?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Linux 3.9-rc4
[linux-2.6/cjktty.git] / net / ceph / osd_client.c
blobd730dd4d8eb28bc421babb138f187605ee7c0ec6
1 #include <linux/ceph/ceph_debug.h>
2
3 #include <linux/module.h>
4 #include <linux/err.h>
5 #include <linux/highmem.h>
6 #include <linux/mm.h>
7 #include <linux/pagemap.h>
8 #include <linux/slab.h>
9 #include <linux/uaccess.h>
10 #ifdef CONFIG_BLOCK
11 #include <linux/bio.h>
12 #endif
13
14 #include <linux/ceph/libceph.h>
15 #include <linux/ceph/osd_client.h>
16 #include <linux/ceph/messenger.h>
17 #include <linux/ceph/decode.h>
18 #include <linux/ceph/auth.h>
19 #include <linux/ceph/pagelist.h>
20
21 #define OSD_OP_FRONT_LEN 4096
22 #define OSD_OPREPLY_FRONT_LEN 512
23
24 static const struct ceph_connection_operations osd_con_ops;
25
26 static void __send_queued(struct ceph_osd_client *osdc);
27 static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd);
28 static void __register_request(struct ceph_osd_client *osdc,
29 struct ceph_osd_request *req);
30 static void __unregister_linger_request(struct ceph_osd_client *osdc,
31 struct ceph_osd_request *req);
32 static void __send_request(struct ceph_osd_client *osdc,
33 struct ceph_osd_request *req);
34
35 static int op_has_extent(int op)
36 {
37 return (op == CEPH_OSD_OP_READ ||
38 op == CEPH_OSD_OP_WRITE);
39 }
40
41 /*
42 * Implement client access to distributed object storage cluster.
43 *
44 * All data objects are stored within a cluster/cloud of OSDs, or
45 * "object storage devices." (Note that Ceph OSDs have _nothing_ to
46 * do with the T10 OSD extensions to SCSI.) Ceph OSDs are simply
47 * remote daemons serving up and coordinating consistent and safe
48 * access to storage.
49 *
50 * Cluster membership and the mapping of data objects onto storage devices
51 * are described by the osd map.
52 *
53 * We keep track of pending OSD requests (read, write), resubmit
54 * requests to different OSDs when the cluster topology/data layout
55 * change, or retry the affected requests when the communications
56 * channel with an OSD is reset.
57 */
58
59 /*
60 * calculate the mapping of a file extent onto an object, and fill out the
61 * request accordingly. shorten extent as necessary if it crosses an
62 * object boundary.
63 *
64 * fill osd op in request message.
65 */
66 static int calc_layout(struct ceph_vino vino,
67 struct ceph_file_layout *layout,
68 u64 off, u64 *plen,
69 struct ceph_osd_request *req,
70 struct ceph_osd_req_op *op)
71 {
72 u64 orig_len = *plen;
73 u64 bno = 0;
74 u64 objoff = 0;
75 u64 objlen = 0;
76 int r;
77
78 /* object extent? */
79 r = ceph_calc_file_object_mapping(layout, off, orig_len, &bno,
80 &objoff, &objlen);
81 if (r < 0)
82 return r;
83 if (objlen < orig_len) {
84 *plen = objlen;
85 dout(" skipping last %llu, final file extent %llu~%llu\n",
86 orig_len - *plen, off, *plen);
87 }
88
89 if (op_has_extent(op->op)) {
90 u32 osize = le32_to_cpu(layout->fl_object_size);
91 op->extent.offset = objoff;
92 op->extent.length = objlen;
93 if (op->extent.truncate_size <= off - objoff) {
94 op->extent.truncate_size = 0;
95 } else {
96 op->extent.truncate_size -= off - objoff;
97 if (op->extent.truncate_size > osize)
98 op->extent.truncate_size = osize;
99 }
100 }
101 req->r_num_pages = calc_pages_for(off, *plen);
102 req->r_page_alignment = off & ~PAGE_MASK;
103 if (op->op == CEPH_OSD_OP_WRITE)
104 op->payload_len = *plen;
105
106 dout("calc_layout bno=%llx %llu~%llu (%d pages)\n",
107 bno, objoff, objlen, req->r_num_pages);
108
109 snprintf(req->r_oid, sizeof(req->r_oid), "%llx.%08llx", vino.ino, bno);
110 req->r_oid_len = strlen(req->r_oid);
111
112 return r;
113 }
114
115 /*
116 * requests
117 */
118 void ceph_osdc_release_request(struct kref *kref)
119 {
120 struct ceph_osd_request *req = container_of(kref,
121 struct ceph_osd_request,
122 r_kref);
123
124 if (req->r_request)
125 ceph_msg_put(req->r_request);
126 if (req->r_con_filling_msg) {
127 dout("%s revoking msg %p from con %p\n", __func__,
128 req->r_reply, req->r_con_filling_msg);
129 ceph_msg_revoke_incoming(req->r_reply);
130 req->r_con_filling_msg->ops->put(req->r_con_filling_msg);
131 req->r_con_filling_msg = NULL;
132 }
133 if (req->r_reply)
134 ceph_msg_put(req->r_reply);
135 if (req->r_own_pages)
136 ceph_release_page_vector(req->r_pages,
137 req->r_num_pages);
138 ceph_put_snap_context(req->r_snapc);
139 ceph_pagelist_release(&req->r_trail);
140 if (req->r_mempool)
141 mempool_free(req, req->r_osdc->req_mempool);
142 else
143 kfree(req);
144 }
145 EXPORT_SYMBOL(ceph_osdc_release_request);
146
147 struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc,
148 struct ceph_snap_context *snapc,
149 unsigned int num_ops,
150 bool use_mempool,
151 gfp_t gfp_flags)
152 {
153 struct ceph_osd_request *req;
154 struct ceph_msg *msg;
155 size_t msg_size;
156
157 msg_size = 4 + 4 + 8 + 8 + 4+8;
158 msg_size += 2 + 4 + 8 + 4 + 4; /* oloc */
159 msg_size += 1 + 8 + 4 + 4; /* pg_t */
160 msg_size += 4 + MAX_OBJ_NAME_SIZE;
161 msg_size += 2 + num_ops*sizeof(struct ceph_osd_op);
162 msg_size += 8; /* snapid */
163 msg_size += 8; /* snap_seq */
164 msg_size += 8 * (snapc ? snapc->num_snaps : 0); /* snaps */
165 msg_size += 4;
166
167 if (use_mempool) {
168 req = mempool_alloc(osdc->req_mempool, gfp_flags);
169 memset(req, 0, sizeof(*req));
170 } else {
171 req = kzalloc(sizeof(*req), gfp_flags);
172 }
173 if (req == NULL)
174 return NULL;
175
176 req->r_osdc = osdc;
177 req->r_mempool = use_mempool;
178
179 kref_init(&req->r_kref);
180 init_completion(&req->r_completion);
181 init_completion(&req->r_safe_completion);
182 RB_CLEAR_NODE(&req->r_node);
183 INIT_LIST_HEAD(&req->r_unsafe_item);
184 INIT_LIST_HEAD(&req->r_linger_item);
185 INIT_LIST_HEAD(&req->r_linger_osd);
186 INIT_LIST_HEAD(&req->r_req_lru_item);
187 INIT_LIST_HEAD(&req->r_osd_item);
188
189 /* create reply message */
190 if (use_mempool)
191 msg = ceph_msgpool_get(&osdc->msgpool_op_reply, 0);
192 else
193 msg = ceph_msg_new(CEPH_MSG_OSD_OPREPLY,
194 OSD_OPREPLY_FRONT_LEN, gfp_flags, true);
195 if (!msg) {
196 ceph_osdc_put_request(req);
197 return NULL;
198 }
199 req->r_reply = msg;
200
201 ceph_pagelist_init(&req->r_trail);
202
203 /* create request message; allow space for oid */
204 if (use_mempool)
205 msg = ceph_msgpool_get(&osdc->msgpool_op, 0);
206 else
207 msg = ceph_msg_new(CEPH_MSG_OSD_OP, msg_size, gfp_flags, true);
208 if (!msg) {
209 ceph_osdc_put_request(req);
210 return NULL;
211 }
212
213 memset(msg->front.iov_base, 0, msg->front.iov_len);
214
215 req->r_request = msg;
216
217 return req;
218 }
219 EXPORT_SYMBOL(ceph_osdc_alloc_request);
220
221 static void osd_req_encode_op(struct ceph_osd_request *req,
222 struct ceph_osd_op *dst,
223 struct ceph_osd_req_op *src)
224 {
225 dst->op = cpu_to_le16(src->op);
226
227 switch (src->op) {
228 case CEPH_OSD_OP_STAT:
229 break;
230 case CEPH_OSD_OP_READ:
231 case CEPH_OSD_OP_WRITE:
232 dst->extent.offset =
233 cpu_to_le64(src->extent.offset);
234 dst->extent.length =
235 cpu_to_le64(src->extent.length);
236 dst->extent.truncate_size =
237 cpu_to_le64(src->extent.truncate_size);
238 dst->extent.truncate_seq =
239 cpu_to_le32(src->extent.truncate_seq);
240 break;
241 case CEPH_OSD_OP_CALL:
242 dst->cls.class_len = src->cls.class_len;
243 dst->cls.method_len = src->cls.method_len;
244 dst->cls.indata_len = cpu_to_le32(src->cls.indata_len);
245
246 ceph_pagelist_append(&req->r_trail, src->cls.class_name,
247 src->cls.class_len);
248 ceph_pagelist_append(&req->r_trail, src->cls.method_name,
249 src->cls.method_len);
250 ceph_pagelist_append(&req->r_trail, src->cls.indata,
251 src->cls.indata_len);
252 break;
253 case CEPH_OSD_OP_STARTSYNC:
254 break;
255 case CEPH_OSD_OP_NOTIFY_ACK:
256 case CEPH_OSD_OP_WATCH:
257 dst->watch.cookie = cpu_to_le64(src->watch.cookie);
258 dst->watch.ver = cpu_to_le64(src->watch.ver);
259 dst->watch.flag = src->watch.flag;
260 break;
261 default:
262 pr_err("unrecognized osd opcode %d\n", dst->op);
263 WARN_ON(1);
264 break;
265 case CEPH_OSD_OP_MAPEXT:
266 case CEPH_OSD_OP_MASKTRUNC:
267 case CEPH_OSD_OP_SPARSE_READ:
