1 /* 2 * CDDL HEADER START 3 * 4 * The contents of this file are subject to the terms of the 5 * Common Development and Distribution License (the "License"). 6 * You may not use this file except in compliance with the License. 7 * 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9 * or http://www.opensolaris.org/os/licensing. 10 * See the License for the specific language governing permissions 11 * and limitations under the License. 12 * 13 * When distributing Covered Code, include this CDDL HEADER in each 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15 * If applicable, add the following below this CDDL HEADER, with the 16 * fields enclosed by brackets "[]" replaced with your own identifying 17 * information: Portions Copyright [yyyy] [name of copyright owner] 18 * 19 * CDDL HEADER END 20 */ 21 22 /* 23 * Copyright (c) 2006, 2010, Oracle and/or its affiliates. All rights reserved. 24 */ 25 26 #include <sys/types.h> 27 #include <sys/errno.h> 28 #include <sys/sysmacros.h> 29 #include <sys/param.h> 30 #include <sys/machsystm.h> 31 #include <sys/stream.h> 32 #include <sys/strsubr.h> 33 #include <sys/kmem.h> 34 #include <sys/strsun.h> 35 #include <sys/callb.h> 36 #include <sys/sdt.h> 37 #include <sys/mach_descrip.h> 38 #include <sys/mdeg.h> 39 #include <net/if.h> 40 #include <sys/vsw.h> 41 #include <sys/vio_mailbox.h> 42 #include <sys/vio_common.h> 43 #include <sys/vnet_common.h> 44 #include <sys/vnet_mailbox.h> 45 #include <sys/vio_util.h> 46 47 /* 48 * This file contains the implementation of RxDringData transfer mode of VIO 49 * Protocol in vsw. The functions in this file are invoked from vsw_ldc.c 50 * after RxDringData mode is negotiated with the peer during attribute phase of 51 * handshake. This file contains functions that setup the transmit and receive 52 * descriptor rings, and associated resources in RxDringData mode. It also 53 * contains the transmit and receive data processing functions that are invoked 54 * in RxDringData mode. The data processing routines in this file have the 55 * suffix '_shm' to indicate the shared memory mechanism used in RxDringData 56 * mode. 57 */ 58 59 /* Functions exported to vsw_ldc.c */ 60 vio_dring_reg_msg_t *vsw_create_rx_dring_info(vsw_ldc_t *); 61 void vsw_destroy_rx_dring(vsw_ldc_t *ldcp); 62 dring_info_t *vsw_map_tx_dring(vsw_ldc_t *ldcp, void *pkt); 63 void vsw_unmap_tx_dring(vsw_ldc_t *ldcp); 64 int vsw_dringsend_shm(vsw_ldc_t *, mblk_t *); 65 void vsw_ldc_rcv_worker(void *arg); 66 void vsw_stop_rcv_thread(vsw_ldc_t *ldcp); 67 void vsw_process_dringdata_shm(void *, void *); 68 69 /* Internal functions */ 70 static dring_info_t *vsw_create_rx_dring(vsw_ldc_t *); 71 static int vsw_setup_rx_dring(vsw_ldc_t *ldcp, dring_info_t *dp); 72 static void vsw_process_dringdata_info_shm(vsw_ldc_t *ldcp, 73 vio_dring_msg_t *msg); 74 static void vsw_process_dringdata_ack_shm(vsw_ldc_t *ldcp, 75 vio_dring_msg_t *msg); 76 static void vsw_ldc_rcv_shm(vsw_ldc_t *ldcp); 77 static int vsw_receive_packet(vsw_ldc_t *ldcp, mblk_t **bp); 78 static int vsw_send_msg_shm(vsw_ldc_t *ldcp, void *msgp, int size, 79 boolean_t handle_reset); 80 81 /* Functions imported from vsw_ldc.c */ 82 extern void vsw_process_pkt(void *); 83 extern void vsw_destroy_rxpools(void *); 84 extern dring_info_t *vsw_map_dring_cmn(vsw_ldc_t *ldcp, 85 vio_dring_reg_msg_t *dring_pkt); 86 extern void vsw_process_conn_evt(vsw_ldc_t *, uint16_t); 87 extern mblk_t *vsw_vlan_frame_pretag(void *arg, int type, mblk_t *mp); 88 89 /* Tunables */ 90 extern int vsw_wretries; 91 extern int vsw_recv_delay; 92 extern int vsw_recv_retries; 93 extern uint32_t vsw_chain_len; 94 extern uint32_t vsw_num_descriptors; 95 extern uint32_t vsw_nrbufs_factor; 96 97 #define VSW_SWITCH_FRAMES(vswp, ldcp, bp, bpt, count, total_count) \ 98 { \ 99 DTRACE_PROBE2(vsw_rx_pkts, vsw_ldc_t *, (ldcp), int, (count)); \ 100 (vswp)->vsw_switch_frame((vswp), (bp), VSW_VNETPORT, \ 101 (ldcp)->ldc_port, NULL); \ 102 (bp) = (bpt) = NULL; \ 103 (count) = 0; \ 104 } 105 106 vio_dring_reg_msg_t * 107 vsw_create_rx_dring_info(vsw_ldc_t *ldcp) 108 { 109 vio_dring_reg_msg_t *mp; 110 vio_dring_reg_ext_msg_t *emsg; 111 dring_info_t *dp; 112 uint8_t *buf; 113 vsw_t *vswp = ldcp->ldc_vswp; 114 115 D1(vswp, "%s enter\n", __func__); 116 117 /* 118 * If we can't create a dring, obviously no point sending 119 * a message. 