osd_client.c 57.1 KB
Newer Older
1
#include <linux/ceph/ceph_debug.h>
S
Sage Weil 已提交
2

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

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

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

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

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,
S
Sage Weil 已提交
67
		       struct ceph_osd_request *req,
68
		       struct ceph_osd_req_op *op, u64 *bno)
S
Sage Weil 已提交
69
{
70 71 72
	u64 orig_len = *plen;
	u64 objoff = 0;
	u64 objlen = 0;
S
Sage Weil 已提交
73
	int r;
S
Sage Weil 已提交
74

75
	/* object extent? */
76
	r = ceph_calc_file_object_mapping(layout, off, orig_len, bno,
77
					  &objoff, &objlen);
S
Sage Weil 已提交
78 79
	if (r < 0)
		return r;
80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103
	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;
		}
	}
	req->r_num_pages = calc_pages_for(off, *plen);
	req->r_page_alignment = off & ~PAGE_MASK;
	if (op->op == CEPH_OSD_OP_WRITE)
		op->payload_len = *plen;

	dout("calc_layout bno=%llx %llu~%llu (%d pages)\n",
104
	     *bno, objoff, objlen, req->r_num_pages);
S
Sage Weil 已提交
105

106
	return 0;
S
Sage Weil 已提交
107 108 109 110 111
}

/*
 * requests
 */
S
Sage Weil 已提交
112
void ceph_osdc_release_request(struct kref *kref)
S
Sage Weil 已提交
113
{
S
Sage Weil 已提交
114 115 116 117 118 119
	struct ceph_osd_request *req = container_of(kref,
						    struct ceph_osd_request,
						    r_kref);

	if (req->r_request)
		ceph_msg_put(req->r_request);
120
	if (req->r_con_filling_msg) {
121 122
		dout("%s revoking msg %p from con %p\n", __func__,
		     req->r_reply, req->r_con_filling_msg);
123
		ceph_msg_revoke_incoming(req->r_reply);
124
		req->r_con_filling_msg->ops->put(req->r_con_filling_msg);
125
		req->r_con_filling_msg = NULL;
126
	}
127 128
	if (req->r_reply)
		ceph_msg_put(req->r_reply);
S
Sage Weil 已提交
129 130 131 132
	if (req->r_own_pages)
		ceph_release_page_vector(req->r_pages,
					 req->r_num_pages);
	ceph_put_snap_context(req->r_snapc);
133
	ceph_pagelist_release(&req->r_trail);
S
Sage Weil 已提交
134 135 136 137
	if (req->r_mempool)
		mempool_free(req, req->r_osdc->req_mempool);
	else
		kfree(req);
S
Sage Weil 已提交
138
}
139
EXPORT_SYMBOL(ceph_osdc_release_request);
140

141
struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc,
S
Sage Weil 已提交
142
					       struct ceph_snap_context *snapc,
143
					       unsigned int num_ops,
144
					       bool use_mempool,
145
					       gfp_t gfp_flags)
S
Sage Weil 已提交
146 147 148
{
	struct ceph_osd_request *req;
	struct ceph_msg *msg;
149 150 151 152 153 154 155 156 157 158 159
	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
160 161

	if (use_mempool) {
162
		req = mempool_alloc(osdc->req_mempool, gfp_flags);
S
Sage Weil 已提交
163 164
		memset(req, 0, sizeof(*req));
	} else {
165
		req = kzalloc(sizeof(*req), gfp_flags);
S
Sage Weil 已提交
166 167
	}
	if (req == NULL)
168
		return NULL;
S
Sage Weil 已提交
169 170 171

	req->r_osdc = osdc;
	req->r_mempool = use_mempool;
172

S
Sage Weil 已提交
173
	kref_init(&req->r_kref);
S
Sage Weil 已提交
174 175
	init_completion(&req->r_completion);
	init_completion(&req->r_safe_completion);
176
	RB_CLEAR_NODE(&req->r_node);
S
Sage Weil 已提交
177
	INIT_LIST_HEAD(&req->r_unsafe_item);
178 179
	INIT_LIST_HEAD(&req->r_linger_item);
	INIT_LIST_HEAD(&req->r_linger_osd);
180
	INIT_LIST_HEAD(&req->r_req_lru_item);
181 182
	INIT_LIST_HEAD(&req->r_osd_item);

S
Sage Weil 已提交
183 184 185 186 187
	/* create reply message */
	if (use_mempool)
		msg = ceph_msgpool_get(&osdc->msgpool_op_reply, 0);
	else
		msg = ceph_msg_new(CEPH_MSG_OSD_OPREPLY,
188
				   OSD_OPREPLY_FRONT_LEN, gfp_flags, true);
189
	if (!msg) {
S
Sage Weil 已提交
190
		ceph_osdc_put_request(req);
191
		return NULL;
S
Sage Weil 已提交
192 193 194
	}
	req->r_reply = msg;

195
	ceph_pagelist_init(&req->r_trail);
196

S
Sage Weil 已提交
197
	/* create request message; allow space for oid */
S
Sage Weil 已提交
198
	if (use_mempool)
199
		msg = ceph_msgpool_get(&osdc->msgpool_op, 0);
S
Sage Weil 已提交
200
	else
201
		msg = ceph_msg_new(CEPH_MSG_OSD_OP, msg_size, gfp_flags, true);
202
	if (!msg) {
S
Sage Weil 已提交
203
		ceph_osdc_put_request(req);
204
		return NULL;
S
Sage Weil 已提交
205
	}
206

S
Sage Weil 已提交
207
	memset(msg->front.iov_base, 0, msg->front.iov_len);
208 209 210 211 212

	req->r_request = msg;

	return req;
}
213
EXPORT_SYMBOL(ceph_osdc_alloc_request);
214

215 216 217 218 219 220
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);

