osd_client.c 58.0 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

A
Alex Elder 已提交
26
static void __send_queued(struct ceph_osd_client *osdc);
27
static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd);
28 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);
}

S
Sage Weil 已提交
41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
/*
 * 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.
 */
66
static int calc_layout(struct ceph_file_layout *layout, u64 off, u64 *plen,
67
		       struct ceph_osd_req_op *op, u64 *bno)
S
Sage Weil 已提交
68
{
69 70 71
	u64 orig_len = *plen;
	u64 objoff = 0;
	u64 objlen = 0;
S
Sage Weil 已提交
72
	int r;
S
Sage Weil 已提交
73

74
	/* object extent? */
75
	r = ceph_calc_file_object_mapping(layout, off, orig_len, bno,
76
					  &objoff, &objlen);
S
Sage Weil 已提交
77 78
	if (r < 0)
		return r;
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99
	if (objlen < orig_len) {
		*plen = objlen;
		dout(" skipping last %llu, final file extent %llu~%llu\n",
		     orig_len - *plen, off, *plen);
	}

	if (op_has_extent(op->op)) {
		u32 osize = le32_to_cpu(layout->fl_object_size);
		op->extent.offset = objoff;
		op->extent.length = objlen;
		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;
		}
	}
	if (op->op == CEPH_OSD_OP_WRITE)
		op->payload_len = *plen;

100
	dout("calc_layout bno=%llx %llu~%llu\n", *bno, objoff, objlen);
S
Sage Weil 已提交
101

102
	return 0;
S
Sage Weil 已提交
103 104 105 106 107
}

/*
 * requests
 */
S
Sage Weil 已提交
108
void ceph_osdc_release_request(struct kref *kref)
S
Sage Weil 已提交
109
{
110
	int num_pages;
S
Sage Weil 已提交
111 112 113 114 115 116
	struct ceph_osd_request *req = container_of(kref,
						    struct ceph_osd_request,
						    r_kref);

	if (req->r_request)
		ceph_msg_put(req->r_request);
117
	if (req->r_con_filling_msg) {
118 119
		dout("%s revoking msg %p from con %p\n", __func__,
		     req->r_reply, req->r_con_filling_msg);
120
		ceph_msg_revoke_incoming(req->r_reply);
121
		req->r_con_filling_msg->ops->put(req->r_con_filling_msg);
122
		req->r_con_filling_msg = NULL;
123
	}
124 125
	if (req->r_reply)
		ceph_msg_put(req->r_reply);
126 127

	if (req->r_data_in.type == CEPH_OSD_DATA_TYPE_PAGES &&
128 129 130 131 132
			req->r_data_in.own_pages) {
		num_pages = calc_pages_for((u64)req->r_data_in.alignment,
						(u64)req->r_data_in.length);
		ceph_release_page_vector(req->r_data_in.pages, num_pages);
	}
133
	if (req->r_data_out.type == CEPH_OSD_DATA_TYPE_PAGES &&
134 135 136 137 138
			req->r_data_out.own_pages) {
		num_pages = calc_pages_for((u64)req->r_data_out.alignment,
						(u64)req->r_data_out.length);
		ceph_release_page_vector(req->r_data_out.pages, num_pages);
	}
139

S
Sage Weil 已提交
140
	ceph_put_snap_context(req->r_snapc);
141
	ceph_pagelist_release(&req->r_trail);
S
Sage Weil 已提交
142 143 144 145
	if (req->r_mempool)
		mempool_free(req, req->r_osdc->req_mempool);
	else
		kfree(req);
S
Sage Weil 已提交
146
}
147
EXPORT_SYMBOL(ceph_osdc_release_request);
148

149
struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc,
S
Sage Weil 已提交
150
					       struct ceph_snap_context *snapc,
151
					       unsigned int num_ops,
152
					       bool use_mempool,
153
					       gfp_t gfp_flags)
S
Sage Weil 已提交
154 155 156
{
	struct ceph_osd_request *req;
	struct ceph_msg *msg;
157 158 159 160 161 162 163 164 165 166 167
	size_t msg_size;

	msg_size = 4 + 4 + 8 + 8 + 4+8;
	msg_size += 2 + 4 + 8 + 4 + 4; /* oloc */
	msg_size += 1 + 8 + 4 + 4;     /* pg_t */
	msg_size += 4 + MAX_OBJ_NAME_SIZE;
	msg_size += 2 + num_ops*sizeof(struct ceph_osd_op);
	msg_size += 8;  /* snapid */
	msg_size += 8;  /* snap_seq */
	msg_size += 8 * (snapc ? snapc->num_snaps : 0);  /* snaps */
	msg_size += 4;
S
Sage Weil 已提交
168 169

	if (use_mempool) {
170
		req = mempool_alloc(osdc->req_mempool, gfp_flags);
S
Sage Weil 已提交
171 172
		memset(req, 0, sizeof(*req));
	} else {
173
		req = kzalloc(sizeof(*req), gfp_flags);
S
Sage Weil 已提交
174 175
	}
	if (req == NULL)
176
		return NULL;
S
Sage Weil 已提交
177 178 179

	req->r_osdc = osdc;
	req->r_mempool = use_mempool;
180

S
Sage Weil 已提交
181
	kref_init(&req->r_kref);
S
Sage Weil 已提交
182 183
	init_completion(&req->r_completion);
	init_completion(&req->r_safe_completion);
184
	RB_CLEAR_NODE(&req->r_node);
S
Sage Weil 已提交
185
	INIT_LIST_HEAD(&req->r_unsafe_item);
186 187
	INIT_LIST_HEAD(&req->r_linger_item);
	INIT_LIST_HEAD(&req->r_linger_osd);
188
	INIT_LIST_HEAD(&req->r_req_lru_item);
189 190
	INIT_LIST_HEAD(&req->r_osd_item);

S
Sage Weil 已提交
191 192 193 194 195
	/* create reply message */
	if (use_mempool)
		msg = ceph_msgpool_get(&osdc->msgpool_op_reply, 0);
	else
		msg = ceph_msg_new(CEPH_MSG_OSD_OPREPLY,
196
				   OSD_OPREPLY_FRONT_LEN, gfp_flags, true);
197
	if (!msg) {
S
Sage Weil 已提交
198
		ceph_osdc_put_request(req);
199
		return NULL;
S
Sage Weil 已提交
200 201 202
	}
	req->r_reply = msg;

203 204
	req->r_data_in.type = CEPH_OSD_DATA_TYPE_NONE;
	req->r_data_out.type = CEPH_OSD_DATA_TYPE_NONE;
205
	ceph_pagelist_init(&req->r_trail);
206

S
Sage Weil 已提交
207
	/* create request message; allow space for oid */
S
Sage Weil 已提交
208
	if (use_mempool)
209
		msg = ceph_msgpool_get(&osdc->msgpool_op, 0);
S
Sage Weil 已提交
210
	else
211
		msg = ceph_msg_new(CEPH_MSG_OSD_OP, msg_size, gfp_flags, true);
212
	if (!msg) {
S
Sage Weil 已提交
213
		ceph_osdc_put_request(req);
214
		return NULL;
S
Sage Weil 已提交
215
	}
216

S
Sage Weil 已提交
217
	memset(msg->front.iov_base, 0, msg->front.iov_len);
218 219 220 221 222

	req->r_request = msg;

	return req;
}
223
EXPORT_SYMBOL(ceph_osdc_alloc_request);
224

225 226 227 228 229 230
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);

231
	switch (src->op) {
A
Alex Elder 已提交
232 233
	case CEPH_OSD_OP_STAT:
		break;
234 235 236 237 238 239 240 241 242 243 244
	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;
245 246 247 248 249
	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);