120 */ 121 if ((dp = vsw_create_rx_dring(ldcp)) == NULL) 122 return (NULL); 123 124 mp = kmem_zalloc(VNET_DRING_REG_EXT_MSG_SIZE(dp->data_ncookies), 125 KM_SLEEP); 126 127 mp->tag.vio_msgtype = VIO_TYPE_CTRL; 128 mp->tag.vio_subtype = VIO_SUBTYPE_INFO; 129 mp->tag.vio_subtype_env = VIO_DRING_REG; 130 mp->tag.vio_sid = ldcp->local_session; 131 132 /* payload */ 133 mp->num_descriptors = dp->num_descriptors; 134 mp->descriptor_size = dp->descriptor_size; 135 mp->options = dp->options; 136 mp->ncookies = dp->dring_ncookies; 137 bcopy(&dp->dring_cookie[0], &mp->cookie[0], 138 sizeof (ldc_mem_cookie_t)); 139 140 mp->dring_ident = 0; 141 142 buf = (uint8_t *)mp->cookie; 143 144 /* skip over dring cookies */ 145 ASSERT(mp->ncookies == 1); 146 buf += (mp->ncookies * sizeof (ldc_mem_cookie_t)); 147 148 emsg = (vio_dring_reg_ext_msg_t *)buf; 149 150 /* copy data_ncookies in the msg */ 151 emsg->data_ncookies = dp->data_ncookies; 152 153 /* copy data area size in the msg */ 154 emsg->data_area_size = dp->data_sz; 155 156 /* copy data area cookies in the msg */ 157 bcopy(dp->data_cookie, (ldc_mem_cookie_t *)emsg->data_cookie, 158 sizeof (ldc_mem_cookie_t) * dp->data_ncookies); 159 160 D1(vswp, "%s exit\n", __func__); 161 162 return (mp); 163 } 164 165 /* 166 * Allocate receive resources for the channel. The resources consist of a 167 * receive descriptor ring and an associated receive buffer area. 168 */ 169 static dring_info_t * 170 vsw_create_rx_dring(vsw_ldc_t *ldcp) 171 { 172 vsw_t *vswp = ldcp->ldc_vswp; 173 ldc_mem_info_t minfo; 174 dring_info_t *dp; 175 176 dp = (dring_info_t *)kmem_zalloc(sizeof (dring_info_t), KM_SLEEP); 177 mutex_init(&dp->dlock, NULL, MUTEX_DRIVER, NULL); 178 ldcp->lane_out.dringp = dp; 179 180 /* Create the receive descriptor ring */ 181 if ((ldc_mem_dring_create(vsw_num_descriptors, 182 sizeof (vnet_rx_dringdata_desc_t), &dp->dring_handle)) != 0) { 183 DERR(vswp, "vsw_create_rx_dring(%lld): ldc dring create " 184 "failed", ldcp->ldc_id); 185 goto fail; 186 } 187 188 ASSERT(dp->dring_handle != NULL); 189 190 /* Get the addr of descriptor ring */ 191 if ((ldc_mem_dring_info(dp->dring_handle, &minfo)) != 0) { 192 DERR(vswp, "vsw_create_rx_dring(%lld): dring info failed\n", 193 ldcp->ldc_id); 194 goto fail; 195 } else { 196 ASSERT(minfo.vaddr != 0); 197 dp->pub_addr = minfo.vaddr; 198 } 199 200 dp->num_descriptors = vsw_num_descriptors; 201 dp->descriptor_size = sizeof (vnet_rx_dringdata_desc_t); 202 dp->options = VIO_RX_DRING_DATA; 203 dp->dring_ncookies = 1; /* guaranteed by ldc */ 204 dp->num_bufs = vsw_num_descriptors * vsw_nrbufs_factor; 205 206 /* 207 * Allocate a table that maps descriptor to its associated buffer; 208 * used while receiving to validate that the peer has not changed the 209 * buffer offset provided in the descriptor. 210 */ 211 dp->rxdp_to_vmp = kmem_zalloc(dp->num_descriptors * sizeof (uintptr_t), 212 KM_SLEEP); 213 214 /* Setup the descriptor ring */ 215 if (vsw_setup_rx_dring(ldcp, dp)) { 216 DERR(vswp, "%s: unable to setup ring", __func__); 217 goto fail; 218 } 219 220 /* 221 * The descriptors and the associated buffers are all ready; 222 * now bind descriptor ring to the channel. 223 */ 224 if ((ldc_mem_dring_bind(ldcp->ldc_handle, dp->dring_handle, 225 LDC_DIRECT_MAP | LDC_SHADOW_MAP, LDC_MEM_RW, 226 &dp->dring_cookie[0], &dp->dring_ncookies)) != 0) { 227 DERR(vswp, "vsw_create_rx_dring: unable to bind to channel " 228 "%lld", ldcp->ldc_id); 229 goto fail; 230 } 231 232 /* haven't used any descriptors yet */ 233 dp->end_idx = 0; 234 dp->last_ack_recv = -1; 235 dp->next_rxi = 0; 236 return (dp); 237 238 fail: 239 vsw_destroy_rx_dring(ldcp); 240 return (NULL); 241 } 242 243 /* 244 * Setup the descriptors in the rx dring. 245 * Returns 0 on success, 1 on failure. 246 */ 247 static int 248 vsw_setup_rx_dring(vsw_ldc_t *ldcp, dring_info_t *dp) 249 { 250 int i, j; 251 int rv; 252 size_t data_sz; 253 vio_mblk_t *vmp; 254 vio_mblk_t **rxdp_to_vmp; 255 vnet_rx_dringdata_desc_t *rxdp; 256 vnet_rx_dringdata_desc_t *pub_addr; 257 vsw_t *vswp = ldcp->ldc_vswp; 258 uint32_t ncookies = 0; 259 static char *name = "vsw_setup_rx_dring"; 260 void *data_addr = NULL; 261 262 /* 263 * Allocate a single large buffer that serves as the rx buffer area. 264 * We allocate a ldc memory handle and export the buffer area as shared 265 * memory. We send the ldc memcookie for this buffer space to the peer, 266 * as part of dring registration phase during handshake. We manage this 267 * buffer area as individual buffers of max_frame_size and provide 268 * specific buffer offsets in each descriptor to the peer. Note that 269 * the factor used to compute the # of buffers (above) must be > 1 to 270 * ensure that there are more buffers than the # of descriptors. This 271 * is needed because, while the shared memory buffers are sent up our 272 * stack during receive, the sender needs additional buffers that can 273 * be used for further transmits. This also means there is no one to 274 * one correspondence between the descriptor index and buffer offset. 275 * The sender has to read the buffer offset in the descriptor and use 276 * the specified offset to copy the tx data into the shared buffer. We 277 * (receiver) manage the individual buffers and their state (see 278 * VIO_MBLK_STATEs in vio_util.h). 