osd_client.c 55.1 KB
Newer Older
1
#include <linux/ceph/ceph_debug.h>
S
Sage Weil 已提交
2

3
#include <linux/module.h>
S
Sage Weil 已提交
4 5 6 7 8 9
#include <linux/err.h>
#include <linux/highmem.h>
#include <linux/mm.h>
#include <linux/pagemap.h>
#include <linux/slab.h>
#include <linux/uaccess.h>
10 11 12
#ifdef CONFIG_BLOCK
#include <linux/bio.h>
#endif
S
Sage Weil 已提交
13

14 15 16 17 18 19
#include <linux/ceph/libceph.h>
#include <linux/ceph/osd_client.h>
#include <linux/ceph/messenger.h>
#include <linux/ceph/decode.h>
#include <linux/ceph/auth.h>
#include <linux/ceph/pagelist.h>
S
Sage Weil 已提交
20

S
Sage Weil 已提交
21 22
#define OSD_OP_FRONT_LEN	4096
#define OSD_OPREPLY_FRONT_LEN	512
23

24
static const struct ceph_connection_operations osd_con_ops;
S
Sage Weil 已提交
25

26 27
static void send_queued(struct ceph_osd_client *osdc);
static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd);
28 29 30 31
static void __register_request(struct ceph_osd_client *osdc,
			       struct ceph_osd_request *req);
static void __unregister_linger_request(struct ceph_osd_client *osdc,
					struct ceph_osd_request *req);
32 33
static void __send_request(struct ceph_osd_client *osdc,
			   struct ceph_osd_request *req);
S
Sage Weil 已提交
34

35 36 37 38 39 40
static int op_has_extent(int op)
{
	return (op == CEPH_OSD_OP_READ ||
		op == CEPH_OSD_OP_WRITE);
}

41
int ceph_calc_raw_layout(struct ceph_file_layout *layout,
42 43 44
			u64 off, u64 *plen, u64 *bno,
			struct ceph_osd_request *req,
			struct ceph_osd_req_op *op)
45
{
46
	u64 orig_len = *plen;
47
	u64 objoff, objlen;    /* extent in object */
S
Sage Weil 已提交
48
	int r;
49 50

	/* object extent? */
51
	r = ceph_calc_file_object_mapping(layout, off, orig_len, bno,
S
Sage Weil 已提交
52 53 54
					  &objoff, &objlen);
	if (r < 0)
		return r;
55 56
	if (objlen < orig_len) {
		*plen = objlen;
57
		dout(" skipping last %llu, final file extent %llu~%llu\n",
58
		     orig_len - *plen, off, *plen);
59
	}
60

61
	if (op_has_extent(op->op)) {
62
		u32 osize = le32_to_cpu(layout->fl_object_size);
63 64
		op->extent.offset = objoff;
		op->extent.length = objlen;
65 66 67 68 69 70 71
		if (op->extent.truncate_size <= off - objoff) {
			op->extent.truncate_size = 0;
		} else {
			op->extent.truncate_size -= off - objoff;
			if (op->extent.truncate_size > osize)
				op->extent.truncate_size = osize;
		}
72 73
	}
	req->r_num_pages = calc_pages_for(off, *plen);
74
	req->r_page_alignment = off & ~PAGE_MASK;
75 76
	if (op->op == CEPH_OSD_OP_WRITE)
		op->payload_len = *plen;
77 78 79

	dout("calc_layout bno=%llx %llu~%llu (%d pages)\n",
	     *bno, objoff, objlen, req->r_num_pages);
S
Sage Weil 已提交
80
	return 0;
81
}
82
EXPORT_SYMBOL(ceph_calc_raw_layout);
83

S
Sage Weil 已提交
84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108
/*
 * Implement client access to distributed object storage cluster.
 *
 * All data objects are stored within a cluster/cloud of OSDs, or
 * "object storage devices."  (Note that Ceph OSDs have _nothing_ to
 * do with the T10 OSD extensions to SCSI.)  Ceph OSDs are simply
 * remote daemons serving up and coordinating consistent and safe
 * access to storage.
 *
 * Cluster membership and the mapping of data objects onto storage devices
 * are described by the osd map.
 *
 * We keep track of pending OSD requests (read, write), resubmit
 * requests to different OSDs when the cluster topology/data layout
 * change, or retry the affected requests when the communications
 * channel with an OSD is reset.
 */

/*
 * calculate the mapping of a file extent onto an object, and fill out the
 * request accordingly.  shorten extent as necessary if it crosses an
 * object boundary.
 *
 * fill osd op in request message.
 */
109
static int calc_layout(struct ceph_vino vino,
S
Sage Weil 已提交
110 111 112 113
		       struct ceph_file_layout *layout,
		       u64 off, u64 *plen,
		       struct ceph_osd_request *req,
		       struct ceph_osd_req_op *op)
S
Sage Weil 已提交
114 115
{
	u64 bno;
S
Sage Weil 已提交
116
	int r;
S
Sage Weil 已提交
117

118
	r = ceph_calc_raw_layout(layout, off, plen, &bno, req, op);
S
Sage Weil 已提交
119 120
	if (r < 0)
		return r;
S
Sage Weil 已提交
121

122
	snprintf(req->r_oid, sizeof(req->r_oid), "%llx.%08llx", vino.ino, bno);
S
Sage Weil 已提交
123
	req->r_oid_len = strlen(req->r_oid);
S
Sage Weil 已提交
124 125

	return r;
S
Sage Weil 已提交
126 127 128 129 130
}

/*
 * requests
 */
S
Sage Weil 已提交
131
void ceph_osdc_release_request(struct kref *kref)
S
Sage Weil 已提交
132
{
S
Sage Weil 已提交
133 134 135 136 137 138
	struct ceph_osd_request *req = container_of(kref,
						    struct ceph_osd_request,
						    r_kref);

	if (req->r_request)
		ceph_msg_put(req->r_request);
139
	if (req->r_con_filling_msg) {
140
		dout("%s revoking pages %p from con %p\n", __func__,
141
		     req->r_pages, req->r_con_filling_msg);
142
		ceph_msg_revoke_incoming(req->r_reply);
143
		req->r_con_filling_msg->ops->put(req->r_con_filling_msg);
144
	}
145 146
	if (req->r_reply)
		ceph_msg_put(req->r_reply);
S
Sage Weil 已提交
147 148 149
	if (req->r_own_pages)
		ceph_release_page_vector(req->r_pages,
					 req->r_num_pages);
150 151 152 153
#ifdef CONFIG_BLOCK
	if (req->r_bio)
		bio_put(req->r_bio);
#endif
S
Sage Weil 已提交
154
	ceph_put_snap_context(req->r_snapc);
155
	ceph_pagelist_release(&req->r_trail);
S
Sage Weil 已提交
156 157 158 159
	if (req->r_mempool)
		mempool_free(req, req->r_osdc->req_mempool);
	else
		kfree(req);
S
Sage Weil 已提交
160
}
161
EXPORT_SYMBOL(ceph_osdc_release_request);
162

A
Alex Elder 已提交
163
static int get_num_ops(struct ceph_osd_req_op *ops)
164 165 166
{
	int i = 0;

A
Alex Elder 已提交
167
	while (ops[i].op)
168 169 170 171 172
		i++;

	return i;
}

173
struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc,
S
Sage Weil 已提交
174
					       struct ceph_snap_context *snapc,
175
					       struct ceph_osd_req_op *ops,
176 177
					       bool use_mempool,
					       gfp_t gfp_flags,
178 179
					       struct page **pages,
					       struct bio *bio)
S
Sage Weil 已提交
180 181 182
{
	struct ceph_osd_request *req;
	struct ceph_msg *msg;
A
Alex Elder 已提交
183
	int num_op = get_num_ops(ops);
184
	size_t msg_size = sizeof(struct ceph_osd_request_head);
185

186
	msg_size += num_op*sizeof(struct ceph_osd_op);
S
Sage Weil 已提交
187 188

	if (use_mempool) {
189
		req = mempool_alloc(osdc->req_mempool, gfp_flags);
S
Sage Weil 已提交
190 191
		memset(req, 0, sizeof(*req));
	} else {
192
		req = kzalloc(sizeof(*req), gfp_flags);
S
Sage Weil 已提交
193 194
	}
	if (req == NULL)
195
		return NULL;
S
Sage Weil 已提交
196 197 198

	req->r_osdc = osdc;
	req->r_mempool = use_mempool;
199

S
Sage Weil 已提交
200
	kref_init(&req->r_kref);
S
Sage Weil 已提交
201 202
	init_completion(&req->r_completion);
	init_completion(&req->r_safe_completion);
203
	RB_CLEAR_NODE(&req->r_node);
S
Sage Weil 已提交
204
	INIT_LIST_HEAD(&req->r_unsafe_item);
205 206
	INIT_LIST_HEAD(&req->r_linger_item);
	INIT_LIST_HEAD(&req->r_linger_osd);
207
	INIT_LIST_HEAD(&req->r_req_lru_item);
208 209
	INIT_LIST_HEAD(&req->r_osd_item);

S
Sage Weil 已提交
210 211 212 213 214
	/* create reply message */
	if (use_mempool)
		msg = ceph_msgpool_get(&osdc->msgpool_op_reply, 0);
	else
		msg = ceph_msg_new(CEPH_MSG_OSD_OPREPLY,
215
				   OSD_OPREPLY_FRONT_LEN, gfp_flags, true);
216
	if (!msg) {
S
Sage Weil 已提交
217
		ceph_osdc_put_request(req);
218
		return NULL;
S
Sage Weil 已提交
219 220 221
	}
	req->r_reply = msg;

222
	ceph_pagelist_init(&req->r_trail);
223

S
Sage Weil 已提交
224
	/* create request message; allow space for oid */
225
	msg_size += MAX_OBJ_NAME_SIZE;
S
Sage Weil 已提交
226 227 228
	if (snapc)
		msg_size += sizeof(u64) * snapc->num_snaps;
	if (use_mempool)
229
		msg = ceph_msgpool_get(&osdc->msgpool_op, 0);
S
Sage Weil 已提交
230
	else
231
		msg = ceph_msg_new(CEPH_MSG_OSD_OP, msg_size, gfp_flags, true);
232
	if (!msg) {
S
Sage Weil 已提交
233
		ceph_osdc_put_request(req);
234
		return NULL;
S
Sage Weil 已提交
235
	}
236

S
Sage Weil 已提交
237
	memset(msg->front.iov_base, 0, msg->front.iov_len);
238 239 240

	req->r_request = msg;
	req->r_pages = pages;
241 242 243 244 245 246
#ifdef CONFIG_BLOCK
	if (bio) {
		req->r_bio = bio;
		bio_get(req->r_bio);
	}
#endif
247 248 249

	return req;
}
250
EXPORT_SYMBOL(ceph_osdc_alloc_request);
251

252 253 254 255 256 257
static void osd_req_encode_op(struct ceph_osd_request *req,
			      struct ceph_osd_op *dst,
			      struct ceph_osd_req_op *src)
{
	dst->op = cpu_to_le16(src->op);

