summaryrefslogtreecommitdiff
path: root/drivers/infiniband/hw
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/infiniband/hw')
-rw-r--r--drivers/infiniband/hw/bnxt_re/ib_verbs.c20
-rw-r--r--drivers/infiniband/hw/bnxt_re/ib_verbs.h6
-rw-r--r--drivers/infiniband/hw/bnxt_re/main.c2
-rw-r--r--drivers/infiniband/hw/bnxt_re/qplib_sp.c2
-rw-r--r--drivers/infiniband/hw/cxgb3/Kconfig9
-rw-r--r--drivers/infiniband/hw/cxgb3/Makefile2
-rw-r--r--drivers/infiniband/hw/cxgb3/cxio_dbg.c206
-rw-r--r--drivers/infiniband/hw/cxgb3/cxio_hal.h9
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch_cq.c6
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch_provider.c5
-rw-r--r--drivers/infiniband/hw/cxgb4/device.c24
-rw-r--r--drivers/infiniband/hw/cxgb4/mem.c3
-rw-r--r--drivers/infiniband/hw/cxgb4/provider.c24
-rw-r--r--drivers/infiniband/hw/hfi1/driver.c1
-rw-r--r--drivers/infiniband/hw/hfi1/hfi.h1
-rw-r--r--drivers/infiniband/hw/hfi1/qp.c1
-rw-r--r--drivers/infiniband/hw/hfi1/user_exp_rcv.c4
-rw-r--r--drivers/infiniband/hw/hfi1/verbs.c2
-rw-r--r--drivers/infiniband/hw/hns/Makefile2
-rw-r--r--drivers/infiniband/hw/hns/hns_roce_ah.c8
-rw-r--r--drivers/infiniband/hw/hns/hns_roce_cq.c54
-rw-r--r--drivers/infiniband/hw/hns/hns_roce_db.c180
-rw-r--r--drivers/infiniband/hw/hns/hns_roce_device.h59
-rw-r--r--drivers/infiniband/hw/hns/hns_roce_hw_v1.c8
-rw-r--r--drivers/infiniband/hw/hns/hns_roce_hw_v2.c80
-rw-r--r--drivers/infiniband/hw/hns/hns_roce_hw_v2.h3
-rw-r--r--drivers/infiniband/hw/hns/hns_roce_main.c41
-rw-r--r--drivers/infiniband/hw/hns/hns_roce_mr.c2
-rw-r--r--drivers/infiniband/hw/hns/hns_roce_pd.c5
-rw-r--r--drivers/infiniband/hw/hns/hns_roce_qp.c70
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw.h11
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw_cm.c129
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw_cm.h5
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw_ctrl.c56
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw_d.h5
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw_hw.c35
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw_main.c2
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw_puda.c8
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw_type.h11
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw_ucontext.h107
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw_utils.c10
-rw-r--r--drivers/infiniband/hw/i40iw/i40iw_verbs.c57
-rw-r--r--drivers/infiniband/hw/mlx4/ah.c10
-rw-r--r--drivers/infiniband/hw/mlx4/main.c102
-rw-r--r--drivers/infiniband/hw/mlx4/mlx4_ib.h19
-rw-r--r--drivers/infiniband/hw/mlx4/mr.c3
-rw-r--r--drivers/infiniband/hw/mlx4/qp.c27
-rw-r--r--drivers/infiniband/hw/mlx5/ah.c12
-rw-r--r--drivers/infiniband/hw/mlx5/cmd.c104
-rw-r--r--drivers/infiniband/hw/mlx5/cmd.h4
-rw-r--r--drivers/infiniband/hw/mlx5/main.c572
-rw-r--r--drivers/infiniband/hw/mlx5/mlx5_ib.h71
-rw-r--r--drivers/infiniband/hw/mlx5/mr.c257
-rw-r--r--drivers/infiniband/hw/mlx5/qp.c156
-rw-r--r--drivers/infiniband/hw/mthca/mthca_provider.c1
-rw-r--r--drivers/infiniband/hw/nes/nes_verbs.c1
-rw-r--r--drivers/infiniband/hw/ocrdma/ocrdma_ah.c8
-rw-r--r--drivers/infiniband/hw/ocrdma/ocrdma_hw.c22
-rw-r--r--drivers/infiniband/hw/ocrdma/ocrdma_main.c4
-rw-r--r--drivers/infiniband/hw/ocrdma/ocrdma_verbs.c34
-rw-r--r--drivers/infiniband/hw/ocrdma/ocrdma_verbs.h12
-rw-r--r--drivers/infiniband/hw/qedr/main.c9
-rw-r--r--drivers/infiniband/hw/qedr/qedr_roce_cm.c16
-rw-r--r--drivers/infiniband/hw/qedr/verbs.c155
-rw-r--r--drivers/infiniband/hw/qedr/verbs.h6
-rw-r--r--drivers/infiniband/hw/qib/qib.h3
-rw-r--r--drivers/infiniband/hw/qib/qib_diag.c2
-rw-r--r--drivers/infiniband/hw/qib/qib_file_ops.c8
-rw-r--r--drivers/infiniband/hw/qib/qib_iba7322.c10
-rw-r--r--drivers/infiniband/hw/qib/qib_init.c4
-rw-r--r--drivers/infiniband/hw/qib/qib_sdma.c24
-rw-r--r--drivers/infiniband/hw/qib/qib_verbs.c2
-rw-r--r--drivers/infiniband/hw/usnic/usnic_ib_main.c1
-rw-r--r--drivers/infiniband/hw/usnic/usnic_transport.c4
-rw-r--r--drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c32
-rw-r--r--drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c2
76 files changed, 1956 insertions, 1016 deletions
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index 8301d7e5fa8c..a76e206704d4 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -314,12 +314,11 @@ int bnxt_re_query_gid(struct ib_device *ibdev, u8 port_num,
return rc;
}
-int bnxt_re_del_gid(struct ib_device *ibdev, u8 port_num,
- unsigned int index, void **context)
+int bnxt_re_del_gid(const struct ib_gid_attr *attr, void **context)
{
int rc = 0;
struct bnxt_re_gid_ctx *ctx, **ctx_tbl;
- struct bnxt_re_dev *rdev = to_bnxt_re_dev(ibdev, ibdev);
+ struct bnxt_re_dev *rdev = to_bnxt_re_dev(attr->device, ibdev);
struct bnxt_qplib_sgid_tbl *sgid_tbl = &rdev->qplib_res.sgid_tbl;
struct bnxt_qplib_gid *gid_to_del;
@@ -365,15 +364,14 @@ int bnxt_re_del_gid(struct ib_device *ibdev, u8 port_num,
return rc;
}
-int bnxt_re_add_gid(struct ib_device *ibdev, u8 port_num,
- unsigned int index, const union ib_gid *gid,
+int bnxt_re_add_gid(const union ib_gid *gid,
const struct ib_gid_attr *attr, void **context)
{
int rc;
u32 tbl_idx = 0;
u16 vlan_id = 0xFFFF;
struct bnxt_re_gid_ctx *ctx, **ctx_tbl;
- struct bnxt_re_dev *rdev = to_bnxt_re_dev(ibdev, ibdev);
+ struct bnxt_re_dev *rdev = to_bnxt_re_dev(attr->device, ibdev);
struct bnxt_qplib_sgid_tbl *sgid_tbl = &rdev->qplib_res.sgid_tbl;
if ((attr->ndev) && is_vlan_dev(attr->ndev))
@@ -718,8 +716,7 @@ struct ib_ah *bnxt_re_create_ah(struct ib_pd *ib_pd,
grh->sgid_index);
goto fail;
}
- if (sgid_attr.ndev)
- dev_put(sgid_attr.ndev);
+ dev_put(sgid_attr.ndev);
/* Get network header type for this GID */
nw_type = ib_gid_to_network_type(sgid_attr.gid_type, &sgid);
switch (nw_type) {
@@ -1540,14 +1537,13 @@ int bnxt_re_post_srq_recv(struct ib_srq *ib_srq, struct ib_recv_wr *wr,
ib_srq);
struct bnxt_qplib_swqe wqe;
unsigned long flags;
- int rc = 0, payload_sz = 0;
+ int rc = 0;
spin_lock_irqsave(&srq->lock, flags);
while (wr) {
/* Transcribe each ib_recv_wr to qplib_swqe */
wqe.num_sge = wr->num_sge;
- payload_sz = bnxt_re_build_sgl(wr->sg_list, wqe.sg_list,
- wr->num_sge);
+ bnxt_re_build_sgl(wr->sg_list, wqe.sg_list, wr->num_sge);
wqe.wr_id = wr->wr_id;
wqe.type = BNXT_QPLIB_SWQE_TYPE_RECV;
@@ -1698,7 +1694,7 @@ int bnxt_re_modify_qp(struct ib_qp *ib_qp, struct ib_qp_attr *qp_attr,
status = ib_get_cached_gid(&rdev->ibdev, 1,
grh->sgid_index,
&sgid, &sgid_attr);
- if (!status && sgid_attr.ndev) {
+ if (!status) {
memcpy(qp->qplib_qp.smac, sgid_attr.ndev->dev_addr,
ETH_ALEN);
dev_put(sgid_attr.ndev);
diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.h b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
index e62b7c2c7da6..5c6414cad4af 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.h
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.h
@@ -157,10 +157,8 @@ int bnxt_re_get_port_immutable(struct ib_device *ibdev, u8 port_num,
void bnxt_re_query_fw_str(struct ib_device *ibdev, char *str);
int bnxt_re_query_pkey(struct ib_device *ibdev, u8 port_num,
u16 index, u16 *pkey);
-int bnxt_re_del_gid(struct ib_device *ibdev, u8 port_num,
- unsigned int index, void **context);
-int bnxt_re_add_gid(struct ib_device *ibdev, u8 port_num,
- unsigned int index, const union ib_gid *gid,
+int bnxt_re_del_gid(const struct ib_gid_attr *attr, void **context);
+int bnxt_re_add_gid(const union ib_gid *gid,
const struct ib_gid_attr *attr, void **context);
int bnxt_re_query_gid(struct ib_device *ibdev, u8 port_num,
int index, union ib_gid *gid);
diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c
index f6e361750466..f6c739ec8b62 100644
--- a/drivers/infiniband/hw/bnxt_re/main.c
+++ b/drivers/infiniband/hw/bnxt_re/main.c
@@ -574,7 +574,6 @@ static int bnxt_re_register_ib(struct bnxt_re_dev *rdev)
ibdev->get_port_immutable = bnxt_re_get_port_immutable;
ibdev->get_dev_fw_str = bnxt_re_query_fw_str;
ibdev->query_pkey = bnxt_re_query_pkey;
- ibdev->query_gid = bnxt_re_query_gid;
ibdev->get_netdev = bnxt_re_get_netdev;
ibdev->add_gid = bnxt_re_add_gid;
ibdev->del_gid = bnxt_re_del_gid;
@@ -619,6 +618,7 @@ static int bnxt_re_register_ib(struct bnxt_re_dev *rdev)
ibdev->get_hw_stats = bnxt_re_ib_get_hw_stats;
ibdev->alloc_hw_stats = bnxt_re_ib_alloc_hw_stats;
+ ibdev->driver_id = RDMA_DRIVER_BNXT_RE;
return ib_register_device(ibdev, NULL);
}
diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.c b/drivers/infiniband/hw/bnxt_re/qplib_sp.c
index ee98e5efef84..2f3f32eaa1d5 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_sp.c
+++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.c
@@ -154,7 +154,7 @@ int bnxt_qplib_get_dev_attr(struct bnxt_qplib_rcfw *rcfw,
attr->tqm_alloc_reqs[i * 4 + 3] = *(++tqm_alloc);
}
- attr->is_atomic = 0;
+ attr->is_atomic = false;
bail:
bnxt_qplib_rcfw_free_sbuf(rcfw, sbuf);
return rc;
diff --git a/drivers/infiniband/hw/cxgb3/Kconfig b/drivers/infiniband/hw/cxgb3/Kconfig
index 431be733fbbe..a7b77cb3d5d5 100644
--- a/drivers/infiniband/hw/cxgb3/Kconfig
+++ b/drivers/infiniband/hw/cxgb3/Kconfig
@@ -16,12 +16,3 @@ config INFINIBAND_CXGB3
To compile this driver as a module, choose M here: the module
will be called iw_cxgb3.
-
-config INFINIBAND_CXGB3_DEBUG
- bool "Verbose debugging output"
- depends on INFINIBAND_CXGB3
- default n
- ---help---
- This option causes the Chelsio RDMA driver to produce copious
- amounts of debug messages. Select this if you are developing
- the driver or trying to diagnose a problem.
diff --git a/drivers/infiniband/hw/cxgb3/Makefile b/drivers/infiniband/hw/cxgb3/Makefile
index 2c66d35d19bd..66fe0917aba0 100644
--- a/drivers/infiniband/hw/cxgb3/Makefile
+++ b/drivers/infiniband/hw/cxgb3/Makefile
@@ -5,5 +5,3 @@ obj-$(CONFIG_INFINIBAND_CXGB3) += iw_cxgb3.o
iw_cxgb3-y := iwch_cm.o iwch_ev.o iwch_cq.o iwch_qp.o iwch_mem.o \
iwch_provider.o iwch.o cxio_hal.o cxio_resource.o
-
-ccflags-$(CONFIG_INFINIBAND_CXGB3_DEBUG) += -DDEBUG
diff --git a/drivers/infiniband/hw/cxgb3/cxio_dbg.c b/drivers/infiniband/hw/cxgb3/cxio_dbg.c
deleted file mode 100644
index 97dbe728520a..000000000000
--- a/drivers/infiniband/hw/cxgb3/cxio_dbg.c
+++ /dev/null
@@ -1,206 +0,0 @@
-/*
- * Copyright (c) 2006 Chelsio, Inc. All rights reserved.
- *
- * This software is available to you under a choice of one of two
- * licenses. You may choose to be licensed under the terms of the GNU
- * General Public License (GPL) Version 2, available from the file
- * COPYING in the main directory of this source tree, or the
- * OpenIB.org BSD license below:
- *
- * Redistribution and use in source and binary forms, with or
- * without modification, are permitted provided that the following
- * conditions are met:
- *
- * - Redistributions of source code must retain the above
- * copyright notice, this list of conditions and the following
- * disclaimer.
- *
- * - Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following
- * disclaimer in the documentation and/or other materials
- * provided with the distribution.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
- * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
- * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-#ifdef DEBUG
-#include <linux/types.h>
-#include <linux/slab.h>
-#include "common.h"
-#include "cxgb3_ioctl.h"
-#include "cxio_hal.h"
-#include "cxio_wr.h"
-
-void cxio_dump_tpt(struct cxio_rdev *rdev, u32 stag)
-{
- struct ch_mem_range *m;
- u64 *data;
- int rc;
- int size = 32;
-
- m = kmalloc(sizeof(*m) + size, GFP_ATOMIC);
- if (!m)
- return;
-
- m->mem_id = MEM_PMRX;
- m->addr = (stag>>8) * 32 + rdev->rnic_info.tpt_base;
- m->len = size;
- pr_debug("%s TPT addr 0x%x len %d\n", __func__, m->addr, m->len);
- rc = rdev->t3cdev_p->ctl(rdev->t3cdev_p, RDMA_GET_MEM, m);
- if (rc) {
- pr_debug("%s toectl returned error %d\n", __func__, rc);
- kfree(m);
- return;
- }
-
- data = (u64 *)m->buf;
- while (size > 0) {
- pr_debug("TPT %08x: %016llx\n",
- m->addr, (unsigned long long)*data);
- size -= 8;
- data++;
- m->addr += 8;
- }
- kfree(m);
-}
-
-void cxio_dump_pbl(struct cxio_rdev *rdev, u32 pbl_addr, uint len, u8 shift)
-{
- struct ch_mem_range *m;
- u64 *data;
- int rc;
- int size, npages;
-
- shift += 12;
- npages = (len + (1ULL << shift) - 1) >> shift;
- size = npages * sizeof(u64);
-
- m = kmalloc(sizeof(*m) + size, GFP_ATOMIC);
- if (!m)
- return;
-
- m->mem_id = MEM_PMRX;
- m->addr = pbl_addr;
- m->len = size;
- pr_debug("%s PBL addr 0x%x len %d depth %d\n",
- __func__, m->addr, m->len, npages);
- rc = rdev->t3cdev_p->ctl(rdev->t3cdev_p, RDMA_GET_MEM, m);
- if (rc) {
- pr_debug("%s toectl returned error %d\n", __func__, rc);
- kfree(m);
- return;
- }
-
- data = (u64 *)m->buf;
- while (size > 0) {
- pr_debug("PBL %08x: %016llx\n",
- m->addr, (unsigned long long)*data);
- size -= 8;
- data++;
- m->addr += 8;
- }
- kfree(m);
-}
-
-void cxio_dump_wqe(union t3_wr *wqe)
-{
- __be64 *data = (__be64 *)wqe;
- uint size = (uint)(be64_to_cpu(*data) & 0xff);
-
- if (size == 0)
- size = 8;
- while (size > 0) {
- pr_debug("WQE %p: %016llx\n",
- data, (unsigned long long)be64_to_cpu(*data));
- size--;
- data++;
- }
-}
-
-void cxio_dump_wce(struct t3_cqe *wce)
-{
- __be64 *data = (__be64 *)wce;
- int size = sizeof(*wce);
-
- while (size > 0) {
- pr_debug("WCE %p: %016llx\n",
- data, (unsigned long long)be64_to_cpu(*data));
- size -= 8;
- data++;
- }
-}
-
-void cxio_dump_rqt(struct cxio_rdev *rdev, u32 hwtid, int nents)
-{
- struct ch_mem_range *m;
- int size = nents * 64;
- u64 *data;
- int rc;
-
- m = kmalloc(sizeof(*m) + size, GFP_ATOMIC);
- if (!m)
- return;
-
- m->mem_id = MEM_PMRX;
- m->addr = ((hwtid)<<10) + rdev->rnic_info.rqt_base;
- m->len = size;
- pr_debug("%s RQT addr 0x%x len %d\n", __func__, m->addr, m->len);
- rc = rdev->t3cdev_p->ctl(rdev->t3cdev_p, RDMA_GET_MEM, m);
- if (rc) {
- pr_debug("%s toectl returned error %d\n", __func__, rc);
- kfree(m);
- return;
- }
-
- data = (u64 *)m->buf;
- while (size > 0) {
- pr_debug("RQT %08x: %016llx\n",
- m->addr, (unsigned long long)*data);
- size -= 8;
- data++;
- m->addr += 8;
- }
- kfree(m);
-}
-
-void cxio_dump_tcb(struct cxio_rdev *rdev, u32 hwtid)
-{
- struct ch_mem_range *m;
- int size = TCB_SIZE;
- u32 *data;
- int rc;
-
- m = kmalloc(sizeof(*m) + size, GFP_ATOMIC);
- if (!m)
- return;
-
- m->mem_id = MEM_CM;
- m->addr = hwtid * size;
- m->len = size;
- pr_debug("%s TCB %d len %d\n", __func__, m->addr, m->len);
- rc = rdev->t3cdev_p->ctl(rdev->t3cdev_p, RDMA_GET_MEM, m);
- if (rc) {
- pr_debug("%s toectl returned error %d\n", __func__, rc);
- kfree(m);
- return;
- }
-
- data = (u32 *)m->buf;
- while (size > 0) {
- printk("%2u: %08x %08x %08x %08x %08x %08x %08x %08x\n",
- m->addr,
- *(data+2), *(data+3), *(data),*(data+1),
- *(data+6), *(data+7), *(data+4), *(data+5));
- size -= 32;
- data += 8;
- m->addr += 32;
- }
- kfree(m);
-}
-#endif
diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.h b/drivers/infiniband/hw/cxgb3/cxio_hal.h
index 7e70c5492262..c64e50b5a548 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_hal.h
+++ b/drivers/infiniband/hw/cxgb3/cxio_hal.h
@@ -202,13 +202,4 @@ int iwch_cxgb3_ofld_send(struct t3cdev *tdev, struct sk_buff *skb);
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
-#ifdef DEBUG
-void cxio_dump_tpt(struct cxio_rdev *rev, u32 stag);
-void cxio_dump_pbl(struct cxio_rdev *rev, u32 pbl_addr, uint len, u8 shift);
-void cxio_dump_wqe(union t3_wr *wqe);
-void cxio_dump_wce(struct t3_cqe *wce);
-void cxio_dump_rqt(struct cxio_rdev *rdev, u32 hwtid, int nents);
-void cxio_dump_tcb(struct cxio_rdev *rdev, u32 hwtid);
-#endif
-
#endif
diff --git a/drivers/infiniband/hw/cxgb3/iwch_cq.c b/drivers/infiniband/hw/cxgb3/iwch_cq.c
index dd5348e48806..0a8542c20804 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_cq.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_cq.c
@@ -200,9 +200,6 @@ int iwch_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc)
spin_lock_irqsave(&chp->lock, flags);
for (npolled = 0; npolled < num_entries; ++npolled) {
-#ifdef DEBUG
- int i=0;
-#endif
/*
* Because T3 can post CQEs that are _not_ associated
@@ -211,9 +208,6 @@ int iwch_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc)
*/
do {
err = iwch_poll_cq_one(rhp, chp, wc + npolled);
-#ifdef DEBUG
- BUG_ON(++i > 1000);
-#endif
} while (err == -EAGAIN);
if (err <= 0)
break;
diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index a578ca559e11..be097c6723c0 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -440,7 +440,9 @@ static struct ib_pd *iwch_allocate_pd(struct ib_device *ibdev,
php->pdid = pdid;
php->rhp = rhp;
if (context) {
- if (ib_copy_to_udata(udata, &php->pdid, sizeof (__u32))) {
+ struct iwch_alloc_pd_resp resp = {.pdid = php->pdid};
+
+ if (ib_copy_to_udata(udata, &resp, sizeof(resp))) {
iwch_deallocate_pd(&php->ibpd);
return ERR_PTR(-EFAULT);
}
@@ -1439,6 +1441,7 @@ int iwch_register_device(struct iwch_dev *dev)
memcpy(dev->ibdev.iwcm->ifname, dev->rdev.t3cdev_p->lldev->name,
sizeof(dev->ibdev.iwcm->ifname));
+ dev->ibdev.driver_id = RDMA_DRIVER_CXGB3;
ret = ib_register_device(&dev->ibdev, NULL);
if (ret)
goto bail1;
diff --git a/drivers/infiniband/hw/cxgb4/device.c b/drivers/infiniband/hw/cxgb4/device.c
index e96771ddc9a7..feeb8ee6f4a2 100644
--- a/drivers/infiniband/hw/cxgb4/device.c
+++ b/drivers/infiniband/hw/cxgb4/device.c
@@ -220,14 +220,14 @@ static void set_ep_sin_addrs(struct c4iw_ep *ep,
{
struct iw_cm_id *id = ep->com.cm_id;
- *lsin = (struct sockaddr_in *)&ep->com.local_addr;
- *rsin = (struct sockaddr_in *)&ep->com.remote_addr;
+ *m_lsin = (struct sockaddr_in *)&ep->com.local_addr;
+ *m_rsin = (struct sockaddr_in *)&ep->com.remote_addr;
if (id) {
- *m_lsin = (struct sockaddr_in *)&id->m_local_addr;
- *m_rsin = (struct sockaddr_in *)&id->m_remote_addr;
+ *lsin = (struct sockaddr_in *)&id->local_addr;
+ *rsin = (struct sockaddr_in *)&id->remote_addr;
} else {
- *m_lsin = &zero_sin;
- *m_rsin = &zero_sin;
+ *lsin = &zero_sin;
+ *rsin = &zero_sin;
}
}
@@ -239,14 +239,14 @@ static void set_ep_sin6_addrs(struct c4iw_ep *ep,
{
struct iw_cm_id *id = ep->com.cm_id;
- *lsin6 = (struct sockaddr_in6 *)&ep->com.local_addr;
- *rsin6 = (struct sockaddr_in6 *)&ep->com.remote_addr;
+ *m_lsin6 = (struct sockaddr_in6 *)&ep->com.local_addr;
+ *m_rsin6 = (struct sockaddr_in6 *)&ep->com.remote_addr;
if (id) {
- *m_lsin6 = (struct sockaddr_in6 *)&id->m_local_addr;
- *m_rsin6 = (struct sockaddr_in6 *)&id->m_remote_addr;
+ *lsin6 = (struct sockaddr_in6 *)&id->local_addr;
+ *rsin6 = (struct sockaddr_in6 *)&id->remote_addr;
} else {
- *m_lsin6 = &zero_sin6;
- *m_rsin6 = &zero_sin6;
+ *lsin6 = &zero_sin6;
+ *rsin6 = &zero_sin6;
}
}
diff --git a/drivers/infiniband/hw/cxgb4/mem.c b/drivers/infiniband/hw/cxgb4/mem.c
index 7e0eb201cc26..e90f2fd8dc16 100644
--- a/drivers/infiniband/hw/cxgb4/mem.c
+++ b/drivers/infiniband/hw/cxgb4/mem.c
@@ -391,6 +391,9 @@ static int finish_mem_reg(struct c4iw_mr *mhp, u32 stag)
mhp->attr.stag = stag;
mmid = stag >> 8;
mhp->ibmr.rkey = mhp->ibmr.lkey = stag;
+ mhp->ibmr.length = mhp->attr.len;
+ mhp->ibmr.iova = mhp->attr.va_fbo;
+ mhp->ibmr.page_size = 1U << (mhp->attr.page_size + 12);
pr_debug("mmid 0x%x mhp %p\n", mmid, mhp);
return insert_handle(mhp->rhp, &mhp->rhp->mmidr, mhp, mmid);
}
diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c
index 1b5c6cd2ac4d..0b9cc73c3ded 100644
--- a/drivers/infiniband/hw/cxgb4/provider.c
+++ b/drivers/infiniband/hw/cxgb4/provider.c
@@ -281,7 +281,9 @@ static struct ib_pd *c4iw_allocate_pd(struct ib_device *ibdev,
php->pdid = pdid;
php->rhp = rhp;
if (context) {
- if (ib_copy_to_udata(udata, &php->pdid, sizeof(u32))) {
+ struct c4iw_alloc_pd_resp uresp = {.pdid = php->pdid};
+
+ if (ib_copy_to_udata(udata, &uresp, sizeof(uresp))) {
c4iw_deallocate_pd(&php->ibpd);
return ERR_PTR(-EFAULT);
}
@@ -531,6 +533,24 @@ static void get_dev_fw_str(struct ib_device *dev, char *str)
FW_HDR_FW_VER_BUILD_G(c4iw_dev->rdev.lldi.fw_vers));
}
+static struct net_device *get_netdev(struct ib_device *dev, u8 port)
+{
+ struct c4iw_dev *c4iw_dev = container_of(dev, struct c4iw_dev, ibdev);
+ struct c4iw_rdev *rdev = &c4iw_dev->rdev;
+ struct net_device *ndev;
+
+ if (!port || port > rdev->lldi.nports)
+ return NULL;
+
+ rcu_read_lock();
+ ndev = rdev->lldi.ports[port - 1];
+ if (ndev)
+ dev_hold(ndev);
+ rcu_read_unlock();
+
+ return ndev;
+}
+
void c4iw_register_device(struct work_struct *work)
{
int ret;
@@ -609,6 +629,7 @@ void c4iw_register_device(struct work_struct *work)
dev->ibdev.uverbs_abi_ver = C4IW_UVERBS_ABI_VERSION;
dev->ibdev.get_port_immutable = c4iw_port_immutable;
dev->ibdev.get_dev_fw_str = get_dev_fw_str;
+ dev->ibdev.get_netdev = get_netdev;
dev->ibdev.iwcm = kmalloc(sizeof(struct iw_cm_verbs), GFP_KERNEL);
if (!dev->ibdev.iwcm) {
@@ -627,6 +648,7 @@ void c4iw_register_device(struct work_struct *work)
memcpy(dev->ibdev.iwcm->ifname, dev->rdev.lldi.ports[0]->name,
sizeof(dev->ibdev.iwcm->ifname));
+ dev->ibdev.driver_id = RDMA_DRIVER_CXGB4;
ret = ib_register_device(&dev->ibdev, NULL);
if (ret)
goto err_kfree_iwcm;
diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c
index addc68e83606..46d1475b2154 100644
--- a/drivers/infiniband/hw/hfi1/driver.c
+++ b/drivers/infiniband/hw/hfi1/driver.c
@@ -390,6 +390,7 @@ static void rcv_hdrerr(struct hfi1_ctxtdata *rcd, struct hfi1_pportdata *ppd,
svc_type = IB_CC_SVCTYPE_UC;
break;
default:
+ rcu_read_unlock();
goto drop;
}
diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h
index 90bc8c76d2ca..32c48265405e 100644
--- a/drivers/infiniband/hw/hfi1/hfi.h
+++ b/drivers/infiniband/hw/hfi1/hfi.h
@@ -70,7 +70,6 @@
#include <linux/rhashtable.h>
#include <linux/netdevice.h>
#include <rdma/rdma_vt.h>
-#include <rdma/opa_addr.h>
#include "chip_registers.h"
#include "common.h"
diff --git a/drivers/infiniband/hw/hfi1/qp.c b/drivers/infiniband/hw/hfi1/qp.c
index d30dd1a5b0a6..1697d96151bd 100644
--- a/drivers/infiniband/hw/hfi1/qp.c
+++ b/drivers/infiniband/hw/hfi1/qp.c
@@ -481,7 +481,6 @@ static void iowait_sdma_drained(struct iowait *wait)
}
/**
- *
* qp_to_sdma_engine - map a qp to a send engine
* @qp: the QP
* @sc5: the 5 bit sc
diff --git a/drivers/infiniband/hw/hfi1/user_exp_rcv.c b/drivers/infiniband/hw/hfi1/user_exp_rcv.c
index c1c596adcd01..0d5330b7353d 100644
--- a/drivers/infiniband/hw/hfi1/user_exp_rcv.c
+++ b/drivers/infiniband/hw/hfi1/user_exp_rcv.c
@@ -473,7 +473,7 @@ nomem:
tinfo->tidcnt = tididx;
tinfo->length = mapped_pages * PAGE_SIZE;
- if (copy_to_user((void __user *)(unsigned long)tinfo->tidlist,
+ if (copy_to_user(u64_to_user_ptr(tinfo->tidlist),
tidlist, sizeof(tidlist[0]) * tididx)) {
/*
* On failure to copy to the user level, we need to undo
@@ -513,7 +513,7 @@ int hfi1_user_exp_rcv_clear(struct hfi1_filedata *fd,
if (unlikely(tinfo->tidcnt > fd->tid_used))
return -EINVAL;
- tidinfo = memdup_user((void __user *)(unsigned long)tinfo->tidlist,
+ tidinfo = memdup_user(u64_to_user_ptr(tinfo->tidlist),
sizeof(tidinfo[0]) * tinfo->tidcnt);
if (IS_ERR(tidinfo))
return PTR_ERR(tidinfo);
diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c
index 471d55c50066..c8cf4d4984d3 100644
--- a/drivers/infiniband/hw/hfi1/verbs.c
+++ b/drivers/infiniband/hw/hfi1/verbs.c
@@ -1960,7 +1960,7 @@ int hfi1_register_ib_device(struct hfi1_devdata *dd)
i,
ppd->pkeys);
- ret = rvt_register_device(&dd->verbs_dev.rdi);
+ ret = rvt_register_device(&dd->verbs_dev.rdi, RDMA_DRIVER_HFI1);
if (ret)
goto err_verbs_txreq;
diff --git a/drivers/infiniband/hw/hns/Makefile b/drivers/infiniband/hw/hns/Makefile
index 97bf2cd1cacb..cf03404b9d58 100644
--- a/drivers/infiniband/hw/hns/Makefile
+++ b/drivers/infiniband/hw/hns/Makefile
@@ -7,7 +7,7 @@ ccflags-y := -Idrivers/net/ethernet/hisilicon/hns3
obj-$(CONFIG_INFINIBAND_HNS) += hns-roce.o
hns-roce-objs := hns_roce_main.o hns_roce_cmd.o hns_roce_pd.o \
hns_roce_ah.o hns_roce_hem.o hns_roce_mr.o hns_roce_qp.o \
- hns_roce_cq.o hns_roce_alloc.o
+ hns_roce_cq.o hns_roce_alloc.o hns_roce_db.o
obj-$(CONFIG_INFINIBAND_HNS_HIP06) += hns-roce-hw-v1.o
hns-roce-hw-v1-objs := hns_roce_hw_v1.o
obj-$(CONFIG_INFINIBAND_HNS_HIP08) += hns-roce-hw-v2.o
diff --git a/drivers/infiniband/hw/hns/hns_roce_ah.c b/drivers/infiniband/hw/hns/hns_roce_ah.c
index 7dd6a66ea244..d74928621559 100644
--- a/drivers/infiniband/hw/hns/hns_roce_ah.c
+++ b/drivers/infiniband/hw/hns/hns_roce_ah.c
@@ -68,11 +68,9 @@ struct ib_ah *hns_roce_create_ah(struct ib_pd *ibpd,
return ERR_PTR(ret);
}
- if (gid_attr.ndev) {
- if (is_vlan_dev(gid_attr.ndev))
- vlan_tag = vlan_dev_vlan_id(gid_attr.ndev);
- dev_put(gid_attr.ndev);
- }
+ if (is_vlan_dev(gid_attr.ndev))
+ vlan_tag = vlan_dev_vlan_id(gid_attr.ndev);
+ dev_put(gid_attr.ndev);
if (vlan_tag < 0x1000)
vlan_tag |= (rdma_ah_get_sl(ah_attr) &
diff --git a/drivers/infiniband/hw/hns/hns_roce_cq.c b/drivers/infiniband/hw/hns/hns_roce_cq.c
index bccc9b54c9ce..14734d0d0b76 100644
--- a/drivers/infiniband/hw/hns/hns_roce_cq.c
+++ b/drivers/infiniband/hw/hns/hns_roce_cq.c
@@ -315,6 +315,7 @@ struct ib_cq *hns_roce_ib_create_cq(struct ib_device *ib_dev,
struct hns_roce_dev *hr_dev = to_hr_dev(ib_dev);
struct device *dev = hr_dev->dev;
struct hns_roce_ib_create_cq ucmd;
+ struct hns_roce_ib_create_cq_resp resp = {};
struct hns_roce_cq *hr_cq = NULL;
struct hns_roce_uar *uar = NULL;
int vector = attr->comp_vector;
@@ -354,15 +355,36 @@ struct ib_cq *hns_roce_ib_create_cq(struct ib_device *ib_dev,
goto err_cq;
}
+ if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) &&
+ (udata->outlen >= sizeof(resp))) {
+ ret = hns_roce_db_map_user(to_hr_ucontext(context),
+ ucmd.db_addr, &hr_cq->db);
+ if (ret) {
+ dev_err(dev, "cq record doorbell map failed!\n");
+ goto err_mtt;
+ }
+ hr_cq->db_en = 1;
+ resp.cap_flags |= HNS_ROCE_SUPPORT_CQ_RECORD_DB;
+ }
+
/* Get user space parameters */
uar = &to_hr_ucontext(context)->uar;
} else {
+ if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) {
+ ret = hns_roce_alloc_db(hr_dev, &hr_cq->db, 1);
+ if (ret)
+ goto err_cq;
+
+ hr_cq->set_ci_db = hr_cq->db.db_record;
+ *hr_cq->set_ci_db = 0;
+ }
+
/* Init mmt table and write buff address to mtt table */
ret = hns_roce_ib_alloc_cq_buf(hr_dev, &hr_cq->hr_buf,
cq_entries);
if (ret) {
dev_err(dev, "Failed to alloc_cq_buf.\n");
- goto err_cq;
+ goto err_db;
}
uar = &hr_dev->priv_uar;
@@ -375,7 +397,7 @@ struct ib_cq *hns_roce_ib_create_cq(struct ib_device *ib_dev,
hr_cq, vector);
if (ret) {
dev_err(dev, "Creat CQ .Failed to cq_alloc.\n");
- goto err_mtt;
+ goto err_dbmap;
}
/*
@@ -393,10 +415,10 @@ struct ib_cq *hns_roce_ib_create_cq(struct ib_device *ib_dev,
hr_cq->cq_depth = cq_entries;
if (context) {
- if (ib_copy_to_udata(udata, &hr_cq->cqn, sizeof(u64))) {
- ret = -EFAULT;
+ resp.cqn = hr_cq->cqn;
+ ret = ib_copy_to_udata(udata, &resp, sizeof(resp));
+ if (ret)
goto err_cqc;
- }
}
return &hr_cq->ib_cq;
@@ -404,6 +426,12 @@ struct ib_cq *hns_roce_ib_create_cq(struct ib_device *ib_dev,
err_cqc:
hns_roce_free_cq(hr_dev, hr_cq);
+err_dbmap:
+ if (context && (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) &&
+ (udata->outlen >= sizeof(resp)))
+ hns_roce_db_unmap_user(to_hr_ucontext(context),
+ &hr_cq->db);
+
err_mtt:
hns_roce_mtt_cleanup(hr_dev, &hr_cq->hr_buf.hr_mtt);
if (context)
@@ -412,6 +440,10 @@ err_mtt:
hns_roce_ib_free_cq_buf(hr_dev, &hr_cq->hr_buf,
hr_cq->ib_cq.cqe);
+err_db:
+ if (!context && (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB))
+ hns_roce_free_db(hr_dev, &hr_cq->db);
+
err_cq:
kfree(hr_cq);
return ERR_PTR(ret);
@@ -430,12 +462,20 @@ int hns_roce_ib_destroy_cq(struct ib_cq *ib_cq)
hns_roce_free_cq(hr_dev, hr_cq);
hns_roce_mtt_cleanup(hr_dev, &hr_cq->hr_buf.hr_mtt);
- if (ib_cq->uobject)
+ if (ib_cq->uobject) {
ib_umem_release(hr_cq->umem);
- else
+
+ if (hr_cq->db_en == 1)
+ hns_roce_db_unmap_user(
+ to_hr_ucontext(ib_cq->uobject->context),
+ &hr_cq->db);
+ } else {
/* Free the buff of stored cq */
hns_roce_ib_free_cq_buf(hr_dev, &hr_cq->hr_buf,
ib_cq->cqe);
+ if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB)
+ hns_roce_free_db(hr_dev, &hr_cq->db);
+ }
kfree(hr_cq);
}
diff --git a/drivers/infiniband/hw/hns/hns_roce_db.c b/drivers/infiniband/hw/hns/hns_roce_db.c
new file mode 100644
index 000000000000..ebee2782a573
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_db.c
@@ -0,0 +1,180 @@
+/* SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause) */
+/*
+ * Copyright (c) 2017 Hisilicon Limited.