268 case CEPH_OSD_OP_NOTIFY:
269 case CEPH_OSD_OP_ASSERT_VER:
270 case CEPH_OSD_OP_WRITEFULL:
271 case CEPH_OSD_OP_TRUNCATE:
272 case CEPH_OSD_OP_ZERO:
273 case CEPH_OSD_OP_DELETE:
274 case CEPH_OSD_OP_APPEND:
275 case CEPH_OSD_OP_SETTRUNC:
276 case CEPH_OSD_OP_TRIMTRUNC:
277 case CEPH_OSD_OP_TMAPUP:
278 case CEPH_OSD_OP_TMAPPUT:
279 case CEPH_OSD_OP_TMAPGET:
280 case CEPH_OSD_OP_CREATE:
281 case CEPH_OSD_OP_ROLLBACK:
282 case CEPH_OSD_OP_OMAPGETKEYS:
283 case CEPH_OSD_OP_OMAPGETVALS:
284 case CEPH_OSD_OP_OMAPGETHEADER:
285 case CEPH_OSD_OP_OMAPGETVALSBYKEYS:
286 case CEPH_OSD_OP_MODE_RD:
287 case CEPH_OSD_OP_OMAPSETVALS:
288 case CEPH_OSD_OP_OMAPSETHEADER:
289 case CEPH_OSD_OP_OMAPCLEAR:
290 case CEPH_OSD_OP_OMAPRMKEYS:
291 case CEPH_OSD_OP_OMAP_CMP:
292 case CEPH_OSD_OP_CLONERANGE:
293 case CEPH_OSD_OP_ASSERT_SRC_VERSION:
294 case CEPH_OSD_OP_SRC_CMPXATTR:
295 case CEPH_OSD_OP_GETXATTR:
296 case CEPH_OSD_OP_GETXATTRS:
297 case CEPH_OSD_OP_CMPXATTR:
298 case CEPH_OSD_OP_SETXATTR:
299 case CEPH_OSD_OP_SETXATTRS:
300 case CEPH_OSD_OP_RESETXATTRS:
301 case CEPH_OSD_OP_RMXATTR:
302 case CEPH_OSD_OP_PULL:
303 case CEPH_OSD_OP_PUSH:
304 case CEPH_OSD_OP_BALANCEREADS:
305 case CEPH_OSD_OP_UNBALANCEREADS:
306 case CEPH_OSD_OP_SCRUB:
307 case CEPH_OSD_OP_SCRUB_RESERVE:
308 case CEPH_OSD_OP_SCRUB_UNRESERVE:
309 case CEPH_OSD_OP_SCRUB_STOP:
310 case CEPH_OSD_OP_SCRUB_MAP:
311 case CEPH_OSD_OP_WRLOCK:
312 case CEPH_OSD_OP_WRUNLOCK:
313 case CEPH_OSD_OP_RDLOCK:
314 case CEPH_OSD_OP_RDUNLOCK:
315 case CEPH_OSD_OP_UPLOCK:
316 case CEPH_OSD_OP_DNLOCK:
317 case CEPH_OSD_OP_PGLS:
318 case CEPH_OSD_OP_PGLS_FILTER:
319 pr_err("unsupported osd opcode %s\n",
320 ceph_osd_op_name(dst->op));
321 WARN_ON(1);
322 break;
323 }
324 dst->payload_len = cpu_to_le32(src->payload_len);
325 }
326
327 /*
328 * build new request AND message
329 *
330 */
331 void ceph_osdc_build_request(struct ceph_osd_request *req,
332 u64 off, u64 len, unsigned int num_ops,
333 struct ceph_osd_req_op *src_ops,
334 struct ceph_snap_context *snapc, u64 snap_id,
335 struct timespec *mtime)
336 {
337 struct ceph_msg *msg = req->r_request;
338 struct ceph_osd_req_op *src_op;
339 void *p;
340 size_t msg_size;
341 int flags = req->r_flags;
342 u64 data_len;
343 int i;
344
345 req->r_num_ops = num_ops;
346 req->r_snapid = snap_id;
347 req->r_snapc = ceph_get_snap_context(snapc);
348
349 /* encode request */
350 msg->hdr.version = cpu_to_le16(4);
351
352 p = msg->front.iov_base;
353 ceph_encode_32(&p, 1); /* client_inc is always 1 */
354 req->r_request_osdmap_epoch = p;
355 p += 4;
356 req->r_request_flags = p;
357 p += 4;
358 if (req->r_flags & CEPH_OSD_FLAG_WRITE)
359 ceph_encode_timespec(p, mtime);
360 p += sizeof(struct ceph_timespec);
361 req->r_request_reassert_version = p;
362 p += sizeof(struct ceph_eversion); /* will get filled in */
363
364 /* oloc */
365 ceph_encode_8(&p, 4);
366 ceph_encode_8(&p, 4);
367 ceph_encode_32(&p, 8 + 4 + 4);
368 req->r_request_pool = p;
369 p += 8;
370 ceph_encode_32(&p, -1); /* preferred */
371 ceph_encode_32(&p, 0); /* key len */
372
373 ceph_encode_8(&p, 1);
374 req->r_request_pgid = p;
375 p += 8 + 4;
376 ceph_encode_32(&p, -1); /* preferred */
377
378 /* oid */
379 ceph_encode_32(&p, req->r_oid_len);
380 memcpy(p, req->r_oid, req->r_oid_len);
381 dout("oid '%.*s' len %d\n", req->r_oid_len, req->r_oid, req->r_oid_len);
382 p += req->r_oid_len;
383
384 /* ops */
385 ceph_encode_16(&p, num_ops);
386 src_op = src_ops;
387 req->r_request_ops = p;
388 for (i = 0; i < num_ops; i++, src_op++) {
389 osd_req_encode_op(req, p, src_op);
390 p += sizeof(struct ceph_osd_op);
391 }
392
393 /* snaps */
394 ceph_encode_64(&p, req->r_snapid);
395 ceph_encode_64(&p, req->r_snapc ? req->r_snapc->seq : 0);
396 ceph_encode_32(&p, req->r_snapc ? req->r_snapc->num_snaps : 0);
397 if (req->r_snapc) {
398 for (i = 0; i < snapc->num_snaps; i++) {
399 ceph_encode_64(&p, req->r_snapc->snaps[i]);
400 }
401 }
402
403 req->r_request_attempts = p;
404 p += 4;
405
406 data_len = req->r_trail.length;
407 if (flags & CEPH_OSD_FLAG_WRITE) {
408 req->r_request->hdr.data_off = cpu_to_le16(off);
409 data_len += len;
410 }
411 req->r_request->hdr.data_len = cpu_to_le32(data_len);
412 req->r_request->page_alignment = req->r_page_alignment;
413
414 BUG_ON(p > msg->front.iov_base + msg->front.iov_len);
415 msg_size = p - msg->front.iov_base;
416 msg->front.iov_len = msg_size;
417 msg->hdr.front_len = cpu_to_le32(msg_size);
418
419 dout("build_request msg_size was %d num_ops %d\n", (int)msg_size,
420 num_ops);
421 return;
422 }
423 EXPORT_SYMBOL(ceph_osdc_build_request);
424
425 /*
426 * build new request AND message, calculate layout, and adjust file
427 * extent as needed.
428 *
429 * if the file was recently truncated, we include information about its
430 * old and new size so that the object can be updated appropriately. (we
431 * avoid synchronously deleting truncated objects because it's slow.)
432 *
433 * if @do_sync, include a 'startsync' command so that the osd will flush
434 * data quickly.
435 */
436 struct ceph_osd_request *ceph_osdc_new_request(struct ceph_osd_client *osdc,
437 struct ceph_file_layout *layout,
438 struct ceph_vino vino,
439 u64 off, u64 *plen,
440 int opcode, int flags,
441 struct ceph_snap_context *snapc,
442 int do_sync,
443 u32 truncate_seq,
444 u64 truncate_size,
445 struct timespec *mtime,
446 bool use_mempool,
447 int page_align)
448 {
449 struct ceph_osd_req_op ops[2];
450 struct ceph_osd_request *req;
451 unsigned int num_op = 1;
452 int r;
453
454 memset(&ops, 0, sizeof ops);
455
456 ops[0].op = opcode;
457 ops[0].extent.truncate_seq = truncate_seq;
458 ops[0].extent.truncate_size = truncate_size;
459
460 if (do_sync) {
461 ops[1].op = CEPH_OSD_OP_STARTSYNC;
462 num_op++;
463 }
464
465 req = ceph_osdc_alloc_request(osdc, snapc, num_op, use_mempool,
466 GFP_NOFS);
467 if (!req)
468 return ERR_PTR(-ENOMEM);
469 req->r_flags = flags;
470
471 /* calculate max write size */
472 r = calc_layout(vino, layout, off, plen, req, ops);
473 if (r < 0)
474 return ERR_PTR(r);
475 req->r_file_layout = *layout; /* keep a copy */
476
477 /* in case it differs from natural (file) alignment that
478 calc_layout filled in for us */
479 req->r_num_pages = calc_pages_for(page_align, *plen);
480 req->r_page_alignment = page_align;
481
482 ceph_osdc_build_request(req, off, *plen, num_op, ops,
483 snapc, vino.snap, mtime);
484
485 return req;
486 }
487 EXPORT_SYMBOL(ceph_osdc_new_request);
488
489 /*
490 * We keep osd requests in an rbtree, sorted by ->r_tid.
491 */
492 static void __insert_request(struct ceph_osd_client *osdc,
493 struct ceph_osd_request *new)
494 {
495 struct rb_node **p = &osdc->requests.rb_node;
496 struct rb_node *parent = NULL;
497 struct ceph_osd_request *req = NULL;
498
499 while (*p) {
500 parent = *p;
501 req = rb_entry(parent, struct ceph_osd_request, r_node);
502 if (new->r_tid < req->r_tid)
503 p = &(*p)->rb_left;
504 else if (new->r_tid > req->r_tid)
505 p = &(*p)->rb_right;
506 else
507 BUG();
508 }
509
510 rb_link_node(&new->r_node, parent, p);
511 rb_insert_color(&new->r_node, &osdc->requests);
512 }
513
514 static struct ceph_osd_request *__lookup_request(struct ceph_osd_client *osdc,
515 u64 tid)
516 {
517 struct ceph_osd_request *req;
518 struct rb_node *n = osdc->requests.rb_node;
519
520 while (n) {
521 req = rb_entry(n, struct ceph_osd_request, r_node);
522 if (tid < req->r_tid)
523 n = n->rb_left;
524 else if (tid > req->r_tid)
525 n = n->rb_right;
526 else
527 return req;
528 }
529 return NULL;
530 }
531
532 static struct ceph_osd_request *
533 __lookup_request_ge(struct ceph_osd_client *osdc,
534 u64 tid)
535 {
536 struct ceph_osd_request *req;
537 struct rb_node *n = osdc->requests.rb_node;
538
539 while (n) {
540 req = rb_entry(n, struct ceph_osd_request, r_node);
541 if (tid < req->r_tid) {
542 if (!n->rb_left)
543 return req;
544 n = n->rb_left;
545 } else if (tid > req->r_tid) {
546 n = n->rb_right;
547 } else {
548 return req;
549 }
550 }
551 return NULL;
552 }
553
554 /*
555 * Resubmit requests pending on the given osd.