250
		ceph_pagelist_append(&req->r_trail, src->cls.class_name,
251
				     src->cls.class_len);
252
		ceph_pagelist_append(&req->r_trail, src->cls.method_name,
253
				     src->cls.method_len);
254
		ceph_pagelist_append(&req->r_trail, src->cls.indata,
255 256
				     src->cls.indata_len);
		break;
257 258
	case CEPH_OSD_OP_STARTSYNC:
		break;
259 260 261 262 263 264
	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;
265
	default:
266
		pr_err("unrecognized osd opcode %d\n", src->op);
267 268
		WARN_ON(1);
		break;
269 270 271
	case CEPH_OSD_OP_MAPEXT:
	case CEPH_OSD_OP_MASKTRUNC:
	case CEPH_OSD_OP_SPARSE_READ:
272
	case CEPH_OSD_OP_NOTIFY:
273 274 275 276 277 278 279 280 281 282 283 284
	case CEPH_OSD_OP_ASSERT_VER:
	case CEPH_OSD_OP_WRITEFULL:
	case CEPH_OSD_OP_TRUNCATE:
	case CEPH_OSD_OP_ZERO:
	case CEPH_OSD_OP_DELETE:
	case CEPH_OSD_OP_APPEND:
	case CEPH_OSD_OP_SETTRUNC:
	case CEPH_OSD_OP_TRIMTRUNC:
	case CEPH_OSD_OP_TMAPUP:
	case CEPH_OSD_OP_TMAPPUT:
	case CEPH_OSD_OP_TMAPGET:
	case CEPH_OSD_OP_CREATE:
285
	case CEPH_OSD_OP_ROLLBACK:
286 287 288 289 290 291 292 293 294 295 296 297 298
	case CEPH_OSD_OP_OMAPGETKEYS:
	case CEPH_OSD_OP_OMAPGETVALS:
	case CEPH_OSD_OP_OMAPGETHEADER:
	case CEPH_OSD_OP_OMAPGETVALSBYKEYS:
	case CEPH_OSD_OP_MODE_RD:
	case CEPH_OSD_OP_OMAPSETVALS:
	case CEPH_OSD_OP_OMAPSETHEADER:
	case CEPH_OSD_OP_OMAPCLEAR:
	case CEPH_OSD_OP_OMAPRMKEYS:
	case CEPH_OSD_OP_OMAP_CMP:
	case CEPH_OSD_OP_CLONERANGE:
	case CEPH_OSD_OP_ASSERT_SRC_VERSION:
	case CEPH_OSD_OP_SRC_CMPXATTR:
299
	case CEPH_OSD_OP_GETXATTR:
300
	case CEPH_OSD_OP_GETXATTRS:
301 302
	case CEPH_OSD_OP_CMPXATTR:
	case CEPH_OSD_OP_SETXATTR:
303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323
	case CEPH_OSD_OP_SETXATTRS:
	case CEPH_OSD_OP_RESETXATTRS:
	case CEPH_OSD_OP_RMXATTR:
	case CEPH_OSD_OP_PULL:
	case CEPH_OSD_OP_PUSH:
	case CEPH_OSD_OP_BALANCEREADS:
	case CEPH_OSD_OP_UNBALANCEREADS:
	case CEPH_OSD_OP_SCRUB:
	case CEPH_OSD_OP_SCRUB_RESERVE:
	case CEPH_OSD_OP_SCRUB_UNRESERVE:
	case CEPH_OSD_OP_SCRUB_STOP:
	case CEPH_OSD_OP_SCRUB_MAP:
	case CEPH_OSD_OP_WRLOCK:
	case CEPH_OSD_OP_WRUNLOCK:
	case CEPH_OSD_OP_RDLOCK:
	case CEPH_OSD_OP_RDUNLOCK:
	case CEPH_OSD_OP_UPLOCK:
	case CEPH_OSD_OP_DNLOCK:
	case CEPH_OSD_OP_PGLS:
	case CEPH_OSD_OP_PGLS_FILTER:
		pr_err("unsupported osd opcode %s\n",
324
			ceph_osd_op_name(src->op));
325 326
		WARN_ON(1);
		break;
327 328 329 330
	}
	dst->payload_len = cpu_to_le32(src->payload_len);
}

331 332 333 334 335
/*
 * build new request AND message
 *
 */
void ceph_osdc_build_request(struct ceph_osd_request *req,
336
			     u64 off, u64 len, unsigned int num_ops,
337
			     struct ceph_osd_req_op *src_ops,
338
			     struct ceph_snap_context *snapc, u64 snap_id,
339
			     struct timespec *mtime)
340 341
{
	struct ceph_msg *msg = req->r_request;
342
	struct ceph_osd_req_op *src_op;
343
	void *p;
344
	size_t msg_size;
345
	int flags = req->r_flags;
346
	u64 data_len;
347
	int i;
348

349 350
	req->r_num_ops = num_ops;
	req->r_snapid = snap_id;
S
Sage Weil 已提交
351 352
	req->r_snapc = ceph_get_snap_context(snapc);

353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383
	/* encode request */
	msg->hdr.version = cpu_to_le16(4);

	p = msg->front.iov_base;
	ceph_encode_32(&p, 1);   /* client_inc  is always 1 */
	req->r_request_osdmap_epoch = p;
	p += 4;
	req->r_request_flags = p;
	p += 4;
	if (req->r_flags & CEPH_OSD_FLAG_WRITE)
		ceph_encode_timespec(p, mtime);
	p += sizeof(struct ceph_timespec);
	req->r_request_reassert_version = p;
	p += sizeof(struct ceph_eversion); /* will get filled in */

	/* oloc */
	ceph_encode_8(&p, 4);
	ceph_encode_8(&p, 4);
	ceph_encode_32(&p, 8 + 4 + 4);
	req->r_request_pool = p;
	p += 8;
	ceph_encode_32(&p, -1);  /* preferred */
	ceph_encode_32(&p, 0);   /* key len */

	ceph_encode_8(&p, 1);
	req->r_request_pgid = p;
	p += 8 + 4;
	ceph_encode_32(&p, -1);  /* preferred */

	/* oid */
	ceph_encode_32(&p, req->r_oid_len);
384
	memcpy(p, req->r_oid, req->r_oid_len);
385
	dout("oid '%.*s' len %d\n", req->r_oid_len, req->r_oid, req->r_oid_len);
386
	p += req->r_oid_len;
S
Sage Weil 已提交
387

388 389
	/* ops */
	ceph_encode_16(&p, num_ops);
390
	src_op = src_ops;
391 392 393 394 395
	req->r_request_ops = p;
	for (i = 0; i < num_ops; i++, src_op++) {
		osd_req_encode_op(req, p, src_op);
		p += sizeof(struct ceph_osd_op);
	}
396

397 398 399 400 401
	/* snaps */
	ceph_encode_64(&p, req->r_snapid);
	ceph_encode_64(&p, req->r_snapc ? req->r_snapc->seq : 0);
	ceph_encode_32(&p, req->r_snapc ? req->r_snapc->num_snaps : 0);
	if (req->r_snapc) {
S
Sage Weil 已提交
402
		for (i = 0; i < snapc->num_snaps; i++) {
403
			ceph_encode_64(&p, req->r_snapc->snaps[i]);
S
Sage Weil 已提交
404 405 406
		}
	}

407 408 409
	req->r_request_attempts = p;
	p += 4;

410
	data_len = req->r_trail.length;
411 412
	if (flags & CEPH_OSD_FLAG_WRITE) {
		req->r_request->hdr.data_off = cpu_to_le16(off);
413
		data_len += len;
414
	}
415
	req->r_request->hdr.data_len = cpu_to_le32(data_len);
416