279 */ 280 data_sz = vswp->max_frame_size + VNET_IPALIGN + VNET_LDCALIGN; 281 data_sz = VNET_ROUNDUP_2K(data_sz); 282 283 dp->desc_data_sz = data_sz; 284 dp->data_sz = (dp->num_bufs * data_sz); 285 data_addr = kmem_zalloc(dp->data_sz, KM_SLEEP); 286 dp->data_addr = data_addr; 287 288 D2(vswp, "%s: allocated %lld bytes at 0x%llx\n", name, 289 dp->data_sz, dp->data_addr); 290 291 /* Allocate a ldc memhandle for the entire rx data area */ 292 rv = ldc_mem_alloc_handle(ldcp->ldc_handle, &dp->data_handle); 293 if (rv != 0) { 294 DERR(vswp, "%s: alloc mem handle failed", name); 295 goto fail; 296 } 297 298 /* Allocate memory for the data cookies */ 299 dp->data_cookie = kmem_zalloc(VNET_DATA_AREA_COOKIES * 300 sizeof (ldc_mem_cookie_t), KM_SLEEP); 301 302 /* 303 * Bind ldc memhandle to the corresponding rx data area. 304 */ 305 rv = ldc_mem_bind_handle(dp->data_handle, (caddr_t)data_addr, 306 dp->data_sz, LDC_DIRECT_MAP, LDC_MEM_W, 307 dp->data_cookie, &ncookies); 308 if (rv != 0) { 309 DERR(vswp, "%s(%lld): ldc_mem_bind_handle failed " 310 "(rv %d)", name, ldcp->ldc_id, rv); 311 goto fail; 312 } 313 if ((ncookies == 0) || (ncookies > VNET_DATA_AREA_COOKIES)) { 314 goto fail; 315 } 316 dp->data_ncookies = ncookies; 317 318 for (j = 1; j < ncookies; j++) { 319 rv = ldc_mem_nextcookie(dp->data_handle, 320 &(dp->data_cookie[j])); 321 if (rv != 0) { 322 DERR(vswp, "%s: ldc_mem_nextcookie " 323 "failed rv (%d)", name, rv); 324 goto fail; 325 } 326 } 327 328 /* 329 * Successful in binding the handle to rx data area. Now setup mblks 330 * around each data buffer and setup the descriptors to point to these 331 * rx data buffers. We associate each descriptor with a buffer 332 * by specifying the buffer offset in the descriptor. When the peer 333 * needs to transmit data, this offset is read by the peer to determine 334 * the buffer in the mapped buffer area where the data to be 335 * transmitted should be copied, for a specific descriptor. 336 */ 337 rv = vio_create_mblks(dp->num_bufs, data_sz, (uint8_t *)data_addr, 338 &dp->rx_vmp); 339 if (rv != 0) { 340 goto fail; 341 } 342 343 pub_addr = dp->pub_addr; 344 rxdp_to_vmp = dp->rxdp_to_vmp; 345 for (i = 0; i < dp->num_descriptors; i++) { 346 rxdp = &pub_addr[i]; 347 /* allocate an mblk around this data buffer */ 348 vmp = vio_allocb(dp->rx_vmp); 349 ASSERT(vmp != NULL); 350 rxdp->data_buf_offset = VIO_MBLK_DATA_OFF(vmp) + VNET_IPALIGN; 351 rxdp->dstate = VIO_DESC_FREE; 352 rxdp_to_vmp[i] = vmp; 353 } 354 355 return (0); 356 357 fail: 358 /* return failure; caller will cleanup */ 359 return (1); 360 } 361 362 /* 363 * Free receive resources for the channel. 364 */ 365 void 366 vsw_destroy_rx_dring(vsw_ldc_t *ldcp) 367 { 368 vsw_t *vswp = ldcp->ldc_vswp; 369 lane_t *lp = &ldcp->lane_out; 370 dring_info_t *dp; 371 372 dp = lp->dringp; 373 if (dp == NULL) { 374 return; 375 } 376 377 mutex_enter(&dp->dlock); 378 379 if (dp->rx_vmp != NULL) { 380 vio_clobber_pool(dp->rx_vmp); 381 /* 382 * If we can't destroy the rx pool for this channel, dispatch a 383 * task to retry and clean up those rx pools. Note that we 384 * don't need to wait for the task to complete. If the vsw 385 * device itself gets detached (vsw_detach()), it will wait for 386 * the task to complete implicitly in ddi_taskq_destroy(). 387 */ 388 if (vio_destroy_mblks(dp->rx_vmp) != 0) { 389 (void) ddi_taskq_dispatch(vswp->rxp_taskq, 390 vsw_destroy_rxpools, dp->rx_vmp, DDI_SLEEP); 391 } 392 } 393 394 /* Free rx data area cookies */ 395 if (dp->data_cookie != NULL) { 396 kmem_free(dp->data_cookie, VNET_DATA_AREA_COOKIES * 397 sizeof (ldc_mem_cookie_t)); 398 dp->data_cookie = NULL; 399 } 400 401 /* Unbind rx data area memhandle */ 402 if (dp->data_ncookies != 0) { 403 (void) ldc_mem_unbind_handle(dp->data_handle); 404 dp->data_ncookies = 0; 405 } 406 407 /* Free rx data area memhandle */ 408 if (dp->data_handle) { 409 (void) ldc_mem_free_handle(dp->data_handle); 410 dp->data_handle = 0; 411 } 412 413 /* Now free the rx data area itself */ 414 if (dp->data_addr != NULL) { 415 kmem_free(dp->data_addr, dp->data_sz); 416 } 417 418 /* Finally, free the receive descriptor ring */ 419 if (dp->dring_handle != NULL) { 420 (void) ldc_mem_dring_unbind(dp->dring_handle); 421 (void) ldc_mem_dring_destroy(dp->dring_handle); 422 } 423 424 if (dp->rxdp_to_vmp != NULL) { 425 kmem_free(dp->rxdp_to_vmp, 426 dp->num_descriptors * sizeof (uintptr_t)); 427 dp->rxdp_to_vmp = NULL; 428 } 429 430 mutex_exit(&dp->dlock); 431 mutex_destroy(&dp->dlock); 432 mutex_destroy(&dp->restart_lock); 433 kmem_free(dp, sizeof (dring_info_t)); 434 lp->dringp = NULL; 435 } 436 437 /* 438 * Map the receive descriptor ring exported by the peer, as