aboutsummaryrefslogtreecommitdiffstats
path: root/net/rds
diff options
context:
space:
mode:
authorAndy Grover <andy.grover@oracle.com>2009-02-24 10:30:36 -0500
committerDavid S. Miller <davem@davemloft.net>2009-02-27 02:39:33 -0500
commitfcd8b7c0ecf792dd824b2edcd63cb2c08563c340 (patch)
tree9ba10fe626cc9129f9138bc81b1fa61c74097e4f /net/rds
parente6babe4cc4ce48577d743cc0de19a214f2527956 (diff)
RDS: Add iWARP support
Support for iWARP NICs is implemented as a separate RDS transport from IB. The code, however, is very similar to IB (it was forked, basically.) so let's keep it in one changeset. The reason for this duplicationis that despite its similarity to IB, there are a number of places where it has different semantics. iwarp zcopy support is still under development, and giving it its own sandbox ensures that IB code isn't disrupted while iwarp changes. Over time these transports will re-converge. Signed-off-by: Andy Grover <andy.grover@oracle.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/rds')
-rw-r--r--net/rds/iw.c333
-rw-r--r--net/rds/iw.h395
-rw-r--r--net/rds/iw_cm.c750
-rw-r--r--net/rds/iw_rdma.c888
-rw-r--r--net/rds/iw_recv.c869
-rw-r--r--net/rds/iw_ring.c169
-rw-r--r--net/rds/iw_send.c975
-rw-r--r--net/rds/iw_stats.c95
-rw-r--r--net/rds/iw_sysctl.c137
9 files changed, 4611 insertions, 0 deletions
diff --git a/net/rds/iw.c b/net/rds/iw.c
new file mode 100644
index 000000000000..1b56905c4c08
--- /dev/null
+++ b/net/rds/iw.c
@@ -0,0 +1,333 @@
1/*
2 * Copyright (c) 2006 Oracle. All rights reserved.
3 *
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
9 *
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
12 * conditions are met:
13 *
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
16 * disclaimer.
17 *
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
22 *
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30 * SOFTWARE.
31 *
32 */
33#include <linux/kernel.h>
34#include <linux/in.h>
35#include <linux/if.h>
36#include <linux/netdevice.h>
37#include <linux/inetdevice.h>
38#include <linux/if_arp.h>
39#include <linux/delay.h>
40
41#include "rds.h"
42#include "iw.h"
43
44unsigned int fastreg_pool_size = RDS_FASTREG_POOL_SIZE;
45unsigned int fastreg_message_size = RDS_FASTREG_SIZE + 1; /* +1 allows for unaligned MRs */
46
47module_param(fastreg_pool_size, int, 0444);
48MODULE_PARM_DESC(fastreg_pool_size, " Max number of fastreg MRs per device");
49module_param(fastreg_message_size, int, 0444);
50MODULE_PARM_DESC(fastreg_message_size, " Max size of a RDMA transfer (fastreg MRs)");
51
52struct list_head rds_iw_devices;
53
54DEFINE_SPINLOCK(iw_nodev_conns_lock);
55LIST_HEAD(iw_nodev_conns);
56
57void rds_iw_add_one(struct ib_device *device)
58{
59 struct rds_iw_device *rds_iwdev;
60 struct ib_device_attr *dev_attr;
61
62 /* Only handle iwarp devices */
63 if (device->node_type != RDMA_NODE_RNIC)
64 return;
65
66 dev_attr = kmalloc(sizeof *dev_attr, GFP_KERNEL);
67 if (!dev_attr)
68 return;
69
70 if (ib_query_device(device, dev_attr)) {
71 rdsdebug("Query device failed for %s\n", device->name);
72 goto free_attr;
73 }
74
75 rds_iwdev = kmalloc(sizeof *rds_iwdev, GFP_KERNEL);
76 if (!rds_iwdev)
77 goto free_attr;
78
79 spin_lock_init(&rds_iwdev->spinlock);
80
81 rds_iwdev->dma_local_lkey = !!(dev_attr->device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY);
82 rds_iwdev->max_wrs = dev_attr->max_qp_wr;
83 rds_iwdev->max_sge = min(dev_attr->max_sge, RDS_IW_MAX_SGE);
84
85 rds_iwdev->page_shift = max(PAGE_SHIFT, ffs(dev_attr->page_size_cap) - 1);
86
87 rds_iwdev->dev = device;
88 rds_iwdev->pd = ib_alloc_pd(device);
89 if (IS_ERR(rds_iwdev->pd))
90 goto free_dev;
91
92 if (!rds_iwdev->dma_local_lkey) {
93 if (device->node_type != RDMA_NODE_RNIC) {
94 rds_iwdev->mr = ib_get_dma_mr(rds_iwdev->pd,
95 IB_ACCESS_LOCAL_WRITE);
96 } else {
97 rds_iwdev->mr = ib_get_dma_mr(rds_iwdev->pd,
98 IB_ACCESS_REMOTE_READ |
99 IB_ACCESS_REMOTE_WRITE |
100 IB_ACCESS_LOCAL_WRITE);
101 }
102 if (IS_ERR(rds_iwdev->mr))
103 goto err_pd;
104 } else
105 rds_iwdev->mr = NULL;
106
107 rds_iwdev->mr_pool = rds_iw_create_mr_pool(rds_iwdev);
108 if (IS_ERR(rds_iwdev->mr_pool)) {
109 rds_iwdev->mr_pool = NULL;
110 goto err_mr;
111 }
112
113 INIT_LIST_HEAD(&rds_iwdev->cm_id_list);
114 INIT_LIST_HEAD(&rds_iwdev->conn_list);
115 list_add_tail(&rds_iwdev->list, &rds_iw_devices);
116
117 ib_set_client_data(device, &rds_iw_client, rds_iwdev);
118
119 goto free_attr;
120
121err_mr:
122 if (rds_iwdev->mr)
123 ib_dereg_mr(rds_iwdev->mr);
124err_pd:
125 ib_dealloc_pd(rds_iwdev->pd);
126free_dev:
127 kfree(rds_iwdev);
128free_attr:
129 kfree(dev_attr);
130}
131
132void rds_iw_remove_one(struct ib_device *device)
133{
134 struct rds_iw_device *rds_iwdev;
135 struct rds_iw_cm_id *i_cm_id, *next;
136
137 rds_iwdev = ib_get_client_data(device, &rds_iw_client);
138 if (!rds_iwdev)
139 return;
140
141 spin_lock_irq(&rds_iwdev->spinlock);
142 list_for_each_entry_safe(i_cm_id, next, &rds_iwdev->cm_id_list, list) {
143 list_del(&i_cm_id->list);
144 kfree(i_cm_id);
145 }
146 spin_unlock_irq(&rds_iwdev->spinlock);
147
148 rds_iw_remove_conns(rds_iwdev);
149
150 if (rds_iwdev->mr_pool)
151 rds_iw_destroy_mr_pool(rds_iwdev->mr_pool);
152
153 if (rds_iwdev->mr)
154 ib_dereg_mr(rds_iwdev->mr);
155
156 while (ib_dealloc_pd(rds_iwdev->pd)) {
157 rdsdebug("Failed to dealloc pd %p\n", rds_iwdev->pd);
158 msleep(1);
159 }
160
161 list_del(&rds_iwdev->list);
162 kfree(rds_iwdev);
163}
164
165struct ib_client rds_iw_client = {
166 .name = "rds_iw",
167 .add = rds_iw_add_one,
168 .remove = rds_iw_remove_one
169};
170
171static int rds_iw_conn_info_visitor(struct rds_connection *conn,
172 void *buffer)
173{
174 struct rds_info_rdma_connection *iinfo = buffer;
175 struct rds_iw_connection *ic;
176
177 /* We will only ever look at IB transports */
178 if (conn->c_trans != &rds_iw_transport)
179 return 0;
180
181 iinfo->src_addr = conn->c_laddr;
182 iinfo->dst_addr = conn->c_faddr;
183
184 memset(&iinfo->src_gid, 0, sizeof(iinfo->src_gid));
185 memset(&iinfo->dst_gid, 0, sizeof(iinfo->dst_gid));
186 if (rds_conn_state(conn) == RDS_CONN_UP) {
187 struct rds_iw_device *rds_iwdev;
188 struct rdma_dev_addr *dev_addr;
189
190 ic = conn->c_transport_data;
191 dev_addr = &ic->i_cm_id->route.addr.dev_addr;
192
193 ib_addr_get_sgid(dev_addr, (union ib_gid *) &iinfo->src_gid);
194 ib_addr_get_dgid(dev_addr, (union ib_gid *) &iinfo->dst_gid);
195
196 rds_iwdev = ib_get_client_data(ic->i_cm_id->device, &rds_iw_client);
197 iinfo->max_send_wr = ic->i_send_ring.w_nr;
198 iinfo->max_recv_wr = ic->i_recv_ring.w_nr;
199 iinfo->max_send_sge = rds_iwdev->max_sge;
200 rds_iw_get_mr_info(rds_iwdev, iinfo);
201 }
202 return 1;
203}
204
205static void rds_iw_ic_info(struct socket *sock, unsigned int len,
206 struct rds_info_iterator *iter,
207 struct rds_info_lengths *lens)
208{
209 rds_for_each_conn_info(sock, len, iter, lens,
210 rds_iw_conn_info_visitor,
211 sizeof(struct rds_info_rdma_connection));
212}
213
214
215/*
216 * Early RDS/IB was built to only bind to an address if there is an IPoIB
217 * device with that address set.
218 *
219 * If it were me, I'd advocate for something more flexible. Sending and
220 * receiving should be device-agnostic. Transports would try and maintain
221 * connections between peers who have messages queued. Userspace would be
222 * allowed to influence which paths have priority. We could call userspace
223 * asserting this policy "routing".
224 */
225static int rds_iw_laddr_check(__be32 addr)
226{
227 int ret;
228 struct rdma_cm_id *cm_id;
229 struct sockaddr_in sin;
230
231 /* Create a CMA ID and try to bind it. This catches both
232 * IB and iWARP capable NICs.
233 */
234 cm_id = rdma_create_id(NULL, NULL, RDMA_PS_TCP);
235 if (!cm_id)
236 return -EADDRNOTAVAIL;
237
238 memset(&sin, 0, sizeof(sin));
239 sin.sin_family = AF_INET;
240 sin.sin_addr.s_addr = addr;
241
242 /* rdma_bind_addr will only succeed for IB & iWARP devices */
243 ret = rdma_bind_addr(cm_id, (struct sockaddr *)&sin);
244 /* due to this, we will claim to support IB devices unless we
245 check node_type. */
246 if (ret || cm_id->device->node_type != RDMA_NODE_RNIC)
247 ret = -EADDRNOTAVAIL;
248
249 rdsdebug("addr %pI4 ret %d node type %d\n",
250 &addr, ret,
251 cm_id->device ? cm_id->device->node_type : -1);
252
253 rdma_destroy_id(cm_id);
254
255 return ret;
256}
257
258void rds_iw_exit(void)
259{
260 rds_info_deregister_func(RDS_INFO_IWARP_CONNECTIONS, rds_iw_ic_info);
261 rds_iw_remove_nodev_conns();
262 ib_unregister_client(&rds_iw_client);
263 rds_iw_sysctl_exit();
264 rds_iw_recv_exit();
265 rds_trans_unregister(&rds_iw_transport);
266}
267
268struct rds_transport rds_iw_transport = {
269 .laddr_check = rds_iw_laddr_check,
270 .xmit_complete = rds_iw_xmit_complete,
271 .xmit = rds_iw_xmit,
272 .xmit_cong_map = NULL,
273 .xmit_rdma = rds_iw_xmit_rdma,
274 .recv = rds_iw_recv,
275 .conn_alloc = rds_iw_conn_alloc,
276 .conn_free = rds_iw_conn_free,
277 .conn_connect = rds_iw_conn_connect,
278 .conn_shutdown = rds_iw_conn_shutdown,
279 .inc_copy_to_user = rds_iw_inc_copy_to_user,
280 .inc_purge = rds_iw_inc_purge,
281 .inc_free = rds_iw_inc_free,
282 .cm_initiate_connect = rds_iw_cm_initiate_connect,
283 .cm_handle_connect = rds_iw_cm_handle_connect,
284 .cm_connect_complete = rds_iw_cm_connect_complete,
285 .stats_info_copy = rds_iw_stats_info_copy,
286 .exit = rds_iw_exit,
287 .get_mr = rds_iw_get_mr,
288 .sync_mr = rds_iw_sync_mr,
289 .free_mr = rds_iw_free_mr,
290 .flush_mrs = rds_iw_flush_mrs,
291 .t_owner = THIS_MODULE,
292 .t_name = "iwarp",
293 .t_prefer_loopback = 1,
294};
295
296int __init rds_iw_init(void)
297{
298 int ret;
299
300 INIT_LIST_HEAD(&rds_iw_devices);
301
302 ret = ib_register_client(&rds_iw_client);
303 if (ret)
304 goto out;
305
306 ret = rds_iw_sysctl_init();
307 if (ret)
308 goto out_ibreg;
309
310 ret = rds_iw_recv_init();
311 if (ret)
312 goto out_sysctl;
313
314 ret = rds_trans_register(&rds_iw_transport);
315 if (ret)
316 goto out_recv;
317
318 rds_info_register_func(RDS_INFO_IWARP_CONNECTIONS, rds_iw_ic_info);
319
320 goto out;
321
322out_recv:
323 rds_iw_recv_exit();
324out_sysctl:
325 rds_iw_sysctl_exit();
326out_ibreg:
327 ib_unregister_client(&rds_iw_client);
328out:
329 return ret;
330}
331
332MODULE_LICENSE("GPL");
333
diff --git a/net/rds/iw.h b/net/rds/iw.h
new file mode 100644
index 000000000000..0ddda34f2a1c
--- /dev/null
+++ b/net/rds/iw.h
@@ -0,0 +1,395 @@
1#ifndef _RDS_IW_H
2#define _RDS_IW_H
3
4#include <rdma/ib_verbs.h>
5#include <rdma/rdma_cm.h>
6#include "rds.h"
7#include "rdma_transport.h"
8
9#define RDS_FASTREG_SIZE 20
10#define RDS_FASTREG_POOL_SIZE 2048
11
12#define RDS_IW_MAX_SGE 8
13#define RDS_IW_RECV_SGE 2
14
15#define RDS_IW_DEFAULT_RECV_WR 1024
16#define RDS_IW_DEFAULT_SEND_WR 256
17
18#define RDS_IW_SUPPORTED_PROTOCOLS 0x00000003 /* minor versions supported */
19
20extern struct list_head rds_iw_devices;
21
22/*
23 * IB posts RDS_FRAG_SIZE fragments of pages to the receive queues to
24 * try and minimize the amount of memory tied up both the device and
25 * socket receive queues.
26 */
27/* page offset of the final full frag that fits in the page */
28#define RDS_PAGE_LAST_OFF (((PAGE_SIZE / RDS_FRAG_SIZE) - 1) * RDS_FRAG_SIZE)
29struct rds_page_frag {
30 struct list_head f_item;
31 struct page *f_page;
32 unsigned long f_offset;
33 dma_addr_t f_mapped;
34};
35
36struct rds_iw_incoming {
37 struct list_head ii_frags;
38 struct rds_incoming ii_inc;
39};
40
41struct rds_iw_connect_private {
42 /* Add new fields at the end, and don't permute existing fields. */
43 __be32 dp_saddr;
44 __be32 dp_daddr;
45 u8 dp_protocol_major;
46 u8 dp_protocol_minor;
47 __be16 dp_protocol_minor_mask; /* bitmask */
48 __be32 dp_reserved1;
49 __be64 dp_ack_seq;
50 __be32 dp_credit; /* non-zero enables flow ctl */
51};
52
53struct rds_iw_scatterlist {
54 struct scatterlist *list;
55 unsigned int len;
56 int dma_len;
57 unsigned int dma_npages;
58 unsigned int bytes;
59};
60
61struct rds_iw_mapping {
62 spinlock_t m_lock; /* protect the mapping struct */
63 struct list_head m_list;
64 struct rds_iw_mr *m_mr;
65 uint32_t m_rkey;
66 struct rds_iw_scatterlist m_sg;
67};
68
69struct rds_iw_send_work {
70 struct rds_message *s_rm;
71
72 /* We should really put these into a union: */
73 struct rds_rdma_op *s_op;
74 struct rds_iw_mapping *s_mapping;
75 struct ib_mr *s_mr;
76 struct ib_fast_reg_page_list *s_page_list;
77 unsigned char s_remap_count;
78
79 struct ib_send_wr s_wr;
80 struct ib_sge s_sge[RDS_IW_MAX_SGE];
81 unsigned long s_queued;
82};
83
84struct rds_iw_recv_work {
85 struct rds_iw_incoming *r_iwinc;
86 struct rds_page_frag *r_frag;
87 struct ib_recv_wr r_wr;
88 struct ib_sge r_sge[2];
89};
90
91struct rds_iw_work_ring {
92 u32 w_nr;
93 u32 w_alloc_ptr;
94 u32 w_alloc_ctr;
95 u32 w_free_ptr;
96 atomic_t w_free_ctr;
97};
98
99struct rds_iw_device;
100
101struct rds_iw_connection {
102
103 struct list_head iw_node;
104 struct rds_iw_device *rds_iwdev;
105 struct rds_connection *conn;
106
107 /* alphabet soup, IBTA style */
108 struct rdma_cm_id *i_cm_id;
109 struct ib_pd *i_pd;
110 struct ib_mr *i_mr;
111 struct ib_cq *i_send_cq;
112 struct ib_cq *i_recv_cq;
113
114 /* tx */
115 struct rds_iw_work_ring i_send_ring;
116 struct rds_message *i_rm;
117 struct rds_header *i_send_hdrs;
118 u64 i_send_hdrs_dma;
119 struct rds_iw_send_work *i_sends;
120
121 /* rx */
122 struct mutex i_recv_mutex;
123 struct rds_iw_work_ring i_recv_ring;
124 struct rds_iw_incoming *i_iwinc;
125 u32 i_recv_data_rem;
126 struct rds_header *i_recv_hdrs;
127 u64 i_recv_hdrs_dma;
128 struct rds_iw_recv_work *i_recvs;
129 struct rds_page_frag i_frag;
130 u64 i_ack_recv; /* last ACK received */
131
132 /* sending acks */
133 unsigned long i_ack_flags;
134 u64 i_ack_next; /* next ACK to send */
135 struct rds_header *i_ack;
136 struct ib_send_wr i_ack_wr;
137 struct ib_sge i_ack_sge;
138 u64 i_ack_dma;
139 unsigned long i_ack_queued;
140
141 /* Flow control related information
142 *
143 * Our algorithm uses a pair variables that we need to access
144 * atomically - one for the send credits, and one posted
145 * recv credits we need to transfer to remote.
146 * Rather than protect them using a slow spinlock, we put both into
147 * a single atomic_t and update it using cmpxchg
148 */
149 atomic_t i_credits;
150
151 /* Protocol version specific information */
152 unsigned int i_flowctl:1; /* enable/disable flow ctl */
153 unsigned int i_dma_local_lkey:1;
154 unsigned int i_fastreg_posted:1; /* fastreg posted on this connection */
155 /* Batched completions */
156 unsigned int i_unsignaled_wrs;
157 long i_unsignaled_bytes;
158};
159
160/* This assumes that atomic_t is at least 32 bits */
161#define IB_GET_SEND_CREDITS(v) ((v) & 0xffff)
162#define IB_GET_POST_CREDITS(v) ((v) >> 16)
163#define IB_SET_SEND_CREDITS(v) ((v) & 0xffff)
164#define IB_SET_POST_CREDITS(v) ((v) << 16)
165
166struct rds_iw_cm_id {
167 struct list_head list;
168 struct rdma_cm_id *cm_id;
169};
170
171struct rds_iw_device {
172 struct list_head list;
173 struct list_head cm_id_list;
174 struct list_head conn_list;
175 struct ib_device *dev;
176 struct ib_pd *pd;
177 struct ib_mr *mr;
178 struct rds_iw_mr_pool *mr_pool;
179 int page_shift;
180 int max_sge;
181 unsigned int max_wrs;
182 unsigned int dma_local_lkey:1;
183 spinlock_t spinlock; /* protect the above */
184};
185
186/* bits for i_ack_flags */
187#define IB_ACK_IN_FLIGHT 0
188#define IB_ACK_REQUESTED 1
189
190/* Magic WR_ID for ACKs */
191#define RDS_IW_ACK_WR_ID ((u64)0xffffffffffffffffULL)
192#define RDS_IW_FAST_REG_WR_ID ((u64)0xefefefefefefefefULL)
193#define RDS_IW_LOCAL_INV_WR_ID ((u64)0xdfdfdfdfdfdfdfdfULL)
194
195struct rds_iw_statistics {
196 uint64_t s_iw_connect_raced;
197 uint64_t s_iw_listen_closed_stale;
198 uint64_t s_iw_tx_cq_call;
199 uint64_t s_iw_tx_cq_event;
200 uint64_t s_iw_tx_ring_full;
201 uint64_t s_iw_tx_throttle;
202 uint64_t s_iw_tx_sg_mapping_failure;
203 uint64_t s_iw_tx_stalled;
204 uint64_t s_iw_tx_credit_updates;
205 uint64_t s_iw_rx_cq_call;
206 uint64_t s_iw_rx_cq_event;
207 uint64_t s_iw_rx_ring_empty;
208 uint64_t s_iw_rx_refill_from_cq;
209 uint64_t s_iw_rx_refill_from_thread;
210 uint64_t s_iw_rx_alloc_limit;
211 uint64_t s_iw_rx_credit_updates;
212 uint64_t s_iw_ack_sent;
213 uint64_t s_iw_ack_send_failure;
214 uint64_t s_iw_ack_send_delayed;
215 uint64_t s_iw_ack_send_piggybacked;
216 uint64_t s_iw_ack_received;
217 uint64_t s_iw_rdma_mr_alloc;
218 uint64_t s_iw_rdma_mr_free;
219 uint64_t s_iw_rdma_mr_used;
220 uint64_t s_iw_rdma_mr_pool_flush;
221 uint64_t s_iw_rdma_mr_pool_wait;
222 uint64_t s_iw_rdma_mr_pool_depleted;
223};
224
225extern struct workqueue_struct *rds_iw_wq;
226
227/*
228 * Fake ib_dma_sync_sg_for_{cpu,device} as long as ib_verbs.h
229 * doesn't define it.