258
	switch (src->op) {
259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277
	case CEPH_OSD_OP_READ:
	case CEPH_OSD_OP_WRITE:
		dst->extent.offset =
			cpu_to_le64(src->extent.offset);
		dst->extent.length =
			cpu_to_le64(src->extent.length);
		dst->extent.truncate_size =
			cpu_to_le64(src->extent.truncate_size);
		dst->extent.truncate_seq =
			cpu_to_le32(src->extent.truncate_seq);
		break;

	case CEPH_OSD_OP_GETXATTR:
	case CEPH_OSD_OP_SETXATTR:
	case CEPH_OSD_OP_CMPXATTR:
		dst->xattr.name_len = cpu_to_le32(src->xattr.name_len);
		dst->xattr.value_len = cpu_to_le32(src->xattr.value_len);
		dst->xattr.cmp_op = src->xattr.cmp_op;
		dst->xattr.cmp_mode = src->xattr.cmp_mode;
278
		ceph_pagelist_append(&req->r_trail, src->xattr.name,
279
				     src->xattr.name_len);
280
		ceph_pagelist_append(&req->r_trail, src->xattr.val,
281 282
				     src->xattr.value_len);
		break;
283 284 285 286 287
	case CEPH_OSD_OP_CALL:
		dst->cls.class_len = src->cls.class_len;
		dst->cls.method_len = src->cls.method_len;
		dst->cls.indata_len = cpu_to_le32(src->cls.indata_len);

288
		ceph_pagelist_append(&req->r_trail, src->cls.class_name,
289
				     src->cls.class_len);
290
		ceph_pagelist_append(&req->r_trail, src->cls.method_name,
291
				     src->cls.method_len);
292
		ceph_pagelist_append(&req->r_trail, src->cls.indata,
293 294 295 296 297
				     src->cls.indata_len);
		break;
	case CEPH_OSD_OP_ROLLBACK:
		dst->snap.snapid = cpu_to_le64(src->snap.snapid);
		break;
298 299
	case CEPH_OSD_OP_STARTSYNC:
		break;
300 301 302 303 304
	case CEPH_OSD_OP_NOTIFY:
		{
			__le32 prot_ver = cpu_to_le32(src->watch.prot_ver);
			__le32 timeout = cpu_to_le32(src->watch.timeout);

305
			ceph_pagelist_append(&req->r_trail,
306
						&prot_ver, sizeof(prot_ver));
307
			ceph_pagelist_append(&req->r_trail,
308 309 310 311 312 313 314 315
						&timeout, sizeof(timeout));
		}
	case CEPH_OSD_OP_NOTIFY_ACK:
	case CEPH_OSD_OP_WATCH:
		dst->watch.cookie = cpu_to_le64(src->watch.cookie);
		dst->watch.ver = cpu_to_le64(src->watch.ver);
		dst->watch.flag = src->watch.flag;
		break;
316 317 318 319 320 321 322 323
	default:
		pr_err("unrecognized osd opcode %d\n", dst->op);
		WARN_ON(1);
		break;
	}
	dst->payload_len = cpu_to_le32(src->payload_len);
}

324 325 326 327 328
/*
 * build new request AND message
 *
 */
void ceph_osdc_build_request(struct ceph_osd_request *req,
329
			     u64 off, u64 len,
330
			     struct ceph_osd_req_op *src_ops,
331
			     struct ceph_snap_context *snapc, u64 snap_id,
332
			     struct timespec *mtime)
333 334 335
{
	struct ceph_msg *msg = req->r_request;
	struct ceph_osd_request_head *head;
336
	struct ceph_osd_req_op *src_op;
337 338
	struct ceph_osd_op *op;
	void *p;
A
Alex Elder 已提交
339
	int num_op = get_num_ops(src_ops);
340 341
	size_t msg_size = sizeof(*head) + num_op*sizeof(*op);
	int flags = req->r_flags;
342 343
	u64 data_len = 0;
	int i;
344

345 346
	WARN_ON((flags & (CEPH_OSD_FLAG_READ|CEPH_OSD_FLAG_WRITE)) == 0);

S
Sage Weil 已提交
347
	head = msg->front.iov_base;
348
	head->snapid = cpu_to_le64(snap_id);
S
Sage Weil 已提交
349 350 351 352 353 354 355 356 357 358 359 360 361
	op = (void *)(head + 1);
	p = (void *)(op + num_op);

	req->r_snapc = ceph_get_snap_context(snapc);

	head->client_inc = cpu_to_le32(1); /* always, for now. */
	head->flags = cpu_to_le32(flags);
	if (flags & CEPH_OSD_FLAG_WRITE)
		ceph_encode_timespec(&head->mtime, mtime);
	head->num_ops = cpu_to_le16(num_op);


	/* fill in oid */
362 363 364
	head->object_len = cpu_to_le32(req->r_oid_len);
	memcpy(p, req->r_oid, req->r_oid_len);
	p += req->r_oid_len;
S
Sage Weil 已提交
365

366 367 368 369
	src_op = src_ops;
	while (src_op->op) {
		osd_req_encode_op(req, op, src_op);
		src_op++;
S
Sage Weil 已提交
370 371
		op++;
	}
372

373
	data_len += req->r_trail.length;
374

S
Sage Weil 已提交
375 376 377 378 379 380 381 382 383
	if (snapc) {
		head->snap_seq = cpu_to_le64(snapc->seq);
		head->num_snaps = cpu_to_le32(snapc->num_snaps);
		for (i = 0; i < snapc->num_snaps; i++) {
			put_unaligned_le64(snapc->snaps[i], p);
			p += sizeof(u64);
		}
	}

384 385
	if (flags & CEPH_OSD_FLAG_WRITE) {
		req->r_request->hdr.data_off = cpu_to_le16(off);
386
		req->r_request->hdr.data_len = cpu_to_le32(len + data_len);
387 388 389 390 391
	} else if (data_len) {
		req->r_request->hdr.data_off = 0;
		req->r_request->hdr.data_len = cpu_to_le32(data_len);
	}

392 393
	req->r_request->page_alignment = req->r_page_alignment;

S
Sage Weil 已提交
394
	BUG_ON(p > msg->front.iov_base + msg->front.iov_len);
395 396 397
	msg_size = p - msg->front.iov_base;
	msg->front.iov_len = msg_size;
	msg->hdr.front_len = cpu_to_le32(msg_size);
398 399
	return;
}
400
EXPORT_SYMBOL(ceph_osdc_build_request);
401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422

/*
 * build new request AND message, calculate layout, and adjust file
 * extent as needed.
 *
 * if the file was recently truncated, we include information about its
 * old and new size so that the object can be updated appropriately.  (we
 * avoid synchronously deleting truncated objects because it's slow.)
 *
 * if @do_sync, include a 'startsync' command so that the osd will flush
 * data quickly.
 */
struct ceph_osd_request *ceph_osdc_new_request(struct ceph_osd_client *osdc,
					       struct ceph_file_layout *layout,
					       struct ceph_vino vino,
					       u64 off, u64 *plen,
					       int opcode, int flags,
					       struct ceph_snap_context *snapc,
					       int do_sync,
					       u32 truncate_seq,
					       u64 truncate_size,
					       struct timespec *mtime,
423 424
					       bool use_mempool, int num_reply,
					       int page_align)
425
{
426 427
	struct ceph_osd_req_op ops[3];
	struct ceph_osd_request *req;
428
	int r;
429 430 431 432 433 434 435 436 437 438 439 440 441

	ops[0].op = opcode;
	ops[0].extent.truncate_seq = truncate_seq;
	ops[0].extent.truncate_size = truncate_size;
	ops[0].payload_len = 0;

	if (do_sync) {
		ops[1].op = CEPH_OSD_OP_STARTSYNC;
		ops[1].payload_len = 0;
		ops[2].op = 0;
	} else
		ops[1].op = 0;

442
	req = ceph_osdc_alloc_request(osdc, snapc, ops,
443
					 use_mempool,
444
					 GFP_NOFS, NULL, NULL);
445
	if (!req)
446
		return ERR_PTR(-ENOMEM);
447
	req->r_flags = flags;
448 449

	/* calculate max write size */
450
	r = calc_layout(vino, layout, off, plen, req, ops);
451 452
	if (r < 0)
		return ERR_PTR(r);
453 454
	req->r_file_layout = *layout;  /* keep a copy */

455 456 457
	/* in case it differs from natural (file) alignment that
	   calc_layout filled in for us */
	req->r_num_pages = calc_pages_for(page_align, *plen);
458 459
	req->r_page_alignment = page_align;

460
	ceph_osdc_build_request(req, off, *plen, ops, snapc, vino.snap, mtime);
461

S
Sage Weil 已提交
462 463
	return req;
}
464
EXPORT_SYMBOL(ceph_osdc_new_request);
S
Sage Weil 已提交
465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530

/*
 * We keep osd requests in an rbtree, sorted by ->r_tid.
 */
static void __insert_request(struct ceph_osd_client *osdc,
			     struct ceph_osd_request *new)
{
	struct rb_node **p = &osdc->requests.rb_node;
	struct rb_node *parent = NULL;
	struct ceph_osd_request *req = NULL;

	while (*p) {
		parent = *p;
		req = rb_entry(parent, struct ceph_osd_request, r_node);
		if (new->r_tid < req->r_tid)
			p = &(*p)->rb_left;
		else if (new->r_tid > req->r_tid)
			p = &(*p)->rb_right;
		else
			BUG();
	}

	rb_link_node(&new->r_node, parent, p);
	rb_insert_color(&new->r_node, &osdc->requests);
}

static struct ceph_osd_request *__lookup_request(struct ceph_osd_client *osdc,
						 u64 tid)
{
	struct ceph_osd_request *req;
	struct rb_node *n = osdc->requests.rb_node;

	while (n) {
		req = rb_entry(n, struct ceph_osd_request, r_node);
		if (tid < req->r_tid)
			n = n->rb_left;
		else if (tid > req->r_tid)
			n = n->rb_right;
		else
			return req;
	}
	return NULL;
}

static struct ceph_osd_request *
__lookup_request_ge(struct ceph_osd_client *osdc,
		    u64 tid)
{
	struct ceph_osd_request *req;
	struct rb_node *n = osdc->requests.rb_node;

	while (n) {
		req = rb_entry(n, struct ceph_osd_request, r_node);
		if (tid < req->r_tid) {
			if (!n->rb_left)
				return req;
			n = n->rb_left;
		} else if (tid > req->r_tid) {
			n = n->rb_right;
		} else {
			return req;
		}
	}
	return NULL;
}

