Commit | Line | Data |
---|---|---|
e4c4625f JM |
1 | /* |
2 | * librpma_gpspm: IO engine that uses PMDK librpma to write data, | |
3 | * based on General Purpose Server Persistency Method | |
4 | * | |
fb04caf1 | 5 | * Copyright 2020-2022, Intel Corporation |
e4c4625f JM |
6 | * |
7 | * This program is free software; you can redistribute it and/or | |
8 | * modify it under the terms of the GNU General Public License, | |
9 | * version 2 as published by the Free Software Foundation.. | |
10 | * | |
11 | * This program is distributed in the hope that it will be useful, | |
12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
14 | * GNU General Public License for more details. | |
15 | */ | |
16 | ||
17 | #include "librpma_fio.h" | |
18 | ||
fb04caf1 KS |
19 | #ifdef CONFIG_LIBPMEM2_INSTALLED |
20 | #include <libpmem2.h> | |
21 | #else | |
e4c4625f | 22 | #include <libpmem.h> |
fb04caf1 | 23 | #endif |
e4c4625f JM |
24 | |
25 | /* Generated by the protocol buffer compiler from: librpma_gpspm_flush.proto */ | |
26 | #include "librpma_gpspm_flush.pb-c.h" | |
27 | ||
28 | #define MAX_MSG_SIZE (512) | |
29 | #define IO_U_BUF_LEN (2 * MAX_MSG_SIZE) | |
30 | #define SEND_OFFSET (0) | |
31 | #define RECV_OFFSET (SEND_OFFSET + MAX_MSG_SIZE) | |
32 | ||
33 | #define GPSPM_FLUSH_REQUEST__LAST \ | |
34 | { PROTOBUF_C_MESSAGE_INIT(&gpspm_flush_request__descriptor), 0, 0, 0 } | |
35 | ||
36 | /* | |
37 | * 'Flush_req_last' is the last flush request | |
38 | * the client has to send to server to indicate | |
39 | * that the client is done. | |
40 | */ | |
41 | static const GPSPMFlushRequest Flush_req_last = GPSPM_FLUSH_REQUEST__LAST; | |
42 | ||
43 | #define IS_NOT_THE_LAST_MESSAGE(flush_req) \ | |
44 | (flush_req->length != Flush_req_last.length || \ | |
45 | flush_req->offset != Flush_req_last.offset) | |
46 | ||
47 | /* client side implementation */ | |
48 | ||
49 | /* get next io_u message buffer in the round-robin fashion */ | |
50 | #define IO_U_NEXT_BUF_OFF_CLIENT(cd) \ | |
51 | (IO_U_BUF_LEN * ((cd->msg_curr++) % cd->msg_num)) | |
52 | ||
53 | struct client_data { | |
54 | /* memory for sending and receiving buffered */ | |
55 | char *io_us_msgs; | |
56 | ||
57 | /* resources for messaging buffer */ | |
58 | uint32_t msg_num; | |
59 | uint32_t msg_curr; | |
60 | struct rpma_mr_local *msg_mr; | |
61 | }; | |
62 | ||
63 | static inline int client_io_flush(struct thread_data *td, | |
64 | struct io_u *first_io_u, struct io_u *last_io_u, | |
65 | unsigned long long int len); | |
66 | ||
4ef7dd21 | 67 | static int client_get_io_u_index(struct ibv_wc *wc, unsigned int *io_u_index); |
e4c4625f JM |
68 | |
69 | static int client_init(struct thread_data *td) | |
70 | { | |
71 | struct librpma_fio_client_data *ccd; | |
72 | struct client_data *cd; | |
73 | uint32_t write_num; | |
74 | struct rpma_conn_cfg *cfg = NULL; | |
75 | int ret; | |
76 | ||
77 | /* | |
78 | * not supported: | |
79 | * - readwrite = read / trim / randread / randtrim / | |
80 | * / rw / randrw / trimwrite | |
81 | */ | |
82 | if (td_read(td) || td_trim(td)) { | |
83 | td_verror(td, EINVAL, "Not supported mode."); | |
84 | return -1; | |
85 | } | |
86 | ||
87 | /* allocate client's data */ | |
88 | cd = calloc(1, sizeof(*cd)); | |
89 | if (cd == NULL) { | |
90 | td_verror(td, errno, "calloc"); | |
91 | return -1; | |
92 | } | |
93 | ||
94 | /* | |
95 | * Calculate the required number of WRITEs and FLUSHes. | |
96 | * | |
97 | * Note: Each flush is a request (SEND) and response (RECV) pair. | |
98 | */ | |
99 | if (td_random(td)) { | |