230 */
231static inline void rds_iw_dma_sync_sg_for_cpu(struct ib_device *dev,
232 struct scatterlist *sg, unsigned int sg_dma_len, int direction)
233{
234 unsigned int i;
235
236 for (i = 0; i < sg_dma_len; ++i) {
237 ib_dma_sync_single_for_cpu(dev,
238 ib_sg_dma_address(dev, &sg[i]),
239 ib_sg_dma_len(dev, &sg[i]),
240 direction);
241 }
242}
243#define ib_dma_sync_sg_for_cpu rds_iw_dma_sync_sg_for_cpu
244
245static inline void rds_iw_dma_sync_sg_for_device(struct ib_device *dev,
246 struct scatterlist *sg, unsigned int sg_dma_len, int direction)
247{
248 unsigned int i;
249
250 for (i = 0; i < sg_dma_len; ++i) {
251 ib_dma_sync_single_for_device(dev,
252 ib_sg_dma_address(dev, &sg[i]),
253 ib_sg_dma_len(dev, &sg[i]),
254 direction);
255 }
256}
257#define ib_dma_sync_sg_for_device rds_iw_dma_sync_sg_for_device
258
259static inline u32 rds_iw_local_dma_lkey(struct rds_iw_connection *ic)
260{
261 return ic->i_dma_local_lkey ? ic->i_cm_id->device->local_dma_lkey : ic->i_mr->lkey;
262}
263
264/* ib.c */
265extern struct rds_transport rds_iw_transport;
266extern void rds_iw_add_one(struct ib_device *device);
267extern void rds_iw_remove_one(struct ib_device *device);
268extern struct ib_client rds_iw_client;
269
270extern unsigned int fastreg_pool_size;
271extern unsigned int fastreg_message_size;
272
273extern spinlock_t iw_nodev_conns_lock;
274extern struct list_head iw_nodev_conns;
275
276/* ib_cm.c */
277int rds_iw_conn_alloc(struct rds_connection *conn, gfp_t gfp);
278void rds_iw_conn_free(void *arg);
279int rds_iw_conn_connect(struct rds_connection *conn);
280void rds_iw_conn_shutdown(struct rds_connection *conn);
281void rds_iw_state_change(struct sock *sk);
282int __init rds_iw_listen_init(void);
283void rds_iw_listen_stop(void);
284void __rds_iw_conn_error(struct rds_connection *conn, const char *, ...);
285int rds_iw_cm_handle_connect(struct rdma_cm_id *cm_id,
286 struct rdma_cm_event *event);
287int rds_iw_cm_initiate_connect(struct rdma_cm_id *cm_id);
288void rds_iw_cm_connect_complete(struct rds_connection *conn,
289 struct rdma_cm_event *event);
290
291
292#define rds_iw_conn_error(conn, fmt...) \
293 __rds_iw_conn_error(conn, KERN_WARNING "RDS/IW: " fmt)
294
295/* ib_rdma.c */
296int rds_iw_update_cm_id(struct rds_iw_device *rds_iwdev, struct rdma_cm_id *cm_id);
297int rds_iw_add_conn(struct rds_iw_device *rds_iwdev, struct rds_connection *conn);
298void rds_iw_remove_nodev_conns(void);
299void rds_iw_remove_conns(struct rds_iw_device *rds_iwdev);
300struct rds_iw_mr_pool *rds_iw_create_mr_pool(struct rds_iw_device *);
301void rds_iw_get_mr_info(struct rds_iw_device *rds_iwdev, struct rds_info_rdma_connection *iinfo);
302void rds_iw_destroy_mr_pool(struct rds_iw_mr_pool *);
303void *rds_iw_get_mr(struct scatterlist *sg, unsigned long nents,
304 struct rds_sock *rs, u32 *key_ret);
305void rds_iw_sync_mr(void *trans_private, int dir);
306void rds_iw_free_mr(void *trans_private, int invalidate);
307void rds_iw_flush_mrs(void);
308void rds_iw_remove_cm_id(struct rds_iw_device *rds_iwdev, struct rdma_cm_id *cm_id);
309
310/* ib_recv.c */
311int __init rds_iw_recv_init(void);
312void rds_iw_recv_exit(void);
313int rds_iw_recv(struct rds_connection *conn);
314int rds_iw_recv_refill(struct rds_connection *conn, gfp_t kptr_gfp,
315 gfp_t page_gfp, int prefill);
316void rds_iw_inc_purge(struct rds_incoming *inc);
317void rds_iw_inc_free(struct rds_incoming *inc);
318int rds_iw_inc_copy_to_user(struct rds_incoming *inc, struct iovec *iov,
319 size_t size);
320void rds_iw_recv_cq_comp_handler(struct ib_cq *cq, void *context);
321void rds_iw_recv_init_ring(struct rds_iw_connection *ic);
322void rds_iw_recv_clear_ring(struct rds_iw_connection *ic);
323void rds_iw_recv_init_ack(struct rds_iw_connection *ic);
324void rds_iw_attempt_ack(struct rds_iw_connection *ic);
325void rds_iw_ack_send_complete(struct rds_iw_connection *ic);
326u64 rds_iw_piggyb_ack(struct rds_iw_connection *ic);
327
328/* ib_ring.c */
329void rds_iw_ring_init(struct rds_iw_work_ring *ring, u32 nr);
330void rds_iw_ring_resize(struct rds_iw_work_ring *ring, u32 nr);
331u32 rds_iw_ring_alloc(struct rds_iw_work_ring *ring, u32 val, u32 *pos);
332void rds_iw_ring_free(struct rds_iw_work_ring *ring, u32 val);
333void rds_iw_ring_unalloc(struct rds_iw_work_ring *ring, u32 val);
334int rds_iw_ring_empty(struct rds_iw_work_ring *ring);
335int rds_iw_ring_low(struct rds_iw_work_ring *ring);
336u32 rds_iw_ring_oldest(struct rds_iw_work_ring *ring);
337u32 rds_iw_ring_completed(struct rds_iw_work_ring *ring, u32 wr_id, u32 oldest);
338extern wait_queue_head_t rds_iw_ring_empty_wait;
339
340/* ib_send.c */
341void rds_iw_xmit_complete(struct rds_connection *conn);
342int rds_iw_xmit(struct rds_connection *conn, struct rds_message *rm,
343 unsigned int hdr_off, unsigned int sg, unsigned int off);
344void rds_iw_send_cq_comp_handler(struct ib_cq *cq, void *context);
345void rds_iw_send_init_ring(struct rds_iw_connection *ic);
346void rds_iw_send_clear_ring(struct rds_iw_connection *ic);
347int rds_iw_xmit_rdma(struct rds_connection *conn, struct rds_rdma_op *op);
348void rds_iw_send_add_credits(struct rds_connection *conn, unsigned int credits);
349void rds_iw_advertise_credits(struct rds_connection *conn, unsigned int posted);
350int rds_iw_send_grab_credits(struct rds_iw_connection *ic, u32 wanted,
351 u32 *adv_credits, int need_posted);
352
353/* ib_stats.c */
354DECLARE_PER_CPU(struct rds_iw_statistics, rds_iw_stats);
355#define rds_iw_stats_inc(member) rds_stats_inc_which(rds_iw_stats, member)
356unsigned int rds_iw_stats_info_copy(struct rds_info_iterator *iter,
357 unsigned int avail);
358
359/* ib_sysctl.c */
360int __init rds_iw_sysctl_init(void);
361void rds_iw_sysctl_exit(void);
362extern unsigned long rds_iw_sysctl_max_send_wr;
363extern unsigned long rds_iw_sysctl_max_recv_wr;
364extern unsigned long rds_iw_sysctl_max_unsig_wrs;
365extern unsigned long rds_iw_sysctl_max_unsig_bytes;
366extern unsigned long rds_iw_sysctl_max_recv_allocation;
367extern unsigned int rds_iw_sysctl_flow_control;
368extern ctl_table rds_iw_sysctl_table[];
369
370/*
371 * Helper functions for getting/setting the header and data SGEs in
372 * RDS packets (not RDMA)
373 */
374static inline struct ib_sge *
375rds_iw_header_sge(struct rds_iw_connection *ic, struct ib_sge *sge)
376{
377 return &sge[0];
378}
379
380static inline struct ib_sge *
381rds_iw_data_sge(struct rds_iw_connection *ic, struct ib_sge *sge)
382{
383 return &sge[1];
384}
385
386static inline void rds_iw_set_64bit(u64 *ptr, u64 val)
387{
388#if BITS_PER_LONG == 64
389 *ptr = val;
390#else
391 set_64bit(ptr, val);
392#endif
393}
394
395#endif
diff --git a/net/rds/iw_cm.c b/net/rds/iw_cm.c
new file mode 100644
index 000000000000..57ecb3d4b8a5
--- /dev/null
+++ b/net/rds/iw_cm.c
@@ -0,0 +1,750 @@
1/*
2 * Copyright (c) 2006 Oracle. All rights reserved.
3 *
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
9 *
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
12 * conditions are met:
13 *
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
16 * disclaimer.
17 *
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
22 *
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30 * SOFTWARE.
31 *
32 */
33#include <linux/kernel.h>
34#include <linux/in.h>
35#include <linux/vmalloc.h>
36
37#include "rds.h"
38#include "iw.h"
39
40/*
41 * Set the selected protocol version
42 */
43static void rds_iw_set_protocol(struct rds_connection *conn, unsigned int version)
44{
45 conn->c_version = version;
46}
47
48/*
49 * Set up flow control
50 */
51static void rds_iw_set_flow_control(struct rds_connection *conn, u32 credits)
52{
53 struct rds_iw_connection *ic = conn->c_transport_data;
54
55 if (rds_iw_sysctl_flow_control && credits != 0) {
56 /* We're doing flow control */
57 ic->i_flowctl = 1;
58 rds_iw_send_add_credits(conn, credits);
59 } else {
60 ic->i_flowctl = 0;
61 }
62}
63
64/*
65 * Connection established.
66 * We get here for both outgoing and incoming connection.
67 */
68void rds_iw_cm_connect_complete(struct rds_connection *conn, struct rdma_cm_event *event)
69{
70 const struct rds_iw_connect_private *dp = NULL;
71 struct rds_iw_connection *ic = conn->c_transport_data;
72 struct rds_iw_device *rds_iwdev;
73 int err;
74
75 if (event->param.conn.private_data_len) {
76 dp = event->param.conn.private_data;
77
78 rds_iw_set_protocol(conn,
79 RDS_PROTOCOL(dp->dp_protocol_major,
80 dp->dp_protocol_minor));
81 rds_iw_set_flow_control(conn, be32_to_cpu(dp->dp_credit));
82 }
83
84 /* update ib_device with this local ipaddr & conn */
85 rds_iwdev = ib_get_client_data(ic->i_cm_id->device, &rds_iw_client);
86 err = rds_iw_update_cm_id(rds_iwdev, ic->i_cm_id);
87 if (err)
88 printk(KERN_ERR "rds_iw_update_ipaddr failed (%d)\n", err);
89 err = rds_iw_add_conn(rds_iwdev, conn);
90 if (err)
91 printk(KERN_ERR "rds_iw_add_conn failed (%d)\n", err);
92
93 /* If the peer gave us the last packet it saw, process this as if
94 * we had received a regular ACK. */
95 if (dp && dp->dp_ack_seq)
96 rds_send_drop_acked(conn, be64_to_cpu(dp->dp_ack_seq), NULL);
97
98 printk(KERN_NOTICE "RDS/IW: connected to %pI4<->%pI4 version %u.%u%s\n",
99 &conn->c_laddr, &conn->c_faddr,
100 RDS_PROTOCOL_MAJOR(conn->c_version),
101 RDS_PROTOCOL_MINOR(conn->c_version),
102 ic->i_flowctl ? ", flow control" : "");
103
104 rds_connect_complete(conn);
105}
106
107static void rds_iw_cm_fill_conn_param(struct rds_connection *conn,
108 struct rdma_conn_param *conn_param,
109 struct rds_iw_connect_private *dp,
110 u32 protocol_version)
111{
112 struct rds_iw_connection *ic = conn->c_transport_data;
113
114 memset(conn_param, 0, sizeof(struct rdma_conn_param));
115 /* XXX tune these? */
116 conn_param->responder_resources = 1;
117 conn_param->initiator_depth = 1;
118
119 if (dp) {
120 memset(dp, 0, sizeof(*dp));
121 dp->dp_saddr = conn->c_laddr;
122 dp->dp_daddr = conn->c_faddr;
123 dp->dp_protocol_major = RDS_PROTOCOL_MAJOR(protocol_version);
124 dp->dp_protocol_minor = RDS_PROTOCOL_MINOR(protocol_version);
125 dp->dp_protocol_minor_mask = cpu_to_be16(RDS_IW_SUPPORTED_PROTOCOLS);
126 dp->dp_ack_seq = rds_iw_piggyb_ack(ic);
127
128 /* Advertise flow control */
129 if (ic->i_flowctl) {
130 unsigned int credits;
131
132 credits = IB_GET_POST_CREDITS(atomic_read(&ic->i_credits));
133 dp->dp_credit = cpu_to_be32(credits);
134 atomic_sub(IB_SET_POST_CREDITS(credits), &ic->i_credits);
135 }
136
137 conn_param->private_data = dp;
138 conn_param->private_data_len = sizeof(*dp);
139 }
140}
141
142static void rds_iw_cq_event_handler(struct ib_event *event, void *data)
143{
144 rdsdebug("event %u data %p\n", event->event, data);
145}
146
147static void rds_iw_qp_event_handler(struct ib_event *event, void *data)
148{
149 struct rds_connection *conn = data;
150 struct rds_iw_connection *ic = conn->c_transport_data;
151
152 rdsdebug("conn %p ic %p event %u\n", conn, ic, event->event);
153
154 switch (event->event) {
155 case IB_EVENT_COMM_EST:
156 rdma_notify(ic->i_cm_id, IB_EVENT_COMM_EST);
157 break;
158 case IB_EVENT_QP_REQ_ERR:
159 case IB_EVENT_QP_FATAL:
160 default:
161 rds_iw_conn_error(conn, "RDS/IW: Fatal QP Event %u - connection %pI4->%pI4...reconnecting\n",
162 event->event, &conn->c_laddr,
163 &conn->c_faddr);
164 break;
165 }
166}
167
168/*
169 * Create a QP
170 */
171static int rds_iw_init_qp_attrs(struct ib_qp_init_attr *attr,
172 struct rds_iw_device *rds_iwdev,
173 struct rds_iw_work_ring *send_ring,
174 void (*send_cq_handler)(struct ib_cq *, void *),
175 struct rds_iw_work_ring *recv_ring,
176 void (*recv_cq_handler)(struct ib_cq *, void *),
177 void *context)
178{
179 struct ib_device *dev = rds_iwdev->dev;
180 unsigned int send_size, recv_size;
181 int ret;
182
183 /* The offset of 1 is to accomodate the additional ACK WR. */
184 send_size = min_t(unsigned int, rds_iwdev->max_wrs, rds_iw_sysctl_max_send_wr + 1);
185 recv_size = min_t(unsigned int, rds_iwdev->max_wrs, rds_iw_sysctl_max_recv_wr + 1);
186 rds_iw_ring_resize(send_ring, send_size - 1);
187 rds_iw_ring_resize(recv_ring, recv_size - 1);
188
189 memset(attr, 0, sizeof(*attr));
190 attr->event_handler = rds_iw_qp_event_handler;
191 attr->qp_context = context;
192 attr->cap.max_send_wr = send_size;
193 attr->cap.max_recv_wr = recv_size;
194 attr->cap.max_send_sge = rds_iwdev->max_sge;
195 attr->cap.max_recv_sge = RDS_IW_RECV_SGE;
196 attr->sq_sig_type = IB_SIGNAL_REQ_WR;
197 attr->qp_type = IB_QPT_RC;
198
199 attr->send_cq = ib_create_cq(dev, send_cq_handler,
200 rds_iw_cq_event_handler,
201 context, send_size, 0);
202 if (IS_ERR(attr->send_cq)) {
203 ret = PTR_ERR(attr->send_cq);
204 attr->send_cq = NULL;
205 rdsdebug("ib_create_cq send failed: %d\n", ret);
206 goto out;
207 }
208
209 attr->recv_cq = ib_create_cq(dev, recv_cq_handler,
210 rds_iw_cq_event_handler,
211 context, recv_size, 0);
212 if (IS_ERR(attr->recv_cq)) {
213 ret = PTR_ERR(attr->recv_cq);
214 attr->recv_cq = NULL;
215 rdsdebug("ib_create_cq send failed: %d\n", ret);
216 goto out;
217 }
218
219 ret = ib_req_notify_cq(attr->send_cq, IB_CQ_NEXT_COMP);
220 if (ret) {
221 rdsdebug("ib_req_notify_cq send failed: %d\n", ret);
222 goto out;
223 }
224
225 ret = ib_req_notify_cq(attr->recv_cq, IB_CQ_SOLICITED);
226 if (ret) {
227 rdsdebug("ib_req_notify_cq recv failed: %d\n", ret);
228 goto out;
229 }
230
231out:
232 if (ret) {
233 if (attr->send_cq)
234 ib_destroy_cq(attr->send_cq);
235 if (attr->recv_cq)
236 ib_destroy_cq(attr->recv_cq);
237 }
238 return ret;
239}
240
241/*
242 * This needs to be very careful to not leave IS_ERR pointers around for
243 * cleanup to trip over.
244 */
245static int rds_iw_setup_qp(struct rds_connection *conn)
246{
247 struct rds_iw_connection *ic = conn->c_transport_data;
248 struct ib_device *dev = ic->i_cm_id->device;
249 struct ib_qp_init_attr attr;
250 struct rds_iw_device *rds_iwdev;
251 int ret;
252
253 /* rds_iw_add_one creates a rds_iw_device object per IB device,
254 * and allocates a protection domain, memory range and MR pool
255 * for each. If that fails for any reason, it will not register
256 * the rds_iwdev at all.
257 */
258 rds_iwdev = ib_get_client_data(dev, &rds_iw_client);
259 if (rds_iwdev == NULL) {
260 if (printk_ratelimit())
261 printk(KERN_NOTICE "RDS/IW: No client_data for device %s\n",
262 dev->name);
263 return -EOPNOTSUPP;
264 }
265
266 /* Protection domain and memory range */
267 ic->i_pd = rds_iwdev->pd;
268 ic->i_mr = rds_iwdev->mr;
269
270 ret = rds_iw_init_qp_attrs(&attr, rds_iwdev,
271 &ic->i_send_ring, rds_iw_send_cq_comp_handler,
272 &ic->i_recv_ring, rds_iw_recv_cq_comp_handler,
273 conn);
274 if (ret < 0)
275 goto out;
276
277 ic->i_send_cq = attr.send_cq;
278 ic->i_recv_cq = attr.recv_cq;
279
280 /*
281 * XXX this can fail if max_*_wr is too large? Are we supposed
282 * to back off until we get a value that the hardware can support?
283 */
284 ret = rdma_create_qp(ic->i_cm_id, ic->i_pd, &attr);
285 if (ret) {
286 rdsdebug("rdma_create_qp failed: %d\n", ret);
287 goto out;
288 }
289
290 ic->i_send_hdrs = ib_dma_alloc_coherent(dev,
291 ic->i_send_ring.w_nr *
292 sizeof(struct rds_header),
293 &ic->i_send_hdrs_dma, GFP_KERNEL);
294 if (ic->i_send_hdrs == NULL) {
295 ret = -ENOMEM;
296 rdsdebug("ib_dma_alloc_coherent send failed\n");
297 goto out;
298 }
299
300 ic->i_recv_hdrs = ib_dma_alloc_coherent(dev,
301 ic->i_recv_ring.w_nr *
302 sizeof(struct rds_header),
303 &ic->i_recv_hdrs_dma, GFP_KERNEL);
304 if (ic->i_recv_hdrs == NULL) {
305 ret = -ENOMEM;
306 rdsdebug("ib_dma_alloc_coherent recv failed\n");
307 goto out;
308 }
309
310 ic->i_ack = ib_dma_alloc_coherent(dev, sizeof(struct rds_header),
311 &ic->i_ack_dma, GFP_KERNEL);
312 if (ic->i_ack == NULL) {
313 ret = -ENOMEM;
314 rdsdebug("ib_dma_alloc_coherent ack failed\n");
315 goto out;
316 }
317
318 ic->i_sends = vmalloc(ic->i_send_ring.w_nr * sizeof(struct rds_iw_send_work));
319 if (ic->i_sends == NULL) {
320 ret = -ENOMEM;
321 rdsdebug("send allocation failed\n");
322 goto out;
323 }
324 rds_iw_send_init_ring(ic);
325
326 ic->i_recvs = vmalloc(ic->i_recv_ring.w_nr * sizeof(struct rds_iw_recv_work));
327 if (ic->i_recvs == NULL) {
328 ret = -ENOMEM;
329 rdsdebug("recv allocation failed\n");
330 goto out;
331 }
332
333 rds_iw_recv_init_ring(ic);
334 rds_iw_recv_init_ack(ic);
335
336 /* Post receive buffers - as a side effect, this will update
337 * the posted credit count. */
338 rds_iw_recv_refill(conn, GFP_KERNEL, GFP_HIGHUSER, 1);
339
340 rdsdebug("conn %p pd %p mr %p cq %p %p\n", conn, ic->i_pd, ic->i_mr,
341 ic->i_send_cq, ic->i_recv_cq);
342
343out:
344 return ret;
345}
346
347static u32 rds_iw_protocol_compatible(const struct rds_iw_connect_private *dp)
348{
349 u16 common;
350 u32 version = 0;
351
352 /* rdma_cm private data is odd - when there is any private data in the
353 * request, we will be given a pretty large buffer without telling us the
354 * original size. The only way to tell the difference is by looking at
355 * the contents, which are initialized to zero.