531 532 533 534 535 536
/*
 * Resubmit requests pending on the given osd.
 */
static void __kick_osd_requests(struct ceph_osd_client *osdc,
				struct ceph_osd *osd)
{
537
	struct ceph_osd_request *req, *nreq;
538 539 540 541
	int err;

	dout("__kick_osd_requests osd%d\n", osd->o_osd);
	err = __reset_osd(osdc, osd);
542
	if (err)
543 544 545 546 547 548
		return;

	list_for_each_entry(req, &osd->o_requests, r_osd_item) {
		list_move(&req->r_req_lru_item, &osdc->req_unsent);
		dout("requeued %p tid %llu osd%d\n", req, req->r_tid,
		     osd->o_osd);
549 550 551 552 553 554
		if (!req->r_linger)
			req->r_flags |= CEPH_OSD_FLAG_RETRY;
	}

	list_for_each_entry_safe(req, nreq, &osd->o_linger_requests,
				 r_linger_osd) {
555 556 557 558 559
		/*
		 * reregister request prior to unregistering linger so
		 * that r_osd is preserved.
		 */
		BUG_ON(!list_empty(&req->r_req_lru_item));
560
		__register_request(osdc, req);
561 562 563
		list_add(&req->r_req_lru_item, &osdc->req_unsent);
		list_add(&req->r_osd_item, &req->r_osd->o_requests);
		__unregister_linger_request(osdc, req);
564 565
		dout("requeued lingering %p tid %llu osd%d\n", req, req->r_tid,
		     osd->o_osd);
566 567 568
	}
}

S
Sage Weil 已提交
569
/*
570
 * If the osd connection drops, we need to resubmit all requests.
S
Sage Weil 已提交
571 572 573 574 575 576 577 578 579 580 581
 */
static void osd_reset(struct ceph_connection *con)
{
	struct ceph_osd *osd = con->private;
	struct ceph_osd_client *osdc;

	if (!osd)
		return;
	dout("osd_reset osd%d\n", osd->o_osd);
	osdc = osd->o_osdc;
	down_read(&osdc->map_sem);
S
Sage Weil 已提交
582 583 584
	mutex_lock(&osdc->request_mutex);
	__kick_osd_requests(osdc, osd);
	mutex_unlock(&osdc->request_mutex);
585
	send_queued(osdc);
S
Sage Weil 已提交
586 587 588 589 590 591
	up_read(&osdc->map_sem);
}

/*
 * Track open sessions with osds.
 */
592
static struct ceph_osd *create_osd(struct ceph_osd_client *osdc, int onum)
S
Sage Weil 已提交
593 594 595 596 597 598 599 600 601
{
	struct ceph_osd *osd;

	osd = kzalloc(sizeof(*osd), GFP_NOFS);
	if (!osd)
		return NULL;

	atomic_set(&osd->o_ref, 1);
	osd->o_osdc = osdc;
602
	osd->o_osd = onum;
603
	RB_CLEAR_NODE(&osd->o_node);
S
Sage Weil 已提交
604
	INIT_LIST_HEAD(&osd->o_requests);
605
	INIT_LIST_HEAD(&osd->o_linger_requests);
606
	INIT_LIST_HEAD(&osd->o_osd_lru);
S
Sage Weil 已提交
607 608
	osd->o_incarnation = 1;

609
	ceph_con_init(&osd->o_con, osd, &osd_con_ops, &osdc->client->msgr);
610

611
	INIT_LIST_HEAD(&osd->o_keepalive_item);
S
Sage Weil 已提交
612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630
	return osd;
}

static struct ceph_osd *get_osd(struct ceph_osd *osd)
{
	if (atomic_inc_not_zero(&osd->o_ref)) {
		dout("get_osd %p %d -> %d\n", osd, atomic_read(&osd->o_ref)-1,
		     atomic_read(&osd->o_ref));
		return osd;
	} else {
		dout("get_osd %p FAIL\n", osd);
		return NULL;
	}
}

static void put_osd(struct ceph_osd *osd)
{
	dout("put_osd %p %d -> %d\n", osd, atomic_read(&osd->o_ref),
	     atomic_read(&osd->o_ref) - 1);
631
	if (atomic_dec_and_test(&osd->o_ref) && osd->o_auth.authorizer) {
S
Sage Weil 已提交
632 633
		struct ceph_auth_client *ac = osd->o_osdc->client->monc.auth;

634
		if (ac->ops && ac->ops->destroy_authorizer)
635
			ac->ops->destroy_authorizer(ac, osd->o_auth.authorizer);
S
Sage Weil 已提交
636
		kfree(osd);
S
Sage Weil 已提交
637
	}
S
Sage Weil 已提交
638 639 640 641 642
}

/*
 * remove an osd from our map
 */
643
static void __remove_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
S
Sage Weil 已提交
644
{
645
	dout("__remove_osd %p\n", osd);
S
Sage Weil 已提交
646 647
	BUG_ON(!list_empty(&osd->o_requests));
	rb_erase(&osd->o_node, &osdc->osds);
648
	list_del_init(&osd->o_osd_lru);
S
Sage Weil 已提交
649 650 651 652
	ceph_con_close(&osd->o_con);
	put_osd(osd);
}

653 654
static void remove_all_osds(struct ceph_osd_client *osdc)
{
655
	dout("%s %p\n", __func__, osdc);
656 657 658 659 660 661 662 663 664
	mutex_lock(&osdc->request_mutex);
	while (!RB_EMPTY_ROOT(&osdc->osds)) {
		struct ceph_osd *osd = rb_entry(rb_first(&osdc->osds),
						struct ceph_osd, o_node);
		__remove_osd(osdc, osd);
	}
	mutex_unlock(&osdc->request_mutex);
}

665 666 667 668 669 670
static void __move_osd_to_lru(struct ceph_osd_client *osdc,
			      struct ceph_osd *osd)
{
	dout("__move_osd_to_lru %p\n", osd);
	BUG_ON(!list_empty(&osd->o_osd_lru));
	list_add_tail(&osd->o_osd_lru, &osdc->osd_lru);
671
	osd->lru_ttl = jiffies + osdc->client->options->osd_idle_ttl * HZ;
672 673 674 675 676 677 678 679 680
}

static void __remove_osd_from_lru(struct ceph_osd *osd)
{
	dout("__remove_osd_from_lru %p\n", osd);
	if (!list_empty(&osd->o_osd_lru))
		list_del_init(&osd->o_osd_lru);
}

681
static void remove_old_osds(struct ceph_osd_client *osdc)
682 683 684 685 686 687
{
	struct ceph_osd *osd, *nosd;

	dout("__remove_old_osds %p\n", osdc);
	mutex_lock(&osdc->request_mutex);
	list_for_each_entry_safe(osd, nosd, &osdc->osd_lru, o_osd_lru) {
688
		if (time_before(jiffies, osd->lru_ttl))
689 690 691 692 693 694
			break;
		__remove_osd(osdc, osd);
	}
	mutex_unlock(&osdc->request_mutex);
}

S
Sage Weil 已提交
695 696 697
/*
 * reset osd connect
 */
698
static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
S
Sage Weil 已提交
699
{
A
Alex Elder 已提交
700
	struct ceph_entity_addr *peer_addr;
S
Sage Weil 已提交
701

702
	dout("__reset_osd %p osd%d\n", osd, osd->o_osd);
703 704
	if (list_empty(&osd->o_requests) &&
	    list_empty(&osd->o_linger_requests)) {
705
		__remove_osd(osdc, osd);
A
Alex Elder 已提交
706 707 708 709 710 711 712 713 714

		return -ENODEV;
	}

	peer_addr = &osdc->osdmap->osd_addr[osd->o_osd];
	if (!memcmp(peer_addr, &osd->o_con.peer_addr, sizeof (*peer_addr)) &&
			!ceph_con_opened(&osd->o_con)) {
		struct ceph_osd_request *req;

715 716 717 718 719
		dout(" osd addr hasn't changed and connection never opened,"
		     " letting msgr retry");
		/* touch each r_stamp for handle_timeout()'s benfit */
		list_for_each_entry(req, &osd->o_requests, r_osd_item)
			req->r_stamp = jiffies;
A
Alex Elder 已提交
720 721

		return -EAGAIN;
S
Sage Weil 已提交
722
	}
A
Alex Elder 已提交
723 724 725 726 727 728

	ceph_con_close(&osd->o_con);
	ceph_con_open(&osd->o_con, CEPH_ENTITY_TYPE_OSD, osd->o_osd, peer_addr);
	osd->o_incarnation++;

	return 0;
S
Sage Weil 已提交
729 730 731 732 733 734 735 736
}

static void __insert_osd(struct ceph_osd_client *osdc, struct ceph_osd *new)
{
	struct rb_node **p = &osdc->osds.rb_node;
	struct rb_node *parent = NULL;
	struct ceph_osd *osd = NULL;

737
	dout("__insert_osd %p osd%d\n", new, new->o_osd);
S
Sage Weil 已提交
738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769
	while (*p) {
		parent = *p;
		osd = rb_entry(parent, struct ceph_osd, o_node);
		if (new->o_osd < osd->o_osd)
			p = &(*p)->rb_left;
		else if (new->o_osd > osd->o_osd)
			p = &(*p)->rb_right;
		else
			BUG();
	}

	rb_link_node(&new->o_node, parent, p);
	rb_insert_color(&new->o_node, &osdc->osds);
}

static struct ceph_osd *__lookup_osd(struct ceph_osd_client *osdc, int o)
{
	struct ceph_osd *osd;
	struct rb_node *n = osdc->osds.rb_node;

	while (n) {
		osd = rb_entry(n, struct ceph_osd, o_node);
		if (o < osd->o_osd)
			n = n->rb_left;
		else if (o > osd->o_osd)
			n = n->rb_right;
		else
			return osd;
	}
	return NULL;
}

770 771 772
static void __schedule_osd_timeout(struct ceph_osd_client *osdc)
{
	schedule_delayed_work(&osdc->timeout_work,
773
			osdc->client->options->osd_keepalive_timeout * HZ);
774 775 776 777 778 779
}

static void __cancel_osd_timeout(struct ceph_osd_client *osdc)
{
	cancel_delayed_work(&osdc->timeout_work);
}
S
Sage Weil 已提交
780 781 782 783 784

/*
 * Register request, assign tid.  If this is the first request, set up
 * the timeout event.
 */
785 786
static void __register_request(struct ceph_osd_client *osdc,
			       struct ceph_osd_request *req)
S
Sage Weil 已提交
787 788
{
	req->r_tid = ++osdc->last_tid;
789
	req->r_request->hdr.tid = cpu_to_le64(req->r_tid);
790
	dout("__register_request %p tid %lld\n", req, req->r_tid);
S
Sage Weil 已提交
791 792 793 794
	__insert_request(osdc, req);
	ceph_osdc_get_request(req);
	osdc->num_requests++;
	if (osdc->num_requests == 1) {
795 796
		dout(" first request, scheduling timeout\n");
		__schedule_osd_timeout(osdc);
S
Sage Weil 已提交
797
	}
798 799 800 801 802 803 804
}

static void register_request(struct ceph_osd_client *osdc,
			     struct ceph_osd_request *req)
{
	mutex_lock(&osdc->request_mutex);
	__register_request(osdc, req);
S
Sage Weil 已提交
805 806 807 808 809 810 811 812 813
	mutex_unlock(&osdc->request_mutex);
}

