Commit dcdede04 authored by santosh.shilimkar@oracle.com's avatar santosh.shilimkar@oracle.com Committed by David S. Miller

RDS: Drop stale iWARP RDMA transport

RDS iWarp support code has become stale and non testable. As
indicated earlier, am dropping the support for it.

If new iWarp user(s) shows up in future, we can adapat the RDS IB
transprt for the special RDMA READ sink case. iWarp needs an MR
for the RDMA READ sink.
Signed-off-by: default avatarSantosh Shilimkar <ssantosh@kernel.org>
Signed-off-by: default avatarSantosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parent afc3de92
......@@ -19,9 +19,7 @@ to N*N if you use a connection-oriented socket transport like TCP.
RDS is not Infiniband-specific; it was designed to support different
transports. The current implementation used to support RDS over TCP as well
as IB. Work is in progress to support RDS over iWARP, and using DCE to
guarantee no dropped packets on Ethernet, it may be possible to use RDS over
UDP in the future.
as IB.
The high-level semantics of RDS from the application's point of view are
......
......@@ -4,14 +4,13 @@ config RDS
depends on INET
---help---
The RDS (Reliable Datagram Sockets) protocol provides reliable,
sequenced delivery of datagrams over Infiniband, iWARP,
or TCP.
sequenced delivery of datagrams over Infiniband or TCP.
config RDS_RDMA
tristate "RDS over Infiniband and iWARP"
tristate "RDS over Infiniband"
depends on RDS && INFINIBAND && INFINIBAND_ADDR_TRANS
---help---
Allow RDS to use Infiniband and iWARP as a transport.
Allow RDS to use Infiniband as a transport.
This transport supports RDMA operations.
config RDS_TCP
......
......@@ -6,9 +6,7 @@ rds-y := af_rds.o bind.o cong.o connection.o info.o message.o \
obj-$(CONFIG_RDS_RDMA) += rds_rdma.o
rds_rdma-y := rdma_transport.o \
ib.o ib_cm.o ib_recv.o ib_ring.o ib_send.o ib_stats.o \
ib_sysctl.o ib_rdma.o \
iw.o iw_cm.o iw_recv.o iw_ring.o iw_send.o iw_stats.o \
iw_sysctl.o iw_rdma.o
ib_sysctl.o ib_rdma.o
obj-$(CONFIG_RDS_TCP) += rds_tcp.o
......
/*
* Copyright (c) 2006 Oracle. 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.
*
*/
#include <linux/kernel.h>
#include <linux/in.h>
#include <linux/if.h>
#include <linux/netdevice.h>
#include <linux/inetdevice.h>
#include <linux/if_arp.h>
#include <linux/delay.h>
#include <linux/slab.h>
#include <linux/module.h>
#include "rds.h"
#include "iw.h"
unsigned int fastreg_pool_size = RDS_FASTREG_POOL_SIZE;
unsigned int fastreg_message_size = RDS_FASTREG_SIZE + 1; /* +1 allows for unaligned MRs */
module_param(fastreg_pool_size, int, 0444);
MODULE_PARM_DESC(fastreg_pool_size, " Max number of fastreg MRs per device");
module_param(fastreg_message_size, int, 0444);
MODULE_PARM_DESC(fastreg_message_size, " Max size of a RDMA transfer (fastreg MRs)");
struct list_head rds_iw_devices;
/* NOTE: if also grabbing iwdev lock, grab this first */
DEFINE_SPINLOCK(iw_nodev_conns_lock);
LIST_HEAD(iw_nodev_conns);
static void rds_iw_add_one(struct ib_device *device)
{
struct rds_iw_device *rds_iwdev;
/* Only handle iwarp devices */
if (device->node_type != RDMA_NODE_RNIC)
return;
rds_iwdev = kmalloc(sizeof *rds_iwdev, GFP_KERNEL);
if (!rds_iwdev)
return;
spin_lock_init(&rds_iwdev->spinlock);
rds_iwdev->dma_local_lkey = !!(device->attrs.device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY);
rds_iwdev->max_wrs = device->attrs.max_qp_wr;