356 * If the protocol version fields aren't set, this is a connection attempt
357 * from an older version. This could could be 3.0 or 2.0 - we can't tell.
358 * We really should have changed this for OFED 1.3 :-( */
359 if (dp->dp_protocol_major == 0)
360 return RDS_PROTOCOL_3_0;
361
362 common = be16_to_cpu(dp->dp_protocol_minor_mask) & RDS_IW_SUPPORTED_PROTOCOLS;
363 if (dp->dp_protocol_major == 3 && common) {
364 version = RDS_PROTOCOL_3_0;
365 while ((common >>= 1) != 0)
366 version++;
367 } else if (printk_ratelimit()) {
368 printk(KERN_NOTICE "RDS: Connection from %pI4 using "
369 "incompatible protocol version %u.%u\n",
370 &dp->dp_saddr,
371 dp->dp_protocol_major,
372 dp->dp_protocol_minor);
373 }
374 return version;
375}
376
377int rds_iw_cm_handle_connect(struct rdma_cm_id *cm_id,
378 struct rdma_cm_event *event)
379{
380 const struct rds_iw_connect_private *dp = event->param.conn.private_data;
381 struct rds_iw_connect_private dp_rep;
382 struct rds_connection *conn = NULL;
383 struct rds_iw_connection *ic = NULL;
384 struct rdma_conn_param conn_param;
385 struct rds_iw_device *rds_iwdev;
386 u32 version;
387 int err, destroy = 1;
388
389 /* Check whether the remote protocol version matches ours. */
390 version = rds_iw_protocol_compatible(dp);
391 if (!version)
392 goto out;
393
394 rdsdebug("saddr %pI4 daddr %pI4 RDSv%u.%u\n",
395 &dp->dp_saddr, &dp->dp_daddr,
396 RDS_PROTOCOL_MAJOR(version), RDS_PROTOCOL_MINOR(version));
397
398 conn = rds_conn_create(dp->dp_daddr, dp->dp_saddr, &rds_iw_transport,
399 GFP_KERNEL);
400 if (IS_ERR(conn)) {
401 rdsdebug("rds_conn_create failed (%ld)\n", PTR_ERR(conn));
402 conn = NULL;
403 goto out;
404 }
405
406 /*
407 * The connection request may occur while the
408 * previous connection exist, e.g. in case of failover.
409 * But as connections may be initiated simultaneously
410 * by both hosts, we have a random backoff mechanism -
411 * see the comment above rds_queue_reconnect()
412 */
413 mutex_lock(&conn->c_cm_lock);
414 if (!rds_conn_transition(conn, RDS_CONN_DOWN, RDS_CONN_CONNECTING)) {
415 if (rds_conn_state(conn) == RDS_CONN_UP) {
416 rdsdebug("incoming connect while connecting\n");
417 rds_conn_drop(conn);
418 rds_iw_stats_inc(s_iw_listen_closed_stale);
419 } else
420 if (rds_conn_state(conn) == RDS_CONN_CONNECTING) {
421 /* Wait and see - our connect may still be succeeding */
422 rds_iw_stats_inc(s_iw_connect_raced);
423 }
424 mutex_unlock(&conn->c_cm_lock);
425 goto out;
426 }
427
428 ic = conn->c_transport_data;
429
430 rds_iw_set_protocol(conn, version);
431 rds_iw_set_flow_control(conn, be32_to_cpu(dp->dp_credit));
432
433 /* If the peer gave us the last packet it saw, process this as if
434 * we had received a regular ACK. */
435 if (dp->dp_ack_seq)
436 rds_send_drop_acked(conn, be64_to_cpu(dp->dp_ack_seq), NULL);
437
438 BUG_ON(cm_id->context);
439 BUG_ON(ic->i_cm_id);
440
441 ic->i_cm_id = cm_id;
442 cm_id->context = conn;
443
444 rds_iwdev = ib_get_client_data(cm_id->device, &rds_iw_client);
445 ic->i_dma_local_lkey = rds_iwdev->dma_local_lkey;
446
447 /* We got halfway through setting up the ib_connection, if we
448 * fail now, we have to take the long route out of this mess. */
449 destroy = 0;
450
451 err = rds_iw_setup_qp(conn);
452 if (err) {
453 rds_iw_conn_error(conn, "rds_iw_setup_qp failed (%d)\n", err);
454 goto out;
455 }
456
457 rds_iw_cm_fill_conn_param(conn, &conn_param, &dp_rep, version);
458
459 /* rdma_accept() calls rdma_reject() internally if it fails */
460 err = rdma_accept(cm_id, &conn_param);
461 mutex_unlock(&conn->c_cm_lock);
462 if (err) {
463 rds_iw_conn_error(conn, "rdma_accept failed (%d)\n", err);
464 goto out;
465 }
466
467 return 0;
468
469out:
470 rdma_reject(cm_id, NULL, 0);
471 return destroy;
472}
473
474
475int rds_iw_cm_initiate_connect(struct rdma_cm_id *cm_id)
476{
477 struct rds_connection *conn = cm_id->context;
478 struct rds_iw_connection *ic = conn->c_transport_data;
479 struct rdma_conn_param conn_param;
480 struct rds_iw_connect_private dp;
481 int ret;
482
483 /* If the peer doesn't do protocol negotiation, we must
484 * default to RDSv3.0 */
485 rds_iw_set_protocol(conn, RDS_PROTOCOL_3_0);
486 ic->i_flowctl = rds_iw_sysctl_flow_control; /* advertise flow control */
487
488 ret = rds_iw_setup_qp(conn);
489 if (ret) {
490 rds_iw_conn_error(conn, "rds_iw_setup_qp failed (%d)\n", ret);
491 goto out;
492 }
493
494 rds_iw_cm_fill_conn_param(conn, &conn_param, &dp, RDS_PROTOCOL_VERSION);
495
496 ret = rdma_connect(cm_id, &conn_param);
497 if (ret)
498 rds_iw_conn_error(conn, "rdma_connect failed (%d)\n", ret);
499
500out:
501 /* Beware - returning non-zero tells the rdma_cm to destroy
502 * the cm_id. We should certainly not do it as long as we still
503 * "own" the cm_id. */
504 if (ret) {
505 struct rds_iw_connection *ic = conn->c_transport_data;
506
507 if (ic->i_cm_id == cm_id)
508 ret = 0;
509 }
510 return ret;
511}
512
513int rds_iw_conn_connect(struct rds_connection *conn)
514{
515 struct rds_iw_connection *ic = conn->c_transport_data;
516 struct rds_iw_device *rds_iwdev;
517 struct sockaddr_in src, dest;
518 int ret;
519
520 /* XXX I wonder what affect the port space has */
521 /* delegate cm event handler to rdma_transport */
522 ic->i_cm_id = rdma_create_id(rds_rdma_cm_event_handler, conn,
523 RDMA_PS_TCP);
524 if (IS_ERR(ic->i_cm_id)) {
525 ret = PTR_ERR(ic->i_cm_id);
526 ic->i_cm_id = NULL;
527 rdsdebug("rdma_create_id() failed: %d\n", ret);
528 goto out;
529 }
530
531 rdsdebug("created cm id %p for conn %p\n", ic->i_cm_id, conn);
532
533 src.sin_family = AF_INET;
534 src.sin_addr.s_addr = (__force u32)conn->c_laddr;
535 src.sin_port = (__force u16)htons(0);
536
537 /* First, bind to the local address and device. */
538 ret = rdma_bind_addr(ic->i_cm_id, (struct sockaddr *) &src);
539 if (ret) {
540 rdsdebug("rdma_bind_addr(%pI4) failed: %d\n",
541 &conn->c_laddr, ret);
542 rdma_destroy_id(ic->i_cm_id);
543 ic->i_cm_id = NULL;
544 goto out;
545 }
546
547 rds_iwdev = ib_get_client_data(ic->i_cm_id->device, &rds_iw_client);
548 ic->i_dma_local_lkey = rds_iwdev->dma_local_lkey;
549
550 dest.sin_family = AF_INET;
551 dest.sin_addr.s_addr = (__force u32)conn->c_faddr;
552 dest.sin_port = (__force u16)htons(RDS_PORT);
553
554 ret = rdma_resolve_addr(ic->i_cm_id, (struct sockaddr *)&src,
555 (struct sockaddr *)&dest,
556 RDS_RDMA_RESOLVE_TIMEOUT_MS);
557 if (ret) {
558 rdsdebug("addr resolve failed for cm id %p: %d\n", ic->i_cm_id,
559 ret);
560 rdma_destroy_id(ic->i_cm_id);
561 ic->i_cm_id = NULL;
562 }
563
564out:
565 return ret;
566}
567
568/*
569 * This is so careful about only cleaning up resources that were built up
570 * so that it can be called at any point during startup. In fact it
571 * can be called multiple times for a given connection.
572 */
573void rds_iw_conn_shutdown(struct rds_connection *conn)
574{
575 struct rds_iw_connection *ic = conn->c_transport_data;
576 int err = 0;
577 struct ib_qp_attr qp_attr;
578
579 rdsdebug("cm %p pd %p cq %p %p qp %p\n", ic->i_cm_id,
580 ic->i_pd, ic->i_send_cq, ic->i_recv_cq,
581 ic->i_cm_id ? ic->i_cm_id->qp : NULL);
582
583 if (ic->i_cm_id) {
584 struct ib_device *dev = ic->i_cm_id->device;
585
586 rdsdebug("disconnecting cm %p\n", ic->i_cm_id);
587 err = rdma_disconnect(ic->i_cm_id);
588 if (err) {
589 /* Actually this may happen quite frequently, when
590 * an outgoing connect raced with an incoming connect.
591 */
592 rdsdebug("rds_iw_conn_shutdown: failed to disconnect,"
593 " cm: %p err %d\n", ic->i_cm_id, err);
594 }
595
596 if (ic->i_cm_id->qp) {
597 qp_attr.qp_state = IB_QPS_ERR;
598 ib_modify_qp(ic->i_cm_id->qp, &qp_attr, IB_QP_STATE);
599 }
600
601 wait_event(rds_iw_ring_empty_wait,
602 rds_iw_ring_empty(&ic->i_send_ring) &&
603 rds_iw_ring_empty(&ic->i_recv_ring));
604
605 if (ic->i_send_hdrs)
606 ib_dma_free_coherent(dev,
607 ic->i_send_ring.w_nr *
608 sizeof(struct rds_header),
609 ic->i_send_hdrs,
610 ic->i_send_hdrs_dma);
611
612 if (ic->i_recv_hdrs)
613 ib_dma_free_coherent(dev,
614 ic->i_recv_ring.w_nr *
615 sizeof(struct rds_header),
616 ic->i_recv_hdrs,
617 ic->i_recv_hdrs_dma);
618
619 if (ic->i_ack)
620 ib_dma_free_coherent(dev, sizeof(struct rds_header),
621 ic->i_ack, ic->i_ack_dma);
622
623 if (ic->i_sends)
624 rds_iw_send_clear_ring(ic);
625 if (ic->i_recvs)
626 rds_iw_recv_clear_ring(ic);
627
628 if (ic->i_cm_id->qp)
629 rdma_destroy_qp(ic->i_cm_id);
630 if (ic->i_send_cq)
631 ib_destroy_cq(ic->i_send_cq);
632 if (ic->i_recv_cq)
633 ib_destroy_cq(ic->i_recv_cq);
634
635 /*
636 * If associated with an rds_iw_device:
637 * Move connection back to the nodev list.
638 * Remove cm_id from the device cm_id list.
639 */
640 if (ic->rds_iwdev) {
641
642 spin_lock_irq(&ic->rds_iwdev->spinlock);
643 BUG_ON(list_empty(&ic->iw_node));
644 list_del(&ic->iw_node);
645 spin_unlock_irq(&ic->rds_iwdev->spinlock);
646
647 spin_lock_irq(&iw_nodev_conns_lock);
648 list_add_tail(&ic->iw_node, &iw_nodev_conns);
649 spin_unlock_irq(&iw_nodev_conns_lock);
650 rds_iw_remove_cm_id(ic->rds_iwdev, ic->i_cm_id);
651 ic->rds_iwdev = NULL;
652 }
653
654 rdma_destroy_id(ic->i_cm_id);
655
656 ic->i_cm_id = NULL;
657 ic->i_pd = NULL;
658 ic->i_mr = NULL;
659 ic->i_send_cq = NULL;
660 ic->i_recv_cq = NULL;
661 ic->i_send_hdrs = NULL;
662 ic->i_recv_hdrs = NULL;
663 ic->i_ack = NULL;
664 }
665 BUG_ON(ic->rds_iwdev);
666
667 /* Clear pending transmit */
668 if (ic->i_rm) {
669 rds_message_put(ic->i_rm);
670 ic->i_rm = NULL;
671 }
672
673 /* Clear the ACK state */
674 clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
675 rds_iw_set_64bit(&ic->i_ack_next, 0);
676 ic->i_ack_recv = 0;
677
678 /* Clear flow control state */
679 ic->i_flowctl = 0;
680 atomic_set(&ic->i_credits, 0);
681
682 rds_iw_ring_init(&ic->i_send_ring, rds_iw_sysctl_max_send_wr);
683 rds_iw_ring_init(&ic->i_recv_ring, rds_iw_sysctl_max_recv_wr);
684
685 if (ic->i_iwinc) {
686 rds_inc_put(&ic->i_iwinc->ii_inc);
687 ic->i_iwinc = NULL;
688 }
689
690 vfree(ic->i_sends);
691 ic->i_sends = NULL;
692 vfree(ic->i_recvs);
693 ic->i_recvs = NULL;
694 rdsdebug("shutdown complete\n");
695}
696
697int rds_iw_conn_alloc(struct rds_connection *conn, gfp_t gfp)
698{
699 struct rds_iw_connection *ic;
700 unsigned long flags;
701
702 /* XXX too lazy? */
703 ic = kzalloc(sizeof(struct rds_iw_connection), GFP_KERNEL);
704 if (ic == NULL)
705 return -ENOMEM;
706
707 INIT_LIST_HEAD(&ic->iw_node);
708 mutex_init(&ic->i_recv_mutex);
709
710 /*
711 * rds_iw_conn_shutdown() waits for these to be emptied so they
712 * must be initialized before it can be called.
713 */
714 rds_iw_ring_init(&ic->i_send_ring, rds_iw_sysctl_max_send_wr);
715 rds_iw_ring_init(&ic->i_recv_ring, rds_iw_sysctl_max_recv_wr);
716
717 ic->conn = conn;
718 conn->c_transport_data = ic;
719
720 spin_lock_irqsave(&iw_nodev_conns_lock, flags);
721 list_add_tail(&ic->iw_node, &iw_nodev_conns);
722 spin_unlock_irqrestore(&iw_nodev_conns_lock, flags);
723
724
725 rdsdebug("conn %p conn ic %p\n", conn, conn->c_transport_data);
726 return 0;
727}
728
729void rds_iw_conn_free(void *arg)
730{
731 struct rds_iw_connection *ic = arg;
732 rdsdebug("ic %p\n", ic);
733 list_del(&ic->iw_node);
734 kfree(ic);
735}
736
737/*
738 * An error occurred on the connection
739 */
740void
741__rds_iw_conn_error(struct rds_connection *conn, const char *fmt, ...)
742{
743 va_list ap;
744
745 rds_conn_drop(conn);
746
747 va_start(ap, fmt);
748 vprintk(fmt, ap);
749 va_end(ap);
750}
diff --git a/net/rds/iw_rdma.c b/net/rds/iw_rdma.c
new file mode 100644
index 000000000000..1c02a8f952d0
--- /dev/null
+++ b/net/rds/iw_rdma.c
@@ -0,0 +1,888 @@
1/*
2 * Copyright (c) 2006 Oracle. All rights reserved.
3 *
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
9 *
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
12 * conditions are met:
13 *
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
16 * disclaimer.
17 *
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
22 *
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30 * SOFTWARE.
31 *
32 */
33#include <linux/kernel.h>
34
35#include "rds.h"
36#include "rdma.h"
37#include "iw.h"
38
39
40/*
41 * This is stored as mr->r_trans_private.
42 */
43struct rds_iw_mr {
44 struct rds_iw_device *device;
45 struct rds_iw_mr_pool *pool;
46 struct rdma_cm_id *cm_id;
47
48 struct ib_mr *mr;
49 struct ib_fast_reg_page_list *page_list;
50
51 struct rds_iw_mapping mapping;
52 unsigned char remap_count;
53};
54
55/*
56 * Our own little MR pool
57 */
58struct rds_iw_mr_pool {
59 struct rds_iw_device *device; /* back ptr to the device that owns us */
60
61 struct mutex flush_lock; /* serialize fmr invalidate */
62 struct work_struct flush_worker; /* flush worker */
63
64 spinlock_t list_lock; /* protect variables below */
65 atomic_t item_count; /* total # of MRs */
66 atomic_t dirty_count; /* # dirty of MRs */
67 struct list_head dirty_list; /* dirty mappings */
68 struct list_head clean_list; /* unused & unamapped MRs */
69 atomic_t free_pinned; /* memory pinned by free MRs */
70 unsigned long max_message_size; /* in pages */
71 unsigned long max_items;
72 unsigned long max_items_soft;
73 unsigned long max_free_pinned;
74 int max_pages;
75};
76
77static int rds_iw_flush_mr_pool(struct rds_iw_mr_pool *pool, int free_all);
78static void rds_iw_mr_pool_flush_worker(struct work_struct *work);
79static int rds_iw_init_fastreg(struct rds_iw_mr_pool *pool, struct rds_iw_mr *ibmr);
80static int rds_iw_map_fastreg(struct rds_iw_mr_pool *pool,
81 struct rds_iw_mr *ibmr,
82 struct scatterlist *sg, unsigned int nents);
83static void rds_iw_free_fastreg(struct rds_iw_mr_pool *pool, struct rds_iw_mr *ibmr);
84static unsigned int rds_iw_unmap_fastreg_list(struct rds_iw_mr_pool *pool,
85 struct list_head *unmap_list,
86 struct list_head *kill_list);
87static void rds_iw_destroy_fastreg(struct rds_iw_mr_pool *pool, struct rds_iw_mr *ibmr);
88
89static int rds_iw_get_device(struct rds_sock *rs, struct rds_iw_device **rds_iwdev, struct rdma_cm_id **cm_id)
90{
91 struct rds_iw_device *iwdev;
92 struct rds_iw_cm_id *i_cm_id;
93
94 *rds_iwdev = NULL;
95 *cm_id = NULL;
96
97 list_for_each_entry(iwdev, &rds_iw_devices, list) {
98 spin_lock_irq(&iwdev->spinlock);
99 list_for_each_entry(i_cm_id, &iwdev->cm_id_list, list) {
100 struct sockaddr_in *src_addr, *dst_addr;
101
102 src_addr = (struct sockaddr_in *)&i_cm_id->cm_id->route.addr.src_addr;
103 dst_addr = (struct sockaddr_in *)&i_cm_id->cm_id->route.addr.dst_addr;
104
105 rdsdebug("local ipaddr = %x port %d, "
106 "remote ipaddr = %x port %d"
107 "..looking for %x port %d, "
108 "remote ipaddr = %x port %d\n",
109 src_addr->sin_addr.s_addr,
110 src_addr->sin_port,
111 dst_addr->sin_addr.s_addr,
112 dst_addr->sin_port,
113 rs->rs_bound_addr,
114 rs->rs_bound_port,
115 rs->rs_conn_addr,
116 rs->rs_conn_port);
117#ifdef WORKING_TUPLE_DETECTION
118 if (src_addr->sin_addr.s_addr == rs->rs_bound_addr &&
119 src_addr->sin_port == rs->rs_bound_port &&
120 dst_addr->sin_addr.s_addr == rs->rs_conn_addr &&
121 dst_addr->sin_port == rs->rs_conn_port) {
122#else
123 /* FIXME - needs to compare the local and remote
124 * ipaddr/port tuple, but the ipaddr is the only
125 * available infomation in the rds_sock (as the rest are
126 * zero'ed. It doesn't appear to be properly populated
127 * during connection setup...