+ * Copyright (c) 2007, 2008 Mellanox Technologies. All rights reserved.
+ */
+
+#include <linux/platform_device.h>
+#include <rdma/ib_umem.h>
+#include "hns_roce_device.h"
+
+int hns_roce_db_map_user(struct hns_roce_ucontext *context, unsigned long virt,
+ struct hns_roce_db *db)
+{
+ struct hns_roce_user_db_page *page;
+ int ret = 0;
+
+ mutex_lock(&context->page_mutex);
+
+ list_for_each_entry(page, &context->page_list, list)
+ if (page->user_virt == (virt & PAGE_MASK))
+ goto found;
+
+ page = kmalloc(sizeof(*page), GFP_KERNEL);
+ if (!page) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ refcount_set(&page->refcount, 1);
+ page->user_virt = (virt & PAGE_MASK);
+ page->umem = ib_umem_get(&context->ibucontext, virt & PAGE_MASK,
+ PAGE_SIZE, 0, 0);
+ if (IS_ERR(page->umem)) {
+ ret = PTR_ERR(page->umem);
+ kfree(page);
+ goto out;
+ }
+
+ list_add(&page->list, &context->page_list);
+
+found:
+ db->dma = sg_dma_address(page->umem->sg_head.sgl) +
+ (virt & ~PAGE_MASK);
+ db->u.user_page = page;
+ refcount_inc(&page->refcount);
+
+out:
+ mutex_unlock(&context->page_mutex);
+
+ return ret;
+}
+EXPORT_SYMBOL(hns_roce_db_map_user);
+
+void hns_roce_db_unmap_user(struct hns_roce_ucontext *context,
+ struct hns_roce_db *db)
+{
+ mutex_lock(&context->page_mutex);
+
+ refcount_dec(&db->u.user_page->refcount);
+ if (refcount_dec_if_one(&db->u.user_page->refcount)) {
+ list_del(&db->u.user_page->list);
+ ib_umem_release(db->u.user_page->umem);
+ kfree(db->u.user_page);
+ }
+
+ mutex_unlock(&context->page_mutex);
+}
+EXPORT_SYMBOL(hns_roce_db_unmap_user);
+
+static struct hns_roce_db_pgdir *hns_roce_alloc_db_pgdir(
+ struct device *dma_device)
+{
+ struct hns_roce_db_pgdir *pgdir;
+
+ pgdir = kzalloc(sizeof(*pgdir), GFP_KERNEL);
+ if (!pgdir)
+ return NULL;
+
+ bitmap_fill(pgdir->order1, HNS_ROCE_DB_PER_PAGE / 2);
+ pgdir->bits[0] = pgdir->order0;
+ pgdir->bits[1] = pgdir->order1;
+ pgdir->page = dma_alloc_coherent(dma_device, PAGE_SIZE,
+ &pgdir->db_dma, GFP_KERNEL);
+ if (!pgdir->page) {
+ kfree(pgdir);
+ return NULL;
+ }
+
+ return pgdir;
+}
+
+static int hns_roce_alloc_db_from_pgdir(struct hns_roce_db_pgdir *pgdir,
+ struct hns_roce_db *db, int order)
+{
+ int o;
+ int i;
+
+ for (o = order; o <= 1; ++o) {
+ i = find_first_bit(pgdir->bits[o], HNS_ROCE_DB_PER_PAGE >> o);
+ if (i < HNS_ROCE_DB_PER_PAGE >> o)
+ goto found;
+ }
+
+ return -ENOMEM;
+
+found:
+ clear_bit(i, pgdir->bits[o]);
+
+ i <<= o;
+
+ if (o > order)
+ set_bit(i ^ 1, pgdir->bits[order]);
+
+ db->u.pgdir = pgdir;
+ db->index = i;
+ db->db_record = pgdir->page + db->index;
+ db->dma = pgdir->db_dma + db->index * 4;
+ db->order = order;
+
+ return 0;
+}
+
+int hns_roce_alloc_db(struct hns_roce_dev *hr_dev, struct hns_roce_db *db,
+ int order)
+{
+ struct hns_roce_db_pgdir *pgdir;
+ int ret = 0;
+
+ mutex_lock(&hr_dev->pgdir_mutex);
+
+ list_for_each_entry(pgdir, &hr_dev->pgdir_list, list)
+ if (!hns_roce_alloc_db_from_pgdir(pgdir, db, order))
+ goto out;
+
+ pgdir = hns_roce_alloc_db_pgdir(hr_dev->dev);
+ if (!pgdir) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ list_add(&pgdir->list, &hr_dev->pgdir_list);
+
+ /* This should never fail -- we just allocated an empty page: */
+ WARN_ON(hns_roce_alloc_db_from_pgdir(pgdir, db, order));
+
+out:
+ mutex_unlock(&hr_dev->pgdir_mutex);
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(hns_roce_alloc_db);
+
+void hns_roce_free_db(struct hns_roce_dev *hr_dev, struct hns_roce_db *db)
+{
+ int o;
+ int i;
+
+ mutex_lock(&hr_dev->pgdir_mutex);
+
+ o = db->order;
+ i = db->index;
+
+ if (db->order == 0 && test_bit(i ^ 1, db->u.pgdir->order0)) {
+ clear_bit(i ^ 1, db->u.pgdir->order0);
+ ++o;
+ }
+
+ i >>= o;
+ set_bit(i, db->u.pgdir->bits[o]);
+
+ if (bitmap_full(db->u.pgdir->order1, HNS_ROCE_DB_PER_PAGE / 2)) {
+ dma_free_coherent(hr_dev->dev, PAGE_SIZE, db->u.pgdir->page,
+ db->u.pgdir->db_dma);
+ list_del(&db->u.pgdir->list);
+ kfree(db->u.pgdir);
+ }
+
+ mutex_unlock(&hr_dev->pgdir_mutex);
+}
+EXPORT_SYMBOL_GPL(hns_roce_free_db);
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 165a09b314f6..fb305b7f99a8 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -105,6 +105,14 @@
#define PAGES_SHIFT_24 24
#define PAGES_SHIFT_32 32
+enum {
+ HNS_ROCE_SUPPORT_RQ_RECORD_DB = 1 << 0,
+};
+
+enum {
+ HNS_ROCE_SUPPORT_CQ_RECORD_DB = 1 << 0,
+};
+
enum hns_roce_qp_state {
HNS_ROCE_QP_STATE_RST,
HNS_ROCE_QP_STATE_INIT,
@@ -178,7 +186,8 @@ enum {
enum {
HNS_ROCE_CAP_FLAG_REREG_MR = BIT(0),
HNS_ROCE_CAP_FLAG_ROCE_V1_V2 = BIT(1),
- HNS_ROCE_CAP_FLAG_RQ_INLINE = BIT(2)
+ HNS_ROCE_CAP_FLAG_RQ_INLINE = BIT(2),
+ HNS_ROCE_CAP_FLAG_RECORD_DB = BIT(3)
};
enum hns_roce_mtt_type {
@@ -186,6 +195,10 @@ enum hns_roce_mtt_type {
MTT_TYPE_CQE,
};
+enum {
+ HNS_ROCE_DB_PER_PAGE = PAGE_SIZE / 4
+};
+
#define HNS_ROCE_CMD_SUCCESS 1
#define HNS_ROCE_PORT_DOWN 0
@@ -203,6 +216,8 @@ struct hns_roce_uar {
struct hns_roce_ucontext {
struct ib_ucontext ibucontext;
struct hns_roce_uar uar;
+ struct list_head page_list;
+ struct mutex page_mutex;
};
struct hns_roce_pd {
@@ -335,6 +350,33 @@ struct hns_roce_buf {
int page_shift;
};
+struct hns_roce_db_pgdir {
+ struct list_head list;
+ DECLARE_BITMAP(order0, HNS_ROCE_DB_PER_PAGE);
+ DECLARE_BITMAP(order1, HNS_ROCE_DB_PER_PAGE / 2);
+ unsigned long *bits[2];
+ u32 *page;
+ dma_addr_t db_dma;
+};
+
+struct hns_roce_user_db_page {
+ struct list_head list;
+ struct ib_umem *umem;
+ unsigned long user_virt;
+ refcount_t refcount;
+};
+
+struct hns_roce_db {
+ u32 *db_record;
+ union {
+ struct hns_roce_db_pgdir *pgdir;
+ struct hns_roce_user_db_page *user_page;
+ } u;
+ dma_addr_t dma;
+ int index;
+ int order;
+};
+
struct hns_roce_cq_buf {
struct hns_roce_buf hr_buf;
struct hns_roce_mtt hr_mtt;
@@ -343,6 +385,8 @@ struct hns_roce_cq_buf {
struct hns_roce_cq {
struct ib_cq ib_cq;
struct hns_roce_cq_buf hr_buf;
+ struct hns_roce_db db;
+ u8 db_en;
spinlock_t lock;
struct ib_umem *umem;
void (*comp)(struct hns_roce_cq *cq);
@@ -351,6 +395,7 @@ struct hns_roce_cq {
struct hns_roce_uar *uar;
u32 cq_depth;
u32 cons_index;
+ u32 *set_ci_db;
void __iomem *cq_db_l;
u16 *tptr_addr;
int arm_sn;
@@ -466,6 +511,8 @@ struct hns_roce_qp {
struct ib_qp ibqp;
struct hns_roce_buf hr_buf;
struct hns_roce_wq rq;
+ struct hns_roce_db rdb;
+ u8 rdb_en;
u32 doorbell_qpn;
__le32 sq_signal_bits;
u32 sq_next_wqe;
@@ -725,6 +772,8 @@ struct hns_roce_dev {
spinlock_t bt_cmd_lock;
struct hns_roce_ib_iboe iboe;
+ struct list_head pgdir_list;
+ struct mutex pgdir_mutex;
int irq[HNS_ROCE_MAX_IRQ_NUM];
u8 __iomem *reg_base;
struct hns_roce_caps caps;
@@ -930,6 +979,14 @@ struct ib_cq *hns_roce_ib_create_cq(struct ib_device *ib_dev,
int hns_roce_ib_destroy_cq(struct ib_cq *ib_cq);
void hns_roce_free_cq(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq);
+int hns_roce_db_map_user(struct hns_roce_ucontext *context, unsigned long virt,
+ struct hns_roce_db *db);
+void hns_roce_db_unmap_user(struct hns_roce_ucontext *context,
+ struct hns_roce_db *db);
+int hns_roce_alloc_db(struct hns_roce_dev *hr_dev, struct hns_roce_db *db,
+ int order);
+void hns_roce_free_db(struct hns_roce_dev *hr_dev, struct hns_roce_db *db);
+
void hns_roce_cq_completion(struct hns_roce_dev *hr_dev, u32 cqn);
void hns_roce_cq_event(struct hns_roce_dev *hr_dev, u32 cqn, int event_type);
void hns_roce_qp_event(struct hns_roce_dev *hr_dev, u32 qpn, int event_type);
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
index da13bd7c3ca9..47e1b6ac1e1a 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
@@ -1687,13 +1687,13 @@ static int hns_roce_v1_post_mbox(struct hns_roce_dev *hr_dev, u64 in_param,
roce_set_field(val, ROCEE_MB6_ROCEE_MB_TOKEN_M,
ROCEE_MB6_ROCEE_MB_TOKEN_S, token);
- __raw_writeq(cpu_to_le64(in_param), hcr + 0);
- __raw_writeq(cpu_to_le64(out_param), hcr + 2);
- __raw_writel(cpu_to_le32(in_modifier), hcr + 4);
+ writeq(in_param, hcr + 0);
+ writeq(out_param, hcr + 2);
+ writel(in_modifier, hcr + 4);
/* Memory barrier */
wmb();
- __raw_writel(cpu_to_le32(val), hcr + 5);
+ writel(val, hcr + 5);
mmiowb();
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index ec638778661c..8b84ab7800d8 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -498,7 +498,6 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
struct hns_roce_v2_wqe_data_seg *dseg;
struct hns_roce_rinl_sge *sge_list;
struct device *dev = hr_dev->dev;
- struct hns_roce_v2_db rq_db;
unsigned long flags;
void *wqe = NULL;
int ret = 0;
@@ -509,7 +508,7 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
spin_lock_irqsave(&hr_qp->rq.lock, flags);
ind = hr_qp->rq.head & (hr_qp->rq.wqe_cnt - 1);
- if (hr_qp->state == IB_QPS_RESET || hr_qp->state == IB_QPS_ERR) {
+ if (hr_qp->state == IB_QPS_RESET) {
spin_unlock_irqrestore(&hr_qp->rq.lock, flags);
*bad_wr = wr;
return -EINVAL;
@@ -564,17 +563,7 @@ out:
/* Memory barrier */
wmb();
- rq_db.byte_4 = 0;
- rq_db.parameter = 0;
-
- roce_set_field(rq_db.byte_4, V2_DB_BYTE_4_TAG_M,
- V2_DB_BYTE_4_TAG_S, hr_qp->qpn);
- roce_set_field(rq_db.byte_4, V2_DB_BYTE_4_CMD_M,
- V2_DB_BYTE_4_CMD_S, HNS_ROCE_V2_RQ_DB);
- roce_set_field(rq_db.parameter, V2_DB_PARAMETER_CONS_IDX_M,
- V2_DB_PARAMETER_CONS_IDX_S, hr_qp->rq.head);
-
- hns_roce_write64_k((__le32 *)&rq_db, hr_qp->rq.db_reg_l);
+ *hr_qp->rdb.db_record = hr_qp->rq.head & 0xffff;
}
spin_unlock_irqrestore(&hr_qp->rq.lock, flags);
@@ -1168,7 +1157,8 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev)
caps->flags = HNS_ROCE_CAP_FLAG_REREG_MR |
HNS_ROCE_CAP_FLAG_ROCE_V1_V2 |
- HNS_ROCE_CAP_FLAG_RQ_INLINE;
+ HNS_ROCE_CAP_FLAG_RQ_INLINE |
+ HNS_ROCE_CAP_FLAG_RECORD_DB;
caps->pkey_table_len[0] = 1;
caps->gid_table_len[0] = HNS_ROCE_V2_GID_INDEX_NUM;
caps->ceqe_depth = HNS_ROCE_V2_COMP_EQE_NUM;
@@ -1228,14 +1218,14 @@ static int hns_roce_v2_post_mbox(struct hns_roce_dev *hr_dev, u64 in_param,
roce_set_field(val1, HNS_ROCE_VF_MB5_TOKEN_MASK,
HNS_ROCE_VF_MB5_TOKEN_SHIFT, token);
- __raw_writeq(cpu_to_le64(in_param), hcr + 0);
- __raw_writeq(cpu_to_le64(out_param), hcr + 2);
+ writeq(in_param, hcr + 0);
+ writeq(out_param, hcr + 2);
/* Memory barrier */
wmb();
- __raw_writel(cpu_to_le32(val0), hcr + 4);
- __raw_writel(cpu_to_le32(val1), hcr + 5);
+ writel(val0, hcr + 4);
+ writel(val1, hcr + 5);
mmiowb();
@@ -1507,24 +1497,7 @@ static struct hns_roce_v2_cqe *next_cqe_sw_v2(struct hns_roce_cq *hr_cq)
static void hns_roce_v2_cq_set_ci(struct hns_roce_cq *hr_cq, u32 cons_index)
{
- struct hns_roce_v2_cq_db cq_db;
-
- cq_db.byte_4 = 0;
- cq_db.parameter = 0;
-
- roce_set_field(cq_db.byte_4, V2_CQ_DB_BYTE_4_TAG_M,
- V2_CQ_DB_BYTE_4_TAG_S, hr_cq->cqn);
- roce_set_field(cq_db.byte_4, V2_CQ_DB_BYTE_4_CMD_M,
- V2_CQ_DB_BYTE_4_CMD_S, HNS_ROCE_V2_CQ_DB_PTR);
-
- roce_set_field(cq_db.parameter, V2_CQ_DB_PARAMETER_CONS_IDX_M,
- V2_CQ_DB_PARAMETER_CONS_IDX_S,
- cons_index & ((hr_cq->cq_depth << 1) - 1));
- roce_set_field(cq_db.parameter, V2_CQ_DB_PARAMETER_CMD_SN_M,
- V2_CQ_DB_PARAMETER_CMD_SN_S, 1);
-
- hns_roce_write64_k((__be32 *)&cq_db, hr_cq->cq_db_l);
-
+ *hr_cq->set_ci_db = cons_index & 0xffffff;
}
static void __hns_roce_v2_cq_clean(struct hns_roce_cq *hr_cq, u32 qpn,
@@ -1637,6 +1610,16 @@ static void hns_roce_v2_write_cqc(struct hns_roce_dev *hr_dev,
roce_set_field(cq_context->byte_40_cqe_ba, V2_CQC_BYTE_40_CQE_BA_M,
V2_CQC_BYTE_40_CQE_BA_S, (dma_handle >> (32 + 3)));
+ if (hr_cq->db_en)
+ roce_set_bit(cq_context->byte_44_db_record,
+ V2_CQC_BYTE_44_DB_RECORD_EN_S, 1);
+
+ roce_set_field(cq_context->byte_44_db_record,
+ V2_CQC_BYTE_44_DB_RECORD_ADDR_M,
+ V2_CQC_BYTE_44_DB_RECORD_ADDR_S,
+ ((u32)hr_cq->db.dma) >> 1);
+ cq_context->db_record_addr = hr_cq->db.dma >> 32;
+
roce_set_field(cq_context->byte_56_cqe_period_maxcnt,
V2_CQC_BYTE_56_CQ_MAX_CNT_M,
V2_CQC_BYTE_56_CQ_MAX_CNT_S,
@@ -2274,6 +2257,23 @@ static void modify_qp_reset_to_init(struct ib_qp *ibqp,
hr_qp->qkey = attr->qkey;
}
+ if (hr_qp->rdb_en) {
+ roce_set_bit(context->byte_68_rq_db,
+ V2_QPC_BYTE_68_RQ_RECORD_EN_S, 1);
+ roce_set_bit(qpc_mask->byte_68_rq_db,
+ V2_QPC_BYTE_68_RQ_RECORD_EN_S, 0);
+ }
+
+ roce_set_field(context->byte_68_rq_db,
+ V2_QPC_BYTE_68_RQ_DB_RECORD_ADDR_M,
+ V2_QPC_BYTE_68_RQ_DB_RECORD_ADDR_S,
+ ((u32)hr_qp->rdb.dma) >> 1);
+ roce_set_field(qpc_mask->byte_68_rq_db,
+ V2_QPC_BYTE_68_RQ_DB_RECORD_ADDR_M,
+ V2_QPC_BYTE_68_RQ_DB_RECORD_ADDR_S, 0);
+ context->rq_db_record_addr = hr_qp->rdb.dma >> 32;
+ qpc_mask->rq_db_record_addr = 0;
+
roce_set_bit(context->byte_76_srqn_op_en, V2_QPC_BYTE_76_RQIE_S, 1);
roce_set_bit(qpc_mask->byte_76_srqn_op_en, V2_QPC_BYTE_76_RQIE_S, 0);
@@ -3211,6 +3211,8 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp,
hr_qp->sq.tail = 0;
hr_qp->sq_next_wqe = 0;
hr_qp->next_sge = 0;
+ if (hr_qp->rq.wqe_cnt)
+ *hr_qp->rdb.db_record = 0;
}
out:
@@ -3437,11 +3439,17 @@ static int hns_roce_v2_destroy_qp_common(struct hns_roce_dev *hr_dev,
hns_roce_mtt_cleanup(hr_dev, &hr_qp->mtt);
if (is_user) {
+ if (hr_qp->rq.wqe_cnt && (hr_qp->rdb_en == 1))
+ hns_roce_db_unmap_user(
+ to_hr_ucontext(hr_qp->ibqp.uobject->context),
+ &hr_qp->rdb);
ib_umem_release(hr_qp->umem);
} else {
kfree(hr_qp->sq.wrid);
kfree(hr_qp->rq.wrid);
hns_roce_buf_free(hr_dev, hr_qp->buff_size, &hr_qp->hr_buf);
+ if (hr_qp->rq.wqe_cnt)
+ hns_roce_free_db(hr_dev, &hr_qp->rdb);
}
if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) {
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
index 2bf8a47e3de3..182b6726f783 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
@@ -299,6 +299,9 @@ struct hns_roce_v2_cq_context {
#define V2_CQC_BYTE_44_DB_RECORD_EN_S 0
+#define V2_CQC_BYTE_44_DB_RECORD_ADDR_S 1
+#define V2_CQC_BYTE_44_DB_RECORD_ADDR_M GENMASK(31, 1)
+
#define V2_CQC_BYTE_52_CQE_CNT_S 0
#define V2_CQC_BYTE_52_CQE_CNT_M GENMASK(23, 0)
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index eb9a69fc7bec..9d48bc07a9e6 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -74,12 +74,11 @@ static int hns_roce_set_mac(struct hns_roce_dev *hr_dev, u8 port, u8 *addr)
return hr_dev->hw->set_mac(hr_dev, phy_port, addr);
}
-static int hns_roce_add_gid(struct ib_device *device, u8 port_num,
- unsigned int index, const union ib_gid *gid,
+static int hns_roce_add_gid(const union ib_gid *gid,
const struct ib_gid_attr *attr, void **context)
{
- struct hns_roce_dev *hr_dev = to_hr_dev(device);
- u8 port = port_num - 1;
+ struct hns_roce_dev *hr_dev = to_hr_dev(attr->device);
+ u8 port = attr->port_num - 1;
unsigned long flags;
int ret;
@@ -88,20 +87,20 @@ static int hns_roce_add_gid(struct ib_device *device, u8 port_num,
spin_lock_irqsave(&hr_dev->iboe.lock, flags);
- ret = hr_dev->hw->set_gid(hr_dev, port, index, (union ib_gid *)gid,
- attr);
+ ret = hr_dev->hw->set_gid(hr_dev, port, attr->index,
+ (union ib_gid *)gid, attr);
spin_unlock_irqrestore(&hr_dev->iboe.lock, flags);
return ret;
}
-static int hns_roce_del_gid(struct ib_device *device, u8 port_num,
- unsigned int index, void **context)
+static int hns_roce_del_gid(const struct ib_gid_attr *attr, void **context)
{
- struct hns_roce_dev *hr_dev = to_hr_dev(device);
+ struct hns_roce_dev *hr_dev = to_hr_dev(attr->device);
+ struct ib_gid_attr zattr = { };
union ib_gid zgid = { {0} };
- u8 port = port_num - 1;
+ u8 port = attr->port_num - 1;
unsigned long flags;
int ret;
@@ -110,7 +109,7 @@ static int hns_roce_del_gid(struct ib_device *device, u8 port_num,
spin_lock_irqsave(&hr_dev->iboe.lock, flags);
- ret = hr_dev->hw->set_gid(hr_dev, port, index, &zgid, NULL);
+ ret = hr_dev->hw->set_gid(hr_dev, port, attr->index, &zgid, &zattr);
spin_unlock_irqrestore(&hr_dev->iboe.lock, flags);
@@ -295,12 +294,6 @@ static enum rdma_link_layer hns_roce_get_link_layer(struct ib_device *device,
return IB_LINK_LAYER_ETHERNET;
}
-static int hns_roce_query_gid(struct ib_device *ib_dev, u8 port_num, int index,
- union ib_gid *gid)
-{
- return 0;
-}
-
static int hns_roce_query_pkey(struct ib_device *ib_dev, u8 port, u16 index,
u16 *pkey)
{
@@ -337,7 +330,7 @@ static struct ib_ucontext *hns_roce_alloc_ucontext(struct ib_device *ib_dev,
{
int ret = 0;
struct hns_roce_ucontext *context;
- struct hns_roce_ib_alloc_ucontext_resp resp;
+ struct hns_roce_ib_alloc_ucontext_resp resp = {};
struct hns_roce_dev *hr_dev = to_hr_dev(ib_dev);
resp.qp_tab_size = hr_dev->caps.num_qps;
@@ -350,6 +343,11 @@ static struct ib_ucontext *hns_roce_alloc_ucontext(struct ib_device *ib_dev,
if (ret)
goto error_fail_uar_alloc;
+ if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) {
+ INIT_LIST_HEAD(&context->page_list);
+ mutex_init(&context->page_mutex);
+ }
+
ret = ib_copy_to_udata(udata, &resp, sizeof(resp));
if (ret)
goto error_fail_copy_to_udata;
@@ -476,7 +474,6 @@ static int hns_roce_register_device(struct hns_roce_dev *hr_dev)
ib_dev->modify_port = hns_roce_modify_port;
ib_dev->get_link_layer = hns_roce_get_link_layer;
ib_dev->get_netdev = hns_roce_get_netdev;
- ib_dev->query_gid = hns_roce_query_gid;
ib_dev->add_gid = hns_roce_add_gid;
ib_dev->del_gid = hns_roce_del_gid;
ib_dev->query_pkey = hns_roce_query_pkey;
@@ -520,6 +517,7 @@ static int hns_roce_register_device(struct hns_roce_dev *hr_dev)
/* OTHERS */
ib_dev->get_port_immutable = hns_roce_port_immutable;
+ ib_dev->driver_id = RDMA_DRIVER_HNS;
ret = ib_register_device(ib_dev, NULL);
if (ret) {
dev_err(dev, "ib_register_device failed!\n");
@@ -659,6 +657,11 @@ static int hns_roce_setup_hca(struct hns_roce_dev *hr_dev)
spin_lock_init(&hr_dev->sm_lock);
spin_lock_init(&hr_dev->bt_cmd_lock);
+ if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) {
+ INIT_LIST_HEAD(&hr_dev->pgdir_list);
+ mutex_init(&hr_dev->pgdir_mutex);
+ }
+
ret = hns_roce_init_uar_table(hr_dev);
if (ret) {
dev_err(dev, "Failed to initialize uar table. aborting\n");
diff --git a/drivers/infiniband/hw/hns/hns_roce_mr.c b/drivers/infiniband/hw/hns/hns_roce_mr.c
index da86a8117bd5..f7256d88d38f 100644
--- a/drivers/infiniband/hw/hns/hns_roce_mr.c
+++ b/drivers/infiniband/hw/hns/hns_roce_mr.c
@@ -933,7 +933,7 @@ int hns_roce_ib_umem_write_mtt(struct hns_roce_dev *hr_dev,
ret = hns_roce_write_mtt(hr_dev, mtt, n, i, pages);
out:
- free_page((unsigned long) pages);
+ free_pages((unsigned long) pages, order);
return ret;
}
diff --git a/drivers/infiniband/hw/hns/hns_roce_pd.c b/drivers/infiniband/hw/hns/hns_roce_pd.c
index bdab2188c04a..4b41e041799c 100644
--- a/drivers/infiniband/hw/hns/hns_roce_pd.c
+++ b/drivers/infiniband/hw/hns/hns_roce_pd.c
@@ -32,6 +32,7 @@
#include <linux/platform_device.h>
#include <linux/pci.h>
+#include <uapi/rdma/hns-abi.h>
