lustre/ptlrpc/niobuf.c

   1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
   2  * vim:expandtab:shiftwidth=8:tabstop=8:
   3  *
   4  *  Copyright (C) 2002 Cluster File Systems, Inc.
   5  *
   6  *   This file is part of Lustre, http://www.lustre.org.
   7  *
   8  *   Lustre is free software; you can redistribute it and/or
   9  *   modify it under the terms of version 2 of the GNU General Public
  10  *   License as published by the Free Software Foundation.
  11  *
  12  *   Lustre is distributed in the hope that it will be useful,
  13  *   but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  *   GNU General Public License for more details.
  16  *
  17  *   You should have received a copy of the GNU General Public License
  18  *   along with Lustre; if not, write to the Free Software
  19  *   Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  20  *
  21  */
  22
  23 #define DEBUG_SUBSYSTEM S_RPC
  24
  25 #include <linux/obd_support.h>
  26 #include <linux/lustre_net.h>
  27 #include <linux/lustre_lib.h>
  28
  29 extern ptl_handle_eq_t request_out_eq, reply_in_eq, reply_out_eq,
  30         bulk_source_eq, bulk_sink_eq;
  31
  32 static int ptl_send_buf(struct ptlrpc_request *request,
  33                         struct ptlrpc_connection *conn, int portal)
  34 {
  35         int rc;
  36         ptl_process_id_t remote_id;
  37         ptl_handle_md_t md_h;
  38
  39         request->rq_req_md.user_ptr = request;
  40
  41         switch (request->rq_type) {
  42         case PTL_RPC_MSG_REQUEST:
  43                 request->rq_reqmsg->type = HTON__u32(request->rq_type);
  44                 request->rq_req_md.start = request->rq_reqmsg;
  45                 request->rq_req_md.length = request->rq_reqlen;
  46                 request->rq_req_md.eventq = request_out_eq;
  47                 break;
  48         case PTL_RPC_MSG_REPLY:
  49                 request->rq_repmsg->type = HTON__u32(request->rq_type);
  50                 request->rq_req_md.start = request->rq_repmsg;
  51                 request->rq_req_md.length = request->rq_replen;
  52                 request->rq_req_md.eventq = reply_out_eq;
  53                 break;
  54         default:
  55                 LBUG();
  56                 return -1; /* notreached */
  57         }
  58         request->rq_req_md.threshold = 1;
  59         request->rq_req_md.options = PTL_MD_OP_PUT;
  60         request->rq_req_md.user_ptr = request;
  61
  62         rc = PtlMDBind(conn->c_peer.peer_ni, request->rq_req_md, &md_h);
  63         if (rc != 0) {
  64                 CERROR("PtlMDBind failed: %d\n", rc);
  65                 LBUG();
  66                 return rc;
  67         }
  68
  69         remote_id.nid = conn->c_peer.peer_nid;
  70         remote_id.pid = 0;
  71
  72         CDEBUG(D_NET, "Sending %d bytes to portal %d, xid %Ld\n",
  73                request->rq_req_md.length, portal, request->rq_xid);
  74
  75         if (!portal)
  76                 LBUG();
  77         rc = PtlPut(md_h, PTL_NOACK_REQ, remote_id, portal, 0, request->rq_xid,
  78                     0, 0);
  79         if (rc != PTL_OK) {
  80                 CERROR("PtlPut(%Lu, %d, %Ld) failed: %d\n", remote_id.nid,
  81                        portal, request->rq_xid, rc);
  82                 PtlMDUnlink(md_h);
  83         }
  84
  85         return rc;
  86 }
  87
  88 static inline struct iovec *
  89 ptlrpc_get_bulk_iov (struct ptlrpc_bulk_desc *desc)
  90 {
  91         struct iovec *iov;
  92
  93         if (desc->bd_page_count <= sizeof (desc->bd_iov)/sizeof (struct iovec))
  94                 return (desc->bd_iov);
  95
  96         OBD_ALLOC (iov, desc->bd_page_count * sizeof (struct iovec));
  97         if (iov == NULL)
  98                 LBUG();
  99
 100         return (iov);
 101 }
 102
 103 static inline void