S
Sage Weil 已提交
417
	BUG_ON(p > msg->front.iov_base + msg->front.iov_len);
418 419 420
	msg_size = p - msg->front.iov_base;
	msg->front.iov_len = msg_size;
	msg->hdr.front_len = cpu_to_le32(msg_size);
421 422 423

	dout("build_request msg_size was %d num_ops %d\n", (int)msg_size,
	     num_ops);
424 425
	return;
}
426
EXPORT_SYMBOL(ceph_osdc_build_request);
427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448

/*
 * 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,
449
					       bool use_mempool)
450
{
A
Alex Elder 已提交
451
	struct ceph_osd_req_op ops[2];
452
	struct ceph_osd_request *req;
A
Alex Elder 已提交
453
	unsigned int num_op = 1;
454
	u64 bno = 0;
455
	int r;
456

A
Alex Elder 已提交
457 458
	memset(&ops, 0, sizeof ops);

459 460 461 462 463 464
	ops[0].op = opcode;
	ops[0].extent.truncate_seq = truncate_seq;
	ops[0].extent.truncate_size = truncate_size;

	if (do_sync) {
		ops[1].op = CEPH_OSD_OP_STARTSYNC;
A
Alex Elder 已提交
465 466
		num_op++;
	}
467

A
Alex Elder 已提交
468 469
	req = ceph_osdc_alloc_request(osdc, snapc, num_op, use_mempool,
					GFP_NOFS);
470
	if (!req)
471
		return ERR_PTR(-ENOMEM);
472
	req->r_flags = flags;
473 474

	/* calculate max write size */
475
	r = calc_layout(layout, off, plen, ops, &bno);
476 477
	if (r < 0) {
		ceph_osdc_put_request(req);
478
		return ERR_PTR(r);
479
	}
480

481 482
	req->r_file_layout = *layout;  /* keep a copy */

483 484 485
	snprintf(req->r_oid, sizeof(req->r_oid), "%llx.%08llx", vino.ino, bno);
	req->r_oid_len = strlen(req->r_oid);

A
Alex Elder 已提交
486 487
	ceph_osdc_build_request(req, off, *plen, num_op, ops,
				snapc, vino.snap, mtime);
488

S
Sage Weil 已提交
489 490
	return req;
}
491
EXPORT_SYMBOL(ceph_osdc_new_request);
S
Sage Weil 已提交
492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557

/*
 * 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;
}

558 559 560 561 562 563
/*
 * Resubmit requests pending on the given osd.
 */
static void __kick_osd_requests(struct ceph_osd_client *osdc,
				struct ceph_osd *osd)
{
564
	struct ceph_osd_request *req, *nreq;
565 566 567 568
	int err;

	dout("__kick_osd_requests osd%d\n", osd->o_osd);
	err = __reset_osd(osdc, osd);
569
	if (err)
570 571 572 573 574 575
		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);
576 577 578 579 580 581
		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) {
582 583 584 585 586
		/*
		 * reregister request prior to unregistering linger so
		 * that r_osd is preserved.
		 */
		BUG_ON(!list_empty(&req->r_req_lru_item));
587
		__register_request(osdc, req);
588 589 590
		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);
591 592
		dout("requeued lingering %p tid %llu osd%d\n", req, req->r_tid,
		     osd->o_osd);
593 594 595
	}
}

S
Sage Weil 已提交
596
/*
597
 * If the osd connection drops, we need to resubmit all requests.
S
Sage Weil 已提交
598 599 600 601 602 603 604 605 606 607 608
 */
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
609 610
	mutex_lock(&osdc->request_mutex);
	__kick_osd_requests(osdc, osd);
A
Alex Elder 已提交
611
	__send_queued(osdc);
S
Sage Weil 已提交
612
	mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
613 614 615 616 617 618
	up_read(&osdc->map_sem);
}

/*
 * Track open sessions with osds.
 */
619
static struct ceph_osd *create_osd(struct ceph_osd_client *osdc, int onum)
S
Sage Weil 已提交
620 621 622 623 624 625 626 627 628
{
	struct ceph_osd *osd;

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

	atomic_set(&osd->o_ref, 1);
	osd->o_osdc = osdc;
629
	osd->o_osd = onum;
630
	RB_CLEAR_NODE(&osd->o_node);
S
Sage Weil 已提交
631
	INIT_LIST_HEAD(&osd->o_requests);
632
	INIT_LIST_HEAD(&osd->o_linger_requests);
633
	INIT_LIST_HEAD(&osd->o_osd_lru);
S
Sage Weil 已提交
634 635
	osd->o_incarnation = 1;

636
	ceph_con_init(&osd->o_con, osd, &osd_con_ops, &osdc->client->msgr);
637

638
	INIT_LIST_HEAD(&osd->o_keepalive_item);
S
Sage Weil 已提交
639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657
	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);
658
	if (atomic_dec_and_test(&osd->o_ref) && osd->o_auth.authorizer) {
S
Sage Weil 已提交
659 660
		struct ceph_auth_client *ac = osd->o_osdc->client->monc.auth;

661
		if (ac->ops && ac->ops->destroy_authorizer)
662
			ac->ops->destroy_authorizer(ac, osd->o_auth.authorizer);
S
Sage Weil 已提交
663
		kfree(osd);
S
Sage Weil 已提交
664
	}
S
Sage Weil 已提交
665 666 667 668 669
}

/*
 * remove an osd from our map
 */
670
static void __remove_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
S
Sage Weil 已提交
671
{
672
	dout("__remove_osd %p\n", osd);
S
Sage Weil 已提交
673 674
	BUG_ON(!list_empty(&osd->o_requests));
	rb_erase(&osd->o_node, &osdc->osds);
675
	list_del_init(&osd->o_osd_lru);
S
Sage Weil 已提交
676 677 678 679
	ceph_con_close(&osd->o_con);
	put_osd(osd);
}

680 681
static void remove_all_osds(struct ceph_osd_client *osdc)
{
682
	dout("%s %p\n", __func__, osdc);
683 684 685 686 687 688 689 690 691
	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);
}

692 693 694 695 696 697
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);
698
	osd->lru_ttl = jiffies + osdc->client->options->osd_idle_ttl * HZ;
699 700 701 702 703 704 705 706 707
}

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);
}

708
static void remove_old_osds(struct ceph_osd_client *osdc)
709 710 711 712 713 714
{
	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) {
715
		if (time_before(jiffies, osd->lru_ttl))
716 717 718 719 720 721
			break;
		__remove_osd(osdc, osd);
	}
	mutex_unlock(&osdc->request_mutex);
}

S
Sage Weil 已提交
722 723 724
/*
 * reset osd connect
 */
725
static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
S
Sage Weil 已提交
726
{
A
Alex Elder 已提交
727
	struct ceph_entity_addr *peer_addr;
S
Sage Weil 已提交
728

729
	dout("__reset_osd %p osd%d\n", osd, osd->o_osd);
730 731
	if (list_empty(&osd->o_requests) &&
	    list_empty(&osd->o_linger_requests)) {
732
		__remove_osd(osdc, osd);
A
Alex Elder 已提交
733 734 735 736 737 738 739 740 741

		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;

742 743 744 745 746
		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
747 748

		return -EAGAIN;
S
Sage Weil 已提交
749
	}
A
Alex Elder 已提交
750 751 752 753 754 755

	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
756 757 758 759 760 761 762 763
}

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;

764
	dout("__insert_osd %p osd%d\n", new, new->o_osd);
S
Sage Weil 已提交
765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796
	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;
}

797 798 799
static void __schedule_osd_timeout(struct ceph_osd_client *osdc)
{
	schedule_delayed_work(&osdc->timeout_work,
800
			osdc->client->options->osd_keepalive_timeout * HZ);
801 802 803 804 805 806
}

static void __cancel_osd_timeout(struct ceph_osd_client *osdc)
{
	cancel_delayed_work(&osdc->timeout_work);
}
S
Sage Weil 已提交
807 808 809 810 811