our transmit 439 * descriptor ring. 440 */ 441 dring_info_t * 442 vsw_map_tx_dring(vsw_ldc_t *ldcp, void *pkt) 443 { 444 int i; 445 int rv; 446 dring_info_t *dp; 447 vnet_rx_dringdata_desc_t *txdp; 448 on_trap_data_t otd; 449 vio_dring_reg_msg_t *dring_pkt = pkt; 450 451 dp = vsw_map_dring_cmn(ldcp, dring_pkt); 452 if (dp == NULL) { 453 return (NULL); 454 } 455 456 /* RxDringData mode specific initializations */ 457 mutex_init(&dp->txlock, NULL, MUTEX_DRIVER, NULL); 458 mutex_init(&dp->restart_lock, NULL, MUTEX_DRIVER, NULL); 459 dp->next_txi = dp->restart_peer_txi = 0; 460 dp->restart_reqd = B_TRUE; 461 ldcp->dringdata_msgid = 0; 462 ldcp->lane_in.dringp = dp; 463 464 /* 465 * Mark the descriptor state as 'done'. This is implementation specific 466 * and not required by the protocol. In our implementation, we only 467 * need the descripor to be in 'done' state to be used by the transmit 468 * function and the peer is not aware of it. As the protocol requires 469 * that during initial registration the exporting end point mark the 470 * dstate as 'free', we change it 'done' here. After this, the dstate 471 * in our implementation will keep moving between 'ready', set by our 472 * transmit function; and and 'done', set by the peer (per protocol) 473 * after receiving data. 474 * Setup on_trap() protection before accessing dring shared memory area. 475 */ 476 rv = LDC_ON_TRAP(&otd); 477 if (rv != 0) { 478 /* 479 * Data access fault occured down the code path below while 480 * accessing the descriptors. Return failure. 481 */ 482 goto fail; 483 } 484 485 txdp = (vnet_rx_dringdata_desc_t *)dp->pub_addr; 486 for (i = 0; i < dp->num_descriptors; i++) { 487 txdp[i].dstate = VIO_DESC_DONE; 488 } 489 490 (void) LDC_NO_TRAP(); 491 492 return (dp); 493 494 fail: 495 if (dp->dring_handle != NULL) { 496 (void) ldc_mem_dring_unmap(dp->dring_handle); 497 } 498 kmem_free(dp, sizeof (*dp)); 499 return (NULL); 500 } 501 502 /* 503 * Unmap the transmit descriptor ring. 504 */ 505 void 506 vsw_unmap_tx_dring(vsw_ldc_t *ldcp) 507 { 508 lane_t *lp = &ldcp->lane_in; 509 dring_info_t *dp; 510 511 if ((dp = lp->dringp) == NULL) { 512 return; 513 } 514 515 /* Unmap tx data area and free data handle */ 516 if (dp->data_handle != NULL) { 517 (void) ldc_mem_unmap(dp->data_handle); 518 (void) ldc_mem_free_handle(dp->data_handle); 519 dp->data_handle = NULL; 520 } 521 522 /* Free tx data area cookies */ 523 if (dp->data_cookie != NULL) { 524 kmem_free(dp->data_cookie, dp->data_ncookies * 525 sizeof (ldc_mem_cookie_t)); 526 dp->data_cookie = NULL; 527 dp->data_ncookies = 0; 528 } 529 530 /* Unmap peer's dring */ 531 if (dp->dring_handle != NULL) { 532 (void) ldc_mem_dring_unmap(dp->dring_handle); 533 dp->dring_handle = NULL; 534 } 535 536 mutex_destroy(&dp->txlock); 537 kmem_free(dp, sizeof (dring_info_t)); 538 lp->dringp = NULL; 539 } 540 541 /* 542 * A per LDC worker thread to process the rx dring and receive packets. This 543 * thread is woken up by the LDC interrupt handler when a dring data info 544 * message is received. 545 */ 546 void 547 vsw_ldc_rcv_worker(void *arg) 548 { 549 callb_cpr_t cprinfo; 550 vsw_ldc_t *ldcp = (vsw_ldc_t *)arg; 551 vsw_t *vswp = ldcp->ldc_vswp; 552 553 D1(vswp, "%s(%lld):enter\n", __func__, ldcp->ldc_id); 554 CALLB_CPR_INIT(&cprinfo, &ldcp->rcv_thr_lock, callb_generic_cpr, 555 "vsw_rcv_thread"); 556 mutex_enter(&ldcp->rcv_thr_lock); 557 while (!(ldcp->rcv_thr_flags & VSW_WTHR_STOP)) { 558 559 CALLB_CPR_SAFE_BEGIN(&cprinfo); 560 /* 561 * Wait until the data is received or a stop 562 * request is received. 563 */ 564 while (!(ldcp->rcv_thr_flags & 565 (VSW_WTHR_DATARCVD | VSW_WTHR_STOP))) { 566 cv_wait(&ldcp->rcv_thr_cv, &ldcp->rcv_thr_lock); 567 } 568 CALLB_CPR_SAFE_END(&cprinfo, &ldcp->rcv_thr_lock) 569 570 /* 571 * First process the stop request. 572 */ 573 if (ldcp->rcv_thr_flags & VSW_WTHR_STOP) { 574 D2(vswp, "%s(%lld):Rx thread stopped\n", 575 __func__, ldcp->ldc_id); 576 break; 577 } 578 ldcp->rcv_thr_flags &= ~VSW_WTHR_DATARCVD; 579 mutex_exit(&ldcp->rcv_thr_lock); 580 D1(vswp, "%s(%lld):calling vsw_process_pkt\n", 581 __func__, ldcp->ldc_id); 582 vsw_ldc_rcv_shm(ldcp); 583 mutex_enter(&ldcp->rcv_thr_lock); 584 } 585 586 /* 587 * Update the run status and wakeup the thread that 588 * has sent the stop request. 589 */ 590 ldcp->rcv_thr_flags &= ~VSW_WTHR_STOP; 591 ldcp->rcv_thread = NULL; 592 CALLB_CPR_EXIT(&cprinfo); 593 D1(vswp, "%s(%lld):exit\n", __func__, ldcp->ldc_id); 594 thread_exit(); 595 } 596 597 /* 598 * Process the rx descriptor ring in the context of receive worker 599 * thread and switch the received packets to their destinations. 