 104 ptlrpc_put_bulk_iov (struct ptlrpc_bulk_desc *desc, struct iovec *iov)
 105 {
 106         if (desc->bd_page_count <= sizeof (desc->bd_iov)/sizeof (struct iovec))
 107                 return;
 108
 109         OBD_FREE (iov, desc->bd_page_count * sizeof (struct iovec));
 110 }
 111
 112 int ptlrpc_send_bulk(struct ptlrpc_bulk_desc *desc)
 113 {
 114         int rc;
 115         struct list_head *tmp, *next;
 116         ptl_process_id_t remote_id;
 117         __u32 xid = 0;
 118         struct iovec *iov;
 119         ENTRY;
 120
 121         iov = ptlrpc_get_bulk_iov (desc);
 122         if (iov == NULL)
 123                 RETURN (-ENOMEM);
 124
 125         desc->bd_md.start = iov;
 126         desc->bd_md.niov = 0;
 127         desc->bd_md.length = 0;
 128         desc->bd_md.eventq = bulk_source_eq;
 129         desc->bd_md.threshold = 2; /* SENT and ACK */
 130         desc->bd_md.options = PTL_MD_OP_PUT | PTL_MD_IOV;
 131         desc->bd_md.user_ptr = desc;
 132
 133         atomic_set (&desc->bd_source_callback_count, 2);
 134
 135         list_for_each_safe(tmp, next, &desc->bd_page_list) {
 136                 struct ptlrpc_bulk_page *bulk;
 137                 bulk = list_entry(tmp, struct ptlrpc_bulk_page, bp_link);
 138
 139                 LASSERT (desc->bd_md.niov < desc->bd_page_count);
 140
 141                 if (desc->bd_md.niov == 0)
 142                         xid = bulk->bp_xid;
 143                 LASSERT (xid == bulk->bp_xid);   /* should all be the same */
 144
 145                 iov[desc->bd_md.niov].iov_base = bulk->bp_buf;
 146                 iov[desc->bd_md.niov].iov_len = bulk->bp_buflen;
 147                 desc->bd_md.niov++;
 148                 desc->bd_md.length += bulk->bp_buflen;
 149         }
 150
 151         LASSERT (desc->bd_md.niov == desc->bd_page_count);
 152         LASSERT (desc->bd_md.niov != 0);
 153
 154         rc = PtlMDBind(desc->bd_connection->c_peer.peer_ni, desc->bd_md,
 155                        &desc->bd_md_h);
 156
 157         ptlrpc_put_bulk_iov (desc, iov);        /* move down to reduce latency to send */
 158
 159         if (rc != PTL_OK) {
 160                 CERROR("PtlMDBind failed: %d\n", rc);
 161                 LBUG();
 162                 RETURN(rc);
 163         }
 164
 165         remote_id.nid = desc->bd_connection->c_peer.peer_nid;
 166         remote_id.pid = 0;
 167
 168         CDEBUG(D_NET, "Sending %u pages %u bytes to portal %d nid %Lx pid %d xid %d\n",
 169                desc->bd_md.niov, desc->bd_md.length,
 170                desc->bd_portal, remote_id.nid, remote_id.pid, xid);
 171
 172         rc = PtlPut(desc->bd_md_h, PTL_ACK_REQ, remote_id,
 173                     desc->bd_portal, 0, xid, 0, 0);
 174         if (rc != PTL_OK) {
 175                 CERROR("PtlPut(%Lu, %d, %d) failed: %d\n",
 176                        remote_id.nid, desc->bd_portal, xid, rc);
 177                 PtlMDUnlink(desc->bd_md_h);
 178                 LBUG();
 179                 RETURN(rc);
 180         }
 181
 182         RETURN(0);
 183 }
 184
 185 int ptlrpc_register_bulk(struct ptlrpc_bulk_desc *desc)
 186 {
 187         struct list_head *tmp, *next;
 188         int rc;
 189         __u32 xid = 0;
 190         struct iovec *iov;
 191         ptl_process_id_t source_id;
 192         ENTRY;
 193
 194         iov = ptlrpc_get_bulk_iov (desc);
 195         if (iov == NULL)
 196                 return (-ENOMEM);
 197
 198         desc->bd_md.start = iov;
 199         desc->bd_md.niov = 0;
 200         desc->bd_md.length = 0;
 201         desc->bd_md.threshold = 1;
 202         desc->bd_md.options = PTL_MD_OP_PUT | PTL_MD_IOV;
 203         desc->bd_md.user_ptr = desc;
 204         desc->bd_md.eventq = bulk_sink_eq;
 205
 206         list_for_each_safe(tmp, next, &desc->bd_page_list) {
 207                 struct ptlrpc_bulk_page *bulk;
 208                 bulk = list_entry(tmp, struct ptlrpc_bulk_page, bp_link);