/*
 * Register request, assign tid.  If this is the first request, set up
 * the timeout event.
 */
812 813
static void __register_request(struct ceph_osd_client *osdc,
			       struct ceph_osd_request *req)
S
Sage Weil 已提交
814 815
{
	req->r_tid = ++osdc->last_tid;
816
	req->r_request->hdr.tid = cpu_to_le64(req->r_tid);
817
	dout("__register_request %p tid %lld\n", req, req->r_tid);
S
Sage Weil 已提交
818 819 820 821
	__insert_request(osdc, req);
	ceph_osdc_get_request(req);
	osdc->num_requests++;
	if (osdc->num_requests == 1) {
822 823
		dout(" first request, scheduling timeout\n");
		__schedule_osd_timeout(osdc);
S
Sage Weil 已提交
824
	}
825 826 827 828 829 830 831
}

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
832 833 834 835 836 837 838 839 840
	mutex_unlock(&osdc->request_mutex);
}

/*
 * called under osdc->request_mutex
 */
static void __unregister_request(struct ceph_osd_client *osdc,
				 struct ceph_osd_request *req)
{
841 842 843 844 845 846
	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
847 848 849 850
	dout("__unregister_request %p tid %lld\n", req, req->r_tid);
	rb_erase(&req->r_node, &osdc->requests);
	osdc->num_requests--;

851 852
	if (req->r_osd) {
		/* make sure the original request isn't in flight. */
853
		ceph_msg_revoke(req->r_request);
854 855

		list_del_init(&req->r_osd_item);
856 857 858
		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);
859
			__move_osd_to_lru(osdc, req->r_osd);
860
		}
861
		if (list_empty(&req->r_linger_item))
862
			req->r_osd = NULL;
863
	}
S
Sage Weil 已提交
864

A
Alex Elder 已提交
865
	list_del_init(&req->r_req_lru_item);
S
Sage Weil 已提交
866 867
	ceph_osdc_put_request(req);

868 869 870
	if (osdc->num_requests == 0) {
		dout(" no requests, canceling timeout\n");
		__cancel_osd_timeout(osdc);
S
Sage Weil 已提交
871 872 873 874 875 876 877 878
	}
}

/*
 * Cancel a previously queued request message
 */
static void __cancel_request(struct ceph_osd_request *req)
{
879
	if (req->r_sent && req->r_osd) {
880
		ceph_msg_revoke(req->r_request);
S
Sage Weil 已提交
881 882 883 884
		req->r_sent = 0;
	}
}

885 886 887 888 889
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);
890 891 892
	if (req->r_osd)
		list_add_tail(&req->r_linger_osd,
			      &req->r_osd->o_linger_requests);
893 894 895 896 897 898
}

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
899
	list_del_init(&req->r_linger_item);
900 901 902 903 904 905 906 907
	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);
		}
908 909
		if (list_empty(&req->r_osd_item))
			req->r_osd = NULL;
910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939
	}
}

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
940 941 942
/*
 * 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
943
 * no up osd, set r_osd to NULL.  Move the request to the appropriate list
944
 * (unsent, homeless) or leave on in-flight lru.
S
Sage Weil 已提交
945 946 947 948 949
 *
 * Return 0 if unchanged, 1 if changed, or negative on error.
 *
 * Caller should hold map_sem for read and request_mutex.
 */
950
static int __map_request(struct ceph_osd_client *osdc,
951
			 struct ceph_osd_request *req, int force_resend)
S
Sage Weil 已提交
952
{
953
	struct ceph_pg pgid;
954 955
	int acting[CEPH_PG_MAX_SIZE];
	int o = -1, num = 0;
S
Sage Weil 已提交
956 957
	int err;

958
	dout("map_request %p tid %lld\n", req, req->r_tid);
959 960
	err = ceph_calc_ceph_pg(&pgid, req->r_oid, osdc->osdmap,
				ceph_file_layout_pg_pool(req->r_file_layout));
961 962
	if (err) {
		list_move(&req->r_req_lru_item, &osdc->req_notarget);
S
Sage Weil 已提交
963
		return err;
964
	}
965 966
	req->r_pgid = pgid;

967 968 969 970 971
	err = ceph_calc_pg_acting(osdc->osdmap, pgid, acting);
	if (err > 0) {
		o = acting[0];
		num = err;
	}
S
Sage Weil 已提交
972

973 974
	if ((!force_resend &&
	     req->r_osd && req->r_osd->o_osd == o &&
975 976 977
	     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
978 979 980
	    (req->r_osd == NULL && o == -1))
		return 0;  /* no change */

981 982
	dout("map_request tid %llu pgid %lld.%x osd%d (was osd%d)\n",
	     req->r_tid, pgid.pool, pgid.seed, o,
S
Sage Weil 已提交
983 984
	     req->r_osd ? req->r_osd->o_osd : -1);

985 986 987 988
	/* 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
989 990 991 992 993 994 995 996
	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) {
997
		err = -ENOMEM;
998
		req->r_osd = create_osd(osdc, o);
999 1000
		if (!req->r_osd) {
			list_move(&req->r_req_lru_item, &osdc->req_notarget);
1001
			goto out;
1002
		}
S
Sage Weil 已提交
1003

1004
		dout("map_request osd %p is osd%d\n", req->r_osd, o);
S
Sage Weil 已提交
1005 1006
		__insert_osd(osdc, req->r_osd);

1007 1008 1009
		ceph_con_open(&req->r_osd->o_con,
			      CEPH_ENTITY_TYPE_OSD, o,
			      &osdc->osdmap->osd_addr[o]);
S
Sage Weil 已提交
1010 1011
	}

1012 1013
	if (req->r_osd) {
		__remove_osd_from_lru(req->r_osd);
S
Sage Weil 已提交
1014
		list_add(&req->r_osd_item, &req->r_osd->o_requests);
1015 1016 1017
		list_move(&req->r_req_lru_item, &osdc->req_unsent);
	} else {
		list_move(&req->r_req_lru_item, &osdc->req_notarget);
1018
	}
1019
	err = 1;   /* osd or pg changed */
S
Sage Weil 已提交
1020 1021 1022 1023 1024 1025 1026 1027

out:
	return err;
}

/*
 * caller should hold map_sem (for read) and request_mutex
 */
1028 1029
static void __send_request(struct ceph_osd_client *osdc,
			   struct ceph_osd_request *req)
S
Sage Weil 已提交
1030
{
1031
	void *p;
S
Sage Weil 已提交
1032

1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046
	dout("send_request %p tid %llu to osd%d flags %d pg %lld.%x\n",
	     req, req->r_tid, req->r_osd->o_osd, req->r_flags,
	     (unsigned long long)req->r_pgid.pool, req->r_pgid.seed);

	/* fill in message content that changes each time we send it */
	put_unaligned_le32(osdc->osdmap->epoch, req->r_request_osdmap_epoch);
	put_unaligned_le32(req->r_flags, req->r_request_flags);
	put_unaligned_le64(req->r_pgid.pool, req->r_request_pool);
	p = req->r_request_pgid;
	ceph_encode_64(&p, req->r_pgid.pool);
	ceph_encode_32(&p, req->r_pgid.seed);
	put_unaligned_le64(1, req->r_request_attempts);  /* FIXME */
	memcpy(req->r_request_reassert_version, &req->r_reassert_version,
	       sizeof(req->r_reassert_version));
1047

S
Sage Weil 已提交
1048
	req->r_stamp = jiffies;
1049
	list_move_tail(&req->r_req_lru_item, &osdc->req_lru);
S
Sage Weil 已提交
1050 1051 1052 1053 1054 1055