556 */
557 static void __kick_osd_requests(struct ceph_osd_client *osdc,
558 struct ceph_osd *osd)
559 {
560 struct ceph_osd_request *req, *nreq;
561 int err;
562
563 dout("__kick_osd_requests osd%d\n", osd->o_osd);
564 err = __reset_osd(osdc, osd);
565 if (err)
566 return;
567
568 list_for_each_entry(req, &osd->o_requests, r_osd_item) {
569 list_move(&req->r_req_lru_item, &osdc->req_unsent);
570 dout("requeued %p tid %llu osd%d\n", req, req->r_tid,
571 osd->o_osd);
572 if (!req->r_linger)
573 req->r_flags |= CEPH_OSD_FLAG_RETRY;
574 }
575
576 list_for_each_entry_safe(req, nreq, &osd->o_linger_requests,
577 r_linger_osd) {
578 /*
579 * reregister request prior to unregistering linger so
580 * that r_osd is preserved.
581 */
582 BUG_ON(!list_empty(&req->r_req_lru_item));
583 __register_request(osdc, req);
584 list_add(&req->r_req_lru_item, &osdc->req_unsent);
585 list_add(&req->r_osd_item, &req->r_osd->o_requests);
586 __unregister_linger_request(osdc, req);
587 dout("requeued lingering %p tid %llu osd%d\n", req, req->r_tid,
588 osd->o_osd);
589 }
590 }
591
592 /*
593 * If the osd connection drops, we need to resubmit all requests.
594 */
595 static void osd_reset(struct ceph_connection *con)
596 {
597 struct ceph_osd *osd = con->private;
598 struct ceph_osd_client *osdc;
599
600 if (!osd)
601 return;
602 dout("osd_reset osd%d\n", osd->o_osd);
603 osdc = osd->o_osdc;
604 down_read(&osdc->map_sem);
605 mutex_lock(&osdc->request_mutex);
606 __kick_osd_requests(osdc, osd);
607 __send_queued(osdc);
608 mutex_unlock(&osdc->request_mutex);
609 up_read(&osdc->map_sem);
610 }
611
612 /*
613 * Track open sessions with osds.
614 */
615 static struct ceph_osd *create_osd(struct ceph_osd_client *osdc, int onum)
616 {
617 struct ceph_osd *osd;
618
619 osd = kzalloc(sizeof(*osd), GFP_NOFS);
620 if (!osd)
621 return NULL;
622
623 atomic_set(&osd->o_ref, 1);
624 osd->o_osdc = osdc;
625 osd->o_osd = onum;
626 RB_CLEAR_NODE(&osd->o_node);
627 INIT_LIST_HEAD(&osd->o_requests);
628 INIT_LIST_HEAD(&osd->o_linger_requests);
629 INIT_LIST_HEAD(&osd->o_osd_lru);
630 osd->o_incarnation = 1;
631
632 ceph_con_init(&osd->o_con, osd, &osd_con_ops, &osdc->client->msgr);
633
634 INIT_LIST_HEAD(&osd->o_keepalive_item);
635 return osd;
636 }
637
638 static struct ceph_osd *get_osd(struct ceph_osd *osd)
639 {
640 if (atomic_inc_not_zero(&osd->o_ref)) {
641 dout("get_osd %p %d -> %d\n", osd, atomic_read(&osd->o_ref)-1,
642 atomic_read(&osd->o_ref));
643 return osd;
644 } else {
645 dout("get_osd %p FAIL\n", osd);
646 return NULL;
647 }
648 }
649
650 static void put_osd(struct ceph_osd *osd)
651 {
652 dout("put_osd %p %d -> %d\n", osd, atomic_read(&osd->o_ref),
653 atomic_read(&osd->o_ref) - 1);
654 if (atomic_dec_and_test(&osd->o_ref) && osd->o_auth.authorizer) {
655 struct ceph_auth_client *ac = osd->o_osdc->client->monc.auth;
656
657 if (ac->ops && ac->ops->destroy_authorizer)
658 ac->ops->destroy_authorizer(ac, osd->o_auth.authorizer);
659 kfree(osd);
660 }
661 }
662
663 /*
664 * remove an osd from our map
665 */
666 static void __remove_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
667 {
668 dout("__remove_osd %p\n", osd);
669 BUG_ON(!list_empty(&osd->o_requests));
670 rb_erase(&osd->o_node, &osdc->osds);
671 list_del_init(&osd->o_osd_lru);
672 ceph_con_close(&osd->o_con);
673 put_osd(osd);
674 }
675
676 static void remove_all_osds(struct ceph_osd_client *osdc)
677 {
678 dout("%s %p\n", __func__, osdc);
679 mutex_lock(&osdc->request_mutex);
680 while (!RB_EMPTY_ROOT(&osdc->osds)) {
681 struct ceph_osd *osd = rb_entry(rb_first(&osdc->osds),
682 struct ceph_osd, o_node);
683 __remove_osd(osdc, osd);
684 }
685 mutex_unlock(&osdc->request_mutex);
686 }
687
688 static void __move_osd_to_lru(struct ceph_osd_client *osdc,
689 struct ceph_osd *osd)
690 {
691 dout("__move_osd_to_lru %p\n", osd);
692 BUG_ON(!list_empty(&osd->o_osd_lru));
693 list_add_tail(&osd->o_osd_lru, &osdc->osd_lru);
694 osd->lru_ttl = jiffies + osdc->client->options->osd_idle_ttl * HZ;
695 }
696
697 static void __remove_osd_from_lru(struct ceph_osd *osd)
698 {
699 dout("__remove_osd_from_lru %p\n", osd);
700 if (!list_empty(&osd->o_osd_lru))
701 list_del_init(&osd->o_osd_lru);
702 }
703
704 static void remove_old_osds(struct ceph_osd_client *osdc)
705 {
706 struct ceph_osd *osd, *nosd;
707
708 dout("__remove_old_osds %p\n", osdc);
709 mutex_lock(&osdc->request_mutex);
710 list_for_each_entry_safe(osd, nosd, &osdc->osd_lru, o_osd_lru) {
711 if (time_before(jiffies, osd->lru_ttl))
712 break;
713 __remove_osd(osdc, osd);
714 }
715 mutex_unlock(&osdc->request_mutex);
716 }
717
718 /*
719 * reset osd connect
720 */
721 static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
722 {
723 struct ceph_entity_addr *peer_addr;
724
725 dout("__reset_osd %p osd%d\n", osd, osd->o_osd);
726 if (list_empty(&osd->o_requests) &&
727 list_empty(&osd->o_linger_requests)) {
728 __remove_osd(osdc, osd);
729
730 return -ENODEV;
731 }
732
733 peer_addr = &osdc->osdmap->osd_addr[osd->o_osd];
734 if (!memcmp(peer_addr, &osd->o_con.peer_addr, sizeof (*peer_addr)) &&
735 !ceph_con_opened(&osd->o_con)) {
736 struct ceph_osd_request *req;
737
738 dout(" osd addr hasn't changed and connection never opened,"
739 " letting msgr retry");
740 /* touch each r_stamp for handle_timeout()'s benfit */
741 list_for_each_entry(req, &osd->o_requests, r_osd_item)
742 req->r_stamp = jiffies;
743
744 return -EAGAIN;
745 }
746
747 ceph_con_close(&osd->o_con);
748 ceph_con_open(&osd->o_con, CEPH_ENTITY_TYPE_OSD, osd->o_osd, peer_addr);
749 osd->o_incarnation++;
750
751 return 0;
752 }
753
754 static void __insert_osd(struct ceph_osd_client *osdc, struct ceph_osd *new)
755 {
756 struct rb_node **p = &osdc->osds.rb_node;
757 struct rb_node *parent = NULL;
758 struct ceph_osd *osd = NULL;
759
760 dout("__insert_osd %p osd%d\n", new, new->o_osd);
761 while (*p) {
762 parent = *p;
763 osd = rb_entry(parent, struct ceph_osd, o_node);
764 if (new->o_osd < osd->o_osd)
765 p = &(*p)->rb_left;
766 else if (new->o_osd > osd->o_osd)
767 p = &(*p)->rb_right;
768 else
769 BUG();
770 }
771
772 rb_link_node(&new->o_node, parent, p);
773 rb_insert_color(&new->o_node, &osdc->osds);
774 }
775
776 static struct ceph_osd *__lookup_osd(struct ceph_osd_client *osdc, int o)
777 {
778 struct ceph_osd *osd;
779 struct rb_node *n = osdc->osds.rb_node;
780
781 while (n) {
782 osd = rb_entry(n, struct ceph_osd, o_node);
783 if (o < osd->o_osd)
784 n = n->rb_left;
785 else if (o > osd->o_osd)
786 n = n->rb_right;
787 else
788 return osd;
789 }
790 return NULL;
791 }
792
793 static void __schedule_osd_timeout(struct ceph_osd_client *osdc)
794 {
795 schedule_delayed_work(&osdc->timeout_work,
796 osdc->client->options->osd_keepalive_timeout * HZ);
797 }
798
799 static void __cancel_osd_timeout(struct ceph_osd_client *osdc)
800 {
801 cancel_delayed_work(&osdc->timeout_work);
802 }
803
804 /*
805 * Register request, assign tid. If this is the first request, set up
806 * the timeout event.