128 */
129 if (src_addr->sin_addr.s_addr == rs->rs_bound_addr) {
130#endif
131 spin_unlock_irq(&iwdev->spinlock);
132 *rds_iwdev = iwdev;
133 *cm_id = i_cm_id->cm_id;
134 return 0;
135 }
136 }
137 spin_unlock_irq(&iwdev->spinlock);
138 }
139
140 return 1;
141}
142
143static int rds_iw_add_cm_id(struct rds_iw_device *rds_iwdev, struct rdma_cm_id *cm_id)
144{
145 struct rds_iw_cm_id *i_cm_id;
146
147 i_cm_id = kmalloc(sizeof *i_cm_id, GFP_KERNEL);
148 if (!i_cm_id)
149 return -ENOMEM;
150
151 i_cm_id->cm_id = cm_id;
152
153 spin_lock_irq(&rds_iwdev->spinlock);
154 list_add_tail(&i_cm_id->list, &rds_iwdev->cm_id_list);
155 spin_unlock_irq(&rds_iwdev->spinlock);
156
157 return 0;
158}
159
160void rds_iw_remove_cm_id(struct rds_iw_device *rds_iwdev, struct rdma_cm_id *cm_id)
161{
162 struct rds_iw_cm_id *i_cm_id;
163
164 spin_lock_irq(&rds_iwdev->spinlock);
165 list_for_each_entry(i_cm_id, &rds_iwdev->cm_id_list, list) {
166 if (i_cm_id->cm_id == cm_id) {
167 list_del(&i_cm_id->list);
168 kfree(i_cm_id);
169 break;
170 }
171 }
172 spin_unlock_irq(&rds_iwdev->spinlock);
173}
174
175
176int rds_iw_update_cm_id(struct rds_iw_device *rds_iwdev, struct rdma_cm_id *cm_id)
177{
178 struct sockaddr_in *src_addr, *dst_addr;
179 struct rds_iw_device *rds_iwdev_old;
180 struct rds_sock rs;
181 struct rdma_cm_id *pcm_id;
182 int rc;
183
184 src_addr = (struct sockaddr_in *)&cm_id->route.addr.src_addr;
185 dst_addr = (struct sockaddr_in *)&cm_id->route.addr.dst_addr;
186
187 rs.rs_bound_addr = src_addr->sin_addr.s_addr;
188 rs.rs_bound_port = src_addr->sin_port;
189 rs.rs_conn_addr = dst_addr->sin_addr.s_addr;
190 rs.rs_conn_port = dst_addr->sin_port;
191
192 rc = rds_iw_get_device(&rs, &rds_iwdev_old, &pcm_id);
193 if (rc)
194 rds_iw_remove_cm_id(rds_iwdev, cm_id);
195
196 return rds_iw_add_cm_id(rds_iwdev, cm_id);
197}
198
199int rds_iw_add_conn(struct rds_iw_device *rds_iwdev, struct rds_connection *conn)
200{
201 struct rds_iw_connection *ic = conn->c_transport_data;
202
203 /* conn was previously on the nodev_conns_list */
204 spin_lock_irq(&iw_nodev_conns_lock);
205 BUG_ON(list_empty(&iw_nodev_conns));
206 BUG_ON(list_empty(&ic->iw_node));
207 list_del(&ic->iw_node);
208 spin_unlock_irq(&iw_nodev_conns_lock);
209
210 spin_lock_irq(&rds_iwdev->spinlock);
211 list_add_tail(&ic->iw_node, &rds_iwdev->conn_list);
212 spin_unlock_irq(&rds_iwdev->spinlock);
213
214 ic->rds_iwdev = rds_iwdev;
215
216 return 0;
217}
218
219void rds_iw_remove_nodev_conns(void)
220{
221 struct rds_iw_connection *ic, *_ic;
222 LIST_HEAD(tmp_list);
223
224 /* avoid calling conn_destroy with irqs off */
225 spin_lock_irq(&iw_nodev_conns_lock);
226 list_splice(&iw_nodev_conns, &tmp_list);
227 INIT_LIST_HEAD(&iw_nodev_conns);
228 spin_unlock_irq(&iw_nodev_conns_lock);
229
230 list_for_each_entry_safe(ic, _ic, &tmp_list, iw_node) {
231 if (ic->conn->c_passive)
232 rds_conn_destroy(ic->conn->c_passive);
233 rds_conn_destroy(ic->conn);
234 }
235}
236
237void rds_iw_remove_conns(struct rds_iw_device *rds_iwdev)
238{
239 struct rds_iw_connection *ic, *_ic;
240 LIST_HEAD(tmp_list);
241
242 /* avoid calling conn_destroy with irqs off */
243 spin_lock_irq(&rds_iwdev->spinlock);
244 list_splice(&rds_iwdev->conn_list, &tmp_list);
245 INIT_LIST_HEAD(&rds_iwdev->conn_list);
246 spin_unlock_irq(&rds_iwdev->spinlock);
247
248 list_for_each_entry_safe(ic, _ic, &tmp_list, iw_node) {
249 if (ic->conn->c_passive)
250 rds_conn_destroy(ic->conn->c_passive);
251 rds_conn_destroy(ic->conn);
252 }
253}
254
255static void rds_iw_set_scatterlist(struct rds_iw_scatterlist *sg,
256 struct scatterlist *list, unsigned int sg_len)
257{
258 sg->list = list;
259 sg->len = sg_len;
260 sg->dma_len = 0;
261 sg->dma_npages = 0;
262 sg->bytes = 0;
263}
264
265static u64 *rds_iw_map_scatterlist(struct rds_iw_device *rds_iwdev,
266 struct rds_iw_scatterlist *sg,
267 unsigned int dma_page_shift)
268{
269 struct ib_device *dev = rds_iwdev->dev;
270 u64 *dma_pages = NULL;
271 u64 dma_mask;
272 unsigned int dma_page_size;
273 int i, j, ret;
274
275 dma_page_size = 1 << dma_page_shift;
276 dma_mask = dma_page_size - 1;
277
278 WARN_ON(sg->dma_len);
279
280 sg->dma_len = ib_dma_map_sg(dev, sg->list, sg->len, DMA_BIDIRECTIONAL);
281 if (unlikely(!sg->dma_len)) {
282 printk(KERN_WARNING "RDS/IW: dma_map_sg failed!\n");
283 return ERR_PTR(-EBUSY);
284 }
285
286 sg->bytes = 0;
287 sg->dma_npages = 0;
288
289 ret = -EINVAL;
290 for (i = 0; i < sg->dma_len; ++i) {
291 unsigned int dma_len = ib_sg_dma_len(dev, &sg->list[i]);
292 u64 dma_addr = ib_sg_dma_address(dev, &sg->list[i]);
293 u64 end_addr;
294
295 sg->bytes += dma_len;
296
297 end_addr = dma_addr + dma_len;
298 if (dma_addr & dma_mask) {
299 if (i > 0)
300 goto out_unmap;
301 dma_addr &= ~dma_mask;
302 }
303 if (end_addr & dma_mask) {
304 if (i < sg->dma_len - 1)
305 goto out_unmap;
306 end_addr = (end_addr + dma_mask) & ~dma_mask;
307 }
308
309 sg->dma_npages += (end_addr - dma_addr) >> dma_page_shift;
310 }
311
312 /* Now gather the dma addrs into one list */
313 if (sg->dma_npages > fastreg_message_size)
314 goto out_unmap;
315
316 dma_pages = kmalloc(sizeof(u64) * sg->dma_npages, GFP_ATOMIC);
317 if (!dma_pages) {
318 ret = -ENOMEM;
319 goto out_unmap;
320 }
321
322 for (i = j = 0; i < sg->dma_len; ++i) {
323 unsigned int dma_len = ib_sg_dma_len(dev, &sg->list[i]);
324 u64 dma_addr = ib_sg_dma_address(dev, &sg->list[i]);
325 u64 end_addr;
326
327 end_addr = dma_addr + dma_len;
328 dma_addr &= ~dma_mask;
329 for (; dma_addr < end_addr; dma_addr += dma_page_size)
330 dma_pages[j++] = dma_addr;
331 BUG_ON(j > sg->dma_npages);
332 }
333
334 return dma_pages;
335
336out_unmap:
337 ib_dma_unmap_sg(rds_iwdev->dev, sg->list, sg->len, DMA_BIDIRECTIONAL);
338 sg->dma_len = 0;
339 kfree(dma_pages);
340 return ERR_PTR(ret);
341}
342
343
344struct rds_iw_mr_pool *rds_iw_create_mr_pool(struct rds_iw_device *rds_iwdev)
345{
346 struct rds_iw_mr_pool *pool;
347
348 pool = kzalloc(sizeof(*pool), GFP_KERNEL);
349 if (!pool) {
350 printk(KERN_WARNING "RDS/IW: rds_iw_create_mr_pool alloc error\n");
351 return ERR_PTR(-ENOMEM);
352 }
353
354 pool->device = rds_iwdev;
355 INIT_LIST_HEAD(&pool->dirty_list);
356 INIT_LIST_HEAD(&pool->clean_list);
357 mutex_init(&pool->flush_lock);
358 spin_lock_init(&pool->list_lock);
359 INIT_WORK(&pool->flush_worker, rds_iw_mr_pool_flush_worker);
360
361 pool->max_message_size = fastreg_message_size;
362 pool->max_items = fastreg_pool_size;
363 pool->max_free_pinned = pool->max_items * pool->max_message_size / 4;
364 pool->max_pages = fastreg_message_size;
365
366 /* We never allow more than max_items MRs to be allocated.
367 * When we exceed more than max_items_soft, we start freeing
368 * items more aggressively.
369 * Make sure that max_items > max_items_soft > max_items / 2
370 */
371 pool->max_items_soft = pool->max_items * 3 / 4;
372
373 return pool;
374}
375
376void rds_iw_get_mr_info(struct rds_iw_device *rds_iwdev, struct rds_info_rdma_connection *iinfo)
377{
378 struct rds_iw_mr_pool *pool = rds_iwdev->mr_pool;
379
380 iinfo->rdma_mr_max = pool->max_items;
381 iinfo->rdma_mr_size = pool->max_pages;
382}
383
384void rds_iw_destroy_mr_pool(struct rds_iw_mr_pool *pool)
385{
386 flush_workqueue(rds_wq);
387 rds_iw_flush_mr_pool(pool, 1);
388 BUG_ON(atomic_read(&pool->item_count));
389 BUG_ON(atomic_read(&pool->free_pinned));
390 kfree(pool);
391}
392
393static inline struct rds_iw_mr *rds_iw_reuse_fmr(struct rds_iw_mr_pool *pool)
394{
395 struct rds_iw_mr *ibmr = NULL;
396 unsigned long flags;
397
398 spin_lock_irqsave(&pool->list_lock, flags);
399 if (!list_empty(&pool->clean_list)) {
400 ibmr = list_entry(pool->clean_list.next, struct rds_iw_mr, mapping.m_list);
401 list_del_init(&ibmr->mapping.m_list);
402 }
403 spin_unlock_irqrestore(&pool->list_lock, flags);
404
405 return ibmr;
406}
407
408static struct rds_iw_mr *rds_iw_alloc_mr(struct rds_iw_device *rds_iwdev)
409{
410 struct rds_iw_mr_pool *pool = rds_iwdev->mr_pool;
411 struct rds_iw_mr *ibmr = NULL;
412 int err = 0, iter = 0;
413
414 while (1) {
415 ibmr = rds_iw_reuse_fmr(pool);
416 if (ibmr)
417 return ibmr;
418
419 /* No clean MRs - now we have the choice of either
420 * allocating a fresh MR up to the limit imposed by the
421 * driver, or flush any dirty unused MRs.
422 * We try to avoid stalling in the send path if possible,
423 * so we allocate as long as we're allowed to.
424 *
425 * We're fussy with enforcing the FMR limit, though. If the driver
426 * tells us we can't use more than N fmrs, we shouldn't start
427 * arguing with it */
428 if (atomic_inc_return(&pool->item_count) <= pool->max_items)
429 break;
430
431 atomic_dec(&pool->item_count);
432
433 if (++iter > 2) {
434 rds_iw_stats_inc(s_iw_rdma_mr_pool_depleted);
435 return ERR_PTR(-EAGAIN);
436 }
437
438 /* We do have some empty MRs. Flush them out. */
439 rds_iw_stats_inc(s_iw_rdma_mr_pool_wait);
440 rds_iw_flush_mr_pool(pool, 0);
441 }
442
443 ibmr = kzalloc(sizeof(*ibmr), GFP_KERNEL);
444 if (!ibmr) {
445 err = -ENOMEM;
446 goto out_no_cigar;
447 }
448
449 spin_lock_init(&ibmr->mapping.m_lock);
450 INIT_LIST_HEAD(&ibmr->mapping.m_list);
451 ibmr->mapping.m_mr = ibmr;
452
453 err = rds_iw_init_fastreg(pool, ibmr);
454 if (err)
455 goto out_no_cigar;
456
457 rds_iw_stats_inc(s_iw_rdma_mr_alloc);
458 return ibmr;
459
460out_no_cigar:
461 if (ibmr) {
462 rds_iw_destroy_fastreg(pool, ibmr);
463 kfree(ibmr);
464 }
465 atomic_dec(&pool->item_count);
466 return ERR_PTR(err);
467}
468
469void rds_iw_sync_mr(void *trans_private, int direction)
470{
471 struct rds_iw_mr *ibmr = trans_private;
472 struct rds_iw_device *rds_iwdev = ibmr->device;
473
474 switch (direction) {
475 case DMA_FROM_DEVICE:
476 ib_dma_sync_sg_for_cpu(rds_iwdev->dev, ibmr->mapping.m_sg.list,
477 ibmr->mapping.m_sg.dma_len, DMA_BIDIRECTIONAL);
478 break;
479 case DMA_TO_DEVICE:
480 ib_dma_sync_sg_for_device(rds_iwdev->dev, ibmr->mapping.m_sg.list,
481 ibmr->mapping.m_sg.dma_len, DMA_BIDIRECTIONAL);
482 break;
483 }
484}
485
486static inline unsigned int rds_iw_flush_goal(struct rds_iw_mr_pool *pool, int free_all)
487{
488 unsigned int item_count;
489
490 item_count = atomic_read(&pool->item_count);
491 if (free_all)
492 return item_count;
493
494 return 0;
495}
496
497/*
498 * Flush our pool of MRs.
499 * At a minimum, all currently unused MRs are unmapped.
500 * If the number of MRs allocated exceeds the limit, we also try
501 * to free as many MRs as needed to get back to this limit.
502 */
503static int rds_iw_flush_mr_pool(struct rds_iw_mr_pool *pool, int free_all)
504{
505 struct rds_iw_mr *ibmr, *next;
506 LIST_HEAD(unmap_list);
507 LIST_HEAD(kill_list);
508 unsigned long flags;
509 unsigned int nfreed = 0, ncleaned = 0, free_goal;
510 int ret = 0;
511
512 rds_iw_stats_inc(s_iw_rdma_mr_pool_flush);
513
514 mutex_lock(&pool->flush_lock);
515
516 spin_lock_irqsave(&pool->list_lock, flags);
517 /* Get the list of all mappings to be destroyed */
518 list_splice_init(&pool->dirty_list, &unmap_list);
519 if (free_all)
520 list_splice_init(&pool->clean_list, &kill_list);
521 spin_unlock_irqrestore(&pool->list_lock, flags);
522
523 free_goal = rds_iw_flush_goal(pool, free_all);
524
525 /* Batched invalidate of dirty MRs.
526 * For FMR based MRs, the mappings on the unmap list are
527 * actually members of an ibmr (ibmr->mapping). They either
528 * migrate to the kill_list, or have been cleaned and should be
529 * moved to the clean_list.
530 * For fastregs, they will be dynamically allocated, and
531 * will be destroyed by the unmap function.
532 */
533 if (!list_empty(&unmap_list)) {
534 ncleaned = rds_iw_unmap_fastreg_list(pool, &unmap_list, &kill_list);
535 /* If we've been asked to destroy all MRs, move those
536 * that were simply cleaned to the kill list */
537 if (free_all)
538 list_splice_init(&unmap_list, &kill_list);
539 }
540
541 /* Destroy any MRs that are past their best before date */
542 list_for_each_entry_safe(ibmr, next, &kill_list, mapping.m_list) {
543 rds_iw_stats_inc(s_iw_rdma_mr_free);
544 list_del(&ibmr->mapping.m_list);
545 rds_iw_destroy_fastreg(pool, ibmr);
546 kfree(ibmr);
547 nfreed++;
548 }
549
550 /* Anything that remains are laundered ibmrs, which we can add
551 * back to the clean list. */
552 if (!list_empty(&unmap_list)) {
553 spin_lock_irqsave(&pool->list_lock, flags);
554 list_splice(&unmap_list, &pool->clean_list);
555 spin_unlock_irqrestore(&pool->list_lock, flags);
556 }
557
558 atomic_sub(ncleaned, &pool->dirty_count);
559 atomic_sub(nfreed, &pool->item_count);
560
561 mutex_unlock(&pool->flush_lock);
562 return ret;
563}
564
565static void rds_iw_mr_pool_flush_worker(struct work_struct *work)
566{
567 struct rds_iw_mr_pool *pool = container_of(work, struct rds_iw_mr_pool, flush_worker);
568
569 rds_iw_flush_mr_pool(pool, 0);
570}
571
572void rds_iw_free_mr(void *trans_private, int invalidate)
573{
574 struct rds_iw_mr *ibmr = trans_private;
575 struct rds_iw_mr_pool *pool = ibmr->device->mr_pool;
576
577 rdsdebug("RDS/IW: free_mr nents %u\n", ibmr->mapping.m_sg.len);
578 if (!pool)
579 return;
580
581 /* Return it to the pool's free list */
582 rds_iw_free_fastreg(pool, ibmr);
583
584 /* If we've pinned too many pages, request a flush */
585 if (atomic_read(&pool->free_pinned) >= pool->max_free_pinned
586 || atomic_read(&pool->dirty_count) >= pool->max_items / 10)
587 queue_work(rds_wq, &pool->flush_worker);
588
589 if (invalidate) {
590 if (likely(!in_interrupt())) {
591 rds_iw_flush_mr_pool(pool, 0);
592 } else {
593 /* We get here if the user created a MR marked
594 * as use_once and invalidate at the same time. */
595 queue_work(rds_wq, &pool->flush_worker);
596 }
597 }
598}
599
600void rds_iw_flush_mrs(void)
601{
602 struct rds_iw_device *rds_iwdev;
603
604 list_for_each_entry(rds_iwdev, &rds_iw_devices, list) {
605 struct rds_iw_mr_pool *pool = rds_iwdev->mr_pool;
606
607 if (pool)
608 rds_iw_flush_mr_pool(pool, 0);
609 }
610}
611
612void *rds_iw_get_mr(struct scatterlist *sg, unsigned long nents,
613 struct rds_sock *rs, u32 *key_ret)
614{
615 struct rds_iw_device *rds_iwdev;
616 struct rds_iw_mr *ibmr = NULL;
617 struct rdma_cm_id *cm_id;
618 int ret;
619
620 ret = rds_iw_get_device(rs, &rds_iwdev, &cm_id);
621 if (ret || !cm_id) {
622 ret = -ENODEV;
623 goto out;
624 }
625
626 if (!rds_iwdev->mr_pool) {
627 ret = -ENODEV;
628 goto out;
629 }
630
631 ibmr = rds_iw_alloc_mr(rds_iwdev);
632 if (IS_ERR(ibmr))
633 return ibmr;
634
635 ibmr->cm_id = cm_id;
636 ibmr->device = rds_iwdev;
637
638 ret = rds_iw_map_fastreg(rds_iwdev->mr_pool, ibmr, sg, nents);
639 if (ret == 0)
640 *key_ret = ibmr->mr->rkey;
641 else
642 printk(KERN_WARNING "RDS/IW: failed to map mr (errno=%d)\n", ret);
643
644out:
645 if (ret) {
646 if (ibmr)
647 rds_iw_free_mr(ibmr, 0);
648 ibmr = ERR_PTR(ret);
649 }
650 return ibmr;
651}
652
653/*
654 * iWARP fastreg handling
655 *
656 * The life cycle of a fastreg registration is a bit different from
657 * FMRs.
658 * The idea behind fastreg is to have one MR, to which we bind different
659 * mappings over time. To avoid stalling on the expensive map and invalidate
660 * operations, these operations are pipelined on the same send queue on
661 * which we want to send the message containing the r_key.
662 *
663 * This creates a bit of a problem for us, as we do not have the destination
664 * IP in GET_MR, so the connection must be setup prior to the GET_MR call for
665 * RDMA to be correctly setup. If a fastreg request is present, rds_iw_xmit
666 * will try to queue a LOCAL_INV (if needed) and a FAST_REG_MR work request
667 * before queuing the SEND. When completions for these arrive, they are
668 * dispatched to the MR has a bit set showing that RDMa can be performed.
669 *
670 * There is another interesting aspect that's related to invalidation.
671 * The application can request that a mapping is invalidated in FREE_MR.
672 * The expectation there is that this invalidation step includes ALL
673 * PREVIOUSLY FREED MRs.
674 */
675static int rds_iw_init_fastreg(struct rds_iw_mr_pool *pool,
676 struct rds_iw_mr *ibmr)
677{
678 struct rds_iw_device *rds_iwdev = pool->device;
679 struct ib_fast_reg_page_list *page_list = NULL;
680 struct ib_mr *mr;
681 int err;
682
683 mr = ib_alloc_fast_reg_mr(rds_iwdev->pd, pool->max_message_size);
684 if (IS_ERR(mr)) {
685 err = PTR_ERR(mr);
686
687 printk(KERN_WARNING "RDS/IW: ib_alloc_fast_reg_mr failed (err=%d)\n", err);
688 return err;
689 }
690
691 /* FIXME - this is overkill, but mapping->m_sg.dma_len/mapping->m_sg.dma_npages
692 * is not filled in.
693 */
694 page_list = ib_alloc_fast_reg_page_list(rds_iwdev->dev, pool->max_message_size);
695 if (IS_ERR(page_list)) {
696 err = PTR_ERR(page_list);
697
698 printk(KERN_WARNING "RDS/IW: ib_alloc_fast_reg_page_list failed (err=%d)\n", err);
699 ib_dereg_mr(mr);
700 return err;
701 }
702
703 ibmr->page_list = page_list;
704 ibmr->mr = mr;
705 return 0;
706}
707
708static int rds_iw_rdma_build_fastreg(struct rds_iw_mapping *mapping)
709{
710 struct rds_iw_mr *ibmr = mapping->m_mr;
711 struct ib_send_wr f_wr, *failed_wr;
712 int ret;
713
714 /*
715 * Perform a WR for the fast_reg_mr. Each individual page
716 * in the sg list is added to the fast reg page list and placed
717 * inside the fast_reg_mr WR. The key used is a rolling 8bit
718 * counter, which should guarantee uniqueness.