#include "hns_roce_device.h"
static int hns_roce_pd_alloc(struct hns_roce_dev *hr_dev, unsigned long *pdn)
@@ -77,7 +78,9 @@ struct ib_pd *hns_roce_alloc_pd(struct ib_device *ib_dev,
}
if (context) {
- if (ib_copy_to_udata(udata, &pd->pdn, sizeof(u64))) {
+ struct hns_roce_ib_alloc_pd_resp uresp = {.pdn = pd->pdn};
+
+ if (ib_copy_to_udata(udata, &uresp, sizeof(uresp))) {
hns_roce_pd_free(to_hr_dev(ib_dev), pd->pdn);
dev_err(dev, "[alloc_pd]ib_copy_to_udata failed!\n");
kfree(pd);
diff --git a/drivers/infiniband/hw/hns/hns_roce_qp.c b/drivers/infiniband/hw/hns/hns_roce_qp.c
index 088973a05882..e289a924e789 100644
--- a/drivers/infiniband/hw/hns/hns_roce_qp.c
+++ b/drivers/infiniband/hw/hns/hns_roce_qp.c
@@ -489,6 +489,15 @@ static int hns_roce_set_kernel_sq_size(struct hns_roce_dev *hr_dev,
return 0;
}
+static int hns_roce_qp_has_rq(struct ib_qp_init_attr *attr)
+{
+ if (attr->qp_type == IB_QPT_XRC_INI ||
+ attr->qp_type == IB_QPT_XRC_TGT || attr->srq)
+ return 0;
+
+ return 1;
+}
+
static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
struct ib_pd *ib_pd,
struct ib_qp_init_attr *init_attr,
@@ -497,6 +506,7 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
{
struct device *dev = hr_dev->dev;
struct hns_roce_ib_create_qp ucmd;
+ struct hns_roce_ib_create_qp_resp resp = {};
unsigned long qpn = 0;
int ret = 0;
u32 page_shift;
@@ -602,6 +612,18 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
dev_err(dev, "hns_roce_ib_umem_write_mtt error for create qp\n");
goto err_mtt;
}
+
+ if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) &&
+ (udata->outlen >= sizeof(resp)) &&
+ hns_roce_qp_has_rq(init_attr)) {
+ ret = hns_roce_db_map_user(
+ to_hr_ucontext(ib_pd->uobject->context),
+ ucmd.db_addr, &hr_qp->rdb);
+ if (ret) {
+ dev_err(dev, "rp record doorbell map failed!\n");
+ goto err_mtt;
+ }
+ }
} else {
if (init_attr->create_flags &
IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK) {
@@ -630,6 +652,16 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
hr_qp->rq.db_reg_l = hr_dev->reg_base + hr_dev->odb_offset +
DB_REG_OFFSET * hr_dev->priv_uar.index;
+ if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) &&
+ hns_roce_qp_has_rq(init_attr)) {
+ ret = hns_roce_alloc_db(hr_dev, &hr_qp->rdb, 0);
+ if (ret) {
+ dev_err(dev, "rq record doorbell alloc failed!\n");
+ goto err_rq_sge_list;
+ }
+ *hr_qp->rdb.db_record = 0;
+ }
+
/* Allocate QP buf */
page_shift = PAGE_SHIFT + hr_dev->caps.mtt_buf_pg_sz;
if (hns_roce_buf_alloc(hr_dev, hr_qp->buff_size,
@@ -637,7 +669,7 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
&hr_qp->hr_buf, page_shift)) {
dev_err(dev, "hns_roce_buf_alloc error!\n");
ret = -ENOMEM;
- goto err_rq_sge_list;
+ goto err_db;
}
hr_qp->mtt.mtt_type = MTT_TYPE_WQE;
@@ -698,17 +730,44 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
else
hr_qp->doorbell_qpn = cpu_to_le64(hr_qp->qpn);
+ if (ib_pd->uobject && (udata->outlen >= sizeof(resp)) &&
+ (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB)) {
+
+ /* indicate kernel supports record db */
+ resp.cap_flags |= HNS_ROCE_SUPPORT_RQ_RECORD_DB;
+ ret = ib_copy_to_udata(udata, &resp, sizeof(resp));
+ if (ret)
+ goto err_qp;
+
+ hr_qp->rdb_en = 1;
+ }
hr_qp->event = hns_roce_ib_qp_event;
return 0;
+err_qp:
+ if (init_attr->qp_type == IB_QPT_GSI &&
+ hr_dev->hw_rev == HNS_ROCE_HW_VER1)
+ hns_roce_qp_remove(hr_dev, hr_qp);
+ else
+ hns_roce_qp_free(hr_dev, hr_qp);
+
err_qpn:
if (!sqpn)
hns_roce_release_range_qp(hr_dev, qpn, 1);
err_wrid:
- kfree(hr_qp->sq.wrid);
- kfree(hr_qp->rq.wrid);
+ if (ib_pd->uobject) {
+ if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) &&
+ (udata->outlen >= sizeof(resp)) &&
+ hns_roce_qp_has_rq(init_attr))
+ hns_roce_db_unmap_user(
+ to_hr_ucontext(ib_pd->uobject->context),
+ &hr_qp->rdb);
+ } else {
+ kfree(hr_qp->sq.wrid);
+ kfree(hr_qp->rq.wrid);
+ }
err_mtt:
hns_roce_mtt_cleanup(hr_dev, &hr_qp->mtt);
@@ -719,6 +778,11 @@ err_buf:
else
hns_roce_buf_free(hr_dev, hr_qp->buff_size, &hr_qp->hr_buf);
+err_db:
+ if (!ib_pd->uobject && hns_roce_qp_has_rq(init_attr) &&
+ (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB))
+ hns_roce_free_db(hr_dev, &hr_qp->rdb);
+
err_rq_sge_list:
if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE)
kfree(hr_qp->rq_inl_buf.wqe_list[0].sg_list);
diff --git a/drivers/infiniband/hw/i40iw/i40iw.h b/drivers/infiniband/hw/i40iw/i40iw.h
index bcddd7061fc0..d5d8c1be345a 100644
--- a/drivers/infiniband/hw/i40iw/i40iw.h
+++ b/drivers/infiniband/hw/i40iw/i40iw.h
@@ -60,7 +60,7 @@
#include <i40e_client.h>
#include "i40iw_type.h"
#include "i40iw_p.h"
-#include "i40iw_ucontext.h"
+#include <rdma/i40iw-abi.h>
#include "i40iw_pble.h"
#include "i40iw_verbs.h"
#include "i40iw_cm.h"
@@ -559,18 +559,25 @@ void i40iw_next_iw_state(struct i40iw_qp *iwqp,
u8 state, u8 del_hash,
u8 term, u8 term_len);
int i40iw_send_syn(struct i40iw_cm_node *cm_node, u32 sendack);
+int i40iw_send_reset(struct i40iw_cm_node *cm_node);
struct i40iw_cm_node *i40iw_find_node(struct i40iw_cm_core *cm_core,
u16 rem_port,
u32 *rem_addr,
u16 loc_port,
u32 *loc_addr,
- bool add_refcnt);
+ bool add_refcnt,
+ bool accelerated_list);
enum i40iw_status_code i40iw_hw_flush_wqes(struct i40iw_device *iwdev,
struct i40iw_sc_qp *qp,
struct i40iw_qp_flush_info *info,
bool wait);
+void i40iw_gen_ae(struct i40iw_device *iwdev,
+ struct i40iw_sc_qp *qp,
+ struct i40iw_gen_ae_info *info,
+ bool wait);
+
void i40iw_copy_ip_ntohl(u32 *dst, __be32 *src);
struct ib_mr *i40iw_reg_phys_mr(struct ib_pd *ib_pd,
u64 addr,
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c
index abf4cd897849..4cfa8f4647e2 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c
@@ -539,7 +539,7 @@ static struct i40iw_puda_buf *i40iw_form_cm_frame(struct i40iw_cm_node *cm_node,
* i40iw_send_reset - Send RST packet
* @cm_node: connection's node
*/
-static int i40iw_send_reset(struct i40iw_cm_node *cm_node)
+int i40iw_send_reset(struct i40iw_cm_node *cm_node)
{
struct i40iw_puda_buf *sqbuf;
int flags = SET_RST | SET_ACK;
@@ -1183,6 +1183,26 @@ static void i40iw_handle_close_entry(struct i40iw_cm_node *cm_node, u32 rem_node
}
/**
+ * i40iw_build_timer_list - Add cm_nodes to timer list
+ * @timer_list: ptr to timer list
+ * @hte: ptr to accelerated or non-accelerated list
+ */
+static void i40iw_build_timer_list(struct list_head *timer_list,
+ struct list_head *hte)
+{
+ struct i40iw_cm_node *cm_node;
+ struct list_head *list_core_temp, *list_node;
+
+ list_for_each_safe(list_node, list_core_temp, hte) {
+ cm_node = container_of(list_node, struct i40iw_cm_node, list);
+ if (cm_node->close_entry || cm_node->send_entry) {
+ atomic_inc(&cm_node->ref_count);
+ list_add(&cm_node->timer_entry, timer_list);
+ }
+ }
+}
+
+/**
* i40iw_cm_timer_tick - system's timer expired callback
* @pass: Pointing to cm_core
*/
@@ -1202,15 +1222,10 @@ static void i40iw_cm_timer_tick(struct timer_list *t)
struct list_head timer_list;
INIT_LIST_HEAD(&timer_list);
- spin_lock_irqsave(&cm_core->ht_lock, flags);
- list_for_each_safe(list_node, list_core_temp, &cm_core->connected_nodes) {
- cm_node = container_of(list_node, struct i40iw_cm_node, list);
- if (cm_node->close_entry || cm_node->send_entry) {
- atomic_inc(&cm_node->ref_count);
- list_add(&cm_node->timer_entry, &timer_list);
- }
- }
+ spin_lock_irqsave(&cm_core->ht_lock, flags);
+ i40iw_build_timer_list(&timer_list, &cm_core->non_accelerated_list);
+ i40iw_build_timer_list(&timer_list, &cm_core->accelerated_list);
spin_unlock_irqrestore(&cm_core->ht_lock, flags);
list_for_each_safe(list_node, list_core_temp, &timer_list) {
@@ -1406,19 +1421,22 @@ static int i40iw_send_fin(struct i40iw_cm_node *cm_node)
* @loc_port: local tcp port num
* @loc_addr: loc ip addr
* @add_refcnt: flag to increment refcount of cm_node
+ * @accelerated_list: flag for accelerated vs non-accelerated list to search
*/
struct i40iw_cm_node *i40iw_find_node(struct i40iw_cm_core *cm_core,
u16 rem_port,
u32 *rem_addr,
u16 loc_port,
u32 *loc_addr,
- bool add_refcnt)
+ bool add_refcnt,
+ bool accelerated_list)
{
struct list_head *hte;
struct i40iw_cm_node *cm_node;
unsigned long flags;
- hte = &cm_core->connected_nodes;
+ hte = accelerated_list ?
+ &cm_core->accelerated_list : &cm_core->non_accelerated_list;
/* walk list and find cm_node associated with this session ID */
spin_lock_irqsave(&cm_core->ht_lock, flags);
@@ -1487,22 +1505,40 @@ static struct i40iw_cm_listener *i40iw_find_listener(
static void i40iw_add_hte_node(struct i40iw_cm_core *cm_core,
struct i40iw_cm_node *cm_node)
{
- struct list_head *hte;
unsigned long flags;
if (!cm_node || !cm_core) {
i40iw_pr_err("cm_node or cm_core == NULL\n");
return;
}
- spin_lock_irqsave(&cm_core->ht_lock, flags);
- /* get a handle on the hash table element (list head for this slot) */
- hte = &cm_core->connected_nodes;
- list_add_tail(&cm_node->list, hte);
+ spin_lock_irqsave(&cm_core->ht_lock, flags);
+ list_add_tail(&cm_node->list, &cm_core->non_accelerated_list);
spin_unlock_irqrestore(&cm_core->ht_lock, flags);
}
/**
+ * i40iw_find_port - find port that matches reference port
+ * @port: port number
+ * @accelerated_list: flag for accelerated vs non-accelerated list
+ */
+static bool i40iw_find_port(struct i40iw_cm_core *cm_core, u16 port,
+ bool accelerated_list)
+{
+ struct list_head *hte;
+ struct i40iw_cm_node *cm_node;
+
+ hte = accelerated_list ?
+ &cm_core->accelerated_list : &cm_core->non_accelerated_list;
+
+ list_for_each_entry(cm_node, hte, list) {
+ if (cm_node->loc_port == port)
+ return true;
+ }
+ return false;
+}
+
+/**
* i40iw_port_in_use - determine if port is in use
* @port: port number
* @active_side: flag for listener side vs active side
@@ -1510,19 +1546,14 @@ static void i40iw_add_hte_node(struct i40iw_cm_core *cm_core,
static bool i40iw_port_in_use(struct i40iw_cm_core *cm_core, u16 port, bool active_side)
{
struct i40iw_cm_listener *listen_node;
- struct i40iw_cm_node *cm_node;
unsigned long flags;
bool ret = false;
if (active_side) {
- /* search connected node list */
spin_lock_irqsave(&cm_core->ht_lock, flags);
- list_for_each_entry(cm_node, &cm_core->connected_nodes, list) {
- if (cm_node->loc_port == port) {
- ret = true;
- break;
- }
- }
+ ret = i40iw_find_port(cm_core, port, true);
+ if (!ret)
+ ret = i40iw_find_port(cm_core, port, false);
if (!ret)
clear_bit(port, cm_core->active_side_ports);
spin_unlock_irqrestore(&cm_core->ht_lock, flags);
@@ -1829,9 +1860,11 @@ static int i40iw_dec_refcnt_listen(struct i40iw_cm_core *cm_core,
INIT_LIST_HEAD(&reset_list);
if (free_hanging_nodes) {
spin_lock_irqsave(&cm_core->ht_lock, flags);
- list_for_each_safe(list_pos, list_temp, &cm_core->connected_nodes) {
+ list_for_each_safe(list_pos,
+ list_temp, &cm_core->non_accelerated_list) {
cm_node = container_of(list_pos, struct i40iw_cm_node, list);
- if ((cm_node->listener == listener) && !cm_node->accelerated) {
+ if ((cm_node->listener == listener) &&
+ !cm_node->accelerated) {
atomic_inc(&cm_node->ref_count);
list_add(&cm_node->reset_entry, &reset_list);
}
@@ -3144,7 +3177,8 @@ void i40iw_receive_ilq(struct i40iw_sc_vsi *vsi, struct i40iw_puda_buf *rbuf)
cm_info.rem_addr,
cm_info.loc_port,
cm_info.loc_addr,
- true);
+ true,
+ false);
if (!cm_node) {
/* Only type of packet accepted are for */
@@ -3202,7 +3236,8 @@ void i40iw_setup_cm_core(struct i40iw_device *iwdev)
cm_core->iwdev = iwdev;
cm_core->dev = &iwdev->sc_dev;
- INIT_LIST_HEAD(&cm_core->connected_nodes);
+ INIT_LIST_HEAD(&cm_core->accelerated_list);
+ INIT_LIST_HEAD(&cm_core->non_accelerated_list);
INIT_LIST_HEAD(&cm_core->listen_nodes);
timer_setup(&cm_core->tcp_timer, i40iw_cm_timer_tick, 0);
@@ -3585,6 +3620,7 @@ int i40iw_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
struct i40iw_qp *iwqp;
struct i40iw_device *iwdev;
struct i40iw_sc_dev *dev;
+ struct i40iw_cm_core *cm_core;
struct i40iw_cm_node *cm_node;
struct ib_qp_attr attr;
int passive_state;
@@ -3594,6 +3630,7 @@ int i40iw_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
struct i40iw_kmem_info accept;
enum i40iw_status_code status;
u64 tagged_offset;
+ unsigned long flags;
memset(&attr, 0, sizeof(attr));
ibqp = i40iw_get_qp(cm_id->device, conn_param->qpn);
@@ -3603,6 +3640,7 @@ int i40iw_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
iwqp = to_iwqp(ibqp);
iwdev = iwqp->iwdev;
dev = &iwdev->sc_dev;
+ cm_core = &iwdev->cm_core;
cm_node = (struct i40iw_cm_node *)cm_id->provider_data;
if (((struct sockaddr_in *)&cm_id->local_addr)->sin_family == AF_INET) {
@@ -3697,6 +3735,10 @@ int i40iw_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
i40iw_modify_qp(&iwqp->ibqp, &attr, IB_QP_STATE, NULL);
cm_node->accelerated = true;
+ spin_lock_irqsave(&cm_core->ht_lock, flags);
+ list_move_tail(&cm_node->list, &cm_core->accelerated_list);
+ spin_unlock_irqrestore(&cm_core->ht_lock, flags);
+
status =
i40iw_send_cm_event(cm_node, cm_id, IW_CM_EVENT_ESTABLISHED, 0);
if (status)
@@ -4026,10 +4068,12 @@ static void i40iw_cm_event_connected(struct i40iw_cm_event *event)
{
struct i40iw_qp *iwqp;
struct i40iw_device *iwdev;
+ struct i40iw_cm_core *cm_core;
struct i40iw_cm_node *cm_node;
struct i40iw_sc_dev *dev;
struct ib_qp_attr attr;
struct iw_cm_id *cm_id;
+ unsigned long flags;
int status;
bool read0;
@@ -4038,6 +4082,7 @@ static void i40iw_cm_event_connected(struct i40iw_cm_event *event)
iwqp = (struct i40iw_qp *)cm_id->provider_data;
iwdev = to_iwdev(iwqp->ibqp.device);
dev = &iwdev->sc_dev;
+ cm_core = &iwdev->cm_core;
if (iwqp->destroyed) {
status = -ETIMEDOUT;
@@ -4057,6 +4102,9 @@ static void i40iw_cm_event_connected(struct i40iw_cm_event *event)
i40iw_modify_qp(&iwqp->ibqp, &attr, IB_QP_STATE, NULL);
cm_node->accelerated = true;
+ spin_lock_irqsave(&cm_core->ht_lock, flags);
+ list_move_tail(&cm_node->list, &cm_core->accelerated_list);
+ spin_unlock_irqrestore(&cm_core->ht_lock, flags);
status = i40iw_send_cm_event(cm_node, cm_id, IW_CM_EVENT_CONNECT_REPLY,
0);
if (status)
@@ -4256,25 +4304,38 @@ void i40iw_cm_teardown_connections(struct i40iw_device *iwdev, u32 *ipaddr,
struct list_head *list_node;
struct i40iw_cm_node *cm_node;
unsigned long flags;
- struct list_head connected_list;
+ struct list_head teardown_list;
struct ib_qp_attr attr;
- INIT_LIST_HEAD(&connected_list);
+ INIT_LIST_HEAD(&teardown_list);
spin_lock_irqsave(&cm_core->ht_lock, flags);
- list_for_each_safe(list_node, list_core_temp, &cm_core->connected_nodes) {
+ list_for_each_safe(list_node, list_core_temp,
+ &cm_core->accelerated_list) {
+ cm_node = container_of(list_node, struct i40iw_cm_node, list);
+ if (disconnect_all ||
+ (nfo->vlan_id == cm_node->vlan_id &&
+ (!memcmp(cm_node->loc_addr, ipaddr, nfo->ipv4 ? 4 : 16) ||
+ !memcmp(cm_node->rem_addr, ipaddr, nfo->ipv4 ? 4 : 16)))) {
+ atomic_inc(&cm_node->ref_count);
+ list_add(&cm_node->teardown_entry, &teardown_list);
+ }
+ }
+ list_for_each_safe(list_node, list_core_temp,
+ &cm_core->non_accelerated_list) {
cm_node = container_of(list_node, struct i40iw_cm_node, list);
if (disconnect_all ||
(nfo->vlan_id == cm_node->vlan_id &&
(!memcmp(cm_node->loc_addr, ipaddr, nfo->ipv4 ? 4 : 16) ||
!memcmp(cm_node->rem_addr, ipaddr, nfo->ipv4 ? 4 : 16)))) {
atomic_inc(&cm_node->ref_count);
- list_add(&cm_node->connected_entry, &connected_list);
+ list_add(&cm_node->teardown_entry, &teardown_list);
}
}
spin_unlock_irqrestore(&cm_core->ht_lock, flags);
- list_for_each_safe(list_node, list_core_temp, &connected_list) {
- cm_node = container_of(list_node, struct i40iw_cm_node, connected_entry);
+ list_for_each_safe(list_node, list_core_temp, &teardown_list) {
+ cm_node = container_of(list_node, struct i40iw_cm_node,
+ teardown_entry);
attr.qp_state = IB_QPS_ERR;
i40iw_modify_qp(&cm_node->iwqp->ibqp, &attr, IB_QP_STATE, NULL);
if (iwdev->reset)
diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.h b/drivers/infiniband/hw/i40iw/i40iw_cm.h
index cf60c451e071..78ba36ae2bbe 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_cm.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_cm.h
@@ -341,7 +341,7 @@ struct i40iw_cm_node {
int accept_pend;
struct list_head timer_entry;
struct list_head reset_entry;
- struct list_head connected_entry;
+ struct list_head teardown_entry;
atomic_t passive_state;
bool qhash_set;
u8 user_pri;
@@ -403,7 +403,8 @@ struct i40iw_cm_core {
struct i40iw_sc_dev *dev;
struct list_head listen_nodes;
- struct list_head connected_nodes;
+ struct list_head accelerated_list;
+ struct list_head non_accelerated_list;
struct timer_list tcp_timer;
diff --git a/drivers/infiniband/hw/i40iw/i40iw_ctrl.c b/drivers/infiniband/hw/i40iw/i40iw_ctrl.c
index c74fd3309b93..4d841a3c68f3 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_ctrl.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_ctrl.c
@@ -2614,10 +2614,8 @@ static enum i40iw_status_code i40iw_sc_qp_flush_wqes(
qp->flush_sq |= flush_sq;
qp->flush_rq |= flush_rq;
- if (!flush_sq && !flush_rq) {
- if (info->ae_code != I40IW_AE_LLP_RECEIVED_MPA_CRC_ERROR)
- return 0;
- }
+ if (!flush_sq && !flush_rq)
+ return 0;
cqp = qp->pd->dev->cqp;
wqe = i40iw_sc_cqp_get_next_send_wqe(cqp, scratch);
@@ -2659,6 +2657,49 @@ static enum i40iw_status_code i40iw_sc_qp_flush_wqes(
}
/**
+ * i40iw_sc_gen_ae - generate AE, currently uses flush WQE CQP OP
+ * @qp: sc qp
+ * @info: gen ae information
+ * @scratch: u64 saved to be used during cqp completion
+ * @post_sq: flag for cqp db to ring
+ */
+static enum i40iw_status_code i40iw_sc_gen_ae(
+ struct i40iw_sc_qp *qp,
+ struct i40iw_gen_ae_info *info,
+ u64 scratch,
+ bool post_sq)
+{
+ u64 temp;
+ u64 *wqe;
+ struct i40iw_sc_cqp *cqp;
+ u64 header;
+
+ cqp = qp->pd->dev->cqp;
+ wqe = i40iw_sc_cqp_get_next_send_wqe(cqp, scratch);
+ if (!wqe)
+ return I40IW_ERR_RING_FULL;
+
+ temp = info->ae_code |
+ LS_64(info->ae_source, I40IW_CQPSQ_FWQE_AESOURCE);
+
+ set_64bit_val(wqe, 8, temp);
+
+ header = qp->qp_uk.qp_id |
+ LS_64(I40IW_CQP_OP_GEN_AE, I40IW_CQPSQ_OPCODE) |
+ LS_64(1, I40IW_CQPSQ_FWQE_GENERATE_AE) |
+ LS_64(cqp->polarity, I40IW_CQPSQ_WQEVALID);
+
+ i40iw_insert_wqe_hdr(wqe, header);
+
+ i40iw_debug_buf(cqp->dev, I40IW_DEBUG_WQE, "GEN_AE WQE",
+ wqe, I40IW_CQP_WQE_SIZE * 8);
+
+ if (post_sq)
+ i40iw_sc_cqp_post_sq(cqp);
+ return 0;
+}
+
+/**
* i40iw_sc_qp_upload_context - upload qp's context
* @dev: sc device struct
* @info: upload context info ptr for return
@@ -4148,6 +4189,13 @@ static enum i40iw_status_code i40iw_exec_cqp_cmd(struct i40iw_sc_dev *dev,
pcmdinfo->in.u.qp_flush_wqes.