 209
 210                 LASSERT (desc->bd_md.niov < desc->bd_page_count);
 211
 212                 if (desc->bd_md.niov == 0)
 213                         xid = bulk->bp_xid;
 214                 LASSERT (xid == bulk->bp_xid);   /* should all be the same */
 215
 216                 iov[desc->bd_md.niov].iov_base = bulk->bp_buf;
 217                 iov[desc->bd_md.niov].iov_len = bulk->bp_buflen;
 218                 desc->bd_md.niov++;
 219                 desc->bd_md.length += bulk->bp_buflen;
 220         }
 221
 222         LASSERT (desc->bd_md.niov == desc->bd_page_count);
 223         LASSERT (desc->bd_md.niov != 0);
 224
 225         source_id.nid = desc->bd_connection->c_peer.peer_nid;
 226         source_id.pid = PTL_PID_ANY;
 227
 228         rc = PtlMEAttach(desc->bd_connection->c_peer.peer_ni,
 229                          desc->bd_portal, source_id, xid, 0,
 230                          PTL_UNLINK, PTL_INS_AFTER, &desc->bd_me_h);
 231
 232         ptlrpc_put_bulk_iov (desc, iov);
 233
 234         if (rc != PTL_OK) {
 235                 CERROR("PtlMEAttach failed: %d\n", rc);
 236                 LBUG();
 237                 GOTO(cleanup, rc);
 238         }
 239
 240         rc = PtlMDAttach(desc->bd_me_h, desc->bd_md, PTL_UNLINK,
 241                          &desc->bd_md_h);
 242         if (rc != PTL_OK) {
 243                 CERROR("PtlMDAttach failed: %d\n", rc);
 244                 LBUG();
 245                 GOTO(cleanup, rc);
 246         }
 247
 248         CDEBUG(D_NET, "Setup bulk sink buffers: %u pages %u bytes, xid %u, "
 249                "portal %u\n", desc->bd_md.niov, desc->bd_md.length,
 250                xid, desc->bd_portal);
 251
 252         RETURN(0);
 253
 254  cleanup:
 255         ptlrpc_abort_bulk(desc);
 256
 257         return rc;
 258 }
 259
 260 int ptlrpc_abort_bulk(struct ptlrpc_bulk_desc *desc)
 261 {
 262         /* This should be safe: these handles are initialized to be
 263          * invalid in ptlrpc_prep_bulk() */
 264         PtlMDUnlink(desc->bd_md_h);
 265         PtlMEUnlink(desc->bd_me_h);
 266
 267         return 0;
 268 }
 269
 270 int ptlrpc_reply(struct ptlrpc_service *svc, struct ptlrpc_request *req)
 271 {
 272         if (req->rq_repmsg == NULL) {
 273                 CERROR("bad: someone called ptlrpc_reply when they meant "
 274                        "ptlrpc_error\n");
 275                 return -EINVAL;
 276         }
 277
 278         /* FIXME: we need to increment the count of handled events */
 279         if (req->rq_type != PTL_RPC_MSG_ERR)
 280                 req->rq_type = PTL_RPC_MSG_REPLY;
 281         //req->rq_repmsg->conn = req->rq_connection->c_remote_conn;
 282         //req->rq_repmsg->token = req->rq_connection->c_remote_token;
 283         req->rq_repmsg->status = HTON__u32(req->rq_status);
 284         return ptl_send_buf(req, req->rq_connection, svc->srv_rep_portal);
 285 }
 286
 287 int ptlrpc_error(struct ptlrpc_service *svc, struct ptlrpc_request *req)
 288 {
 289         int rc;
 290         ENTRY;
 291
 292         if (req->rq_repmsg) {
 293                 CERROR("req already has repmsg\n");
 294                 LBUG();
 295         }
 296
 297         rc = lustre_pack_msg(0, NULL, NULL, &req->rq_replen, &req->rq_repmsg);
 298         if (rc)
 299                 RETURN(rc);
 300
 301         req->rq_repmsg->type = HTON__u32(PTL_RPC_MSG_ERR);
 302
 303         rc = ptlrpc_reply(svc, req);
 304         RETURN(rc);
 305 }
 306
 307 int ptl_send_rpc(struct ptlrpc_request *request)
 308 {
 309         int rc;
 310         char *repbuf;
 311         ptl_process_id_t source_id;
 312
 313         ENTRY;
 314
 315         if (request->rq_type != PTL_RPC_MSG_REQUEST) {
 316                 CERROR("wrong packet type sent %d\n",
 317                        NTOH__u32(request->rq_reqmsg->type));
 318                 LBUG();