/*
 * called under osdc->request_mutex
 */
static void __unregister_request(struct ceph_osd_client *osdc,
				 struct ceph_osd_request *req)
{
814 815 816 817 818 819
	if (RB_EMPTY_NODE(&req->r_node)) {
		dout("__unregister_request %p tid %lld not registered\n",
			req, req->r_tid);
		return;
	}

S
Sage Weil 已提交
820 821 822 823
	dout("__unregister_request %p tid %lld\n", req, req->r_tid);
	rb_erase(&req->r_node, &osdc->requests);
	osdc->num_requests--;

824 825
	if (req->r_osd) {
		/* make sure the original request isn't in flight. */
826
		ceph_msg_revoke(req->r_request);
827 828

		list_del_init(&req->r_osd_item);
829 830 831
		if (list_empty(&req->r_osd->o_requests) &&
		    list_empty(&req->r_osd->o_linger_requests)) {
			dout("moving osd to %p lru\n", req->r_osd);
832
			__move_osd_to_lru(osdc, req->r_osd);
833
		}
834
		if (list_empty(&req->r_linger_item))
835
			req->r_osd = NULL;
836
	}
S
Sage Weil 已提交
837

A
Alex Elder 已提交
838
	list_del_init(&req->r_req_lru_item);
S
Sage Weil 已提交
839 840
	ceph_osdc_put_request(req);

841 842 843
	if (osdc->num_requests == 0) {
		dout(" no requests, canceling timeout\n");
		__cancel_osd_timeout(osdc);
S
Sage Weil 已提交
844 845 846 847 848 849 850 851
	}
}

/*
 * Cancel a previously queued request message
 */
static void __cancel_request(struct ceph_osd_request *req)
{
852
	if (req->r_sent && req->r_osd) {
853
		ceph_msg_revoke(req->r_request);
S
Sage Weil 已提交
854 855 856 857
		req->r_sent = 0;
	}
}

858 859 860 861 862
static void __register_linger_request(struct ceph_osd_client *osdc,
				    struct ceph_osd_request *req)
{
	dout("__register_linger_request %p\n", req);
	list_add_tail(&req->r_linger_item, &osdc->req_linger);
863 864 865
	if (req->r_osd)
		list_add_tail(&req->r_linger_osd,
			      &req->r_osd->o_linger_requests);
866 867 868 869 870 871
}

static void __unregister_linger_request(struct ceph_osd_client *osdc,
					struct ceph_osd_request *req)
{
	dout("__unregister_linger_request %p\n", req);
A
Alex Elder 已提交
872
	list_del_init(&req->r_linger_item);
873 874 875 876 877 878 879 880
	if (req->r_osd) {
		list_del_init(&req->r_linger_osd);

		if (list_empty(&req->r_osd->o_requests) &&
		    list_empty(&req->r_osd->o_linger_requests)) {
			dout("moving osd to %p lru\n", req->r_osd);
			__move_osd_to_lru(osdc, req->r_osd);
		}
881 882
		if (list_empty(&req->r_osd_item))
			req->r_osd = NULL;
883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912
	}
}

void ceph_osdc_unregister_linger_request(struct ceph_osd_client *osdc,
					 struct ceph_osd_request *req)
{
	mutex_lock(&osdc->request_mutex);
	if (req->r_linger) {
		__unregister_linger_request(osdc, req);
		ceph_osdc_put_request(req);
	}
	mutex_unlock(&osdc->request_mutex);
}
EXPORT_SYMBOL(ceph_osdc_unregister_linger_request);

void ceph_osdc_set_request_linger(struct ceph_osd_client *osdc,
				  struct ceph_osd_request *req)
{
	if (!req->r_linger) {
		dout("set_request_linger %p\n", req);
		req->r_linger = 1;
		/*
		 * caller is now responsible for calling
		 * unregister_linger_request
		 */
		ceph_osdc_get_request(req);
	}
}
EXPORT_SYMBOL(ceph_osdc_set_request_linger);

S
Sage Weil 已提交
913 914 915
/*
 * Pick an osd (the first 'up' osd in the pg), allocate the osd struct
 * (as needed), and set the request r_osd appropriately.  If there is
L
Lucas De Marchi 已提交
916
 * no up osd, set r_osd to NULL.  Move the request to the appropriate list
917
 * (unsent, homeless) or leave on in-flight lru.
S
Sage Weil 已提交
918 919 920 921 922
 *
 * Return 0 if unchanged, 1 if changed, or negative on error.
 *
 * Caller should hold map_sem for read and request_mutex.
 */
923
static int __map_request(struct ceph_osd_client *osdc,
924
			 struct ceph_osd_request *req, int force_resend)
S
Sage Weil 已提交
925 926
{
	struct ceph_osd_request_head *reqhead = req->r_request->front.iov_base;
927
	struct ceph_pg pgid;
928 929
	int acting[CEPH_PG_MAX_SIZE];
	int o = -1, num = 0;
S
Sage Weil 已提交
930 931
	int err;

932
	dout("map_request %p tid %lld\n", req, req->r_tid);
S
Sage Weil 已提交
933 934
	err = ceph_calc_object_layout(&reqhead->layout, req->r_oid,
				      &req->r_file_layout, osdc->osdmap);
935 936
	if (err) {
		list_move(&req->r_req_lru_item, &osdc->req_notarget);
S
Sage Weil 已提交
937
		return err;
938
	}
939
	pgid = reqhead->layout.ol_pgid;
940 941
	req->r_pgid = pgid;

942 943 944 945 946
	err = ceph_calc_pg_acting(osdc->osdmap, pgid, acting);
	if (err > 0) {
		o = acting[0];
		num = err;
	}
S
Sage Weil 已提交
947

948 949
	if ((!force_resend &&
	     req->r_osd && req->r_osd->o_osd == o &&
950 951 952
	     req->r_sent >= req->r_osd->o_incarnation &&
	     req->r_num_pg_osds == num &&
	     memcmp(req->r_pg_osds, acting, sizeof(acting[0])*num) == 0) ||
S
Sage Weil 已提交
953 954 955
	    (req->r_osd == NULL && o == -1))
		return 0;  /* no change */

956
	dout("map_request tid %llu pgid %d.%x osd%d (was osd%d)\n",
957
	     req->r_tid, le32_to_cpu(pgid.pool), le16_to_cpu(pgid.ps), o,
S
Sage Weil 已提交
958 959
	     req->r_osd ? req->r_osd->o_osd : -1);

960 961 962 963
	/* record full pg acting set */
	memcpy(req->r_pg_osds, acting, sizeof(acting[0]) * num);
	req->r_num_pg_osds = num;

S
Sage Weil 已提交
964 965 966 967 968 969 970 971
	if (req->r_osd) {
		__cancel_request(req);
		list_del_init(&req->r_osd_item);
		req->r_osd = NULL;
	}

	req->r_osd = __lookup_osd(osdc, o);
	if (!req->r_osd && o >= 0) {
972
		err = -ENOMEM;
973
		req->r_osd = create_osd(osdc, o);
974 975
		if (!req->r_osd) {
			list_move(&req->r_req_lru_item, &osdc->req_notarget);
976
			goto out;
977
		}
S
Sage Weil 已提交
978

979
		dout("map_request osd %p is osd%d\n", req->r_osd, o);
S
Sage Weil 已提交
980 981
		__insert_osd(osdc, req->r_osd);

982 983 984
		ceph_con_open(&req->r_osd->o_con,
			      CEPH_ENTITY_TYPE_OSD, o,
			      &osdc->osdmap->osd_addr[o]);
S
Sage Weil 已提交
985 986
	}

987 988
	if (req->r_osd) {
		__remove_osd_from_lru(req->r_osd);
S
Sage Weil 已提交
989
		list_add(&req->r_osd_item, &req->r_osd->o_requests);
990 991 992
		list_move(&req->r_req_lru_item, &osdc->req_unsent);
	} else {
		list_move(&req->r_req_lru_item, &osdc->req_notarget);
993
	}
994
	err = 1;   /* osd or pg changed */
S
Sage Weil 已提交
995 996 997 998 999 1000 1001 1002

out:
	return err;
}

/*
 * caller should hold map_sem (for read) and request_mutex
 */
1003 1004
static void __send_request(struct ceph_osd_client *osdc,
			   struct ceph_osd_request *req)
S
Sage Weil 已提交
1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015
{
	struct ceph_osd_request_head *reqhead;

	dout("send_request %p tid %llu to osd%d flags %d\n",
	     req, req->r_tid, req->r_osd->o_osd, req->r_flags);

	reqhead = req->r_request->front.iov_base;
	reqhead->osdmap_epoch = cpu_to_le32(osdc->osdmap->epoch);
	reqhead->flags |= cpu_to_le32(req->r_flags);  /* e.g., RETRY */
	reqhead->reassert_version = req->r_reassert_version;

S
Sage Weil 已提交
1016
	req->r_stamp = jiffies;
1017
	list_move_tail(&req->r_req_lru_item, &osdc->req_lru);
S
Sage Weil 已提交
1018 1019 1020 1021 1022 1023

	ceph_msg_get(req->r_request); /* send consumes a ref */
	ceph_con_send(&req->r_osd->o_con, req->r_request);
	req->r_sent = req->r_osd->o_incarnation;
}

1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038
/*
 * Send any requests in the queue (req_unsent).
 */
static void send_queued(struct ceph_osd_client *osdc)
{
	struct ceph_osd_request *req, *tmp;

	dout("send_queued\n");
	mutex_lock(&osdc->request_mutex);
	list_for_each_entry_safe(req, tmp, &osdc->req_unsent, r_req_lru_item) {
		__send_request(osdc, req);
	}
	mutex_unlock(&osdc->request_mutex);
}

S
Sage Weil 已提交
1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051
/*
 * Timeout callback, called every N seconds when 1 or more osd
 * requests has been active for more than N seconds.  When this
 * happens, we ping all OSDs with requests who have timed out to
 * ensure any communications channel reset is detected.  Reset the
 * request timeouts another N seconds in the future as we go.
 * Reschedule the timeout event another N seconds in future (unless
 * there are no open requests).
 */
static void handle_timeout(struct work_struct *work)
{
	struct ceph_osd_client *osdc =
		container_of(work, struct ceph_osd_client, timeout_work.work);
S
Sage Weil 已提交
1052
	struct ceph_osd_request *req;
S
Sage Weil 已提交
1053
	struct ceph_osd *osd;
1054
	unsigned long keepalive =
1055
		osdc->client->options->osd_keepalive_timeout * HZ;
1056
	struct list_head slow_osds;
S
Sage Weil 已提交
1057 1058 1059 1060 1061 1062 1063
	dout("timeout\n");
	down_read(&osdc->map_sem);

	ceph_monc_request_next_osdmap(&osdc->client->monc);

	mutex_lock(&osdc->request_mutex);

