osd_client.c 44.9 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;
25 26
static int __kick_requests(struct ceph_osd_client *osdc,
			  struct ceph_osd *kickosd);
S
Sage Weil 已提交
27 28 29

static void kick_requests(struct ceph_osd_client *osdc, struct ceph_osd *osd);

30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
static int op_needs_trail(int op)
{
	switch (op) {
	case CEPH_OSD_OP_GETXATTR:
	case CEPH_OSD_OP_SETXATTR:
	case CEPH_OSD_OP_CMPXATTR:
	case CEPH_OSD_OP_CALL:
		return 1;
	default:
		return 0;
	}
}

static int op_has_extent(int op)
{
	return (op == CEPH_OSD_OP_READ ||
		op == CEPH_OSD_OP_WRITE);
}

49 50 51
void ceph_calc_raw_layout(struct ceph_osd_client *osdc,
			struct ceph_file_layout *layout,
			u64 snapid,
52 53 54
			u64 off, u64 *plen, u64 *bno,
			struct ceph_osd_request *req,
			struct ceph_osd_req_op *op)
55 56
{
	struct ceph_osd_request_head *reqhead = req->r_request->front.iov_base;
57
	u64 orig_len = *plen;
58 59 60 61 62
	u64 objoff, objlen;    /* extent in object */

	reqhead->snapid = cpu_to_le64(snapid);

	/* object extent? */
63
	ceph_calc_file_object_mapping(layout, off, plen, bno,
64
				      &objoff, &objlen);
65
	if (*plen < orig_len)
66
		dout(" skipping last %llu, final file extent %llu~%llu\n",
67
		     orig_len - *plen, off, *plen);
68

69 70 71 72 73
	if (op_has_extent(op->op)) {
		op->extent.offset = objoff;
		op->extent.length = objlen;
	}
	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 80 81

	dout("calc_layout bno=%llx %llu~%llu (%d pages)\n",
	     *bno, objoff, objlen, req->r_num_pages);

}
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 99 100 101 102 103 104 105 106 107 108 109
/*
 * 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.
 */
static void calc_layout(struct ceph_osd_client *osdc,
110 111
			struct ceph_vino vino,
			struct ceph_file_layout *layout,
S
Sage Weil 已提交
112
			u64 off, u64 *plen,
113 114
			struct ceph_osd_request *req,
			struct ceph_osd_req_op *op)
S
Sage Weil 已提交
115 116 117
{
	u64 bno;

118 119
	ceph_calc_raw_layout(osdc, layout, vino.snap, off,
			     plen, &bno, req, op);
S
Sage Weil 已提交
120 121 122 123 124 125 126 127

	sprintf(req->r_oid, "%llx.%08llx", vino.ino, bno);
	req->r_oid_len = strlen(req->r_oid);
}

/*
 * requests
 */
S
Sage Weil 已提交
128
void ceph_osdc_release_request(struct kref *kref)
S
Sage Weil 已提交
129
{
S
Sage Weil 已提交
130 131 132 133 134 135 136 137
	struct ceph_osd_request *req = container_of(kref,
						    struct ceph_osd_request,
						    r_kref);

	if (req->r_request)
		ceph_msg_put(req->r_request);
	if (req->r_reply)
		ceph_msg_put(req->r_reply);
138
	if (req->r_con_filling_msg) {
139
		dout("release_request revoking pages %p from con %p\n",
140 141 142 143
		     req->r_pages, req->r_con_filling_msg);
		ceph_con_revoke_message(req->r_con_filling_msg,
				      req->r_reply);
		ceph_con_put(req->r_con_filling_msg);
144
	}
S
Sage Weil 已提交
145 146 147
	if (req->r_own_pages)
		ceph_release_page_vector(req->r_pages,
					 req->r_num_pages);
148 149 150 151
#ifdef CONFIG_BLOCK
	if (req->r_bio)
		bio_put(req->r_bio);
#endif
S
Sage Weil 已提交
152
	ceph_put_snap_context(req->r_snapc);
153 154 155 156
	if (req->r_trail) {
		ceph_pagelist_release(req->r_trail);
		kfree(req->r_trail);
	}
S
Sage Weil 已提交
157 158 159 160
	if (req->r_mempool)
		mempool_free(req, req->r_osdc->req_mempool);
	else
		kfree(req);
S
Sage Weil 已提交
161
}
162
EXPORT_SYMBOL(ceph_osdc_release_request);
163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178

static int get_num_ops(struct ceph_osd_req_op *ops, int *needs_trail)
{
	int i = 0;

	if (needs_trail)
		*needs_trail = 0;
	while (ops[i].op) {
		if (needs_trail && op_needs_trail(ops[i].op))
			*needs_trail = 1;
		i++;
	}

	return i;
}

179 180
struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc,
					       int flags,
S
Sage Weil 已提交
181
					       struct ceph_snap_context *snapc,
182
					       struct ceph_osd_req_op *ops,
183 184
					       bool use_mempool,
					       gfp_t gfp_flags,
185 186
					       struct page **pages,
					       struct bio *bio)
S
Sage Weil 已提交
187 188 189
{
	struct ceph_osd_request *req;
	struct ceph_msg *msg;
190 191 192
	int needs_trail;
	int num_op = get_num_ops(ops, &needs_trail);
	size_t msg_size = sizeof(struct ceph_osd_request_head);
193

194
	msg_size += num_op*sizeof(struct ceph_osd_op);
S
Sage Weil 已提交
195 196

	if (use_mempool) {
197
		req = mempool_alloc(osdc->req_mempool, gfp_flags);
S
Sage Weil 已提交
198 199
		memset(req, 0, sizeof(*req));
	} else {
200
		req = kzalloc(sizeof(*req), gfp_flags);
S
Sage Weil 已提交
201 202
	}
	if (req == NULL)
203
		return NULL;
S
Sage Weil 已提交
204 205 206

	req->r_osdc = osdc;
	req->r_mempool = use_mempool;
207

S
Sage Weil 已提交
208
	kref_init(&req->r_kref);
S
Sage Weil 已提交
209 210 211 212 213 214 215
	init_completion(&req->r_completion);
	init_completion(&req->r_safe_completion);
	INIT_LIST_HEAD(&req->r_unsafe_item);
	req->r_flags = flags;