100 | write_num = td->o.iodepth; /* WRITE * N */ | |
101 | cd->msg_num = td->o.iodepth; /* FLUSH * N */ | |
102 | } else { | |
103 | if (td->o.sync_io) { | |
104 | write_num = 1; /* WRITE */ | |
105 | cd->msg_num = 1; /* FLUSH */ | |
106 | } else { | |
107 | write_num = td->o.iodepth; /* WRITE * N */ | |
108 | /* | |
109 | * FLUSH * B where: | |
110 | * - B == ceil(iodepth / iodepth_batch) | |
111 | * which is the number of batches for N writes | |
112 | */ | |
113 | cd->msg_num = LIBRPMA_FIO_CEIL(td->o.iodepth, | |
114 | td->o.iodepth_batch); | |
115 | } | |
116 | } | |
117 | ||
118 | /* create a connection configuration object */ | |
119 | if ((ret = rpma_conn_cfg_new(&cfg))) { | |
120 | librpma_td_verror(td, ret, "rpma_conn_cfg_new"); | |
121 | goto err_free_cd; | |
122 | } | |
123 | ||
124 | /* | |
125 | * Calculate the required queue sizes where: | |
126 | * - the send queue (SQ) has to be big enough to accommodate | |
127 | * all io_us (WRITEs) and all flush requests (SENDs) | |
128 | * - the receive queue (RQ) has to be big enough to accommodate | |
129 | * all flush responses (RECVs) | |
130 | * - the completion queue (CQ) has to be big enough to accommodate all | |
131 | * success and error completions (sq_size + rq_size) | |
132 | */ | |
133 | if ((ret = rpma_conn_cfg_set_sq_size(cfg, write_num + cd->msg_num))) { | |
134 | librpma_td_verror(td, ret, "rpma_conn_cfg_set_sq_size"); | |
135 | goto err_cfg_delete; | |
136 | } | |
137 | if ((ret = rpma_conn_cfg_set_rq_size(cfg, cd->msg_num))) { | |
138 | librpma_td_verror(td, ret, "rpma_conn_cfg_set_rq_size"); | |
139 | goto err_cfg_delete; | |
140 | } | |
141 | if ((ret = rpma_conn_cfg_set_cq_size(cfg, write_num + cd->msg_num * 2))) { | |
142 | librpma_td_verror(td, ret, "rpma_conn_cfg_set_cq_size"); | |
143 | goto err_cfg_delete; | |
144 | } | |
145 | ||
146 | if (librpma_fio_client_init(td, cfg)) | |
147 | goto err_cfg_delete; | |
148 | ||
149 | ccd = td->io_ops_data; | |
150 | ||
151 | if (ccd->ws->direct_write_to_pmem && | |
152 | ccd->server_mr_flush_type == RPMA_FLUSH_TYPE_PERSISTENT && | |
153 | td->thread_number == 1) { | |
154 | /* XXX log_info mixes with the JSON output */ | |
155 | log_err( | |
156 | "Note: The server side supports Direct Write to PMem and it is equipped with PMem (direct_write_to_pmem).\n" | |
157 | "You can use librpma_client and librpma_server engines for better performance instead of GPSPM.\n"); | |
158 | } | |
159 | ||
160 | /* validate the server's RQ capacity */ | |
161 | if (cd->msg_num > ccd->ws->max_msg_num) { | |
162 | log_err( | |
163 | "server's RQ size (iodepth) too small to handle the client's workspace requirements (%u < %u)\n", | |
164 | ccd->ws->max_msg_num, cd->msg_num); | |
165 | goto err_cleanup_common; | |
166 | } | |
167 | ||
168 | if ((ret = rpma_conn_cfg_delete(&cfg))) { | |
169 | librpma_td_verror(td, ret, "rpma_conn_cfg_delete"); | |
170 | /* non fatal error - continue */ | |
171 | } | |
172 | ||
173 | ccd->flush = client_io_flush; | |
174 | ccd->get_io_u_index = client_get_io_u_index; | |
175 | ccd->client_data = cd; | |
176 | ||
177 | return 0; | |
178 | ||
179 | err_cleanup_common: | |
180 | librpma_fio_client_cleanup(td); | |
181 | ||
182 | err_cfg_delete: | |
183 | (void) rpma_conn_cfg_delete(&cfg); | |
184 | ||
185 | err_free_cd: | |
186 | free(cd); | |
187 | ||
188 | return -1; | |
189 | } | |
190 | ||
191 | static int client_post_init(struct thread_data *td) | |
192 | { | |
193 | struct librpma_fio_client_data *ccd = td->io_ops_data; | |
194 | struct client_data *cd = ccd->client_data; | |