1064 1065 1066 1067 1068 1069 1070
	/*
	 * ping osds that are a bit slow.  this ensures that if there
	 * is a break in the TCP connection we will notice, and reopen
	 * a connection with that osd (from the fault callback).
	 */
	INIT_LIST_HEAD(&slow_osds);
	list_for_each_entry(req, &osdc->req_lru, r_req_lru_item) {
S
Sage Weil 已提交
1071
		if (time_before(jiffies, req->r_stamp + keepalive))
1072 1073 1074 1075 1076
			break;

		osd = req->r_osd;
		BUG_ON(!osd);
		dout(" tid %llu is slow, will send keepalive on osd%d\n",
S
Sage Weil 已提交
1077
		     req->r_tid, osd->o_osd);
1078 1079 1080 1081 1082 1083
		list_move_tail(&osd->o_keepalive_item, &slow_osds);
	}
	while (!list_empty(&slow_osds)) {
		osd = list_entry(slow_osds.next, struct ceph_osd,
				 o_keepalive_item);
		list_del_init(&osd->o_keepalive_item);
S
Sage Weil 已提交
1084 1085 1086
		ceph_con_keepalive(&osd->o_con);
	}

1087
	__schedule_osd_timeout(osdc);
S
Sage Weil 已提交
1088
	mutex_unlock(&osdc->request_mutex);
1089
	send_queued(osdc);
S
Sage Weil 已提交
1090 1091 1092
	up_read(&osdc->map_sem);
}

1093 1094 1095 1096 1097 1098
static void handle_osds_timeout(struct work_struct *work)
{
	struct ceph_osd_client *osdc =
		container_of(work, struct ceph_osd_client,
			     osds_timeout_work.work);
	unsigned long delay =
1099
		osdc->client->options->osd_idle_ttl * HZ >> 2;
1100 1101 1102

	dout("osds timeout\n");
	down_read(&osdc->map_sem);
1103
	remove_old_osds(osdc);
1104 1105 1106 1107 1108 1109
	up_read(&osdc->map_sem);

	schedule_delayed_work(&osdc->osds_timeout_work,
			      round_jiffies_relative(delay));
}

S
Sage Weil 已提交
1110 1111 1112 1113 1114 1115 1116
static void complete_request(struct ceph_osd_request *req)
{
	if (req->r_safe_callback)
		req->r_safe_callback(req, NULL);
	complete_all(&req->r_safe_completion);  /* fsync waiter */
}

S
Sage Weil 已提交
1117 1118 1119 1120
/*
 * handle osd op reply.  either call the callback if it is specified,
 * or do the completion to wake up the waiting thread.
 */
1121 1122
static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
			 struct ceph_connection *con)
S
Sage Weil 已提交
1123 1124 1125 1126 1127
{
	struct ceph_osd_reply_head *rhead = msg->front.iov_base;
	struct ceph_osd_request *req;
	u64 tid;
	int numops, object_len, flags;
1128
	s32 result;
S
Sage Weil 已提交
1129

1130
	tid = le64_to_cpu(msg->hdr.tid);
S
Sage Weil 已提交
1131 1132 1133 1134
	if (msg->front.iov_len < sizeof(*rhead))
		goto bad;
	numops = le32_to_cpu(rhead->num_ops);
	object_len = le32_to_cpu(rhead->object_len);
1135
	result = le32_to_cpu(rhead->result);
S
Sage Weil 已提交
1136 1137 1138
	if (msg->front.iov_len != sizeof(*rhead) + object_len +
	    numops * sizeof(struct ceph_osd_op))
		goto bad;
1139
	dout("handle_reply %p tid %llu result %d\n", msg, tid, (int)result);
S
Sage Weil 已提交
1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150
	/* lookup */
	mutex_lock(&osdc->request_mutex);
	req = __lookup_request(osdc, tid);
	if (req == NULL) {
		dout("handle_reply tid %llu dne\n", tid);
		mutex_unlock(&osdc->request_mutex);
		return;
	}
	ceph_osdc_get_request(req);
	flags = le32_to_cpu(rhead->flags);

1151
	/*
1152
	 * if this connection filled our message, drop our reference now, to
1153 1154
	 * avoid a (safe but slower) revoke later.
	 */
1155
	if (req->r_con_filling_msg == con && req->r_reply == msg) {
S
Sage Weil 已提交
1156
		dout(" dropping con_filling_msg ref %p\n", con);
1157
		req->r_con_filling_msg = NULL;
1158
		con->ops->put(con);
1159 1160
	}

S
Sage Weil 已提交
1161
	if (!req->r_got_reply) {
1162
		unsigned int bytes;
S
Sage Weil 已提交
1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176

		req->r_result = le32_to_cpu(rhead->result);
		bytes = le32_to_cpu(msg->hdr.data_len);
		dout("handle_reply result %d bytes %d\n", req->r_result,
		     bytes);
		if (req->r_result == 0)
			req->r_result = bytes;

		/* in case this is a write and we need to replay, */
		req->r_reassert_version = rhead->reassert_version;

		req->r_got_reply = 1;
	} else if ((flags & CEPH_OSD_FLAG_ONDISK) == 0) {
		dout("handle_reply tid %llu dup ack\n", tid);
S
Sage Weil 已提交
1177
		mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1178 1179 1180 1181 1182
		goto done;
	}

	dout("handle_reply tid %llu flags %d\n", tid, flags);

1183 1184 1185
	if (req->r_linger && (flags & CEPH_OSD_FLAG_ONDISK))
		__register_linger_request(osdc, req);

S
Sage Weil 已提交
1186
	/* either this is a read, or we got the safe response */
1187 1188
	if (result < 0 ||
	    (flags & CEPH_OSD_FLAG_ONDISK) ||
S
Sage Weil 已提交
1189 1190 1191 1192 1193 1194 1195 1196
	    ((flags & CEPH_OSD_FLAG_WRITE) == 0))
		__unregister_request(osdc, req);

	mutex_unlock(&osdc->request_mutex);

	if (req->r_callback)
		req->r_callback(req, msg);
	else
1197
		complete_all(&req->r_completion);
S
Sage Weil 已提交
1198

S
Sage Weil 已提交
1199 1200
	if (flags & CEPH_OSD_FLAG_ONDISK)
		complete_request(req);
S
Sage Weil 已提交
1201 1202

done:
1203
	dout("req=%p req->r_linger=%d\n", req, req->r_linger);
S
Sage Weil 已提交
1204 1205 1206 1207 1208 1209 1210
	ceph_osdc_put_request(req);
	return;

bad:
	pr_err("corrupt osd_op_reply got %d %d expected %d\n",
	       (int)msg->front.iov_len, le32_to_cpu(msg->hdr.front_len),
	       (int)sizeof(*rhead));
1211
	ceph_msg_dump(msg);
S
Sage Weil 已提交
1212 1213
}

1214
static void reset_changed_osds(struct ceph_osd_client *osdc)
S
Sage Weil 已提交
1215 1216 1217
{
	struct rb_node *p, *n;

1218 1219
	for (p = rb_first(&osdc->osds); p; p = n) {
		struct ceph_osd *osd = rb_entry(p, struct ceph_osd, o_node);
S
Sage Weil 已提交
1220

1221 1222 1223 1224 1225 1226 1227
		n = rb_next(p);
		if (!ceph_osd_is_up(osdc->osdmap, osd->o_osd) ||
		    memcmp(&osd->o_con.peer_addr,
			   ceph_osd_addr(osdc->osdmap,
					 osd->o_osd),
			   sizeof(struct ceph_entity_addr)) != 0)
			__reset_osd(osdc, osd);
S
Sage Weil 已提交
1228
	}
1229 1230 1231
}

/*
1232 1233
 * Requeue requests whose mapping to an OSD has changed.  If requests map to
 * no osd, request a new map.
1234
 *
1235
 * Caller should hold map_sem for read.
1236
 */
1237
static void kick_requests(struct ceph_osd_client *osdc, int force_resend)
1238
{
1239
	struct ceph_osd_request *req, *nreq;
1240 1241 1242
	struct rb_node *p;
	int needmap = 0;
	int err;
1243

1244
	dout("kick_requests %s\n", force_resend ? " (force resend)" : "");
1245
	mutex_lock(&osdc->request_mutex);
1246
	for (p = rb_first(&osdc->requests); p; ) {
1247
		req = rb_entry(p, struct ceph_osd_request, r_node);
1248
		p = rb_next(p);
1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266

		/*
		 * For linger requests that have not yet been
		 * registered, move them to the linger list; they'll
		 * be sent to the osd in the loop below.  Unregister
		 * the request before re-registering it as a linger
		 * request to ensure the __map_request() below
		 * will decide it needs to be sent.
		 */
		if (req->r_linger && list_empty(&req->r_linger_item)) {
			dout("%p tid %llu restart on osd%d\n",
			     req, req->r_tid,
			     req->r_osd ? req->r_osd->o_osd : -1);
			__unregister_request(osdc, req);
			__register_linger_request(osdc, req);
			continue;
		}

1267
		err = __map_request(osdc, req, force_resend);
1268 1269 1270 1271 1272 1273
		if (err < 0)
			continue;  /* error */
		if (req->r_osd == NULL) {
			dout("%p tid %llu maps to no osd\n", req, req->r_tid);
			needmap++;  /* request a newer map */
		} else if (err > 0) {
1274 1275 1276 1277
			if (!req->r_linger) {
				dout("%p tid %llu requeued on osd%d\n", req,
				     req->r_tid,
				     req->r_osd ? req->r_osd->o_osd : -1);
1278
				req->r_flags |= CEPH_OSD_FLAG_RETRY;
1279 1280
			}
		}
1281 1282 1283 1284 1285 1286
	}

	list_for_each_entry_safe(req, nreq, &osdc->req_linger,
				 r_linger_item) {
		dout("linger req=%p req->r_osd=%p\n", req, req->r_osd);

1287
		err = __map_request(osdc, req, force_resend);
1288
		dout("__map_request returned %d\n", err);
1289 1290 1291 1292 1293 1294 1295 1296
		if (err == 0)
			continue;  /* no change and no osd was specified */
		if (err < 0)
			continue;  /* hrm! */
		if (req->r_osd == NULL) {
			dout("tid %llu maps to no valid osd\n", req->r_tid);
			needmap++;  /* request a newer map */
			continue;
1297
		}
1298 1299 1300 1301

		dout("kicking lingering %p tid %llu osd%d\n", req, req->r_tid,
		     req->r_osd ? req->r_osd->o_osd : -1);
		__register_request(osdc, req);
1302
		__unregister_linger_request(osdc, req);
1303
	}
S
Sage Weil 已提交
1304 1305 1306 1307 1308 1309
	mutex_unlock(&osdc->request_mutex);

	if (needmap) {
		dout("%d requests for down osds, need new map\n", needmap);
		ceph_monc_request_next_osdmap(&osdc->client->monc);
	}
1310
	reset_changed_osds(osdc);
1311
}
1312 1313


