernel client using kva */
memcpy(paddr,
(const void *)(uintptr_t)sge->laddr, bytes);
memcpy(paddr, ib_virt_dma_to_ptr(sge->laddr), bytes);
} else if (c_tx->in_syscall) {
if (copy_from_user(paddr, u64_to_user_ptr(sge->laddr),
bytes))
@ -477,7 +476,7 @@ static int siw_tx_hdt(struct siw_iwarp_tx *c_tx, struct socket *s)
* or memory region with assigned kernel buffer
*/
iov[seg].iov_base =
(void *)(uintptr_t)(sge->laddr + sge_off);
ib_virt_dma_to_ptr(sge->laddr + sge_off);
iov[seg].iov_len = sge_len;
if (do_crc)
@ -537,19 +536,13 @@ static int siw_tx_hdt(struct siw_iwarp_tx *c_tx, struct socket *s)
* Cast to an uintptr_t to preserve all 64 bits
* in sge->laddr.
*/
uintptr_t va = (uintptr_t)(sge->laddr + sge_off);
u64 va = sge->laddr + sge_off;
/*
* virt_to_page() takes a (void *) pointer
* so cast to a (void *) meaning it will be 64
* bits on a 64 bit platform and 32 bits on a
* 32 bit platform.
*/
page_array[seg] = virt_to_page((void *)(va & PAGE_MASK));
page_array[seg] = ib_virt_dma_to_page(va);
if (do_crc)
crypto_shash_update(
c_tx->mpa_crc_hd,
(void *)va,
ib_virt_dma_to_ptr(va),
plen);
}
@ -558,7 +551,7 @@ static int siw_tx_hdt(struct siw_iwarp_tx *c_tx, struct socket *s)
data_len -= plen;
fp_off = 0;
if (++seg > (int)MAX_ARRAY) {
if (++seg >= (int)MAX_ARRAY) {
siw_dbg_qp(tx_qp(c_tx), "to many fragments\n");
siw_unmap_pages(iov, kmap_mask, seg-1);
wqe->processed -= c_tx->bytes_unsent;

4
drivers/infiniband/sw/siw/siw_verbs.c
View File

@ -660,7 +660,7 @@ static int siw_copy_inline_sgl(const struct ib_send_wr *core_wr,
bytes = -EINVAL;
break;
}
memcpy(kbuf, (void *)(uintptr_t)core_sge->addr,
memcpy(kbuf, ib_virt_dma_to_ptr(core_sge->addr),
core_sge->length);
kbuf += core_sge->length;
@ -1523,7 +1523,7 @@ int siw_map_mr_sg(struct ib_mr *base_mr, struct scatterlist *sl, int num_sle,
}
siw_dbg_mem(mem,
"sge[%d], size %u, addr 0x%p, total %lu\n",
i, pble->size, (void *)(uintptr_t)pble->addr,
i, pble->size, ib_virt_dma_to_ptr(pble->addr),
pbl_size);
}
rv = ib_sg_to_pages(base_mr, sl, num_sle, sg_off, siw_set_pbl_page);

17
drivers/infiniband/ulp/iser/iser_initiator.c
View File

@ -141,10 +141,14 @@ out_err:
/* creates a new tx descriptor and adds header regd buffer */
static void iser_create_send_desc(struct iser_conn *iser_conn,
struct iser_tx_desc *tx_desc)
struct iser_tx_desc *tx_desc, enum iser_desc_type type,
void (*done)(struct ib_cq *cq, struct ib_wc *wc))
{
struct iser_device *device = iser_conn->ib_conn.device;
tx_desc->type = type;
tx_desc->cqe.done = done;
ib_dma_sync_single_for_cpu(device->ib_device,
tx_desc->dma_addr, ISER_HEADERS_LEN, DMA_TO_DEVICE);
@ -349,9 +353,8 @@ int iser_send_command(struct iscsi_conn *conn, struct iscsi_task *task)
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;
tx_desc->cqe.done = iser_cmd_comp;
iser_create_send_desc(iser_conn, tx_desc);
iser_create_send_desc(iser_conn, tx_desc, ISCSI_TX_SCSI_COMMAND,
iser_cmd_comp);
if (hdr->flags & ISCSI_FLAG_CMD_READ) {
data_buf = &iser_task->data[ISER_DIR_IN];
@ -457,7 +460,6 @@ int iser_send_data_out(struct iscsi_conn *conn, struct iscsi_task *task,
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);
if (!err)
return 0;
@ -478,9 +480,8 @@ int iser_send_control(struct iscsi_conn *conn, struct iscsi_task *task)
struct iser_device *device;
/* build the tx desc regd header and add it to the tx desc dto */
mdesc->type = ISCSI_TX_CONTROL;
mdesc->cqe.done = iser_ctrl_comp;
iser_create_send_desc(iser_conn, mdesc);
iser_create_send_desc(iser_conn, mdesc, ISCSI_TX_CONTROL,
iser_ctrl_comp);
device = iser_conn->ib_conn.device;