807 */
808 static void __register_request(struct ceph_osd_client *osdc,
809 struct ceph_osd_request *req)
810 {
811 req->r_tid = ++osdc->last_tid;
812 req->r_request->hdr.tid = cpu_to_le64(req->r_tid);
813 dout("__register_request %p tid %lld\n", req, req->r_tid);
814 __insert_request(osdc, req);
815 ceph_osdc_get_request(req);
816 osdc->num_requests++;
817 if (osdc->num_requests == 1) {
818 dout(" first request, scheduling timeout\n");
819 __schedule_osd_timeout(osdc);
820 }
821 }
822
823 static void register_request(struct ceph_osd_client *osdc,
824 struct ceph_osd_request *req)
825 {
826 mutex_lock(&osdc->request_mutex);
827 __register_request(osdc, req);
828 mutex_unlock(&osdc->request_mutex);
829 }
830
831 /*
832 * called under osdc->request_mutex
833 */
834 static void __unregister_request(struct ceph_osd_client *osdc,
835 struct ceph_osd_request *req)
836 {
837 if (RB_EMPTY_NODE(&req->r_node)) {
838 dout("__unregister_request %p tid %lld not registered\n",
839 req, req->r_tid);
840 return;
841 }
842
843 dout("__unregister_request %p tid %lld\n", req, req->r_tid);
844 rb_erase(&req->r_node, &osdc->requests);
845 osdc->num_requests--;
846
847 if (req->r_osd) {
848 /* make sure the original request isn't in flight. */
849 ceph_msg_revoke(req->r_request);
850
851 list_del_init(&req->r_osd_item);
852 if (list_empty(&req->r_osd->o_requests) &&
853 list_empty(&req->r_osd->o_linger_requests)) {
854 dout("moving osd to %p lru\n", req->r_osd);
855 __move_osd_to_lru(osdc, req->r_osd);
856 }
857 if (list_empty(&req->r_linger_item))
858 req->r_osd = NULL;
859 }
860
861 list_del_init(&req->r_req_lru_item);
862 ceph_osdc_put_request(req);
863
864 if (osdc->num_requests == 0) {
865 dout(" no requests, canceling timeout\n");
866 __cancel_osd_timeout(osdc);
867 }
868 }
869
870 /*
871 * Cancel a previously queued request message
872 */
873 static void __cancel_request(struct ceph_osd_request *req)
874 {
875 if (req->r_sent && req->r_osd) {
876 ceph_msg_revoke(req->r_request);
877 req->r_sent = 0;
878 }
879 }
880
881 static void __register_linger_request(struct ceph_osd_client *osdc,
882 struct ceph_osd_request *req)
883 {
884 dout("__register_linger_request %p\n", req);
885 list_add_tail(&req->r_linger_item, &osdc->req_linger);
886 if (req->r_osd)
887 list_add_tail(&req->r_linger_osd,
888 &req->r_osd->o_linger_requests);
889 }
890
891 static void __unregister_linger_request(struct ceph_osd_client *osdc,
892 struct ceph_osd_request *req)
893 {
894 dout("__unregister_linger_request %p\n", req);
895 list_del_init(&req->r_linger_item);
896 if (req->r_osd) {
897 list_del_init(&req->r_linger_osd);
898
899 if (list_empty(&req->r_osd->o_requests) &&
900 list_empty(&req->r_osd->o_linger_requests)) {
901 dout("moving osd to %p lru\n", req->r_osd);
902 __move_osd_to_lru(osdc, req->r_osd);
903 }
904 if (list_empty(&req->r_osd_item))
905 req->r_osd = NULL;
906 }
907 }
908
909 void ceph_osdc_unregister_linger_request(struct ceph_osd_client *osdc,
910 struct ceph_osd_request *req)
911 {
912 mutex_lock(&osdc->request_mutex);
913 if (req->r_linger) {
914 __unregister_linger_request(osdc, req);
915 ceph_osdc_put_request(req);
916 }
917 mutex_unlock(&osdc->request_mutex);
918 }
919 EXPORT_SYMBOL(ceph_osdc_unregister_linger_request);
920
921 void ceph_osdc_set_request_linger(struct ceph_osd_client *osdc,
922 struct ceph_osd_request *req)
923 {
924 if (!req->r_linger) {
925 dout("set_request_linger %p\n", req);
926 req->r_linger = 1;
927 /*
928 * caller is now responsible for calling
929 * unregister_linger_request
930 */
931 ceph_osdc_get_request(req);
932 }
933 }
934 EXPORT_SYMBOL(ceph_osdc_set_request_linger);
935
936 /*
937 * Pick an osd (the first 'up' osd in the pg), allocate the osd struct
938 * (as needed), and set the request r_osd appropriately. If there is
939 * no up osd, set r_osd to NULL. Move the request to the appropriate list
940 * (unsent, homeless) or leave on in-flight lru.
941 *
942 * Return 0 if unchanged, 1 if changed, or negative on error.
943 *
944 * Caller should hold map_sem for read and request_mutex.
945 */
946 static int __map_request(struct ceph_osd_client *osdc,
947 struct ceph_osd_request *req, int force_resend)
948 {
949 struct ceph_pg pgid;
950 int acting[CEPH_PG_MAX_SIZE];
951 int o = -1, num = 0;
952 int err;
953
954 dout("map_request %p tid %lld\n", req, req->r_tid);
955 err = ceph_calc_object_layout(&pgid, req->r_oid,
956 &req->r_file_layout, osdc->osdmap);
957 if (err) {
958 list_move(&req->r_req_lru_item, &osdc->req_notarget);
959 return err;
960 }
961 req->r_pgid = pgid;
962
963 err = ceph_calc_pg_acting(osdc->osdmap, pgid, acting);
964 if (err > 0) {
965 o = acting[0];
966 num = err;
967 }
968
969 if ((!force_resend &&
970 req->r_osd && req->r_osd->o_osd == o &&
971 req->r_sent >= req->r_osd->o_incarnation &&
972 req->r_num_pg_osds == num &&
973 memcmp(req->r_pg_osds, acting, sizeof(acting[0])*num) == 0) ||
974 (req->r_osd == NULL && o == -1))
975 return 0; /* no change */
976
977 dout("map_request tid %llu pgid %lld.%x osd%d (was osd%d)\n",
978 req->r_tid, pgid.pool, pgid.seed, o,
979 req->r_osd ? req->r_osd->o_osd : -1);
980
981 /* record full pg acting set */
982 memcpy(req->r_pg_osds, acting, sizeof(acting[0]) * num);
983 req->r_num_pg_osds = num;
984
985 if (req->r_osd) {
986 __cancel_request(req);
987 list_del_init(&req->r_osd_item);
988 req->r_osd = NULL;
989 }
990
991 req->r_osd = __lookup_osd(osdc, o);
992 if (!req->r_osd && o >= 0) {
993 err = -ENOMEM;
994 req->r_osd = create_osd(osdc, o);
995 if (!req->r_osd) {
996 list_move(&req->r_req_lru_item, &osdc->req_notarget);
997 goto out;
998 }
999
1000 dout("map_request osd %p is osd%d\n", req->r_osd, o);
1001 __insert_osd(osdc, req->r_osd);
1002
1003 ceph_con_open(&req->r_osd->o_con,
1004 CEPH_ENTITY_TYPE_OSD, o,
1005 &osdc->osdmap->osd_addr[o]);
1006 }
1007
1008 if (req->r_osd) {
1009 __remove_osd_from_lru(req->r_osd);
1010 list_add(&req->r_osd_item, &req->r_osd->o_requests);
1011 list_move(&req->r_req_lru_item, &osdc->req_unsent);
1012 } else {
1013 list_move(&req->r_req_lru_item, &osdc->req_notarget);
1014 }
1015 err = 1; /* osd or pg changed */
1016
1017 out:
1018 return err;
1019 }
1020
1021 /*
1022 * caller should hold map_sem (for read) and request_mutex
1023 */
1024 static void __send_request(struct ceph_osd_client *osdc,
1025 struct ceph_osd_request *req)
1026 {
1027 void *p;
1028
1029 dout("send_request %p tid %llu to osd%d flags %d pg %lld.%x\n",
1030 req, req->r_tid, req->r_osd->o_osd, req->r_flags,
1031 (unsigned long long)req->r_pgid.pool, req->r_pgid.seed);
1032
1033 /* fill in message content that changes each time we send it */
1034 put_unaligned_le32(osdc->osdmap->epoch, req->r_request_osdmap_epoch);
1035 put_unaligned_le32(req->r_flags, req->r_request_flags);
1036 put_unaligned_le64(req->r_pgid.pool, req->r_request_pool);
1037 p = req->r_request_pgid;
1038 ceph_encode_64(&p, req->r_pgid.pool);
1039 ceph_encode_32(&p, req->r_pgid.seed);
1040 put_unaligned_le64(1, req->r_request_attempts); /* FIXME */
1041 memcpy(req->r_request_reassert_version, &req->r_reassert_version,
1042 sizeof(req->r_reassert_version));
1043
1044 req->r_stamp = jiffies;
1045 list_move_tail(&req->r_req_lru_item, &osdc->req_lru);
1046
1047 ceph_msg_get(req->r_request); /* send consumes a ref */
1048 ceph_con_send(&req->r_osd->o_con, req->r_request);
1049 req->r_sent = req->r_osd->o_incarnation;
1050 }
1051
1052 /*
1053 * Send any requests in the queue (req_unsent).
1054 */
1055 static void __send_queued(struct ceph_osd_client *osdc)
1056 {
1057 struct ceph_osd_request *req, *tmp;
1058
1059 dout("__send_queued\n");
1060 list_for_each_entry_safe(req, tmp, &osdc->req_unsent, r_req_lru_item)
1061 __send_request(osdc, req);
1062 }
1063
1064 /*
1065 * Timeout callback, called every N seconds when 1 or more osd
1066 * requests has been active for more than N seconds. When this
1067 * happens, we ping all OSDs with requests who have timed out to
1068 * ensure any communications channel reset is detected. Reset the
1069 * request timeouts another N seconds in the future as we go.
1070 * Reschedule the timeout event another N seconds in future (unless
1071 * there are no open requests).
1072 */
1073 static void handle_timeout(struct work_struct *work)
1074 {
1075 struct ceph_osd_client *osdc =
1076 container_of(work, struct ceph_osd_client, timeout_work.work);
1077 struct ceph_osd_request *req;
1078 struct ceph_osd *osd;
1079 unsigned long keepalive =
1080 osdc->client->options->osd_keepalive_timeout * HZ;
1081 struct list_head slow_osds;
1082 dout("timeout\n");
1083 down_read(&osdc->map_sem);
1084
1085 ceph_monc_request_next_osdmap(&osdc->client->monc);
1086
1087 mutex_lock(&osdc->request_mutex);
1088
1089 /*
1090 * ping osds that are a bit slow. this ensures that if there
1091 * is a break in the TCP connection we will notice, and reopen
1092 * a connection with that osd (from the fault callback).