rds_iwdev->max_sge = min(device->attrs.max_sge, RDS_IW_MAX_SGE);
rds_iwdev->dev = device;
rds_iwdev->pd = ib_alloc_pd(device);
if (IS_ERR(rds_iwdev->pd))
goto free_dev;
if (!rds_iwdev->dma_local_lkey) {
rds_iwdev->mr = ib_get_dma_mr(rds_iwdev->pd,
IB_ACCESS_REMOTE_READ |
IB_ACCESS_REMOTE_WRITE |
IB_ACCESS_LOCAL_WRITE);
if (IS_ERR(rds_iwdev->mr))
goto err_pd;
} else
rds_iwdev->mr = NULL;
rds_iwdev->mr_pool = rds_iw_create_mr_pool(rds_iwdev);
if (IS_ERR(rds_iwdev->mr_pool)) {
rds_iwdev->mr_pool = NULL;
goto err_mr;
}
INIT_LIST_HEAD(&rds_iwdev->cm_id_list);
INIT_LIST_HEAD(&rds_iwdev->conn_list);
list_add_tail(&rds_iwdev->list, &rds_iw_devices);
ib_set_client_data(device, &rds_iw_client, rds_iwdev);
return;
err_mr:
if (rds_iwdev->mr)
ib_dereg_mr(rds_iwdev->mr);
err_pd:
ib_dealloc_pd(rds_iwdev->pd);
free_dev:
kfree(rds_iwdev);
}
static void rds_iw_remove_one(struct ib_device *device, void *client_data)
{
struct rds_iw_device *rds_iwdev = client_data;
struct rds_iw_cm_id *i_cm_id, *next;
if (!rds_iwdev)
return;
spin_lock_irq(&rds_iwdev->spinlock);
list_for_each_entry_safe(i_cm_id, next, &rds_iwdev->cm_id_list, list) {
list_del(&i_cm_id->list);
kfree(i_cm_id);
}
spin_unlock_irq(&rds_iwdev->spinlock);
rds_iw_destroy_conns(rds_iwdev);
if (rds_iwdev->mr_pool)
rds_iw_destroy_mr_pool(rds_iwdev->mr_pool);
if (rds_iwdev->mr)
ib_dereg_mr(rds_iwdev->mr);
ib_dealloc_pd(rds_iwdev->pd);
list_del(&rds_iwdev->list);
kfree(rds_iwdev);
}
struct ib_client rds_iw_client = {
.name = "rds_iw",
.add = rds_iw_add_one,
.remove = rds_iw_remove_one
};
static int rds_iw_conn_info_visitor(struct rds_connection *conn,
void *buffer)
{
struct rds_info_rdma_connection *iinfo = buffer;
struct rds_iw_connection *ic;
/* We will only ever look at IB transports */
if (conn->c_trans != &rds_iw_transport)
return 0;
iinfo->src_addr = conn->c_laddr;
iinfo->dst_addr = conn->c_faddr;
memset(&iinfo->src_gid, 0, sizeof(iinfo->src_gid));
memset(&iinfo->dst_gid, 0, sizeof(iinfo->dst_gid));
if (rds_conn_state(conn) == RDS_CONN_UP) {
struct rds_iw_device *rds_iwdev;
struct rdma_dev_addr *dev_addr;
ic = conn->c_transport_data;
dev_addr = &ic->i_cm_id->route.addr.dev_addr;
rdma_addr_get_sgid(dev_addr, (union ib_gid *) &iinfo->src_gid);
rdma_addr_get_dgid(dev_addr, (union ib_gid *) &iinfo->dst_gid);
rds_iwdev = ib_get_client_data(ic->i_cm_id->device, &rds_iw_client);
iinfo->max_send_wr = ic->i_send_ring.w_nr;
iinfo->max_recv_wr = ic->i_recv_ring.w_nr;
iinfo->max_send_sge = rds_iwdev->max_sge;
rds_iw_get_mr_info(rds_iwdev, iinfo);
}
return 1;
}
static void rds_iw_ic_info(struct socket *sock, unsigned int len,
struct rds_info_iterator *iter,
struct rds_info_lengths *lens)
{
rds_for_each_conn_info(sock, len, iter, lens,
rds_iw_conn_info_visitor,
sizeof(struct rds_info_rdma_connection));
}
/*
* Early RDS/IB was built to only bind to an address if there is an IPoIB
* device with that address set.
*
* If it were me, I'd advocate for something more flexible. Sending and
* receiving should be device-agnostic. Transports would try and maintain
* connections between peers who have messages queued. Userspace would be
* allowed to influence which paths have priority. We could call userspace
* asserting this policy "routing".
*/
static int rds_iw_laddr_check(struct net *net, __be32 addr)
{
int ret;
struct rdma_cm_id *cm_id;
struct sockaddr_in sin;
/* Create a CMA ID and try to bind it. This catches both
* IB and iWARP capable NICs.