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

S
Sage Weil 已提交
216 217 218 219 220
	/* create reply message */
	if (use_mempool)
		msg = ceph_msgpool_get(&osdc->msgpool_op_reply, 0);
	else
		msg = ceph_msg_new(CEPH_MSG_OSD_OPREPLY,
221
				   OSD_OPREPLY_FRONT_LEN, gfp_flags);
222
	if (!msg) {
S
Sage Weil 已提交
223
		ceph_osdc_put_request(req);
224
		return NULL;
S
Sage Weil 已提交
225 226 227
	}
	req->r_reply = msg;

228 229 230 231 232 233 234 235 236
	/* allocate space for the trailing data */
	if (needs_trail) {
		req->r_trail = kmalloc(sizeof(struct ceph_pagelist), gfp_flags);
		if (!req->r_trail) {
			ceph_osdc_put_request(req);
			return NULL;
		}
		ceph_pagelist_init(req->r_trail);
	}
S
Sage Weil 已提交
237
	/* create request message; allow space for oid */
S
Sage Weil 已提交
238 239 240 241
	msg_size += 40;
	if (snapc)
		msg_size += sizeof(u64) * snapc->num_snaps;
	if (use_mempool)
242
		msg = ceph_msgpool_get(&osdc->msgpool_op, 0);
S
Sage Weil 已提交
243
	else
244
		msg = ceph_msg_new(CEPH_MSG_OSD_OP, msg_size, gfp_flags);
245
	if (!msg) {
S
Sage Weil 已提交
246
		ceph_osdc_put_request(req);
247
		return NULL;
S
Sage Weil 已提交
248
	}
249

S
Sage Weil 已提交
250 251
	msg->hdr.type = cpu_to_le16(CEPH_MSG_OSD_OP);
	memset(msg->front.iov_base, 0, msg->front.iov_len);
252 253 254

	req->r_request = msg;
	req->r_pages = pages;
255 256 257 258 259 260
#ifdef CONFIG_BLOCK
	if (bio) {
		req->r_bio = bio;
		bio_get(req->r_bio);
	}
#endif
261 262 263

	return req;
}
264
EXPORT_SYMBOL(ceph_osdc_alloc_request);
265

266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298
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);

	switch (dst->op) {
	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:
		BUG_ON(!req->r_trail);

		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;
		ceph_pagelist_append(req->r_trail, src->xattr.name,
				     src->xattr.name_len);
		ceph_pagelist_append(req->r_trail, src->xattr.val,
				     src->xattr.value_len);
		break;
299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315
	case CEPH_OSD_OP_CALL:
		BUG_ON(!req->r_trail);

		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);

		ceph_pagelist_append(req->r_trail, src->cls.class_name,
				     src->cls.class_len);
		ceph_pagelist_append(req->r_trail, src->cls.method_name,
				     src->cls.method_len);
		ceph_pagelist_append(req->r_trail, src->cls.indata,
				     src->cls.indata_len);
		break;
	case CEPH_OSD_OP_ROLLBACK:
		dst->snap.snapid = cpu_to_le64(src->snap.snapid);
		break;
316 317 318 319 320 321 322 323 324 325
	case CEPH_OSD_OP_STARTSYNC:
		break;
	default:
		pr_err("unrecognized osd opcode %d\n", dst->op);
		WARN_ON(1);
		break;
	}
	dst->payload_len = cpu_to_le32(src->payload_len);
}

326 327 328 329 330
/*
 * build new request AND message
 *
 */
void ceph_osdc_build_request(struct ceph_osd_request *req,
331 332 333 334 335 336
			     u64 off, u64 *plen,
			     struct ceph_osd_req_op *src_ops,
			     struct ceph_snap_context *snapc,
			     struct timespec *mtime,
			     const char *oid,
			     int oid_len)
337 338 339
{
	struct ceph_msg *msg = req->r_request;
	struct ceph_osd_request_head *head;
340
	struct ceph_osd_req_op *src_op;
341 342
	struct ceph_osd_op *op;
	void *p;
343
	int num_op = get_num_ops(src_ops, NULL);
344 345
	size_t msg_size = sizeof(*head) + num_op*sizeof(*op);
	int flags = req->r_flags;
346 347
	u64 data_len = 0;
	int i;
348

S
Sage Weil 已提交
349 350 351 352 353 354 355 356 357 358 359 360 361 362
	head = msg->front.iov_base;
	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 */
363 364 365
	head->object_len = cpu_to_le32(oid_len);
	memcpy(p, oid, oid_len);
	p += oid_len;
S
Sage Weil 已提交
366

367 368 369 370
	src_op = src_ops;
	while (src_op->op) {
		osd_req_encode_op(req, op, src_op);
		src_op++;
S
Sage Weil 已提交
371 372
		op++;
	}
373 374 375 376

	if (req->r_trail)
		data_len += req->r_trail->length;

S
Sage Weil 已提交
377 378 379 380 381 382 383 384 385
	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);
		}
	}

386 387 388 389 390 391 392 393
	if (flags & CEPH_OSD_FLAG_WRITE) {
		req->r_request->hdr.data_off = cpu_to_le16(off);
		req->r_request->hdr.data_len = cpu_to_le32(*plen + data_len);
	} else if (data_len) {
		req->r_request->hdr.data_off = 0;
		req->r_request->hdr.data_len = cpu_to_le32(data_len);
	}

394 395
	req->r_request->page_alignment = req->r_page_alignment;