	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;
}

1056 1057 1058
/*
 * Send any requests in the queue (req_unsent).
 */
A
Alex Elder 已提交
1059
static void __send_queued(struct ceph_osd_client *osdc)
1060 1061 1062
{
	struct ceph_osd_request *req, *tmp;

A
Alex Elder 已提交
1063 1064
	dout("__send_queued\n");
	list_for_each_entry_safe(req, tmp, &osdc->req_unsent, r_req_lru_item)
1065 1066 1067
		__send_request(osdc, req);
}

S
Sage Weil 已提交
1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080
/*
 * 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
1081
	struct ceph_osd_request *req;
S
Sage Weil 已提交
1082
	struct ceph_osd *osd;
1083
	unsigned long keepalive =
1084
		osdc->client->options->osd_keepalive_timeout * HZ;
1085
	struct list_head slow_osds;
S
Sage Weil 已提交
1086 1087 1088 1089 1090 1091 1092
	dout("timeout\n");
	down_read(&osdc->map_sem);

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

	mutex_lock(&osdc->request_mutex);

1093 1094 1095 1096 1097 1098 1099
	/*
	 * 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
1100
		if (time_before(jiffies, req->r_stamp + keepalive))
1101 1102 1103 1104 1105
			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
1106
		     req->r_tid, osd->o_osd);
1107 1108 1109 1110 1111 1112
		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
1113 1114 1115
		ceph_con_keepalive(&osd->o_con);
	}

1116
	__schedule_osd_timeout(osdc);
A
Alex Elder 已提交
1117
	__send_queued(osdc);
S
Sage Weil 已提交
1118 1119 1120 1121
	mutex_unlock(&osdc->request_mutex);
	up_read(&osdc->map_sem);
}

1122 1123 1124 1125 1126 1127
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 =
1128
		osdc->client->options->osd_idle_ttl * HZ >> 2;
1129 1130 1131

	dout("osds timeout\n");
	down_read(&osdc->map_sem);
1132
	remove_old_osds(osdc);
1133 1134 1135 1136 1137 1138
	up_read(&osdc->map_sem);

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

S
Sage Weil 已提交
1139 1140 1141 1142 1143 1144 1145
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 */
}

1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165
static int __decode_pgid(void **p, void *end, struct ceph_pg *pgid)
{
	__u8 v;

	ceph_decode_need(p, end, 1 + 8 + 4 + 4, bad);
	v = ceph_decode_8(p);
	if (v > 1) {
		pr_warning("do not understand pg encoding %d > 1", v);
		return -EINVAL;
	}
	pgid->pool = ceph_decode_64(p);
	pgid->seed = ceph_decode_32(p);
	*p += 4;
	return 0;

bad:
	pr_warning("incomplete pg encoding");
	return -EINVAL;
}

S
Sage Weil 已提交
1166 1167 1168 1169
/*
 * handle osd op reply.  either call the callback if it is specified,
 * or do the completion to wake up the waiting thread.
 */
1170 1171
static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
			 struct ceph_connection *con)
S
Sage Weil 已提交
1172
{
1173
	void *p, *end;
S
Sage Weil 已提交
1174 1175
	struct ceph_osd_request *req;
	u64 tid;
1176 1177
	int object_len;
	int numops, payload_len, flags;
1178
	s32 result;
1179 1180 1181 1182 1183 1184
	s32 retry_attempt;
	struct ceph_pg pg;
	int err;
	u32 reassert_epoch;
	u64 reassert_version;
	u32 osdmap_epoch;
1185
	int already_completed;
1186
	int i;
S
Sage Weil 已提交
1187

1188
	tid = le64_to_cpu(msg->hdr.tid);
1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200
	dout("handle_reply %p tid %llu\n", msg, tid);

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

	ceph_decode_need(&p, end, 4, bad);
	object_len = ceph_decode_32(&p);
	ceph_decode_need(&p, end, object_len, bad);
	p += object_len;

	err = __decode_pgid(&p, end, &pg);
	if (err)
S
Sage Weil 已提交
1201
		goto bad;
1202 1203 1204 1205 1206 1207 1208 1209

	ceph_decode_need(&p, end, 8 + 4 + 4 + 8 + 4, bad);
	flags = ceph_decode_64(&p);
	result = ceph_decode_32(&p);
	reassert_epoch = ceph_decode_32(&p);
	reassert_version = ceph_decode_64(&p);
	osdmap_epoch = ceph_decode_32(&p);

S
Sage Weil 已提交
1210 1211 1212 1213 1214 1215 1216 1217 1218
	/* 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);
1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250

	dout("handle_reply %p tid %llu req %p result %d\n", msg, tid,
	     req, result);

	ceph_decode_need(&p, end, 4, bad);
	numops = ceph_decode_32(&p);
	if (numops > CEPH_OSD_MAX_OP)
		goto bad_put;
	if (numops != req->r_num_ops)
		goto bad_put;
	payload_len = 0;
	ceph_decode_need(&p, end, numops * sizeof(struct ceph_osd_op), bad);
	for (i = 0; i < numops; i++) {
		struct ceph_osd_op *op = p;
		int len;

		len = le32_to_cpu(op->payload_len);
		req->r_reply_op_len[i] = len;
		dout(" op %d has %d bytes\n", i, len);
		payload_len += len;
		p += sizeof(*op);
	}
	if (payload_len != le32_to_cpu(msg->hdr.data_len)) {
		pr_warning("sum of op payload lens %d != data_len %d",
			   payload_len, le32_to_cpu(msg->hdr.data_len));
		goto bad_put;
	}

	ceph_decode_need(&p, end, 4 + numops * 4, bad);
	retry_attempt = ceph_decode_32(&p);
	for (i = 0; i < numops; i++)
		req->r_reply_op_result[i] = ceph_decode_32(&p);
S
Sage Weil 已提交
1251

1252
	/*
1253
	 * if this connection filled our message, drop our reference now, to
1254 1255
	 * avoid a (safe but slower) revoke later.
	 */
1256
	if (req->r_con_filling_msg == con && req->r_reply == msg) {
S
Sage Weil 已提交
1257
		dout(" dropping con_filling_msg ref %p\n", con);
1258
		req->r_con_filling_msg = NULL;
1259
		con->ops->put(con);
1260 1261
	}

S
Sage Weil 已提交
1262
	if (!req->r_got_reply) {
1263
		unsigned int bytes;
S
Sage Weil 已提交
1264

1265
		req->r_result = result;
S
Sage Weil 已提交
1266 1267 1268 1269 1270 1271 1272
		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, */
1273 1274
		req->r_reassert_version.epoch = cpu_to_le32(reassert_epoch);
		req->r_reassert_version.version = cpu_to_le64(reassert_version);
S
Sage Weil 已提交
1275 1276 1277 1278

		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
1279
		mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1280 1281 1282 1283 1284
		goto done;
	}

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

1285 1286 1287
	if (req->r_linger && (flags & CEPH_OSD_FLAG_ONDISK))
		__register_linger_request(osdc, req);

S
Sage Weil 已提交
1288
	/* either this is a read, or we got the safe response */
1289 1290
	if (result < 0 ||
	    (flags & CEPH_OSD_FLAG_ONDISK) ||
S
Sage Weil 已提交
1291 1292 1293
	    ((flags & CEPH_OSD_FLAG_WRITE) == 0))
		__unregister_request(osdc, req);

1294 1295
	already_completed = req->r_completed;
	req->r_completed = 1;
S
Sage Weil 已提交
1296
	mutex_unlock(&osdc->request_mutex);
1297 1298
	if (already_completed)
		goto done;
S
Sage Weil 已提交
1299 1300 1301 1302

	if (req->r_callback)
		req->r_callback(req, msg);
	else
1303
		complete_all(&req->r_completion);
S
Sage Weil 已提交
1304