600 */ 601 static void 602 vsw_ldc_rcv_shm(vsw_ldc_t *ldcp) 603 { 604 int rv; 605 uint32_t end_ix; 606 vio_dring_msg_t msg; 607 vio_dring_msg_t *msgp = &msg; 608 int count = 0; 609 int total_count = 0; 610 uint32_t retries = 0; 611 mblk_t *bp = NULL; 612 mblk_t *bpt = NULL; 613 mblk_t *mp = NULL; 614 vsw_t *vswp = ldcp->ldc_vswp; 615 lane_t *lp = &ldcp->lane_out; 616 dring_info_t *dp = lp->dringp; 617 618 do { 619 again: 620 rv = vsw_receive_packet(ldcp, &mp); 621 if (rv != 0) { 622 if (rv == EINVAL) { 623 /* Invalid descriptor error; get next */ 624 continue; 625 } 626 if (rv != EAGAIN) { 627 break; 628 } 629 630 /* Descriptor not ready for processsing */ 631 if (retries == vsw_recv_retries) { 632 DTRACE_PROBE1(vsw_noready_rxds, 633 vsw_ldc_t *, ldcp); 634 break; 635 } 636 637 /* Switch packets received so far before retrying */ 638 if (bp != NULL) { 639 VSW_SWITCH_FRAMES(vswp, ldcp, bp, bpt, count, 640 total_count); 641 } 642 retries++; 643 drv_usecwait(vsw_recv_delay); 644 goto again; 645 } 646 retries = 0; 647 648 /* Build a chain of received packets */ 649 if (bp == NULL) { 650 /* first pkt */ 651 bp = mp; 652 bpt = bp; 653 bpt->b_next = NULL; 654 } else { 655 mp->b_next = NULL; 656 bpt->b_next = mp; 657 bpt = mp; 658 } 659 660 total_count++; 661 count++; 662 663 /* 664 * If we have gathered vsw_chain_len (tunable) 665 * # of packets in the chain, switch them. 666 */ 667 if (count == vsw_chain_len) { 668 VSW_SWITCH_FRAMES(vswp, ldcp, bp, bpt, count, 669 total_count); 670 } 671 672 /* 673 * Stop further processing if we processed the entire dring 674 * once; otherwise continue. 675 */ 676 } while (total_count < dp->num_bufs); 677 678 DTRACE_PROBE2(vsw_rx_total_count, vsw_ldc_t *, ldcp, 679 int, (total_count)); 680 if (bp != NULL) { 681 VSW_SWITCH_FRAMES(vswp, ldcp, bp, bpt, count, 682 total_count); 683 } 684 685 /* Send stopped signal to peer (sender) */ 686 end_ix = lp->dringp->next_rxi; 687 DECR_RXI(dp, end_ix); 688 msgp->tag.vio_msgtype = VIO_TYPE_DATA; 689 msgp->tag.vio_subtype = VIO_SUBTYPE_ACK; 690 msgp->tag.vio_subtype_env = VIO_DRING_DATA; 691 msgp->dring_ident = ldcp->lane_in.dringp->ident; 692 msgp->tag.vio_sid = ldcp->local_session; 693 msgp->dring_process_state = VIO_DP_STOPPED; 694 msgp->start_idx = VNET_START_IDX_UNSPEC; 695 msgp->end_idx = end_ix; 696 697 (void) vsw_send_msg_shm(ldcp, (void *)msgp, 698 sizeof (vio_dring_msg_t), B_TRUE); 699 700 ldcp->ldc_stats.dring_data_acks_sent++; 701 ldcp->ldc_stats.dring_stopped_acks_sent++; 702 } 703 704 /* 705 * Process the next index in the rx dring and receive the associated packet. 706 * 707 * Returns: 708 * bp: Success: The received packet. 709 * Failure: NULL 710 * retval: 711 * Success: 0 712 * Failure: EAGAIN: Descriptor not ready 713 * EIO: Descriptor contents invalid. 714 */ 715 static int 716 vsw_receive_packet(vsw_ldc_t *ldcp, mblk_t **bp) 717 { 718 uint32_t rxi; 719 vio_mblk_t *vmp; 720 vio_mblk_t *new_vmp; 721 struct ether_header *ehp; 722 vnet_rx_dringdata_desc_t *rxdp; 723 int err = 0; 724 uint_t nbytes = 0; 725 mblk_t *mp = NULL; 726 mblk_t *dmp = NULL; 727 vgen_stats_t *statsp = &ldcp->ldc_stats; 728 dring_info_t *dp = ldcp->lane_out.dringp; 729 vnet_rx_dringdata_desc_t *pub_addr = dp->pub_addr; 730 731 rxi = dp->next_rxi; 732 rxdp = &(pub_addr[rxi]); 733 vmp = dp->rxdp_to_vmp[rxi]; 734 735 if (rxdp->dstate != VIO_DESC_READY) { 736 /* 737 * Descriptor is not ready. 738 */ 739 return (EAGAIN); 740 } 741 742 /* 743 * Ensure load ordering of dstate and nbytes. 744 */ 745 MEMBAR_CONSUMER(); 746 747 if ((rxdp->nbytes < ETHERMIN) || 748 (rxdp->nbytes > ldcp->lane_in.mtu) || 749 (rxdp->data_buf_offset != 750 (VIO_MBLK_DATA_OFF(vmp) + VNET_IPALIGN))) { 751 /* 752 * Descriptor contents invalid. 753 */ 754 statsp->ierrors++; 755 rxdp->dstate = VIO_DESC_DONE; 756 err = EIO; 757 goto done; 758 } 759 760 /* 761 * Now allocate a new buffer for this descriptor before sending up the 762 * buffer being processed. If that fails, stop processing; as we are 763 * out of receive buffers. 764 */ 765 new_vmp = vio_allocb(dp->rx_vmp); 766 767 /* 768 * Process the current buffer being received. 769 */ 770 nbytes = rxdp->nbytes; 771 mp = vmp->mp; 772 773 if (new_vmp == NULL) { 774 /* 775 * We failed to get a new mapped buffer that is needed to 776 * refill the descriptor. In that case, leave the current 777 * buffer bound to the descriptor; allocate an mblk dynamically 778 * and copy the contents of the buffer to the mblk. Then send 779 * up this mblk. This way the sender has the same buffer as 780 * before that can be used to send new data. 781 */ 782 statsp->norcvbuf++; 783 dmp = allocb(nbytes + VNET_IPALIGN, BPRI_MED); 784 bcopy(mp->b_rptr + VNET_IPALIGN, 785 dmp->b_rptr + VNET_IPALIGN, nbytes); 786 mp = dmp; 787 } else { 788 /* Mark the status of the current rbuf */ 789 vmp->state = VIO_MBLK_HAS_DATA; 790 791 /* Set the offset of the new buffer in the descriptor */ 792 