6
drivers/infiniband/ulp/iser/iser_verbs.c
View File

@ -37,12 +37,6 @@
#include "iscsi_iser.h"
#define ISCSI_ISER_MAX_CONN 8
#define ISER_MAX_RX_LEN (ISER_QP_MAX_RECV_DTOS * ISCSI_ISER_MAX_CONN)
#define ISER_MAX_TX_LEN (ISER_QP_MAX_REQ_DTOS * ISCSI_ISER_MAX_CONN)
#define ISER_MAX_CQ_LEN (ISER_MAX_RX_LEN + ISER_MAX_TX_LEN + \
ISCSI_ISER_MAX_CONN)
static void iser_qp_event_callback(struct ib_event *cause, void *context)
{
iser_err("qp event %s (%d)\n",

23
drivers/infiniband/ulp/srpt/ib_srpt.c
View File

@ -549,6 +549,7 @@ static int srpt_format_guid(char *buf, unsigned int size, const __be64 *guid)
*/
static int srpt_refresh_port(struct srpt_port *sport)
{
struct ib_mad_agent *mad_agent;
struct ib_mad_reg_req reg_req;
struct ib_port_modify port_modify;
struct ib_port_attr port_attr;
@ -593,24 +594,26 @@ static int srpt_refresh_port(struct srpt_port *sport)
set_bit(IB_MGMT_METHOD_GET, reg_req.method_mask);
set_bit(IB_MGMT_METHOD_SET, reg_req.method_mask);
sport->mad_agent = ib_register_mad_agent(sport->sdev->device,
sport->port,
IB_QPT_GSI,
&reg_req, 0,
srpt_mad_send_handler,
srpt_mad_recv_handler,
sport, 0);
if (IS_ERR(sport->mad_agent)) {
mad_agent = ib_register_mad_agent(sport->sdev->device,
sport->port,
IB_QPT_GSI,
&reg_req, 0,
srpt_mad_send_handler,
srpt_mad_recv_handler,
sport, 0);
if (IS_ERR(mad_agent)) {
pr_err("%s-%d: MAD agent registration failed (%ld). Note: this is expected if SR-IOV is enabled.\n",
dev_name(&sport->sdev->device->dev), sport->port,
PTR_ERR(sport->mad_agent));
PTR_ERR(mad_agent));
sport->mad_agent = NULL;
memset(&port_modify, 0, sizeof(port_modify));
port_modify.clr_port_cap_mask = IB_PORT_DEVICE_MGMT_SUP;
ib_modify_port(sport->sdev->device, sport->port, 0,
&port_modify);
return 0;
}
sport->mad_agent = mad_agent;
}
return 0;