S
Sage Weil 已提交
1305 1306
	if (flags & CEPH_OSD_FLAG_ONDISK)
		complete_request(req);
S
Sage Weil 已提交
1307 1308

done:
1309
	dout("req=%p req->r_linger=%d\n", req, req->r_linger);
S
Sage Weil 已提交
1310 1311 1312
	ceph_osdc_put_request(req);
	return;

1313 1314
bad_put:
	ceph_osdc_put_request(req);
S
Sage Weil 已提交
1315
bad:
1316 1317
	pr_err("corrupt osd_op_reply got %d %d\n",
	       (int)msg->front.iov_len, le32_to_cpu(msg->hdr.front_len));
1318
	ceph_msg_dump(msg);
S
Sage Weil 已提交
1319 1320
}

1321
static void reset_changed_osds(struct ceph_osd_client *osdc)
S
Sage Weil 已提交
1322 1323 1324
{
	struct rb_node *p, *n;

1325 1326
	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
1327

1328 1329 1330 1331 1332 1333 1334
		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
1335
	}
1336 1337 1338
}

/*
1339 1340
 * Requeue requests whose mapping to an OSD has changed.  If requests map to
 * no osd, request a new map.
1341
 *
1342
 * Caller should hold map_sem for read.
1343
 */
1344
static void kick_requests(struct ceph_osd_client *osdc, int force_resend)
1345
{
1346
	struct ceph_osd_request *req, *nreq;
1347 1348 1349
	struct rb_node *p;
	int needmap = 0;
	int err;
1350

1351
	dout("kick_requests %s\n", force_resend ? " (force resend)" : "");
1352
	mutex_lock(&osdc->request_mutex);
1353
	for (p = rb_first(&osdc->requests); p; ) {
1354
		req = rb_entry(p, struct ceph_osd_request, r_node);
1355
		p = rb_next(p);
1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373

		/*
		 * 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;
		}

1374
		err = __map_request(osdc, req, force_resend);
1375 1376 1377 1378 1379 1380
		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) {
1381 1382 1383 1384
			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);
1385
				req->r_flags |= CEPH_OSD_FLAG_RETRY;
1386 1387
			}
		}
1388 1389 1390 1391 1392 1393
	}

	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);

1394
		err = __map_request(osdc, req, force_resend);
1395
		dout("__map_request returned %d\n", err);
1396 1397 1398 1399 1400 1401 1402 1403
		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;
1404
		}
1405 1406 1407 1408

		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);
1409
		__unregister_linger_request(osdc, req);
1410
	}
S
Sage Weil 已提交
1411 1412 1413 1414 1415 1416
	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);
	}
1417
	reset_changed_osds(osdc);
1418
}
1419 1420


S
Sage Weil 已提交
1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443
/*
 * 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));
1444 1445
	if (ceph_check_fsid(osdc->client, &fsid) < 0)
		return;
S
Sage Weil 已提交
1446 1447 1448 1449 1450 1451 1452 1453

	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);
1454 1455
		epoch = ceph_decode_32(&p);
		maplen = ceph_decode_32(&p);
S
Sage Weil 已提交
1456 1457 1458 1459 1460 1461 1462
		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,
1463
							  &osdc->client->msgr);
S
Sage Weil 已提交
1464 1465 1466 1467
			if (IS_ERR(newmap)) {
				err = PTR_ERR(newmap);
				goto bad;
			}
1468
			BUG_ON(!newmap);
S
Sage Weil 已提交
1469 1470 1471 1472
			if (newmap != osdc->osdmap) {
				ceph_osdmap_destroy(osdc->osdmap);
				osdc->osdmap = newmap;
			}
1473
			kick_requests(osdc, 0);
S
Sage Weil 已提交
1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488
		} 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);
1489 1490
		epoch = ceph_decode_32(&p);
		maplen = ceph_decode_32(&p);
S
Sage Weil 已提交
1491 1492 1493 1494 1495 1496 1497 1498 1499
		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 {
1500 1501
			int skipped_map = 0;

S
Sage Weil 已提交
1502 1503 1504 1505 1506 1507
			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;
			}
1508
			BUG_ON(!newmap);
S
Sage Weil 已提交
1509 1510
			oldmap = osdc->osdmap;
			osdc->osdmap = newmap;
1511 1512 1513
			if (oldmap) {
				if (oldmap->epoch + 1 < newmap->epoch)
					skipped_map = 1;
S
Sage Weil 已提交
1514
				ceph_osdmap_destroy(oldmap);
1515 1516
			}
			kick_requests(osdc, skipped_map);
S
Sage Weil 已提交
1517 1518 1519 1520 1521 1522 1523 1524
		}
		p += maplen;
		nr_maps--;
	}

done:
	downgrade_write(&osdc->map_sem);
	ceph_monc_got_osdmap(&osdc->client->monc, osdc->osdmap->epoch);
1525 1526 1527 1528 1529 1530 1531 1532 1533

	/*
	 * 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);

A
Alex Elder 已提交
1534 1535 1536
	mutex_lock(&osdc->request_mutex);
	__send_queued(osdc);
	mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1537
	up_read(&osdc->map_sem);
1538
	wake_up_all(&osdc->client->auth_wq);
S
Sage Weil 已提交
1539 1540 1541 1542
	return;

bad:
	pr_err("osdc handle_map corrupt msg\n");
1543
	ceph_msg_dump(msg);
S
Sage Weil 已提交
1544 1545 1546 1547
	up_write(&osdc->map_sem);
	return;
}

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 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629
/*
 * 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 *),
1630
			   void *data, struct ceph_osd_event **pevent)
1631 1632 1633 1634 1635 1636 1637 1638 1639
{
	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;
1640
	event->one_shot = 0;
1641 1642 1643
	event->data = data;
	event->osdc = osdc;
	INIT_LIST_HEAD(&event->osd_node);
1644
	RB_CLEAR_NODE(&event->node);
1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690
	kref_init(&event->kref);   /* one ref for us */
	kref_get(&event->kref);    /* one ref for the caller */

	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);
	dout("do_event_work completed %p\n", event);
	ceph_osdc_put_event(event);
	kfree(event_work);
}


/*
 * Process osd watch notifications
 */
1691 1692
static void handle_watch_notify(struct ceph_osd_client *osdc,
				struct ceph_msg *msg)
1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712
{
	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) {
1713
		BUG_ON(event->one_shot);
1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724
		get_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;
		}
1725
		INIT_WORK(&event_work->work, do_event_work);
1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746
		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:
	ceph_osdc_put_event(event);
	return;

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

1747 1748
static void ceph_osdc_msg_data_set(struct ceph_msg *msg,
				struct ceph_osd_data *osd_data)
S
Sage Weil 已提交
1749
{
1750
	if (osd_data->type == CEPH_OSD_DATA_TYPE_PAGES) {
1751
		BUG_ON(osd_data->length > (u64) SIZE_MAX);
1752
		if (osd_data->length)
1753 1754
			ceph_msg_data_set_pages(msg, osd_data->pages,
				osd_data->length, osd_data->alignment);
1755
#ifdef CONFIG_BLOCK
1756
	} else if (osd_data->type == CEPH_OSD_DATA_TYPE_BIO) {
1757
		ceph_msg_data_set_bio(msg, osd_data->bio);
1758
#endif
1759
	} else {
1760
		BUG_ON(osd_data->type != CEPH_OSD_DATA_TYPE_NONE);
1761
	}
1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776
}