rxdp->data_buf_offset = 793 VIO_MBLK_DATA_OFF(new_vmp) + VNET_IPALIGN; 794 dp->rxdp_to_vmp[rxi] = new_vmp; 795 } 796 mp->b_rptr += VNET_IPALIGN; 797 mp->b_wptr = mp->b_rptr + nbytes; 798 799 /* 800 * Ensure store ordering of data_buf_offset and dstate; so that the 801 * peer sees the right data_buf_offset after it checks that the dstate 802 * is DONE. 803 */ 804 MEMBAR_PRODUCER(); 805 806 /* Now mark the descriptor 'done' */ 807 rxdp->dstate = VIO_DESC_DONE; 808 809 /* Update stats */ 810 statsp->ipackets++; 811 statsp->rbytes += rxdp->nbytes; 812 ehp = (struct ether_header *)mp->b_rptr; 813 if (IS_BROADCAST(ehp)) 814 statsp->brdcstrcv++; 815 else if (IS_MULTICAST(ehp)) 816 statsp->multircv++; 817 done: 818 /* Update the next index to be processed */ 819 INCR_RXI(dp, rxi); 820 821 /* Save the new recv index */ 822 dp->next_rxi = rxi; 823 824 /* Return the packet received */ 825 *bp = mp; 826 return (err); 827 } 828 829 void 830 vsw_stop_rcv_thread(vsw_ldc_t *ldcp) 831 { 832 kt_did_t tid = 0; 833 vsw_t *vswp = ldcp->ldc_vswp; 834 835 D1(vswp, "%s(%lld):enter\n", __func__, ldcp->ldc_id); 836 /* 837 * Send a stop request by setting the stop flag and 838 * wait until the rcv process thread stops. 839 */ 840 mutex_enter(&ldcp->rcv_thr_lock); 841 if (ldcp->rcv_thread != NULL) { 842 tid = ldcp->rcv_thread->t_did; 843 ldcp->rcv_thr_flags |= VSW_WTHR_STOP; 844 cv_signal(&ldcp->rcv_thr_cv); 845 } 846 mutex_exit(&ldcp->rcv_thr_lock); 847 848 if (tid != 0) { 849 thread_join(tid); 850 } 851 D1(vswp, "%s(%lld):exit\n", __func__, ldcp->ldc_id); 852 } 853 854 int 855 vsw_dringsend_shm(vsw_ldc_t *ldcp, mblk_t *mp) 856 { 857 uint32_t next_txi; 858 uint32_t txi; 859 vnet_rx_dringdata_desc_t *txdp; 860 struct ether_header *ehp; 861 size_t mblksz; 862 caddr_t dst; 863 mblk_t *bp; 864 size_t size; 865 on_trap_data_t otd; 866 uint32_t buf_offset; 867 vnet_rx_dringdata_desc_t *pub_addr; 868 vio_dring_msg_t msg; 869 vio_dring_msg_t *msgp = &msg; 870 int rv = 0; 871 boolean_t resched_peer = B_FALSE; 872 boolean_t is_bcast = B_FALSE; 873 boolean_t is_mcast = B_FALSE; 874 vgen_stats_t *statsp = &ldcp->ldc_stats; 875 lane_t *lane_in = &ldcp->lane_in; 876 lane_t *lane_out = &ldcp->lane_out; 877 dring_info_t *dp = lane_in->dringp; 878 vsw_t *vswp = ldcp->ldc_vswp; 879 880 if ((!(lane_in->lstate & VSW_LANE_ACTIVE)) || 881 (ldcp->ldc_status != LDC_UP) || (ldcp->ldc_handle == NULL)) { 882 DWARN(vswp, "%s(%lld) status(%d) lstate(0x%llx), dropping " 883 "packet\n", __func__, ldcp->ldc_id, ldcp->ldc_status, 884 lane_in->lstate); 885 statsp->oerrors++; 886 return (LDC_TX_FAILURE); 887 } 888 889 if (dp == NULL) { 890 DERR(vswp, "%s(%lld): no dring for outbound lane on" 891 " channel %d", __func__, ldcp->ldc_id, ldcp->ldc_id); 892 statsp->oerrors++; 893 return (LDC_TX_FAILURE); 894 } 895 pub_addr = dp->pub_addr; 896 897 size = msgsize(mp); 898 899 /* 900 * Note: In RxDringData mode, lane_in is associated with transmit and 901 * lane_out is associated with receive. However, we still keep the 902 * negotiated mtu in lane_out (our exported attributes). 903 */ 904 if (size > (size_t)lane_out->mtu) { 905 DERR(vswp, "%s(%lld) invalid size (%ld)\n", __func__, 906 ldcp->ldc_id, size); 907 statsp->oerrors++; 908 return (LDC_TX_FAILURE); 909 } 910 911 if (size < ETHERMIN) 912 size = ETHERMIN; 913 914 ehp = (struct ether_header *)mp->b_rptr; 915 is_bcast = IS_BROADCAST(ehp); 916 is_mcast = IS_MULTICAST(ehp); 917 918 /* 919 * Setup on_trap() protection before accessing shared memory areas 920 * (descriptor and data buffer). Note that we enable this protection a 921 * little early and turn it off slightly later, than keeping it enabled 922 * strictly at the points in code below where the descriptor and data 923 * buffer are accessed. This is done for performance reasons: 924 * (a) to avoid calling the trap protection code while holding mutex. 925 * (b) to avoid multiple on/off steps for descriptor and data accesses. 926 */ 927 rv = LDC_ON_TRAP(&otd); 928 if (rv != 0) { 929 /* 930 * Data access fault occured down the code path below while 931 * accessing either the descriptor or the data buffer. Release 932 * any locks that we might have acquired in the code below and 933 * return failure. 