S
Sage Weil 已提交
1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336
/*
 * Process updated osd map.
 *
 * The message contains any number of incremental and full maps, normally
 * indicating some sort of topology change in the cluster.  Kick requests
 * off to different OSDs as needed.
 */
void ceph_osdc_handle_map(struct ceph_osd_client *osdc, struct ceph_msg *msg)
{
	void *p, *end, *next;
	u32 nr_maps, maplen;
	u32 epoch;
	struct ceph_osdmap *newmap = NULL, *oldmap;
	int err;
	struct ceph_fsid fsid;

	dout("handle_map have %u\n", osdc->osdmap ? osdc->osdmap->epoch : 0);
	p = msg->front.iov_base;
	end = p + msg->front.iov_len;

	/* verify fsid */
	ceph_decode_need(&p, end, sizeof(fsid), bad);
	ceph_decode_copy(&p, &fsid, sizeof(fsid));
1337 1338
	if (ceph_check_fsid(osdc->client, &fsid) < 0)
		return;
S
Sage Weil 已提交
1339 1340 1341 1342 1343 1344 1345 1346

	down_write(&osdc->map_sem);

	/* incremental maps */
	ceph_decode_32_safe(&p, end, nr_maps, bad);
	dout(" %d inc maps\n", nr_maps);
	while (nr_maps > 0) {
		ceph_decode_need(&p, end, 2*sizeof(u32), bad);
1347 1348
		epoch = ceph_decode_32(&p);
		maplen = ceph_decode_32(&p);
S
Sage Weil 已提交
1349 1350 1351 1352 1353 1354 1355
		ceph_decode_need(&p, end, maplen, bad);
		next = p + maplen;
		if (osdc->osdmap && osdc->osdmap->epoch+1 == epoch) {
			dout("applying incremental map %u len %d\n",
			     epoch, maplen);
			newmap = osdmap_apply_incremental(&p, next,
							  osdc->osdmap,
1356
							  &osdc->client->msgr);
S
Sage Weil 已提交
1357 1358 1359 1360
			if (IS_ERR(newmap)) {
				err = PTR_ERR(newmap);
				goto bad;
			}
1361
			BUG_ON(!newmap);
S
Sage Weil 已提交
1362 1363 1364 1365
			if (newmap != osdc->osdmap) {
				ceph_osdmap_destroy(osdc->osdmap);
				osdc->osdmap = newmap;
			}
1366
			kick_requests(osdc, 0);
S
Sage Weil 已提交
1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381
		} else {
			dout("ignoring incremental map %u len %d\n",
			     epoch, maplen);
		}
		p = next;
		nr_maps--;
	}
	if (newmap)
		goto done;

	/* full maps */
	ceph_decode_32_safe(&p, end, nr_maps, bad);
	dout(" %d full maps\n", nr_maps);
	while (nr_maps) {
		ceph_decode_need(&p, end, 2*sizeof(u32), bad);
1382 1383
		epoch = ceph_decode_32(&p);
		maplen = ceph_decode_32(&p);
S
Sage Weil 已提交
1384 1385 1386 1387 1388 1389 1390 1391 1392
		ceph_decode_need(&p, end, maplen, bad);
		if (nr_maps > 1) {
			dout("skipping non-latest full map %u len %d\n",
			     epoch, maplen);
		} else if (osdc->osdmap && osdc->osdmap->epoch >= epoch) {
			dout("skipping full map %u len %d, "
			     "older than our %u\n", epoch, maplen,
			     osdc->osdmap->epoch);
		} else {
1393 1394
			int skipped_map = 0;

S
Sage Weil 已提交
1395 1396 1397 1398 1399 1400
			dout("taking full map %u len %d\n", epoch, maplen);
			newmap = osdmap_decode(&p, p+maplen);
			if (IS_ERR(newmap)) {
				err = PTR_ERR(newmap);
				goto bad;
			}
1401
			BUG_ON(!newmap);
S
Sage Weil 已提交
1402 1403
			oldmap = osdc->osdmap;
			osdc->osdmap = newmap;
1404 1405 1406
			if (oldmap) {
				if (oldmap->epoch + 1 < newmap->epoch)
					skipped_map = 1;
S
Sage Weil 已提交
1407
				ceph_osdmap_destroy(oldmap);
1408 1409
			}
			kick_requests(osdc, skipped_map);
S
Sage Weil 已提交
1410 1411 1412 1413 1414 1415 1416 1417
		}
		p += maplen;
		nr_maps--;
	}

done:
	downgrade_write(&osdc->map_sem);
	ceph_monc_got_osdmap(&osdc->client->monc, osdc->osdmap->epoch);
1418 1419 1420 1421 1422 1423 1424 1425 1426

	/*
	 * subscribe to subsequent osdmap updates if full to ensure
	 * we find out when we are no longer full and stop returning
	 * ENOSPC.
	 */
	if (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL))
		ceph_monc_request_next_osdmap(&osdc->client->monc);

1427
	send_queued(osdc);
S
Sage Weil 已提交
1428
	up_read(&osdc->map_sem);
1429
	wake_up_all(&osdc->client->auth_wq);
S
Sage Weil 已提交
1430 1431 1432 1433
	return;

bad:
	pr_err("osdc handle_map corrupt msg\n");
1434
	ceph_msg_dump(msg);
S
Sage Weil 已提交
1435 1436 1437 1438
	up_write(&osdc->map_sem);
	return;
}

1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535
/*
 * watch/notify callback event infrastructure
 *
 * These callbacks are used both for watch and notify operations.
 */
static void __release_event(struct kref *kref)
{
	struct ceph_osd_event *event =
		container_of(kref, struct ceph_osd_event, kref);

	dout("__release_event %p\n", event);
	kfree(event);
}

static void get_event(struct ceph_osd_event *event)
{
	kref_get(&event->kref);
}

void ceph_osdc_put_event(struct ceph_osd_event *event)
{
	kref_put(&event->kref, __release_event);
}
EXPORT_SYMBOL(ceph_osdc_put_event);

static void __insert_event(struct ceph_osd_client *osdc,
			     struct ceph_osd_event *new)
{
	struct rb_node **p = &osdc->event_tree.rb_node;
	struct rb_node *parent = NULL;
	struct ceph_osd_event *event = NULL;

	while (*p) {
		parent = *p;
		event = rb_entry(parent, struct ceph_osd_event, node);
		if (new->cookie < event->cookie)
			p = &(*p)->rb_left;
		else if (new->cookie > event->cookie)
			p = &(*p)->rb_right;
		else
			BUG();
	}

	rb_link_node(&new->node, parent, p);
	rb_insert_color(&new->node, &osdc->event_tree);
}

static struct ceph_osd_event *__find_event(struct ceph_osd_client *osdc,
					        u64 cookie)
{
	struct rb_node **p = &osdc->event_tree.rb_node;
	struct rb_node *parent = NULL;
	struct ceph_osd_event *event = NULL;

	while (*p) {
		parent = *p;
		event = rb_entry(parent, struct ceph_osd_event, node);
		if (cookie < event->cookie)
			p = &(*p)->rb_left;
		else if (cookie > event->cookie)
			p = &(*p)->rb_right;
		else
			return event;
	}
	return NULL;
}

static void __remove_event(struct ceph_osd_event *event)
{
	struct ceph_osd_client *osdc = event->osdc;

	if (!RB_EMPTY_NODE(&event->node)) {
		dout("__remove_event removed %p\n", event);
		rb_erase(&event->node, &osdc->event_tree);
		ceph_osdc_put_event(event);
	} else {
		dout("__remove_event didn't remove %p\n", event);
	}
}

int ceph_osdc_create_event(struct ceph_osd_client *osdc,
			   void (*event_cb)(u64, u64, u8, void *),
			   int one_shot, void *data,
			   struct ceph_osd_event **pevent)
{
	struct ceph_osd_event *event;

	event = kmalloc(sizeof(*event), GFP_NOIO);
	if (!event)
		return -ENOMEM;

	dout("create_event %p\n", event);
	event->cb = event_cb;
	event->one_shot = one_shot;
	event->data = data;
	event->osdc = osdc;
	INIT_LIST_HEAD(&event->osd_node);
1536
	RB_CLEAR_NODE(&event->node);
1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618
	kref_init(&event->kref);   /* one ref for us */
	kref_get(&event->kref);    /* one ref for the caller */
	init_completion(&event->completion);

	spin_lock(&osdc->event_lock);
	event->cookie = ++osdc->event_count;
	__insert_event(osdc, event);
	spin_unlock(&osdc->event_lock);

	*pevent = event;
	return 0;
}
EXPORT_SYMBOL(ceph_osdc_create_event);

void ceph_osdc_cancel_event(struct ceph_osd_event *event)
{
	struct ceph_osd_client *osdc = event->osdc;

	dout("cancel_event %p\n", event);
	spin_lock(&osdc->event_lock);
	__remove_event(event);
	spin_unlock(&osdc->event_lock);
	ceph_osdc_put_event(event); /* caller's */
}
EXPORT_SYMBOL(ceph_osdc_cancel_event);


static void do_event_work(struct work_struct *work)
{
	struct ceph_osd_event_work *event_work =
		container_of(work, struct ceph_osd_event_work, work);
	struct ceph_osd_event *event = event_work->event;
	u64 ver = event_work->ver;
	u64 notify_id = event_work->notify_id;
	u8 opcode = event_work->opcode;

	dout("do_event_work completing %p\n", event);
	event->cb(ver, notify_id, opcode, event->data);
	complete(&event->completion);
	dout("do_event_work completed %p\n", event);
	ceph_osdc_put_event(event);
	kfree(event_work);
}


/*
 * Process osd watch notifications
 */
void handle_watch_notify(struct ceph_osd_client *osdc, struct ceph_msg *msg)
{
	void *p, *end;
	u8 proto_ver;
	u64 cookie, ver, notify_id;
	u8 opcode;
	struct ceph_osd_event *event;
	struct ceph_osd_event_work *event_work;

	p = msg->front.iov_base;
	end = p + msg->front.iov_len;

	ceph_decode_8_safe(&p, end, proto_ver, bad);
	ceph_decode_8_safe(&p, end, opcode, bad);
	ceph_decode_64_safe(&p, end, cookie, bad);
	ceph_decode_64_safe(&p, end, ver, bad);
	ceph_decode_64_safe(&p, end, notify_id, bad);

	spin_lock(&osdc->event_lock);
	event = __find_event(osdc, cookie);
	if (event) {
		get_event(event);
		if (event->one_shot)
			__remove_event(event);
	}
	spin_unlock(&osdc->event_lock);
	dout("handle_watch_notify cookie %lld ver %lld event %p\n",
	     cookie, ver, event);
	if (event) {
		event_work = kmalloc(sizeof(*event_work), GFP_NOIO);
		if (!event_work) {
			dout("ERROR: could not allocate event_work\n");
			goto done_err;
		}
1619
		INIT_WORK(&event_work->work, do_event_work);
1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656
		event_work->event = event;
		event_work->ver = ver;
		event_work->notify_id = notify_id;
		event_work->opcode = opcode;
		if (!queue_work(osdc->notify_wq, &event_work->work)) {
			dout("WARNING: failed to queue notify event work\n");
			goto done_err;
		}
	}

	return;

done_err:
	complete(&event->completion);
	ceph_osdc_put_event(event);
	return;

bad:
	pr_err("osdc handle_watch_notify corrupt msg\n");
	return;
}

int ceph_osdc_wait_event(struct ceph_osd_event *event, unsigned long timeout)
{
	int err;

	dout("wait_event %p\n", event);
	err = wait_for_completion_interruptible_timeout(&event->completion,
							timeout * HZ);
	ceph_osdc_put_event(event);
	if (err > 0)
		err = 0;
	dout("wait_event %p returns %d\n", event, err);
	return err;
}
EXPORT_SYMBOL(ceph_osdc_wait_event);