1093 */
1094 INIT_LIST_HEAD(&slow_osds);
1095 list_for_each_entry(req, &osdc->req_lru, r_req_lru_item) {
1096 if (time_before(jiffies, req->r_stamp + keepalive))
1097 break;
1098
1099 osd = req->r_osd;
1100 BUG_ON(!osd);
1101 dout(" tid %llu is slow, will send keepalive on osd%d\n",
1102 req->r_tid, osd->o_osd);
1103 list_move_tail(&osd->o_keepalive_item, &slow_osds);
1104 }
1105 while (!list_empty(&slow_osds)) {
1106 osd = list_entry(slow_osds.next, struct ceph_osd,
1107 o_keepalive_item);
1108 list_del_init(&osd->o_keepalive_item);
1109 ceph_con_keepalive(&osd->o_con);
1110 }
1111
1112 __schedule_osd_timeout(osdc);
1113 __send_queued(osdc);
1114 mutex_unlock(&osdc->request_mutex);
1115 up_read(&osdc->map_sem);
1116 }
1117
1118 static void handle_osds_timeout(struct work_struct *work)
1119 {
1120 struct ceph_osd_client *osdc =
1121 container_of(work, struct ceph_osd_client,
1122 osds_timeout_work.work);
1123 unsigned long delay =
1124 osdc->client->options->osd_idle_ttl * HZ >> 2;
1125
1126 dout("osds timeout\n");
1127 down_read(&osdc->map_sem);
1128 remove_old_osds(osdc);
1129 up_read(&osdc->map_sem);
1130
1131 schedule_delayed_work(&osdc->osds_timeout_work,
1132 round_jiffies_relative(delay));
1133 }
1134
1135 static void complete_request(struct ceph_osd_request *req)
1136 {
1137 if (req->r_safe_callback)
1138 req->r_safe_callback(req, NULL);
1139 complete_all(&req->r_safe_completion); /* fsync waiter */
1140 }
1141
1142 static int __decode_pgid(void **p, void *end, struct ceph_pg *pgid)
1143 {
1144 __u8 v;
1145
1146 ceph_decode_need(p, end, 1 + 8 + 4 + 4, bad);
1147 v = ceph_decode_8(p);
1148 if (v > 1) {
1149 pr_warning("do not understand pg encoding %d > 1", v);
1150 return -EINVAL;
1151 }
1152 pgid->pool = ceph_decode_64(p);
1153 pgid->seed = ceph_decode_32(p);
1154 *p += 4;
1155 return 0;
1156
1157 bad:
1158 pr_warning("incomplete pg encoding");
1159 return -EINVAL;
1160 }
1161
1162 /*
1163 * handle osd op reply. either call the callback if it is specified,
1164 * or do the completion to wake up the waiting thread.
1165 */
1166 static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
1167 struct ceph_connection *con)
1168 {
1169 void *p, *end;
1170 struct ceph_osd_request *req;
1171 u64 tid;
1172 int object_len;
1173 int numops, payload_len, flags;
1174 s32 result;
1175 s32 retry_attempt;
1176 struct ceph_pg pg;
1177 int err;
1178 u32 reassert_epoch;
1179 u64 reassert_version;
1180 u32 osdmap_epoch;
1181 int i;
1182
1183 tid = le64_to_cpu(msg->hdr.tid);
1184 dout("handle_reply %p tid %llu\n", msg, tid);
1185
1186 p = msg->front.iov_base;
1187 end = p + msg->front.iov_len;
1188
1189 ceph_decode_need(&p, end, 4, bad);
1190 object_len = ceph_decode_32(&p);
1191 ceph_decode_need(&p, end, object_len, bad);
1192 p += object_len;
1193
1194 err = __decode_pgid(&p, end, &pg);
1195 if (err)
1196 goto bad;
1197
1198 ceph_decode_need(&p, end, 8 + 4 + 4 + 8 + 4, bad);
1199 flags = ceph_decode_64(&p);
1200 result = ceph_decode_32(&p);
1201 reassert_epoch = ceph_decode_32(&p);
1202 reassert_version = ceph_decode_64(&p);
1203 osdmap_epoch = ceph_decode_32(&p);
1204
1205 /* lookup */
1206 mutex_lock(&osdc->request_mutex);
1207 req = __lookup_request(osdc, tid);
1208 if (req == NULL) {
1209 dout("handle_reply tid %llu dne\n", tid);
1210 mutex_unlock(&osdc->request_mutex);
1211 return;
1212 }
1213 ceph_osdc_get_request(req);
1214
1215 dout("handle_reply %p tid %llu req %p result %d\n", msg, tid,
1216 req, result);
1217
1218 ceph_decode_need(&p, end, 4, bad);
1219 numops = ceph_decode_32(&p);
1220 if (numops > CEPH_OSD_MAX_OP)
1221 goto bad_put;
1222 if (numops != req->r_num_ops)
1223 goto bad_put;
1224 payload_len = 0;
1225 ceph_decode_need(&p, end, numops * sizeof(struct ceph_osd_op), bad);
1226 for (i = 0; i < numops; i++) {
1227 struct ceph_osd_op *op = p;
1228 int len;
1229
1230 len = le32_to_cpu(op->payload_len);
1231 req->r_reply_op_len[i] = len;
1232 dout(" op %d has %d bytes\n", i, len);
1233 payload_len += len;
1234 p += sizeof(*op);
1235 }
1236 if (payload_len != le32_to_cpu(msg->hdr.data_len)) {
1237 pr_warning("sum of op payload lens %d != data_len %d",
1238 payload_len, le32_to_cpu(msg->hdr.data_len));
1239 goto bad_put;
1240 }
1241
1242 ceph_decode_need(&p, end, 4 + numops * 4, bad);
1243 retry_attempt = ceph_decode_32(&p);
1244 for (i = 0; i < numops; i++)
1245 req->r_reply_op_result[i] = ceph_decode_32(&p);
1246
1247 /*
1248 * if this connection filled our message, drop our reference now, to
1249 * avoid a (safe but slower) revoke later.
1250 */
1251 if (req->r_con_filling_msg == con && req->r_reply == msg) {
1252 dout(" dropping con_filling_msg ref %p\n", con);
1253 req->r_con_filling_msg = NULL;
1254 con->ops->put(con);
1255 }
1256
1257 if (!req->r_got_reply) {
1258 unsigned int bytes;
1259
1260 req->r_result = result;
1261 bytes = le32_to_cpu(msg->hdr.data_len);
1262 dout("handle_reply result %d bytes %d\n", req->r_result,
1263 bytes);
1264 if (req->r_result == 0)
1265 req->r_result = bytes;
1266
1267 /* in case this is a write and we need to replay, */
1268 req->r_reassert_version.epoch = cpu_to_le32(reassert_epoch);
1269 req->r_reassert_version.version = cpu_to_le64(reassert_version);
1270
1271 req->r_got_reply = 1;
1272 } else if ((flags & CEPH_OSD_FLAG_ONDISK) == 0) {
1273 dout("handle_reply tid %llu dup ack\n", tid);
1274 mutex_unlock(&osdc->request_mutex);
1275 goto done;
1276 }
1277
1278 dout("handle_reply tid %llu flags %d\n", tid, flags);
1279
1280 if (req->r_linger && (flags & CEPH_OSD_FLAG_ONDISK))
1281 __register_linger_request(osdc, req);
1282
1283 /* either this is a read, or we got the safe response */
1284 if (result < 0 ||
1285 (flags & CEPH_OSD_FLAG_ONDISK) ||
1286 ((flags & CEPH_OSD_FLAG_WRITE) == 0))
1287 __unregister_request(osdc, req);
1288
1289 mutex_unlock(&osdc->request_mutex);
1290
1291 if (req->r_callback)
1292 req->r_callback(req, msg);
1293 else
1294 complete_all(&req->r_completion);
1295
1296 if (flags & CEPH_OSD_FLAG_ONDISK)
1297 complete_request(req);
1298
1299 done:
1300 dout("req=%p req->r_linger=%d\n", req, req->r_linger);
1301 ceph_osdc_put_request(req);
1302 return;
1303
1304 bad_put:
1305 ceph_osdc_put_request(req);
1306 bad:
1307 pr_err("corrupt osd_op_reply got %d %d\n",
1308 (int)msg->front.iov_len, le32_to_cpu(msg->hdr.front_len));
1309 ceph_msg_dump(msg);
1310 }
1311
1312 static void reset_changed_osds(struct ceph_osd_client *osdc)
1313 {
1314 struct rb_node *p, *n;
1315
1316 for (p = rb_first(&osdc->osds); p; p = n) {
1317 struct ceph_osd *osd = rb_entry(p, struct ceph_osd, o_node);
1318
1319 n = rb_next(p);
1320 if (!ceph_osd_is_up(osdc->osdmap, osd->o_osd) ||
1321 memcmp(&osd->o_con.peer_addr,
1322 ceph_osd_addr(osdc->osdmap,
1323 osd->o_osd),
1324 sizeof(struct ceph_entity_addr)) != 0)
1325 __reset_osd(osdc, osd);
1326 }
1327 }
1328
1329 /*
1330 * Requeue requests whose mapping to an OSD has changed. If requests map to
1331 * no osd, request a new map.
1332 *
1333 * Caller should hold map_sem for read.
1334 */
1335 static void kick_requests(struct ceph_osd_client *osdc, int force_resend)
1336 {
1337 struct ceph_osd_request *req, *nreq;
1338 struct rb_node *p;
1339 int needmap = 0;
1340 int err;
1341
1342 dout("kick_requests %s\n", force_resend ? " (force resend)" : "");
1343 mutex_lock(&osdc->request_mutex);
1344 for (p = rb_first(&osdc->requests); p; ) {
1345 req = rb_entry(p, struct ceph_osd_request, r_node);
1346 p = rb_next(p);
1347
1348 /*
1349 * For linger requests that have not yet been
1350 * registered, move them to the linger list; they'll
1351 * be sent to the osd in the loop below. Unregister
1352 * the request before re-registering it as a linger
1353 * request to ensure the __map_request() below
1354 * will decide it needs to be sent.