195 | unsigned int io_us_msgs_size; | |
196 | int ret; | |
197 | ||
198 | /* message buffers initialization and registration */ | |
199 | io_us_msgs_size = cd->msg_num * IO_U_BUF_LEN; | |
200 | if ((ret = posix_memalign((void **)&cd->io_us_msgs, page_size, | |
201 | io_us_msgs_size))) { | |
202 | td_verror(td, ret, "posix_memalign"); | |
203 | return ret; | |
204 | } | |
205 | if ((ret = rpma_mr_reg(ccd->peer, cd->io_us_msgs, io_us_msgs_size, | |
206 | RPMA_MR_USAGE_SEND | RPMA_MR_USAGE_RECV, | |
207 | &cd->msg_mr))) { | |
208 | librpma_td_verror(td, ret, "rpma_mr_reg"); | |
209 | return ret; | |
210 | } | |
211 | ||
212 | return librpma_fio_client_post_init(td); | |
213 | } | |
214 | ||
215 | static void client_cleanup(struct thread_data *td) | |
216 | { | |
217 | struct librpma_fio_client_data *ccd = td->io_ops_data; | |
218 | struct client_data *cd; | |
219 | size_t flush_req_size; | |
220 | size_t io_u_buf_off; | |
221 | size_t send_offset; | |
222 | void *send_ptr; | |
223 | int ret; | |
224 | ||
225 | if (ccd == NULL) | |
226 | return; | |
227 | ||
228 | cd = ccd->client_data; | |
229 | if (cd == NULL) { | |
230 | librpma_fio_client_cleanup(td); | |
231 | return; | |
232 | } | |
233 | ||
234 | /* | |
235 | * Make sure all SEND completions are collected ergo there are free | |
236 | * slots in the SQ for the last SEND message. | |
237 | * | |
238 | * Note: If any operation will fail we still can send the termination | |
239 | * notice. | |
240 | */ | |
241 | (void) librpma_fio_client_io_complete_all_sends(td); | |
242 | ||
243 | /* prepare the last flush message and pack it to the send buffer */ | |
244 | flush_req_size = gpspm_flush_request__get_packed_size(&Flush_req_last); | |
245 | if (flush_req_size > MAX_MSG_SIZE) { | |
246 | log_err( | |
247 | "Packed flush request size is bigger than available send buffer space (%zu > %d\n", | |
248 | flush_req_size, MAX_MSG_SIZE); | |
249 | } else { | |
250 | io_u_buf_off = IO_U_NEXT_BUF_OFF_CLIENT(cd); | |
251 | send_offset = io_u_buf_off + SEND_OFFSET; | |
252 | send_ptr = cd->io_us_msgs + send_offset; | |
253 | (void) gpspm_flush_request__pack(&Flush_req_last, send_ptr); | |
254 | ||
255 | /* send the flush message */ | |
256 | if ((ret = rpma_send(ccd->conn, cd->msg_mr, send_offset, | |
257 | flush_req_size, RPMA_F_COMPLETION_ALWAYS, | |
258 | NULL))) | |
259 | librpma_td_verror(td, ret, "rpma_send"); | |
260 | ||
261 | ++ccd->op_send_posted; | |
262 | ||
263 | /* Wait for the SEND to complete */ | |
264 | (void) librpma_fio_client_io_complete_all_sends(td); | |
265 | } | |
266 | ||
267 | /* deregister the messaging buffer memory */ | |
268 | if ((ret = rpma_mr_dereg(&cd->msg_mr))) | |
269 | librpma_td_verror(td, ret, "rpma_mr_dereg"); | |
270 | ||
271 | free(ccd->client_data); | |
272 | ||
273 | librpma_fio_client_cleanup(td); | |
274 | } | |
275 | ||
276 | static inline int client_io_flush(struct thread_data *td, | |
277 | struct io_u *first_io_u, struct io_u *last_io_u, | |
278 | unsigned long long int len) | |
279 | { | |
280 | struct librpma_fio_client_data *ccd = td->io_ops_data; | |
281 | struct client_data *cd = ccd->client_data; | |
282 | size_t io_u_buf_off = IO_U_NEXT_BUF_OFF_CLIENT(cd); | |
283 | size_t send_offset = io_u_buf_off + SEND_OFFSET; | |
284 | size_t recv_offset = io_u_buf_off + RECV_OFFSET; | |
285 | void *send_ptr = cd->io_us_msgs + send_offset; | |
286 | void *recv_ptr = cd->io_us_msgs + recv_offset; | |
287 | GPSPMFlushRequest flush_req = GPSPM_FLUSH_REQUEST__INIT; | |