*/
cm_id = rdma_create_id(&init_net, NULL, NULL, RDMA_PS_TCP, IB_QPT_RC);
if (IS_ERR(cm_id))
return PTR_ERR(cm_id);
memset(&sin, 0, sizeof(sin));
sin.sin_family = AF_INET;
sin.sin_addr.s_addr = addr;
/* rdma_bind_addr will only succeed for IB & iWARP devices */
ret = rdma_bind_addr(cm_id, (struct sockaddr *)&sin);
/* due to this, we will claim to support IB devices unless we
check node_type. */
if (ret || !cm_id->device ||
cm_id->device->node_type != RDMA_NODE_RNIC)
ret = -EADDRNOTAVAIL;
rdsdebug("addr %pI4 ret %d node type %d\n",
&addr, ret,
cm_id->device ? cm_id->device->node_type : -1);
rdma_destroy_id(cm_id);
return ret;
}
void rds_iw_exit(void)
{
rds_info_deregister_func(RDS_INFO_IWARP_CONNECTIONS, rds_iw_ic_info);
rds_iw_destroy_nodev_conns();
ib_unregister_client(&rds_iw_client);
rds_iw_sysctl_exit();
rds_iw_recv_exit();
rds_trans_unregister(&rds_iw_transport);
}
struct rds_transport rds_iw_transport = {
.laddr_check = rds_iw_laddr_check,
.xmit_complete = rds_iw_xmit_complete,
.xmit = rds_iw_xmit,
.xmit_rdma = rds_iw_xmit_rdma,
.recv = rds_iw_recv,
.conn_alloc = rds_iw_conn_alloc,
.conn_free = rds_iw_conn_free,
.conn_connect = rds_iw_conn_connect,
.conn_shutdown = rds_iw_conn_shutdown,
.inc_copy_to_user = rds_iw_inc_copy_to_user,
.inc_free = rds_iw_inc_free,
.cm_initiate_connect = rds_iw_cm_initiate_connect,
.cm_handle_connect = rds_iw_cm_handle_connect,
.cm_connect_complete = rds_iw_cm_connect_complete,
.stats_info_copy = rds_iw_stats_info_copy,
.exit = rds_iw_exit,
.get_mr = rds_iw_get_mr,
.sync_mr = rds_iw_sync_mr,
.free_mr = rds_iw_free_mr,
.flush_mrs = rds_iw_flush_mrs,
.t_owner = THIS_MODULE,
.t_name = "iwarp",
.t_type = RDS_TRANS_IWARP,
.t_prefer_loopback = 1,
};
int rds_iw_init(void)
{
int ret;
INIT_LIST_HEAD(&rds_iw_devices);
ret = ib_register_client(&rds_iw_client);
if (ret)
goto out;
ret = rds_iw_sysctl_init();
if (ret)
goto out_ibreg;
ret = rds_iw_recv_init();
if (ret)
goto out_sysctl;
ret = rds_trans_register(&rds_iw_transport);
if (ret)
goto out_recv;
rds_info_register_func(RDS_INFO_IWARP_CONNECTIONS, rds_iw_ic_info);
goto out;
out_recv:
rds_iw_recv_exit();
out_sysctl:
rds_iw_sysctl_exit();
out_ibreg:
ib_unregister_client(&rds_iw_client);
out:
return ret;
}
MODULE_LICENSE("GPL");
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
/*
* Copyright (c) 2006 Oracle. 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.
*
*/
#include <linux/kernel.h>
#include "rds.h"
#include "iw.h"
/*
* Locking for IB rings.
* We assume that allocation is always protected by a mutex
* in the caller (this is a valid assumption for the current
* implementation).
*
* Freeing always happens in an interrupt, and hence only
* races with allocations, but not with other free()s.
*
* The interaction between allocation and freeing is that
* the alloc code has to determine the number of free entries.
* To this end, we maintain two counters; an allocation counter
* and a free counter. Both are allowed to run freely, and wrap
* around.
* The number of used entries is always (alloc_ctr - free_ctr) % NR.
*
* The current implementation makes free_ctr atomic. When the
* caller finds an allocation fails, it should set an "alloc fail"
* bit and retry the allocation. The "alloc fail" bit essentially tells
* the CQ completion handlers to wake it up after freeing some
* more entries.
*/
/*
* This only happens on shutdown.