S
Sage Weil 已提交
1657 1658 1659 1660 1661 1662 1663
/*
 * Register request, send initial attempt.
 */
int ceph_osdc_start_request(struct ceph_osd_client *osdc,
			    struct ceph_osd_request *req,
			    bool nofail)
{
S
Sage Weil 已提交
1664
	int rc = 0;
S
Sage Weil 已提交
1665 1666 1667

	req->r_request->pages = req->r_pages;
	req->r_request->nr_pages = req->r_num_pages;
1668 1669 1670
#ifdef CONFIG_BLOCK
	req->r_request->bio = req->r_bio;
#endif
1671
	req->r_request->trail = &req->r_trail;
S
Sage Weil 已提交
1672 1673 1674 1675 1676

	register_request(osdc, req);

	down_read(&osdc->map_sem);
	mutex_lock(&osdc->request_mutex);
S
Sage Weil 已提交
1677 1678 1679 1680 1681 1682
	/*
	 * a racing kick_requests() may have sent the message for us
	 * while we dropped request_mutex above, so only send now if
	 * the request still han't been touched yet.
	 */
	if (req->r_sent == 0) {
1683
		rc = __map_request(osdc, req, 0);
1684 1685 1686 1687 1688 1689
		if (rc < 0) {
			if (nofail) {
				dout("osdc_start_request failed map, "
				     " will retry %lld\n", req->r_tid);
				rc = 0;
			}
1690
			goto out_unlock;
1691
		}
1692 1693 1694 1695
		if (req->r_osd == NULL) {
			dout("send_request %p no up osds in pg\n", req);
			ceph_monc_request_next_osdmap(&osdc->client->monc);
		} else {
1696
			__send_request(osdc, req);
S
Sage Weil 已提交
1697
		}
1698
		rc = 0;
S
Sage Weil 已提交
1699
	}
1700 1701

out_unlock:
S
Sage Weil 已提交
1702 1703 1704 1705
	mutex_unlock(&osdc->request_mutex);
	up_read(&osdc->map_sem);
	return rc;
}
1706
EXPORT_SYMBOL(ceph_osdc_start_request);
S
Sage Weil 已提交
1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719

/*
 * wait for a request to complete
 */
int ceph_osdc_wait_request(struct ceph_osd_client *osdc,
			   struct ceph_osd_request *req)
{
	int rc;

	rc = wait_for_completion_interruptible(&req->r_completion);
	if (rc < 0) {
		mutex_lock(&osdc->request_mutex);
		__cancel_request(req);
1720
		__unregister_request(osdc, req);
S
Sage Weil 已提交
1721
		mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1722
		complete_request(req);
1723
		dout("wait_request tid %llu canceled/timed out\n", req->r_tid);
S
Sage Weil 已提交
1724 1725 1726 1727 1728 1729
		return rc;
	}

	dout("wait_request tid %llu result %d\n", req->r_tid, req->r_result);
	return req->r_result;
}
1730
EXPORT_SYMBOL(ceph_osdc_wait_request);
S
Sage Weil 已提交
1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763

/*
 * sync - wait for all in-flight requests to flush.  avoid starvation.
 */
void ceph_osdc_sync(struct ceph_osd_client *osdc)
{
	struct ceph_osd_request *req;
	u64 last_tid, next_tid = 0;

	mutex_lock(&osdc->request_mutex);
	last_tid = osdc->last_tid;
	while (1) {
		req = __lookup_request_ge(osdc, next_tid);
		if (!req)
			break;
		if (req->r_tid > last_tid)
			break;

		next_tid = req->r_tid + 1;
		if ((req->r_flags & CEPH_OSD_FLAG_WRITE) == 0)
			continue;

		ceph_osdc_get_request(req);
		mutex_unlock(&osdc->request_mutex);
		dout("sync waiting on tid %llu (last is %llu)\n",
		     req->r_tid, last_tid);
		wait_for_completion(&req->r_safe_completion);
		mutex_lock(&osdc->request_mutex);
		ceph_osdc_put_request(req);
	}
	mutex_unlock(&osdc->request_mutex);
	dout("sync done (thru tid %llu)\n", last_tid);
}
1764
EXPORT_SYMBOL(ceph_osdc_sync);
S
Sage Weil 已提交
1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781

/*
 * init, shutdown
 */
int ceph_osdc_init(struct ceph_osd_client *osdc, struct ceph_client *client)
{
	int err;

	dout("init\n");
	osdc->client = client;
	osdc->osdmap = NULL;
	init_rwsem(&osdc->map_sem);
	init_completion(&osdc->map_waiters);
	osdc->last_requested_map = 0;
	mutex_init(&osdc->request_mutex);
	osdc->last_tid = 0;
	osdc->osds = RB_ROOT;
1782
	INIT_LIST_HEAD(&osdc->osd_lru);
S
Sage Weil 已提交
1783
	osdc->requests = RB_ROOT;
1784
	INIT_LIST_HEAD(&osdc->req_lru);
1785 1786
	INIT_LIST_HEAD(&osdc->req_unsent);
	INIT_LIST_HEAD(&osdc->req_notarget);
1787
	INIT_LIST_HEAD(&osdc->req_linger);
S
Sage Weil 已提交
1788 1789
	osdc->num_requests = 0;
	INIT_DELAYED_WORK(&osdc->timeout_work, handle_timeout);
1790
	INIT_DELAYED_WORK(&osdc->osds_timeout_work, handle_osds_timeout);
1791 1792 1793
	spin_lock_init(&osdc->event_lock);
	osdc->event_tree = RB_ROOT;
	osdc->event_count = 0;
1794 1795

	schedule_delayed_work(&osdc->osds_timeout_work,
1796
	   round_jiffies_relative(osdc->client->options->osd_idle_ttl * HZ));
S
Sage Weil 已提交
1797

1798
	err = -ENOMEM;
S
Sage Weil 已提交
1799 1800 1801
	osdc->req_mempool = mempool_create_kmalloc_pool(10,
					sizeof(struct ceph_osd_request));
	if (!osdc->req_mempool)
1802
		goto out;
S
Sage Weil 已提交
1803

1804 1805
	err = ceph_msgpool_init(&osdc->msgpool_op, CEPH_MSG_OSD_OP,
				OSD_OP_FRONT_LEN, 10, true,
1806
				"osd_op");
S
Sage Weil 已提交
1807
	if (err < 0)
1808
		goto out_mempool;
1809
	err = ceph_msgpool_init(&osdc->msgpool_op_reply, CEPH_MSG_OSD_OPREPLY,
1810 1811
				OSD_OPREPLY_FRONT_LEN, 10, true,
				"osd_op_reply");
S
Sage Weil 已提交
1812 1813
	if (err < 0)
		goto out_msgpool;
1814 1815 1816 1817 1818 1819 1820

	osdc->notify_wq = create_singlethread_workqueue("ceph-watch-notify");
	if (IS_ERR(osdc->notify_wq)) {
		err = PTR_ERR(osdc->notify_wq);
		osdc->notify_wq = NULL;
		goto out_msgpool;
	}
S
Sage Weil 已提交
1821
	return 0;
1822

S
Sage Weil 已提交
1823 1824
out_msgpool:
	ceph_msgpool_destroy(&osdc->msgpool_op);
1825 1826 1827 1828
out_mempool:
	mempool_destroy(osdc->req_mempool);
out:
	return err;
S
Sage Weil 已提交
1829
}
1830
EXPORT_SYMBOL(ceph_osdc_init);
S
Sage Weil 已提交
1831 1832 1833

void ceph_osdc_stop(struct ceph_osd_client *osdc)
{
1834 1835
	flush_workqueue(osdc->notify_wq);
	destroy_workqueue(osdc->notify_wq);
S
Sage Weil 已提交
1836
	cancel_delayed_work_sync(&osdc->timeout_work);
1837
	cancel_delayed_work_sync(&osdc->osds_timeout_work);
S
Sage Weil 已提交
1838 1839 1840 1841
	if (osdc->osdmap) {
		ceph_osdmap_destroy(osdc->osdmap);
		osdc->osdmap = NULL;
	}
1842
	remove_all_osds(osdc);
S
Sage Weil 已提交
1843 1844
	mempool_destroy(osdc->req_mempool);
	ceph_msgpool_destroy(&osdc->msgpool_op);
S
Sage Weil 已提交
1845
	ceph_msgpool_destroy(&osdc->msgpool_op_reply);
S
Sage Weil 已提交
1846
}
1847
EXPORT_SYMBOL(ceph_osdc_stop);
S
Sage Weil 已提交
1848 1849 1850 1851 1852 1853 1854 1855 1856

/*
 * Read some contiguous pages.  If we cross a stripe boundary, shorten
 * *plen.  Return number of bytes read, or error.
 */
int ceph_osdc_readpages(struct ceph_osd_client *osdc,
			struct ceph_vino vino, struct ceph_file_layout *layout,
			u64 off, u64 *plen,
			u32 truncate_seq, u64 truncate_size,
1857
			struct page **pages, int num_pages, int page_align)
S
Sage Weil 已提交
1858 1859 1860 1861 1862 1863 1864 1865 1866
{
	struct ceph_osd_request *req;
	int rc = 0;

	dout("readpages on ino %llx.%llx on %llu~%llu\n", vino.ino,
	     vino.snap, off, *plen);
	req = ceph_osdc_new_request(osdc, layout, vino, off, plen,
				    CEPH_OSD_OP_READ, CEPH_OSD_FLAG_READ,
				    NULL, 0, truncate_seq, truncate_size, NULL,
1867
				    false, 1, page_align);
1868 1869
	if (IS_ERR(req))
		return PTR_ERR(req);
S
Sage Weil 已提交
1870 1871 1872 1873

	/* it may be a short read due to an object boundary */
	req->r_pages = pages;