scratch, pcmdinfo->post_sq);
break;
+ case OP_GEN_AE:
+ status = i40iw_sc_gen_ae(
+ pcmdinfo->in.u.gen_ae.qp,
+ &pcmdinfo->in.u.gen_ae.info,
+ pcmdinfo->in.u.gen_ae.scratch,
+ pcmdinfo->post_sq);
+ break;
case OP_ADD_ARP_CACHE_ENTRY:
status = i40iw_sc_add_arp_cache_entry(
pcmdinfo->in.u.add_arp_cache_entry.cqp,
diff --git a/drivers/infiniband/hw/i40iw/i40iw_d.h b/drivers/infiniband/hw/i40iw/i40iw_d.h
index 4b65e4140bd7..6ddaeec87d2f 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_d.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_d.h
@@ -418,6 +418,8 @@
#define I40IW_CQP_OP_QUERY_FPM_VALUES 0x20
#define I40IW_CQP_OP_COMMIT_FPM_VALUES 0x21
#define I40IW_CQP_OP_FLUSH_WQES 0x22
+/* I40IW_CQP_OP_GEN_AE is the same value as I40IW_CQP_OP_FLUSH_WQES */
+#define I40IW_CQP_OP_GEN_AE 0x22
#define I40IW_CQP_OP_MANAGE_APBVT 0x23
#define I40IW_CQP_OP_NOP 0x24
#define I40IW_CQP_OP_MANAGE_QUAD_HASH_TABLE_ENTRY 0x25
@@ -1729,6 +1731,7 @@ enum i40iw_alignment {
#define OP_COMMIT_FPM_VALUES 30
#define OP_REQUESTED_COMMANDS 31
#define OP_COMPLETED_COMMANDS 32
-#define OP_SIZE_CQP_STAT_ARRAY 33
+#define OP_GEN_AE 33
+#define OP_SIZE_CQP_STAT_ARRAY 34
#endif
diff --git a/drivers/infiniband/hw/i40iw/i40iw_hw.c b/drivers/infiniband/hw/i40iw/i40iw_hw.c
index 61540e14e4b9..6139836fb533 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_hw.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_hw.c
@@ -352,6 +352,8 @@ void i40iw_process_aeq(struct i40iw_device *iwdev)
else
i40iw_cm_disconn(iwqp);
break;
+ case I40IW_AE_BAD_CLOSE:
+ /* fall through */
case I40IW_AE_RESET_SENT:
i40iw_next_iw_state(iwqp, I40IW_QP_STATE_ERROR, 1, 0, 0);
i40iw_cm_disconn(iwqp);
@@ -668,6 +670,39 @@ enum i40iw_status_code i40iw_hw_flush_wqes(struct i40iw_device *iwdev,
}
/**
+ * i40iw_gen_ae - generate AE
+ * @iwdev: iwarp device
+ * @qp: qp associated with AE
+ * @info: info for ae
+ * @wait: wait for completion
+ */
+void i40iw_gen_ae(struct i40iw_device *iwdev,
+ struct i40iw_sc_qp *qp,
+ struct i40iw_gen_ae_info *info,
+ bool wait)
+{
+ struct i40iw_gen_ae_info *ae_info;
+ struct i40iw_cqp_request *cqp_request;
+ struct cqp_commands_info *cqp_info;
+
+ cqp_request = i40iw_get_cqp_request(&iwdev->cqp, wait);
+ if (!cqp_request)
+ return;
+
+ cqp_info = &cqp_request->info;
+ ae_info = &cqp_request->info.in.u.gen_ae.info;
+ memcpy(ae_info, info, sizeof(*ae_info));
+
+ cqp_info->cqp_cmd = OP_GEN_AE;
+ cqp_info->post_sq = 1;
+ cqp_info->in.u.gen_ae.qp = qp;
+ cqp_info->in.u.gen_ae.scratch = (uintptr_t)cqp_request;
+ if (i40iw_handle_cqp_op(iwdev, cqp_request))
+ i40iw_pr_err("CQP OP failed attempting to generate ae_code=0x%x\n",
+ info->ae_code);
+}
+
+/**
* i40iw_hw_manage_vf_pble_bp - manage vf pbles
* @iwdev: iwarp device
* @info: info for managing pble
diff --git a/drivers/infiniband/hw/i40iw/i40iw_main.c b/drivers/infiniband/hw/i40iw/i40iw_main.c
index b08862978de8..9cd0d3ef9057 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_main.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_main.c
@@ -1560,8 +1560,6 @@ static enum i40iw_status_code i40iw_setup_init_state(struct i40iw_handler *hdl,
enum i40iw_status_code status;
memcpy(&hdl->ldev, ldev, sizeof(*ldev));
- if (resource_profile == 1)
- resource_profile = 2;
iwdev->mpa_version = mpa_version;
iwdev->resource_profile = (resource_profile < I40IW_HMC_PROFILE_EQUAL) ?
diff --git a/drivers/infiniband/hw/i40iw/i40iw_puda.c b/drivers/infiniband/hw/i40iw/i40iw_puda.c
index 4c21197830b3..d9c7ae6a7030 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_puda.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_puda.c
@@ -348,8 +348,8 @@ enum i40iw_status_code i40iw_puda_poll_completion(struct i40iw_sc_dev *dev,
spin_lock_irqsave(&rsrc->bufpool_lock, flags);
rsrc->tx_wqe_avail_cnt++;
spin_unlock_irqrestore(&rsrc->bufpool_lock, flags);
- if (!list_empty(&rsrc->vsi->ilq->txpend))
- i40iw_puda_send_buf(rsrc->vsi->ilq, NULL);
+ if (!list_empty(&rsrc->txpend))
+ i40iw_puda_send_buf(rsrc, NULL);
}
done:
@@ -1471,10 +1471,6 @@ static void i40iw_ieq_tx_compl(struct i40iw_sc_vsi *vsi, void *sqwrid)
struct i40iw_puda_buf *buf = (struct i40iw_puda_buf *)sqwrid;
i40iw_puda_ret_bufpool(ieq, buf);
- if (!list_empty(&ieq->txpend)) {
- buf = i40iw_puda_get_listbuf(&ieq->txpend);
- i40iw_puda_send_buf(ieq, buf);
- }
}
/**
diff --git a/drivers/infiniband/hw/i40iw/i40iw_type.h b/drivers/infiniband/hw/i40iw/i40iw_type.h
index a27d392c92a2..adc8d2ec523d 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_type.h
+++ b/drivers/infiniband/hw/i40iw/i40iw_type.h
@@ -1004,6 +1004,11 @@ struct i40iw_cqp_query_fpm_values {
u32 pbl_max;
};
+struct i40iw_gen_ae_info {
+ u16 ae_code;
+ u8 ae_source;
+};
+
struct i40iw_cqp_ops {
enum i40iw_status_code (*cqp_init)(struct i40iw_sc_cqp *,
struct i40iw_cqp_init_info *);
@@ -1291,6 +1296,12 @@ struct cqp_info {
} qp_flush_wqes;
struct {
+ struct i40iw_sc_qp *qp;
+ struct i40iw_gen_ae_info info;
+ u64 scratch;
+ } gen_ae;
+
+ struct {
struct i40iw_sc_cqp *cqp;
void *fpm_values_va;
u64 fpm_values_pa;
diff --git a/drivers/infiniband/hw/i40iw/i40iw_ucontext.h b/drivers/infiniband/hw/i40iw/i40iw_ucontext.h
deleted file mode 100644
index 57d3f1d11ff1..000000000000
--- a/drivers/infiniband/hw/i40iw/i40iw_ucontext.h
+++ /dev/null
@@ -1,107 +0,0 @@
-/*
- * Copyright (c) 2006 - 2016 Intel Corporation. All rights reserved.
- * Copyright (c) 2005 Topspin Communications. All rights reserved.
- * Copyright (c) 2005 Cisco Systems. All rights reserved.
- * Copyright (c) 2005 Open Grid Computing, Inc. All rights reserved.
- *
- * This software is available to you under a choice of one of two
- * licenses. You may choose to be licensed under the terms of the GNU
- * General Public License (GPL) Version 2, available from the file
- * COPYING in the main directory of this source tree, or the
- * OpenIB.org BSD license below:
- *
- * Redistribution and use in source and binary forms, with or
- * without modification, are permitted provided that the following
- * conditions are met:
- *
- * - Redistributions of source code must retain the above
- * copyright notice, this list of conditions and the following
- * disclaimer.
- *
- * - Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following
- * disclaimer in the documentation and/or other materials
- * provided with the distribution.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
- * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
- * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- *
- */
-
-#ifndef I40IW_USER_CONTEXT_H
-#define I40IW_USER_CONTEXT_H
-
-#include <linux/types.h>
-
-#define I40IW_ABI_VER 5
-
-struct i40iw_alloc_ucontext_req {
- __u32 reserved32;
- __u8 userspace_ver;
- __u8 reserved8[3];
-};
-
-struct i40iw_alloc_ucontext_resp {
- __u32 max_pds; /* maximum pds allowed for this user process */
- __u32 max_qps; /* maximum qps allowed for this user process */
- __u32 wq_size; /* size of the WQs (sq+rq) allocated to the mmaped area */
- __u8 kernel_ver;
- __u8 reserved[3];
-};
-
-struct i40iw_alloc_pd_resp {
- __u32 pd_id;
- __u8 reserved[4];
-};
-
-struct i40iw_create_cq_req {
- __u64 user_cq_buffer;
- __u64 user_shadow_area;
-};
-
-struct i40iw_create_qp_req {
- __u64 user_wqe_buffers;
- __u64 user_compl_ctx;
-
- /* UDA QP PHB */
- __u64 user_sq_phb; /* place for VA of the sq phb buff */
- __u64 user_rq_phb; /* place for VA of the rq phb buff */
-};
-
-enum i40iw_memreg_type {
- IW_MEMREG_TYPE_MEM = 0x0000,
- IW_MEMREG_TYPE_QP = 0x0001,
- IW_MEMREG_TYPE_CQ = 0x0002,
-};
-
-struct i40iw_mem_reg_req {
- __u16 reg_type; /* Memory, QP or CQ */
- __u16 cq_pages;
- __u16 rq_pages;
- __u16 sq_pages;
-};
-
-struct i40iw_create_cq_resp {
- __u32 cq_id;
- __u32 cq_size;
- __u32 mmap_db_index;
- __u32 reserved;
-};
-
-struct i40iw_create_qp_resp {
- __u32 qp_id;
- __u32 actual_sq_size;
- __u32 actual_rq_size;
- __u32 i40iw_drv_opt;
- __u16 push_idx;
- __u8 lsmm;
- __u8 rsvd2;
-};
-
-#endif
diff --git a/drivers/infiniband/hw/i40iw/i40iw_utils.c b/drivers/infiniband/hw/i40iw/i40iw_utils.c
index ddc1056b0b4e..a9ea966877f2 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_utils.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_utils.c
@@ -1284,15 +1284,13 @@ void i40iw_cqp_qp_destroy_cmd(struct i40iw_sc_dev *dev, struct i40iw_sc_qp *qp)
*/
void i40iw_ieq_mpa_crc_ae(struct i40iw_sc_dev *dev, struct i40iw_sc_qp *qp)
{
- struct i40iw_qp_flush_info info;
+ struct i40iw_gen_ae_info info;
struct i40iw_device *iwdev = (struct i40iw_device *)dev->back_dev;
i40iw_debug(dev, I40IW_DEBUG_AEQ, "%s entered\n", __func__);
- memset(&info, 0, sizeof(info));
info.ae_code = I40IW_AE_LLP_RECEIVED_MPA_CRC_ERROR;
- info.generate_ae = true;
- info.ae_source = 0x3;
- (void)i40iw_hw_flush_wqes(iwdev, qp, &info, false);
+ info.ae_source = I40IW_AE_SOURCE_RQ;
+ i40iw_gen_ae(iwdev, qp, &info, false);
}
/**
@@ -1407,7 +1405,7 @@ struct i40iw_sc_qp *i40iw_ieq_get_qp(struct i40iw_sc_dev *dev,
rem_port = ntohs(tcph->source);
cm_node = i40iw_find_node(&iwdev->cm_core, rem_port, rem_addr, loc_port,
- loc_addr, false);
+ loc_addr, false, true);
if (!cm_node)
return NULL;
iwqp = cm_node->iwqp;
diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
index 70024e8e2692..40e4f5ab2b46 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c
@@ -38,6 +38,7 @@
#include <linux/highmem.h>
#include <linux/time.h>
#include <linux/hugetlb.h>
+#include <linux/irq.h>
#include <asm/byteorder.h>
#include <net/ip.h>
#include <rdma/ib_verbs.h>
@@ -830,10 +831,10 @@ static int i40iw_query_qp(struct ib_qp *ibqp,
void i40iw_hw_modify_qp(struct i40iw_device *iwdev, struct i40iw_qp *iwqp,
struct i40iw_modify_qp_info *info, bool wait)
{
- enum i40iw_status_code status;
struct i40iw_cqp_request *cqp_request;
struct cqp_commands_info *cqp_info;
struct i40iw_modify_qp_info *m_info;
+ struct i40iw_gen_ae_info ae_info;
cqp_request = i40iw_get_cqp_request(&iwdev->cqp, wait);
if (!cqp_request)
@@ -846,9 +847,25 @@ void i40iw_hw_modify_qp(struct i40iw_device *iwdev, struct i40iw_qp *iwqp,
cqp_info->post_sq = 1;
cqp_info->in.u.qp_modify.qp = &iwqp->sc_qp;
cqp_info->in.u.qp_modify.scratch = (uintptr_t)cqp_request;
- status = i40iw_handle_cqp_op(iwdev, cqp_request);
- if (status)
- i40iw_pr_err("CQP-OP Modify QP fail");
+ if (!i40iw_handle_cqp_op(iwdev, cqp_request))
+ return;
+
+ switch (m_info->next_iwarp_state) {
+ case I40IW_QP_STATE_RTS:
+ if (iwqp->iwarp_state == I40IW_QP_STATE_IDLE)
+ i40iw_send_reset(iwqp->cm_node);
+ /* fall through */
+ case I40IW_QP_STATE_IDLE:
+ case I40IW_QP_STATE_TERMINATE:
+ case I40IW_QP_STATE_CLOSING:
+ ae_info.ae_code = I40IW_AE_BAD_CLOSE;
+ ae_info.ae_source = 0;
+ i40iw_gen_ae(iwdev, &iwqp->sc_qp, &ae_info, false);
+ break;
+ case I40IW_QP_STATE_ERROR:
+ default:
+ break;
+ }
}
/**
@@ -961,10 +978,6 @@ int i40iw_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
iwqp->ibqp_state = attr->qp_state;
- if (issue_modify_qp)
- iwqp->iwarp_state = info.next_iwarp_state;
- else
- info.next_iwarp_state = iwqp->iwarp_state;
}
if (attr_mask & IB_QP_ACCESS_FLAGS) {
ctx_info->iwarp_info_valid = true;
@@ -1002,9 +1015,14 @@ int i40iw_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
spin_unlock_irqrestore(&iwqp->lock, flags);
- if (issue_modify_qp)
+ if (issue_modify_qp) {
i40iw_hw_modify_qp(iwdev, iwqp, &info, true);
+ spin_lock_irqsave(&iwqp->lock, flags);
+ iwqp->iwarp_state = info.next_iwarp_state;
+ spin_unlock_irqrestore(&iwqp->lock, flags);
+ }
+
if (issue_modify_qp && (iwqp->ibqp_state > IB_QPS_RTS)) {
if (dont_wait) {
if (iwqp->cm_id && iwqp->hw_tcp_state) {
@@ -2729,6 +2747,25 @@ static int i40iw_destroy_ah(struct ib_ah *ah)
}
/**
+ * i40iw_get_vector_affinity - report IRQ affinity mask
+ * @ibdev: IB device
+ * @comp_vector: completion vector index
+ */
+static const struct cpumask *i40iw_get_vector_affinity(struct ib_device *ibdev,
+ int comp_vector)
+{
+ struct i40iw_device *iwdev = to_iwdev(ibdev);
+ struct i40iw_msix_vector *msix_vec;
+
+ if (iwdev->msix_shared)
+ msix_vec = &iwdev->iw_msixtbl[comp_vector];
+ else
+ msix_vec = &iwdev->iw_msixtbl[comp_vector + 1];
+
+ return irq_get_affinity_mask(msix_vec->irq);
+}
+
+/**
* i40iw_init_rdma_device - initialization of iwarp device
* @iwdev: iwarp device
*/
@@ -2824,6 +2861,7 @@ static struct i40iw_ib_device *i40iw_init_rdma_device(struct i40iw_device *iwdev
iwibdev->ibdev.req_notify_cq = i40iw_req_notify_cq;
iwibdev->ibdev.post_send = i40iw_post_send;
iwibdev->ibdev.post_recv = i40iw_post_recv;
+ iwibdev->ibdev.get_vector_affinity = i40iw_get_vector_affinity;
return iwibdev;
}
@@ -2889,6 +2927,7 @@ int i40iw_register_rdma_device(struct i40iw_device *iwdev)
return -ENOMEM;
iwibdev = iwdev->iwibdev;
+ iwibdev->ibdev.driver_id = RDMA_DRIVER_I40IW;
ret = ib_register_device(&iwibdev->ibdev, NULL);
if (ret)
goto error;
diff --git a/drivers/infiniband/hw/mlx4/ah.c b/drivers/infiniband/hw/mlx4/ah.c
index 6dee4fdc5d67..9345d5b546d1 100644
--- a/drivers/infiniband/hw/mlx4/ah.c
+++ b/drivers/infiniband/hw/mlx4/ah.c
@@ -101,12 +101,10 @@ static struct ib_ah *create_iboe_ah(struct ib_pd *pd,
if (ret)
return ERR_PTR(ret);
eth_zero_addr(ah->av.eth.s_mac);
- if (gid_attr.ndev) {
- if (is_vlan_dev(gid_attr.ndev))
- vlan_tag = vlan_dev_vlan_id(gid_attr.ndev);
- memcpy(ah->av.eth.s_mac, gid_attr.ndev->dev_addr, ETH_ALEN);
- dev_put(gid_attr.ndev);
- }
+ if (is_vlan_dev(gid_attr.ndev))
+ vlan_tag = vlan_dev_vlan_id(gid_attr.ndev);
+ memcpy(ah->av.eth.s_mac, gid_attr.ndev->dev_addr, ETH_ALEN);
+ dev_put(gid_attr.ndev);
if (vlan_tag < 0x1000)
vlan_tag |= (rdma_ah_get_sl(ah_attr) & 7) << 13;
ah->av.eth.port_pd = cpu_to_be32(to_mpd(pd)->pdn |
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 5a0e4fc4785a..5b70744f414a 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -246,14 +246,11 @@ static int mlx4_ib_update_gids(struct gid_entry *gids,
return mlx4_ib_update_gids_v1(gids, ibdev, port_num);
}
-static int mlx4_ib_add_gid(struct ib_device *device,
- u8 port_num,
- unsigned int index,
- const union ib_gid *gid,
+static int mlx4_ib_add_gid(const union ib_gid *gid,
const struct ib_gid_attr *attr,
void **context)
{
- struct mlx4_ib_dev *ibdev = to_mdev(device);
+ struct mlx4_ib_dev *ibdev = to_mdev(attr->device);
struct mlx4_ib_iboe *iboe = &ibdev->iboe;
struct mlx4_port_gid_table *port_gid_table;
int free = -1, found = -1;
@@ -262,16 +259,16 @@ static int mlx4_ib_add_gid(struct ib_device *device,
int i;
struct gid_entry *gids = NULL;
- if (!rdma_cap_roce_gid_table(device, port_num))
+ if (!rdma_cap_roce_gid_table(attr->device, attr->port_num))
return -EINVAL;
- if (port_num > MLX4_MAX_PORTS)
+ if (attr->port_num > MLX4_MAX_PORTS)
return -EINVAL;
if (!context)
return -EINVAL;
- port_gid_table = &iboe->gids[port_num - 1];
+ port_gid_table = &iboe->gids[attr->port_num - 1];
spin_lock_bh(&iboe->lock);
for (i = 0; i < MLX4_MAX_PORT_GIDS; ++i) {
if (!memcmp(&port_gid_table->gids[i].gid, gid, sizeof(*gid)) &&
@@ -318,33 +315,30 @@ static int mlx4_ib_add_gid(struct ib_device *device,
spin_unlock_bh(&iboe->lock);
if (!ret && hw_update) {
- ret = mlx4_ib_update_gids(gids, ibdev, port_num);
+ ret = mlx4_ib_update_gids(gids, ibdev, attr->port_num);
kfree(gids);
}
return ret;
}
-static int mlx4_ib_del_gid(struct ib_device *device,
- u8 port_num,
- unsigned int index,
- void **context)
+static int mlx4_ib_del_gid(const struct ib_gid_attr *attr, void **context)
{
struct gid_cache_context *ctx = *context;
- struct mlx4_ib_dev *ibdev = to_mdev(device);
+ struct mlx4_ib_dev *ibdev = to_mdev(attr->device);
struct mlx4_ib_iboe *iboe = &ibdev->iboe;
struct mlx4_port_gid_table *port_gid_table;
int ret = 0;
int hw_update = 0;
struct gid_entry *gids = NULL;
- if (!rdma_cap_roce_gid_table(device, port_num))
+ if (!rdma_cap_roce_gid_table(attr->device, attr->port_num))
return -EINVAL;
- if (port_num > MLX4_MAX_PORTS)
+ if (attr->port_num > MLX4_MAX_PORTS)
return -EINVAL;
- port_gid_table = &iboe->gids[port_num - 1];
+ port_gid_table = &iboe->gids[attr->port_num - 1];
spin_lock_bh(&iboe->lock);
if (ctx) {
ctx->refcount--;
@@ -376,7 +370,7 @@ static int mlx4_ib_del_gid(struct ib_device *device,
spin_unlock_bh(&iboe->lock);
if (!ret && hw_update) {
- ret = mlx4_ib_update_gids(gids, ibdev, port_num);
+ ret = mlx4_ib_update_gids(gids, ibdev, attr->port_num);
kfree(gids);
}
return ret;
@@ -411,9 +405,6 @@ int mlx4_ib_gid_index_to_real_index(struct mlx4_ib_dev *ibdev,
if (attr.ndev)
dev_put(attr.ndev);
- if (!memcmp(&gid, &zgid, sizeof(gid)))
- return -EINVAL;
-
spin_lock_irqsave(&iboe->lock, flags);
port_gid_table = &iboe->gids[port_num - 1];
@@ -429,6 +420,9 @@ int mlx4_ib_gid_index_to_real_index(struct mlx4_ib_dev *ibdev,
return real_index;
}
+#define field_avail(type, fld, sz) (offsetof(type, fld) + \
+ sizeof(((type *)0)->fld) <= (sz))
+
static int mlx4_ib_query_device(struct ib_device *ibdev,
struct ib_device_attr *props,
struct ib_udata *uhw)
@@ -556,14 +550,19 @@ static int mlx4_ib_query_device(struct ib_device *ibdev,
props->timestamp_mask = 0xFFFFFFFFFFFFULL;
props->max_ah = INT_MAX;
- if ((dev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_RSS) &&
- (mlx4_ib_port_link_layer(ibdev, 1) == IB_LINK_LAYER_ETHERNET ||
- mlx4_ib_port_link_layer(ibdev, 2) == IB_LINK_LAYER_ETHERNET)) {
- props->rss_caps.max_rwq_indirection_tables = props->max_qp;
- props->rss_caps.max_rwq_indirection_table_size =
- dev->dev->caps.max_rss_tbl_sz;
- props->rss_caps.supported_qpts = 1 << IB_QPT_RAW_PACKET;
- props->max_wq_type_rq = props->max_qp;
+ if (mlx4_ib_port_link_layer(ibdev, 1) == IB_LINK_LAYER_ETHERNET ||
+ mlx4_ib_port_link_layer(ibdev, 2) == IB_LINK_LAYER_ETHERNET) {
+ if (dev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_RSS) {
+ props->rss_caps.max_rwq_indirection_tables =
+ props->max_qp;
+ props->rss_caps.max_rwq_indirection_table_size =
+ dev->dev->caps.max_rss_tbl_sz;
+ props->rss_caps.supported_qpts = 1 << IB_QPT_RAW_PACKET;
+ props->max_wq_type_rq = props->max_qp;
+ }
+
+ if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_FCS_KEEP)
+ props->raw_packet_caps |= IB_RAW_PACKET_CAP_SCATTER_FCS;
}
props->cq_caps.max_cq_moderation_count = MLX4_MAX_CQ_COUNT;
@@ -575,7 +574,7 @@ static int mlx4_ib_query_device(struct ib_device *ibdev,
if (uhw->outlen >= resp.response_length + sizeof(resp.hca_core_clock_offset)) {
resp.response_length += sizeof(resp.hca_core_clock_offset);
if (!err && !mlx4_is_slave(dev->dev)) {
- resp.comp_mask |= QUERY_DEVICE_RESP_MASK_TIMESTAMP;
+ resp.comp_mask |= MLX4_IB_QUERY_DEV_RESP_MASK_CORE_CLOCK_OFFSET;
resp.hca_core_clock_offset = clock_params.offset % PAGE_SIZE;
}
}
@@ -587,8 +586,7 @@ static int mlx4_ib_query_device(struct ib_device *ibdev,
sizeof(struct mlx4_wqe_data_seg);
}
- if (uhw->outlen >= resp.response_length + sizeof(resp.rss_caps)) {
- resp.response_length += sizeof(resp.rss_caps);
+ if (field_avail(typeof(resp), rss_caps, uhw->outlen)) {
if (props->rss_caps.supported_qpts) {
resp.rss_caps.rx_hash_function =
MLX4_IB_RX_HASH_FUNC_TOEPLITZ;
@@ -608,6 +606,22 @@ static int mlx4_ib_query_device(struct ib_device *ibdev,
resp.rss_caps.rx_hash_fields_mask |=
MLX4_IB_RX_HASH_INNER;
}
+ resp.response_length = offsetof(typeof(resp), rss_caps) +
+ sizeof(resp.rss_caps);
+ }
+
+ if (field_avail(typeof(resp), tso_caps, uhw->outlen)) {
+ if (dev->dev->caps.max_gso_sz &&
+ ((mlx4_ib_port_link_layer(ibdev, 1) ==
+ IB_LINK_LAYER_ETHERNET) ||
+ (mlx4_ib_port_link_layer(ibdev, 2) ==
+ IB_LINK_LAYER_ETHERNET))) {
+ resp.tso_caps.max_tso = dev->dev->caps.max_gso_sz;
+ resp.tso_caps.supported_qpts |=
+ 1 << IB_QPT_RAW_PACKET;
+ }
+ resp.response_length = offsetof(typeof(resp), tso_caps) +
+ sizeof(resp.tso_caps);
}
if (uhw->outlen) {
@@ -865,24 +879,9 @@ out:
static int mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
union ib_gid *gid)
{
- int ret;
-
if (rdma_protocol_ib(ibdev, port))
return __mlx4_ib_query_gid(ibdev, port, index, gid, 0);
-
- if (!rdma_protocol_roce(ibdev, port))
- return -ENODEV;
-
- if (!rdma_cap_roce_gid_table(ibdev, port))
- return -ENODEV;
-
- ret = ib_get_cached_gid(ibdev, port, index, gid, NULL);
- if (ret == -EAGAIN) {
- memcpy(gid, &zgid, sizeof(*gid));
- return 0;
- }
-
- return ret;
+ return 0;
}
static int mlx4_ib_query_sl2vl(struct ib_device *ibdev, u8 port, u64 *sl2vl_tbl)
@@ -1330,7 +1329,7 @@ static struct ib_pd *mlx4_ib_alloc_pd(struct ib_device *ibdev,
struct mlx4_ib_pd *pd;
int err;
- pd = kmalloc(sizeof *pd, GFP_KERNEL);
+ pd = kzalloc(sizeof(*pd), GFP_KERNEL);
if (!pd)
return ERR_PTR(-ENOMEM);
@@ -1346,7 +1345,6 @@ static struct ib_pd *mlx4_ib_alloc_pd(struct ib_device *ibdev,
kfree(pd);
return ERR_PTR(-EFAULT);
}
-
return &pd->ibpd;
}
@@ -1860,6 +1858,9 @@ static struct ib_flow *mlx4_ib_create_flow(struct ib_qp *qp,
if (flow_attr->port < 1 || flow_attr->port > qp->device->phys_port_cnt)
return ERR_PTR(-EINVAL);
+ if (flow_attr->flags & ~IB_FLOW_ATTR_FLAGS_DONT_TRAP)
+ return ERR_PTR(-EOPNOTSUPP);
+
if ((flow_attr->flags & IB_FLOW_ATTR_FLAGS_DONT_TRAP) &&
(flow_attr->type != IB_FLOW_ATTR_NORMAL))
return ERR_PTR(-EOPNOTSUPP);
@@ -2933,6 +2934,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
if (mlx4_ib_alloc_diag_counters(ibdev))
goto err_steer_free_bitmap;
+ ibdev->ib_dev.driver_id = RDMA_DRIVER_MLX4;
if (ib_register_device(&ibdev->ib_dev, NULL))
goto err_diag_counters;
diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h
index e14919c15b06..7b1429917aba 100644
--- a/drivers/infiniband/hw/mlx4/mlx4_ib.h
+++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h
@@ -189,6 +189,7 @@ enum mlx4_ib_qp_flags {
MLX4_IB_QP_LSO = IB_QP_CREATE_IPOIB_UD_LSO,
MLX4_IB_QP_BLOCK_MULTICAST_LOOPBACK = IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK,
MLX4_IB_QP_NETIF = IB_QP_CREATE_NETIF_QP,
+ MLX4_IB_QP_SCATTER_FCS = IB_QP_CREATE_SCATTER_FCS,
/* Mellanox specific flags start from IB_QP_CREATE_RESERVED_START */
MLX4_IB_ROCE_V2_GSI_QP = MLX4_IB_QP_CREATE_ROCE_V2_GSI,
@@ -641,24 +642,6 @@ struct mlx4_uverbs_ex_query_device {
__u32 reserved;
};
-enum query_device_resp_mask {
- QUERY_DEVICE_RESP_MASK_TIMESTAMP = 1UL << 0,
-};
-
-struct mlx4_ib_rss_caps {
- __u64 rx_hash_fields_mask; /* enum mlx4_rx_hash_fields */
- __u8 rx_hash_function; /* enum mlx4_rx_hash_function_flags */
- __u8 reserved[7];
-};
-
-struct mlx4_uverbs_ex_query_device_resp {
- __u32 comp_mask;
- __u32 response_length;
- __u64 hca_core_clock_offset;
- __u32 max_inl_recv_sz;
- struct mlx4_ib_rss_caps rss_caps;
-};
-
static inline struct mlx4_ib_dev *to_mdev(struct ib_device *ibdev)
{
return container_of(ibdev, struct mlx4_ib_dev, ib_dev);
diff --git a/drivers/infiniband/hw/mlx4/mr.c b/drivers/infiniband/hw/mlx4/mr.c
index 4975f3e6596e..17f4f151a97f 100644
--- a/drivers/infiniband/hw/mlx4/mr.c
+++ b/drivers/infiniband/hw/mlx4/mr.c
@@ -407,6 +407,9 @@ struct ib_mr *mlx4_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
goto err_mr;
mr->ibmr.rkey = mr->ibmr.lkey = mr->mmr.key;
+ mr->ibmr.length = length;
+ mr->ibmr.iova = virt_addr;
+ mr->ibmr.page_size = 1U << shift;
return &mr->ibmr;
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index f045491f2c14..50af8915e7ec 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -1096,6 +1096,17 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd,
qp->inl_recv_sz = ucmd.qp.inl_recv_sz;
}
+ if (init_attr->create_flags & IB_QP_CREATE_SCATTER_FCS) {
+ if (!(dev->dev->caps.flags &
+ MLX4_DEV_CAP_FLAG_FCS_KEEP)) {
+ pr_debug("scatter FCS is unsupported\n");
+ err = -EOPNOTSUPP;
+ goto err;
+ }
+
+ qp->flags |= MLX4_IB_QP_SCATTER_FCS;
+ }
+
err = set_rq_size(dev, &init_attr->cap, !!pd->uobject,
qp_has_rq(init_attr), qp, qp->inl_recv_sz);
if (err)
@@ -2234,6 +2245,9 @@ static int __mlx4_ib_modify_qp(void *src, enum mlx4_ib_source_type src_type,
if (qp->inl_recv_sz)
context->param3 |= cpu_to_be32(1 << 25);
+ if (qp->flags & MLX4_IB_QP_SCATTER_FCS)
+ context->param3 |= cpu_to_be32(1 << 29);
+
if (qp_type == IB_QPT_GSI || qp_type == IB_QPT_SMI)
context->mtu_msgmax = (IB_MTU_4096 << 5) | 11;
else if (qp_type == IB_QPT_RAW_PACKET)
@@ -2356,9 +2370,7 @@ static int __mlx4_ib_modify_qp(void *src, enum mlx4_ib_source_type src_type,
status = ib_get_cached_gid(&dev->ib_dev, port_num,
index, &gid, &gid_attr);
- if (!status && !memcmp(&gid, &zgid, sizeof(gid)))
- status = -ENOENT;
- if (!status && gid_attr.ndev) {
+ if (!status) {
vlan = rdma_vlan_dev_vlan_id(gid_attr.ndev);
memcpy(smac, gid_attr.ndev->dev_addr, ETH_ALEN);
dev_put(gid_attr.ndev);
@@ -3880,8 +3892,8 @@ out:
*/
wmb();
- writel(qp->doorbell_qpn,
- to_mdev(ibqp->device)->uar_map + MLX4_SEND_DOORBELL);
+ writel_relaxed(qp->doorbell_qpn,
+ to_mdev(ibqp->device)->uar_map + MLX4_SEND_DOORBELL);
/*
* Make sure doorbells don't leak out of SQ spinlock
@@ -4204,7 +4216,7 @@ struct ib_wq *mlx4_ib_create_wq(struct ib_pd *pd,
return ERR_PTR(-EOPNOTSUPP);
}
- if (init_attr->create_flags) {
+ if (init_attr->create_flags & ~IB_WQ_FLAGS_SCATTER_FCS) {
pr_debug("unsupported create_flags %u\n",
init_attr->create_flags);
return ERR_PTR(-EOPNOTSUPP);
@@ -4225,6 +4237,9 @@ struct ib_wq *mlx4_ib_create_wq(struct ib_pd *pd,
ib_qp_init_attr.recv_cq = init_attr->cq;
ib_qp_init_attr.send_cq = ib_qp_init_attr.recv_cq; /* Dummy CQ */
+ if (init_attr->create_flags & IB_WQ_FLAGS_SCATTER_FCS)
+ ib_qp_init_attr.create_flags |= IB_QP_CREATE_SCATTER_FCS;
+
err = create_qp_common(dev, pd, MLX4_IB_RWQ_SRC, &ib_qp_init_attr,
udata, 0, &qp);
if (err) {
diff --git a/drivers/infiniband/hw/mlx5/ah.c b/drivers/infiniband/hw/mlx5/ah.c
index fe269f680103..e6bde32a83f3 100644
--- a/drivers/infiniband/hw/mlx5/ah.c
+++ b/drivers/infiniband/hw/mlx5/ah.c
@@ -36,6 +36,9 @@ static struct ib_ah *create_ib_ah(struct mlx5_ib_dev *dev,
struct mlx5_ib_ah *ah,
struct rdma_ah_attr *ah_attr)
{
+ enum ib_gid_type gid_type;
+ int err;
+
if (rdma_ah_get_ah_flags(ah_attr) & IB_AH_GRH) {
const struct ib_global_route *grh = rdma_ah_read_grh(ah_attr);
@@ -50,6 +53,12 @@ static struct ib_ah *create_ib_ah(struct mlx5_ib_dev *dev,
ah->av.stat_rate_sl = (rdma_ah_get_static_rate(ah_attr) << 4);
if (ah_attr->type == RDMA_AH_ATTR_TYPE_ROCE) {
+ err = mlx5_get_roce_gid_type(dev, ah_attr->port_num,
+ ah_attr->grh.sgid_index,
+ &gid_type);
+ if (err)
+ return ERR_PTR(err);
+
memcpy(ah->av.rmac, ah_attr->roce.dmac,
sizeof(ah_attr->roce.dmac));
ah->av.udp_sport =
@@ -57,6 +66,9 @@ static struct ib_ah *create_ib_ah(struct mlx5_ib_dev *dev,
rdma_ah_get_port_num(ah_attr),
rdma_ah_read_grh(ah_attr)->sgid_index);
ah->av.stat_rate_sl |= (rdma_ah_get_sl(ah_attr) & 0x7) << 1;
+ if (gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP)
+#define MLX5_ECN_ENABLED BIT(1)
+ ah->av.tclass |= MLX5_ECN_ENABLED;
} else {
ah->av.rlid = cpu_to_be16(rdma_ah_get_dlid(ah_attr));
ah->av.fl_mlid = rdma_ah_get_path_bits(ah_attr) & 0x7f;
diff --git a/drivers/infiniband/hw/mlx5/cmd.c b/drivers/infiniband/hw/mlx5/cmd.c
index 6f6712f87a73..188512bf46e6 100644
--- a/drivers/infiniband/hw/mlx5/cmd.c
+++ b/drivers/infiniband/hw/mlx5/cmd.c
@@ -66,3 +66,107 @@ int mlx5_cmd_modify_cong_params(struct mlx5_core_dev *dev,
return mlx5_cmd_exec(dev, in, in_size, out, sizeof(out));
}
+
+int mlx5_cmd_alloc_memic(struct mlx5_memic *memic, phys_addr_t *addr,
+ u64 length, u32 alignment)
+{
+ struct mlx5_core_dev *dev = memic->dev;
+ u64 num_memic_hw_pages = MLX5_CAP_DEV_MEM(dev, memic_bar_size)
+ >> PAGE_SHIFT;
+ u64 hw_start_addr = MLX5_CAP64_DEV_MEM(dev, memic_bar_start_addr);
+ u32 max_alignment = MLX5_CAP_DEV_MEM(dev, log_max_memic_addr_alignment);
+ u32 num_pages = DIV_ROUND_UP(length, PAGE_SIZE);
+ u32 out[MLX5_ST_SZ_DW(alloc_memic_out)] = {};
+ u32 in[MLX5_ST_SZ_DW(alloc_memic_in)] = {};
+ u32 mlx5_alignment;
+ u64 page_idx = 0;
+ int ret = 0;
+
+ if (!length || (length & MLX5_MEMIC_ALLOC_SIZE_MASK))
+ return -EINVAL;
+
+ /* mlx5 device sets alignment as 64*2^driver_value
+ * so normalizing is needed.