*/
DECLARE_WAIT_QUEUE_HEAD(rds_iw_ring_empty_wait);
void rds_iw_ring_init(struct rds_iw_work_ring *ring, u32 nr)
{
memset(ring, 0, sizeof(*ring));
ring->w_nr = nr;
rdsdebug("ring %p nr %u\n", ring, ring->w_nr);
}
static inline u32 __rds_iw_ring_used(struct rds_iw_work_ring *ring)
{
u32 diff;
/* This assumes that atomic_t has at least as many bits as u32 */
diff = ring->w_alloc_ctr - (u32) atomic_read(&ring->w_free_ctr);
BUG_ON(diff > ring->w_nr);
return diff;
}
void rds_iw_ring_resize(struct rds_iw_work_ring *ring, u32 nr)
{
/* We only ever get called from the connection setup code,
* prior to creating the QP. */
BUG_ON(__rds_iw_ring_used(ring));
ring->w_nr = nr;
}
static int __rds_iw_ring_empty(struct rds_iw_work_ring *ring)
{
return __rds_iw_ring_used(ring) == 0;
}
u32 rds_iw_ring_alloc(struct rds_iw_work_ring *ring, u32 val, u32 *pos)
{
u32 ret = 0, avail;
avail = ring->w_nr - __rds_iw_ring_used(ring);
rdsdebug("ring %p val %u next %u free %u\n", ring, val,
ring->w_alloc_ptr, avail);
if (val && avail) {
ret = min(val, avail);
*pos = ring->w_alloc_ptr;
ring->w_alloc_ptr = (ring->w_alloc_ptr + ret) % ring->w_nr;
ring->w_alloc_ctr += ret;
}
return ret;
}
void rds_iw_ring_free(struct rds_iw_work_ring *ring, u32 val)
{
ring->w_free_ptr = (ring->w_free_ptr + val) % ring->w_nr;
atomic_add(val, &ring->w_free_ctr);
if (__rds_iw_ring_empty(ring) &&
waitqueue_active(&rds_iw_ring_empty_wait))
wake_up(&rds_iw_ring_empty_wait);
}
void rds_iw_ring_unalloc(struct rds_iw_work_ring *ring, u32 val)
{
ring->w_alloc_ptr = (ring->w_alloc_ptr - val) % ring->w_nr;
ring->w_alloc_ctr -= val;
}
int rds_iw_ring_empty(struct rds_iw_work_ring *ring)
{
return __rds_iw_ring_empty(ring);
}
int rds_iw_ring_low(struct rds_iw_work_ring *ring)
{
return __rds_iw_ring_used(ring) <= (ring->w_nr >> 1);
}
/*
* returns the oldest alloced ring entry. This will be the next one
* freed. This can't be called if there are none allocated.
*/
u32 rds_iw_ring_oldest(struct rds_iw_work_ring *ring)
{
return ring->w_free_ptr;
}
/*
* returns the number of completed work requests.
*/
u32 rds_iw_ring_completed(struct rds_iw_work_ring *ring, u32 wr_id, u32 oldest)
{
u32 ret;
if (oldest <= (unsigned long long)wr_id)
ret = (unsigned long long)wr_id - oldest + 1;
else
ret = ring->w_nr - oldest + (unsigned long long)wr_id + 1;
rdsdebug("ring %p ret %u wr_id %u oldest %u\n", ring, ret,
wr_id, oldest);
return ret;
}
This diff is collapsed.
/*
* Copyright (c) 2006 Oracle. 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.
*
*/
#include <linux/percpu.h>
#include <linux/seq_file.h>
#include <linux/proc_fs.h>
#include "rds.h"
#include "iw.h"
DEFINE_PER_CPU_SHARED_ALIGNED(struct rds_iw_statistics, rds_iw_stats);
static const char *const rds_iw_stat_names[] = {
"iw_connect_raced",
"iw_listen_closed_stale",
"iw_tx_cq_call",
"iw_tx_cq_event",
"iw_tx_ring_full",
"iw_tx_throttle",
"iw_tx_sg_mapping_failure",
"iw_tx_stalled",
"iw_tx_credit_updates",
"iw_rx_cq_call",
"iw_rx_cq_event",
"iw_rx_ring_empty",
"iw_rx_refill_from_cq",
"iw_rx_refill_from_thread",
"iw_rx_alloc_limit",
"iw_rx_credit_updates",
"iw_ack_sent",
"iw_ack_send_failure",
"iw_ack_send_delayed",
"iw_ack_send_piggybacked",
"iw_ack_received",
"iw_rdma_mr_alloc",
"iw_rdma_mr_free",
"iw_rdma_mr_used",
"iw_rdma_mr_pool_flush",
"iw_rdma_mr_pool_wait",
"iw_rdma_mr_pool_depleted",
};
unsigned int rds_iw_stats_info_copy(struct rds_info_iterator *iter,
unsigned int avail)
{
struct rds_iw_statistics stats = {0, };
uint64_t *src;
uint64_t *sum;
size_t i;
int cpu;
if (avail < ARRAY_SIZE(rds_iw_stat_names))
goto out;
for_each_online_cpu(cpu) {
src = (uint64_t *)&(per_cpu(rds_iw_stats, cpu));
sum = (uint64_t *)&stats;
for (i = 0; i < sizeof(stats) / sizeof(uint64_t); i++)
*(sum++) += *(src++);
}
rds_stats_info_copy(iter, (uint64_t *)&stats, rds_iw_stat_names,
ARRAY_SIZE(rds_iw_stat_names));
out:
return ARRAY_SIZE(rds_iw_stat_names);
}
/*
* Copyright (c) 2006 Oracle. 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.