S
Sage Weil 已提交
396
	BUG_ON(p > msg->front.iov_base + msg->front.iov_len);
397 398 399
	msg_size = p - msg->front.iov_base;
	msg->front.iov_len = msg_size;
	msg->hdr.front_len = cpu_to_le32(msg_size);
400 401
	return;
}
402
EXPORT_SYMBOL(ceph_osdc_build_request);
403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424

/*
 * 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,
425 426
					       bool use_mempool, int num_reply,
					       int page_align)
427
{
428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
	struct ceph_osd_req_op ops[3];
	struct ceph_osd_request *req;

	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;

	req = ceph_osdc_alloc_request(osdc, flags,
					 snapc, ops,
445
					 use_mempool,
446
					 GFP_NOFS, NULL, NULL);
447 448 449 450
	if (IS_ERR(req))
		return req;

	/* calculate max write size */
451
	calc_layout(osdc, vino, layout, off, plen, req, ops);
452 453
	req->r_file_layout = *layout;  /* keep a copy */

454 455 456 457
	/* in case it differs from natural alignment that calc_layout
	   filled in for us */
	req->r_page_alignment = page_align;

458 459
	ceph_osdc_build_request(req, off, plen, ops,
				snapc,
460 461 462
				mtime,
				req->r_oid, req->r_oid_len);

S
Sage Weil 已提交
463 464
	return req;
}
465
EXPORT_SYMBOL(ceph_osdc_new_request);
S
Sage Weil 已提交
466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533

/*
 * 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;
}


/*
534
 * If the osd connection drops, we need to resubmit all requests.
S
Sage Weil 已提交
535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563
 */
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);
	kick_requests(osdc, osd);
	up_read(&osdc->map_sem);
}

/*
 * Track open sessions with osds.
 */
static struct ceph_osd *create_osd(struct ceph_osd_client *osdc)
{
	struct ceph_osd *osd;

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

	atomic_set(&osd->o_ref, 1);
	osd->o_osdc = osdc;
	INIT_LIST_HEAD(&osd->o_requests);
564
	INIT_LIST_HEAD(&osd->o_osd_lru);
S
Sage Weil 已提交
565 566 567 568 569 570
	osd->o_incarnation = 1;

	ceph_con_init(osdc->client->msgr, &osd->o_con);
	osd->o_con.private = osd;
	osd->o_con.ops = &osd_con_ops;
	osd->o_con.peer_name.type = CEPH_ENTITY_TYPE_OSD;
571

572
	INIT_LIST_HEAD(&osd->o_keepalive_item);
S
Sage Weil 已提交
573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591
	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);
S
Sage Weil 已提交
592 593 594 595 596
	if (atomic_dec_and_test(&osd->o_ref)) {
		struct ceph_auth_client *ac = osd->o_osdc->client->monc.auth;

		if (osd->o_authorizer)
			ac->ops->destroy_authorizer(ac, osd->o_authorizer);
S
Sage Weil 已提交
597
		kfree(osd);
S
Sage Weil 已提交
598
	}
S
Sage Weil 已提交
599 600 601 602 603
}

/*
 * remove an osd from our map
 */
604
static void __remove_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
S
Sage Weil 已提交
605
{
606
	dout("__remove_osd %p\n", osd);
S
Sage Weil 已提交
607 608
	BUG_ON(!list_empty(&osd->o_requests));
	rb_erase(&osd->o_node, &osdc->osds);
609
	list_del_init(&osd->o_osd_lru);
S
Sage Weil 已提交
610 611 612 613
	ceph_con_close(&osd->o_con);
	put_osd(osd);
}

614 615 616 617 618 619
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);
620
	osd->lru_ttl = jiffies + osdc->client->options->osd_idle_ttl * HZ;
621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643
}

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);
}

static void remove_old_osds(struct ceph_osd_client *osdc, int remove_all)
{
	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) {
		if (!remove_all && time_before(jiffies, osd->lru_ttl))
			break;
		__remove_osd(osdc, osd);
	}
	mutex_unlock(&osdc->request_mutex);
}

S
Sage Weil 已提交
644 645 646
/*
 * reset osd connect
 */
647
static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
S
Sage Weil 已提交
648
{
649
	struct ceph_osd_request *req;
S
Sage Weil 已提交
650 651
	int ret = 0;

652
	dout("__reset_osd %p osd%d\n", osd, osd->o_osd);
S
Sage Weil 已提交
653
	if (list_empty(&osd->o_requests)) {
654
		__remove_osd(osdc, osd);
655 656 657 658 659 660 661 662 663 664
	} else if (memcmp(&osdc->osdmap->osd_addr[osd->o_osd],
			  &osd->o_con.peer_addr,
			  sizeof(osd->o_con.peer_addr)) == 0 &&
		   !ceph_con_opened(&osd->o_con)) {
		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;
		ret = -EAGAIN;
S
Sage Weil 已提交
665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710
	} else {
		ceph_con_close(&osd->o_con);
		ceph_con_open(&osd->o_con, &osdc->osdmap->osd_addr[osd->o_osd]);
		osd->o_incarnation++;
	}
	return ret;
}

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;

	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;
}

711 712 713
static void __schedule_osd_timeout(struct ceph_osd_client *osdc)
{
	schedule_delayed_work(&osdc->timeout_work,
714
			osdc->client->options->osd_keepalive_timeout * HZ);
715 716 717 718 719 720
}

static void __cancel_osd_timeout(struct ceph_osd_client *osdc)
{
	cancel_delayed_work(&osdc->timeout_work);
}
S
Sage Weil 已提交
721 722 723 724 725 726 727 728 729 730