719 */
720 ib_update_fast_reg_key(ibmr->mr, ibmr->remap_count++);
721 mapping->m_rkey = ibmr->mr->rkey;
722
723 memset(&f_wr, 0, sizeof(f_wr));
724 f_wr.wr_id = RDS_IW_FAST_REG_WR_ID;
725 f_wr.opcode = IB_WR_FAST_REG_MR;
726 f_wr.wr.fast_reg.length = mapping->m_sg.bytes;
727 f_wr.wr.fast_reg.rkey = mapping->m_rkey;
728 f_wr.wr.fast_reg.page_list = ibmr->page_list;
729 f_wr.wr.fast_reg.page_list_len = mapping->m_sg.dma_len;
730 f_wr.wr.fast_reg.page_shift = ibmr->device->page_shift;
731 f_wr.wr.fast_reg.access_flags = IB_ACCESS_LOCAL_WRITE |
732 IB_ACCESS_REMOTE_READ |
733 IB_ACCESS_REMOTE_WRITE;
734 f_wr.wr.fast_reg.iova_start = 0;
735 f_wr.send_flags = IB_SEND_SIGNALED;
736
737 failed_wr = &f_wr;
738 ret = ib_post_send(ibmr->cm_id->qp, &f_wr, &failed_wr);
739 BUG_ON(failed_wr != &f_wr);
740 if (ret && printk_ratelimit())
741 printk(KERN_WARNING "RDS/IW: %s:%d ib_post_send returned %d\n",
742 __func__, __LINE__, ret);
743 return ret;
744}
745
746static int rds_iw_rdma_fastreg_inv(struct rds_iw_mr *ibmr)
747{
748 struct ib_send_wr s_wr, *failed_wr;
749 int ret = 0;
750
751 if (!ibmr->cm_id->qp || !ibmr->mr)
752 goto out;
753
754 memset(&s_wr, 0, sizeof(s_wr));
755 s_wr.wr_id = RDS_IW_LOCAL_INV_WR_ID;
756 s_wr.opcode = IB_WR_LOCAL_INV;
757 s_wr.ex.invalidate_rkey = ibmr->mr->rkey;
758 s_wr.send_flags = IB_SEND_SIGNALED;
759
760 failed_wr = &s_wr;
761 ret = ib_post_send(ibmr->cm_id->qp, &s_wr, &failed_wr);
762 if (ret && printk_ratelimit()) {
763 printk(KERN_WARNING "RDS/IW: %s:%d ib_post_send returned %d\n",
764 __func__, __LINE__, ret);
765 goto out;
766 }
767out:
768 return ret;
769}
770
771static int rds_iw_map_fastreg(struct rds_iw_mr_pool *pool,
772 struct rds_iw_mr *ibmr,
773 struct scatterlist *sg,
774 unsigned int sg_len)
775{
776 struct rds_iw_device *rds_iwdev = pool->device;
777 struct rds_iw_mapping *mapping = &ibmr->mapping;
778 u64 *dma_pages;
779 int i, ret = 0;
780
781 rds_iw_set_scatterlist(&mapping->m_sg, sg, sg_len);
782
783 dma_pages = rds_iw_map_scatterlist(rds_iwdev,
784 &mapping->m_sg,
785 rds_iwdev->page_shift);
786 if (IS_ERR(dma_pages)) {
787 ret = PTR_ERR(dma_pages);
788 dma_pages = NULL;
789 goto out;
790 }
791
792 if (mapping->m_sg.dma_len > pool->max_message_size) {
793 ret = -EMSGSIZE;
794 goto out;
795 }
796
797 for (i = 0; i < mapping->m_sg.dma_npages; ++i)
798 ibmr->page_list->page_list[i] = dma_pages[i];
799
800 ret = rds_iw_rdma_build_fastreg(mapping);
801 if (ret)
802 goto out;
803
804 rds_iw_stats_inc(s_iw_rdma_mr_used);
805
806out:
807 kfree(dma_pages);
808
809 return ret;
810}
811
812/*
813 * "Free" a fastreg MR.
814 */
815static void rds_iw_free_fastreg(struct rds_iw_mr_pool *pool,
816 struct rds_iw_mr *ibmr)
817{
818 unsigned long flags;
819 int ret;
820
821 if (!ibmr->mapping.m_sg.dma_len)
822 return;
823
824 ret = rds_iw_rdma_fastreg_inv(ibmr);
825 if (ret)
826 return;
827
828 /* Try to post the LOCAL_INV WR to the queue. */
829 spin_lock_irqsave(&pool->list_lock, flags);
830
831 list_add_tail(&ibmr->mapping.m_list, &pool->dirty_list);
832 atomic_add(ibmr->mapping.m_sg.len, &pool->free_pinned);
833 atomic_inc(&pool->dirty_count);
834
835 spin_unlock_irqrestore(&pool->list_lock, flags);
836}
837
838static unsigned int rds_iw_unmap_fastreg_list(struct rds_iw_mr_pool *pool,
839 struct list_head *unmap_list,
840 struct list_head *kill_list)
841{
842 struct rds_iw_mapping *mapping, *next;
843 unsigned int ncleaned = 0;
844 LIST_HEAD(laundered);
845
846 /* Batched invalidation of fastreg MRs.
847 * Why do we do it this way, even though we could pipeline unmap
848 * and remap? The reason is the application semantics - when the
849 * application requests an invalidation of MRs, it expects all
850 * previously released R_Keys to become invalid.
851 *
852 * If we implement MR reuse naively, we risk memory corruption
853 * (this has actually been observed). So the default behavior
854 * requires that a MR goes through an explicit unmap operation before
855 * we can reuse it again.
856 *
857 * We could probably improve on this a little, by allowing immediate
858 * reuse of a MR on the same socket (eg you could add small
859 * cache of unused MRs to strct rds_socket - GET_MR could grab one
860 * of these without requiring an explicit invalidate).
861 */
862 while (!list_empty(unmap_list)) {
863 unsigned long flags;
864
865 spin_lock_irqsave(&pool->list_lock, flags);
866 list_for_each_entry_safe(mapping, next, unmap_list, m_list) {
867 list_move(&mapping->m_list, &laundered);
868 ncleaned++;
869 }
870 spin_unlock_irqrestore(&pool->list_lock, flags);
871 }
872
873 /* Move all laundered mappings back to the unmap list.
874 * We do not kill any WRs right now - it doesn't seem the
875 * fastreg API has a max_remap limit. */
876 list_splice_init(&laundered, unmap_list);
877
878 return ncleaned;
879}
880
881static void rds_iw_destroy_fastreg(struct rds_iw_mr_pool *pool,
882 struct rds_iw_mr *ibmr)
883{
884 if (ibmr->page_list)
885 ib_free_fast_reg_page_list(ibmr->page_list);
886 if (ibmr->mr)
887 ib_dereg_mr(ibmr->mr);
888}
diff --git a/net/rds/iw_recv.c b/net/rds/iw_recv.c
new file mode 100644
index 000000000000..a1931f0027a2
--- /dev/null
+++ b/net/rds/iw_recv.c
@@ -0,0 +1,869 @@
1/*
2 * Copyright (c) 2006 Oracle. All rights reserved.
3 *
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
9 *
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
12 * conditions are met:
13 *
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
16 * disclaimer.
17 *
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
22 *
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30 * SOFTWARE.
31 *
32 */
33#include <linux/kernel.h>
34#include <linux/pci.h>
35#include <linux/dma-mapping.h>
36#include <rdma/rdma_cm.h>
37
38#include "rds.h"
39#include "iw.h"
40
41static struct kmem_cache *rds_iw_incoming_slab;
42static struct kmem_cache *rds_iw_frag_slab;
43static atomic_t rds_iw_allocation = ATOMIC_INIT(0);
44
45static void rds_iw_frag_drop_page(struct rds_page_frag *frag)
46{
47 rdsdebug("frag %p page %p\n", frag, frag->f_page);
48 __free_page(frag->f_page);
49 frag->f_page = NULL;
50}
51
52static void rds_iw_frag_free(struct rds_page_frag *frag)
53{
54 rdsdebug("frag %p page %p\n", frag, frag->f_page);
55 BUG_ON(frag->f_page != NULL);
56 kmem_cache_free(rds_iw_frag_slab, frag);
57}
58
59/*
60 * We map a page at a time. Its fragments are posted in order. This
61 * is called in fragment order as the fragments get send completion events.
62 * Only the last frag in the page performs the unmapping.
63 *
64 * It's OK for ring cleanup to call this in whatever order it likes because
65 * DMA is not in flight and so we can unmap while other ring entries still
66 * hold page references in their frags.
67 */
68static void rds_iw_recv_unmap_page(struct rds_iw_connection *ic,
69 struct rds_iw_recv_work *recv)
70{
71 struct rds_page_frag *frag = recv->r_frag;
72
73 rdsdebug("recv %p frag %p page %p\n", recv, frag, frag->f_page);
74 if (frag->f_mapped)
75 ib_dma_unmap_page(ic->i_cm_id->device,
76 frag->f_mapped,
77 RDS_FRAG_SIZE, DMA_FROM_DEVICE);
78 frag->f_mapped = 0;
79}
80
81void rds_iw_recv_init_ring(struct rds_iw_connection *ic)
82{
83 struct rds_iw_recv_work *recv;
84 u32 i;
85
86 for (i = 0, recv = ic->i_recvs; i < ic->i_recv_ring.w_nr; i++, recv++) {
87 struct ib_sge *sge;
88
89 recv->r_iwinc = NULL;
90 recv->r_frag = NULL;
91
92 recv->r_wr.next = NULL;
93 recv->r_wr.wr_id = i;
94 recv->r_wr.sg_list = recv->r_sge;
95 recv->r_wr.num_sge = RDS_IW_RECV_SGE;
96
97 sge = rds_iw_data_sge(ic, recv->r_sge);
98 sge->addr = 0;
99 sge->length = RDS_FRAG_SIZE;
100 sge->lkey = 0;
101
102 sge = rds_iw_header_sge(ic, recv->r_sge);
103 sge->addr = ic->i_recv_hdrs_dma + (i * sizeof(struct rds_header));
104 sge->length = sizeof(struct rds_header);
105 sge->lkey = 0;
106 }
107}
108
109static void rds_iw_recv_clear_one(struct rds_iw_connection *ic,
110 struct rds_iw_recv_work *recv)
111{
112 if (recv->r_iwinc) {
113 rds_inc_put(&recv->r_iwinc->ii_inc);
114 recv->r_iwinc = NULL;
115 }
116 if (recv->r_frag) {
117 rds_iw_recv_unmap_page(ic, recv);
118 if (recv->r_frag->f_page)
119 rds_iw_frag_drop_page(recv->r_frag);
120 rds_iw_frag_free(recv->r_frag);
121 recv->r_frag = NULL;
122 }
123}
124
125void rds_iw_recv_clear_ring(struct rds_iw_connection *ic)
126{
127 u32 i;
128
129 for (i = 0; i < ic->i_recv_ring.w_nr; i++)
130 rds_iw_recv_clear_one(ic, &ic->i_recvs[i]);
131
132 if (ic->i_frag.f_page)
133 rds_iw_frag_drop_page(&ic->i_frag);
134}
135
136static int rds_iw_recv_refill_one(struct rds_connection *conn,
137 struct rds_iw_recv_work *recv,
138 gfp_t kptr_gfp, gfp_t page_gfp)
139{
140 struct rds_iw_connection *ic = conn->c_transport_data;
141 dma_addr_t dma_addr;
142 struct ib_sge *sge;
143 int ret = -ENOMEM;
144
145 if (recv->r_iwinc == NULL) {
146 if (atomic_read(&rds_iw_allocation) >= rds_iw_sysctl_max_recv_allocation) {
147 rds_iw_stats_inc(s_iw_rx_alloc_limit);
148 goto out;
149 }
150 recv->r_iwinc = kmem_cache_alloc(rds_iw_incoming_slab,
151 kptr_gfp);
152 if (recv->r_iwinc == NULL)
153 goto out;
154 atomic_inc(&rds_iw_allocation);
155 INIT_LIST_HEAD(&recv->r_iwinc->ii_frags);
156 rds_inc_init(&recv->r_iwinc->ii_inc, conn, conn->c_faddr);
157 }
158
159 if (recv->r_frag == NULL) {
160 recv->r_frag = kmem_cache_alloc(rds_iw_frag_slab, kptr_gfp);
161 if (recv->r_frag == NULL)
162 goto out;
163 INIT_LIST_HEAD(&recv->r_frag->f_item);
164 recv->r_frag->f_page = NULL;
165 }
166
167 if (ic->i_frag.f_page == NULL) {
168 ic->i_frag.f_page = alloc_page(page_gfp);
169 if (ic->i_frag.f_page == NULL)
170 goto out;
171 ic->i_frag.f_offset = 0;
172 }
173
174 dma_addr = ib_dma_map_page(ic->i_cm_id->device,
175 ic->i_frag.f_page,
176 ic->i_frag.f_offset,
177 RDS_FRAG_SIZE,
178 DMA_FROM_DEVICE);
179 if (ib_dma_mapping_error(ic->i_cm_id->device, dma_addr))
180 goto out;
181
182 /*
183 * Once we get the RDS_PAGE_LAST_OFF frag then rds_iw_frag_unmap()
184 * must be called on this recv. This happens as completions hit
185 * in order or on connection shutdown.
186 */
187 recv->r_frag->f_page = ic->i_frag.f_page;
188 recv->r_frag->f_offset = ic->i_frag.f_offset;
189 recv->r_frag->f_mapped = dma_addr;
190
191 sge = rds_iw_data_sge(ic, recv->r_sge);
192 sge->addr = dma_addr;
193 sge->length = RDS_FRAG_SIZE;
194
195 sge = rds_iw_header_sge(ic, recv->r_sge);
196 sge->addr = ic->i_recv_hdrs_dma + (recv - ic->i_recvs) * sizeof(struct rds_header);
197 sge->length = sizeof(struct rds_header);
198
199 get_page(recv->r_frag->f_page);
200
201 if (ic->i_frag.f_offset < RDS_PAGE_LAST_OFF) {
202 ic->i_frag.f_offset += RDS_FRAG_SIZE;
203 } else {
204 put_page(ic->i_frag.f_page);
205 ic->i_frag.f_page = NULL;
206 ic->i_frag.f_offset = 0;
207 }
208
209 ret = 0;
210out:
211 return ret;
212}
213
214/*
215 * This tries to allocate and post unused work requests after making sure that
216 * they have all the allocations they need to queue received fragments into
217 * sockets. The i_recv_mutex is held here so that ring_alloc and _unalloc
218 * pairs don't go unmatched.
219 *
220 * -1 is returned if posting fails due to temporary resource exhaustion.
221 */
222int rds_iw_recv_refill(struct rds_connection *conn, gfp_t kptr_gfp,
223 gfp_t page_gfp, int prefill)
224{
225 struct rds_iw_connection *ic = conn->c_transport_data;
226 struct rds_iw_recv_work *recv;
227 struct ib_recv_wr *failed_wr;
228 unsigned int posted = 0;
229 int ret = 0;
230 u32 pos;
231
232 while ((prefill || rds_conn_up(conn))
233 && rds_iw_ring_alloc(&ic->i_recv_ring, 1, &pos)) {
234 if (pos >= ic->i_recv_ring.w_nr) {
235 printk(KERN_NOTICE "Argh - ring alloc returned pos=%u\n",
236 pos);
237 ret = -EINVAL;
238 break;
239 }
240
241 recv = &ic->i_recvs[pos];
242 ret = rds_iw_recv_refill_one(conn, recv, kptr_gfp, page_gfp);
243 if (ret) {
244 ret = -1;
245 break;
246 }
247
248 /* XXX when can this fail? */
249 ret = ib_post_recv(ic->i_cm_id->qp, &recv->r_wr, &failed_wr);
250 rdsdebug("recv %p iwinc %p page %p addr %lu ret %d\n", recv,
251 recv->r_iwinc, recv->r_frag->f_page,
252 (long) recv->r_frag->f_mapped, ret);
253 if (ret) {
254 rds_iw_conn_error(conn, "recv post on "
255 "%pI4 returned %d, disconnecting and "
256 "reconnecting\n", &conn->c_faddr,
257 ret);
258 ret = -1;
259 break;
260 }
261
262 posted++;
263 }
264
265 /* We're doing flow control - update the window. */
266 if (ic->i_flowctl && posted)
267 rds_iw_advertise_credits(conn, posted);
268
269 if (ret)
270 rds_iw_ring_unalloc(&ic->i_recv_ring, 1);
271 return ret;
272}
273
274void rds_iw_inc_purge(struct rds_incoming *inc)
275{
276 struct rds_iw_incoming *iwinc;
277 struct rds_page_frag *frag;
278 struct rds_page_frag *pos;
279
280 iwinc = container_of(inc, struct rds_iw_incoming, ii_inc);
281 rdsdebug("purging iwinc %p inc %p\n", iwinc, inc);
282
283 list_for_each_entry_safe(frag, pos, &iwinc->ii_frags, f_item) {
284 list_del_init(&frag->f_item);
285 rds_iw_frag_drop_page(frag);
286 rds_iw_frag_free(frag);
287 }
288}
289
290void rds_iw_inc_free(struct rds_incoming *inc)
291{
292 struct rds_iw_incoming *iwinc;
293
294 iwinc = container_of(inc, struct rds_iw_incoming, ii_inc);
295
296 rds_iw_inc_purge(inc);
297 rdsdebug("freeing iwinc %p inc %p\n", iwinc, inc);
298 BUG_ON(!list_empty(&iwinc->ii_frags));
299 kmem_cache_free(rds_iw_incoming_slab, iwinc);
300 atomic_dec(&rds_iw_allocation);
301 BUG_ON(atomic_read(&rds_iw_allocation) < 0);
302}
303
304int rds_iw_inc_copy_to_user(struct rds_incoming *inc, struct iovec *first_iov,
305 size_t size)
306{
307 struct rds_iw_incoming *iwinc;
308 struct rds_page_frag *frag;
309 struct iovec *iov = first_iov;
310 unsigned long to_copy;
311 unsigned long frag_off = 0;
312 unsigned long iov_off = 0;
313 int copied = 0;
314 int ret;
315 u32 len;
316
317 iwinc = container_of(inc, struct rds_iw_incoming, ii_inc);
318 frag = list_entry(iwinc->ii_frags.next, struct rds_page_frag, f_item);
319 len = be32_to_cpu(inc->i_hdr.h_len);
320
321 while (copied < size && copied < len) {
322 if (frag_off == RDS_FRAG_SIZE) {
323 frag = list_entry(frag->f_item.next,
324 struct rds_page_frag, f_item);
325 frag_off = 0;
326 }
327 while (iov_off == iov->iov_len) {
328 iov_off = 0;
329 iov++;
330 }
331
332 to_copy = min(iov->iov_len - iov_off, RDS_FRAG_SIZE - frag_off);
333 to_copy = min_t(size_t, to_copy, size - copied);
334 to_copy = min_t(unsigned long, to_copy, len - copied);
335
336 rdsdebug("%lu bytes to user [%p, %zu] + %lu from frag "
337 "[%p, %lu] + %lu\n",
338 to_copy, iov->iov_base, iov->iov_len, iov_off,
339 frag->f_page, frag->f_offset, frag_off);
340
341 /* XXX needs + offset for multiple recvs per page */
342 ret = rds_page_copy_to_user(frag->f_page,
343 frag->f_offset + frag_off,
344 iov->iov_base + iov_off,
345 to_copy);
346 if (ret) {
347 copied = ret;
348 break;
349 }
350
351 iov_off += to_copy;
352 frag_off += to_copy;
353 copied += to_copy;
354 }
355
356 return copied;
357}
358
359/* ic starts out kzalloc()ed */
360void rds_iw_recv_init_ack(struct rds_iw_connection *ic)
361{
362 struct ib_send_wr *wr = &ic->i_ack_wr;
363 struct ib_sge *sge = &ic->i_ack_sge;
364
365 sge->addr = ic->i_ack_dma;
366 sge->length = sizeof(struct rds_header);
367 sge->lkey = rds_iw_local_dma_lkey(ic);
368
369 wr->sg_list = sge;
370 wr->num_sge = 1;
371 wr->opcode = IB_WR_SEND;
372 wr->wr_id = RDS_IW_ACK_WR_ID;
373 wr->send_flags = IB_SEND_SIGNALED | IB_SEND_SOLICITED;
374}
375
376/*
377 * You'd think that with reliable IB connections you wouldn't need to ack
378 * messages that have been received. The problem is that IB hardware generates
379 * an ack message before it has DMAed the message into memory. This creates a
380 * potential message loss if the HCA is disabled for any reason between when it
381 * sends the ack and before the message is DMAed and processed. This is only a
382 * potential issue if another HCA is available for fail-over.
383 *
384 * When the remote host receives our ack they'll free the sent message from
385 * their send queue. To decrease the latency of this we always send an ack
386 * immediately after we've received messages.
387 *
388 * For simplicity, we only have one ack in flight at a time. This puts
389 * pressure on senders to have deep enough send queues to absorb the latency of
390 * a single ack frame being in flight. This might not be good enough.
391 *
392 * This is implemented by have a long-lived send_wr and sge which point to a
393 * statically allocated ack frame. This ack wr does not fall under the ring
394 * accounting that the tx and rx wrs do. The QP attribute specifically makes
395 * room for it beyond the ring size. Send completion notices its special
396 * wr_id and avoids working with the ring in that case.
397 */
398static void rds_iw_set_ack(struct rds_iw_connection *ic, u64 seq,
399 int ack_required)
400{
401 rds_iw_set_64bit(&ic->i_ack_next, seq);
402 if (ack_required) {
403 smp_mb__before_clear_bit();
404 set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
405 }
406}
407
408static u64 rds_iw_get_ack(struct rds_iw_connection *ic)
409{
410 clear_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
411 smp_mb__after_clear_bit();
412
413 return ic->i_ack_next;
414}
415
416static void rds_iw_send_ack(struct rds_iw_connection *ic, unsigned int adv_credits)
417{
418 struct rds_header *hdr = ic->i_ack;
419 struct ib_send_wr *failed_wr;
420 u64 seq;
421 int ret;
422
423 seq = rds_iw_get_ack(ic);
424
425 rdsdebug("send_ack: ic %p ack %llu\n", ic, (unsigned long long) seq);
426 rds_message_populate_header(hdr, 0, 0, 0);
427 hdr->h_ack = cpu_to_be64(seq);
428 hdr->h_credit = adv_credits;
429 rds_message_make_checksum(hdr);
430 ic->i_ack_queued = jiffies;
431
432 ret = ib_post_send(ic->i_cm_id->qp, &ic->i_ack_wr, &failed_wr);
433 if (unlikely(ret)) {
434 /* Failed to send. Release the WR, and
435 * force another ACK.
436 */
437 clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
438 set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
439
440 rds_iw_stats_inc(s_iw_ack_send_failure);
441 /* Need to finesse this later. */
442 BUG();
443 } else
444 rds_iw_stats_inc(s_iw_ack_sent);
445}
446
447/*
448 * There are 3 ways of getting acknowledgements to the peer:
449 * 1. We call rds_iw_attempt_ack from the recv completion handler
450 * to send an ACK-only frame.
451 * However, there can be only one such frame in the send queue
452 * at any time, so we may have to postpone it.
453 * 2. When another (data) packet is transmitted while there's
454 * an ACK in the queue, we piggyback the ACK sequence number
455 * on the data packet.
456 * 3. If the ACK WR is done sending, we get called from the
457 * send queue completion handler, and check whether there's
458 * another ACK pending (postponed because the WR was on the
459 * queue). If so, we transmit it.