288 | size_t flush_req_size = 0; | |
289 | int ret; | |
290 | ||
291 | /* prepare a response buffer */ | |
292 | if ((ret = rpma_recv(ccd->conn, cd->msg_mr, recv_offset, MAX_MSG_SIZE, | |
293 | recv_ptr))) { | |
294 | librpma_td_verror(td, ret, "rpma_recv"); | |
295 | return -1; | |
296 | } | |
297 | ||
298 | /* prepare a flush message and pack it to a send buffer */ | |
299 | flush_req.offset = first_io_u->offset; | |
300 | flush_req.length = len; | |
301 | flush_req.op_context = last_io_u->index; | |
302 | flush_req_size = gpspm_flush_request__get_packed_size(&flush_req); | |
303 | if (flush_req_size > MAX_MSG_SIZE) { | |
304 | log_err( | |
305 | "Packed flush request size is bigger than available send buffer space (%" | |
306 | PRIu64 " > %d\n", flush_req_size, MAX_MSG_SIZE); | |
307 | return -1; | |
308 | } | |
309 | (void) gpspm_flush_request__pack(&flush_req, send_ptr); | |
310 | ||
311 | /* send the flush message */ | |
312 | if ((ret = rpma_send(ccd->conn, cd->msg_mr, send_offset, flush_req_size, | |
313 | RPMA_F_COMPLETION_ALWAYS, NULL))) { | |
314 | librpma_td_verror(td, ret, "rpma_send"); | |
315 | return -1; | |
316 | } | |
317 | ||
318 | ++ccd->op_send_posted; | |
319 | ||
320 | return 0; | |
321 | } | |
322 | ||
4ef7dd21 | 323 | static int client_get_io_u_index(struct ibv_wc *wc, unsigned int *io_u_index) |
e4c4625f JM |
324 | { |
325 | GPSPMFlushResponse *flush_resp; | |
326 | ||
4ef7dd21 | 327 | if (wc->opcode != IBV_WC_RECV) |
e4c4625f JM |
328 | return 0; |
329 | ||
330 | /* unpack a response from the received buffer */ | |
331 | flush_resp = gpspm_flush_response__unpack(NULL, | |
4ef7dd21 | 332 | wc->byte_len, (void *)wc->wr_id); |
e4c4625f JM |
333 | if (flush_resp == NULL) { |
334 | log_err("Cannot unpack the flush response buffer\n"); | |
335 | return -1; | |
336 | } | |
337 | ||
338 | memcpy(io_u_index, &flush_resp->op_context, sizeof(*io_u_index)); | |
339 | ||
340 | gpspm_flush_response__free_unpacked(flush_resp, NULL); | |
341 | ||
342 | return 1; | |
343 | } | |
344 | ||
345 | FIO_STATIC struct ioengine_ops ioengine_client = { | |
346 | .name = "librpma_gpspm_client", | |
347 | .version = FIO_IOOPS_VERSION, | |
348 | .init = client_init, | |
349 | .post_init = client_post_init, | |
350 | .get_file_size = librpma_fio_client_get_file_size, | |
351 | .open_file = librpma_fio_file_nop, | |
352 | .queue = librpma_fio_client_queue, | |
353 | .commit = librpma_fio_client_commit, | |
354 | .getevents = librpma_fio_client_getevents, | |
355 | .event = librpma_fio_client_event, | |
356 | .errdetails = librpma_fio_client_errdetails, | |
357 | .close_file = librpma_fio_file_nop, | |
358 | .cleanup = client_cleanup, | |
2b82135e | 359 | .flags = FIO_DISKLESSIO | FIO_ASYNCIO_SETS_ISSUE_TIME, |
e4c4625f JM |
360 | .options = librpma_fio_options, |
361 | .option_struct_size = sizeof(struct librpma_fio_options_values), | |
362 | }; | |
363 | ||
364 | /* server side implementation */ | |
365 | ||
366 | #define IO_U_BUFF_OFF_SERVER(i) (i * IO_U_BUF_LEN) | |
367 | ||
fb04caf1 KS |
368 | typedef void (*librpma_fio_persist_fn)(const void *ptr, size_t size); |
369 | ||
e4c4625f JM |
370 | struct server_data { |
371 | /* aligned td->orig_buffer */ | |
372 | char *orig_buffer_aligned; | |
373 | ||
374 | /* resources for messaging buffer from DRAM allocated by fio */ | |
375 | struct rpma_mr_local *msg_mr; | |
376 | ||
377 | uint32_t msg_sqe_available; /* # of free SQ slots */ | |
378 | ||
379 | /* in-memory queues */ | |
4ef7dd21 | 380 | struct ibv_wc *msgs_queued; |
e4c4625f | 381 | uint32_t msg_queued_nr; |