/*
 * Register request, assign tid.  If this is the first request, set up
 * the timeout event.
 */
static void register_request(struct ceph_osd_client *osdc,
			     struct ceph_osd_request *req)
{
	mutex_lock(&osdc->request_mutex);
	req->r_tid = ++osdc->last_tid;
731
	req->r_request->hdr.tid = cpu_to_le64(req->r_tid);
732
	INIT_LIST_HEAD(&req->r_req_lru_item);
S
Sage Weil 已提交
733 734 735 736 737 738 739

	dout("register_request %p tid %lld\n", req, req->r_tid);
	__insert_request(osdc, req);
	ceph_osdc_get_request(req);
	osdc->num_requests++;

	if (osdc->num_requests == 1) {
740 741
		dout(" first request, scheduling timeout\n");
		__schedule_osd_timeout(osdc);
S
Sage Weil 已提交
742 743 744 745 746 747 748 749 750 751 752 753 754 755
	}
	mutex_unlock(&osdc->request_mutex);
}

/*
 * called under osdc->request_mutex
 */
static void __unregister_request(struct ceph_osd_client *osdc,
				 struct ceph_osd_request *req)
{
	dout("__unregister_request %p tid %lld\n", req, req->r_tid);
	rb_erase(&req->r_node, &osdc->requests);
	osdc->num_requests--;

756 757 758 759 760 761
	if (req->r_osd) {
		/* make sure the original request isn't in flight. */
		ceph_con_revoke(&req->r_osd->o_con, req->r_request);

		list_del_init(&req->r_osd_item);
		if (list_empty(&req->r_osd->o_requests))
762
			__move_osd_to_lru(osdc, req->r_osd);
763 764
		req->r_osd = NULL;
	}
S
Sage Weil 已提交
765 766 767

	ceph_osdc_put_request(req);

768 769 770 771
	list_del_init(&req->r_req_lru_item);
	if (osdc->num_requests == 0) {
		dout(" no requests, canceling timeout\n");
		__cancel_osd_timeout(osdc);
S
Sage Weil 已提交
772 773 774 775 776 777 778 779
	}
}

/*
 * Cancel a previously queued request message
 */
static void __cancel_request(struct ceph_osd_request *req)
{
780
	if (req->r_sent && req->r_osd) {
S
Sage Weil 已提交
781 782 783
		ceph_con_revoke(&req->r_osd->o_con, req->r_request);
		req->r_sent = 0;
	}
784
	list_del_init(&req->r_req_lru_item);
S
Sage Weil 已提交
785 786 787 788 789 790 791 792 793 794 795 796 797 798 799
}

/*
 * 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
 * no up osd, set r_osd to NULL.
 *
 * Return 0 if unchanged, 1 if changed, or negative on error.
 *
 * Caller should hold map_sem for read and request_mutex.
 */
static int __map_osds(struct ceph_osd_client *osdc,
		      struct ceph_osd_request *req)
{
	struct ceph_osd_request_head *reqhead = req->r_request->front.iov_base;
800
	struct ceph_pg pgid;
801 802
	int acting[CEPH_PG_MAX_SIZE];
	int o = -1, num = 0;
S
Sage Weil 已提交
803 804 805 806 807 808 809
	int err;

	dout("map_osds %p tid %lld\n", req, req->r_tid);
	err = ceph_calc_object_layout(&reqhead->layout, req->r_oid,
				      &req->r_file_layout, osdc->osdmap);
	if (err)
		return err;
810
	pgid = reqhead->layout.ol_pgid;
811 812
	req->r_pgid = pgid;

813 814 815 816 817
	err = ceph_calc_pg_acting(osdc->osdmap, pgid, acting);
	if (err > 0) {
		o = acting[0];
		num = err;
	}
S
Sage Weil 已提交
818 819

	if ((req->r_osd && req->r_osd->o_osd == o &&
820 821 822
	     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
823 824 825
	    (req->r_osd == NULL && o == -1))
		return 0;  /* no change */

826 827
	dout("map_osds tid %llu pgid %d.%x osd%d (was osd%d)\n",
	     req->r_tid, le32_to_cpu(pgid.pool), le16_to_cpu(pgid.ps), o,
S
Sage Weil 已提交
828 829
	     req->r_osd ? req->r_osd->o_osd : -1);

830 831 832 833
	/* 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
834 835 836 837 838 839 840 841
	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) {
842 843 844 845
		err = -ENOMEM;
		req->r_osd = create_osd(osdc);
		if (!req->r_osd)
			goto out;
S
Sage Weil 已提交
846 847 848 849 850 851 852 853 854

		dout("map_osds osd %p is osd%d\n", req->r_osd, o);
		req->r_osd->o_osd = o;
		req->r_osd->o_con.peer_name.num = cpu_to_le64(o);
		__insert_osd(osdc, req->r_osd);

		ceph_con_open(&req->r_osd->o_con, &osdc->osdmap->osd_addr[o]);
	}

855 856
	if (req->r_osd) {
		__remove_osd_from_lru(req->r_osd);
S
Sage Weil 已提交
857
		list_add(&req->r_osd_item, &req->r_osd->o_requests);
858
	}
859
	err = 1;   /* osd or pg changed */
S
Sage Weil 已提交
860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890

out:
	return err;
}

/*
 * caller should hold map_sem (for read) and request_mutex
 */
static int __send_request(struct ceph_osd_client *osdc,
			  struct ceph_osd_request *req)
{
	struct ceph_osd_request_head *reqhead;
	int err;

	err = __map_osds(osdc, req);
	if (err < 0)
		return err;
	if (req->r_osd == NULL) {
		dout("send_request %p no up osds in pg\n", req);
		ceph_monc_request_next_osdmap(&osdc->client->monc);
		return 0;
	}

	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
891
	req->r_stamp = jiffies;
892
	list_move_tail(&req->r_req_lru_item, &osdc->req_lru);
S
Sage Weil 已提交
893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912

	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;
	return 0;
}

/*
 * 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);
913
	struct ceph_osd_request *req, *last_req = NULL;
S
Sage Weil 已提交
914
	struct ceph_osd *osd;
915
	unsigned long timeout = osdc->client->options->osd_timeout * HZ;
916
	unsigned long keepalive =
917
		osdc->client->options->osd_keepalive_timeout * HZ;
S
Sage Weil 已提交
918
	unsigned long last_stamp = 0;
S
Sage Weil 已提交
919
	struct rb_node *p;
920
	struct list_head slow_osds;
S
Sage Weil 已提交
921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943

	dout("timeout\n");
	down_read(&osdc->map_sem);