1874 1875
	dout("readpages  final extent is %llu~%llu (%d pages align %d)\n",
	     off, *plen, req->r_num_pages, page_align);
S
Sage Weil 已提交
1876 1877 1878 1879 1880 1881 1882 1883 1884

	rc = ceph_osdc_start_request(osdc, req, false);
	if (!rc)
		rc = ceph_osdc_wait_request(osdc, req);

	ceph_osdc_put_request(req);
	dout("readpages result %d\n", rc);
	return rc;
}
1885
EXPORT_SYMBOL(ceph_osdc_readpages);
S
Sage Weil 已提交
1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900

/*
 * do a synchronous write on N pages
 */
int ceph_osdc_writepages(struct ceph_osd_client *osdc, struct ceph_vino vino,
			 struct ceph_file_layout *layout,
			 struct ceph_snap_context *snapc,
			 u64 off, u64 len,
			 u32 truncate_seq, u64 truncate_size,
			 struct timespec *mtime,
			 struct page **pages, int num_pages,
			 int flags, int do_sync, bool nofail)
{
	struct ceph_osd_request *req;
	int rc = 0;
1901
	int page_align = off & ~PAGE_MASK;
S
Sage Weil 已提交
1902 1903 1904 1905 1906 1907 1908 1909

	BUG_ON(vino.snap != CEPH_NOSNAP);
	req = ceph_osdc_new_request(osdc, layout, vino, off, &len,
				    CEPH_OSD_OP_WRITE,
				    flags | CEPH_OSD_FLAG_ONDISK |
					    CEPH_OSD_FLAG_WRITE,
				    snapc, do_sync,
				    truncate_seq, truncate_size, mtime,
1910
				    nofail, 1, page_align);
1911 1912
	if (IS_ERR(req))
		return PTR_ERR(req);
S
Sage Weil 已提交
1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928

	/* it may be a short write due to an object boundary */
	req->r_pages = pages;
	dout("writepages %llu~%llu (%d pages)\n", off, len,
	     req->r_num_pages);

	rc = ceph_osdc_start_request(osdc, req, nofail);
	if (!rc)
		rc = ceph_osdc_wait_request(osdc, req);

	ceph_osdc_put_request(req);
	if (rc == 0)
		rc = len;
	dout("writepages result %d\n", rc);
	return rc;
}
1929
EXPORT_SYMBOL(ceph_osdc_writepages);
S
Sage Weil 已提交
1930 1931 1932 1933 1934 1935 1936

/*
 * handle incoming message
 */
static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
{
	struct ceph_osd *osd = con->private;
1937
	struct ceph_osd_client *osdc;
S
Sage Weil 已提交
1938 1939 1940
	int type = le16_to_cpu(msg->hdr.type);

	if (!osd)
S
Sage Weil 已提交
1941
		goto out;
1942
	osdc = osd->o_osdc;
S
Sage Weil 已提交
1943 1944 1945 1946 1947 1948

	switch (type) {
	case CEPH_MSG_OSD_MAP:
		ceph_osdc_handle_map(osdc, msg);
		break;
	case CEPH_MSG_OSD_OPREPLY:
1949
		handle_reply(osdc, msg, con);
S
Sage Weil 已提交
1950
		break;
1951 1952 1953
	case CEPH_MSG_WATCH_NOTIFY:
		handle_watch_notify(osdc, msg);
		break;
S
Sage Weil 已提交
1954 1955 1956 1957 1958

	default:
		pr_err("received unknown message type %d %s\n", type,
		       ceph_msg_type_name(type));
	}
S
Sage Weil 已提交
1959
out:
S
Sage Weil 已提交
1960 1961 1962
	ceph_msg_put(msg);
}

1963
/*
1964 1965
 * lookup and return message for incoming reply.  set up reply message
 * pages.
1966 1967
 */
static struct ceph_msg *get_reply(struct ceph_connection *con,
1968 1969
				  struct ceph_msg_header *hdr,
				  int *skip)
S
Sage Weil 已提交
1970 1971 1972
{
	struct ceph_osd *osd = con->private;
	struct ceph_osd_client *osdc = osd->o_osdc;
1973
	struct ceph_msg *m;
1974
	struct ceph_osd_request *req;
1975 1976
	int front = le32_to_cpu(hdr->front_len);
	int data_len = le32_to_cpu(hdr->data_len);
1977
	u64 tid;
S
Sage Weil 已提交
1978

1979 1980 1981 1982 1983 1984
	tid = le64_to_cpu(hdr->tid);
	mutex_lock(&osdc->request_mutex);
	req = __lookup_request(osdc, tid);
	if (!req) {
		*skip = 1;
		m = NULL;
1985 1986
		dout("get_reply unknown tid %llu from osd%d\n", tid,
		     osd->o_osd);
1987 1988
		goto out;
	}
S
Sage Weil 已提交
1989 1990

	if (req->r_con_filling_msg) {
1991
		dout("%s revoking msg %p from old con %p\n", __func__,
S
Sage Weil 已提交
1992
		     req->r_reply, req->r_con_filling_msg);
1993
		ceph_msg_revoke_incoming(req->r_reply);
1994
		req->r_con_filling_msg->ops->put(req->r_con_filling_msg);
1995
		req->r_con_filling_msg = NULL;
1996 1997
	}

S
Sage Weil 已提交
1998 1999 2000
	if (front > req->r_reply->front.iov_len) {
		pr_warning("get_reply front %d > preallocated %d\n",
			   front, (int)req->r_reply->front.iov_len);
2001
		m = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, front, GFP_NOFS, false);
2002
		if (!m)
S
Sage Weil 已提交
2003 2004 2005 2006 2007 2008
			goto out;
		ceph_msg_put(req->r_reply);
		req->r_reply = m;
	}
	m = ceph_msg_get(req->r_reply);

2009
	if (data_len > 0) {
2010
		int want = calc_pages_for(req->r_page_alignment, data_len);
2011 2012

		if (unlikely(req->r_num_pages < want)) {
2013 2014 2015
			pr_warning("tid %lld reply has %d bytes %d pages, we"
				   " had only %d pages ready\n", tid, data_len,
				   want, req->r_num_pages);
2016 2017
			*skip = 1;
			ceph_msg_put(m);
2018
			m = NULL;
2019
			goto out;
2020
		}
2021 2022
		m->pages = req->r_pages;
		m->nr_pages = req->r_num_pages;
2023
		m->page_alignment = req->r_page_alignment;
2024 2025 2026
#ifdef CONFIG_BLOCK
		m->bio = req->r_bio;
#endif
2027
	}
2028
	*skip = 0;
2029
	req->r_con_filling_msg = con->ops->get(con);
S
Sage Weil 已提交
2030
	dout("get_reply tid %lld %p\n", tid, m);
2031 2032 2033

out:
	mutex_unlock(&osdc->request_mutex);
2034
	return m;
2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045

}

static struct ceph_msg *alloc_msg(struct ceph_connection *con,
				  struct ceph_msg_header *hdr,
				  int *skip)
{
	struct ceph_osd *osd = con->private;
	int type = le16_to_cpu(hdr->type);
	int front = le32_to_cpu(hdr->front_len);

A
Alex Elder 已提交
2046
	*skip = 0;
2047 2048
	switch (type) {
	case CEPH_MSG_OSD_MAP:
2049
	case CEPH_MSG_WATCH_NOTIFY:
2050
		return ceph_msg_new(type, front, GFP_NOFS, false);
2051 2052 2053 2054 2055 2056 2057 2058
	case CEPH_MSG_OSD_OPREPLY:
		return get_reply(con, hdr, skip);
	default:
		pr_info("alloc_msg unexpected msg type %d from osd%d\n", type,
			osd->o_osd);
		*skip = 1;
		return NULL;
	}
S
Sage Weil 已提交
2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077
}

/*
 * Wrappers to refcount containing ceph_osd struct
 */
static struct ceph_connection *get_osd_con(struct ceph_connection *con)
{
	struct ceph_osd *osd = con->private;
	if (get_osd(osd))
		return con;
	return NULL;
}

static void put_osd_con(struct ceph_connection *con)
{
	struct ceph_osd *osd = con->private;
	put_osd(osd);
}

2078 2079 2080
/*
 * authentication
 */
2081 2082 2083 2084 2085
/*
 * Note: returned pointer is the address of a structure that's
 * managed separately.  Caller must *not* attempt to free it.
 */
static struct ceph_auth_handshake *get_authorizer(struct ceph_connection *con,
2086
					int *proto, int force_new)
2087 2088 2089 2090
{
	struct ceph_osd *o = con->private;
	struct ceph_osd_client *osdc = o->o_osdc;
	struct ceph_auth_client *ac = osdc->client->monc.auth;
2091
	struct ceph_auth_handshake *auth = &o->o_auth;
2092

2093
	if (force_new && auth->authorizer) {
2094 2095
		if (ac->ops && ac->ops->destroy_authorizer)
			ac->ops->destroy_authorizer(ac, auth->authorizer);
2096 2097
		auth->authorizer = NULL;
	}
2098
	if (!auth->authorizer && ac->ops && ac->ops->create_authorizer) {
2099 2100
		int ret = ac->ops->create_authorizer(ac, CEPH_ENTITY_TYPE_OSD,
							auth);
2101
		if (ret)
2102
			return ERR_PTR(ret);
2103 2104
	}
	*proto = ac->protocol;
2105

2106
	return auth;
2107 2108 2109 2110 2111 2112 2113 2114 2115
}


static int verify_authorizer_reply(struct ceph_connection *con, int len)
{
	struct ceph_osd *o = con->private;
	struct ceph_osd_client *osdc = o->o_osdc;
	struct ceph_auth_client *ac = osdc->client->monc.auth;

2116 2117 2118 2119
	/*
	 * XXX If ac->ops or ac->ops->verify_authorizer_reply is null,
	 * XXX which do we do:  succeed or fail?
	 */
2120
	return ac->ops->verify_authorizer_reply(ac, o->o_auth.authorizer, len);
2121 2122
}

2123 2124 2125 2126 2127 2128
static int invalidate_authorizer(struct ceph_connection *con)
{
	struct ceph_osd *o = con->private;
	struct ceph_osd_client *osdc = o->o_osdc;
	struct ceph_auth_client *ac = osdc->client->monc.auth;

2129
	if (ac->ops && ac->ops->invalidate_authorizer)
2130 2131 2132 2133
		ac->ops->invalidate_authorizer(ac, CEPH_ENTITY_TYPE_OSD);

	return ceph_monc_validate_auth(&osdc->client->monc);
}
2134

2135
static const struct ceph_connection_operations osd_con_ops = {
S
Sage Weil 已提交
2136 2137 2138
	.get = get_osd_con,
	.put = put_osd_con,
	.dispatch = dispatch,
2139 2140
	.get_authorizer = get_authorizer,
	.verify_authorizer_reply = verify_authorizer_reply,
2141
	.invalidate_authorizer = invalidate_authorizer,
S
Sage Weil 已提交
2142
	.alloc_msg = alloc_msg,
2143
	.fault = osd_reset,
S
Sage Weil 已提交
2144
};