fb04caf1 KS |
382 | |
383 | librpma_fio_persist_fn persist; | |
e4c4625f JM |
384 | }; |
385 | ||
386 | static int server_init(struct thread_data *td) | |
387 | { | |
388 | struct librpma_fio_server_data *csd; | |
389 | struct server_data *sd; | |
390 | int ret = -1; | |
391 | ||
392 | if ((ret = librpma_fio_server_init(td))) | |
393 | return ret; | |
394 | ||
395 | csd = td->io_ops_data; | |
396 | ||
397 | /* allocate server's data */ | |
398 | sd = calloc(1, sizeof(*sd)); | |
399 | if (sd == NULL) { | |
400 | td_verror(td, errno, "calloc"); | |
401 | goto err_server_cleanup; | |
402 | } | |
403 | ||
404 | /* allocate in-memory queue */ | |
405 | sd->msgs_queued = calloc(td->o.iodepth, sizeof(*sd->msgs_queued)); | |
406 | if (sd->msgs_queued == NULL) { | |
407 | td_verror(td, errno, "calloc"); | |
408 | goto err_free_sd; | |
409 | } | |
410 | ||
fb04caf1 KS |
411 | #ifdef CONFIG_LIBPMEM2_INSTALLED |
412 | /* get libpmem2 persist function from pmem2_map */ | |
413 | sd->persist = pmem2_get_persist_fn(csd->mem.map); | |
414 | #else | |
415 | sd->persist = pmem_persist; | |
416 | #endif | |
417 | ||
e4c4625f JM |
418 | /* |
419 | * Assure a single io_u buffer can store both SEND and RECV messages and | |
420 | * an io_us buffer allocation is page-size-aligned which is required | |
421 | * to register for RDMA. User-provided values are intentionally ignored. | |
422 | */ | |
423 | td->o.max_bs[DDIR_READ] = IO_U_BUF_LEN; | |
424 | td->o.mem_align = page_size; | |
425 | ||
426 | csd->server_data = sd; | |
427 | ||
428 | return 0; | |
429 | ||
430 | err_free_sd: | |
431 | free(sd); | |
432 | ||
433 | err_server_cleanup: | |
434 | librpma_fio_server_cleanup(td); | |
435 | ||
436 | return -1; | |
437 | } | |
438 | ||
439 | static int server_post_init(struct thread_data *td) | |
440 | { | |
441 | struct librpma_fio_server_data *csd = td->io_ops_data; | |
442 | struct server_data *sd = csd->server_data; | |
443 | size_t io_us_size; | |
444 | size_t io_u_buflen; | |
445 | int ret; | |
446 | ||
447 | /* | |
448 | * td->orig_buffer is not aligned. The engine requires aligned io_us | |
fc002f14 | 449 | * so FIO aligns up the address using the formula below. |
e4c4625f JM |
450 | */ |
451 | sd->orig_buffer_aligned = PTR_ALIGN(td->orig_buffer, page_mask) + | |
452 | td->o.mem_align; | |
453 | ||
454 | /* | |
455 | * XXX | |
456 | * Each io_u message buffer contains recv and send messages. | |
457 | * Aligning each of those buffers may potentially give | |
458 | * some performance benefits. | |
459 | */ | |
460 | io_u_buflen = td_max_bs(td); | |
461 | ||
462 | /* check whether io_u buffer is big enough */ | |
463 | if (io_u_buflen < IO_U_BUF_LEN) { | |
464 | log_err( | |
465 | "blocksize too small to accommodate assumed maximal request/response pair size (%" PRIu64 " < %d)\n", | |
466 | io_u_buflen, IO_U_BUF_LEN); | |
467 | return -1; | |
468 | } | |
469 | ||
470 | /* | |
471 | * td->orig_buffer_size beside the space really consumed by io_us | |
472 | * has paddings which can be omitted for the memory registration. | |
473 | */ | |
474 | io_us_size = (unsigned long long)io_u_buflen * | |
475 | (unsigned long long)td->o.iodepth; | |
476 | ||
477 | if ((ret = rpma_mr_reg(csd->peer, sd->orig_buffer_aligned, io_us_size, | |
478 | RPMA_MR_USAGE_SEND | RPMA_MR_USAGE_RECV, | |
479 | &sd->msg_mr))) { | |
480 | librpma_td_verror(td, ret, "rpma_mr_reg"); | |
481 | return -1; | |
482 | } | |
483 | ||
484 | return 0; | |
485 | } | |
486 | ||
487 | static void server_cleanup(struct thread_data *td) | |