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

	mutex_lock(&osdc->request_mutex);
	for (p = rb_first(&osdc->requests); p; p = rb_next(p)) {
		req = rb_entry(p, struct ceph_osd_request, r_node);

		if (req->r_resend) {
			int err;

			dout("osdc resending prev failed %lld\n", req->r_tid);
			err = __send_request(osdc, req);
			if (err)
				dout("osdc failed again on %lld\n", req->r_tid);
			else
				req->r_resend = false;
			continue;
		}
	}

944 945 946 947 948 949 950
	/*
	 * reset osds that appear to be _really_ unresponsive.  this
	 * is a failsafe measure.. we really shouldn't be getting to
	 * this point if the system is working properly.  the monitors
	 * should mark the osd as failed and we should find out about
	 * it from an updated osd map.
	 */
S
Sage Weil 已提交
951
	while (timeout && !list_empty(&osdc->req_lru)) {
952 953 954
		req = list_entry(osdc->req_lru.next, struct ceph_osd_request,
				 r_req_lru_item);

S
Sage Weil 已提交
955
		if (time_before(jiffies, req->r_stamp + timeout))
956 957
			break;

S
Sage Weil 已提交
958
		BUG_ON(req == last_req && req->r_stamp == last_stamp);
959
		last_req = req;
S
Sage Weil 已提交
960
		last_stamp = req->r_stamp;
961 962 963 964 965 966 967 968 969 970 971 972 973 974 975

		osd = req->r_osd;
		BUG_ON(!osd);
		pr_warning(" tid %llu timed out on osd%d, will reset osd\n",
			   req->r_tid, osd->o_osd);
		__kick_requests(osdc, osd);
	}

	/*
	 * 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
976
		if (time_before(jiffies, req->r_stamp + keepalive))
977 978 979 980 981
			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
982
		     req->r_tid, osd->o_osd);
983 984 985 986 987 988
		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
989 990 991
		ceph_con_keepalive(&osd->o_con);
	}

992
	__schedule_osd_timeout(osdc);
S
Sage Weil 已提交
993 994 995 996 997
	mutex_unlock(&osdc->request_mutex);

	up_read(&osdc->map_sem);
}

998 999 1000 1001 1002 1003
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 =
1004
		osdc->client->options->osd_idle_ttl * HZ >> 2;
1005 1006 1007 1008 1009 1010 1011 1012 1013 1014

	dout("osds timeout\n");
	down_read(&osdc->map_sem);
	remove_old_osds(osdc, 0);
	up_read(&osdc->map_sem);

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

S
Sage Weil 已提交
1015 1016 1017 1018
/*
 * handle osd op reply.  either call the callback if it is specified,
 * or do the completion to wake up the waiting thread.
 */
1019 1020
static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
			 struct ceph_connection *con)
S
Sage Weil 已提交
1021 1022 1023 1024 1025
{
	struct ceph_osd_reply_head *rhead = msg->front.iov_base;
	struct ceph_osd_request *req;
	u64 tid;
	int numops, object_len, flags;
1026
	s32 result;
S
Sage Weil 已提交
1027

1028
	tid = le64_to_cpu(msg->hdr.tid);
S
Sage Weil 已提交
1029 1030 1031 1032
	if (msg->front.iov_len < sizeof(*rhead))
		goto bad;
	numops = le32_to_cpu(rhead->num_ops);
	object_len = le32_to_cpu(rhead->object_len);
1033
	result = le32_to_cpu(rhead->result);
S
Sage Weil 已提交
1034 1035 1036
	if (msg->front.iov_len != sizeof(*rhead) + object_len +
	    numops * sizeof(struct ceph_osd_op))
		goto bad;
1037
	dout("handle_reply %p tid %llu result %d\n", msg, tid, (int)result);
S
Sage Weil 已提交
1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049

	/* 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);

1050
	/*
1051
	 * if this connection filled our message, drop our reference now, to
1052 1053
	 * avoid a (safe but slower) revoke later.
	 */
1054
	if (req->r_con_filling_msg == con && req->r_reply == msg) {
S
Sage Weil 已提交
1055
		dout(" dropping con_filling_msg ref %p\n", con);
1056
		req->r_con_filling_msg = NULL;
1057 1058 1059
		ceph_con_put(con);
	}

S
Sage Weil 已提交
1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075
	if (!req->r_got_reply) {
		unsigned bytes;

		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
1076
		mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1077 1078 1079 1080 1081 1082
		goto done;
	}

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

	/* either this is a read, or we got the safe response */
1083 1084
	if (result < 0 ||
	    (flags & CEPH_OSD_FLAG_ONDISK) ||
S
Sage Weil 已提交
1085 1086 1087 1088 1089 1090 1091 1092
	    ((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
1093
		complete_all(&req->r_completion);
S
Sage Weil 已提交
1094 1095 1096 1097

	if (flags & CEPH_OSD_FLAG_ONDISK) {
		if (req->r_safe_callback)
			req->r_safe_callback(req, msg);
1098
		complete_all(&req->r_safe_completion);  /* fsync waiter */
S
Sage Weil 已提交
1099 1100 1101 1102 1103 1104 1105 1106 1107 1108
	}

done:
	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));
1109
	ceph_msg_dump(msg);
S
Sage Weil 已提交
1110 1111 1112
}