460 *
461 * We maintain 2 variables:
462 * - i_ack_flags, which keeps track of whether the ACK WR
463 * is currently in the send queue or not (IB_ACK_IN_FLIGHT)
464 * - i_ack_next, which is the last sequence number we received
465 *
466 * Potentially, send queue and receive queue handlers can run concurrently.
467 *
468 * Reconnecting complicates this picture just slightly. When we
469 * reconnect, we may be seeing duplicate packets. The peer
470 * is retransmitting them, because it hasn't seen an ACK for
471 * them. It is important that we ACK these.
472 *
473 * ACK mitigation adds a header flag "ACK_REQUIRED"; any packet with
474 * this flag set *MUST* be acknowledged immediately.
475 */
476
477/*
478 * When we get here, we're called from the recv queue handler.
479 * Check whether we ought to transmit an ACK.
480 */
481void rds_iw_attempt_ack(struct rds_iw_connection *ic)
482{
483 unsigned int adv_credits;
484
485 if (!test_bit(IB_ACK_REQUESTED, &ic->i_ack_flags))
486 return;
487
488 if (test_and_set_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags)) {
489 rds_iw_stats_inc(s_iw_ack_send_delayed);
490 return;
491 }
492
493 /* Can we get a send credit? */
494 if (!rds_iw_send_grab_credits(ic, 1, &adv_credits, 0)) {
495 rds_iw_stats_inc(s_iw_tx_throttle);
496 clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
497 return;
498 }
499
500 clear_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
501 rds_iw_send_ack(ic, adv_credits);
502}
503
504/*
505 * We get here from the send completion handler, when the
506 * adapter tells us the ACK frame was sent.
507 */
508void rds_iw_ack_send_complete(struct rds_iw_connection *ic)
509{
510 clear_bit(IB_ACK_IN_FLIGHT, &ic->i_ack_flags);
511 rds_iw_attempt_ack(ic);
512}
513
514/*
515 * This is called by the regular xmit code when it wants to piggyback
516 * an ACK on an outgoing frame.
517 */
518u64 rds_iw_piggyb_ack(struct rds_iw_connection *ic)
519{
520 if (test_and_clear_bit(IB_ACK_REQUESTED, &ic->i_ack_flags))
521 rds_iw_stats_inc(s_iw_ack_send_piggybacked);
522 return rds_iw_get_ack(ic);
523}
524
525/*
526 * It's kind of lame that we're copying from the posted receive pages into
527 * long-lived bitmaps. We could have posted the bitmaps and rdma written into
528 * them. But receiving new congestion bitmaps should be a *rare* event, so
529 * hopefully we won't need to invest that complexity in making it more
530 * efficient. By copying we can share a simpler core with TCP which has to
531 * copy.
532 */
533static void rds_iw_cong_recv(struct rds_connection *conn,
534 struct rds_iw_incoming *iwinc)
535{
536 struct rds_cong_map *map;
537 unsigned int map_off;
538 unsigned int map_page;
539 struct rds_page_frag *frag;
540 unsigned long frag_off;
541 unsigned long to_copy;
542 unsigned long copied;
543 uint64_t uncongested = 0;
544 void *addr;
545
546 /* catch completely corrupt packets */
547 if (be32_to_cpu(iwinc->ii_inc.i_hdr.h_len) != RDS_CONG_MAP_BYTES)
548 return;
549
550 map = conn->c_fcong;
551 map_page = 0;
552 map_off = 0;
553
554 frag = list_entry(iwinc->ii_frags.next, struct rds_page_frag, f_item);
555 frag_off = 0;
556
557 copied = 0;
558
559 while (copied < RDS_CONG_MAP_BYTES) {
560 uint64_t *src, *dst;
561 unsigned int k;
562
563 to_copy = min(RDS_FRAG_SIZE - frag_off, PAGE_SIZE - map_off);
564 BUG_ON(to_copy & 7); /* Must be 64bit aligned. */
565
566 addr = kmap_atomic(frag->f_page, KM_SOFTIRQ0);
567
568 src = addr + frag_off;
569 dst = (void *)map->m_page_addrs[map_page] + map_off;
570 for (k = 0; k < to_copy; k += 8) {
571 /* Record ports that became uncongested, ie
572 * bits that changed from 0 to 1. */
573 uncongested |= ~(*src) & *dst;
574 *dst++ = *src++;
575 }
576 kunmap_atomic(addr, KM_SOFTIRQ0);
577
578 copied += to_copy;
579
580 map_off += to_copy;
581 if (map_off == PAGE_SIZE) {
582 map_off = 0;
583 map_page++;
584 }
585
586 frag_off += to_copy;
587 if (frag_off == RDS_FRAG_SIZE) {
588 frag = list_entry(frag->f_item.next,
589 struct rds_page_frag, f_item);
590 frag_off = 0;
591 }
592 }
593
594 /* the congestion map is in little endian order */
595 uncongested = le64_to_cpu(uncongested);
596
597 rds_cong_map_updated(map, uncongested);
598}
599
600/*
601 * Rings are posted with all the allocations they'll need to queue the
602 * incoming message to the receiving socket so this can't fail.
603 * All fragments start with a header, so we can make sure we're not receiving
604 * garbage, and we can tell a small 8 byte fragment from an ACK frame.
605 */
606struct rds_iw_ack_state {
607 u64 ack_next;
608 u64 ack_recv;
609 unsigned int ack_required:1;
610 unsigned int ack_next_valid:1;
611 unsigned int ack_recv_valid:1;
612};
613
614static void rds_iw_process_recv(struct rds_connection *conn,
615 struct rds_iw_recv_work *recv, u32 byte_len,
616 struct rds_iw_ack_state *state)
617{
618 struct rds_iw_connection *ic = conn->c_transport_data;
619 struct rds_iw_incoming *iwinc = ic->i_iwinc;
620 struct rds_header *ihdr, *hdr;
621
622 /* XXX shut down the connection if port 0,0 are seen? */
623
624 rdsdebug("ic %p iwinc %p recv %p byte len %u\n", ic, iwinc, recv,
625 byte_len);
626
627 if (byte_len < sizeof(struct rds_header)) {
628 rds_iw_conn_error(conn, "incoming message "
629 "from %pI4 didn't inclue a "
630 "header, disconnecting and "
631 "reconnecting\n",
632 &conn->c_faddr);
633 return;
634 }
635 byte_len -= sizeof(struct rds_header);
636
637 ihdr = &ic->i_recv_hdrs[recv - ic->i_recvs];
638
639 /* Validate the checksum. */
640 if (!rds_message_verify_checksum(ihdr)) {
641 rds_iw_conn_error(conn, "incoming message "
642 "from %pI4 has corrupted header - "
643 "forcing a reconnect\n",
644 &conn->c_faddr);
645 rds_stats_inc(s_recv_drop_bad_checksum);
646 return;
647 }
648
649 /* Process the ACK sequence which comes with every packet */
650 state->ack_recv = be64_to_cpu(ihdr->h_ack);
651 state->ack_recv_valid = 1;
652
653 /* Process the credits update if there was one */
654 if (ihdr->h_credit)
655 rds_iw_send_add_credits(conn, ihdr->h_credit);
656
657 if (ihdr->h_sport == 0 && ihdr->h_dport == 0 && byte_len == 0) {
658 /* This is an ACK-only packet. The fact that it gets
659 * special treatment here is that historically, ACKs
660 * were rather special beasts.
661 */
662 rds_iw_stats_inc(s_iw_ack_received);
663
664 /*
665 * Usually the frags make their way on to incs and are then freed as
666 * the inc is freed. We don't go that route, so we have to drop the
667 * page ref ourselves. We can't just leave the page on the recv
668 * because that confuses the dma mapping of pages and each recv's use
669 * of a partial page. We can leave the frag, though, it will be
670 * reused.
671 *
672 * FIXME: Fold this into the code path below.
673 */
674 rds_iw_frag_drop_page(recv->r_frag);
675 return;
676 }
677
678 /*
679 * If we don't already have an inc on the connection then this
680 * fragment has a header and starts a message.. copy its header
681 * into the inc and save the inc so we can hang upcoming fragments
682 * off its list.
683 */
684 if (iwinc == NULL) {
685 iwinc = recv->r_iwinc;
686 recv->r_iwinc = NULL;
687 ic->i_iwinc = iwinc;
688
689 hdr = &iwinc->ii_inc.i_hdr;
690 memcpy(hdr, ihdr, sizeof(*hdr));
691 ic->i_recv_data_rem = be32_to_cpu(hdr->h_len);
692
693 rdsdebug("ic %p iwinc %p rem %u flag 0x%x\n", ic, iwinc,
694 ic->i_recv_data_rem, hdr->h_flags);
695 } else {
696 hdr = &iwinc->ii_inc.i_hdr;
697 /* We can't just use memcmp here; fragments of a
698 * single message may carry different ACKs */
699 if (hdr->h_sequence != ihdr->h_sequence
700 || hdr->h_len != ihdr->h_len
701 || hdr->h_sport != ihdr->h_sport
702 || hdr->h_dport != ihdr->h_dport) {
703 rds_iw_conn_error(conn,
704 "fragment header mismatch; forcing reconnect\n");
705 return;
706 }
707 }
708
709 list_add_tail(&recv->r_frag->f_item, &iwinc->ii_frags);
710 recv->r_frag = NULL;
711
712 if (ic->i_recv_data_rem > RDS_FRAG_SIZE)
713 ic->i_recv_data_rem -= RDS_FRAG_SIZE;
714 else {
715 ic->i_recv_data_rem = 0;
716 ic->i_iwinc = NULL;
717
718 if (iwinc->ii_inc.i_hdr.h_flags == RDS_FLAG_CONG_BITMAP)
719 rds_iw_cong_recv(conn, iwinc);
720 else {
721 rds_recv_incoming(conn, conn->c_faddr, conn->c_laddr,
722 &iwinc->ii_inc, GFP_ATOMIC,
723 KM_SOFTIRQ0);
724 state->ack_next = be64_to_cpu(hdr->h_sequence);
725 state->ack_next_valid = 1;
726 }
727
728 /* Evaluate the ACK_REQUIRED flag *after* we received
729 * the complete frame, and after bumping the next_rx
730 * sequence. */
731 if (hdr->h_flags & RDS_FLAG_ACK_REQUIRED) {
732 rds_stats_inc(s_recv_ack_required);
733 state->ack_required = 1;
734 }
735
736 rds_inc_put(&iwinc->ii_inc);
737 }
738}
739
740/*
741 * Plucking the oldest entry from the ring can be done concurrently with
742 * the thread refilling the ring. Each ring operation is protected by
743 * spinlocks and the transient state of refilling doesn't change the
744 * recording of which entry is oldest.
745 *
746 * This relies on IB only calling one cq comp_handler for each cq so that
747 * there will only be one caller of rds_recv_incoming() per RDS connection.
748 */
749void rds_iw_recv_cq_comp_handler(struct ib_cq *cq, void *context)
750{
751 struct rds_connection *conn = context;
752 struct rds_iw_connection *ic = conn->c_transport_data;
753 struct ib_wc wc;
754 struct rds_iw_ack_state state = { 0, };
755 struct rds_iw_recv_work *recv;
756
757 rdsdebug("conn %p cq %p\n", conn, cq);
758
759 rds_iw_stats_inc(s_iw_rx_cq_call);
760
761 ib_req_notify_cq(cq, IB_CQ_SOLICITED);
762
763 while (ib_poll_cq(cq, 1, &wc) > 0) {
764 rdsdebug("wc wr_id 0x%llx status %u byte_len %u imm_data %u\n",
765 (unsigned long long)wc.wr_id, wc.status, wc.byte_len,
766 be32_to_cpu(wc.ex.imm_data));
767 rds_iw_stats_inc(s_iw_rx_cq_event);
768
769 recv = &ic->i_recvs[rds_iw_ring_oldest(&ic->i_recv_ring)];
770
771 rds_iw_recv_unmap_page(ic, recv);
772
773 /*
774 * Also process recvs in connecting state because it is possible
775 * to get a recv completion _before_ the rdmacm ESTABLISHED
776 * event is processed.
777 */
778 if (rds_conn_up(conn) || rds_conn_connecting(conn)) {
779 /* We expect errors as the qp is drained during shutdown */
780 if (wc.status == IB_WC_SUCCESS) {
781 rds_iw_process_recv(conn, recv, wc.byte_len, &state);
782 } else {
783 rds_iw_conn_error(conn, "recv completion on "
784 "%pI4 had status %u, disconnecting and "
785 "reconnecting\n", &conn->c_faddr,
786 wc.status);
787 }
788 }
789
790 rds_iw_ring_free(&ic->i_recv_ring, 1);
791 }
792
793 if (state.ack_next_valid)
794 rds_iw_set_ack(ic, state.ack_next, state.ack_required);
795 if (state.ack_recv_valid && state.ack_recv > ic->i_ack_recv) {
796 rds_send_drop_acked(conn, state.ack_recv, NULL);
797 ic->i_ack_recv = state.ack_recv;
798 }
799 if (rds_conn_up(conn))
800 rds_iw_attempt_ack(ic);
801
802 /* If we ever end up with a really empty receive ring, we're
803 * in deep trouble, as the sender will definitely see RNR
804 * timeouts. */
805 if (rds_iw_ring_empty(&ic->i_recv_ring))
806 rds_iw_stats_inc(s_iw_rx_ring_empty);
807
808 /*
809 * If the ring is running low, then schedule the thread to refill.
810 */
811 if (rds_iw_ring_low(&ic->i_recv_ring))
812 queue_delayed_work(rds_wq, &conn->c_recv_w, 0);
813}
814
815int rds_iw_recv(struct rds_connection *conn)
816{
817 struct rds_iw_connection *ic = conn->c_transport_data;
818 int ret = 0;
819
820 rdsdebug("conn %p\n", conn);
821
822 /*
823 * If we get a temporary posting failure in this context then
824 * we're really low and we want the caller to back off for a bit.
825 */
826 mutex_lock(&ic->i_recv_mutex);
827 if (rds_iw_recv_refill(conn, GFP_KERNEL, GFP_HIGHUSER, 0))
828 ret = -ENOMEM;
829 else
830 rds_iw_stats_inc(s_iw_rx_refill_from_thread);
831 mutex_unlock(&ic->i_recv_mutex);
832
833 if (rds_conn_up(conn))
834 rds_iw_attempt_ack(ic);
835
836 return ret;
837}
838
839int __init rds_iw_recv_init(void)
840{
841 struct sysinfo si;
842 int ret = -ENOMEM;
843
844 /* Default to 30% of all available RAM for recv memory */
845 si_meminfo(&si);
846 rds_iw_sysctl_max_recv_allocation = si.totalram / 3 * PAGE_SIZE / RDS_FRAG_SIZE;
847
848 rds_iw_incoming_slab = kmem_cache_create("rds_iw_incoming",
849 sizeof(struct rds_iw_incoming),
850 0, 0, NULL);
851 if (rds_iw_incoming_slab == NULL)
852 goto out;
853
854 rds_iw_frag_slab = kmem_cache_create("rds_iw_frag",
855 sizeof(struct rds_page_frag),
856 0, 0, NULL);
857 if (rds_iw_frag_slab == NULL)
858 kmem_cache_destroy(rds_iw_incoming_slab);
859 else
860 ret = 0;
861out:
862 return ret;
863}
864
865void rds_iw_recv_exit(void)
866{
867 kmem_cache_destroy(rds_iw_incoming_slab);
868 kmem_cache_destroy(rds_iw_frag_slab);
869}
diff --git a/net/rds/iw_ring.c b/net/rds/iw_ring.c
new file mode 100644
index 000000000000..d422d4b5deef
--- /dev/null
+++ b/net/rds/iw_ring.c
@@ -0,0 +1,169 @@
1/*
2 * Copyright (c) 2006 Oracle. All rights reserved.
3 *
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
9 *
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
12 * conditions are met:
13 *
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
16 * disclaimer.
17 *
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
22 *
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30 * SOFTWARE.
31 *
32 */
33#include <linux/kernel.h>
34
35#include "rds.h"
36#include "iw.h"
37
38/*
39 * Locking for IB rings.
40 * We assume that allocation is always protected by a mutex
41 * in the caller (this is a valid assumption for the current
42 * implementation).
43 *
44 * Freeing always happens in an interrupt, and hence only
45 * races with allocations, but not with other free()s.
46 *
47 * The interaction between allocation and freeing is that
48 * the alloc code has to determine the number of free entries.
49 * To this end, we maintain two counters; an allocation counter
50 * and a free counter. Both are allowed to run freely, and wrap
51 * around.
52 * The number of used entries is always (alloc_ctr - free_ctr) % NR.
53 *
54 * The current implementation makes free_ctr atomic. When the
55 * caller finds an allocation fails, it should set an "alloc fail"
56 * bit and retry the allocation. The "alloc fail" bit essentially tells
57 * the CQ completion handlers to wake it up after freeing some
58 * more entries.
59 */
60
61/*
62 * This only happens on shutdown.
63 */
64DECLARE_WAIT_QUEUE_HEAD(rds_iw_ring_empty_wait);
65
66void rds_iw_ring_init(struct rds_iw_work_ring *ring, u32 nr)
67{
68 memset(ring, 0, sizeof(*ring));
69 ring->w_nr = nr;
70 rdsdebug("ring %p nr %u\n", ring, ring->w_nr);
71}
72
73static inline u32 __rds_iw_ring_used(struct rds_iw_work_ring *ring)
74{
75 u32 diff;
76
77 /* This assumes that atomic_t has at least as many bits as u32 */
78 diff = ring->w_alloc_ctr - (u32) atomic_read(&ring->w_free_ctr);
79 BUG_ON(diff > ring->w_nr);
80
81 return diff;
82}
83
84void rds_iw_ring_resize(struct rds_iw_work_ring *ring, u32 nr)
85{
86 /* We only ever get called from the connection setup code,
87 * prior to creating the QP. */
88 BUG_ON(__rds_iw_ring_used(ring));
89 ring->w_nr = nr;
90}
91
92static int __rds_iw_ring_empty(struct rds_iw_work_ring *ring)
93{
94 return __rds_iw_ring_used(ring) == 0;
95}
96
97u32 rds_iw_ring_alloc(struct rds_iw_work_ring *ring, u32 val, u32 *pos)
98{
99 u32 ret = 0, avail;
100
101 avail = ring->w_nr - __rds_iw_ring_used(ring);
102
103 rdsdebug("ring %p val %u next %u free %u\n", ring, val,
104 ring->w_alloc_ptr, avail);
105
106 if (val && avail) {
107 ret = min(val, avail);
108 *pos = ring->w_alloc_ptr;
109
110 ring->w_alloc_ptr = (ring->w_alloc_ptr + ret) % ring->w_nr;
111 ring->w_alloc_ctr += ret;
112 }
113
114 return ret;
115}
116
117void rds_iw_ring_free(struct rds_iw_work_ring *ring, u32 val)
118{
119 ring->w_free_ptr = (ring->w_free_ptr + val) % ring->w_nr;
120 atomic_add(val, &ring->w_free_ctr);
121
122 if (__rds_iw_ring_empty(ring) &&
123 waitqueue_active(&rds_iw_ring_empty_wait))
124 wake_up(&rds_iw_ring_empty_wait);
125}
126
127void rds_iw_ring_unalloc(struct rds_iw_work_ring *ring, u32 val)
128{
129 ring->w_alloc_ptr = (ring->w_alloc_ptr - val) % ring->w_nr;
130 ring->w_alloc_ctr -= val;
131}
132
133int rds_iw_ring_empty(struct rds_iw_work_ring *ring)
134{
135 return __rds_iw_ring_empty(ring);
136}
137
138int rds_iw_ring_low(struct rds_iw_work_ring *ring)
139{
140 return __rds_iw_ring_used(ring) <= (ring->w_nr >> 2);
141}
142
143
144/*
145 * returns the oldest alloced ring entry. This will be the next one
146 * freed. This can't be called if there are none allocated.
147 */
148u32 rds_iw_ring_oldest(struct rds_iw_work_ring *ring)
149{
150 return ring->w_free_ptr;
151}
152
153/*
154 * returns the number of completed work requests.
155 */
156
157u32 rds_iw_ring_completed(struct rds_iw_work_ring *ring, u32 wr_id, u32 oldest)
158{
159 u32 ret;
160
161 if (oldest <= (unsigned long long)wr_id)
162 ret = (unsigned long long)wr_id - oldest + 1;
163 else
164 ret = ring->w_nr - oldest + (unsigned long long)wr_id + 1;
165
166 rdsdebug("ring %p ret %u wr_id %u oldest %u\n", ring, ret,
167 wr_id, oldest);
168 return ret;
169}
diff --git a/net/rds/iw_send.c b/net/rds/iw_send.c
new file mode 100644
index 000000000000..22dd38ffd608
--- /dev/null
+++ b/net/rds/iw_send.c
@@ -0,0 +1,975 @@
1/*
2 * Copyright (c) 2006 Oracle. All rights reserved.
3 *
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
9 *
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
12 * conditions are met:
13 *
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
16 * disclaimer.
17 *
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
22 *
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30 * SOFTWARE.