+ */
+ mlx5_alignment = (alignment < MLX5_MEMIC_BASE_ALIGN) ? 0 :
+ alignment - MLX5_MEMIC_BASE_ALIGN;
+ if (mlx5_alignment > max_alignment)
+ return -EINVAL;
+
+ MLX5_SET(alloc_memic_in, in, opcode, MLX5_CMD_OP_ALLOC_MEMIC);
+ MLX5_SET(alloc_memic_in, in, range_size, num_pages * PAGE_SIZE);
+ MLX5_SET(alloc_memic_in, in, memic_size, length);
+ MLX5_SET(alloc_memic_in, in, log_memic_addr_alignment,
+ mlx5_alignment);
+
+ while (page_idx < num_memic_hw_pages) {
+ spin_lock(&memic->memic_lock);
+ page_idx = bitmap_find_next_zero_area(memic->memic_alloc_pages,
+ num_memic_hw_pages,
+ page_idx,
+ num_pages, 0);
+
+ if (page_idx < num_memic_hw_pages)
+ bitmap_set(memic->memic_alloc_pages,
+ page_idx, num_pages);
+
+ spin_unlock(&memic->memic_lock);
+
+ if (page_idx >= num_memic_hw_pages)
+ break;
+
+ MLX5_SET64(alloc_memic_in, in, range_start_addr,
+ hw_start_addr + (page_idx * PAGE_SIZE));
+
+ ret = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out));
+ if (ret) {
+ spin_lock(&memic->memic_lock);
+ bitmap_clear(memic->memic_alloc_pages,
+ page_idx, num_pages);
+ spin_unlock(&memic->memic_lock);
+
+ if (ret == -EAGAIN) {
+ page_idx++;
+ continue;
+ }
+
+ return ret;
+ }
+
+ *addr = pci_resource_start(dev->pdev, 0) +
+ MLX5_GET64(alloc_memic_out, out, memic_start_addr);
+
+ return 0;
+ }
+
+ return -ENOMEM;
+}
+
+int mlx5_cmd_dealloc_memic(struct mlx5_memic *memic, u64 addr, u64 length)
+{
+ struct mlx5_core_dev *dev = memic->dev;
+ u64 hw_start_addr = MLX5_CAP64_DEV_MEM(dev, memic_bar_start_addr);
+ u32 num_pages = DIV_ROUND_UP(length, PAGE_SIZE);
+ u32 out[MLX5_ST_SZ_DW(dealloc_memic_out)] = {0};
+ u32 in[MLX5_ST_SZ_DW(dealloc_memic_in)] = {0};
+ u64 start_page_idx;
+ int err;
+
+ addr -= pci_resource_start(dev->pdev, 0);
+ start_page_idx = (addr - hw_start_addr) >> PAGE_SHIFT;
+
+ MLX5_SET(dealloc_memic_in, in, opcode, MLX5_CMD_OP_DEALLOC_MEMIC);
+ MLX5_SET64(dealloc_memic_in, in, memic_start_addr, addr);
+ MLX5_SET(dealloc_memic_in, in, memic_size, length);
+
+ err = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out));
+
+ if (!err) {
+ spin_lock(&memic->memic_lock);
+ bitmap_clear(memic->memic_alloc_pages,
+ start_page_idx, num_pages);
+ spin_unlock(&memic->memic_lock);
+ }
+
+ return err;
+}
diff --git a/drivers/infiniband/hw/mlx5/cmd.h b/drivers/infiniband/hw/mlx5/cmd.h
index 78ffded7cc2c..e7206c8a8011 100644
--- a/drivers/infiniband/hw/mlx5/cmd.h
+++ b/drivers/infiniband/hw/mlx5/cmd.h
@@ -33,6 +33,7 @@
#ifndef MLX5_IB_CMD_H
#define MLX5_IB_CMD_H
+#include "mlx5_ib.h"
#include <linux/kernel.h>
#include <linux/mlx5/driver.h>
@@ -41,4 +42,7 @@ int mlx5_cmd_query_cong_params(struct mlx5_core_dev *dev, int cong_point,
void *out, int out_size);
int mlx5_cmd_modify_cong_params(struct mlx5_core_dev *mdev,
void *in, int in_size);
+int mlx5_cmd_alloc_memic(struct mlx5_memic *memic, phys_addr_t *addr,
+ u64 length, u32 alignment);
+int mlx5_cmd_dealloc_memic(struct mlx5_memic *memic, u64 addr, u64 length);
#endif /* MLX5_IB_CMD_H */
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 071fd9a7b919..daa919e5a442 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -38,6 +38,7 @@
#include <linux/pci.h>
#include <linux/dma-mapping.h>
#include <linux/slab.h>
+#include <linux/bitmap.h>
#if defined(CONFIG_X86)
#include <asm/pat.h>
#endif
@@ -51,6 +52,7 @@
#include <linux/mlx5/port.h>
#include <linux/mlx5/vport.h>
#include <linux/mlx5/fs.h>
+#include <linux/mlx5/fs_helpers.h>
#include <linux/list.h>
#include <rdma/ib_smi.h>
#include <rdma/ib_umem.h>
@@ -60,6 +62,13 @@
#include "ib_rep.h"
#include "cmd.h"
#include <linux/mlx5/fs_helpers.h>
+#include <linux/mlx5/accel.h>
+#include <rdma/uverbs_std_types.h>
+#include <rdma/mlx5_user_ioctl_verbs.h>
+#include <rdma/mlx5_user_ioctl_cmds.h>
+
+#define UVERBS_MODULE_NAME mlx5_ib
+#include <rdma/uverbs_named_ioctl.h>
#define DRIVER_NAME "mlx5_ib"
#define DRIVER_VERSION "5.0-0"
@@ -92,6 +101,12 @@ static LIST_HEAD(mlx5_ib_dev_list);
*/
static DEFINE_MUTEX(mlx5_ib_multiport_mutex);
+/* We can't use an array for xlt_emergency_page because dma_map_single
+ * doesn't work on kernel modules memory
+ */
+static unsigned long xlt_emergency_page;
+static struct mutex xlt_emergency_page_mutex;
+
struct mlx5_ib_dev *mlx5_ib_get_ibdev_from_mpi(struct mlx5_ib_multiport_info *mpi)
{
struct mlx5_ib_dev *dev;
@@ -399,6 +414,9 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num,
if (err)
goto out;
+ props->active_width = IB_WIDTH_4X;
+ props->active_speed = IB_SPEED_QDR;
+
translate_eth_proto_oper(eth_prot_oper, &props->active_speed,
&props->active_width);
@@ -493,18 +511,19 @@ static int set_roce_addr(struct mlx5_ib_dev *dev, u8 port_num,
vlan_id, port_num);
}
-static int mlx5_ib_add_gid(struct ib_device *device, u8 port_num,
- unsigned int index, const union ib_gid *gid,
+static int mlx5_ib_add_gid(const union ib_gid *gid,
const struct ib_gid_attr *attr,
__always_unused void **context)
{
- return set_roce_addr(to_mdev(device), port_num, index, gid, attr);
+ return set_roce_addr(to_mdev(attr->device), attr->port_num,
+ attr->index, gid, attr);
}
-static int mlx5_ib_del_gid(struct ib_device *device, u8 port_num,
- unsigned int index, __always_unused void **context)
+static int mlx5_ib_del_gid(const struct ib_gid_attr *attr,
+ __always_unused void **context)
{
- return set_roce_addr(to_mdev(device), port_num, index, NULL, NULL);
+ return set_roce_addr(to_mdev(attr->device), attr->port_num,
+ attr->index, NULL, NULL);
}
__be16 mlx5_get_roce_udp_sport(struct mlx5_ib_dev *dev, u8 port_num,
@@ -516,9 +535,6 @@ __be16 mlx5_get_roce_udp_sport(struct mlx5_ib_dev *dev, u8 port_num,
if (ib_get_cached_gid(&dev->ib_dev, port_num, index, &gid, &attr))
return 0;
- if (!attr.ndev)
- return 0;
-
dev_put(attr.ndev);
if (attr.gid_type != IB_GID_TYPE_ROCE_UDP_ENCAP)
@@ -538,9 +554,6 @@ int mlx5_get_roce_gid_type(struct mlx5_ib_dev *dev, u8 port_num,
if (ret)
return ret;
- if (!attr.ndev)
- return -ENODEV;
-
dev_put(attr.ndev);
*gid_type = attr.gid_type;
@@ -844,6 +857,10 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
MLX5_RX_HASH_SRC_PORT_UDP |
MLX5_RX_HASH_DST_PORT_UDP |
MLX5_RX_HASH_INNER;
+ if (mlx5_accel_ipsec_device_caps(dev->mdev) &
+ MLX5_ACCEL_IPSEC_CAP_DEVICE)
+ resp.rss_caps.rx_hash_fields_mask |=
+ MLX5_RX_HASH_IPSEC_SPI;
resp.response_length += sizeof(resp.rss_caps);
}
} else {
@@ -875,6 +892,11 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
props->raw_packet_caps |= IB_RAW_PACKET_CAP_SCATTER_FCS;
}
+ if (MLX5_CAP_DEV_MEM(mdev, memic)) {
+ props->max_dm_size =
+ MLX5_CAP_DEV_MEM(mdev, max_memic_size);
+ }
+
if (mlx5_get_flow_namespace(dev->mdev, MLX5_FLOW_NAMESPACE_BYPASS))
props->device_cap_flags |= IB_DEVICE_MANAGED_FLOW_STEERING;
@@ -980,6 +1002,10 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
MLX5_CAP_QOS(mdev, packet_pacing_min_rate);
resp.packet_pacing_caps.supported_qpts |=
1 << IB_QPT_RAW_PACKET;
+ if (MLX5_CAP_QOS(mdev, packet_pacing_burst_bound) &&
+ MLX5_CAP_QOS(mdev, packet_pacing_typical_size))
+ resp.packet_pacing_caps.cap_flags |=
+ MLX5_IB_PP_SUPPORT_BURST;
}
resp.response_length += sizeof(resp.packet_pacing_caps);
}
@@ -1665,6 +1691,18 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
resp.response_length = min(offsetof(typeof(resp), response_length) +
sizeof(resp.response_length), udata->outlen);
+ if (mlx5_accel_ipsec_device_caps(dev->mdev) & MLX5_ACCEL_IPSEC_CAP_DEVICE) {
+ if (mlx5_get_flow_namespace(dev->mdev, MLX5_FLOW_NAMESPACE_EGRESS))
+ resp.flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM;
+ if (mlx5_accel_ipsec_device_caps(dev->mdev) & MLX5_ACCEL_IPSEC_CAP_REQUIRED_METADATA)
+ resp.flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_REQ_METADATA;
+ if (MLX5_CAP_FLOWTABLE(dev->mdev, flow_table_properties_nic_receive.ft_field_support.outer_esp_spi))
+ resp.flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_SPI_STEERING;
+ if (mlx5_accel_ipsec_device_caps(dev->mdev) & MLX5_ACCEL_IPSEC_CAP_TX_IV_IS_ESN)
+ resp.flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_TX_IV_IS_ESN;
+ /* MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_FULL_OFFLOAD is currently always 0 */
+ }
+
context = kzalloc(sizeof(*context), GFP_KERNEL);
if (!context)
return ERR_PTR(-ENOMEM);
@@ -1702,17 +1740,10 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
context->ibucontext.invalidate_range = &mlx5_ib_invalidate_range;
#endif
- context->upd_xlt_page = __get_free_page(GFP_KERNEL);
- if (!context->upd_xlt_page) {
- err = -ENOMEM;
- goto out_uars;
- }
- mutex_init(&context->upd_xlt_page_mutex);
-
if (MLX5_CAP_GEN(dev->mdev, log_max_transport_domain)) {
err = mlx5_ib_alloc_transport_domain(dev, &context->tdn);
if (err)
- goto out_page;
+ goto out_uars;
}
INIT_LIST_HEAD(&context->vma_private_list);
@@ -1789,9 +1820,6 @@ out_td:
if (MLX5_CAP_GEN(dev->mdev, log_max_transport_domain))
mlx5_ib_dealloc_transport_domain(dev, context->tdn);
-out_page:
- free_page(context->upd_xlt_page);
-
out_uars:
deallocate_uars(dev, context);
@@ -1817,7 +1845,6 @@ static int mlx5_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
if (MLX5_CAP_GEN(dev->mdev, log_max_transport_domain))
mlx5_ib_dealloc_transport_domain(dev, context->tdn);
- free_page(context->upd_xlt_page);
deallocate_uars(dev, context);
kfree(bfregi->sys_pages);
kfree(bfregi->count);
@@ -1993,6 +2020,8 @@ static inline char *mmap_cmd2str(enum mlx5_ib_mmap_cmd cmd)
return "best effort WC";
case MLX5_IB_MMAP_NC_PAGE:
return "NC";
+ case MLX5_IB_MMAP_DEVICE_MEM:
+ return "Device Memory";
default:
return NULL;
}
@@ -2151,6 +2180,34 @@ free_bfreg:
return err;
}
+static int dm_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
+{
+ struct mlx5_ib_ucontext *mctx = to_mucontext(context);
+ struct mlx5_ib_dev *dev = to_mdev(context->device);
+ u16 page_idx = get_extended_index(vma->vm_pgoff);
+ size_t map_size = vma->vm_end - vma->vm_start;
+ u32 npages = map_size >> PAGE_SHIFT;
+ phys_addr_t pfn;
+ pgprot_t prot;
+
+ if (find_next_zero_bit(mctx->dm_pages, page_idx + npages, page_idx) !=
+ page_idx + npages)
+ return -EINVAL;
+
+ pfn = ((pci_resource_start(dev->mdev->pdev, 0) +
+ MLX5_CAP64_DEV_MEM(dev->mdev, memic_bar_start_addr)) >>
+ PAGE_SHIFT) +
+ page_idx;
+ prot = pgprot_writecombine(vma->vm_page_prot);
+ vma->vm_page_prot = prot;
+
+ if (io_remap_pfn_range(vma, vma->vm_start, pfn, map_size,
+ vma->vm_page_prot))
+ return -EAGAIN;
+
+ return mlx5_ib_set_vma_data(vma, mctx);
+}
+
static int mlx5_ib_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vma)
{
struct mlx5_ib_ucontext *context = to_mucontext(ibcontext);
@@ -2195,6 +2252,9 @@ static int mlx5_ib_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vm
case MLX5_IB_MMAP_CLOCK_INFO:
return mlx5_ib_mmap_clock_info_page(dev, vma, context);
+ case MLX5_IB_MMAP_DEVICE_MEM:
+ return dm_mmap(ibcontext, vma);
+
default:
return -EINVAL;
}
@@ -2202,6 +2262,87 @@ static int mlx5_ib_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vm
return 0;
}
+struct ib_dm *mlx5_ib_alloc_dm(struct ib_device *ibdev,
+ struct ib_ucontext *context,
+ struct ib_dm_alloc_attr *attr,
+ struct uverbs_attr_bundle *attrs)
+{
+ u64 act_size = roundup(attr->length, MLX5_MEMIC_BASE_SIZE);
+ struct mlx5_memic *memic = &to_mdev(ibdev)->memic;
+ phys_addr_t memic_addr;
+ struct mlx5_ib_dm *dm;
+ u64 start_offset;
+ u32 page_idx;
+ int err;
+
+ dm = kzalloc(sizeof(*dm), GFP_KERNEL);
+ if (!dm)
+ return ERR_PTR(-ENOMEM);
+
+ mlx5_ib_dbg(to_mdev(ibdev), "alloc_memic req: user_length=0x%llx act_length=0x%llx log_alignment=%d\n",
+ attr->length, act_size, attr->alignment);
+
+ err = mlx5_cmd_alloc_memic(memic, &memic_addr,
+ act_size, attr->alignment);
+ if (err)
+ goto err_free;
+
+ start_offset = memic_addr & ~PAGE_MASK;
+ page_idx = (memic_addr - pci_resource_start(memic->dev->pdev, 0) -
+ MLX5_CAP64_DEV_MEM(memic->dev, memic_bar_start_addr)) >>
+ PAGE_SHIFT;
+
+ err = uverbs_copy_to(attrs,
+ MLX5_IB_ATTR_ALLOC_DM_RESP_START_OFFSET,
+ &start_offset, sizeof(start_offset));
+ if (err)
+ goto err_dealloc;
+
+ err = uverbs_copy_to(attrs,
+ MLX5_IB_ATTR_ALLOC_DM_RESP_PAGE_INDEX,
+ &page_idx, sizeof(page_idx));
+ if (err)
+ goto err_dealloc;
+
+ bitmap_set(to_mucontext(context)->dm_pages, page_idx,
+ DIV_ROUND_UP(act_size, PAGE_SIZE));
+
+ dm->dev_addr = memic_addr;
+
+ return &dm->ibdm;
+
+err_dealloc:
+ mlx5_cmd_dealloc_memic(memic, memic_addr,
+ act_size);
+err_free:
+ kfree(dm);
+ return ERR_PTR(err);
+}
+
+int mlx5_ib_dealloc_dm(struct ib_dm *ibdm)
+{
+ struct mlx5_memic *memic = &to_mdev(ibdm->device)->memic;
+ struct mlx5_ib_dm *dm = to_mdm(ibdm);
+ u64 act_size = roundup(dm->ibdm.length, MLX5_MEMIC_BASE_SIZE);
+ u32 page_idx;
+ int ret;
+
+ ret = mlx5_cmd_dealloc_memic(memic, dm->dev_addr, act_size);
+ if (ret)
+ return ret;
+
+ page_idx = (dm->dev_addr - pci_resource_start(memic->dev->pdev, 0) -
+ MLX5_CAP64_DEV_MEM(memic->dev, memic_bar_start_addr)) >>
+ PAGE_SHIFT;
+ bitmap_clear(to_mucontext(ibdm->uobject->context)->dm_pages,
+ page_idx,
+ DIV_ROUND_UP(act_size, PAGE_SIZE));
+
+ kfree(dm);
+
+ return 0;
+}
+
static struct ib_pd *mlx5_ib_alloc_pd(struct ib_device *ibdev,
struct ib_ucontext *context,
struct ib_udata *udata)
@@ -2317,8 +2458,28 @@ static void set_tos(void *outer_c, void *outer_v, u8 mask, u8 val)
offsetof(typeof(filter), field) -\
sizeof(filter.field))
+static int parse_flow_flow_action(const union ib_flow_spec *ib_spec,
+ const struct ib_flow_attr *flow_attr,
+ struct mlx5_flow_act *action)
+{
+ struct mlx5_ib_flow_action *maction = to_mflow_act(ib_spec->action.act);
+
+ switch (maction->ib_action.type) {
+ case IB_FLOW_ACTION_ESP:
+ /* Currently only AES_GCM keymat is supported by the driver */
+ action->esp_id = (uintptr_t)maction->esp_aes_gcm.ctx;
+ action->action |= flow_attr->flags & IB_FLOW_ATTR_FLAGS_EGRESS ?
+ MLX5_FLOW_CONTEXT_ACTION_ENCRYPT :
+ MLX5_FLOW_CONTEXT_ACTION_DECRYPT;
+ return 0;
+ default:
+ return -EOPNOTSUPP;
+ }
+}
+
static int parse_flow_attr(struct mlx5_core_dev *mdev, u32 *match_c,
u32 *match_v, const union ib_flow_spec *ib_spec,
+ const struct ib_flow_attr *flow_attr,
struct mlx5_flow_act *action)
{
void *misc_params_c = MLX5_ADDR_OF(fte_match_param, match_c,
@@ -2328,6 +2489,7 @@ static int parse_flow_attr(struct mlx5_core_dev *mdev, u32 *match_c,
void *headers_c;
void *headers_v;
int match_ipv;
+ int ret;
if (ib_spec->type & IB_FLOW_SPEC_INNER) {
headers_c = MLX5_ADDR_OF(fte_match_param, match_c,
@@ -2478,7 +2640,15 @@ static int parse_flow_attr(struct mlx5_core_dev *mdev, u32 *match_c,
ntohl(ib_spec->ipv6.mask.flow_label),
ntohl(ib_spec->ipv6.val.flow_label),
ib_spec->type & IB_FLOW_SPEC_INNER);
+ break;
+ case IB_FLOW_SPEC_ESP:
+ if (ib_spec->esp.mask.seq)
+ return -EOPNOTSUPP;
+ MLX5_SET(fte_match_set_misc, misc_params_c, outer_esp_spi,
+ ntohl(ib_spec->esp.mask.spi));
+ MLX5_SET(fte_match_set_misc, misc_params_v, outer_esp_spi,
+ ntohl(ib_spec->esp.val.spi));
break;
case IB_FLOW_SPEC_TCP:
if (FIELDS_NOT_SUPPORTED(ib_spec->tcp_udp.mask,
@@ -2546,6 +2716,11 @@ static int parse_flow_attr(struct mlx5_core_dev *mdev, u32 *match_c,
return -EOPNOTSUPP;
action->action |= MLX5_FLOW_CONTEXT_ACTION_DROP;
break;
+ case IB_FLOW_SPEC_ACTION_HANDLE:
+ ret = parse_flow_flow_action(ib_spec, flow_attr, action);
+ if (ret)
+ return ret;
+ break;
default:
return -EINVAL;
}
@@ -2587,6 +2762,46 @@ static bool flow_is_multicast_only(const struct ib_flow_attr *ib_attr)
return false;
}
+enum valid_spec {
+ VALID_SPEC_INVALID,
+ VALID_SPEC_VALID,
+ VALID_SPEC_NA,
+};
+
+static enum valid_spec
+is_valid_esp_aes_gcm(struct mlx5_core_dev *mdev,
+ const struct mlx5_flow_spec *spec,
+ const struct mlx5_flow_act *flow_act,
+ bool egress)
+{
+ const u32 *match_c = spec->match_criteria;
+ bool is_crypto =
+ (flow_act->action & (MLX5_FLOW_CONTEXT_ACTION_ENCRYPT |
+ MLX5_FLOW_CONTEXT_ACTION_DECRYPT));
+ bool is_ipsec = mlx5_fs_is_ipsec_flow(match_c);
+ bool is_drop = flow_act->action & MLX5_FLOW_CONTEXT_ACTION_DROP;
+
+ /*
+ * Currently only crypto is supported in egress, when regular egress
+ * rules would be supported, always return VALID_SPEC_NA.
+ */
+ if (!is_crypto)
+ return egress ? VALID_SPEC_INVALID : VALID_SPEC_NA;
+
+ return is_crypto && is_ipsec &&
+ (!egress || (!is_drop && !flow_act->has_flow_tag)) ?
+ VALID_SPEC_VALID : VALID_SPEC_INVALID;
+}
+
+static bool is_valid_spec(struct mlx5_core_dev *mdev,
+ const struct mlx5_flow_spec *spec,
+ const struct mlx5_flow_act *flow_act,
+ bool egress)
+{
+ /* We curretly only support ipsec egress flow */
+ return is_valid_esp_aes_gcm(mdev, spec, flow_act, egress) != VALID_SPEC_INVALID;
+}
+
static bool is_valid_ethertype(struct mlx5_core_dev *mdev,
const struct ib_flow_attr *flow_attr,
bool check_inner)
@@ -2711,13 +2926,17 @@ static struct mlx5_ib_flow_prio *get_flow_table(struct mlx5_ib_dev *dev,
max_table_size = BIT(MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev,
log_max_ft_size));
if (flow_attr->type == IB_FLOW_ATTR_NORMAL) {
- if (flow_is_multicast_only(flow_attr) &&
- !dont_trap)
+ if (ft_type == MLX5_IB_FT_TX)
+ priority = 0;
+ else if (flow_is_multicast_only(flow_attr) &&
+ !dont_trap)
priority = MLX5_IB_FLOW_MCAST_PRIO;
else
priority = ib_prio_to_core_prio(flow_attr->priority,
dont_trap);
ns = mlx5_get_flow_namespace(dev->mdev,
+ ft_type == MLX5_IB_FT_TX ?