 319                 RETURN(EINVAL);
 320         }
 321         if (request->rq_replen == 0) {
 322                 CERROR("request->rq_replen is 0!\n");
 323                 RETURN(EINVAL);
 324         }
 325
 326         /* request->rq_repmsg is set only when the reply comes in, in
 327          * client_packet_callback() */
 328         if (request->rq_reply_md.start)
 329                 OBD_FREE(request->rq_reply_md.start, request->rq_replen);
 330
 331         OBD_ALLOC(repbuf, request->rq_replen);
 332         if (!repbuf) {
 333                 LBUG();
 334                 RETURN(ENOMEM);
 335         }
 336
 337         // down(&request->rq_client->cli_rpc_sem);
 338
 339         source_id.nid = request->rq_connection->c_peer.peer_nid;
 340         source_id.pid = PTL_PID_ANY;
 341
 342         rc = PtlMEAttach(request->rq_connection->c_peer.peer_ni,
 343                          request->rq_import->imp_client->cli_reply_portal,
 344                          source_id, request->rq_xid, 0, PTL_UNLINK,
 345                          PTL_INS_AFTER, &request->rq_reply_me_h);
 346         if (rc != PTL_OK) {
 347                 CERROR("PtlMEAttach failed: %d\n", rc);
 348                 LBUG();
 349                 GOTO(cleanup, rc);
 350         }
 351
 352         request->rq_reply_md.start = repbuf;
 353         request->rq_reply_md.length = request->rq_replen;
 354         request->rq_reply_md.threshold = 1;
 355         request->rq_reply_md.options = PTL_MD_OP_PUT;
 356         request->rq_reply_md.user_ptr = request;
 357         request->rq_reply_md.eventq = reply_in_eq;
 358
 359         rc = PtlMDAttach(request->rq_reply_me_h, request->rq_reply_md,
 360                          PTL_UNLINK, &request->rq_reply_md_h);
 361         if (rc != PTL_OK) {
 362                 CERROR("PtlMDAttach failed: %d\n", rc);
 363                 LBUG();
 364                 GOTO(cleanup2, rc);
 365         }
 366
 367         CDEBUG(D_NET, "Setup reply buffer: %u bytes, xid %Lu, portal %u\n",
 368                request->rq_replen, request->rq_xid,
 369                request->rq_import->imp_client->cli_reply_portal);
 370
 371         rc = ptl_send_buf(request, request->rq_connection,
 372                           request->rq_import->imp_client->cli_request_portal);
 373         RETURN(rc);
 374
 375  cleanup2:
 376         PtlMEUnlink(request->rq_reply_me_h);
 377  cleanup:
 378         OBD_FREE(repbuf, request->rq_replen);
 379         // up(&request->rq_client->cli_rpc_sem);
 380
 381         return rc;
 382 }
 383
 384 void ptlrpc_link_svc_me(struct ptlrpc_request_buffer_desc *rqbd)
 385 {
 386         struct ptlrpc_service *service = rqbd->rqbd_service;
 387         static ptl_process_id_t match_id = {PTL_NID_ANY, PTL_PID_ANY};
 388         int rc;
 389         ptl_md_t dummy;
 390         ptl_handle_md_t md_h;
 391
 392         /* Attach the leading ME on which we build the ring */
 393         rc = PtlMEAttach(service->srv_self.peer_ni, service->srv_req_portal,
 394                          match_id, 0, ~0,
 395                          PTL_UNLINK, PTL_INS_AFTER, &rqbd->rqbd_me_h);
 396         if (rc != PTL_OK) {
 397                 CERROR("PtlMEAttach failed: %d\n", rc);
 398                 LBUG();
 399         }
 400
 401         dummy.start         = rqbd->rqbd_buffer;
 402         dummy.length        = service->srv_buf_size;
 403         dummy.max_offset    = service->srv_buf_size;
 404         dummy.threshold     = 1;
 405         dummy.options       = PTL_MD_OP_PUT;
 406         dummy.user_ptr      = rqbd;
 407         dummy.eventq        = service->srv_eq_h;
 408
 409         rc = PtlMDAttach(rqbd->rqbd_me_h, dummy, PTL_UNLINK, &md_h);
 410         if (rc != PTL_OK) {
 411                 /* cleanup */
 412                 CERROR("PtlMDAttach failed: %d\n", rc);
 413                 LBUG();
 414         }
 415 }