*
*/
#include <linux/kernel.h>
#include <linux/sysctl.h>
#include <linux/proc_fs.h>
#include "iw.h"
static struct ctl_table_header *rds_iw_sysctl_hdr;
unsigned long rds_iw_sysctl_max_send_wr = RDS_IW_DEFAULT_SEND_WR;
unsigned long rds_iw_sysctl_max_recv_wr = RDS_IW_DEFAULT_RECV_WR;
unsigned long rds_iw_sysctl_max_recv_allocation = (128 * 1024 * 1024) / RDS_FRAG_SIZE;
static unsigned long rds_iw_sysctl_max_wr_min = 1;
/* hardware will fail CQ creation long before this */
static unsigned long rds_iw_sysctl_max_wr_max = (u32)~0;
unsigned long rds_iw_sysctl_max_unsig_wrs = 16;
static unsigned long rds_iw_sysctl_max_unsig_wr_min = 1;
static unsigned long rds_iw_sysctl_max_unsig_wr_max = 64;
unsigned long rds_iw_sysctl_max_unsig_bytes = (16 << 20);
static unsigned long rds_iw_sysctl_max_unsig_bytes_min = 1;
static unsigned long rds_iw_sysctl_max_unsig_bytes_max = ~0UL;
unsigned int rds_iw_sysctl_flow_control = 1;
static struct ctl_table rds_iw_sysctl_table[] = {
{
.procname = "max_send_wr",
.data = &rds_iw_sysctl_max_send_wr,
.maxlen = sizeof(unsigned long),
.mode = 0644,
.proc_handler = proc_doulongvec_minmax,
.extra1 = &rds_iw_sysctl_max_wr_min,
.extra2 = &rds_iw_sysctl_max_wr_max,
},
{
.procname = "max_recv_wr",
.data = &rds_iw_sysctl_max_recv_wr,
.maxlen = sizeof(unsigned long),
.mode = 0644,
.proc_handler = proc_doulongvec_minmax,
.extra1 = &rds_iw_sysctl_max_wr_min,
.extra2 = &rds_iw_sysctl_max_wr_max,
},
{
.procname = "max_unsignaled_wr",
.data = &rds_iw_sysctl_max_unsig_wrs,
.maxlen = sizeof(unsigned long),
.mode = 0644,
.proc_handler = proc_doulongvec_minmax,
.extra1 = &rds_iw_sysctl_max_unsig_wr_min,
.extra2 = &rds_iw_sysctl_max_unsig_wr_max,
},
{
.procname = "max_unsignaled_bytes",
.data = &rds_iw_sysctl_max_unsig_bytes,
.maxlen = sizeof(unsigned long),
.mode = 0644,
.proc_handler = proc_doulongvec_minmax,
.extra1 = &rds_iw_sysctl_max_unsig_bytes_min,
.extra2 = &rds_iw_sysctl_max_unsig_bytes_max,
},
{
.procname = "max_recv_allocation",
.data = &rds_iw_sysctl_max_recv_allocation,
.maxlen = sizeof(unsigned long),
.mode = 0644,
.proc_handler = proc_doulongvec_minmax,
},
{
.procname = "flow_control",
.data = &rds_iw_sysctl_flow_control,
.maxlen = sizeof(rds_iw_sysctl_flow_control),
.mode = 0644,
.proc_handler = proc_dointvec,
},
{ }
};
void rds_iw_sysctl_exit(void)
{
unregister_net_sysctl_table(rds_iw_sysctl_hdr);
}
int rds_iw_sysctl_init(void)
{
rds_iw_sysctl_hdr = register_net_sysctl(&init_net, "net/rds/iw", rds_iw_sysctl_table);
if (!rds_iw_sysctl_hdr)
return -ENOMEM;
return 0;
}
......@@ -49,9 +49,7 @@ int rds_rdma_cm_event_handler(struct rdma_cm_id *cm_id,