31 *
32 */
33#include <linux/kernel.h>
34#include <linux/in.h>
35#include <linux/device.h>
36#include <linux/dmapool.h>
37
38#include "rds.h"
39#include "rdma.h"
40#include "iw.h"
41
42static void rds_iw_send_rdma_complete(struct rds_message *rm,
43 int wc_status)
44{
45 int notify_status;
46
47 switch (wc_status) {
48 case IB_WC_WR_FLUSH_ERR:
49 return;
50
51 case IB_WC_SUCCESS:
52 notify_status = RDS_RDMA_SUCCESS;
53 break;
54
55 case IB_WC_REM_ACCESS_ERR:
56 notify_status = RDS_RDMA_REMOTE_ERROR;
57 break;
58
59 default:
60 notify_status = RDS_RDMA_OTHER_ERROR;
61 break;
62 }
63 rds_rdma_send_complete(rm, notify_status);
64}
65
66static void rds_iw_send_unmap_rdma(struct rds_iw_connection *ic,
67 struct rds_rdma_op *op)
68{
69 if (op->r_mapped) {
70 ib_dma_unmap_sg(ic->i_cm_id->device,
71 op->r_sg, op->r_nents,
72 op->r_write ? DMA_TO_DEVICE : DMA_FROM_DEVICE);
73 op->r_mapped = 0;
74 }
75}
76
77static void rds_iw_send_unmap_rm(struct rds_iw_connection *ic,
78 struct rds_iw_send_work *send,
79 int wc_status)
80{
81 struct rds_message *rm = send->s_rm;
82
83 rdsdebug("ic %p send %p rm %p\n", ic, send, rm);
84
85 ib_dma_unmap_sg(ic->i_cm_id->device,
86 rm->m_sg, rm->m_nents,
87 DMA_TO_DEVICE);
88
89 if (rm->m_rdma_op != NULL) {
90 rds_iw_send_unmap_rdma(ic, rm->m_rdma_op);
91
92 /* If the user asked for a completion notification on this
93 * message, we can implement three different semantics:
94 * 1. Notify when we received the ACK on the RDS message
95 * that was queued with the RDMA. This provides reliable
96 * notification of RDMA status at the expense of a one-way
97 * packet delay.
98 * 2. Notify when the IB stack gives us the completion event for
99 * the RDMA operation.
100 * 3. Notify when the IB stack gives us the completion event for
101 * the accompanying RDS messages.
102 * Here, we implement approach #3. To implement approach #2,
103 * call rds_rdma_send_complete from the cq_handler. To implement #1,
104 * don't call rds_rdma_send_complete at all, and fall back to the notify
105 * handling in the ACK processing code.
106 *
107 * Note: There's no need to explicitly sync any RDMA buffers using
108 * ib_dma_sync_sg_for_cpu - the completion for the RDMA
109 * operation itself unmapped the RDMA buffers, which takes care
110 * of synching.
111 */
112 rds_iw_send_rdma_complete(rm, wc_status);
113
114 if (rm->m_rdma_op->r_write)
115 rds_stats_add(s_send_rdma_bytes, rm->m_rdma_op->r_bytes);
116 else
117 rds_stats_add(s_recv_rdma_bytes, rm->m_rdma_op->r_bytes);
118 }
119
120 /* If anyone waited for this message to get flushed out, wake
121 * them up now */
122 rds_message_unmapped(rm);
123
124 rds_message_put(rm);
125 send->s_rm = NULL;
126}
127
128void rds_iw_send_init_ring(struct rds_iw_connection *ic)
129{
130 struct rds_iw_send_work *send;
131 u32 i;
132
133 for (i = 0, send = ic->i_sends; i < ic->i_send_ring.w_nr; i++, send++) {
134 struct ib_sge *sge;
135
136 send->s_rm = NULL;
137 send->s_op = NULL;
138 send->s_mapping = NULL;
139
140 send->s_wr.next = NULL;
141 send->s_wr.wr_id = i;
142 send->s_wr.sg_list = send->s_sge;
143 send->s_wr.num_sge = 1;
144 send->s_wr.opcode = IB_WR_SEND;
145 send->s_wr.send_flags = 0;
146 send->s_wr.ex.imm_data = 0;
147
148 sge = rds_iw_data_sge(ic, send->s_sge);
149 sge->lkey = 0;
150
151 sge = rds_iw_header_sge(ic, send->s_sge);
152 sge->addr = ic->i_send_hdrs_dma + (i * sizeof(struct rds_header));
153 sge->length = sizeof(struct rds_header);
154 sge->lkey = 0;
155
156 send->s_mr = ib_alloc_fast_reg_mr(ic->i_pd, fastreg_message_size);
157 if (IS_ERR(send->s_mr)) {
158 printk(KERN_WARNING "RDS/IW: ib_alloc_fast_reg_mr failed\n");
159 break;
160 }
161
162 send->s_page_list = ib_alloc_fast_reg_page_list(
163 ic->i_cm_id->device, fastreg_message_size);
164 if (IS_ERR(send->s_page_list)) {
165 printk(KERN_WARNING "RDS/IW: ib_alloc_fast_reg_page_list failed\n");
166 break;
167 }
168 }
169}
170
171void rds_iw_send_clear_ring(struct rds_iw_connection *ic)
172{
173 struct rds_iw_send_work *send;
174 u32 i;
175
176 for (i = 0, send = ic->i_sends; i < ic->i_send_ring.w_nr; i++, send++) {
177 BUG_ON(!send->s_mr);
178 ib_dereg_mr(send->s_mr);
179 BUG_ON(!send->s_page_list);
180 ib_free_fast_reg_page_list(send->s_page_list);
181 if (send->s_wr.opcode == 0xdead)
182 continue;
183 if (send->s_rm)
184 rds_iw_send_unmap_rm(ic, send, IB_WC_WR_FLUSH_ERR);
185 if (send->s_op)
186 rds_iw_send_unmap_rdma(ic, send->s_op);
187 }
188}
189
190/*
191 * The _oldest/_free ring operations here race cleanly with the alloc/unalloc
192 * operations performed in the send path. As the sender allocs and potentially
193 * unallocs the next free entry in the ring it doesn't alter which is
194 * the next to be freed, which is what this is concerned with.
195 */
196void rds_iw_send_cq_comp_handler(struct ib_cq *cq, void *context)
197{
198 struct rds_connection *conn = context;
199 struct rds_iw_connection *ic = conn->c_transport_data;
200 struct ib_wc wc;
201 struct rds_iw_send_work *send;
202 u32 completed;
203 u32 oldest;
204 u32 i;
205 int ret;
206
207 rdsdebug("cq %p conn %p\n", cq, conn);
208 rds_iw_stats_inc(s_iw_tx_cq_call);
209 ret = ib_req_notify_cq(cq, IB_CQ_NEXT_COMP);
210 if (ret)
211 rdsdebug("ib_req_notify_cq send failed: %d\n", ret);
212
213 while (ib_poll_cq(cq, 1, &wc) > 0) {
214 rdsdebug("wc wr_id 0x%llx status %u byte_len %u imm_data %u\n",
215 (unsigned long long)wc.wr_id, wc.status, wc.byte_len,
216 be32_to_cpu(wc.ex.imm_data));
217 rds_iw_stats_inc(s_iw_tx_cq_event);
218
219 if (wc.status != IB_WC_SUCCESS) {
220 printk(KERN_ERR "WC Error: status = %d opcode = %d\n", wc.status, wc.opcode);
221 break;
222 }
223
224 if (wc.opcode == IB_WC_LOCAL_INV && wc.wr_id == RDS_IW_LOCAL_INV_WR_ID) {
225 ic->i_fastreg_posted = 0;
226 continue;
227 }
228
229 if (wc.opcode == IB_WC_FAST_REG_MR && wc.wr_id == RDS_IW_FAST_REG_WR_ID) {
230 ic->i_fastreg_posted = 1;
231 continue;
232 }
233
234 if (wc.wr_id == RDS_IW_ACK_WR_ID) {
235 if (ic->i_ack_queued + HZ/2 < jiffies)
236 rds_iw_stats_inc(s_iw_tx_stalled);
237 rds_iw_ack_send_complete(ic);
238 continue;
239 }
240
241 oldest = rds_iw_ring_oldest(&ic->i_send_ring);
242
243 completed = rds_iw_ring_completed(&ic->i_send_ring, wc.wr_id, oldest);
244
245 for (i = 0; i < completed; i++) {
246 send = &ic->i_sends[oldest];
247
248 /* In the error case, wc.opcode sometimes contains garbage */
249 switch (send->s_wr.opcode) {
250 case IB_WR_SEND:
251 if (send->s_rm)
252 rds_iw_send_unmap_rm(ic, send, wc.status);
253 break;
254 case IB_WR_FAST_REG_MR:
255 case IB_WR_RDMA_WRITE:
256 case IB_WR_RDMA_READ:
257 case IB_WR_RDMA_READ_WITH_INV:
258 /* Nothing to be done - the SG list will be unmapped
259 * when the SEND completes. */
260 break;
261 default:
262 if (printk_ratelimit())
263 printk(KERN_NOTICE
264 "RDS/IW: %s: unexpected opcode 0x%x in WR!\n",
265 __func__, send->s_wr.opcode);
266 break;
267 }
268
269 send->s_wr.opcode = 0xdead;
270 send->s_wr.num_sge = 1;
271 if (send->s_queued + HZ/2 < jiffies)
272 rds_iw_stats_inc(s_iw_tx_stalled);
273
274 /* If a RDMA operation produced an error, signal this right
275 * away. If we don't, the subsequent SEND that goes with this
276 * RDMA will be canceled with ERR_WFLUSH, and the application
277 * never learn that the RDMA failed. */
278 if (unlikely(wc.status == IB_WC_REM_ACCESS_ERR && send->s_op)) {
279 struct rds_message *rm;
280
281 rm = rds_send_get_message(conn, send->s_op);
282 if (rm)
283 rds_iw_send_rdma_complete(rm, wc.status);
284 }
285
286 oldest = (oldest + 1) % ic->i_send_ring.w_nr;
287 }
288
289 rds_iw_ring_free(&ic->i_send_ring, completed);
290
291 if (test_and_clear_bit(RDS_LL_SEND_FULL, &conn->c_flags)
292 || test_bit(0, &conn->c_map_queued))
293 queue_delayed_work(rds_wq, &conn->c_send_w, 0);
294
295 /* We expect errors as the qp is drained during shutdown */
296 if (wc.status != IB_WC_SUCCESS && rds_conn_up(conn)) {
297 rds_iw_conn_error(conn,
298 "send completion on %pI4 "
299 "had status %u, disconnecting and reconnecting\n",
300 &conn->c_faddr, wc.status);
301 }
302 }
303}
304
305/*
306 * This is the main function for allocating credits when sending
307 * messages.
308 *
309 * Conceptually, we have two counters:
310 * - send credits: this tells us how many WRs we're allowed
311 * to submit without overruning the reciever's queue. For
312 * each SEND WR we post, we decrement this by one.
313 *
314 * - posted credits: this tells us how many WRs we recently
315 * posted to the receive queue. This value is transferred
316 * to the peer as a "credit update" in a RDS header field.
317 * Every time we transmit credits to the peer, we subtract
318 * the amount of transferred credits from this counter.
319 *
320 * It is essential that we avoid situations where both sides have
321 * exhausted their send credits, and are unable to send new credits
322 * to the peer. We achieve this by requiring that we send at least
323 * one credit update to the peer before exhausting our credits.
324 * When new credits arrive, we subtract one credit that is withheld
325 * until we've posted new buffers and are ready to transmit these
326 * credits (see rds_iw_send_add_credits below).
327 *
328 * The RDS send code is essentially single-threaded; rds_send_xmit
329 * grabs c_send_lock to ensure exclusive access to the send ring.
330 * However, the ACK sending code is independent and can race with
331 * message SENDs.
332 *
333 * In the send path, we need to update the counters for send credits
334 * and the counter of posted buffers atomically - when we use the
335 * last available credit, we cannot allow another thread to race us
336 * and grab the posted credits counter. Hence, we have to use a
337 * spinlock to protect the credit counter, or use atomics.
338 *
339 * Spinlocks shared between the send and the receive path are bad,
340 * because they create unnecessary delays. An early implementation
341 * using a spinlock showed a 5% degradation in throughput at some
342 * loads.
343 *
344 * This implementation avoids spinlocks completely, putting both
345 * counters into a single atomic, and updating that atomic using
346 * atomic_add (in the receive path, when receiving fresh credits),
347 * and using atomic_cmpxchg when updating the two counters.
348 */
349int rds_iw_send_grab_credits(struct rds_iw_connection *ic,
350 u32 wanted, u32 *adv_credits, int need_posted)
351{
352 unsigned int avail, posted, got = 0, advertise;
353 long oldval, newval;
354
355 *adv_credits = 0;
356 if (!ic->i_flowctl)
357 return wanted;
358
359try_again:
360 advertise = 0;
361 oldval = newval = atomic_read(&ic->i_credits);
362 posted = IB_GET_POST_CREDITS(oldval);
363 avail = IB_GET_SEND_CREDITS(oldval);
364
365 rdsdebug("rds_iw_send_grab_credits(%u): credits=%u posted=%u\n",
366 wanted, avail, posted);
367
368 /* The last credit must be used to send a credit update. */
369 if (avail && !posted)
370 avail--;
371
372 if (avail < wanted) {
373 struct rds_connection *conn = ic->i_cm_id->context;
374
375 /* Oops, there aren't that many credits left! */
376 set_bit(RDS_LL_SEND_FULL, &conn->c_flags);
377 got = avail;
378 } else {
379 /* Sometimes you get what you want, lalala. */
380 got = wanted;
381 }
382 newval -= IB_SET_SEND_CREDITS(got);
383
384 /*
385 * If need_posted is non-zero, then the caller wants
386 * the posted regardless of whether any send credits are
387 * available.
388 */
389 if (posted && (got || need_posted)) {
390 advertise = min_t(unsigned int, posted, RDS_MAX_ADV_CREDIT);
391 newval -= IB_SET_POST_CREDITS(advertise);
392 }
393
394 /* Finally bill everything */
395 if (atomic_cmpxchg(&ic->i_credits, oldval, newval) != oldval)
396 goto try_again;
397
398 *adv_credits = advertise;
399 return got;
400}
401
402void rds_iw_send_add_credits(struct rds_connection *conn, unsigned int credits)
403{
404 struct rds_iw_connection *ic = conn->c_transport_data;
405
406 if (credits == 0)
407 return;
408
409 rdsdebug("rds_iw_send_add_credits(%u): current=%u%s\n",
410 credits,
411 IB_GET_SEND_CREDITS(atomic_read(&ic->i_credits)),
412 test_bit(RDS_LL_SEND_FULL, &conn->c_flags) ? ", ll_send_full" : "");
413
414 atomic_add(IB_SET_SEND_CREDITS(credits), &ic->i_credits);
415 if (test_and_clear_bit(RDS_LL_SEND_FULL, &conn->c_flags))
416 queue_delayed_work(rds_wq, &conn->c_send_w, 0);
417
418 WARN_ON(IB_GET_SEND_CREDITS(credits) >= 16384);
419
420 rds_iw_stats_inc(s_iw_rx_credit_updates);
421}
422
423void rds_iw_advertise_credits(struct rds_connection *conn, unsigned int posted)
424{
425 struct rds_iw_connection *ic = conn->c_transport_data;
426
427 if (posted == 0)
428 return;
429
430 atomic_add(IB_SET_POST_CREDITS(posted), &ic->i_credits);
431
432 /* Decide whether to send an update to the peer now.
433 * If we would send a credit update for every single buffer we
434 * post, we would end up with an ACK storm (ACK arrives,
435 * consumes buffer, we refill the ring, send ACK to remote
436 * advertising the newly posted buffer... ad inf)
437 *
438 * Performance pretty much depends on how often we send
439 * credit updates - too frequent updates mean lots of ACKs.
440 * Too infrequent updates, and the peer will run out of
441 * credits and has to throttle.
442 * For the time being, 16 seems to be a good compromise.
443 */
444 if (IB_GET_POST_CREDITS(atomic_read(&ic->i_credits)) >= 16)
445 set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
446}
447
448static inline void
449rds_iw_xmit_populate_wr(struct rds_iw_connection *ic,
450 struct rds_iw_send_work *send, unsigned int pos,
451 unsigned long buffer, unsigned int length,
452 int send_flags)
453{
454 struct ib_sge *sge;
455
456 WARN_ON(pos != send - ic->i_sends);
457
458 send->s_wr.send_flags = send_flags;
459 send->s_wr.opcode = IB_WR_SEND;
460 send->s_wr.num_sge = 2;
461 send->s_wr.next = NULL;
462 send->s_queued = jiffies;
463 send->s_op = NULL;
464
465 if (length != 0) {
466 sge = rds_iw_data_sge(ic, send->s_sge);
467 sge->addr = buffer;
468 sge->length = length;
469 sge->lkey = rds_iw_local_dma_lkey(ic);
470
471 sge = rds_iw_header_sge(ic, send->s_sge);
472 } else {
473 /* We're sending a packet with no payload. There is only
474 * one SGE */
475 send->s_wr.num_sge = 1;
476 sge = &send->s_sge[0];
477 }
478
479 sge->addr = ic->i_send_hdrs_dma + (pos * sizeof(struct rds_header));
480 sge->length = sizeof(struct rds_header);
481 sge->lkey = rds_iw_local_dma_lkey(ic);
482}
483
484/*
485 * This can be called multiple times for a given message. The first time
486 * we see a message we map its scatterlist into the IB device so that
487 * we can provide that mapped address to the IB scatter gather entries
488 * in the IB work requests. We translate the scatterlist into a series
489 * of work requests that fragment the message. These work requests complete
490 * in order so we pass ownership of the message to the completion handler
491 * once we send the final fragment.
492 *
493 * The RDS core uses the c_send_lock to only enter this function once
494 * per connection. This makes sure that the tx ring alloc/unalloc pairs
495 * don't get out of sync and confuse the ring.
496 */
497int rds_iw_xmit(struct rds_connection *conn, struct rds_message *rm,
498 unsigned int hdr_off, unsigned int sg, unsigned int off)
499{
500 struct rds_iw_connection *ic = conn->c_transport_data;
501 struct ib_device *dev = ic->i_cm_id->device;
502 struct rds_iw_send_work *send = NULL;
503 struct rds_iw_send_work *first;
504 struct rds_iw_send_work *prev;
505 struct ib_send_wr *failed_wr;
506 struct scatterlist *scat;
507 u32 pos;
508 u32 i;
509 u32 work_alloc;
510 u32 credit_alloc;
511 u32 posted;
512 u32 adv_credits = 0;
513 int send_flags = 0;
514 int sent;
515 int ret;
516 int flow_controlled = 0;
517
518 BUG_ON(off % RDS_FRAG_SIZE);
519 BUG_ON(hdr_off != 0 && hdr_off != sizeof(struct rds_header));
520
521 /* Fastreg support */
522 if (rds_rdma_cookie_key(rm->m_rdma_cookie)
523 && !ic->i_fastreg_posted) {
524 ret = -EAGAIN;
525 goto out;
526 }
527
528 /* FIXME we may overallocate here */
529 if (be32_to_cpu(rm->m_inc.i_hdr.h_len) == 0)
530 i = 1;
531 else
532 i = ceil(be32_to_cpu(rm->m_inc.i_hdr.h_len), RDS_FRAG_SIZE);
533
534 work_alloc = rds_iw_ring_alloc(&ic->i_send_ring, i, &pos);
535 if (work_alloc == 0) {
536 set_bit(RDS_LL_SEND_FULL, &conn->c_flags);
537 rds_iw_stats_inc(s_iw_tx_ring_full);
538 ret = -ENOMEM;
539 goto out;
540 }
541
542 credit_alloc = work_alloc;
543 if (ic->i_flowctl) {
544 credit_alloc = rds_iw_send_grab_credits(ic, work_alloc, &posted, 0);
545 adv_credits += posted;
546 if (credit_alloc < work_alloc) {
547 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc - credit_alloc);
548 work_alloc = credit_alloc;
549 flow_controlled++;
550 }
551 if (work_alloc == 0) {
552 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
553 rds_iw_stats_inc(s_iw_tx_throttle);
554 ret = -ENOMEM;
555 goto out;
556 }
557 }
558
559 /* map the message the first time we see it */
560 if (ic->i_rm == NULL) {
561 /*
562 printk(KERN_NOTICE "rds_iw_xmit prep msg dport=%u flags=0x%x len=%d\n",
563 be16_to_cpu(rm->m_inc.i_hdr.h_dport),
564 rm->m_inc.i_hdr.h_flags,
565 be32_to_cpu(rm->m_inc.i_hdr.h_len));
566 */
567 if (rm->m_nents) {
568 rm->m_count = ib_dma_map_sg(dev,
569 rm->m_sg, rm->m_nents, DMA_TO_DEVICE);
570 rdsdebug("ic %p mapping rm %p: %d\n", ic, rm, rm->m_count);
571 if (rm->m_count == 0) {
572 rds_iw_stats_inc(s_iw_tx_sg_mapping_failure);
573 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
574 ret = -ENOMEM; /* XXX ? */
575 goto out;
576 }
577 } else {
578 rm->m_count = 0;
579 }
580
581 ic->i_unsignaled_wrs = rds_iw_sysctl_max_unsig_wrs;
582 ic->i_unsignaled_bytes = rds_iw_sysctl_max_unsig_bytes;
583 rds_message_addref(rm);
584 ic->i_rm = rm;
585
586 /* Finalize the header */
587 if (test_bit(RDS_MSG_ACK_REQUIRED, &rm->m_flags))
588 rm->m_inc.i_hdr.h_flags |= RDS_FLAG_ACK_REQUIRED;
589 if (test_bit(RDS_MSG_RETRANSMITTED, &rm->m_flags))
590 rm->m_inc.i_hdr.h_flags |= RDS_FLAG_RETRANSMITTED;
591
592 /* If it has a RDMA op, tell the peer we did it. This is
593 * used by the peer to release use-once RDMA MRs. */
594 if (rm->m_rdma_op) {
595 struct rds_ext_header_rdma ext_hdr;
596
597 ext_hdr.h_rdma_rkey = cpu_to_be32(rm->m_rdma_op->r_key);
598 rds_message_add_extension(&rm->m_inc.i_hdr,
599 RDS_EXTHDR_RDMA, &ext_hdr, sizeof(ext_hdr));
600 }
601 if (rm->m_rdma_cookie) {
602 rds_message_add_rdma_dest_extension(&rm->m_inc.i_hdr,
603 rds_rdma_cookie_key(rm->m_rdma_cookie),
604 rds_rdma_cookie_offset(rm->m_rdma_cookie));
605 }
606
607 /* Note - rds_iw_piggyb_ack clears the ACK_REQUIRED bit, so
608 * we should not do this unless we have a chance of at least
609 * sticking the header into the send ring. Which is why we
610 * should call rds_iw_ring_alloc first. */
611 rm->m_inc.i_hdr.h_ack = cpu_to_be64(rds_iw_piggyb_ack(ic));
612 rds_message_make_checksum(&rm->m_inc.i_hdr);
613
614 /*
615 * Update adv_credits since we reset the ACK_REQUIRED bit.