/*
 * Register request, send initial attempt.
 */
int ceph_osdc_start_request(struct ceph_osd_client *osdc,
			    struct ceph_osd_request *req,
			    bool nofail)
{
	int rc = 0;

	/* Set up response incoming data and request outgoing data fields */

	ceph_osdc_msg_data_set(req->r_reply, &req->r_data_in);
	ceph_osdc_msg_data_set(req->r_request, &req->r_data_out);
1777 1778
	if (req->r_trail.length)
		ceph_msg_data_set_trail(req->r_request, &req->r_trail);
S
Sage Weil 已提交
1779 1780 1781 1782 1783

	register_request(osdc, req);

	down_read(&osdc->map_sem);
	mutex_lock(&osdc->request_mutex);
S
Sage Weil 已提交
1784 1785 1786 1787 1788 1789
	/*
	 * 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) {
1790
		rc = __map_request(osdc, req, 0);
1791 1792 1793 1794 1795 1796
		if (rc < 0) {
			if (nofail) {
				dout("osdc_start_request failed map, "
				     " will retry %lld\n", req->r_tid);
				rc = 0;
			}
1797
			goto out_unlock;
1798
		}
1799 1800 1801 1802
		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 {
1803
			__send_request(osdc, req);
S
Sage Weil 已提交
1804
		}
1805
		rc = 0;
S
Sage Weil 已提交
1806
	}
1807 1808

out_unlock:
S
Sage Weil 已提交
1809 1810 1811 1812
	mutex_unlock(&osdc->request_mutex);
	up_read(&osdc->map_sem);
	return rc;
}
1813
EXPORT_SYMBOL(ceph_osdc_start_request);
S
Sage Weil 已提交
1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826

/*
 * 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);
1827
		__unregister_request(osdc, req);
S
Sage Weil 已提交
1828
		mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1829
		complete_request(req);
1830
		dout("wait_request tid %llu canceled/timed out\n", req->r_tid);
S
Sage Weil 已提交
1831 1832 1833 1834 1835 1836
		return rc;
	}

	dout("wait_request tid %llu result %d\n", req->r_tid, req->r_result);
	return req->r_result;
}
1837
EXPORT_SYMBOL(ceph_osdc_wait_request);
S
Sage Weil 已提交
1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870

/*
 * 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);
}
1871
EXPORT_SYMBOL(ceph_osdc_sync);
S
Sage Weil 已提交
1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888

/*
 * 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;
1889
	INIT_LIST_HEAD(&osdc->osd_lru);
S
Sage Weil 已提交
1890
	osdc->requests = RB_ROOT;
1891
	INIT_LIST_HEAD(&osdc->req_lru);
1892 1893
	INIT_LIST_HEAD(&osdc->req_unsent);
	INIT_LIST_HEAD(&osdc->req_notarget);
1894
	INIT_LIST_HEAD(&osdc->req_linger);
S
Sage Weil 已提交
1895 1896
	osdc->num_requests = 0;
	INIT_DELAYED_WORK(&osdc->timeout_work, handle_timeout);
1897
	INIT_DELAYED_WORK(&osdc->osds_timeout_work, handle_osds_timeout);
1898 1899 1900
	spin_lock_init(&osdc->event_lock);
	osdc->event_tree = RB_ROOT;
	osdc->event_count = 0;
1901 1902

	schedule_delayed_work(&osdc->osds_timeout_work,
1903
	   round_jiffies_relative(osdc->client->options->osd_idle_ttl * HZ));
S
Sage Weil 已提交
1904

1905
	err = -ENOMEM;
S
Sage Weil 已提交
1906 1907 1908
	osdc->req_mempool = mempool_create_kmalloc_pool(10,
					sizeof(struct ceph_osd_request));
	if (!osdc->req_mempool)
1909
		goto out;
S
Sage Weil 已提交
1910

1911 1912
	err = ceph_msgpool_init(&osdc->msgpool_op, CEPH_MSG_OSD_OP,
				OSD_OP_FRONT_LEN, 10, true,
1913
				"osd_op");
S
Sage Weil 已提交
1914
	if (err < 0)
1915
		goto out_mempool;
1916
	err = ceph_msgpool_init(&osdc->msgpool_op_reply, CEPH_MSG_OSD_OPREPLY,
1917 1918
				OSD_OPREPLY_FRONT_LEN, 10, true,
				"osd_op_reply");
S
Sage Weil 已提交
1919 1920
	if (err < 0)
		goto out_msgpool;
1921 1922 1923 1924 1925 1926 1927

	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
1928
	return 0;
1929

S
Sage Weil 已提交
1930 1931
out_msgpool:
	ceph_msgpool_destroy(&osdc->msgpool_op);
1932 1933 1934 1935
out_mempool:
	mempool_destroy(osdc->req_mempool);
out:
	return err;
S
Sage Weil 已提交
1936 1937 1938 1939
}

void ceph_osdc_stop(struct ceph_osd_client *osdc)
{
1940 1941
	flush_workqueue(osdc->notify_wq);
	destroy_workqueue(osdc->notify_wq);
S
Sage Weil 已提交
1942
	cancel_delayed_work_sync(&osdc->timeout_work);
1943
	cancel_delayed_work_sync(&osdc->osds_timeout_work);
S
Sage Weil 已提交
1944 1945 1946 1947
	if (osdc->osdmap) {
		ceph_osdmap_destroy(osdc->osdmap);
		osdc->osdmap = NULL;
	}
1948
	remove_all_osds(osdc);
S
Sage Weil 已提交
1949 1950
	mempool_destroy(osdc->req_mempool);
	ceph_msgpool_destroy(&osdc->msgpool_op);
S
Sage Weil 已提交
1951
	ceph_msgpool_destroy(&osdc->msgpool_op_reply);
S
Sage Weil 已提交
1952 1953 1954 1955 1956 1957 1958 1959 1960 1961
}

/*
 * 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,
1962
			struct page **pages, int num_pages, int page_align)
S
Sage Weil 已提交
1963 1964
{
	struct ceph_osd_request *req;
1965
	struct ceph_osd_data *osd_data;
S
Sage Weil 已提交
1966 1967 1968 1969 1970 1971 1972
	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,
1973
				    false);
1974 1975
	if (IS_ERR(req))
		return PTR_ERR(req);
S
Sage Weil 已提交
1976 1977

	/* it may be a short read due to an object boundary */
1978 1979 1980 1981

	osd_data = &req->r_data_in;
	osd_data->type = CEPH_OSD_DATA_TYPE_PAGES;
	osd_data->pages = pages;
1982
	osd_data->length = *plen;
1983
	osd_data->alignment = page_align;
S
Sage Weil 已提交
1984

1985 1986
	dout("readpages  final extent is %llu~%llu (%llu bytes align %d)\n",
	     off, *plen, osd_data->length, page_align);
S
Sage Weil 已提交
1987 1988 1989 1990 1991 1992 1993 1994 1995

	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;
}
1996
EXPORT_SYMBOL(ceph_osdc_readpages);
S
Sage Weil 已提交
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

/*
 * 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,
2007
			 struct page **pages, int num_pages)
S
Sage Weil 已提交
2008 2009
{
	struct ceph_osd_request *req;
2010
	struct ceph_osd_data *osd_data;
S
Sage Weil 已提交
2011
	int rc = 0;
2012
	int page_align = off & ~PAGE_MASK;
S
Sage Weil 已提交
2013 2014 2015 2016

	BUG_ON(vino.snap != CEPH_NOSNAP);
	req = ceph_osdc_new_request(osdc, layout, vino, off, &len,
				    CEPH_OSD_OP_WRITE,
2017
				    CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE,
2018
				    snapc, 0,
S
Sage Weil 已提交
2019
				    truncate_seq, truncate_size, mtime,
2020
				    true);
2021 2022
	if (IS_ERR(req))
		return PTR_ERR(req);
S
Sage Weil 已提交
2023 2024