1113
static int __kick_requests(struct ceph_osd_client *osdc,
S
Sage Weil 已提交
1114 1115 1116 1117 1118 1119 1120 1121
			  struct ceph_osd *kickosd)
{
	struct ceph_osd_request *req;
	struct rb_node *p, *n;
	int needmap = 0;
	int err;

	dout("kick_requests osd%d\n", kickosd ? kickosd->o_osd : -1);
1122
	if (kickosd) {
1123 1124 1125
		err = __reset_osd(osdc, kickosd);
		if (err == -EAGAIN)
			return 1;
1126
	} else {
S
Sage Weil 已提交
1127 1128 1129 1130 1131 1132
		for (p = rb_first(&osdc->osds); p; p = n) {
			struct ceph_osd *osd =
				rb_entry(p, struct ceph_osd, o_node);

			n = rb_next(p);
			if (!ceph_osd_is_up(osdc->osdmap, osd->o_osd) ||
S
Sage Weil 已提交
1133 1134 1135 1136
			    memcmp(&osd->o_con.peer_addr,
				   ceph_osd_addr(osdc->osdmap,
						 osd->o_osd),
				   sizeof(struct ceph_entity_addr)) != 0)
1137
				__reset_osd(osdc, osd);
S
Sage Weil 已提交
1138 1139 1140 1141 1142 1143 1144 1145
		}
	}

	for (p = rb_first(&osdc->requests); p; p = rb_next(p)) {
		req = rb_entry(p, struct ceph_osd_request, r_node);

		if (req->r_resend) {
			dout(" r_resend set on tid %llu\n", req->r_tid);
1146
			__cancel_request(req);
S
Sage Weil 已提交
1147 1148
			goto kick;
		}
1149 1150
		if (req->r_osd && kickosd == req->r_osd) {
			__cancel_request(req);
S
Sage Weil 已提交
1151
			goto kick;
1152
		}
S
Sage Weil 已提交
1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174

		err = __map_osds(osdc, req);
		if (err == 0)
			continue;  /* no change */
		if (err < 0) {
			/*
			 * FIXME: really, we should set the request
			 * error and fail if this isn't a 'nofail'
			 * request, but that's a fair bit more
			 * complicated to do.  So retry!
			 */
			dout(" setting r_resend on %llu\n", req->r_tid);
			req->r_resend = true;
			continue;
		}
		if (req->r_osd == NULL) {
			dout("tid %llu maps to no valid osd\n", req->r_tid);
			needmap++;  /* request a newer map */
			continue;
		}

kick:
S
Sage Weil 已提交
1175
		dout("kicking %p tid %llu osd%d\n", req, req->r_tid,
1176
		     req->r_osd ? req->r_osd->o_osd : -1);
S
Sage Weil 已提交
1177 1178 1179 1180 1181 1182 1183
		req->r_flags |= CEPH_OSD_FLAG_RETRY;
		err = __send_request(osdc, req);
		if (err) {
			dout(" setting r_resend on %llu\n", req->r_tid);
			req->r_resend = true;
		}
	}
1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205

	return needmap;
}

/*
 * Resubmit osd requests whose osd or osd address has changed.  Request
 * a new osd map if osds are down, or we are otherwise unable to determine
 * how to direct a request.
 *
 * Close connections to down osds.
 *
 * If @who is specified, resubmit requests for that specific osd.
 *
 * Caller should hold map_sem for read and request_mutex.
 */
static void kick_requests(struct ceph_osd_client *osdc,
			  struct ceph_osd *kickosd)
{
	int needmap;

	mutex_lock(&osdc->request_mutex);
	needmap = __kick_requests(osdc, kickosd);
S
Sage Weil 已提交
1206 1207 1208 1209 1210 1211 1212
	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);
	}

1213
}
S
Sage Weil 已提交
1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236
/*
 * 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));
1237 1238
	if (ceph_check_fsid(osdc->client, &fsid) < 0)
		return;
S
Sage Weil 已提交
1239 1240 1241 1242 1243 1244 1245 1246

	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);
1247 1248
		epoch = ceph_decode_32(&p);
		maplen = ceph_decode_32(&p);
S
Sage Weil 已提交
1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260
		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,
							  osdc->client->msgr);
			if (IS_ERR(newmap)) {
				err = PTR_ERR(newmap);
				goto bad;
			}
1261
			BUG_ON(!newmap);
S
Sage Weil 已提交
1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
			if (newmap != osdc->osdmap) {
				ceph_osdmap_destroy(osdc->osdmap);
				osdc->osdmap = newmap;
			}
		} 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);
1281 1282
		epoch = ceph_decode_32(&p);
		maplen = ceph_decode_32(&p);
S
Sage Weil 已提交
1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297
		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 {
			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;
			}
1298
			BUG_ON(!newmap);
S
Sage Weil 已提交
1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313
			oldmap = osdc->osdmap;
			osdc->osdmap = newmap;
			if (oldmap)
				ceph_osdmap_destroy(oldmap);
		}
		p += maplen;
		nr_maps--;
	}

done:
	downgrade_write(&osdc->map_sem);
	ceph_monc_got_osdmap(&osdc->client->monc, osdc->osdmap->epoch);
	if (newmap)
		kick_requests(osdc, NULL);
	up_read(&osdc->map_sem);
1314
	wake_up_all(&osdc->client->auth_wq);
S
Sage Weil 已提交
1315 1316 1317 1318
	return;

bad:
	pr_err("osdc handle_map corrupt msg\n");
1319
	ceph_msg_dump(msg);
S
Sage Weil 已提交
1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330
	up_write(&osdc->map_sem);
	return;
}

/*
 * 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
1331
	int rc = 0;
S
Sage Weil 已提交
1332 1333 1334

	req->r_request->pages = req->r_pages;
	req->r_request->nr_pages = req->r_num_pages;
1335 1336 1337 1338
#ifdef CONFIG_BLOCK
	req->r_request->bio = req->r_bio;
#endif
	req->r_request->trail = req->r_trail;
S
Sage Weil 已提交
1339 1340 1341 1342 1343

	register_request(osdc, req);

	down_read(&osdc->map_sem);
	mutex_lock(&osdc->request_mutex);
S
Sage Weil 已提交
1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359
	/*
	 * 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) {
		rc = __send_request(osdc, req);
		if (rc) {
			if (nofail) {
				dout("osdc_start_request failed send, "
				     " marking %lld\n", req->r_tid);
				req->r_resend = true;
				rc = 0;
			} else {
				__unregister_request(osdc, req);
			}
S
Sage Weil 已提交
1360 1361 1362 1363 1364 1365
		}
	}
	mutex_unlock(&osdc->request_mutex);
	up_read(&osdc->map_sem);
	return rc;
}
1366
EXPORT_SYMBOL(ceph_osdc_start_request);
S
Sage Weil 已提交
1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379