616 */
617 rds_iw_send_grab_credits(ic, 0, &posted, 1);
618 adv_credits += posted;
619 BUG_ON(adv_credits > 255);
620 } else if (ic->i_rm != rm)
621 BUG();
622
623 send = &ic->i_sends[pos];
624 first = send;
625 prev = NULL;
626 scat = &rm->m_sg[sg];
627 sent = 0;
628 i = 0;
629
630 /* Sometimes you want to put a fence between an RDMA
631 * READ and the following SEND.
632 * We could either do this all the time
633 * or when requested by the user. Right now, we let
634 * the application choose.
635 */
636 if (rm->m_rdma_op && rm->m_rdma_op->r_fence)
637 send_flags = IB_SEND_FENCE;
638
639 /*
640 * We could be copying the header into the unused tail of the page.
641 * That would need to be changed in the future when those pages might
642 * be mapped userspace pages or page cache pages. So instead we always
643 * use a second sge and our long-lived ring of mapped headers. We send
644 * the header after the data so that the data payload can be aligned on
645 * the receiver.
646 */
647
648 /* handle a 0-len message */
649 if (be32_to_cpu(rm->m_inc.i_hdr.h_len) == 0) {
650 rds_iw_xmit_populate_wr(ic, send, pos, 0, 0, send_flags);
651 goto add_header;
652 }
653
654 /* if there's data reference it with a chain of work reqs */
655 for (; i < work_alloc && scat != &rm->m_sg[rm->m_count]; i++) {
656 unsigned int len;
657
658 send = &ic->i_sends[pos];
659
660 len = min(RDS_FRAG_SIZE, ib_sg_dma_len(dev, scat) - off);
661 rds_iw_xmit_populate_wr(ic, send, pos,
662 ib_sg_dma_address(dev, scat) + off, len,
663 send_flags);
664
665 /*
666 * We want to delay signaling completions just enough to get
667 * the batching benefits but not so much that we create dead time
668 * on the wire.
669 */
670 if (ic->i_unsignaled_wrs-- == 0) {
671 ic->i_unsignaled_wrs = rds_iw_sysctl_max_unsig_wrs;
672 send->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
673 }
674
675 ic->i_unsignaled_bytes -= len;
676 if (ic->i_unsignaled_bytes <= 0) {
677 ic->i_unsignaled_bytes = rds_iw_sysctl_max_unsig_bytes;
678 send->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
679 }
680
681 /*
682 * Always signal the last one if we're stopping due to flow control.
683 */
684 if (flow_controlled && i == (work_alloc-1))
685 send->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
686
687 rdsdebug("send %p wr %p num_sge %u next %p\n", send,
688 &send->s_wr, send->s_wr.num_sge, send->s_wr.next);
689
690 sent += len;
691 off += len;
692 if (off == ib_sg_dma_len(dev, scat)) {
693 scat++;
694 off = 0;
695 }
696
697add_header:
698 /* Tack on the header after the data. The header SGE should already
699 * have been set up to point to the right header buffer. */
700 memcpy(&ic->i_send_hdrs[pos], &rm->m_inc.i_hdr, sizeof(struct rds_header));
701
702 if (0) {
703 struct rds_header *hdr = &ic->i_send_hdrs[pos];
704
705 printk(KERN_NOTICE "send WR dport=%u flags=0x%x len=%d\n",
706 be16_to_cpu(hdr->h_dport),
707 hdr->h_flags,
708 be32_to_cpu(hdr->h_len));
709 }
710 if (adv_credits) {
711 struct rds_header *hdr = &ic->i_send_hdrs[pos];
712
713 /* add credit and redo the header checksum */
714 hdr->h_credit = adv_credits;
715 rds_message_make_checksum(hdr);
716 adv_credits = 0;
717 rds_iw_stats_inc(s_iw_tx_credit_updates);
718 }
719
720 if (prev)
721 prev->s_wr.next = &send->s_wr;
722 prev = send;
723
724 pos = (pos + 1) % ic->i_send_ring.w_nr;
725 }
726
727 /* Account the RDS header in the number of bytes we sent, but just once.
728 * The caller has no concept of fragmentation. */
729 if (hdr_off == 0)
730 sent += sizeof(struct rds_header);
731
732 /* if we finished the message then send completion owns it */
733 if (scat == &rm->m_sg[rm->m_count]) {
734 prev->s_rm = ic->i_rm;
735 prev->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
736 ic->i_rm = NULL;
737 }
738
739 if (i < work_alloc) {
740 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc - i);
741 work_alloc = i;
742 }
743 if (ic->i_flowctl && i < credit_alloc)
744 rds_iw_send_add_credits(conn, credit_alloc - i);
745
746 /* XXX need to worry about failed_wr and partial sends. */
747 failed_wr = &first->s_wr;
748 ret = ib_post_send(ic->i_cm_id->qp, &first->s_wr, &failed_wr);
749 rdsdebug("ic %p first %p (wr %p) ret %d wr %p\n", ic,
750 first, &first->s_wr, ret, failed_wr);
751 BUG_ON(failed_wr != &first->s_wr);
752 if (ret) {
753 printk(KERN_WARNING "RDS/IW: ib_post_send to %pI4 "
754 "returned %d\n", &conn->c_faddr, ret);
755 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
756 if (prev->s_rm) {
757 ic->i_rm = prev->s_rm;
758 prev->s_rm = NULL;
759 }
760 goto out;
761 }
762
763 ret = sent;
764out:
765 BUG_ON(adv_credits);
766 return ret;
767}
768
769static void rds_iw_build_send_fastreg(struct rds_iw_device *rds_iwdev, struct rds_iw_connection *ic, struct rds_iw_send_work *send, int nent, int len, u64 sg_addr)
770{
771 BUG_ON(nent > send->s_page_list->max_page_list_len);
772 /*
773 * Perform a WR for the fast_reg_mr. Each individual page
774 * in the sg list is added to the fast reg page list and placed
775 * inside the fast_reg_mr WR.
776 */
777 send->s_wr.opcode = IB_WR_FAST_REG_MR;
778 send->s_wr.wr.fast_reg.length = len;
779 send->s_wr.wr.fast_reg.rkey = send->s_mr->rkey;
780 send->s_wr.wr.fast_reg.page_list = send->s_page_list;
781 send->s_wr.wr.fast_reg.page_list_len = nent;
782 send->s_wr.wr.fast_reg.page_shift = rds_iwdev->page_shift;
783 send->s_wr.wr.fast_reg.access_flags = IB_ACCESS_REMOTE_WRITE;
784 send->s_wr.wr.fast_reg.iova_start = sg_addr;
785
786 ib_update_fast_reg_key(send->s_mr, send->s_remap_count++);
787}
788
789int rds_iw_xmit_rdma(struct rds_connection *conn, struct rds_rdma_op *op)
790{
791 struct rds_iw_connection *ic = conn->c_transport_data;
792 struct rds_iw_send_work *send = NULL;
793 struct rds_iw_send_work *first;
794 struct rds_iw_send_work *prev;
795 struct ib_send_wr *failed_wr;
796 struct rds_iw_device *rds_iwdev;
797 struct scatterlist *scat;
798 unsigned long len;
799 u64 remote_addr = op->r_remote_addr;
800 u32 pos, fr_pos;
801 u32 work_alloc;
802 u32 i;
803 u32 j;
804 int sent;
805 int ret;
806 int num_sge;
807
808 rds_iwdev = ib_get_client_data(ic->i_cm_id->device, &rds_iw_client);
809
810 /* map the message the first time we see it */
811 if (!op->r_mapped) {
812 op->r_count = ib_dma_map_sg(ic->i_cm_id->device,
813 op->r_sg, op->r_nents, (op->r_write) ?
814 DMA_TO_DEVICE : DMA_FROM_DEVICE);
815 rdsdebug("ic %p mapping op %p: %d\n", ic, op, op->r_count);
816 if (op->r_count == 0) {
817 rds_iw_stats_inc(s_iw_tx_sg_mapping_failure);
818 ret = -ENOMEM; /* XXX ? */
819 goto out;
820 }
821
822 op->r_mapped = 1;
823 }
824
825 if (!op->r_write) {
826 /* Alloc space on the send queue for the fastreg */
827 work_alloc = rds_iw_ring_alloc(&ic->i_send_ring, 1, &fr_pos);
828 if (work_alloc != 1) {
829 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
830 rds_iw_stats_inc(s_iw_tx_ring_full);
831 ret = -ENOMEM;
832 goto out;
833 }
834 }
835
836 /*
837 * Instead of knowing how to return a partial rdma read/write we insist that there
838 * be enough work requests to send the entire message.
839 */
840 i = ceil(op->r_count, rds_iwdev->max_sge);
841
842 work_alloc = rds_iw_ring_alloc(&ic->i_send_ring, i, &pos);
843 if (work_alloc != i) {
844 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
845 rds_iw_stats_inc(s_iw_tx_ring_full);
846 ret = -ENOMEM;
847 goto out;
848 }
849
850 send = &ic->i_sends[pos];
851 if (!op->r_write) {
852 first = prev = &ic->i_sends[fr_pos];
853 } else {
854 first = send;
855 prev = NULL;
856 }
857 scat = &op->r_sg[0];
858 sent = 0;
859 num_sge = op->r_count;
860
861 for (i = 0; i < work_alloc && scat != &op->r_sg[op->r_count]; i++) {
862 send->s_wr.send_flags = 0;
863 send->s_queued = jiffies;
864
865 /*
866 * We want to delay signaling completions just enough to get
867 * the batching benefits but not so much that we create dead time on the wire.
868 */
869 if (ic->i_unsignaled_wrs-- == 0) {
870 ic->i_unsignaled_wrs = rds_iw_sysctl_max_unsig_wrs;
871 send->s_wr.send_flags = IB_SEND_SIGNALED;
872 }
873
874 /* To avoid the need to have the plumbing to invalidate the fastreg_mr used
875 * for local access after RDS is finished with it, using
876 * IB_WR_RDMA_READ_WITH_INV will invalidate it after the read has completed.
877 */
878 if (op->r_write)
879 send->s_wr.opcode = IB_WR_RDMA_WRITE;
880 else
881 send->s_wr.opcode = IB_WR_RDMA_READ_WITH_INV;
882
883 send->s_wr.wr.rdma.remote_addr = remote_addr;
884 send->s_wr.wr.rdma.rkey = op->r_key;
885 send->s_op = op;
886
887 if (num_sge > rds_iwdev->max_sge) {
888 send->s_wr.num_sge = rds_iwdev->max_sge;
889 num_sge -= rds_iwdev->max_sge;
890 } else
891 send->s_wr.num_sge = num_sge;
892
893 send->s_wr.next = NULL;
894
895 if (prev)
896 prev->s_wr.next = &send->s_wr;
897
898 for (j = 0; j < send->s_wr.num_sge && scat != &op->r_sg[op->r_count]; j++) {
899 len = ib_sg_dma_len(ic->i_cm_id->device, scat);
900
901 if (send->s_wr.opcode == IB_WR_RDMA_READ_WITH_INV)
902 send->s_page_list->page_list[j] = ib_sg_dma_address(ic->i_cm_id->device, scat);
903 else {
904 send->s_sge[j].addr = ib_sg_dma_address(ic->i_cm_id->device, scat);
905 send->s_sge[j].length = len;
906 send->s_sge[j].lkey = rds_iw_local_dma_lkey(ic);
907 }
908
909 sent += len;
910 rdsdebug("ic %p sent %d remote_addr %llu\n", ic, sent, remote_addr);
911 remote_addr += len;
912
913 scat++;
914 }
915
916 if (send->s_wr.opcode == IB_WR_RDMA_READ_WITH_INV) {
917 send->s_wr.num_sge = 1;
918 send->s_sge[0].addr = conn->c_xmit_rm->m_rs->rs_user_addr;
919 send->s_sge[0].length = conn->c_xmit_rm->m_rs->rs_user_bytes;
920 send->s_sge[0].lkey = ic->i_sends[fr_pos].s_mr->lkey;
921 }
922
923 rdsdebug("send %p wr %p num_sge %u next %p\n", send,
924 &send->s_wr, send->s_wr.num_sge, send->s_wr.next);
925
926 prev = send;
927 if (++send == &ic->i_sends[ic->i_send_ring.w_nr])
928 send = ic->i_sends;
929 }
930
931 /* if we finished the message then send completion owns it */
932 if (scat == &op->r_sg[op->r_count])
933 first->s_wr.send_flags = IB_SEND_SIGNALED;
934
935 if (i < work_alloc) {
936 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc - i);
937 work_alloc = i;
938 }
939
940 /* On iWARP, local memory access by a remote system (ie, RDMA Read) is not
941 * recommended. Putting the lkey on the wire is a security hole, as it can
942 * allow for memory access to all of memory on the remote system. Some
943 * adapters do not allow using the lkey for this at all. To bypass this use a
944 * fastreg_mr (or possibly a dma_mr)
945 */
946 if (!op->r_write) {
947 rds_iw_build_send_fastreg(rds_iwdev, ic, &ic->i_sends[fr_pos],
948 op->r_count, sent, conn->c_xmit_rm->m_rs->rs_user_addr);
949 work_alloc++;
950 }
951
952 failed_wr = &first->s_wr;
953 ret = ib_post_send(ic->i_cm_id->qp, &first->s_wr, &failed_wr);
954 rdsdebug("ic %p first %p (wr %p) ret %d wr %p\n", ic,
955 first, &first->s_wr, ret, failed_wr);
956 BUG_ON(failed_wr != &first->s_wr);
957 if (ret) {
958 printk(KERN_WARNING "RDS/IW: rdma ib_post_send to %pI4 "
959 "returned %d\n", &conn->c_faddr, ret);
960 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
961 goto out;
962 }
963
964out:
965 return ret;
966}
967
968void rds_iw_xmit_complete(struct rds_connection *conn)
969{
970 struct rds_iw_connection *ic = conn->c_transport_data;
971
972 /* We may have a pending ACK or window update we were unable
973 * to send previously (due to flow control). Try again. */
974 rds_iw_attempt_ack(ic);
975}
diff --git a/net/rds/iw_stats.c b/net/rds/iw_stats.c
new file mode 100644
index 000000000000..ccc7e8f0bf0e
--- /dev/null
+++ b/net/rds/iw_stats.c
@@ -0,0 +1,95 @@
1/*
2 * Copyright (c) 2006 Oracle. All rights reserved.
3 *
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
9 *
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
12 * conditions are met:
13 *
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
16 * disclaimer.
17 *
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
22 *
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30 * SOFTWARE.
31 *
32 */
33#include <linux/percpu.h>
34#include <linux/seq_file.h>
35#include <linux/proc_fs.h>
36
37#include "rds.h"
38#include "iw.h"
39
40DEFINE_PER_CPU(struct rds_iw_statistics, rds_iw_stats) ____cacheline_aligned;
41
42static char *rds_iw_stat_names[] = {
43 "iw_connect_raced",
44 "iw_listen_closed_stale",
45 "iw_tx_cq_call",
46 "iw_tx_cq_event",
47 "iw_tx_ring_full",
48 "iw_tx_throttle",
49 "iw_tx_sg_mapping_failure",
50 "iw_tx_stalled",
51 "iw_tx_credit_updates",
52 "iw_rx_cq_call",
53 "iw_rx_cq_event",
54 "iw_rx_ring_empty",
55 "iw_rx_refill_from_cq",
56 "iw_rx_refill_from_thread",
57 "iw_rx_alloc_limit",
58 "iw_rx_credit_updates",
59 "iw_ack_sent",
60 "iw_ack_send_failure",
61 "iw_ack_send_delayed",
62 "iw_ack_send_piggybacked",
63 "iw_ack_received",
64 "iw_rdma_mr_alloc",
65 "iw_rdma_mr_free",
66 "iw_rdma_mr_used",
67 "iw_rdma_mr_pool_flush",
68 "iw_rdma_mr_pool_wait",
69 "iw_rdma_mr_pool_depleted",
70};
71
72unsigned int rds_iw_stats_info_copy(struct rds_info_iterator *iter,
73 unsigned int avail)
74{
75 struct rds_iw_statistics stats = {0, };
76 uint64_t *src;
77 uint64_t *sum;
78 size_t i;
79 int cpu;
80
81 if (avail < ARRAY_SIZE(rds_iw_stat_names))
82 goto out;
83
84 for_each_online_cpu(cpu) {
85 src = (uint64_t *)&(per_cpu(rds_iw_stats, cpu));
86 sum = (uint64_t *)&stats;
87 for (i = 0; i < sizeof(stats) / sizeof(uint64_t); i++)
88 *(sum++) += *(src++);
89 }
90
91 rds_stats_info_copy(iter, (uint64_t *)&stats, rds_iw_stat_names,
92 ARRAY_SIZE(rds_iw_stat_names));
93out:
94 return ARRAY_SIZE(rds_iw_stat_names);
95}
diff --git a/net/rds/iw_sysctl.c b/net/rds/iw_sysctl.c
new file mode 100644
index 000000000000..9590678cd616
--- /dev/null
+++ b/net/rds/iw_sysctl.c
@@ -0,0 +1,137 @@
1/*
2 * Copyright (c) 2006 Oracle. All rights reserved.
3 *
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
9 *
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
12 * conditions are met:
13 *
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
16 * disclaimer.
17 *
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
22 *
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30 * SOFTWARE.
31 *
32 */
33#include <linux/kernel.h>
34#include <linux/sysctl.h>
35#include <linux/proc_fs.h>
36
37#include "iw.h"
38
39static struct ctl_table_header *rds_iw_sysctl_hdr;
40
41unsigned long rds_iw_sysctl_max_send_wr = RDS_IW_DEFAULT_SEND_WR;
42unsigned long rds_iw_sysctl_max_recv_wr = RDS_IW_DEFAULT_RECV_WR;
43unsigned long rds_iw_sysctl_max_recv_allocation = (128 * 1024 * 1024) / RDS_FRAG_SIZE;
44static unsigned long rds_iw_sysctl_max_wr_min = 1;
45/* hardware will fail CQ creation long before this */
46static unsigned long rds_iw_sysctl_max_wr_max = (u32)~0;
47
48unsigned long rds_iw_sysctl_max_unsig_wrs = 16;
49static unsigned long rds_iw_sysctl_max_unsig_wr_min = 1;
50static unsigned long rds_iw_sysctl_max_unsig_wr_max = 64;
51
52unsigned long rds_iw_sysctl_max_unsig_bytes = (16 << 20);
53static unsigned long rds_iw_sysctl_max_unsig_bytes_min = 1;
54static unsigned long rds_iw_sysctl_max_unsig_bytes_max = ~0UL;
55
56unsigned int rds_iw_sysctl_flow_control = 1;
57
58ctl_table rds_iw_sysctl_table[] = {
59 {
60 .ctl_name = CTL_UNNUMBERED,
61 .procname = "max_send_wr",
62 .data = &rds_iw_sysctl_max_send_wr,
63 .maxlen = sizeof(unsigned long),
64 .mode = 0644,
65 .proc_handler = &proc_doulongvec_minmax,
66 .extra1 = &rds_iw_sysctl_max_wr_min,
67 .extra2 = &rds_iw_sysctl_max_wr_max,
68 },
69 {
70 .ctl_name = CTL_UNNUMBERED,
71 .procname = "max_recv_wr",
72 .data = &rds_iw_sysctl_max_recv_wr,
73 .maxlen = sizeof(unsigned long),
74 .mode = 0644,
75 .proc_handler = &proc_doulongvec_minmax,
76 .extra1 = &rds_iw_sysctl_max_wr_min,
77 .extra2 = &rds_iw_sysctl_max_wr_max,
78 },
79 {
80 .ctl_name = CTL_UNNUMBERED,
81 .procname = "max_unsignaled_wr",
82 .data = &rds_iw_sysctl_max_unsig_wrs,
83 .maxlen = sizeof(unsigned long),
84 .mode = 0644,
85 .proc_handler = &proc_doulongvec_minmax,
86 .extra1 = &rds_iw_sysctl_max_unsig_wr_min,
87 .extra2 = &rds_iw_sysctl_max_unsig_wr_max,
88 },
89 {
90 .ctl_name = CTL_UNNUMBERED,
91 .procname = "max_unsignaled_bytes",
92 .data = &rds_iw_sysctl_max_unsig_bytes,
93 .maxlen = sizeof(unsigned long),
94 .mode = 0644,
95 .proc_handler = &proc_doulongvec_minmax,
96 .extra1 = &rds_iw_sysctl_max_unsig_bytes_min,
97 .extra2 = &rds_iw_sysctl_max_unsig_bytes_max,
98 },
99 {
100 .ctl_name = CTL_UNNUMBERED,
101 .procname = "max_recv_allocation",
102 .data = &rds_iw_sysctl_max_recv_allocation,
103 .maxlen = sizeof(unsigned long),
104 .mode = 0644,
105 .proc_handler = &proc_doulongvec_minmax,
106 },
107 {
108 .ctl_name = CTL_UNNUMBERED,
109 .procname = "flow_control",
110 .data = &rds_iw_sysctl_flow_control,
111 .maxlen = sizeof(rds_iw_sysctl_flow_control),
112 .mode = 0644,
113 .proc_handler = &proc_dointvec,
114 },
115 { .ctl_name = 0}
116};
117
118static struct ctl_path rds_iw_sysctl_path[] = {
119 { .procname = "net", .ctl_name = CTL_NET, },
120 { .procname = "rds", .ctl_name = CTL_UNNUMBERED, },
121 { .procname = "iw", .ctl_name = CTL_UNNUMBERED, },
122 { }
123};
124
125void rds_iw_sysctl_exit(void)
126{
127 if (rds_iw_sysctl_hdr)
128 unregister_sysctl_table(rds_iw_sysctl_hdr);
129}
130
131int __init rds_iw_sysctl_init(void)
132{
133 rds_iw_sysctl_hdr = register_sysctl_paths(rds_iw_sysctl_path, rds_iw_sysctl_table);
134 if (rds_iw_sysctl_hdr == NULL)
135 return -ENOMEM;
136 return 0;
137}
generated by cgit v1.2.2 (git 2.25.0) at 2024-11-07 06:31:17 -0500