488 | { | |
489 | struct librpma_fio_server_data *csd = td->io_ops_data; | |
490 | struct server_data *sd; | |
491 | int ret; | |
492 | ||
493 | if (csd == NULL) | |
494 | return; | |
495 | ||
496 | sd = csd->server_data; | |
497 | ||
498 | if (sd != NULL) { | |
499 | /* rpma_mr_dereg(messaging buffer from DRAM) */ | |
500 | if ((ret = rpma_mr_dereg(&sd->msg_mr))) | |
501 | librpma_td_verror(td, ret, "rpma_mr_dereg"); | |
502 | ||
503 | free(sd->msgs_queued); | |
504 | free(sd); | |
505 | } | |
506 | ||
507 | librpma_fio_server_cleanup(td); | |
508 | } | |
509 | ||
510 | static int prepare_connection(struct thread_data *td, | |
511 | struct rpma_conn_req *conn_req) | |
512 | { | |
513 | struct librpma_fio_server_data *csd = td->io_ops_data; | |
514 | struct server_data *sd = csd->server_data; | |
515 | int ret; | |
516 | int i; | |
517 | ||
518 | /* prepare buffers for a flush requests */ | |
519 | sd->msg_sqe_available = td->o.iodepth; | |
520 | for (i = 0; i < td->o.iodepth; i++) { | |
521 | size_t offset_recv_msg = IO_U_BUFF_OFF_SERVER(i) + RECV_OFFSET; | |
522 | if ((ret = rpma_conn_req_recv(conn_req, sd->msg_mr, | |
523 | offset_recv_msg, MAX_MSG_SIZE, | |
524 | (const void *)(uintptr_t)i))) { | |
525 | librpma_td_verror(td, ret, "rpma_conn_req_recv"); | |
526 | return ret; | |
527 | } | |
528 | } | |
529 | ||
530 | return 0; | |
531 | } | |
532 | ||
533 | static int server_open_file(struct thread_data *td, struct fio_file *f) | |
534 | { | |
535 | struct librpma_fio_server_data *csd = td->io_ops_data; | |
536 | struct rpma_conn_cfg *cfg = NULL; | |
537 | uint16_t max_msg_num = td->o.iodepth; | |
538 | int ret; | |
539 | ||
540 | csd->prepare_connection = prepare_connection; | |
541 | ||
542 | /* create a connection configuration object */ | |
543 | if ((ret = rpma_conn_cfg_new(&cfg))) { | |
544 | librpma_td_verror(td, ret, "rpma_conn_cfg_new"); | |
545 | return -1; | |
546 | } | |
547 | ||
548 | /* | |
549 | * Calculate the required queue sizes where: | |
550 | * - the send queue (SQ) has to be big enough to accommodate | |
551 | * all possible flush requests (SENDs) | |
552 | * - the receive queue (RQ) has to be big enough to accommodate | |
553 | * all flush responses (RECVs) | |
554 | * - the completion queue (CQ) has to be big enough to accommodate | |
555 | * all success and error completions (sq_size + rq_size) | |
556 | */ | |
557 | if ((ret = rpma_conn_cfg_set_sq_size(cfg, max_msg_num))) { | |
558 | librpma_td_verror(td, ret, "rpma_conn_cfg_set_sq_size"); | |
559 | goto err_cfg_delete; | |
560 | } | |
561 | if ((ret = rpma_conn_cfg_set_rq_size(cfg, max_msg_num))) { | |
562 | librpma_td_verror(td, ret, "rpma_conn_cfg_set_rq_size"); | |
563 | goto err_cfg_delete; | |
564 | } | |
565 | if ((ret = rpma_conn_cfg_set_cq_size(cfg, max_msg_num * 2))) { | |
566 | librpma_td_verror(td, ret, "rpma_conn_cfg_set_cq_size"); | |
567 | goto err_cfg_delete; | |
568 | } | |
569 | ||
570 | ret = librpma_fio_server_open_file(td, f, cfg); | |
571 | ||
572 | err_cfg_delete: | |
573 | (void) rpma_conn_cfg_delete(&cfg); | |
574 | ||
575 | return ret; | |
576 | } | |
577 | ||
4ef7dd21 | 578 | static int server_qe_process(struct thread_data *td, struct ibv_wc *wc) |
e4c4625f JM |
579 | { |
580 | struct librpma_fio_server_data *csd = td->io_ops_data; | |
581 | struct server_data *sd = csd->server_data; | |
582 | GPSPMFlushRequest *flush_req; | |
583 | GPSPMFlushResponse flush_resp = GPSPM_FLUSH_RESPONSE__INIT; | |
584 | size_t flush_resp_size = 0; | |
585 | size_t send_buff_offset; | |