1355 */
1356 if (req->r_linger && list_empty(&req->r_linger_item)) {
1357 dout("%p tid %llu restart on osd%d\n",
1358 req, req->r_tid,
1359 req->r_osd ? req->r_osd->o_osd : -1);
1360 __unregister_request(osdc, req);
1361 __register_linger_request(osdc, req);
1362 continue;
1363 }
1364
1365 err = __map_request(osdc, req, force_resend);
1366 if (err < 0)
1367 continue; /* error */
1368 if (req->r_osd == NULL) {
1369 dout("%p tid %llu maps to no osd\n", req, req->r_tid);
1370 needmap++; /* request a newer map */
1371 } else if (err > 0) {
1372 if (!req->r_linger) {
1373 dout("%p tid %llu requeued on osd%d\n", req,
1374 req->r_tid,
1375 req->r_osd ? req->r_osd->o_osd : -1);
1376 req->r_flags |= CEPH_OSD_FLAG_RETRY;
1377 }
1378 }
1379 }
1380
1381 list_for_each_entry_safe(req, nreq, &osdc->req_linger,
1382 r_linger_item) {
1383 dout("linger req=%p req->r_osd=%p\n", req, req->r_osd);
1384
1385 err = __map_request(osdc, req, force_resend);
1386 dout("__map_request returned %d\n", err);
1387 if (err == 0)
1388 continue; /* no change and no osd was specified */
1389 if (err < 0)
1390 continue; /* hrm! */
1391 if (req->r_osd == NULL) {
1392 dout("tid %llu maps to no valid osd\n", req->r_tid);
1393 needmap++; /* request a newer map */
1394 continue;
1395 }
1396
1397 dout("kicking lingering %p tid %llu osd%d\n", req, req->r_tid,
1398 req->r_osd ? req->r_osd->o_osd : -1);
1399 __register_request(osdc, req);
1400 __unregister_linger_request(osdc, req);
1401 }
1402 mutex_unlock(&osdc->request_mutex);
1403
1404 if (needmap) {
1405 dout("%d requests for down osds, need new map\n", needmap);
1406 ceph_monc_request_next_osdmap(&osdc->client->monc);
1407 }
1408 reset_changed_osds(osdc);
1409 }
1410
1411
1412 /*
1413 * Process updated osd map.
1414 *
1415 * The message contains any number of incremental and full maps, normally
1416 * indicating some sort of topology change in the cluster. Kick requests
1417 * off to different OSDs as needed.
1418 */
1419 void ceph_osdc_handle_map(struct ceph_osd_client *osdc, struct ceph_msg *msg)
1420 {
1421 void *p, *end, *next;
1422 u32 nr_maps, maplen;
1423 u32 epoch;
1424 struct ceph_osdmap *newmap = NULL, *oldmap;
1425 int err;
1426 struct ceph_fsid fsid;
1427
1428 dout("handle_map have %u\n", osdc->osdmap ? osdc->osdmap->epoch : 0);
1429 p = msg->front.iov_base;
1430 end = p + msg->front.iov_len;
1431
1432 /* verify fsid */
1433 ceph_decode_need(&p, end, sizeof(fsid), bad);
1434 ceph_decode_copy(&p, &fsid, sizeof(fsid));
1435 if (ceph_check_fsid(osdc->client, &fsid) < 0)
1436 return;
1437
1438 down_write(&osdc->map_sem);
1439
1440 /* incremental maps */
1441 ceph_decode_32_safe(&p, end, nr_maps, bad);
1442 dout(" %d inc maps\n", nr_maps);
1443 while (nr_maps > 0) {
1444 ceph_decode_need(&p, end, 2*sizeof(u32), bad);
1445 epoch = ceph_decode_32(&p);
1446 maplen = ceph_decode_32(&p);
1447 ceph_decode_need(&p, end, maplen, bad);
1448 next = p + maplen;
1449 if (osdc->osdmap && osdc->osdmap->epoch+1 == epoch) {
1450 dout("applying incremental map %u len %d\n",
1451 epoch, maplen);
1452 newmap = osdmap_apply_incremental(&p, next,
1453 osdc->osdmap,
1454 &osdc->client->msgr);
1455 if (IS_ERR(newmap)) {
1456 err = PTR_ERR(newmap);
1457 goto bad;
1458 }
1459 BUG_ON(!newmap);
1460 if (newmap != osdc->osdmap) {
1461 ceph_osdmap_destroy(osdc->osdmap);
1462 osdc->osdmap = newmap;
1463 }
1464 kick_requests(osdc, 0);
1465 } else {
1466 dout("ignoring incremental map %u len %d\n",
1467 epoch, maplen);
1468 }
1469 p = next;
1470 nr_maps--;
1471 }
1472 if (newmap)
1473 goto done;
1474
1475 /* full maps */
1476 ceph_decode_32_safe(&p, end, nr_maps, bad);
1477 dout(" %d full maps\n", nr_maps);
1478 while (nr_maps) {
1479 ceph_decode_need(&p, end, 2*sizeof(u32), bad);
1480 epoch = ceph_decode_32(&p);
1481 maplen = ceph_decode_32(&p);
1482 ceph_decode_need(&p, end, maplen, bad);
1483 if (nr_maps > 1) {
1484 dout("skipping non-latest full map %u len %d\n",
1485 epoch, maplen);
1486 } else if (osdc->osdmap && osdc->osdmap->epoch >= epoch) {
1487 dout("skipping full map %u len %d, "
1488 "older than our %u\n", epoch, maplen,
1489 osdc->osdmap->epoch);
1490 } else {
1491 int skipped_map = 0;
1492
1493 dout("taking full map %u len %d\n", epoch, maplen);
1494 newmap = osdmap_decode(&p, p+maplen);
1495 if (IS_ERR(newmap)) {
1496 err = PTR_ERR(newmap);
1497 goto bad;
1498 }
1499 BUG_ON(!newmap);
1500 oldmap = osdc->osdmap;
1501 osdc->osdmap = newmap;
1502 if (oldmap) {
1503 if (oldmap->epoch + 1 < newmap->epoch)
1504 skipped_map = 1;
1505 ceph_osdmap_destroy(oldmap);
1506 }
1507 kick_requests(osdc, skipped_map);
1508 }
1509 p += maplen;
1510 nr_maps--;
1511 }
1512
1513 done:
1514 downgrade_write(&osdc->map_sem);
1515 ceph_monc_got_osdmap(&osdc->client->monc, osdc->osdmap->epoch);
1516
1517 /*
1518 * subscribe to subsequent osdmap updates if full to ensure
1519 * we find out when we are no longer full and stop returning
1520 * ENOSPC.
1521 */
1522 if (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL))
1523 ceph_monc_request_next_osdmap(&osdc->client->monc);
1524
1525 mutex_lock(&osdc->request_mutex);
1526 __send_queued(osdc);
1527 mutex_unlock(&osdc->request_mutex);
1528 up_read(&osdc->map_sem);
1529 wake_up_all(&osdc->client->auth_wq);
1530 return;
1531
1532 bad:
1533 pr_err("osdc handle_map corrupt msg\n");
1534 ceph_msg_dump(msg);
1535 up_write(&osdc->map_sem);
1536 return;
1537 }
1538
1539 /*
1540 * watch/notify callback event infrastructure
1541 *
1542 * These callbacks are used both for watch and notify operations.
1543 */
1544 static void __release_event(struct kref *kref)
1545 {
1546 struct ceph_osd_event *event =
1547 container_of(kref, struct ceph_osd_event, kref);
1548
1549 dout("__release_event %p\n", event);
1550 kfree(event);
1551 }
1552
1553 static void get_event(struct ceph_osd_event *event)
1554 {
1555 kref_get(&event->kref);
1556 }
1557
1558 void ceph_osdc_put_event(struct ceph_osd_event *event)
1559 {
1560 kref_put(&event->kref, __release_event);
1561 }
1562 EXPORT_SYMBOL(ceph_osdc_put_event);
1563
1564 static void __insert_event(struct ceph_osd_client *osdc,
1565 struct ceph_osd_event *new)
1566 {
1567 struct rb_node **p = &osdc->event_tree.rb_node;
1568 struct rb_node *parent = NULL;
1569 struct ceph_osd_event *event = NULL;
1570
1571 while (*p) {
1572 parent = *p;
1573 event = rb_entry(parent, struct ceph_osd_event, node);
1574 if (new->cookie < event->cookie)
1575 p = &(*p)->rb_left;
1576 else if (new->cookie > event->cookie)
1577 p = &(*p)->rb_right;
1578 else
1579 BUG();
1580 }
1581
1582 rb_link_node(&new->node, parent, p);
1583 rb_insert_color(&new->node, &osdc->event_tree);
1584 }
1585
1586 static struct ceph_osd_event *__find_event(struct ceph_osd_client *osdc,
1587 u64 cookie)
1588 {
1589 struct rb_node **p = &osdc->event_tree.rb_node;
1590 struct rb_node *parent = NULL;
1591 struct ceph_osd_event *event = NULL;
1592
1593 while (*p) {
1594 parent = *p;
1595 event = rb_entry(parent, struct ceph_osd_event, node);
1596 if (cookie < event->cookie)
1597 p = &(*p)->rb_left;
1598 else if (cookie > event->cookie)
1599 p = &(*p)->rb_right;
1600 else
1601 return event;
1602 }
1603 return NULL;
1604 }
1605
1606 static void __remove_event(struct ceph_osd_event *event)
1607 {
1608 struct ceph_osd_client *osdc = event->osdc;
1609
1610 if (!RB_EMPTY_NODE(&event->node)) {
1611 dout("__remove_event removed %p\n", event);
1612 rb_erase(&event->node, &osdc->event_tree);
1613 ceph_osdc_put_event(event);
1614 } else {
1615 dout("__remove_event didn't remove %p\n", event);
1616 }
1617 }
1618
1619 int ceph_osdc_create_event(struct ceph_osd_client *osdc,
1620 void (*event_cb)(u64, u64, u8, void *),
1621 void *data, struct ceph_osd_event **pevent)
1622 {
1623 struct ceph_osd_event *event;
1624
1625 event = kmalloc(sizeof(*event), GFP_NOIO);
1626 if (!event)
1627 return -ENOMEM;
1628
1629 dout("create_event %p\n", event);
1630 event->cb = event_cb;
1631 event->one_shot = 0;
1632 event->data = data;
1633 event->osdc = osdc;
1634 INIT_LIST_HEAD(&event->osd_node);
1635 RB_CLEAR_NODE(&event->node);
1636 kref_init(&event->kref); /* one ref for us */
1637 kref_get(&event->kref); /* one ref for the caller */
1638
1639 spin_lock(&osdc->event_lock);
1640 event->cookie = ++osdc->event_count;
1641 __insert_event(osdc, event);
1642 spin_unlock(&osdc->event_lock);
1643
1644 *pevent = event;
1645 return 0;
1646 }
1647 EXPORT_SYMBOL(ceph_osdc_create_event);
1648
1649 void ceph_osdc_cancel_event(struct ceph_osd_event *event)
1650 {
1651 struct ceph_osd_client *osdc = event->osdc;
1652
1653 dout("cancel_event %p\n", event);
1654 spin_lock(&osdc->event_lock);
1655 __remove_event(event);
1656 spin_unlock(&osdc->event_lock);
1657 ceph_osdc_put_event(event); /* caller's */
1658 }
1659 EXPORT_SYMBOL(ceph_osdc_cancel_event);
1660
1661
1662 static void do_event_work(struct work_struct *work)
1663 {
1664 struct ceph_osd_event_work *event_work =
1665 container_of(work, struct ceph_osd_event_work, work);
1666 struct ceph_osd_event *event = event_work->event;
1667 u64 ver = event_work->ver;
1668 u64 notify_id = event_work->notify_id;
1669 u8 opcode = event_work->opcode;
1670
1671 dout("do_event_work completing %p\n", event);
1672 event->cb(ver, notify_id, opcode, event->data);
1673 dout("do_event_work completed %p\n", event);
1674 ceph_osdc_put_event(event);
1675 kfree(event_work);
1676 }
1677
1678
1679 /*
1680 * Process osd watch notifications
1681 */
1682 static void handle_watch_notify(struct ceph_osd_client *osdc,
1683 struct ceph_msg *msg)
1684 {
1685 void *p, *end;
1686 u8 proto_ver;
1687 u64 cookie, ver, notify_id;
1688 u8 opcode;
1689 struct ceph_osd_event *event;
1690 struct ceph_osd_event_work *event_work;
1691
1692 p = msg->front.iov_base;
1693 end = p + msg->front.iov_len;
1694
1695 ceph_decode_8_safe(&p, end, proto_ver, bad);
1696 ceph_decode_8_safe(&p, end, opcode, bad);
1697 ceph_decode_64_safe(&p, end, cookie, bad);
1698 ceph_decode_64_safe(&p, end, ver, bad);
1699 ceph_decode_64_safe(&p, end, notify_id, bad);
1700
1701 spin_lock(&osdc->event_lock);
1702 event = __find_event(osdc, cookie);
1703 if (event) {
1704 BUG_ON(event->one_shot);
1705 get_event(event);
1706 }
1707 spin_unlock(&osdc->event_lock);
1708 dout("handle_watch_notify cookie %lld ver %lld event %p\n",
1709 cookie, ver, event);
1710 if (event) {
1711 event_work = kmalloc(sizeof(*event_work), GFP_NOIO);
1712 if (!event_work) {
1713 dout("ERROR: could not allocate event_work\n");
1714 goto done_err;
1715 }
1716 INIT_WORK(&event_work->work, do_event_work);
1717 event_work->event = event;
1718 event_work->ver = ver;
1719 event_work->notify_id = notify_id;
1720 event_work->opcode = opcode;
1721 if (!queue_work(osdc->notify_wq, &event_work->work)) {
1722 dout("WARNING: failed to queue notify event work\n");
1723 goto done_err;
1724 }
1725 }
1726
1727 return;
1728
1729 done_err:
1730 ceph_osdc_put_event(event);
1731 return;
1732
1733 bad:
1734 pr_err("osdc handle_watch_notify corrupt msg\n");
1735 return;
1736 }
1737
1738 /*
1739 * Register request, send initial attempt.