+ MLX5_FLOW_NAMESPACE_EGRESS :
MLX5_FLOW_NAMESPACE_BYPASS);
num_entries = MLX5_FS_MAX_ENTRIES;
num_groups = MLX5_FS_MAX_TYPES;
@@ -2804,6 +3023,7 @@ static struct mlx5_ib_flow_handler *_create_flow_rule(struct mlx5_ib_dev *dev,
unsigned int spec_index;
int err = 0;
int dest_num = 1;
+ bool is_egress = flow_attr->flags & IB_FLOW_ATTR_FLAGS_EGRESS;
if (!is_valid_attr(dev->mdev, flow_attr))
return ERR_PTR(-EINVAL);
@@ -2820,7 +3040,7 @@ static struct mlx5_ib_flow_handler *_create_flow_rule(struct mlx5_ib_dev *dev,
for (spec_index = 0; spec_index < flow_attr->num_of_specs; spec_index++) {
err = parse_flow_attr(dev->mdev, spec->match_criteria,
spec->match_value,
- ib_flow, &flow_act);
+ ib_flow, flow_attr, &flow_act);
if (err < 0)
goto free;
@@ -2843,12 +3063,23 @@ static struct mlx5_ib_flow_handler *_create_flow_rule(struct mlx5_ib_dev *dev,
}
spec->match_criteria_enable = get_match_criteria_enable(spec->match_criteria);
+
+ if (is_egress &&
+ !is_valid_spec(dev->mdev, spec, &flow_act, is_egress)) {
+ err = -EINVAL;
+ goto free;
+ }
+
if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_DROP) {
rule_dst = NULL;
dest_num = 0;
} else {
- flow_act.action = dst ? MLX5_FLOW_CONTEXT_ACTION_FWD_DEST :
- MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO;
+ if (is_egress)
+ flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_ALLOW;
+ else
+ flow_act.action |=
+ dst ? MLX5_FLOW_CONTEXT_ACTION_FWD_DEST :
+ MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO;
}
if (flow_act.has_flow_tag &&
@@ -3022,6 +3253,7 @@ static struct ib_flow *mlx5_ib_create_flow(struct ib_qp *qp,
struct mlx5_flow_destination *dst = NULL;
struct mlx5_ib_flow_prio *ft_prio_tx = NULL;
struct mlx5_ib_flow_prio *ft_prio;
+ bool is_egress = flow_attr->flags & IB_FLOW_ATTR_FLAGS_EGRESS;
int err;
int underlay_qpn;
@@ -3030,7 +3262,13 @@ static struct ib_flow *mlx5_ib_create_flow(struct ib_qp *qp,
if (domain != IB_FLOW_DOMAIN_USER ||
flow_attr->port > dev->num_ports ||
- (flow_attr->flags & ~IB_FLOW_ATTR_FLAGS_DONT_TRAP))
+ (flow_attr->flags & ~(IB_FLOW_ATTR_FLAGS_DONT_TRAP |
+ IB_FLOW_ATTR_FLAGS_EGRESS)))
+ return ERR_PTR(-EINVAL);
+
+ if (is_egress &&
+ (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT ||
+ flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT))
return ERR_PTR(-EINVAL);
dst = kzalloc(sizeof(*dst), GFP_KERNEL);
@@ -3039,7 +3277,8 @@ static struct ib_flow *mlx5_ib_create_flow(struct ib_qp *qp,
mutex_lock(&dev->flow_db->lock);
- ft_prio = get_flow_table(dev, flow_attr, MLX5_IB_FT_RX);
+ ft_prio = get_flow_table(dev, flow_attr,
+ is_egress ? MLX5_IB_FT_TX : MLX5_IB_FT_RX);
if (IS_ERR(ft_prio)) {
err = PTR_ERR(ft_prio);
goto unlock;
@@ -3053,11 +3292,15 @@ static struct ib_flow *mlx5_ib_create_flow(struct ib_qp *qp,
}
}
- dst->type = MLX5_FLOW_DESTINATION_TYPE_TIR;
- if (mqp->flags & MLX5_IB_QP_RSS)
- dst->tir_num = mqp->rss_qp.tirn;
- else
- dst->tir_num = mqp->raw_packet_qp.rq.tirn;
+ if (is_egress) {
+ dst->type = MLX5_FLOW_DESTINATION_TYPE_PORT;
+ } else {
+ dst->type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+ if (mqp->flags & MLX5_IB_QP_RSS)
+ dst->tir_num = mqp->rss_qp.tirn;
+ else
+ dst->tir_num = mqp->raw_packet_qp.rq.tirn;
+ }
if (flow_attr->type == IB_FLOW_ATTR_NORMAL) {
if (flow_attr->flags & IB_FLOW_ATTR_FLAGS_DONT_TRAP) {
@@ -3102,6 +3345,170 @@ unlock:
return ERR_PTR(err);
}
+static u32 mlx5_ib_flow_action_flags_to_accel_xfrm_flags(u32 mlx5_flags)
+{
+ u32 flags = 0;
+
+ if (mlx5_flags & MLX5_IB_UAPI_FLOW_ACTION_FLAGS_REQUIRE_METADATA)
+ flags |= MLX5_ACCEL_XFRM_FLAG_REQUIRE_METADATA;
+
+ return flags;
+}
+
+#define MLX5_FLOW_ACTION_ESP_CREATE_LAST_SUPPORTED MLX5_IB_UAPI_FLOW_ACTION_FLAGS_REQUIRE_METADATA
+static struct ib_flow_action *
+mlx5_ib_create_flow_action_esp(struct ib_device *device,
+ const struct ib_flow_action_attrs_esp *attr,
+ struct uverbs_attr_bundle *attrs)
+{
+ struct mlx5_ib_dev *mdev = to_mdev(device);
+ struct ib_uverbs_flow_action_esp_keymat_aes_gcm *aes_gcm;
+ struct mlx5_accel_esp_xfrm_attrs accel_attrs = {};
+ struct mlx5_ib_flow_action *action;
+ u64 action_flags;
+ u64 flags;
+ int err = 0;
+
+ if (IS_UVERBS_COPY_ERR(uverbs_copy_from(&action_flags, attrs,
+ MLX5_IB_ATTR_CREATE_FLOW_ACTION_FLAGS)))
+ return ERR_PTR(-EFAULT);
+
+ if (action_flags >= (MLX5_FLOW_ACTION_ESP_CREATE_LAST_SUPPORTED << 1))
+ return ERR_PTR(-EOPNOTSUPP);
+
+ flags = mlx5_ib_flow_action_flags_to_accel_xfrm_flags(action_flags);
+
+ /* We current only support a subset of the standard features. Only a
+ * keymat of type AES_GCM, with icv_len == 16, iv_algo == SEQ and esn
+ * (with overlap). Full offload mode isn't supported.
+ */
+ if (!attr->keymat || attr->replay || attr->encap ||
+ attr->spi || attr->seq || attr->tfc_pad ||
+ attr->hard_limit_pkts ||
+ (attr->flags & ~(IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED |
+ IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ENCRYPT)))
+ return ERR_PTR(-EOPNOTSUPP);
+
+ if (attr->keymat->protocol !=
+ IB_UVERBS_FLOW_ACTION_ESP_KEYMAT_AES_GCM)
+ return ERR_PTR(-EOPNOTSUPP);
+
+ aes_gcm = &attr->keymat->keymat.aes_gcm;
+
+ if (aes_gcm->icv_len != 16 ||
+ aes_gcm->iv_algo != IB_UVERBS_FLOW_ACTION_IV_ALGO_SEQ)
+ return ERR_PTR(-EOPNOTSUPP);
+
+ action = kmalloc(sizeof(*action), GFP_KERNEL);
+ if (!action)
+ return ERR_PTR(-ENOMEM);
+
+ action->esp_aes_gcm.ib_flags = attr->flags;
+ memcpy(&accel_attrs.keymat.aes_gcm.aes_key, &aes_gcm->aes_key,
+ sizeof(accel_attrs.keymat.aes_gcm.aes_key));
+ accel_attrs.keymat.aes_gcm.key_len = aes_gcm->key_len * 8;
+ memcpy(&accel_attrs.keymat.aes_gcm.salt, &aes_gcm->salt,
+ sizeof(accel_attrs.keymat.aes_gcm.salt));
+ memcpy(&accel_attrs.keymat.aes_gcm.seq_iv, &aes_gcm->iv,
+ sizeof(accel_attrs.keymat.aes_gcm.seq_iv));
+ accel_attrs.keymat.aes_gcm.icv_len = aes_gcm->icv_len * 8;
+ accel_attrs.keymat.aes_gcm.iv_algo = MLX5_ACCEL_ESP_AES_GCM_IV_ALGO_SEQ;
+ accel_attrs.keymat_type = MLX5_ACCEL_ESP_KEYMAT_AES_GCM;
+
+ accel_attrs.esn = attr->esn;
+ if (attr->flags & IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED)
+ accel_attrs.flags |= MLX5_ACCEL_ESP_FLAGS_ESN_TRIGGERED;
+ if (attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW)
+ accel_attrs.flags |= MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP;
+
+ if (attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ENCRYPT)
+ accel_attrs.action |= MLX5_ACCEL_ESP_ACTION_ENCRYPT;
+
+ action->esp_aes_gcm.ctx =
+ mlx5_accel_esp_create_xfrm(mdev->mdev, &accel_attrs, flags);
+ if (IS_ERR(action->esp_aes_gcm.ctx)) {
+ err = PTR_ERR(action->esp_aes_gcm.ctx);
+ goto err_parse;
+ }
+
+ action->esp_aes_gcm.ib_flags = attr->flags;
+
+ return &action->ib_action;
+
+err_parse:
+ kfree(action);
+ return ERR_PTR(err);
+}
+
+static int
+mlx5_ib_modify_flow_action_esp(struct ib_flow_action *action,
+ const struct ib_flow_action_attrs_esp *attr,
+ struct uverbs_attr_bundle *attrs)
+{
+ struct mlx5_ib_flow_action *maction = to_mflow_act(action);
+ struct mlx5_accel_esp_xfrm_attrs accel_attrs;
+ int err = 0;
+
+ if (attr->keymat || attr->replay || attr->encap ||
+ attr->spi || attr->seq || attr->tfc_pad ||
+ attr->hard_limit_pkts ||
+ (attr->flags & ~(IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED |
+ IB_FLOW_ACTION_ESP_FLAGS_MOD_ESP_ATTRS |
+ IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW)))
+ return -EOPNOTSUPP;
+
+ /* Only the ESN value or the MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP can
+ * be modified.
+ */
+ if (!(maction->esp_aes_gcm.ib_flags &
+ IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED) &&
+ attr->flags & (IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED |
+ IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW))
+ return -EINVAL;
+
+ memcpy(&accel_attrs, &maction->esp_aes_gcm.ctx->attrs,
+ sizeof(accel_attrs));
+
+ accel_attrs.esn = attr->esn;
+ if (attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW)
+ accel_attrs.flags |= MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP;
+ else
+ accel_attrs.flags &= ~MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP;
+
+ err = mlx5_accel_esp_modify_xfrm(maction->esp_aes_gcm.ctx,
+ &accel_attrs);
+ if (err)
+ return err;
+
+ maction->esp_aes_gcm.ib_flags &=
+ ~IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW;
+ maction->esp_aes_gcm.ib_flags |=
+ attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW;
+
+ return 0;
+}
+
+static int mlx5_ib_destroy_flow_action(struct ib_flow_action *action)
+{
+ struct mlx5_ib_flow_action *maction = to_mflow_act(action);
+
+ switch (action->type) {
+ case IB_FLOW_ACTION_ESP:
+ /*
+ * We only support aes_gcm by now, so we implicitly know this is
+ * the underline crypto.
+ */
+ mlx5_accel_esp_destroy_xfrm(maction->esp_aes_gcm.ctx);
+ break;
+ default:
+ WARN_ON(true);
+ break;
+ }
+
+ kfree(maction);
+ return 0;
+}
+
static int mlx5_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
{
struct mlx5_ib_dev *dev = to_mdev(ibqp->device);
@@ -4553,6 +4960,47 @@ static void mlx5_ib_cleanup_multiport_master(struct mlx5_ib_dev *dev)
mlx5_nic_vport_disable_roce(dev->mdev);
}
+ADD_UVERBS_ATTRIBUTES_SIMPLE(mlx5_ib_dm, UVERBS_OBJECT_DM,
+ UVERBS_METHOD_DM_ALLOC,
+ &UVERBS_ATTR_PTR_OUT(MLX5_IB_ATTR_ALLOC_DM_RESP_START_OFFSET,
+ UVERBS_ATTR_TYPE(u64),
+ UA_FLAGS(UVERBS_ATTR_SPEC_F_MANDATORY)),
+ &UVERBS_ATTR_PTR_OUT(MLX5_IB_ATTR_ALLOC_DM_RESP_PAGE_INDEX,
+ UVERBS_ATTR_TYPE(u16),
+ UA_FLAGS(UVERBS_ATTR_SPEC_F_MANDATORY)));
+
+ADD_UVERBS_ATTRIBUTES_SIMPLE(mlx5_ib_flow_action, UVERBS_OBJECT_FLOW_ACTION,
+ UVERBS_METHOD_FLOW_ACTION_ESP_CREATE,
+ &UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_CREATE_FLOW_ACTION_FLAGS,
+ UVERBS_ATTR_TYPE(u64),
+ UA_FLAGS(UVERBS_ATTR_SPEC_F_MANDATORY)));
+
+#define NUM_TREES 2
+static int populate_specs_root(struct mlx5_ib_dev *dev)
+{
+ const struct uverbs_object_tree_def *default_root[NUM_TREES + 1] = {
+ uverbs_default_get_objects()};
+ size_t num_trees = 1;
+
+ if (mlx5_accel_ipsec_device_caps(dev->mdev) & MLX5_ACCEL_IPSEC_CAP_DEVICE &&
+ !WARN_ON(num_trees >= ARRAY_SIZE(default_root)))
+ default_root[num_trees++] = &mlx5_ib_flow_action;
+
+ if (MLX5_CAP_DEV_MEM(dev->mdev, memic) &&
+ !WARN_ON(num_trees >= ARRAY_SIZE(default_root)))
+ default_root[num_trees++] = &mlx5_ib_dm;
+
+ dev->ib_dev.specs_root =
+ uverbs_alloc_spec_tree(num_trees, default_root);
+
+ return PTR_ERR_OR_ZERO(dev->ib_dev.specs_root);
+}
+
+static void depopulate_specs_root(struct mlx5_ib_dev *dev)
+{
+ uverbs_free_spec_tree(dev->ib_dev.specs_root);
+}
+
void mlx5_ib_stage_init_cleanup(struct mlx5_ib_dev *dev)
{
mlx5_ib_cleanup_multiport_master(dev);
@@ -4616,6 +5064,9 @@ int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev)
INIT_LIST_HEAD(&dev->qp_list);
spin_lock_init(&dev->reset_flow_resource_lock);
+ spin_lock_init(&dev->memic.memic_lock);
+ dev->memic.dev = mdev;
+
#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
err = init_srcu_struct(&dev->mr_srcu);
if (err)
@@ -4778,11 +5229,21 @@ int mlx5_ib_stage_caps_init(struct mlx5_ib_dev *dev)
(1ull << IB_USER_VERBS_CMD_CLOSE_XRCD);
}
+ if (MLX5_CAP_DEV_MEM(mdev, memic)) {
+ dev->ib_dev.alloc_dm = mlx5_ib_alloc_dm;
+ dev->ib_dev.dealloc_dm = mlx5_ib_dealloc_dm;
+ dev->ib_dev.reg_dm_mr = mlx5_ib_reg_dm_mr;
+ }
+
dev->ib_dev.create_flow = mlx5_ib_create_flow;
dev->ib_dev.destroy_flow = mlx5_ib_destroy_flow;
dev->ib_dev.uverbs_ex_cmd_mask |=
(1ull << IB_USER_VERBS_EX_CMD_CREATE_FLOW) |
(1ull << IB_USER_VERBS_EX_CMD_DESTROY_FLOW);
+ dev->ib_dev.create_flow_action_esp = mlx5_ib_create_flow_action_esp;
+ dev->ib_dev.destroy_flow_action = mlx5_ib_destroy_flow_action;
+ dev->ib_dev.modify_flow_action_esp = mlx5_ib_modify_flow_action_esp;
+ dev->ib_dev.driver_id = RDMA_DRIVER_MLX5;
err = init_node_data(dev);
if (err)
@@ -4997,11 +5458,21 @@ void mlx5_ib_stage_bfrag_cleanup(struct mlx5_ib_dev *dev)
mlx5_free_bfreg(dev->mdev, &dev->bfreg);
}
+static int mlx5_ib_stage_populate_specs(struct mlx5_ib_dev *dev)
+{
+ return populate_specs_root(dev);
+}
+
int mlx5_ib_stage_ib_reg_init(struct mlx5_ib_dev *dev)
{
return ib_register_device(&dev->ib_dev, NULL);
}
+static void mlx5_ib_stage_depopulate_specs(struct mlx5_ib_dev *dev)
+{
+ depopulate_specs_root(dev);
+}
+
void mlx5_ib_stage_pre_ib_reg_umr_cleanup(struct mlx5_ib_dev *dev)
{
destroy_umrc_res(dev);
@@ -5136,6 +5607,9 @@ static const struct mlx5_ib_profile pf_profile = {
STAGE_CREATE(MLX5_IB_STAGE_PRE_IB_REG_UMR,
NULL,
mlx5_ib_stage_pre_ib_reg_umr_cleanup),
+ STAGE_CREATE(MLX5_IB_STAGE_SPECS,
+ mlx5_ib_stage_populate_specs,
+ mlx5_ib_stage_depopulate_specs),
STAGE_CREATE(MLX5_IB_STAGE_IB_REG,
mlx5_ib_stage_ib_reg_init,
mlx5_ib_stage_ib_reg_cleanup),
@@ -5181,6 +5655,9 @@ static const struct mlx5_ib_profile nic_rep_profile = {
STAGE_CREATE(MLX5_IB_STAGE_PRE_IB_REG_UMR,
NULL,
mlx5_ib_stage_pre_ib_reg_umr_cleanup),
+ STAGE_CREATE(MLX5_IB_STAGE_SPECS,
+ mlx5_ib_stage_populate_specs,
+ mlx5_ib_stage_depopulate_specs),
STAGE_CREATE(MLX5_IB_STAGE_IB_REG,
mlx5_ib_stage_ib_reg_init,
mlx5_ib_stage_ib_reg_cleanup),
@@ -5301,13 +5778,32 @@ static struct mlx5_interface mlx5_ib_interface = {
.protocol = MLX5_INTERFACE_PROTOCOL_IB,
};
+unsigned long mlx5_ib_get_xlt_emergency_page(void)
+{
+ mutex_lock(&xlt_emergency_page_mutex);
+ return xlt_emergency_page;
+}
+
+void mlx5_ib_put_xlt_emergency_page(void)
+{
+ mutex_unlock(&xlt_emergency_page_mutex);
+}
+
static int __init mlx5_ib_init(void)
{
int err;
+ xlt_emergency_page = __get_free_page(GFP_KERNEL);
+ if (!xlt_emergency_page)
+ return -ENOMEM;
+
+ mutex_init(&xlt_emergency_page_mutex);
+
mlx5_ib_event_wq = alloc_ordered_workqueue("mlx5_ib_event_wq", 0);
- if (!mlx5_ib_event_wq)
+ if (!mlx5_ib_event_wq) {
+ free_page(xlt_emergency_page);
return -ENOMEM;
+ }
mlx5_ib_odp_init();
@@ -5320,6 +5816,8 @@ static void __exit mlx5_ib_cleanup(void)
{
mlx5_unregister_interface(&mlx5_ib_interface);
destroy_workqueue(mlx5_ib_event_wq);
+ mutex_destroy(&xlt_emergency_page_mutex);
+ free_page(xlt_emergency_page);
}
module_init(mlx5_ib_init);
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index c33bf1523d67..49a1aa0ff429 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -45,6 +45,7 @@
#include <linux/mlx5/transobj.h>
#include <rdma/ib_user_verbs.h>
#include <rdma/mlx5-abi.h>
+#include <rdma/uverbs_ioctl.h>
#define mlx5_ib_dbg(dev, format, arg...) \
pr_debug("%s:%s:%d:(pid %d): " format, (dev)->ib_dev.name, __func__, \
@@ -108,6 +109,16 @@ enum {
MLX5_IB_INVALID_BFREG = BIT(31),
};
+enum {
+ MLX5_MAX_MEMIC_PAGES = 0x100,
+ MLX5_MEMIC_ALLOC_SIZE_MASK = 0x3f,
+};
+
+enum {
+ MLX5_MEMIC_BASE_ALIGN = 6,
+ MLX5_MEMIC_BASE_SIZE = 1 << MLX5_MEMIC_BASE_ALIGN,
+};
+
struct mlx5_ib_vma_private_data {
struct list_head list;
struct vm_area_struct *vma;
@@ -130,10 +141,8 @@ struct mlx5_ib_ucontext {
/* protect vma_private_list add/del */
struct mutex vma_private_list_mutex;
- unsigned long upd_xlt_page;
- /* protect ODP/KSM */
- struct mutex upd_xlt_page_mutex;
u64 lib_caps;
+ DECLARE_BITMAP(dm_pages, MLX5_MAX_MEMIC_PAGES);
};
static inline struct mlx5_ib_ucontext *to_mucontext(struct ib_ucontext *ibucontext)
@@ -155,6 +164,7 @@ struct mlx5_ib_pd {
#define MLX5_IB_NUM_FLOW_FT (MLX5_IB_FLOW_LEFTOVERS_PRIO + 1)
#define MLX5_IB_NUM_SNIFFER_FTS 2
+#define MLX5_IB_NUM_EGRESS_FTS 1
struct mlx5_ib_flow_prio {
struct mlx5_flow_table *flow_table;
unsigned int refcount;
@@ -170,6 +180,7 @@ struct mlx5_ib_flow_handler {
struct mlx5_ib_flow_db {
struct mlx5_ib_flow_prio prios[MLX5_IB_NUM_FLOW_FT];
struct mlx5_ib_flow_prio sniffer[MLX5_IB_NUM_SNIFFER_FTS];
+ struct mlx5_ib_flow_prio egress[MLX5_IB_NUM_EGRESS_FTS];
struct mlx5_flow_table *lag_demux_ft;
/* Protect flow steering bypass flow tables
* when add/del flow rules.
@@ -406,7 +417,7 @@ struct mlx5_ib_qp {
struct list_head qps_list;
struct list_head cq_recv_list;
struct list_head cq_send_list;
- u32 rate_limit;
+ struct mlx5_rate_limit rl;
u32 underlay_qpn;
bool tunnel_offload_en;
/* storage for qp sub type when core qp type is IB_QPT_DRIVER */
@@ -522,8 +533,19 @@ enum mlx5_ib_mtt_access_flags {
MLX5_IB_MTT_WRITE = (1 << 1),
};
+struct mlx5_ib_dm {
+ struct ib_dm ibdm;
+ phys_addr_t dev_addr;
+};
+
#define MLX5_IB_MTT_PRESENT (MLX5_IB_MTT_READ | MLX5_IB_MTT_WRITE)
+#define MLX5_IB_DM_ALLOWED_ACCESS (IB_ACCESS_LOCAL_WRITE |\
+ IB_ACCESS_REMOTE_WRITE |\
+ IB_ACCESS_REMOTE_READ |\
+ IB_ACCESS_REMOTE_ATOMIC |\
+ IB_ZERO_BASED)
+
struct mlx5_ib_mr {
struct ib_mr ibmr;
void *descs;
@@ -743,6 +765,7 @@ enum mlx5_ib_stages {
MLX5_IB_STAGE_UAR,
MLX5_IB_STAGE_BFREG,
MLX5_IB_STAGE_PRE_IB_REG_UMR,
+ MLX5_IB_STAGE_SPECS,
MLX5_IB_STAGE_IB_REG,
MLX5_IB_STAGE_POST_IB_REG_UMR,
MLX5_IB_STAGE_DELAY_DROP,
@@ -774,6 +797,22 @@ struct mlx5_ib_multiport_info {
bool unaffiliate;
};
+struct mlx5_ib_flow_action {
+ struct ib_flow_action ib_action;
+ union {
+ struct {
+ u64 ib_flags;
+ struct mlx5_accel_esp_xfrm *ctx;
+ } esp_aes_gcm;
+ };
+};
+
+struct mlx5_memic {
+ struct mlx5_core_dev *dev;
+ spinlock_t memic_lock;
+ DECLARE_BITMAP(memic_alloc_pages, MLX5_MAX_MEMIC_PAGES);
+};
+
struct mlx5_ib_dev {
struct ib_device ib_dev;
struct mlx5_core_dev *mdev;
@@ -820,6 +859,7 @@ struct mlx5_ib_dev {
u8 umr_fence;
struct list_head ib_dev_list;
u64 sys_image_guid;
+ struct mlx5_memic memic;
};
static inline struct mlx5_ib_cq *to_mibcq(struct mlx5_core_cq *mcq)
@@ -887,6 +927,11 @@ static inline struct mlx5_ib_srq *to_mibsrq(struct mlx5_core_srq *msrq)
return container_of(msrq, struct mlx5_ib_srq, msrq);
}
+static inline struct mlx5_ib_dm *to_mdm(struct ib_dm *ibdm)
+{
+ return container_of(ibdm, struct mlx5_ib_dm, ibdm);
+}
+
static inline struct mlx5_ib_mr *to_mmr(struct ib_mr *ibmr)
{
return container_of(ibmr, struct mlx5_ib_mr, ibmr);
@@ -897,6 +942,12 @@ static inline struct mlx5_ib_mw *to_mmw(struct ib_mw *ibmw)
return container_of(ibmw, struct mlx5_ib_mw, ibmw);
}
+static inline struct mlx5_ib_flow_action *
+to_mflow_act(struct ib_flow_action *ibact)
+{
+ return container_of(ibact, struct mlx5_ib_flow_action, ib_action);
+}
+
int mlx5_ib_db_map_user(struct mlx5_ib_ucontext *context, unsigned long virt,
struct mlx5_db *db);
void mlx5_ib_db_unmap_user(struct mlx5_ib_ucontext *context, struct mlx5_db *db);
@@ -1025,7 +1076,14 @@ struct ib_rwq_ind_table *mlx5_ib_create_rwq_ind_table(struct ib_device *device,
struct ib_udata *udata);
int mlx5_ib_destroy_rwq_ind_table(struct ib_rwq_ind_table *wq_ind_table);
bool mlx5_ib_dc_atomic_is_supported(struct mlx5_ib_dev *dev);
-
+struct ib_dm *mlx5_ib_alloc_dm(struct ib_device *ibdev,
+ struct ib_ucontext *context,
+ struct ib_dm_alloc_attr *attr,
+ struct uverbs_attr_bundle *attrs);
+int mlx5_ib_dealloc_dm(struct ib_dm *ibdm);
+struct ib_mr *mlx5_ib_reg_dm_mr(struct ib_pd *pd, struct ib_dm *dm,
+ struct ib_dm_mr_attr *attr,
+ struct uverbs_attr_bundle *attrs);
#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev);
@@ -1221,4 +1279,7 @@ static inline int get_num_static_uars(struct mlx5_ib_dev *dev,
return get_uars_per_sys_page(dev, bfregi->lib_uar_4k) * bfregi->num_static_sys_pages;
}
+unsigned long mlx5_ib_get_xlt_emergency_page(void);
+void mlx5_ib_put_xlt_emergency_page(void);
+
#endif /* MLX5_IB_H */
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index 654bc31bc428..1520a2f20f98 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -47,10 +47,25 @@ enum {
#define MLX5_UMR_ALIGN 2048
-static int clean_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr);
-static int dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr);
+static void clean_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr);
+static void dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr);
static int mr_cache_max_order(struct mlx5_ib_dev *dev);
static int unreg_umr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr);
+static bool umr_can_modify_entity_size(struct mlx5_ib_dev *dev)
+{
+ return !MLX5_CAP_GEN(dev->mdev, umr_modify_entity_size_disabled);
+}
+
+static bool umr_can_use_indirect_mkey(struct mlx5_ib_dev *dev)
+{
+ return !MLX5_CAP_GEN(dev->mdev, umr_indirect_mkey_disabled);