586 | size_t recv_buff_offset; | |
587 | size_t io_u_buff_offset; | |
588 | void *send_buff_ptr; | |
589 | void *recv_buff_ptr; | |
590 | void *op_ptr; | |
591 | int msg_index; | |
592 | int ret; | |
593 | ||
594 | /* calculate SEND/RECV pair parameters */ | |
4ef7dd21 | 595 | msg_index = (int)(uintptr_t)wc->wr_id; |
e4c4625f JM |
596 | io_u_buff_offset = IO_U_BUFF_OFF_SERVER(msg_index); |
597 | send_buff_offset = io_u_buff_offset + SEND_OFFSET; | |
598 | recv_buff_offset = io_u_buff_offset + RECV_OFFSET; | |
599 | send_buff_ptr = sd->orig_buffer_aligned + send_buff_offset; | |
600 | recv_buff_ptr = sd->orig_buffer_aligned + recv_buff_offset; | |
601 | ||
602 | /* unpack a flush request from the received buffer */ | |
4ef7dd21 | 603 | flush_req = gpspm_flush_request__unpack(NULL, wc->byte_len, |
e4c4625f JM |
604 | recv_buff_ptr); |
605 | if (flush_req == NULL) { | |
606 | log_err("cannot unpack the flush request buffer\n"); | |
607 | goto err_terminate; | |
608 | } | |
609 | ||
610 | if (IS_NOT_THE_LAST_MESSAGE(flush_req)) { | |
611 | op_ptr = csd->ws_ptr + flush_req->offset; | |
fb04caf1 | 612 | sd->persist(op_ptr, flush_req->length); |
e4c4625f JM |
613 | } else { |
614 | /* | |
615 | * This is the last message - the client is done. | |
616 | */ | |
617 | gpspm_flush_request__free_unpacked(flush_req, NULL); | |
618 | td->done = true; | |
619 | return 0; | |
620 | } | |
621 | ||
622 | /* initiate the next receive operation */ | |
623 | if ((ret = rpma_recv(csd->conn, sd->msg_mr, recv_buff_offset, | |
624 | MAX_MSG_SIZE, | |
625 | (const void *)(uintptr_t)msg_index))) { | |
626 | librpma_td_verror(td, ret, "rpma_recv"); | |
627 | goto err_free_unpacked; | |
628 | } | |
629 | ||
630 | /* prepare a flush response and pack it to a send buffer */ | |
631 | flush_resp.op_context = flush_req->op_context; | |
632 | flush_resp_size = gpspm_flush_response__get_packed_size(&flush_resp); | |
633 | if (flush_resp_size > MAX_MSG_SIZE) { | |
634 | log_err( | |
635 | "Size of the packed flush response is bigger than the available space of the send buffer (%" | |
636 | PRIu64 " > %i\n", flush_resp_size, MAX_MSG_SIZE); | |
637 | goto err_free_unpacked; | |
638 | } | |
639 | ||
640 | (void) gpspm_flush_response__pack(&flush_resp, send_buff_ptr); | |
641 | ||
642 | /* send the flush response */ | |
643 | if ((ret = rpma_send(csd->conn, sd->msg_mr, send_buff_offset, | |
644 | flush_resp_size, RPMA_F_COMPLETION_ALWAYS, NULL))) { | |
645 | librpma_td_verror(td, ret, "rpma_send"); | |
646 | goto err_free_unpacked; | |
647 | } | |
648 | --sd->msg_sqe_available; | |
649 | ||
650 | gpspm_flush_request__free_unpacked(flush_req, NULL); | |
651 | ||
652 | return 0; | |
653 | ||
654 | err_free_unpacked: | |
655 | gpspm_flush_request__free_unpacked(flush_req, NULL); | |
656 | ||
657 | err_terminate: | |
658 | td->terminate = true; | |
659 | ||
660 | return -1; | |
661 | } | |
662 | ||
663 | static inline int server_queue_process(struct thread_data *td) | |
664 | { | |
665 | struct librpma_fio_server_data *csd = td->io_ops_data; | |
666 | struct server_data *sd = csd->server_data; | |
667 | int ret; | |
668 | int i; | |
669 | ||
670 | /* min(# of queue entries, # of SQ entries available) */ | |
671 | uint32_t qes_to_process = min(sd->msg_queued_nr, sd->msg_sqe_available); | |
672 | if (qes_to_process == 0) | |
673 | return 0; | |
674 | ||
675 | /* process queued completions */ | |
676 | for (i = 0; i < qes_to_process; ++i) { | |
677 | if ((ret = server_qe_process(td, &sd->msgs_queued[i]))) | |
678 | return ret; | |
679 | } | |
680 | ||