1740 */
1741 int ceph_osdc_start_request(struct ceph_osd_client *osdc,
1742 struct ceph_osd_request *req,
1743 bool nofail)
1744 {
1745 int rc = 0;
1746
1747 req->r_request->pages = req->r_pages;
1748 req->r_request->nr_pages = req->r_num_pages;
1749 #ifdef CONFIG_BLOCK
1750 req->r_request->bio = req->r_bio;
1751 #endif
1752 req->r_request->trail = &req->r_trail;
1753
1754 register_request(osdc, req);
1755
1756 down_read(&osdc->map_sem);
1757 mutex_lock(&osdc->request_mutex);
1758 /*
1759 * a racing kick_requests() may have sent the message for us
1760 * while we dropped request_mutex above, so only send now if
1761 * the request still han't been touched yet.
1762 */
1763 if (req->r_sent == 0) {
1764 rc = __map_request(osdc, req, 0);
1765 if (rc < 0) {
1766 if (nofail) {
1767 dout("osdc_start_request failed map, "
1768 " will retry %lld\n", req->r_tid);
1769 rc = 0;
1770 }
1771 goto out_unlock;
1772 }
1773 if (req->r_osd == NULL) {
1774 dout("send_request %p no up osds in pg\n", req);
1775 ceph_monc_request_next_osdmap(&osdc->client->monc);
1776 } else {
1777 __send_request(osdc, req);
1778 }
1779 rc = 0;
1780 }
1781
1782 out_unlock:
1783 mutex_unlock(&osdc->request_mutex);
1784 up_read(&osdc->map_sem);
1785 return rc;
1786 }
1787 EXPORT_SYMBOL(ceph_osdc_start_request);
1788
1789 /*
1790 * wait for a request to complete
1791 */
1792 int ceph_osdc_wait_request(struct ceph_osd_client *osdc,
1793 struct ceph_osd_request *req)
1794 {
1795 int rc;
1796
1797 rc = wait_for_completion_interruptible(&req->r_completion);
1798 if (rc < 0) {
1799 mutex_lock(&osdc->request_mutex);
1800 __cancel_request(req);
1801 __unregister_request(osdc, req);
1802 mutex_unlock(&osdc->request_mutex);
1803 complete_request(req);
1804 dout("wait_request tid %llu canceled/timed out\n", req->r_tid);
1805 return rc;
1806 }
1807
1808 dout("wait_request tid %llu result %d\n", req->r_tid, req->r_result);
1809 return req->r_result;
1810 }
1811 EXPORT_SYMBOL(ceph_osdc_wait_request);
1812
1813 /*
1814 * sync - wait for all in-flight requests to flush. avoid starvation.
1815 */
1816 void ceph_osdc_sync(struct ceph_osd_client *osdc)
1817 {
1818 struct ceph_osd_request *req;
1819 u64 last_tid, next_tid = 0;
1820
1821 mutex_lock(&osdc->request_mutex);
1822 last_tid = osdc->last_tid;
1823 while (1) {
1824 req = __lookup_request_ge(osdc, next_tid);
1825 if (!req)
1826 break;
1827 if (req->r_tid > last_tid)
1828 break;
1829
1830 next_tid = req->r_tid + 1;
1831 if ((req->r_flags & CEPH_OSD_FLAG_WRITE) == 0)
1832 continue;
1833
1834 ceph_osdc_get_request(req);
1835 mutex_unlock(&osdc->request_mutex);
1836 dout("sync waiting on tid %llu (last is %llu)\n",
1837 req->r_tid, last_tid);
1838 wait_for_completion(&req->r_safe_completion);
1839 mutex_lock(&osdc->request_mutex);
1840 ceph_osdc_put_request(req);
1841 }
1842 mutex_unlock(&osdc->request_mutex);
1843 dout("sync done (thru tid %llu)\n", last_tid);
1844 }
1845 EXPORT_SYMBOL(ceph_osdc_sync);
1846
1847 /*
1848 * init, shutdown
1849 */
1850 int ceph_osdc_init(struct ceph_osd_client *osdc, struct ceph_client *client)
1851 {
1852 int err;
1853
1854 dout("init\n");
1855 osdc->client = client;
1856 osdc->osdmap = NULL;
1857 init_rwsem(&osdc->map_sem);
1858 init_completion(&osdc->map_waiters);
1859 osdc->last_requested_map = 0;
1860 mutex_init(&osdc->request_mutex);
1861 osdc->last_tid = 0;
1862 osdc->osds = RB_ROOT;
1863 INIT_LIST_HEAD(&osdc->osd_lru);
1864 osdc->requests = RB_ROOT;
1865 INIT_LIST_HEAD(&osdc->req_lru);
1866 INIT_LIST_HEAD(&osdc->req_unsent);
1867 INIT_LIST_HEAD(&osdc->req_notarget);
1868 INIT_LIST_HEAD(&osdc->req_linger);
1869 osdc->num_requests = 0;
1870 INIT_DELAYED_WORK(&osdc->timeout_work, handle_timeout);
1871 INIT_DELAYED_WORK(&osdc->osds_timeout_work, handle_osds_timeout);
1872 spin_lock_init(&osdc->event_lock);
1873 osdc->event_tree = RB_ROOT;
1874 osdc->event_count = 0;
1875
1876 schedule_delayed_work(&osdc->osds_timeout_work,
1877 round_jiffies_relative(osdc->client->options->osd_idle_ttl * HZ));
1878
1879 err = -ENOMEM;
1880 osdc->req_mempool = mempool_create_kmalloc_pool(10,
1881 sizeof(struct ceph_osd_request));
1882 if (!osdc->req_mempool)
1883 goto out;
1884
1885 err = ceph_msgpool_init(&osdc->msgpool_op, CEPH_MSG_OSD_OP,
1886 OSD_OP_FRONT_LEN, 10, true,
1887 "osd_op");
1888 if (err < 0)
1889 goto out_mempool;
1890 err = ceph_msgpool_init(&osdc->msgpool_op_reply, CEPH_MSG_OSD_OPREPLY,
1891 OSD_OPREPLY_FRONT_LEN, 10, true,
1892 "osd_op_reply");
1893 if (err < 0)
1894 goto out_msgpool;
1895
1896 osdc->notify_wq = create_singlethread_workqueue("ceph-watch-notify");
1897 if (IS_ERR(osdc->notify_wq)) {
1898 err = PTR_ERR(osdc->notify_wq);
1899 osdc->notify_wq = NULL;
1900 goto out_msgpool;
1901 }
1902 return 0;
1903
1904 out_msgpool:
1905 ceph_msgpool_destroy(&osdc->msgpool_op);
1906 out_mempool:
1907 mempool_destroy(osdc->req_mempool);
1908 out:
1909 return err;
1910 }
1911
1912 void ceph_osdc_stop(struct ceph_osd_client *osdc)
1913 {
1914 flush_workqueue(osdc->notify_wq);
1915 destroy_workqueue(osdc->notify_wq);
1916 cancel_delayed_work_sync(&osdc->timeout_work);
1917 cancel_delayed_work_sync(&osdc->osds_timeout_work);
1918 if (osdc->osdmap) {
1919 ceph_osdmap_destroy(osdc->osdmap);
1920 osdc->osdmap = NULL;
1921 }
1922 remove_all_osds(osdc);
1923 mempool_destroy(osdc->req_mempool);
1924 ceph_msgpool_destroy(&osdc->msgpool_op);
1925 ceph_msgpool_destroy(&osdc->msgpool_op_reply);
1926 }
1927
1928 /*
1929 * Read some contiguous pages. If we cross a stripe boundary, shorten
1930 * *plen. Return number of bytes read, or error.