+}
+
+static bool use_umr(struct mlx5_ib_dev *dev, int order)
+{
+ return order <= mr_cache_max_order(dev) &&
+ umr_can_modify_entity_size(dev);
+}
static int destroy_mkey(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
{
@@ -189,7 +204,9 @@ static int add_keys(struct mlx5_ib_dev *dev, int c, int num)
MLX5_SET(mkc, mkc, free, 1);
MLX5_SET(mkc, mkc, umr_en, 1);
- MLX5_SET(mkc, mkc, access_mode, ent->access_mode);
+ MLX5_SET(mkc, mkc, access_mode_1_0, ent->access_mode & 0x3);
+ MLX5_SET(mkc, mkc, access_mode_4_2,
+ (ent->access_mode >> 2) & 0x7);
MLX5_SET(mkc, mkc, qpn, 0xffffff);
MLX5_SET(mkc, mkc, translations_octword_size, ent->xlt);
@@ -220,26 +237,32 @@ static void remove_keys(struct mlx5_ib_dev *dev, int c, int num)
{
struct mlx5_mr_cache *cache = &dev->cache;
struct mlx5_cache_ent *ent = &cache->ent[c];
+ struct mlx5_ib_mr *tmp_mr;
struct mlx5_ib_mr *mr;
- int err;
+ LIST_HEAD(del_list);
int i;
for (i = 0; i < num; i++) {
spin_lock_irq(&ent->lock);
if (list_empty(&ent->head)) {
spin_unlock_irq(&ent->lock);
- return;
+ break;
}
mr = list_first_entry(&ent->head, struct mlx5_ib_mr, list);
- list_del(&mr->list);
+ list_move(&mr->list, &del_list);
ent->cur--;
ent->size--;
spin_unlock_irq(&ent->lock);
- err = destroy_mkey(dev, mr);
- if (err)
- mlx5_ib_warn(dev, "failed destroy mkey\n");
- else
- kfree(mr);
+ mlx5_core_destroy_mkey(dev->mdev, &mr->mmkey);
+ }
+
+#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
+ synchronize_srcu(&dev->mr_srcu);
+#endif
+
+ list_for_each_entry_safe(mr, tmp_mr, &del_list, list) {
+ list_del(&mr->list);
+ kfree(mr);
}
}
@@ -562,26 +585,32 @@ static void clean_keys(struct mlx5_ib_dev *dev, int c)
{
struct mlx5_mr_cache *cache = &dev->cache;
struct mlx5_cache_ent *ent = &cache->ent[c];
+ struct mlx5_ib_mr *tmp_mr;
struct mlx5_ib_mr *mr;
- int err;
+ LIST_HEAD(del_list);
cancel_delayed_work(&ent->dwork);
while (1) {
spin_lock_irq(&ent->lock);
if (list_empty(&ent->head)) {
spin_unlock_irq(&ent->lock);
- return;
+ break;
}
mr = list_first_entry(&ent->head, struct mlx5_ib_mr, list);
- list_del(&mr->list);
+ list_move(&mr->list, &del_list);
ent->cur--;
ent->size--;
spin_unlock_irq(&ent->lock);
- err = destroy_mkey(dev, mr);
- if (err)
- mlx5_ib_warn(dev, "failed destroy mkey\n");
- else
- kfree(mr);
+ mlx5_core_destroy_mkey(dev->mdev, &mr->mmkey);
+ }
+
+#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
+ synchronize_srcu(&dev->mr_srcu);
+#endif
+
+ list_for_each_entry_safe(mr, tmp_mr, &del_list, list) {
+ list_del(&mr->list);
+ kfree(mr);
}
}
@@ -780,7 +809,7 @@ struct ib_mr *mlx5_ib_get_dma_mr(struct ib_pd *pd, int acc)
mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
- MLX5_SET(mkc, mkc, access_mode, MLX5_MKC_ACCESS_MODE_PA);
+ MLX5_SET(mkc, mkc, access_mode_1_0, MLX5_MKC_ACCESS_MODE_PA);
MLX5_SET(mkc, mkc, a, !!(acc & IB_ACCESS_REMOTE_ATOMIC));
MLX5_SET(mkc, mkc, rw, !!(acc & IB_ACCESS_REMOTE_WRITE));
MLX5_SET(mkc, mkc, rr, !!(acc & IB_ACCESS_REMOTE_READ));
@@ -947,7 +976,10 @@ static inline int populate_xlt(struct mlx5_ib_mr *mr, int idx, int npages,
{
struct mlx5_ib_dev *dev = mr->dev;
struct ib_umem *umem = mr->umem;
+
if (flags & MLX5_IB_UPD_XLT_INDIRECT) {
+ if (!umr_can_use_indirect_mkey(dev))
+ return -EPERM;
mlx5_odp_populate_klm(xlt, idx, npages, mr, flags);
return npages;
}
@@ -977,7 +1009,6 @@ int mlx5_ib_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages,
{
struct mlx5_ib_dev *dev = mr->dev;
struct device *ddev = dev->ib_dev.dev.parent;
- struct mlx5_ib_ucontext *uctx = NULL;
int size;
void *xlt;
dma_addr_t dma;
@@ -993,6 +1024,11 @@ int mlx5_ib_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages,
size_t pages_to_map = 0;
size_t pages_iter = 0;
gfp_t gfp;
+ bool use_emergency_page = false;
+
+ if ((flags & MLX5_IB_UPD_XLT_INDIRECT) &&
+ !umr_can_use_indirect_mkey(dev))
+ return -EPERM;
/* UMR copies MTTs in units of MLX5_UMR_MTT_ALIGNMENT bytes,
* so we need to align the offset and length accordingly
@@ -1019,12 +1055,11 @@ int mlx5_ib_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages,
}
if (!xlt) {
- uctx = to_mucontext(mr->ibmr.pd->uobject->context);
mlx5_ib_warn(dev, "Using XLT emergency buffer\n");
+ xlt = (void *)mlx5_ib_get_xlt_emergency_page();
size = PAGE_SIZE;
- xlt = (void *)uctx->upd_xlt_page;
- mutex_lock(&uctx->upd_xlt_page_mutex);
memset(xlt, 0, size);
+ use_emergency_page = true;
}
pages_iter = size / desc_size;
dma = dma_map_single(ddev, xlt, size, DMA_TO_DEVICE);
@@ -1088,8 +1123,8 @@ int mlx5_ib_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages,
dma_unmap_single(ddev, dma, size, DMA_TO_DEVICE);
free_xlt:
- if (uctx)
- mutex_unlock(&uctx->upd_xlt_page_mutex);
+ if (use_emergency_page)
+ mlx5_ib_put_xlt_emergency_page();
else
free_pages((unsigned long)xlt, get_order(size));
@@ -1141,7 +1176,7 @@ static struct mlx5_ib_mr *reg_create(struct ib_mr *ibmr, struct ib_pd *pd,
mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
MLX5_SET(mkc, mkc, free, !populate);
- MLX5_SET(mkc, mkc, access_mode, MLX5_MKC_ACCESS_MODE_MTT);
+ MLX5_SET(mkc, mkc, access_mode_1_0, MLX5_MKC_ACCESS_MODE_MTT);
MLX5_SET(mkc, mkc, a, !!(access_flags & IB_ACCESS_REMOTE_ATOMIC));
MLX5_SET(mkc, mkc, rw, !!(access_flags & IB_ACCESS_REMOTE_WRITE));
MLX5_SET(mkc, mkc, rr, !!(access_flags & IB_ACCESS_REMOTE_READ));
@@ -1197,22 +1232,96 @@ static void set_mr_fileds(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr,
mr->access_flags = access_flags;
}
+static struct ib_mr *mlx5_ib_get_memic_mr(struct ib_pd *pd, u64 memic_addr,
+ u64 length, int acc)
+{
+ struct mlx5_ib_dev *dev = to_mdev(pd->device);
+ int inlen = MLX5_ST_SZ_BYTES(create_mkey_in);
+ struct mlx5_core_dev *mdev = dev->mdev;
+ struct mlx5_ib_mr *mr;
+ void *mkc;
+ u32 *in;
+ int err;
+
+ mr = kzalloc(sizeof(*mr), GFP_KERNEL);
+ if (!mr)
+ return ERR_PTR(-ENOMEM);
+
+ in = kzalloc(inlen, GFP_KERNEL);
+ if (!in) {
+ err = -ENOMEM;
+ goto err_free;
+ }
+
+ mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
+
+ MLX5_SET(mkc, mkc, access_mode_1_0, MLX5_MKC_ACCESS_MODE_MEMIC & 0x3);
+ MLX5_SET(mkc, mkc, access_mode_4_2,
+ (MLX5_MKC_ACCESS_MODE_MEMIC >> 2) & 0x7);
+ MLX5_SET(mkc, mkc, a, !!(acc & IB_ACCESS_REMOTE_ATOMIC));
+ MLX5_SET(mkc, mkc, rw, !!(acc & IB_ACCESS_REMOTE_WRITE));
+ MLX5_SET(mkc, mkc, rr, !!(acc & IB_ACCESS_REMOTE_READ));
+ MLX5_SET(mkc, mkc, lw, !!(acc & IB_ACCESS_LOCAL_WRITE));
+ MLX5_SET(mkc, mkc, lr, 1);
+
+ MLX5_SET64(mkc, mkc, len, length);
+ MLX5_SET(mkc, mkc, pd, to_mpd(pd)->pdn);
+ MLX5_SET(mkc, mkc, qpn, 0xffffff);
+ MLX5_SET64(mkc, mkc, start_addr,
+ memic_addr - pci_resource_start(dev->mdev->pdev, 0));
+
+ err = mlx5_core_create_mkey(mdev, &mr->mmkey, in, inlen);
+ if (err)
+ goto err_in;
+
+ kfree(in);
+
+ mr->umem = NULL;
+ set_mr_fileds(dev, mr, 0, length, acc);
+
+ return &mr->ibmr;
+
+err_in:
+ kfree(in);
+
+err_free:
+ kfree(mr);
+
+ return ERR_PTR(err);
+}
+
+struct ib_mr *mlx5_ib_reg_dm_mr(struct ib_pd *pd, struct ib_dm *dm,
+ struct ib_dm_mr_attr *attr,
+ struct uverbs_attr_bundle *attrs)
+{
+ struct mlx5_ib_dm *mdm = to_mdm(dm);
+ u64 memic_addr;
+
+ if (attr->access_flags & ~MLX5_IB_DM_ALLOWED_ACCESS)
+ return ERR_PTR(-EINVAL);
+
+ memic_addr = mdm->dev_addr + attr->offset;
+
+ return mlx5_ib_get_memic_mr(pd, memic_addr, attr->length,
+ attr->access_flags);
+}
+
struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
u64 virt_addr, int access_flags,
struct ib_udata *udata)
{
struct mlx5_ib_dev *dev = to_mdev(pd->device);
struct mlx5_ib_mr *mr = NULL;
+ bool populate_mtts = false;
struct ib_umem *umem;
int page_shift;
int npages;
int ncont;
int order;
int err;
- bool use_umr = true;
if (!IS_ENABLED(CONFIG_INFINIBAND_USER_MEM))
- return ERR_PTR(-EINVAL);
+ return ERR_PTR(-EOPNOTSUPP);
mlx5_ib_dbg(dev, "start 0x%llx, virt_addr 0x%llx, length 0x%llx, access_flags 0x%x\n",
start, virt_addr, length, access_flags);
@@ -1224,6 +1333,8 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
return ERR_PTR(-EINVAL);
mr = mlx5_ib_alloc_implicit_mr(to_mpd(pd), access_flags);
+ if (IS_ERR(mr))
+ return ERR_CAST(mr);
return &mr->ibmr;
}
#endif
@@ -1234,26 +1345,29 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
if (err < 0)
return ERR_PTR(err);
- if (order <= mr_cache_max_order(dev)) {
+ if (use_umr(dev, order)) {
mr = alloc_mr_from_cache(pd, umem, virt_addr, length, ncont,
page_shift, order, access_flags);
if (PTR_ERR(mr) == -EAGAIN) {
mlx5_ib_dbg(dev, "cache empty for order %d\n", order);
mr = NULL;
}
+ populate_mtts = false;
} else if (!MLX5_CAP_GEN(dev->mdev, umr_extended_translation_offset)) {
if (access_flags & IB_ACCESS_ON_DEMAND) {
err = -EINVAL;
pr_err("Got MR registration for ODP MR > 512MB, not supported for Connect-IB\n");
goto error;
}
- use_umr = false;
+ populate_mtts = true;
}
if (!mr) {
+ if (!umr_can_modify_entity_size(dev))
+ populate_mtts = true;
mutex_lock(&dev->slow_path_mutex);
mr = reg_create(NULL, pd, virt_addr, length, umem, ncont,
- page_shift, access_flags, !use_umr);
+ page_shift, access_flags, populate_mtts);
mutex_unlock(&dev->slow_path_mutex);
}
@@ -1271,7 +1385,7 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
update_odp_mr(mr);
#endif
- if (use_umr) {
+ if (!populate_mtts) {
int update_xlt_flags = MLX5_IB_UPD_XLT_ENABLE;
if (access_flags & IB_ACCESS_ON_DEMAND)
@@ -1286,7 +1400,9 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
}
}
+#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
mr->live = 1;
+#endif
return &mr->ibmr;
error:
ib_umem_release(umem);
@@ -1365,36 +1481,34 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
ib_umem_release(mr->umem);
err = mr_umem_get(pd, addr, len, access_flags, &mr->umem,
&npages, &page_shift, &ncont, &order);
- if (err < 0) {
- clean_mr(dev, mr);
- return err;
- }
+ if (err)
+ goto err;
}
if (flags & IB_MR_REREG_TRANS && !use_umr_mtt_update(mr, addr, len)) {
/*
* UMR can't be used - MKey needs to be replaced.
*/
- if (mr->allocated_from_cache) {
+ if (mr->allocated_from_cache)
err = unreg_umr(dev, mr);
- if (err)
- mlx5_ib_warn(dev, "Failed to unregister MR\n");
- } else {
+ else
err = destroy_mkey(dev, mr);
- if (err)
- mlx5_ib_warn(dev, "Failed to destroy MKey\n");
- }
if (err)
- return err;
+ goto err;
mr = reg_create(ib_mr, pd, addr, len, mr->umem, ncont,
page_shift, access_flags, true);
- if (IS_ERR(mr))
- return PTR_ERR(mr);
+ if (IS_ERR(mr)) {
+ err = PTR_ERR(mr);
+ mr = to_mmr(ib_mr);
+ goto err;
+ }
mr->allocated_from_cache = 0;
+#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
mr->live = 1;
+#endif
} else {
/*
* Send a UMR WQE
@@ -1417,13 +1531,8 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
err = rereg_umr(pd, mr, access_flags, flags);
}
- if (err) {
- mlx5_ib_warn(dev, "Failed to rereg UMR\n");
- ib_umem_release(mr->umem);
- mr->umem = NULL;
- clean_mr(dev, mr);
- return err;
- }
+ if (err)
+ goto err;
}
set_mr_fileds(dev, mr, npages, len, access_flags);
@@ -1432,6 +1541,14 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
update_odp_mr(mr);
#endif
return 0;
+
+err:
+ if (mr->umem) {
+ ib_umem_release(mr->umem);
+ mr->umem = NULL;
+ }
+ clean_mr(dev, mr);
+ return err;
}
static int
@@ -1480,10 +1597,9 @@ mlx5_free_priv_descs(struct mlx5_ib_mr *mr)
}
}
-static int clean_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
+static void clean_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
{
int allocated_from_cache = mr->allocated_from_cache;
- int err;
if (mr->sig) {
if (mlx5_core_destroy_psv(dev->mdev,
@@ -1500,21 +1616,11 @@ static int clean_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
mlx5_free_priv_descs(mr);
- if (!allocated_from_cache) {
- u32 key = mr->mmkey.key;
-
- err = destroy_mkey(dev, mr);
- if (err) {
- mlx5_ib_warn(dev, "failed to destroy mkey 0x%x (%d)\n",
- key, err);
- return err;
- }
- }
-
- return 0;
+ if (!allocated_from_cache)
+ destroy_mkey(dev, mr);
}
-static int dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
+static void dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
{
int npages = mr->npages;
struct ib_umem *umem = mr->umem;
@@ -1555,16 +1661,12 @@ static int dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
kfree(mr);
else
mlx5_mr_cache_free(dev, mr);
-
- return 0;
}
int mlx5_ib_dereg_mr(struct ib_mr *ibmr)
{
- struct mlx5_ib_dev *dev = to_mdev(ibmr->device);
- struct mlx5_ib_mr *mr = to_mmr(ibmr);
-
- return dereg_mr(dev, mr);
+ dereg_mr(to_mdev(ibmr->device), to_mmr(ibmr));
+ return 0;
}
struct ib_mr *mlx5_ib_alloc_mr(struct ib_pd *pd,
@@ -1645,7 +1747,8 @@ struct ib_mr *mlx5_ib_alloc_mr(struct ib_pd *pd,
goto err_free_in;
}
- MLX5_SET(mkc, mkc, access_mode, mr->access_mode);
+ MLX5_SET(mkc, mkc, access_mode_1_0, mr->access_mode & 0x3);
+ MLX5_SET(mkc, mkc, access_mode_4_2, (mr->access_mode >> 2) & 0x7);
MLX5_SET(mkc, mkc, umr_en, 1);
mr->ibmr.device = pd->device;
@@ -1726,7 +1829,7 @@ struct ib_mw *mlx5_ib_alloc_mw(struct ib_pd *pd, enum ib_mw_type type,
MLX5_SET(mkc, mkc, pd, to_mpd(pd)->pdn);
MLX5_SET(mkc, mkc, umr_en, 1);
MLX5_SET(mkc, mkc, lr, 1);
- MLX5_SET(mkc, mkc, access_mode, MLX5_MKC_ACCESS_MODE_KLMS);
+ MLX5_SET(mkc, mkc, access_mode_1_0, MLX5_MKC_ACCESS_MODE_KLMS);
MLX5_SET(mkc, mkc, en_rinval, !!((type == IB_MW_TYPE_2)));
MLX5_SET(mkc, mkc, qpn, 0xffffff);
diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index 0d0b0b8dad98..7ed4b70f6447 100644
--- a/drivers/infiniband/hw/mlx5/qp.c
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -86,7 +86,9 @@ struct mlx5_modify_raw_qp_param {
u16 operation;
u32 set_mask; /* raw_qp_set_mask_map */
- u32 rate_limit;
+
+ struct mlx5_rate_limit rl;
+
u8 rq_q_ctr_id;
};
@@ -878,7 +880,7 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd,
goto err_free;
}
- err = ib_copy_to_udata(udata, resp, sizeof(*resp));
+ err = ib_copy_to_udata(udata, resp, min(udata->outlen, sizeof(*resp)));
if (err) {
mlx5_ib_dbg(dev, "copy failed\n");
goto err_unmap;
@@ -1411,6 +1413,7 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp,
void *tirc;
void *hfso;
u32 selected_fields = 0;
+ u32 outer_l4;
size_t min_resp_len;
u32 tdn = mucontext->tdn;
struct mlx5_ib_create_qp_rss ucmd = {};
@@ -1466,7 +1469,7 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp,
return -EOPNOTSUPP;
}
- err = ib_copy_to_udata(udata, &resp, min_resp_len);
+ err = ib_copy_to_udata(udata, &resp, min(udata->outlen, sizeof(resp)));
if (err) {
mlx5_ib_dbg(dev, "copy failed\n");
return -EINVAL;
@@ -1541,10 +1544,14 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp,
MLX5_SET(rx_hash_field_select, hfso, l3_prot_type,
MLX5_L3_PROT_TYPE_IPV6);
- if (((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_TCP) ||
- (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_TCP)) &&
- ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_UDP) ||
- (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_UDP))) {
+ outer_l4 = ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_TCP) ||
+ (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_TCP)) << 0 |
+ ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_UDP) ||
+ (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_UDP)) << 1 |
+ (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_IPSEC_SPI) << 2;
+
+ /* Check that only one l4 protocol is set */
+ if (outer_l4 & (outer_l4 - 1)) {
err = -EINVAL;
goto err;
}
@@ -1575,6 +1582,9 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp,
(ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_UDP))
selected_fields |= MLX5_HASH_FIELD_SEL_L4_DPORT;
+ if (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_IPSEC_SPI)
+ selected_fields |= MLX5_HASH_FIELD_SEL_IPSEC_SPI;
+
MLX5_SET(rx_hash_field_select, hfso, selected_fields, selected_fields);
create_tir:
@@ -2774,8 +2784,9 @@ static int modify_raw_packet_qp_sq(struct mlx5_core_dev *dev,
const struct mlx5_modify_raw_qp_param *raw_qp_param)
{
struct mlx5_ib_qp *ibqp = sq->base.container_mibqp;
- u32 old_rate = ibqp->rate_limit;
- u32 new_rate = old_rate;
+ struct mlx5_rate_limit old_rl = ibqp->rl;
+ struct mlx5_rate_limit new_rl = old_rl;
+ bool new_rate_added = false;
u16 rl_index = 0;
void *in;
void *sqc;
@@ -2797,39 +2808,43 @@ static int modify_raw_packet_qp_sq(struct mlx5_core_dev *dev,
pr_warn("%s: Rate limit can only be changed when SQ is moving to RDY\n",
__func__);
else
- new_rate = raw_qp_param->rate_limit;
+ new_rl = raw_qp_param->rl;
}
- if (old_rate != new_rate) {
- if (new_rate) {
- err = mlx5_rl_add_rate(dev, new_rate, &rl_index);
+ if (!mlx5_rl_are_equal(&old_rl, &new_rl)) {
+ if (new_rl.rate) {
+ err = mlx5_rl_add_rate(dev, &rl_index, &new_rl);
if (err) {
- pr_err("Failed configuring rate %u: %d\n",
- new_rate, err);
+ pr_err("Failed configuring rate limit(err %d): \
+ rate %u, max_burst_sz %u, typical_pkt_sz %u\n",
+ err, new_rl.rate, new_rl.max_burst_sz,
+ new_rl.typical_pkt_sz);
+
goto out;
}
+ new_rate_added = true;
}
MLX5_SET64(modify_sq_in, in, modify_bitmask, 1);
+ /* index 0 means no limit */
MLX5_SET(sqc, sqc, packet_pacing_rate_limit_index, rl_index);
}
err = mlx5_core_modify_sq(dev, sq->base.mqp.qpn, in, inlen);
if (err) {
/* Remove new rate from table if failed */
- if (new_rate &&
- old_rate != new_rate)
- mlx5_rl_remove_rate(dev, new_rate);
+ if (new_rate_added)
+ mlx5_rl_remove_rate(dev, &new_rl);
goto out;
}
/* Only remove the old rate after new rate was set */
- if ((old_rate &&
- (old_rate != new_rate)) ||
+ if ((old_rl.rate &&
+ !mlx5_rl_are_equal(&old_rl, &new_rl)) ||
(new_state != MLX5_SQC_STATE_RDY))
- mlx5_rl_remove_rate(dev, old_rate);
+ mlx5_rl_remove_rate(dev, &old_rl);
- ibqp->rate_limit = new_rate;
+ ibqp->rl = new_rl;
sq->state = new_state;
out:
@@ -2906,7 +2921,8 @@ static int modify_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp,
static int __mlx5_ib_modify_qp(struct ib_qp *ibqp,
const struct ib_qp_attr *attr, int attr_mask,
- enum ib_qp_state cur_state, enum ib_qp_state new_state)
+ enum ib_qp_state cur_state, enum ib_qp_state new_state,
+ const struct mlx5_ib_modify_qp *ucmd)
{
static const u16 optab[MLX5_QP_NUM_STATE][MLX5_QP_NUM_STATE] = {
[MLX5_QP_STATE_RST] = {
@@ -2959,18 +2975,16 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp,
u16 op;
u8 tx_affinity = 0;
+ mlx5_st = to_mlx5_st(ibqp->qp_type == IB_QPT_DRIVER ?
+ qp->qp_sub_type : ibqp->qp_type);
+ if (mlx5_st < 0)
+ return -EINVAL;
+
context = kzalloc(sizeof(*context), GFP_KERNEL);
if (!context)
return -ENOMEM;
- err = to_mlx5_st(ibqp->qp_type == IB_QPT_DRIVER ?
- qp->qp_sub_type : ibqp->qp_type);
- if (err < 0) {
- mlx5_ib_dbg(dev, "unsupported qp type %d\n", ibqp->qp_type);
- goto out;
- }
-
- context->flags = cpu_to_be32(err << 16);
+ context->flags = cpu_to_be32(mlx5_st << 16);
if (!(attr_mask & IB_QP_PATH_MIG_STATE)) {
context->flags |= cpu_to_be32(MLX5_QP_PM_MIGRATED << 11);
@@ -3124,10 +3138,6 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp,
mlx5_cur = to_mlx5_state(cur_state);
mlx5_new = to_mlx5_state(new_state);
- mlx5_st = to_mlx5_st(ibqp->qp_type == IB_QPT_DRIVER ?
- qp->qp_sub_type : ibqp->qp_type);
- if (mlx5_st < 0)
- goto out;
if (mlx5_cur >= MLX5_QP_NUM_STATE || mlx5_new >= MLX5_QP_NUM_STATE ||
!optab[mlx5_cur][mlx5_new]) {
@@ -3150,7 +3160,30 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp,
}
if (attr_mask & IB_QP_RATE_LIMIT) {
- raw_qp_param.rate_limit = attr->rate_limit;
+ raw_qp_param.rl.rate = attr->rate_limit;
+
+ if (ucmd->burst_info.max_burst_sz) {
+ if (attr->rate_limit &&
+ MLX5_CAP_QOS(dev->mdev, packet_pacing_burst_bound)) {
+ raw_qp_param.rl.max_burst_sz =
+ ucmd->burst_info.max_burst_sz;
+ } else {
+ err = -EINVAL;
+ goto out;
+ }
+ }
+
+ if (ucmd->burst_info.typical_pkt_sz) {
+ if (attr->rate_limit &&
+ MLX5_CAP_QOS(dev->mdev, packet_pacing_typical_size)) {
+ raw_qp_param.rl.typical_pkt_sz =
+ ucmd->burst_info.typical_pkt_sz;
+ } else {
+ err = -EINVAL;
+ goto out;
+ }
+ }
+
raw_qp_param.set_mask |= MLX5_RAW_QP_RATE_LIMIT;
}
@@ -3178,7 +3211,8 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp,
* If we moved a kernel QP to RESET, clean up all old CQ
* entries and reinitialize the QP.