934 */ 935 DERR(vswp, "%s(%lld) data access fault occured\n", 936 __func__, ldcp->ldc_id); 937 statsp->oerrors++; 938 if (mutex_owned(&dp->txlock)) { 939 mutex_exit(&dp->txlock); 940 } 941 if (mutex_owned(&dp->restart_lock)) { 942 mutex_exit(&dp->restart_lock); 943 } 944 goto dringsend_shm_exit; 945 } 946 947 /* 948 * Allocate a descriptor 949 */ 950 mutex_enter(&dp->txlock); 951 txi = next_txi = dp->next_txi; 952 INCR_TXI(dp, next_txi); 953 txdp = &(pub_addr[txi]); 954 if (txdp->dstate != VIO_DESC_DONE) { /* out of descriptors */ 955 statsp->tx_no_desc++; 956 mutex_exit(&dp->txlock); 957 (void) LDC_NO_TRAP(); 958 return (LDC_TX_NORESOURCES); 959 } else { 960 txdp->dstate = VIO_DESC_INITIALIZING; 961 } 962 963 /* Update descriptor ring index */ 964 dp->next_txi = next_txi; 965 mutex_exit(&dp->txlock); 966 967 /* Ensure load ordering of dstate (above) and data_buf_offset. */ 968 MEMBAR_CONSUMER(); 969 970 /* Get the offset of the buffer to be used */ 971 buf_offset = txdp->data_buf_offset; 972 973 /* Access the buffer using the offset */ 974 dst = (caddr_t)dp->data_addr + buf_offset; 975 976 /* Copy data into mapped transmit buffer */ 977 for (bp = mp; bp != NULL; bp = bp->b_cont) { 978 mblksz = MBLKL(bp); 979 bcopy(bp->b_rptr, dst, mblksz); 980 dst += mblksz; 981 } 982 983 /* Set the size of data in the descriptor */ 984 txdp->nbytes = size; 985 986 /* 987 * Ensure store ordering of nbytes and dstate (below); so that the peer 988 * sees the right nbytes value after it checks that the dstate is READY. 989 */ 990 MEMBAR_PRODUCER(); 991 992 mutex_enter(&dp->restart_lock); 993 994 ASSERT(txdp->dstate == VIO_DESC_INITIALIZING); 995 996 /* Mark the descriptor ready */ 997 txdp->dstate = VIO_DESC_READY; 998 999 /* Check if peer needs wake up (handled below) */ 1000 if (dp->restart_reqd == B_TRUE && dp->restart_peer_txi == txi) { 1001 dp->restart_reqd = B_FALSE; 1002 resched_peer = B_TRUE; 1003 } 1004 1005 /* Update tx stats */ 1006 statsp->opackets++; 1007 statsp->obytes += size; 1008 if (is_bcast) 1009 statsp->brdcstxmt++; 1010 else if (is_mcast) 1011 statsp->multixmt++; 1012 1013 mutex_exit(&dp->restart_lock); 1014 1015 /* 1016 * We are done accessing shared memory; clear trap protection. 1017 */ 1018 (void) LDC_NO_TRAP(); 1019 1020 /* 1021 * Need to wake up the peer ? 1022 */ 1023 if (resched_peer == B_TRUE) { 1024 msgp->tag.vio_msgtype = VIO_TYPE_DATA; 1025 msgp->tag.vio_subtype = VIO_SUBTYPE_INFO; 1026 msgp->tag.vio_subtype_env = VIO_DRING_DATA; 1027 msgp->tag.vio_sid = ldcp->local_session; 1028 msgp->dring_ident = lane_out->dringp->ident; 1029 msgp->start_idx = txi; 1030 msgp->end_idx = -1; 1031 1032 rv = vsw_send_msg_shm(ldcp, (void *)msgp, sizeof (*msgp), 1033 B_FALSE); 1034 if (rv != 0) { 1035 /* error: drop the packet */ 1036 DERR(vswp, "%s(%lld) failed sending dringdata msg\n", 1037 __func__, ldcp->ldc_id); 1038 mutex_enter(&dp->restart_lock); 1039 statsp->oerrors++; 1040 dp->restart_reqd = B_TRUE; 1041 mutex_exit(&dp->restart_lock); 1042 } 1043 statsp->dring_data_msgs_sent++; 1044 } 1045 1046 dringsend_shm_exit: 1047 if (rv == ECONNRESET || rv == EACCES) { 1048 vsw_process_conn_evt(ldcp, VSW_CONN_RESET); 1049 } 1050 return (LDC_TX_SUCCESS); 1051 } 1052 1053 void 1054 vsw_process_dringdata_shm(void *arg, void *dpkt) 1055 { 1056 vsw_ldc_t *ldcp = arg; 1057 vsw_t *vswp = ldcp->ldc_vswp; 1058 vio_dring_msg_t *dring_pkt = dpkt; 1059 1060 switch (dring_pkt->tag.vio_subtype) { 1061 case VIO_SUBTYPE_INFO: 1062 D2(vswp, "%s(%lld): VIO_SUBTYPE_INFO", __func__, ldcp->ldc_id); 1063 vsw_process_dringdata_info_shm(ldcp, dring_pkt); 1064 break; 1065 1066 case VIO_SUBTYPE_ACK: 1067 D2(vswp, "%s(%lld): VIO_SUBTYPE_ACK", __func__, ldcp->ldc_id); 1068 vsw_process_dringdata_ack_shm(ldcp, dring_pkt); 1069 break; 1070 1071 case VIO_SUBTYPE_NACK: 1072 DWARN(vswp, "%s(%lld): VIO_SUBTYPE_NACK", 1073 __func__, ldcp->ldc_id); 1074 /* 1075 * Something is badly wrong if we are getting NACK's 1076 * for our data pkts. So reset the channel. 1077 */ 1078 vsw_process_conn_evt(ldcp, VSW_CONN_RESTART); 1079 break; 1080 1081 default: 1082 DERR(vswp, "%s(%lld): Unknown vio_subtype %x\n", __func__, 1083 ldcp->ldc_id, dring_pkt->tag.vio_subtype); 1084 } 1085 } 1086 1087 static void 1088 vsw_process_dringdata_info_shm(vsw_ldc_t *ldcp, vio_dring_msg_t *msg) 1089 { 1090 dring_info_t *dp = ldcp->lane_in.dringp; 1091 vsw_t *vswp = ldcp->ldc_vswp; 1092 vgen_stats_t *statsp = &ldcp->ldc_stats; 1093 1094 if (dp->ident != msg->dring_ident) { 1095 /* drop the message */ 1096 DERR(vswp, "%s(%lld): Invalid dring ident 0x%llx", 1097 __func__, ldcp->ldc_id, msg->dring_ident); 1098 return; 1099 } 1100 1101 statsp->dring_data_msgs_rcvd++; 1102 1103 /* 1104 * Wake up the rcv worker thread to process the rx dring. 1105 */ 1106 ASSERT(MUTEX_HELD(&ldcp->ldc_cblock)); 1107 mutex_exit(&ldcp->ldc_cblock); 1108 mutex_enter(&ldcp->rcv_thr_lock); 1109 if (!