681 | /* progress the queue */ | |
682 | for (i = 0; i < sd->msg_queued_nr - qes_to_process; ++i) { | |
683 | memcpy(&sd->msgs_queued[i], | |
684 | &sd->msgs_queued[qes_to_process + i], | |
685 | sizeof(sd->msgs_queued[i])); | |
686 | } | |
687 | ||
688 | sd->msg_queued_nr -= qes_to_process; | |
689 | ||
690 | return 0; | |
691 | } | |
692 | ||
693 | static int server_cmpl_process(struct thread_data *td) | |
694 | { | |
695 | struct librpma_fio_server_data *csd = td->io_ops_data; | |
696 | struct server_data *sd = csd->server_data; | |
4ef7dd21 | 697 | struct ibv_wc *wc = &sd->msgs_queued[sd->msg_queued_nr]; |
6a229978 | 698 | struct librpma_fio_options_values *o = td->eo; |
e4c4625f JM |
699 | int ret; |
700 | ||
4ef7dd21 | 701 | ret = rpma_cq_get_wc(csd->cq, 1, wc, NULL); |
e4c4625f | 702 | if (ret == RPMA_E_NO_COMPLETION) { |
d3061c18 LD |
703 | if (o->busy_wait_polling) |
704 | return 0; /* lack of completion is not an error */ | |
705 | ||
706 | ret = rpma_cq_wait(csd->cq); | |
707 | if (ret == RPMA_E_NO_COMPLETION) | |
708 | return 0; /* lack of completion is not an error */ | |
709 | if (ret) { | |
710 | librpma_td_verror(td, ret, "rpma_cq_wait"); | |
711 | goto err_terminate; | |
712 | } | |
713 | ||
714 | ret = rpma_cq_get_wc(csd->cq, 1, wc, NULL); | |
715 | if (ret == RPMA_E_NO_COMPLETION) | |
716 | return 0; /* lack of completion is not an error */ | |
717 | if (ret) { | |
718 | librpma_td_verror(td, ret, "rpma_cq_get_wc"); | |
719 | goto err_terminate; | |
6a229978 | 720 | } |
d3061c18 | 721 | } else if (ret) { |
4ef7dd21 | 722 | librpma_td_verror(td, ret, "rpma_cq_get_wc"); |
e4c4625f JM |
723 | goto err_terminate; |
724 | } | |
725 | ||
726 | /* validate the completion */ | |
4ef7dd21 | 727 | if (wc->status != IBV_WC_SUCCESS) |
e4c4625f JM |
728 | goto err_terminate; |
729 | ||
4ef7dd21 | 730 | if (wc->opcode == IBV_WC_RECV) |
e4c4625f | 731 | ++sd->msg_queued_nr; |
4ef7dd21 | 732 | else if (wc->opcode == IBV_WC_SEND) |
e4c4625f JM |
733 | ++sd->msg_sqe_available; |
734 | ||
735 | return 0; | |
736 | ||
737 | err_terminate: | |
738 | td->terminate = true; | |
739 | ||
740 | return -1; | |
741 | } | |
742 | ||
743 | static enum fio_q_status server_queue(struct thread_data *td, struct io_u *io_u) | |
744 | { | |
745 | do { | |
746 | if (server_cmpl_process(td)) | |
747 | return FIO_Q_BUSY; | |
748 | ||
749 | if (server_queue_process(td)) | |
750 | return FIO_Q_BUSY; | |
751 | ||
752 | } while (!td->done); | |
753 | ||
754 | return FIO_Q_COMPLETED; | |
755 | } | |
756 | ||
757 | FIO_STATIC struct ioengine_ops ioengine_server = { | |
758 | .name = "librpma_gpspm_server", | |
759 | .version = FIO_IOOPS_VERSION, | |
760 | .init = server_init, | |
761 | .post_init = server_post_init, | |
762 | .open_file = server_open_file, | |
763 | .close_file = librpma_fio_server_close_file, | |
764 | .queue = server_queue, | |
765 | .invalidate = librpma_fio_file_nop, | |
766 | .cleanup = server_cleanup, | |
767 | .flags = FIO_SYNCIO, | |
768 | .options = librpma_fio_options, | |
769 | .option_struct_size = sizeof(struct librpma_fio_options_values), | |
770 | }; | |
771 | ||
772 | /* register both engines */ | |
773 | ||
774 | static void fio_init fio_librpma_gpspm_register(void) | |
775 | { | |
776 | register_ioengine(&ioengine_client); | |
777 | register_ioengine(&ioengine_server); | |
778 | } | |
779 | ||
780 | static void fio_exit fio_librpma_gpspm_unregister(void) | |
781 | { | |
782 | unregister_ioengine(&ioengine_client); | |
783 | unregister_ioengine(&ioengine_server); | |
784 | } |