*/
- if (new_state == IB_QPS_RESET && !ibqp->uobject) {
+ if (new_state == IB_QPS_RESET &&
+ !ibqp->uobject && ibqp->qp_type != IB_QPT_XRC_TGT) {
mlx5_ib_cq_clean(recv_cq, base->mqp.qpn,
ibqp->srq ? to_msrq(ibqp->srq) : NULL);
if (send_cq != recv_cq)
@@ -3337,8 +3371,10 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
{
struct mlx5_ib_dev *dev = to_mdev(ibqp->device);
struct mlx5_ib_qp *qp = to_mqp(ibqp);
+ struct mlx5_ib_modify_qp ucmd = {};
enum ib_qp_type qp_type;
enum ib_qp_state cur_state, new_state;
+ size_t required_cmd_sz;
int err = -EINVAL;
int port;
enum rdma_link_layer ll = IB_LINK_LAYER_UNSPECIFIED;
@@ -3346,6 +3382,28 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
if (ibqp->rwq_ind_tbl)
return -ENOSYS;
+ if (udata && udata->inlen) {
+ required_cmd_sz = offsetof(typeof(ucmd), reserved) +
+ sizeof(ucmd.reserved);
+ if (udata->inlen < required_cmd_sz)
+ return -EINVAL;
+
+ if (udata->inlen > sizeof(ucmd) &&
+ !ib_is_udata_cleared(udata, sizeof(ucmd),
+ udata->inlen - sizeof(ucmd)))
+ return -EOPNOTSUPP;
+
+ if (ib_copy_from_udata(&ucmd, udata,
+ min(udata->inlen, sizeof(ucmd))))
+ return -EFAULT;
+
+ if (ucmd.comp_mask ||
+ memchr_inv(&ucmd.reserved, 0, sizeof(ucmd.reserved)) ||
+ memchr_inv(&ucmd.burst_info.reserved, 0,
+ sizeof(ucmd.burst_info.reserved)))
+ return -EOPNOTSUPP;
+ }
+
if (unlikely(ibqp->qp_type == IB_QPT_GSI))
return mlx5_ib_gsi_modify_qp(ibqp, attr, attr_mask);
@@ -3426,7 +3484,8 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
goto out;
}
- err = __mlx5_ib_modify_qp(ibqp, attr, attr_mask, cur_state, new_state);
+ err = __mlx5_ib_modify_qp(ibqp, attr, attr_mask, cur_state,
+ new_state, &ucmd);
out:
mutex_unlock(&qp->mutex);
@@ -3646,8 +3705,19 @@ static __be64 get_umr_update_pd_mask(void)
return cpu_to_be64(result);
}
-static void set_reg_umr_segment(struct mlx5_wqe_umr_ctrl_seg *umr,
- struct ib_send_wr *wr, int atomic)
+static int umr_check_mkey_mask(struct mlx5_ib_dev *dev, u64 mask)
+{
+ if ((mask & MLX5_MKEY_MASK_PAGE_SIZE &&
+ MLX5_CAP_GEN(dev->mdev, umr_modify_entity_size_disabled)) ||
+ (mask & MLX5_MKEY_MASK_A &&
+ MLX5_CAP_GEN(dev->mdev, umr_modify_atomic_disabled)))
+ return -EPERM;
+ return 0;
+}
+
+static int set_reg_umr_segment(struct mlx5_ib_dev *dev,
+ struct mlx5_wqe_umr_ctrl_seg *umr,
+ struct ib_send_wr *wr, int atomic)
{
struct mlx5_umr_wr *umrwr = umr_wr(wr);
@@ -3679,6 +3749,8 @@ static void set_reg_umr_segment(struct mlx5_wqe_umr_ctrl_seg *umr,
if (!wr->num_sge)
umr->flags |= MLX5_UMR_INLINE;
+
+ return umr_check_mkey_mask(dev, be64_to_cpu(umr->mkey_mask));
}
static u8 get_umr_flags(int acc)
@@ -4501,7 +4573,9 @@ int mlx5_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
}
qp->sq.wr_data[idx] = MLX5_IB_WR_UMR;
ctrl->imm = cpu_to_be32(umr_wr(wr)->mkey);
- set_reg_umr_segment(seg, wr, !!(MLX5_CAP_GEN(mdev, atomic)));
+ err = set_reg_umr_segment(dev, seg, wr, !!(MLX5_CAP_GEN(mdev, atomic)));
+ if (unlikely(err))
+ goto out;
seg += sizeof(struct mlx5_wqe_umr_ctrl_seg);
size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16;
if (unlikely((seg == qend)))
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c
index 6fee7795d1c8..541f237965c7 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.c
+++ b/drivers/infiniband/hw/mthca/mthca_provider.c
@@ -1295,6 +1295,7 @@ int mthca_register_device(struct mthca_dev *dev)
mutex_init(&dev->cap_mask_mutex);
+ dev->ib_dev.driver_id = RDMA_DRIVER_MTHCA;
ret = ib_register_device(&dev->ib_dev, NULL);
if (ret)
return ret;
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c
index 162475aeeedd..1040a6e34230 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.c
+++ b/drivers/infiniband/hw/nes/nes_verbs.c
@@ -3854,6 +3854,7 @@ int nes_register_ofa_device(struct nes_ib_device *nesibdev)
struct nes_adapter *nesadapter = nesdev->nesadapter;
int i, ret;
+ nesvnic->nesibdev->ibdev.driver_id = RDMA_DRIVER_NES;
ret = ib_register_device(&nesvnic->nesibdev->ibdev, NULL);
if (ret) {
return ret;
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_ah.c b/drivers/infiniband/hw/ocrdma/ocrdma_ah.c
index dec650930ca6..3897b64532e1 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_ah.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_ah.c
@@ -193,11 +193,9 @@ struct ib_ah *ocrdma_create_ah(struct ib_pd *ibpd, struct rdma_ah_attr *attr,
__func__, status);
goto av_conf_err;
}
- if (sgid_attr.ndev) {
- if (is_vlan_dev(sgid_attr.ndev))
- vlan_tag = vlan_dev_vlan_id(sgid_attr.ndev);
- dev_put(sgid_attr.ndev);
- }
+ if (is_vlan_dev(sgid_attr.ndev))
+ vlan_tag = vlan_dev_vlan_id(sgid_attr.ndev);
+ dev_put(sgid_attr.ndev);
/* Get network header type for this GID */
ah->hdr_type = ib_gid_to_network_type(sgid_attr.gid_type, &sgid);
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_hw.c b/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
index 9904918589a4..2c260e1c29d1 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
@@ -2014,7 +2014,7 @@ static int ocrdma_mbx_reg_mr_cont(struct ocrdma_dev *dev,
struct ocrdma_hw_mr *hwmr, u32 pbl_cnt,
u32 pbl_offset, u32 last)
{
- int status = -ENOMEM;
+ int status;
int i;
struct ocrdma_reg_nsmr_cont *cmd;
@@ -2033,9 +2033,7 @@ static int ocrdma_mbx_reg_mr_cont(struct ocrdma_dev *dev,
upper_32_bits(hwmr->pbl_table[i + pbl_offset].pa);
}
status = ocrdma_mbx_cmd(dev, (struct ocrdma_mqe *)cmd);
- if (status)
- goto mbx_err;
-mbx_err:
+
kfree(cmd);
return status;
}
@@ -2496,7 +2494,7 @@ static int ocrdma_set_av_params(struct ocrdma_qp *qp,
{
int status;
struct rdma_ah_attr *ah_attr = &attrs->ah_attr;
- union ib_gid sgid, zgid;
+ union ib_gid sgid;
struct ib_gid_attr sgid_attr;
u32 vlan_id = 0xFFFF;
u8 mac_addr[6], hdr_type;
@@ -2529,16 +2527,12 @@ static int ocrdma_set_av_params(struct ocrdma_qp *qp,
status = ib_get_cached_gid(&dev->ibdev, 1, grh->sgid_index,
&sgid, &sgid_attr);
- if (!status && sgid_attr.ndev) {
+ if (!status) {
vlan_id = rdma_vlan_dev_vlan_id(sgid_attr.ndev);
memcpy(mac_addr, sgid_attr.ndev->dev_addr, ETH_ALEN);
dev_put(sgid_attr.ndev);
}
- memset(&zgid, 0, sizeof(zgid));
- if (!memcmp(&sgid, &zgid, sizeof(zgid)))
- return -EINVAL;
-
qp->sgid_idx = grh->sgid_index;
memcpy(&cmd->params.sgid[0], &sgid.raw[0], sizeof(cmd->params.sgid));
status = ocrdma_resolve_dmac(dev, ah_attr, &mac_addr[0]);
@@ -3133,12 +3127,12 @@ done:
static int ocrdma_mbx_modify_eqd(struct ocrdma_dev *dev, struct ocrdma_eq *eq,
int num)
{
- int i, status = -ENOMEM;
+ int i, status;
struct ocrdma_modify_eqd_req *cmd;
cmd = ocrdma_init_emb_mqe(OCRDMA_CMD_MODIFY_EQ_DELAY, sizeof(*cmd));
if (!cmd)
- return status;
+ return -ENOMEM;
ocrdma_init_mch(&cmd->cmd.req, OCRDMA_CMD_MODIFY_EQ_DELAY,
OCRDMA_SUBSYS_COMMON, sizeof(*cmd));
@@ -3151,9 +3145,7 @@ static int ocrdma_mbx_modify_eqd(struct ocrdma_dev *dev, struct ocrdma_eq *eq,
(eq[i].aic_obj.prev_eqd * 65)/100;
}
status = ocrdma_mbx_cmd(dev, (struct ocrdma_mqe *)cmd);
- if (status)
- goto mbx_err;
-mbx_err:
+
kfree(cmd);
return status;
}
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_main.c b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
index fbfbd9e96147..eb8b6a935016 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_main.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
@@ -158,10 +158,7 @@ static int ocrdma_register_device(struct ocrdma_dev *dev)
dev->ibdev.query_device = ocrdma_query_device;
dev->ibdev.query_port = ocrdma_query_port;
dev->ibdev.modify_port = ocrdma_modify_port;
- dev->ibdev.query_gid = ocrdma_query_gid;
dev->ibdev.get_netdev = ocrdma_get_netdev;
- dev->ibdev.add_gid = ocrdma_add_gid;
- dev->ibdev.del_gid = ocrdma_del_gid;
dev->ibdev.get_link_layer = ocrdma_link_layer;
dev->ibdev.alloc_pd = ocrdma_alloc_pd;
dev->ibdev.dealloc_pd = ocrdma_dealloc_pd;
@@ -217,6 +214,7 @@ static int ocrdma_register_device(struct ocrdma_dev *dev)
dev->ibdev.destroy_srq = ocrdma_destroy_srq;
dev->ibdev.post_srq_recv = ocrdma_post_srq_recv;
}
+ dev->ibdev.driver_id = RDMA_DRIVER_OCRDMA;
return ib_register_device(&dev->ibdev, NULL);
}
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
index 8009bdad4e5b..784ed6b09a46 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
@@ -62,40 +62,6 @@ int ocrdma_query_pkey(struct ib_device *ibdev, u8 port, u16 index, u16 *pkey)
return 0;
}
-int ocrdma_query_gid(struct ib_device *ibdev, u8 port,
- int index, union ib_gid *sgid)
-{
- int ret;
-
- memset(sgid, 0, sizeof(*sgid));
- if (index >= OCRDMA_MAX_SGID)
- return -EINVAL;
-
- ret = ib_get_cached_gid(ibdev, port, index, sgid, NULL);
- if (ret == -EAGAIN) {
- memcpy(sgid, &zgid, sizeof(*sgid));
- return 0;
- }
-
- return ret;
-}
-
-int ocrdma_add_gid(struct ib_device *device,
- u8 port_num,
- unsigned int index,
- const union ib_gid *gid,
- const struct ib_gid_attr *attr,
- void **context) {
- return 0;
-}
-
-int ocrdma_del_gid(struct ib_device *device,
- u8 port_num,
- unsigned int index,
- void **context) {
- return 0;
-}
-
int ocrdma_query_device(struct ib_device *ibdev, struct ib_device_attr *attr,
struct ib_udata *uhw)
{
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h
index 704ef1e9271b..9a9971708646 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h
@@ -61,19 +61,7 @@ enum rdma_protocol_type
ocrdma_query_protocol(struct ib_device *device, u8 port_num);
void ocrdma_get_guid(struct ocrdma_dev *, u8 *guid);
-int ocrdma_query_gid(struct ib_device *, u8 port,
- int index, union ib_gid *gid);
struct net_device *ocrdma_get_netdev(struct ib_device *device, u8 port_num);
-int ocrdma_add_gid(struct ib_device *device,
- u8 port_num,
- unsigned int index,
- const union ib_gid *gid,
- const struct ib_gid_attr *attr,
- void **context);
-int ocrdma_del_gid(struct ib_device *device,
- u8 port_num,
- unsigned int index,
- void **context);
int ocrdma_query_pkey(struct ib_device *, u8 port, u16 index, u16 *pkey);
struct ib_ucontext *ocrdma_alloc_ucontext(struct ib_device *,
diff --git a/drivers/infiniband/hw/qedr/main.c b/drivers/infiniband/hw/qedr/main.c
index f9a645c869ce..f4cb60b658ea 100644
--- a/drivers/infiniband/hw/qedr/main.c
+++ b/drivers/infiniband/hw/qedr/main.c
@@ -162,10 +162,6 @@ static int qedr_iw_register_device(struct qedr_dev *dev)
static void qedr_roce_register_device(struct qedr_dev *dev)
{
dev->ibdev.node_type = RDMA_NODE_IB_CA;
- dev->ibdev.query_gid = qedr_query_gid;
-
- dev->ibdev.add_gid = qedr_add_gid;
- dev->ibdev.del_gid = qedr_del_gid;
dev->ibdev.get_port_immutable = qedr_roce_port_immutable;
}
@@ -257,6 +253,7 @@ static int qedr_register_device(struct qedr_dev *dev)
dev->ibdev.get_link_layer = qedr_link_layer;
dev->ibdev.get_dev_fw_str = qedr_get_dev_fw_str;
+ dev->ibdev.driver_id = RDMA_DRIVER_QEDR;
return ib_register_device(&dev->ibdev, NULL);
}
@@ -707,7 +704,7 @@ static void qedr_affiliated_event(void *context, u8 e_code, void *fw_handle)
"Error: CQ event with NULL pointer ibcq. Handle=%llx\n",
roce_handle64);
}
- DP_ERR(dev, "CQ event %d on hanlde %p\n", e_code, cq);
+ DP_ERR(dev, "CQ event %d on handle %p\n", e_code, cq);
break;
case EVENT_TYPE_QP:
qp = (struct qedr_qp *)(uintptr_t)roce_handle64;
@@ -723,7 +720,7 @@ static void qedr_affiliated_event(void *context, u8 e_code, void *fw_handle)
"Error: QP event with NULL pointer ibqp. Handle=%llx\n",
roce_handle64);
}
- DP_ERR(dev, "QP event %d on hanlde %p\n", e_code, qp);
+ DP_ERR(dev, "QP event %d on handle %p\n", e_code, qp);
break;
default:
break;
diff --git a/drivers/infiniband/hw/qedr/qedr_roce_cm.c b/drivers/infiniband/hw/qedr/qedr_roce_cm.c
index 2bdbb12bfc69..0f14e687bb91 100644
--- a/drivers/infiniband/hw/qedr/qedr_roce_cm.c
+++ b/drivers/infiniband/hw/qedr/qedr_roce_cm.c
@@ -412,19 +412,11 @@ static inline int qedr_gsi_build_header(struct qedr_dev *dev,
return rc;
}
- if (sgid_attr.ndev) {
- vlan_id = rdma_vlan_dev_vlan_id(sgid_attr.ndev);
- if (vlan_id < VLAN_CFI_MASK)
- has_vlan = true;
+ vlan_id = rdma_vlan_dev_vlan_id(sgid_attr.ndev);
+ if (vlan_id < VLAN_CFI_MASK)
+ has_vlan = true;
- dev_put(sgid_attr.ndev);
- }
-
- if (!memcmp(&sgid, &zgid, sizeof(sgid))) {
- DP_ERR(dev, "gsi post send: GID not found GID index %d\n",
- grh->sgid_index);
- return -ENOENT;
- }
+ dev_put(sgid_attr.ndev);
has_udp = (sgid_attr.gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP);
if (!has_udp) {
diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c
index f9c3cc71f5c0..7d3763b2e01c 100644
--- a/drivers/infiniband/hw/qedr/verbs.c
+++ b/drivers/infiniband/hw/qedr/verbs.c
@@ -84,58 +84,6 @@ int qedr_iw_query_gid(struct ib_device *ibdev, u8 port,
return 0;
}
-int qedr_query_gid(struct ib_device *ibdev, u8 port, int index,
- union ib_gid *sgid)
-{
- struct qedr_dev *dev = get_qedr_dev(ibdev);
- int rc = 0;
-
- if (!rdma_cap_roce_gid_table(ibdev, port))
- return -ENODEV;
-
- rc = ib_get_cached_gid(ibdev, port, index, sgid, NULL);
- if (rc == -EAGAIN) {
- memcpy(sgid, &zgid, sizeof(*sgid));
- return 0;
- }
-
- DP_DEBUG(dev, QEDR_MSG_INIT, "query gid: index=%d %llx:%llx\n", index,
- sgid->global.interface_id, sgid->global.subnet_prefix);
-
- return rc;
-}
-
-int qedr_add_gid(struct ib_device *device, u8 port_num,
- unsigned int index, const union ib_gid *gid,
- const struct ib_gid_attr *attr, void **context)
-{
- if (!rdma_cap_roce_gid_table(device, port_num))
- return -EINVAL;
-
- if (port_num > QEDR_MAX_PORT)
- return -EINVAL;
-
- if (!context)
- return -EINVAL;
-
- return 0;
-}
-
-int qedr_del_gid(struct ib_device *device, u8 port_num,
- unsigned int index, void **context)
-{
- if (!rdma_cap_roce_gid_table(device, port_num))
- return -EINVAL;
-
- if (port_num > QEDR_MAX_PORT)
- return -EINVAL;
-
- if (!context)
- return -EINVAL;
-
- return 0;
-}
-
int qedr_query_device(struct ib_device *ibdev,
struct ib_device_attr *attr, struct ib_udata *udata)
{
@@ -525,9 +473,9 @@ struct ib_pd *qedr_alloc_pd(struct ib_device *ibdev,
pd->pd_id = pd_id;
if (udata && context) {
- struct qedr_alloc_pd_uresp uresp;
-
- uresp.pd_id = pd_id;
+ struct qedr_alloc_pd_uresp uresp = {
+ .pd_id = pd_id,
+ };
rc = qedr_ib_copy_to_udata(udata, &uresp, sizeof(uresp));
if (rc) {
@@ -856,8 +804,6 @@ static inline void qedr_init_cq_params(struct qedr_cq *cq,
static void doorbell_cq(struct qedr_cq *cq, u32 cons, u8 flags)
{
- /* Flush data before signalling doorbell */
- wmb();
cq->db.data.agg_flags = flags;
cq->db.data.value = cpu_to_le32(cons);
writeq(cq->db.raw, cq->db_addr);
@@ -1145,46 +1091,41 @@ static inline int get_gid_info_from_table(struct ib_qp *ibqp,
if (rc)
return rc;
- if (!memcmp(&gid, &zgid, sizeof(gid)))
- return -ENOENT;
-
- if (gid_attr.ndev) {
- qp_params->vlan_id = rdma_vlan_dev_vlan_id(gid_attr.ndev);
-
- dev_put(gid_attr.ndev);
- nw_type = ib_gid_to_network_type(gid_attr.gid_type, &gid);
- switch (nw_type) {
- case RDMA_NETWORK_IPV6:
- memcpy(&qp_params->sgid.bytes[0], &gid.raw[0],
- sizeof(qp_params->sgid));
- memcpy(&qp_params->dgid.bytes[0],
- &grh->dgid,
- sizeof(qp_params->dgid));
- qp_params->roce_mode = ROCE_V2_IPV6;
- SET_FIELD(qp_params->modify_flags,
- QED_ROCE_MODIFY_QP_VALID_ROCE_MODE, 1);
- break;
- case RDMA_NETWORK_IB:
- memcpy(&qp_params->sgid.bytes[0], &gid.raw[0],
- sizeof(qp_params->sgid));
- memcpy(&qp_params->dgid.bytes[0],
- &grh->dgid,
- sizeof(qp_params->dgid));
- qp_params->roce_mode = ROCE_V1;
- break;
- case RDMA_NETWORK_IPV4:
- memset(&qp_params->sgid, 0, sizeof(qp_params->sgid));
- memset(&qp_params->dgid, 0, sizeof(qp_params->dgid));
- ipv4_addr = qedr_get_ipv4_from_gid(gid.raw);
- qp_params->sgid.ipv4_addr = ipv4_addr;
- ipv4_addr =
- qedr_get_ipv4_from_gid(grh->dgid.raw);
- qp_params->dgid.ipv4_addr = ipv4_addr;
- SET_FIELD(qp_params->modify_flags,
- QED_ROCE_MODIFY_QP_VALID_ROCE_MODE, 1);
- qp_params->roce_mode = ROCE_V2_IPV4;
- break;
- }
+ qp_params->vlan_id = rdma_vlan_dev_vlan_id(gid_attr.ndev);
+
+ dev_put(gid_attr.ndev);
+ nw_type = ib_gid_to_network_type(gid_attr.gid_type, &gid);
+ switch (nw_type) {
+ case RDMA_NETWORK_IPV6:
+ memcpy(&qp_params->sgid.bytes[0], &gid.raw[0],
+ sizeof(qp_params->sgid));
+ memcpy(&qp_params->dgid.bytes[0],
+ &grh->dgid,
+ sizeof(qp_params->dgid));
+ qp_params->roce_mode = ROCE_V2_IPV6;
+ SET_FIELD(qp_params->modify_flags,
+ QED_ROCE_MODIFY_QP_VALID_ROCE_MODE, 1);
+ break;
+ case RDMA_NETWORK_IB:
+ memcpy(&qp_params->sgid.bytes[0], &gid.raw[0],
+ sizeof(qp_params->sgid));
+ memcpy(&qp_params->dgid.bytes[0],
+ &grh->dgid,
+ sizeof(qp_params->dgid));
+ qp_params->roce_mode = ROCE_V1;
+ break;
+ case RDMA_NETWORK_IPV4:
+ memset(&qp_params->sgid, 0, sizeof(qp_params->sgid));
+ memset(&qp_params->dgid, 0, sizeof(qp_params->dgid));
+ ipv4_addr = qedr_get_ipv4_from_gid(gid.raw);
+ qp_params->sgid.ipv4_addr = ipv4_addr;
+ ipv4_addr =
+ qedr_get_ipv4_from_gid(grh->dgid.raw);
+ qp_params->dgid.ipv4_addr = ipv4_addr;
+ SET_FIELD(qp_params->modify_flags,
+ QED_ROCE_MODIFY_QP_VALID_ROCE_MODE, 1);
+ qp_params->roce_mode = ROCE_V2_IPV4;
+ break;
}
for (i = 0; i < 4; i++) {
@@ -1870,7 +1811,6 @@ static int qedr_update_qp_state(struct qedr_dev *dev,
*/
if (rdma_protocol_roce(&dev->ibdev, 1)) {
- wmb();
writel(qp->rq.db_data.raw, qp->rq.db);
/* Make sure write takes effect */
mmiowb();
@@ -3274,8 +3214,15 @@ int qedr_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
* vane. However this is not harmful (as long as the producer value is
* unchanged). For performance reasons we avoid checking for this
* redundant doorbell.
+ *
+ * qp->wqe_wr_id is accessed during qedr_poll_cq, as
+ * soon as we give the doorbell, we could get a completion
+ * for this wr, therefore we need to make sure that the
+ * memory is updated before giving the doorbell.
+ * During qedr_poll_cq, rmb is called before accessing the
+ * cqe. This covers for the smp_rmb as well.
*/
- wmb();
+ smp_wmb();
writel(qp->sq.db_data.raw, qp->sq.db);
/* Make sure write sticks */
@@ -3362,8 +3309,14 @@ int qedr_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
qedr_inc_sw_prod(&qp->rq);
- /* Flush all the writes before signalling doorbell */
- wmb();
+ /* qp->rqe_wr_id is accessed during qedr_poll_cq, as
+ * soon as we give the doorbell, we could get a completion
+ * for this wr, therefore we need to make sure that the
+ * memory is update before giving the doorbell.
+ * During qedr_poll_cq, rmb is called before accessing the
+ * cqe. This covers for the smp_rmb as well.
+ */
+ smp_wmb();
qp->rq.db_data.data.value++;
diff --git a/drivers/infiniband/hw/qedr/verbs.h b/drivers/infiniband/hw/qedr/verbs.h
index 1a94425dea33..2c57e4c592a6 100644
--- a/drivers/infiniband/hw/qedr/verbs.h
+++ b/drivers/infiniband/hw/qedr/verbs.h
@@ -38,7 +38,6 @@ int qedr_query_port(struct ib_device *, u8 port, struct ib_port_attr *props);
int qedr_modify_port(struct ib_device *, u8 port, int mask,
struct ib_port_modify *props);
-int qedr_query_gid(struct ib_device *, u8 port, int index, union ib_gid *gid);
int qedr_iw_query_gid(struct ib_device *ibdev, u8 port,
int index, union ib_gid *gid);
@@ -48,11 +47,6 @@ struct ib_ucontext *qedr_alloc_ucontext(struct ib_device *, struct ib_udata *);
int qedr_dealloc_ucontext(struct ib_ucontext *);
int qedr_mmap(struct ib_ucontext *, struct vm_area_struct *vma);
-int qedr_del_gid(struct ib_device *device, u8 port_num,
- unsigned int index, void **context);
-int qedr_add_gid(struct ib_device *device, u8 port_num,
- unsigned int index, const union ib_gid *gid,
- const struct ib_gid_attr *attr, void **context);
struct ib_pd *qedr_alloc_pd(struct ib_device *,
struct ib_ucontext *, struct ib_udata *);
int qedr_dealloc_pd(struct ib_pd *pd);
diff --git a/drivers/infiniband/hw/qib/qib.h b/drivers/infiniband/hw/qib/qib.h
index 0235f76bbc72..46072455130c 100644
--- a/drivers/infiniband/hw/qib/qib.h
+++ b/drivers/infiniband/hw/qib/qib.h
@@ -472,9 +472,6 @@ enum qib_sdma_events {
qib_sdma_event_e90_timer_tick,
};
-extern char *qib_sdma_state_names[];
-extern char *qib_sdma_event_names[];
-
struct sdma_set_state_action {
unsigned op_enable:1;
unsigned op_intenable:1;
diff --git a/drivers/infiniband/hw/qib/qib_diag.c b/drivers/infiniband/hw/qib/qib_diag.c
index a9377eee8734..11da796dd1b7 100644
--- a/drivers/infiniband/hw/qib/qib_diag.c
+++ b/drivers/infiniband/hw/qib/qib_diag.c
@@ -614,7 +614,7 @@ static ssize_t qib_diagpkt_write(struct file *fp,
}
if (copy_from_user(tmpbuf,
- (const void __user *) (unsigned long) dp.data,
+ u64_to_user_ptr(dp.data),
dp.len)) {
ret = -EFAULT;
goto bail;
diff --git a/drivers/infiniband/hw/qib/qib_file_ops.c b/drivers/infiniband/hw/qib/qib_file_ops.c
index 52c29db3a2f4..6a8800b65047 100644
--- a/drivers/infiniband/hw/qib/qib_file_ops.c
+++ b/drivers/infiniband/hw/qib/qib_file_ops.c
@@ -443,7 +443,7 @@ cleanup:
ret = -EFAULT;
goto cleanup;
}
- if (copy_to_user((void __user *) (unsigned long) ti->tidmap,
+ if (copy_to_user(u64_to_user_ptr(ti->tidmap),
tidmap, sizeof(tidmap))) {
ret = -EFAULT;
goto cleanup;
@@ -490,7 +490,7 @@ static int qib_tid_free(struct qib_ctxtdata *rcd, unsigned subctxt,
goto done;
}
- if (copy_from_user(tidmap, (void __user *)(unsigned long)ti->tidmap,
+ if (copy_from_user(tidmap, u64_to_user_ptr(ti->tidmap),
sizeof(tidmap))) {
ret = -EFAULT;
goto done;
@@ -2168,8 +2168,8 @@ static ssize_t qib_write(struct file *fp, const char __user *data,
ret = qib_do_user_init(fp, &cmd.cmd.user_info);
if (ret)
goto bail;
- ret = qib_get_base_info(fp, (void __user *) (unsigned long)
- cmd.cmd.user_info.spu_base_info,
+ ret = qib_get_base_info(fp, u64_to_user_ptr(
+ cmd.cmd.user_info.spu_base_info),
cmd.cmd.user_info.spu_base_info_size);
break;
diff --git a/drivers/infiniband/hw/qib/qib_iba7322.c b/drivers/infiniband/hw/qib/qib_iba7322.c
index 6265dac415fc..8414ae44a518 100644
--- a/drivers/infiniband/hw/qib/qib_iba7322.c
+++ b/drivers/infiniband/hw/qib/qib_iba7322.c
@@ -463,6 +463,16 @@ static u8 ib_rate_to_delay[IB_RATE_120_GBPS + 1] = {
[IB_RATE_40_GBPS] = 1
};
+static const char * const qib_sdma_state_names[] = {
+ [qib_sdma_state_s00_hw_down] = "s00_HwDown",
+ [qib_sdma_state_s10_hw_start_up_wait] = "s10_HwStartUpWait",
+ [qib_sdma_state_s20_idle] = "s20_Idle",
+ [qib_sdma_state_s30_sw_clean_up_wait] = "s30_SwCleanUpWait",
+ [qib_sdma_state_s40_hw_clean_up_wait] = "s40_HwCleanUpWait",
+ [qib_sdma_state_s50_hw_halt_wait] = "s50_HwHaltWait",
+ [qib_sdma_state_s99_running] = "s99_Running",
+};
+
#define IBA7322_LINKSPEED_SHIFT SYM_LSB(IBCStatusA_0, LinkSpeedActive)
#define IBA7322_LINKWIDTH_SHIFT SYM_LSB(IBCStatusA_0, LinkWidthActive)
diff --git a/drivers/infiniband/hw/qib/qib_init.c b/drivers/infiniband/hw/qib/qib_init.c
index 3990f386aa32..6c68f8a97018 100644
--- a/drivers/infiniband/hw/qib/qib_init.c
+++ b/drivers/infiniband/hw/qib/qib_init.c
@@ -678,11 +678,9 @@ int qib_init(struct qib_devdata *dd, int reinit)
lastfail = qib_create_rcvhdrq(dd, rcd);
if (!lastfail)
lastfail = qib_setup_eagerbufs(rcd);
- if (lastfail) {
+ if (lastfail)
qib_dev_err(dd,
"failed to allocate kernel ctxt's rcvhdrq and/or egr bufs\n");
- continue;
- }
}
for (pidx = 0; pidx < dd->num_pports; ++pidx) {
diff --git a/drivers/infiniband/hw/qib/qib_sdma.c b/drivers/infiniband/hw/qib/qib_sdma.c
index c3690bd51582..d0723d4aef5c 100644
--- a/drivers/infiniband/hw/qib/qib_sdma.c
+++ b/drivers/infiniband/hw/qib/qib_sdma.c
@@ -54,30 +54,6 @@ MODULE_PARM_DESC(sdma_descq_cnt, "Number of SDMA descq entries");
#define SDMA_DESC_COUNT_LSB 16
#define SDMA_DESC_GEN_LSB 30
-char *qib_sdma_state_names[] = {
- [qib_sdma_state_s00_hw_down] = "s00_HwDown",
- [qib_sdma_state_s10_hw_start_up_wait] = "s10_HwStartUpWait",
- [qib_sdma_state_s20_idle] = "s20_Idle",
- [qib_sdma_state_s30_sw_clean_up_wait] = "s30_SwCleanUpWait",
- [qib_sdma_state_s40_hw_clean_up_wait] = "s40_HwCleanUpWait",
- [qib_sdma_state_s50_hw_halt_wait] = "s50_HwHaltWait",
- [qib_sdma_state_s99_running] = "s99_Running",
-};
-
-char *qib_sdma_event_names[] = {
- [qib_sdma_event_e00_go_hw_down] = "e00_GoHwDown",
- [qib_sdma_event_e10_go_hw_start] = "e10_GoHwStart",
- [qib_sdma_event_e20_hw_started] = "e20_HwStarted",
- [qib_sdma_event_e30_go_running] = "e30_GoRunning",
- [qib_sdma_event_e40_sw_cleaned] = "e40_SwCleaned",
- [qib_sdma_event_e50_hw_cleaned] = "e50_HwCleaned",
- [qib_sdma_event_e60_hw_halted] = "e60_HwHalted",
- [qib_sdma_event_e70_go_idle] = "e70_GoIdle",
- [qib_sdma_event_e7220_err_halted] = "e7220_ErrHalted",
- [qib_sdma_event_e7322_err_halted] = "e7322_ErrHalted",
- [qib_sdma_event_e90_timer_tick] = "e90_TimerTick",
-};
-
/* declare all statics here rather than keep sorting */
static int alloc_sdma(struct qib_pportdata *);
static void sdma_complete(struct kref *);
diff --git a/drivers/infiniband/hw/qib/qib_verbs.c b/drivers/infiniband/hw/qib/qib_verbs.c
index fabee760407e..3977abbc83ad 100644
--- a/drivers/infiniband/hw/qib/qib_verbs.c
+++ b/drivers/infiniband/hw/qib/qib_verbs.c
@@ -1646,7 +1646,7 @@ int qib_register_ib_device(struct qib_devdata *dd)
dd->rcd[ctxt]->pkeys);
}
- ret = rvt_register_device(&dd->verbs_dev.rdi);
+ ret = rvt_register_device(&dd->verbs_dev.rdi, RDMA_DRIVER_QIB);
if (ret)
goto err_tx;
diff --git a/drivers/infiniband/hw/usnic/usnic_ib_main.c b/drivers/infiniband/hw/usnic/usnic_ib_main.c
index ca5638091b55..f0538a460328 100644
--- a/drivers/infiniband/hw/usnic/usnic_ib_main.c
+++ b/drivers/infiniband/hw/usnic/usnic_ib_main.c
@@ -415,6 +415,7 @@ static void *usnic_ib_device_add(struct pci_dev *dev)
us_ibdev->ib_dev.get_dev_fw_str = usnic_get_dev_fw_str;
+ us_ibdev->ib_dev.driver_id = RDMA_DRIVER_USNIC;
if (ib_register_device(&us_ibdev->ib_dev, NULL))
goto err_fwd_dealloc;
diff --git a/drivers/infiniband/hw/usnic/usnic_transport.c b/drivers/infiniband/hw/usnic/usnic_transport.c
index 67de94343cb4..e0a95538c364 100644
--- a/drivers/infiniband/hw/usnic/usnic_transport.c
+++ b/drivers/infiniband/hw/usnic/usnic_transport.c
@@ -200,10 +200,8 @@ int usnic_transport_sock_get_addr(struct socket *sock, int *proto,
int usnic_transport_init(void)
{
roce_bitmap = kzalloc(ROCE_BITMAP_SZ, GFP_KERNEL);
- if (!roce_bitmap) {
- usnic_err("Failed to allocate bit map");
+ if (!roce_bitmap)
return -ENOMEM;
- }
/* Do not ever allocate bit 0, hence set it here */
bitmap_set(roce_bitmap, 0, 1);
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
index d650a9fcde24..0be33a81bbe6 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c
@@ -62,17 +62,10 @@ static DEFINE_MUTEX(pvrdma_device_list_lock);
static LIST_HEAD(pvrdma_device_list);
static struct workqueue_struct *event_wq;
-static int pvrdma_add_gid(struct ib_device *ibdev,
- u8 port_num,
- unsigned int index,
- const union ib_gid *gid,
+static int pvrdma_add_gid(const union ib_gid *gid,
const struct ib_gid_attr *attr,
void **context);
-static int pvrdma_del_gid(struct ib_device *ibdev,
- u8 port_num,
- unsigned int index,
- void **context);
-
+static int pvrdma_del_gid(const struct ib_gid_attr *attr, void **context);
static ssize_t show_hca(struct device *device, struct device_attribute *attr,
char *buf)
@@ -276,6 +269,7 @@ static int pvrdma_register_device(struct pvrdma_dev *dev)
if (!dev->srq_tbl)
goto err_qp_free;
}
+ dev->ib_dev.driver_id = RDMA_DRIVER_VMW_PVRDMA;
spin_lock_init(&dev->srq_tbl_lock);
ret = ib_register_device(&dev->ib_dev, NULL);
@@ -656,18 +650,15 @@ static int pvrdma_add_gid_at_index(struct pvrdma_dev *dev,
return 0;
}
-static int pvrdma_add_gid(struct ib_device *ibdev,
- u8 port_num,
- unsigned int index,
- const union ib_gid *gid,
+static int pvrdma_add_gid(const union ib_gid *gid,
const struct ib_gid_attr *attr,
void **context)
{
- struct pvrdma_dev *dev = to_vdev(ibdev);
+ struct pvrdma_dev *dev = to_vdev(attr->device);
return pvrdma_add_gid_at_index(dev, gid,
ib_gid_type_to_pvrdma(attr->gid_type),
- index);
+ attr->index);
}
static int pvrdma_del_gid_at_index(struct pvrdma_dev *dev, int index)
@@ -697,17 +688,14 @@ static int pvrdma_del_gid_at_index(struct pvrdma_dev *dev, int index)
return 0;
}
-static int pvrdma_del_gid(struct ib_device *ibdev,
- u8 port_num,
- unsigned int index,
- void **context)
+static int pvrdma_del_gid(const struct ib_gid_attr *attr, void **context)
{
- struct pvrdma_dev *dev = to_vdev(ibdev);
+ struct pvrdma_dev *dev = to_vdev(attr->device);
dev_dbg(&dev->pdev->dev, "removing gid at index %u from %s",
- index, dev->netdev->name);
+ attr->index, dev->netdev->name);
- return pvrdma_del_gid_at_index(dev, index);
+ return pvrdma_del_gid_at_index(dev, attr->index);
}
static void pvrdma_netdevice_event_handle(struct pvrdma_dev *dev,
diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c
index 7bf518bdbf21..eb5b1065ec08 100644
--- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c
+++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c
@@ -489,7 +489,7 @@ int pvrdma_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
union pvrdma_cmd_req req;
union pvrdma_cmd_resp rsp;
struct pvrdma_cmd_modify_qp *cmd = &req.modify_qp;
- int cur_state, next_state;
+ enum ib_qp_state cur_state, next_state;
int ret;
/* Sanity checking. Should need lock here */