(ldcp->rcv_thr_flags & VSW_WTHR_DATARCVD)) { 1110 ldcp->rcv_thr_flags |= VSW_WTHR_DATARCVD; 1111 cv_signal(&ldcp->rcv_thr_cv); 1112 } 1113 mutex_exit(&ldcp->rcv_thr_lock); 1114 mutex_enter(&ldcp->ldc_cblock); 1115 } 1116 1117 static void 1118 vsw_process_dringdata_ack_shm(vsw_ldc_t *ldcp, vio_dring_msg_t *msg) 1119 { 1120 dring_info_t *dp; 1121 uint32_t start; 1122 int32_t end; 1123 int rv; 1124 on_trap_data_t otd; 1125 uint32_t txi; 1126 vnet_rx_dringdata_desc_t *txdp; 1127 vnet_rx_dringdata_desc_t *pub_addr; 1128 boolean_t ready_txd = B_FALSE; 1129 vsw_t *vswp = ldcp->ldc_vswp; 1130 vgen_stats_t *statsp = &ldcp->ldc_stats; 1131 1132 dp = ldcp->lane_in.dringp; 1133 start = msg->start_idx; 1134 end = msg->end_idx; 1135 pub_addr = dp->pub_addr; 1136 1137 /* 1138 * In RxDringData mode (v1.6), start index of -1 can be used by the 1139 * peer to indicate that it is unspecified. However, the end index 1140 * must be set correctly indicating the last descriptor index processed. 1141 */ 1142 if (((start != VNET_START_IDX_UNSPEC) && !(CHECK_TXI(dp, start))) || 1143 !(CHECK_TXI(dp, end))) { 1144 /* drop the message if invalid index */ 1145 DWARN(vswp, "%s(%lld): Invalid Tx ack start(%d) or end(%d)\n", 1146 __func__, ldcp->ldc_id, start, end); 1147 return; 1148 } 1149 1150 /* Validate dring_ident */ 1151 if (msg->dring_ident != ldcp->lane_out.dringp->ident) { 1152 /* invalid dring_ident, drop the msg */ 1153 DWARN(vswp, "%s(%lld): Invalid dring ident 0x%x\n", 1154 __func__, ldcp->ldc_id, msg->dring_ident); 1155 return; 1156 } 1157 statsp->dring_data_acks_rcvd++; 1158 1159 if (msg->dring_process_state != VIO_DP_STOPPED) { 1160 /* 1161 * Receiver continued processing 1162 * dring after sending us the ack. 1163 */ 1164 return; 1165 } 1166 1167 statsp->dring_stopped_acks_rcvd++; 1168 1169 /* 1170 * Setup on_trap() protection before accessing dring shared memory area. 1171 */ 1172 rv = LDC_ON_TRAP(&otd); 1173 if (rv != 0) { 1174 /* 1175 * Data access fault occured down the code path below while 1176 * accessing the descriptors. Release any locks that we might 1177 * have acquired in the code below and return failure. 1178 */ 1179 if (mutex_owned(&dp->restart_lock)) { 1180 mutex_exit(&dp->restart_lock); 1181 } 1182 return; 1183 } 1184 1185 /* 1186 * Determine if there are any pending tx descriptors ready to be 1187 * processed by the receiver(peer) and if so, send a message to the 1188 * peer to restart receiving. 1189 */ 1190 mutex_enter(&dp->restart_lock); 1191 1192 ready_txd = B_FALSE; 1193 txi = end; 1194 INCR_TXI(dp, txi); 1195 txdp = &pub_addr[txi]; 1196 if (txdp->dstate == VIO_DESC_READY) { 1197 ready_txd = B_TRUE; 1198 } 1199 1200 /* 1201 * We are done accessing shared memory; clear trap protection. 1202 */ 1203 (void) LDC_NO_TRAP(); 1204 1205 if (ready_txd == B_FALSE) { 1206 /* 1207 * No ready tx descriptors. Set the flag to send a message to 1208 * the peer when tx descriptors are ready in transmit routine. 1209 */ 1210 dp->restart_reqd = B_TRUE; 1211 dp->restart_peer_txi = txi; 1212 mutex_exit(&dp->restart_lock); 1213 return; 1214 } 1215 1216 /* 1217 * We have some tx descriptors ready to be processed by the receiver. 1218 * Send a dring data message to the peer to restart processing. 1219 */ 1220 dp->restart_reqd = B_FALSE; 1221 mutex_exit(&dp->restart_lock); 1222 1223 msg->tag.vio_msgtype = VIO_TYPE_DATA; 1224 msg->tag.vio_subtype = VIO_SUBTYPE_INFO; 1225 msg->tag.vio_subtype_env = VIO_DRING_DATA; 1226 msg->tag.vio_sid = ldcp->local_session; 1227 msg->dring_ident = ldcp->lane_out.dringp->ident; 1228 msg->start_idx = txi; 1229 msg->end_idx = -1; 1230 rv = vsw_send_msg_shm(ldcp, (void *)msg, 1231 sizeof (vio_dring_msg_t), B_FALSE); 1232 statsp->dring_data_msgs_sent++; 1233 if (rv != 0) { 1234 mutex_enter(&dp->restart_lock); 1235 dp->restart_reqd = B_TRUE; 1236 mutex_exit(&dp->restart_lock); 1237 } 1238 1239 if (rv == ECONNRESET) { 1240 vsw_process_conn_evt(ldcp, VSW_CONN_RESET); 1241 } 1242 } 1243 1244 /* 1245 * Send dring data msgs (info/ack/nack) over LDC. 1246 */ 1247 int 1248 vsw_send_msg_shm(vsw_ldc_t *ldcp, void *msgp, int size, boolean_t handle_reset) 1249 { 1250 int rv; 1251 int retries = vsw_wretries; 1252 size_t msglen = size; 1253 vsw_t *vswp = ldcp->ldc_vswp; 1254 vio_dring_msg_t *dmsg = (vio_dring_msg_t *)msgp; 1255 1256 D1(vswp, "vsw_send_msg (%lld) enter : sending %d bytes", 1257 ldcp->ldc_id, size); 1258 1259 dmsg->seq_num = atomic_inc_32_nv(&ldcp->dringdata_msgid); 1260 1261 do { 1262 msglen = size; 1263 rv = ldc_write(ldcp->ldc_handle, (caddr_t)msgp, &msglen); 1264 } while (rv == EWOULDBLOCK && --retries > 0); 1265 1266 if ((rv != 0) || (msglen != size)) { 1267 DERR(vswp, "vsw_send_msg_shm:ldc_write failed: " 1268 "chan(%lld) rv(%d) size (%d) msglen(%d)\n", 1269 ldcp->ldc_id, rv, size, msglen); 1270 ldcp->ldc_stats.oerrors++; 1271 } 1272 1273 /* 1274 * If channel has been reset we either handle it here or 1275 * simply report back that it has been reset and let caller 1276 * decide what to do. 1277 */ 1278 if (rv == ECONNRESET) { 1279 DWARN(vswp, "%s (%lld) channel reset", __func__, ldcp->ldc_id); 1280 1281 if (handle_reset) { 1282 vsw_process_conn_evt(ldcp, VSW_CONN_RESET); 1283 } 1284 } 1285 1286 return (rv); 1287 } 1288