	/* it may be a short write due to an object boundary */
2025 2026 2027
	osd_data = &req->r_data_out;
	osd_data->type = CEPH_OSD_DATA_TYPE_PAGES;
	osd_data->pages = pages;
2028
	osd_data->length = len;
2029
	osd_data->alignment = page_align;
2030
	dout("writepages %llu~%llu (%llu bytes)\n", off, len, osd_data->length);
S
Sage Weil 已提交
2031

2032
	rc = ceph_osdc_start_request(osdc, req, true);
S
Sage Weil 已提交
2033 2034 2035 2036 2037 2038 2039 2040 2041
	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;
}
2042
EXPORT_SYMBOL(ceph_osdc_writepages);
S
Sage Weil 已提交
2043 2044 2045 2046 2047 2048 2049

/*
 * handle incoming message
 */
static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
{
	struct ceph_osd *osd = con->private;
2050
	struct ceph_osd_client *osdc;
S
Sage Weil 已提交
2051 2052 2053
	int type = le16_to_cpu(msg->hdr.type);

	if (!osd)
S
Sage Weil 已提交
2054
		goto out;
2055
	osdc = osd->o_osdc;
S
Sage Weil 已提交
2056 2057 2058 2059 2060 2061

	switch (type) {
	case CEPH_MSG_OSD_MAP:
		ceph_osdc_handle_map(osdc, msg);
		break;
	case CEPH_MSG_OSD_OPREPLY:
2062
		handle_reply(osdc, msg, con);
S
Sage Weil 已提交
2063
		break;
2064 2065 2066
	case CEPH_MSG_WATCH_NOTIFY:
		handle_watch_notify(osdc, msg);
		break;
S
Sage Weil 已提交
2067 2068 2069 2070 2071

	default:
		pr_err("received unknown message type %d %s\n", type,
		       ceph_msg_type_name(type));
	}
S
Sage Weil 已提交
2072
out:
S
Sage Weil 已提交
2073 2074 2075
	ceph_msg_put(msg);
}

2076
/*
2077 2078
 * lookup and return message for incoming reply.  set up reply message
 * pages.
2079 2080
 */
static struct ceph_msg *get_reply(struct ceph_connection *con,
2081 2082
				  struct ceph_msg_header *hdr,
				  int *skip)
S
Sage Weil 已提交
2083 2084 2085
{
	struct ceph_osd *osd = con->private;
	struct ceph_osd_client *osdc = osd->o_osdc;
2086
	struct ceph_msg *m;
2087
	struct ceph_osd_request *req;
2088 2089
	int front = le32_to_cpu(hdr->front_len);
	int data_len = le32_to_cpu(hdr->data_len);
2090
	u64 tid;
S
Sage Weil 已提交
2091

2092 2093 2094 2095 2096 2097
	tid = le64_to_cpu(hdr->tid);
	mutex_lock(&osdc->request_mutex);
	req = __lookup_request(osdc, tid);
	if (!req) {
		*skip = 1;
		m = NULL;
2098 2099
		dout("get_reply unknown tid %llu from osd%d\n", tid,
		     osd->o_osd);
2100 2101
		goto out;
	}
S
Sage Weil 已提交
2102 2103

	if (req->r_con_filling_msg) {
2104
		dout("%s revoking msg %p from old con %p\n", __func__,
S
Sage Weil 已提交
2105
		     req->r_reply, req->r_con_filling_msg);
2106
		ceph_msg_revoke_incoming(req->r_reply);
2107
		req->r_con_filling_msg->ops->put(req->r_con_filling_msg);
2108
		req->r_con_filling_msg = NULL;
2109 2110
	}

S
Sage Weil 已提交
2111 2112 2113
	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);
2114
		m = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, front, GFP_NOFS, false);
2115
		if (!m)
S
Sage Weil 已提交
2116 2117 2118 2119 2120 2121
			goto out;
		ceph_msg_put(req->r_reply);
		req->r_reply = m;
	}
	m = ceph_msg_get(req->r_reply);

2122
	if (data_len > 0) {
2123 2124 2125 2126
		struct ceph_osd_data *osd_data = &req->r_data_in;

		if (osd_data->type == CEPH_OSD_DATA_TYPE_PAGES) {
			if (osd_data->pages &&
2127
				unlikely(osd_data->length < data_len)) {
2128

2129 2130 2131
				pr_warning("tid %lld reply has %d bytes "
					"we had only %llu bytes ready\n",
					tid, data_len, osd_data->length);
2132 2133 2134 2135 2136 2137
				*skip = 1;
				ceph_msg_put(m);
				m = NULL;
				goto out;
			}
		}
2138
	}
2139
	*skip = 0;
2140
	req->r_con_filling_msg = con->ops->get(con);
S
Sage Weil 已提交
2141
	dout("get_reply tid %lld %p\n", tid, m);
2142 2143 2144

out:
	mutex_unlock(&osdc->request_mutex);
2145
	return m;
2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156

}

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
2157
	*skip = 0;
2158 2159
	switch (type) {
	case CEPH_MSG_OSD_MAP:
2160
	case CEPH_MSG_WATCH_NOTIFY:
2161
		return ceph_msg_new(type, front, GFP_NOFS, false);
2162 2163 2164 2165 2166 2167 2168 2169
	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
2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188
}

/*
 * 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);
}

2189 2190 2191
/*
 * authentication
 */
2192 2193 2194 2195 2196
/*
 * 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,
2197
					int *proto, int force_new)
2198 2199 2200 2201
{
	struct ceph_osd *o = con->private;
	struct ceph_osd_client *osdc = o->o_osdc;
	struct ceph_auth_client *ac = osdc->client->monc.auth;
2202
	struct ceph_auth_handshake *auth = &o->o_auth;
2203

2204
	if (force_new && auth->authorizer) {
2205 2206
		if (ac->ops && ac->ops->destroy_authorizer)
			ac->ops->destroy_authorizer(ac, auth->authorizer);
2207 2208
		auth->authorizer = NULL;
	}
2209
	if (!auth->authorizer && ac->ops && ac->ops->create_authorizer) {
2210 2211
		int ret = ac->ops->create_authorizer(ac, CEPH_ENTITY_TYPE_OSD,
							auth);
2212
		if (ret)
2213
			return ERR_PTR(ret);
2214 2215
	}
	*proto = ac->protocol;
2216

2217
	return auth;
2218 2219 2220 2221 2222 2223 2224 2225 2226
}


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;

2227 2228 2229 2230
	/*
	 * XXX If ac->ops or ac->ops->verify_authorizer_reply is null,
	 * XXX which do we do:  succeed or fail?
	 */
2231
	return ac->ops->verify_authorizer_reply(ac, o->o_auth.authorizer, len);
2232 2233
}

2234 2235 2236 2237 2238 2239
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;

2240
	if (ac->ops && ac->ops->invalidate_authorizer)
2241 2242 2243 2244
		ac->ops->invalidate_authorizer(ac, CEPH_ENTITY_TYPE_OSD);

	return ceph_monc_validate_auth(&osdc->client->monc);
}
2245

2246
static const struct ceph_connection_operations osd_con_ops = {
S
Sage Weil 已提交
2247 2248 2249
	.get = get_osd_con,
	.put = put_osd_con,
	.dispatch = dispatch,
2250 2251
	.get_authorizer = get_authorizer,
	.verify_authorizer_reply = verify_authorizer_reply,
2252
	.invalidate_authorizer = invalidate_authorizer,
S
Sage Weil 已提交
2253
	.alloc_msg = alloc_msg,
2254
	.fault = osd_reset,
S
Sage Weil 已提交
2255
};