1931 */
1932 int ceph_osdc_readpages(struct ceph_osd_client *osdc,
1933 struct ceph_vino vino, struct ceph_file_layout *layout,
1934 u64 off, u64 *plen,
1935 u32 truncate_seq, u64 truncate_size,
1936 struct page **pages, int num_pages, int page_align)
1937 {
1938 struct ceph_osd_request *req;
1939 int rc = 0;
1940
1941 dout("readpages on ino %llx.%llx on %llu~%llu\n", vino.ino,
1942 vino.snap, off, *plen);
1943 req = ceph_osdc_new_request(osdc, layout, vino, off, plen,
1944 CEPH_OSD_OP_READ, CEPH_OSD_FLAG_READ,
1945 NULL, 0, truncate_seq, truncate_size, NULL,
1946 false, page_align);
1947 if (IS_ERR(req))
1948 return PTR_ERR(req);
1949
1950 /* it may be a short read due to an object boundary */
1951 req->r_pages = pages;
1952
1953 dout("readpages final extent is %llu~%llu (%d pages align %d)\n",
1954 off, *plen, req->r_num_pages, page_align);
1955
1956 rc = ceph_osdc_start_request(osdc, req, false);
1957 if (!rc)
1958 rc = ceph_osdc_wait_request(osdc, req);
1959
1960 ceph_osdc_put_request(req);
1961 dout("readpages result %d\n", rc);
1962 return rc;
1963 }
1964 EXPORT_SYMBOL(ceph_osdc_readpages);
1965
1966 /*
1967 * do a synchronous write on N pages
1968 */
1969 int ceph_osdc_writepages(struct ceph_osd_client *osdc, struct ceph_vino vino,
1970 struct ceph_file_layout *layout,
1971 struct ceph_snap_context *snapc,
1972 u64 off, u64 len,
1973 u32 truncate_seq, u64 truncate_size,
1974 struct timespec *mtime,
1975 struct page **pages, int num_pages)
1976 {
1977 struct ceph_osd_request *req;
1978 int rc = 0;
1979 int page_align = off & ~PAGE_MASK;
1980
1981 BUG_ON(vino.snap != CEPH_NOSNAP);
1982 req = ceph_osdc_new_request(osdc, layout, vino, off, &len,
1983 CEPH_OSD_OP_WRITE,
1984 CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE,
1985 snapc, 0,
1986 truncate_seq, truncate_size, mtime,
1987 true, page_align);
1988 if (IS_ERR(req))
1989 return PTR_ERR(req);
1990
1991 /* it may be a short write due to an object boundary */
1992 req->r_pages = pages;
1993 dout("writepages %llu~%llu (%d pages)\n", off, len,
1994 req->r_num_pages);
1995
1996 rc = ceph_osdc_start_request(osdc, req, true);
1997 if (!rc)
1998 rc = ceph_osdc_wait_request(osdc, req);
1999
2000 ceph_osdc_put_request(req);
2001 if (rc == 0)
2002 rc = len;
2003 dout("writepages result %d\n", rc);
2004 return rc;
2005 }
2006 EXPORT_SYMBOL(ceph_osdc_writepages);
2007
2008 /*
2009 * handle incoming message
2010 */
2011 static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
2012 {
2013 struct ceph_osd *osd = con->private;
2014 struct ceph_osd_client *osdc;
2015 int type = le16_to_cpu(msg->hdr.type);
2016
2017 if (!osd)
2018 goto out;
2019 osdc = osd->o_osdc;
2020
2021 switch (type) {
2022 case CEPH_MSG_OSD_MAP:
2023 ceph_osdc_handle_map(osdc, msg);
2024 break;
2025 case CEPH_MSG_OSD_OPREPLY:
2026 handle_reply(osdc, msg, con);
2027 break;
2028 case CEPH_MSG_WATCH_NOTIFY:
2029 handle_watch_notify(osdc, msg);
2030 break;
2031
2032 default:
2033 pr_err("received unknown message type %d %s\n", type,
2034 ceph_msg_type_name(type));
2035 }
2036 out:
2037 ceph_msg_put(msg);
2038 }
2039
2040 /*
2041 * lookup and return message for incoming reply. set up reply message
2042 * pages.
2043 */
2044 static struct ceph_msg *get_reply(struct ceph_connection *con,
2045 struct ceph_msg_header *hdr,
2046 int *skip)
2047 {
2048 struct ceph_osd *osd = con->private;
2049 struct ceph_osd_client *osdc = osd->o_osdc;
2050 struct ceph_msg *m;
2051 struct ceph_osd_request *req;
2052 int front = le32_to_cpu(hdr->front_len);
2053 int data_len = le32_to_cpu(hdr->data_len);
2054 u64 tid;
2055
2056 tid = le64_to_cpu(hdr->tid);
2057 mutex_lock(&osdc->request_mutex);
2058 req = __lookup_request(osdc, tid);
2059 if (!req) {
2060 *skip = 1;
2061 m = NULL;
2062 dout("get_reply unknown tid %llu from osd%d\n", tid,
2063 osd->o_osd);
2064 goto out;
2065 }
2066
2067 if (req->r_con_filling_msg) {
2068 dout("%s revoking msg %p from old con %p\n", __func__,
2069 req->r_reply, req->r_con_filling_msg);
2070 ceph_msg_revoke_incoming(req->r_reply);
2071 req->r_con_filling_msg->ops->put(req->r_con_filling_msg);
2072 req->r_con_filling_msg = NULL;
2073 }
2074
2075 if (front > req->r_reply->front.iov_len) {
2076 pr_warning("get_reply front %d > preallocated %d\n",
2077 front, (int)req->r_reply->front.iov_len);
2078 m = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, front, GFP_NOFS, false);
2079 if (!m)
2080 goto out;
2081 ceph_msg_put(req->r_reply);
2082 req->r_reply = m;
2083 }
2084 m = ceph_msg_get(req->r_reply);
2085
2086 if (data_len > 0) {
2087 int want = calc_pages_for(req->r_page_alignment, data_len);
2088
2089 if (req->r_pages && unlikely(req->r_num_pages < want)) {
2090 pr_warning("tid %lld reply has %d bytes %d pages, we"
2091 " had only %d pages ready\n", tid, data_len,
2092 want, req->r_num_pages);
2093 *skip = 1;
2094 ceph_msg_put(m);
2095 m = NULL;
2096 goto out;
2097 }
2098 m->pages = req->r_pages;
2099 m->nr_pages = req->r_num_pages;
2100 m->page_alignment = req->r_page_alignment;
2101 #ifdef CONFIG_BLOCK
2102 m->bio = req->r_bio;
2103 #endif
2104 }
2105 *skip = 0;
2106 req->r_con_filling_msg = con->ops->get(con);
2107 dout("get_reply tid %lld %p\n", tid, m);
2108
2109 out:
2110 mutex_unlock(&osdc->request_mutex);
2111 return m;
2112
2113 }
2114
2115 static struct ceph_msg *alloc_msg(struct ceph_connection *con,
2116 struct ceph_msg_header *hdr,
2117 int *skip)
2118 {
2119 struct ceph_osd *osd = con->private;
2120 int type = le16_to_cpu(hdr->type);
2121 int front = le32_to_cpu(hdr->front_len);
2122
2123 *skip = 0;
2124 switch (type) {
2125 case CEPH_MSG_OSD_MAP:
2126 case CEPH_MSG_WATCH_NOTIFY:
2127 return ceph_msg_new(type, front, GFP_NOFS, false);
2128 case CEPH_MSG_OSD_OPREPLY:
2129 return get_reply(con, hdr, skip);
2130 default:
2131 pr_info("alloc_msg unexpected msg type %d from osd%d\n", type,
2132 osd->o_osd);
2133 *skip = 1;
2134 return NULL;
2135 }
2136 }
2137
2138 /*
2139 * Wrappers to refcount containing ceph_osd struct
2140 */
2141 static struct ceph_connection *get_osd_con(struct ceph_connection *con)
2142 {
2143 struct ceph_osd *osd = con->private;
2144 if (get_osd(osd))
2145 return con;
2146 return NULL;
2147 }
2148
2149 static void put_osd_con(struct ceph_connection *con)
2150 {
2151 struct ceph_osd *osd = con->private;
2152 put_osd(osd);
2153 }
2154
2155 /*
2156 * authentication
2157 */
2158 /*
2159 * Note: returned pointer is the address of a structure that's
2160 * managed separately. Caller must *not* attempt to free it.
2161 */
2162 static struct ceph_auth_handshake *get_authorizer(struct ceph_connection *con,
2163 int *proto, int force_new)
2164 {
2165 struct ceph_osd *o = con->private;
2166 struct ceph_osd_client *osdc = o->o_osdc;
2167 struct ceph_auth_client *ac = osdc->client->monc.auth;
2168 struct ceph_auth_handshake *auth = &o->o_auth;
2169
2170 if (force_new && auth->authorizer) {
2171 if (ac->ops && ac->ops->destroy_authorizer)
2172 ac->ops->destroy_authorizer(ac, auth->authorizer);
2173 auth->authorizer = NULL;
2174 }
2175 if (!auth->authorizer && ac->ops && ac->ops->create_authorizer) {
2176 int ret = ac->ops->create_authorizer(ac, CEPH_ENTITY_TYPE_OSD,
2177 auth);
2178 if (ret)
2179 return ERR_PTR(ret);
2180 }
2181 *proto = ac->protocol;
2182
2183 return auth;
2184 }
2185
2186
2187 static int verify_authorizer_reply(struct ceph_connection *con, int len)
2188 {
2189 struct ceph_osd *o = con->private;
2190 struct ceph_osd_client *osdc = o->o_osdc;
2191 struct ceph_auth_client *ac = osdc->client->monc.auth;
2192
2193 /*
2194 * XXX If ac->ops or ac->ops->verify_authorizer_reply is null,
2195 * XXX which do we do: succeed or fail?
2196 */
2197 return ac->ops->verify_authorizer_reply(ac, o->o_auth.authorizer, len);
2198 }
2199
2200 static int invalidate_authorizer(struct ceph_connection *con)
2201 {
2202 struct ceph_osd *o = con->private;
2203 struct ceph_osd_client *osdc = o->o_osdc;
2204 struct ceph_auth_client *ac = osdc->client->monc.auth;
2205
2206 if (ac->ops && ac->ops->invalidate_authorizer)
2207 ac->ops->invalidate_authorizer(ac, CEPH_ENTITY_TYPE_OSD);
2208
2209 return ceph_monc_validate_auth(&osdc->client->monc);
2210 }
2211
2212 static const struct ceph_connection_operations osd_con_ops = {
2213 .get = get_osd_con,
2214 .put = put_osd_con,
2215 .dispatch = dispatch,
2216 .get_authorizer = get_authorizer,
2217 .verify_authorizer_reply = verify_authorizer_reply,
2218 .invalidate_authorizer = invalidate_authorizer,
2219 .alloc_msg = alloc_msg,
2220 .fault = osd_reset,
2221 };
CJK support for tty framebuffer vt