/*
 * 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);
1380
		__unregister_request(osdc, req);
S
Sage Weil 已提交
1381
		mutex_unlock(&osdc->request_mutex);
1382
		dout("wait_request tid %llu canceled/timed out\n", req->r_tid);
S
Sage Weil 已提交
1383 1384 1385 1386 1387 1388
		return rc;
	}

	dout("wait_request tid %llu result %d\n", req->r_tid, req->r_result);
	return req->r_result;
}
1389
EXPORT_SYMBOL(ceph_osdc_wait_request);
S
Sage Weil 已提交
1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422

/*
 * 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);
}
1423
EXPORT_SYMBOL(ceph_osdc_sync);
S
Sage Weil 已提交
1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440

/*
 * 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;
1441
	INIT_LIST_HEAD(&osdc->osd_lru);
S
Sage Weil 已提交
1442
	osdc->requests = RB_ROOT;
1443
	INIT_LIST_HEAD(&osdc->req_lru);
S
Sage Weil 已提交
1444 1445
	osdc->num_requests = 0;
	INIT_DELAYED_WORK(&osdc->timeout_work, handle_timeout);
1446 1447 1448
	INIT_DELAYED_WORK(&osdc->osds_timeout_work, handle_osds_timeout);

	schedule_delayed_work(&osdc->osds_timeout_work,
1449
	   round_jiffies_relative(osdc->client->options->osd_idle_ttl * HZ));
S
Sage Weil 已提交
1450

1451
	err = -ENOMEM;
S
Sage Weil 已提交
1452 1453 1454
	osdc->req_mempool = mempool_create_kmalloc_pool(10,
					sizeof(struct ceph_osd_request));
	if (!osdc->req_mempool)
1455
		goto out;
S
Sage Weil 已提交
1456

1457 1458
	err = ceph_msgpool_init(&osdc->msgpool_op, OSD_OP_FRONT_LEN, 10, true,
				"osd_op");
S
Sage Weil 已提交
1459
	if (err < 0)
1460
		goto out_mempool;
S
Sage Weil 已提交
1461
	err = ceph_msgpool_init(&osdc->msgpool_op_reply,
1462 1463
				OSD_OPREPLY_FRONT_LEN, 10, true,
				"osd_op_reply");
S
Sage Weil 已提交
1464 1465
	if (err < 0)
		goto out_msgpool;
S
Sage Weil 已提交
1466
	return 0;
1467

S
Sage Weil 已提交
1468 1469
out_msgpool:
	ceph_msgpool_destroy(&osdc->msgpool_op);
1470 1471 1472 1473
out_mempool:
	mempool_destroy(osdc->req_mempool);
out:
	return err;
S
Sage Weil 已提交
1474
}
1475
EXPORT_SYMBOL(ceph_osdc_init);
S
Sage Weil 已提交
1476 1477 1478 1479

void ceph_osdc_stop(struct ceph_osd_client *osdc)
{
	cancel_delayed_work_sync(&osdc->timeout_work);
1480
	cancel_delayed_work_sync(&osdc->osds_timeout_work);
S
Sage Weil 已提交
1481 1482 1483 1484
	if (osdc->osdmap) {
		ceph_osdmap_destroy(osdc->osdmap);
		osdc->osdmap = NULL;
	}
1485
	remove_old_osds(osdc, 1);
S
Sage Weil 已提交
1486 1487
	mempool_destroy(osdc->req_mempool);
	ceph_msgpool_destroy(&osdc->msgpool_op);
S
Sage Weil 已提交
1488
	ceph_msgpool_destroy(&osdc->msgpool_op_reply);
S
Sage Weil 已提交
1489
}
1490
EXPORT_SYMBOL(ceph_osdc_stop);
S
Sage Weil 已提交
1491 1492 1493 1494 1495 1496 1497 1498 1499

/*
 * 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,
1500
			struct page **pages, int num_pages, int page_align)
S
Sage Weil 已提交
1501 1502 1503 1504 1505 1506 1507 1508 1509
{
	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,
1510
				    false, 1, page_align);
1511 1512
	if (!req)
		return -ENOMEM;
S
Sage Weil 已提交
1513 1514 1515 1516

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

1517 1518
	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
1519 1520 1521 1522 1523 1524 1525 1526 1527

	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;
}
1528
EXPORT_SYMBOL(ceph_osdc_readpages);
S
Sage Weil 已提交
1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543

/*
 * 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;
1544
	int page_align = off & ~PAGE_MASK;
S
Sage Weil 已提交
1545 1546 1547 1548 1549 1550 1551 1552

	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,
1553
				    nofail, 1, page_align);
1554 1555
	if (!req)
		return -ENOMEM;
S
Sage Weil 已提交
1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571

	/* 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;
}
1572
EXPORT_SYMBOL(ceph_osdc_writepages);
S
Sage Weil 已提交
1573 1574 1575 1576 1577 1578 1579

/*
 * handle incoming message
 */
static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
{
	struct ceph_osd *osd = con->private;
1580
	struct ceph_osd_client *osdc;
S
Sage Weil 已提交
1581 1582 1583
	int type = le16_to_cpu(msg->hdr.type);

	if (!osd)
S
Sage Weil 已提交
1584
		goto out;
1585
	osdc = osd->o_osdc;
S
Sage Weil 已提交
1586 1587 1588 1589 1590 1591

	switch (type) {
	case CEPH_MSG_OSD_MAP:
		ceph_osdc_handle_map(osdc, msg);
		break;
	case CEPH_MSG_OSD_OPREPLY:
1592
		handle_reply(osdc, msg, con);
S
Sage Weil 已提交
1593 1594 1595 1596 1597 1598
		break;

	default:
		pr_err("received unknown message type %d %s\n", type,
		       ceph_msg_type_name(type));
	}
S
Sage Weil 已提交
1599
out:
S
Sage Weil 已提交
1600 1601 1602
	ceph_msg_put(msg);
}