81
drivers/net/ethernet/broadcom/bnxt/bnxt_hsi.h
View File

@ -10321,6 +10321,87 @@ struct hwrm_selftest_irq_output {
u8 valid;
};
/* dbc_dbc (size:64b/8B) */
struct dbc_dbc {
u32 index;
#define DBC_DBC_INDEX_MASK 0xffffffUL
#define DBC_DBC_INDEX_SFT 0
#define DBC_DBC_EPOCH 0x1000000UL
#define DBC_DBC_TOGGLE_MASK 0x6000000UL
#define DBC_DBC_TOGGLE_SFT 25
u32 type_path_xid;
#define DBC_DBC_XID_MASK 0xfffffUL
#define DBC_DBC_XID_SFT 0
#define DBC_DBC_PATH_MASK 0x3000000UL
#define DBC_DBC_PATH_SFT 24
#define DBC_DBC_PATH_ROCE (0x0UL << 24)
#define DBC_DBC_PATH_L2 (0x1UL << 24)
#define DBC_DBC_PATH_ENGINE (0x2UL << 24)
#define DBC_DBC_PATH_LAST DBC_DBC_PATH_ENGINE
#define DBC_DBC_VALID 0x4000000UL
#define DBC_DBC_DEBUG_TRACE 0x8000000UL
#define DBC_DBC_TYPE_MASK 0xf0000000UL
#define DBC_DBC_TYPE_SFT 28
#define DBC_DBC_TYPE_SQ (0x0UL << 28)
#define DBC_DBC_TYPE_RQ (0x1UL << 28)
#define DBC_DBC_TYPE_SRQ (0x2UL << 28)
#define DBC_DBC_TYPE_SRQ_ARM (0x3UL << 28)
#define DBC_DBC_TYPE_CQ (0x4UL << 28)
#define DBC_DBC_TYPE_CQ_ARMSE (0x5UL << 28)
#define DBC_DBC_TYPE_CQ_ARMALL (0x6UL << 28)
#define DBC_DBC_TYPE_CQ_ARMENA (0x7UL << 28)
#define DBC_DBC_TYPE_SRQ_ARMENA (0x8UL << 28)
#define DBC_DBC_TYPE_CQ_CUTOFF_ACK (0x9UL << 28)
#define DBC_DBC_TYPE_NQ (0xaUL << 28)
#define DBC_DBC_TYPE_NQ_ARM (0xbUL << 28)
#define DBC_DBC_TYPE_NQ_MASK (0xeUL << 28)
#define DBC_DBC_TYPE_NULL (0xfUL << 28)
#define DBC_DBC_TYPE_LAST DBC_DBC_TYPE_NULL
};
/* db_push_start (size:64b/8B) */
struct db_push_start {
u64 db;
#define DB_PUSH_START_DB_INDEX_MASK 0xffffffUL
#define DB_PUSH_START_DB_INDEX_SFT 0
#define DB_PUSH_START_DB_PI_LO_MASK 0xff000000UL
#define DB_PUSH_START_DB_PI_LO_SFT 24
#define DB_PUSH_START_DB_XID_MASK 0xfffff00000000ULL
#define DB_PUSH_START_DB_XID_SFT 32
#define DB_PUSH_START_DB_PI_HI_MASK 0xf0000000000000ULL
#define DB_PUSH_START_DB_PI_HI_SFT 52
#define DB_PUSH_START_DB_TYPE_MASK 0xf000000000000000ULL
#define DB_PUSH_START_DB_TYPE_SFT 60
#define DB_PUSH_START_DB_TYPE_PUSH_START (0xcULL << 60)
#define DB_PUSH_START_DB_TYPE_PUSH_END (0xdULL << 60)
#define DB_PUSH_START_DB_TYPE_LAST DB_PUSH_START_DB_TYPE_PUSH_END
};
/* db_push_end (size:64b/8B) */
struct db_push_end {
u64 db;
#define DB_PUSH_END_DB_INDEX_MASK 0xffffffUL
#define DB_PUSH_END_DB_INDEX_SFT 0
#define DB_PUSH_END_DB_PI_LO_MASK 0xff000000UL
#define DB_PUSH_END_DB_PI_LO_SFT 24
#define DB_PUSH_END_DB_XID_MASK 0xfffff00000000ULL
#define DB_PUSH_END_DB_XID_SFT 32
#define DB_PUSH_END_DB_PI_HI_MASK 0xf0000000000000ULL
#define DB_PUSH_END_DB_PI_HI_SFT 52
#define DB_PUSH_END_DB_PATH_MASK 0x300000000000000ULL
#define DB_PUSH_END_DB_PATH_SFT 56
#define DB_PUSH_END_DB_PATH_ROCE (0x0ULL << 56)
#define DB_PUSH_END_DB_PATH_L2 (0x1ULL << 56)
#define DB_PUSH_END_DB_PATH_ENGINE (0x2ULL << 56)
#define DB_PUSH_END_DB_PATH_LAST DB_PUSH_END_DB_PATH_ENGINE
#define DB_PUSH_END_DB_DEBUG_TRACE 0x800000000000000ULL
#define DB_PUSH_END_DB_TYPE_MASK 0xf000000000000000ULL
#define DB_PUSH_END_DB_TYPE_SFT 60
#define DB_PUSH_END_DB_TYPE_PUSH_START (0xcULL << 60)
#define DB_PUSH_END_DB_TYPE_PUSH_END (0xdULL << 60)
#define DB_PUSH_END_DB_TYPE_LAST DB_PUSH_END_DB_TYPE_PUSH_END
};
/* db_push_info (size:64b/8B) */
struct db_push_info {
u32 push_size_push_index;

3
drivers/net/ethernet/mellanox/mlx5/core/en/params.c
View File

@ -930,8 +930,7 @@ static void mlx5e_build_rx_cq_param(struct mlx5_core_dev *mdev,
static u8 rq_end_pad_mode(struct mlx5_core_dev *mdev, struct mlx5e_params *params)
{
bool lro_en = params->packet_merge.type == MLX5E_PACKET_MERGE_LRO;
bool ro = pcie_relaxed_ordering_enabled(mdev->pdev) &&
MLX5_CAP_GEN(mdev, relaxed_ordering_write);
bool ro = MLX5_CAP_GEN(mdev, relaxed_ordering_write);
return ro && lro_en ?
MLX5_WQ_END_PAD_MODE_NONE : MLX5_WQ_END_PAD_MODE_ALIGN;

Some files were not shown because too many files have changed in this diff Show More