1603
/*
1604 1605
 * lookup and return message for incoming reply.  set up reply message
 * pages.
1606 1607
 */
static struct ceph_msg *get_reply(struct ceph_connection *con,
1608 1609
				  struct ceph_msg_header *hdr,
				  int *skip)
S
Sage Weil 已提交
1610 1611 1612
{
	struct ceph_osd *osd = con->private;
	struct ceph_osd_client *osdc = osd->o_osdc;
1613
	struct ceph_msg *m;
1614
	struct ceph_osd_request *req;
1615 1616
	int front = le32_to_cpu(hdr->front_len);
	int data_len = le32_to_cpu(hdr->data_len);
1617
	u64 tid;
S
Sage Weil 已提交
1618

1619 1620 1621 1622 1623 1624
	tid = le64_to_cpu(hdr->tid);
	mutex_lock(&osdc->request_mutex);
	req = __lookup_request(osdc, tid);
	if (!req) {
		*skip = 1;
		m = NULL;
S
Sage Weil 已提交
1625
		pr_info("get_reply unknown tid %llu from osd%d\n", tid,
1626
			osd->o_osd);
1627 1628
		goto out;
	}
S
Sage Weil 已提交
1629 1630 1631 1632 1633 1634

	if (req->r_con_filling_msg) {
		dout("get_reply revoking msg %p from old con %p\n",
		     req->r_reply, req->r_con_filling_msg);
		ceph_con_revoke_message(req->r_con_filling_msg, req->r_reply);
		ceph_con_put(req->r_con_filling_msg);
1635
		req->r_con_filling_msg = NULL;
1636 1637
	}

S
Sage Weil 已提交
1638 1639 1640
	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);
1641
		m = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, front, GFP_NOFS);
1642
		if (!m)
S
Sage Weil 已提交
1643 1644 1645 1646 1647 1648
			goto out;
		ceph_msg_put(req->r_reply);
		req->r_reply = m;
	}
	m = ceph_msg_get(req->r_reply);

1649
	if (data_len > 0) {
1650
		int want = calc_pages_for(req->r_page_alignment, data_len);
1651 1652 1653 1654

		if (unlikely(req->r_num_pages < want)) {
			pr_warning("tid %lld reply %d > expected %d pages\n",
				   tid, want, m->nr_pages);
1655 1656
			*skip = 1;
			ceph_msg_put(m);
1657
			m = NULL;
1658
			goto out;
1659
		}
1660 1661
		m->pages = req->r_pages;
		m->nr_pages = req->r_num_pages;
1662
		m->page_alignment = req->r_page_alignment;
1663 1664 1665
#ifdef CONFIG_BLOCK
		m->bio = req->r_bio;
#endif
1666
	}
1667
	*skip = 0;
S
Sage Weil 已提交
1668 1669
	req->r_con_filling_msg = ceph_con_get(con);
	dout("get_reply tid %lld %p\n", tid, m);
1670 1671 1672

out:
	mutex_unlock(&osdc->request_mutex);
1673
	return m;
1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686

}

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);

	switch (type) {
	case CEPH_MSG_OSD_MAP:
1687
		return ceph_msg_new(type, front, GFP_NOFS);
1688 1689 1690 1691 1692 1693 1694 1695
	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
1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714
}

/*
 * 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);
}

1715 1716 1717 1718
/*
 * authentication
 */
static int get_authorizer(struct ceph_connection *con,
S
Sage Weil 已提交
1719 1720
			  void **buf, int *len, int *proto,
			  void **reply_buf, int *reply_len, int force_new)
1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739
{
	struct ceph_osd *o = con->private;
	struct ceph_osd_client *osdc = o->o_osdc;
	struct ceph_auth_client *ac = osdc->client->monc.auth;
	int ret = 0;

	if (force_new && o->o_authorizer) {
		ac->ops->destroy_authorizer(ac, o->o_authorizer);
		o->o_authorizer = NULL;
	}
	if (o->o_authorizer == NULL) {
		ret = ac->ops->create_authorizer(
			ac, CEPH_ENTITY_TYPE_OSD,
			&o->o_authorizer,
			&o->o_authorizer_buf,
			&o->o_authorizer_buf_len,
			&o->o_authorizer_reply_buf,
			&o->o_authorizer_reply_buf_len);
		if (ret)
S
Sage Weil 已提交
1740
			return ret;
1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760
	}

	*proto = ac->protocol;
	*buf = o->o_authorizer_buf;
	*len = o->o_authorizer_buf_len;
	*reply_buf = o->o_authorizer_reply_buf;
	*reply_len = o->o_authorizer_reply_buf_len;
	return 0;
}


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;

	return ac->ops->verify_authorizer_reply(ac, o->o_authorizer, len);
}

1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771
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;

	if (ac->ops->invalidate_authorizer)
		ac->ops->invalidate_authorizer(ac, CEPH_ENTITY_TYPE_OSD);

	return ceph_monc_validate_auth(&osdc->client->monc);
}
1772

1773
static const struct ceph_connection_operations osd_con_ops = {
S
Sage Weil 已提交
1774 1775 1776
	.get = get_osd_con,
	.put = put_osd_con,
	.dispatch = dispatch,
1777 1778
	.get_authorizer = get_authorizer,
	.verify_authorizer_reply = verify_authorizer_reply,
1779
	.invalidate_authorizer = invalidate_authorizer,
S
Sage Weil 已提交
1780
	.alloc_msg = alloc_msg,
1781
	.fault = osd_reset,
S
Sage Weil 已提交
1782
};