221
	switch (src->op) {
A
Alex Elder 已提交
222 223
	case CEPH_OSD_OP_STAT:
		break;
224 225 226 227 228 229 230 231 232 233 234
	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;
235 236 237 238 239
	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);

240
		ceph_pagelist_append(&req->r_trail, src->cls.class_name,
241
				     src->cls.class_len);
242
		ceph_pagelist_append(&req->r_trail, src->cls.method_name,
243
				     src->cls.method_len);
244
		ceph_pagelist_append(&req->r_trail, src->cls.indata,
245 246
				     src->cls.indata_len);
		break;
247 248
	case CEPH_OSD_OP_STARTSYNC:
		break;
249 250 251 252 253 254
	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;
255 256 257 258
	default:
		pr_err("unrecognized osd opcode %d\n", dst->op);
		WARN_ON(1);
		break;
259 260 261
	case CEPH_OSD_OP_MAPEXT:
	case CEPH_OSD_OP_MASKTRUNC:
	case CEPH_OSD_OP_SPARSE_READ:
262
	case CEPH_OSD_OP_NOTIFY:
263 264 265 266 267 268 269 270 271 272 273 274
	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:
275
	case CEPH_OSD_OP_ROLLBACK:
276 277 278 279 280 281 282 283 284 285 286 287 288
	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:
289
	case CEPH_OSD_OP_GETXATTR:
290
	case CEPH_OSD_OP_GETXATTRS:
291 292
	case CEPH_OSD_OP_CMPXATTR:
	case CEPH_OSD_OP_SETXATTR:
293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316
	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",
			ceph_osd_op_name(dst->op));
		WARN_ON(1);
		break;
317 318 319 320
	}
	dst->payload_len = cpu_to_le32(src->payload_len);
}

321 322 323 324 325
/*
 * build new request AND message
 *
 */
void ceph_osdc_build_request(struct ceph_osd_request *req,
326
			     u64 off, u64 len, unsigned int num_ops,
327
			     struct ceph_osd_req_op *src_ops,
328
			     struct ceph_snap_context *snapc, u64 snap_id,
329
			     struct timespec *mtime)
330 331
{
	struct ceph_msg *msg = req->r_request;
332
	struct ceph_osd_req_op *src_op;
333
	void *p;
334
	size_t msg_size;
335
	int flags = req->r_flags;
336
	u64 data_len;
337
	int i;
338

339 340
	req->r_num_ops = num_ops;
	req->r_snapid = snap_id;
S
Sage Weil 已提交
341 342
	req->r_snapc = ceph_get_snap_context(snapc);

343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373
	/* 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);
374
	memcpy(p, req->r_oid, req->r_oid_len);
375
	dout("oid '%.*s' len %d\n", req->r_oid_len, req->r_oid, req->r_oid_len);
376
	p += req->r_oid_len;
S
Sage Weil 已提交
377

378 379
	/* ops */
	ceph_encode_16(&p, num_ops);
380
	src_op = src_ops;
381 382 383 384 385
	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);
	}
386

387 388 389 390 391
	/* 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
392
		for (i = 0; i < snapc->num_snaps; i++) {
393
			ceph_encode_64(&p, req->r_snapc->snaps[i]);
S
Sage Weil 已提交
394 395 396
		}
	}

397 398 399
	req->r_request_attempts = p;
	p += 4;

400
	data_len = req->r_trail.length;
401 402
	if (flags & CEPH_OSD_FLAG_WRITE) {
		req->r_request->hdr.data_off = cpu_to_le16(off);
403
		data_len += len;
404
	}
405
	req->r_request->hdr.data_len = cpu_to_le32(data_len);
406 407
	req->r_request->page_alignment = req->r_page_alignment;

S
Sage Weil 已提交
408
	BUG_ON(p > msg->front.iov_base + msg->front.iov_len);
409 410 411
	msg_size = p - msg->front.iov_base;
	msg->front.iov_len = msg_size;
	msg->hdr.front_len = cpu_to_le32(msg_size);
412 413 414

	dout("build_request msg_size was %d num_ops %d\n", (int)msg_size,
	     num_ops);
415 416
	return;
}
417
EXPORT_SYMBOL(ceph_osdc_build_request);
418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439

/*
 * 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,
440
					       bool use_mempool,
441
					       int page_align)
442
{
A
Alex Elder 已提交
443
	struct ceph_osd_req_op ops[2];
444
	struct ceph_osd_request *req;
A
Alex Elder 已提交
445
	unsigned int num_op = 1;
446
	u64 bno = 0;
447
	int r;
448

A
Alex Elder 已提交
449 450
	memset(&ops, 0, sizeof ops);

451 452 453 454 455 456
	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
457 458
		num_op++;
	}
459

A
Alex Elder 已提交
460 461
	req = ceph_osdc_alloc_request(osdc, snapc, num_op, use_mempool,
					GFP_NOFS);
462
	if (!req)
463
		return ERR_PTR(-ENOMEM);
464
	req->r_flags = flags;
465 466

	/* calculate max write size */
467
	r = calc_layout(layout, off, plen, req, ops, &bno);
468 469
	if (r < 0) {
		ceph_osdc_put_request(req);
470
		return ERR_PTR(r);
471
	}
472

473 474
	req->r_file_layout = *layout;  /* keep a copy */

475 476 477
	snprintf(req->r_oid, sizeof(req->r_oid), "%llx.%08llx", vino.ino, bno);
	req->r_oid_len = strlen(req->r_oid);

478 479 480
	/* in case it differs from natural (file) alignment that
	   calc_layout filled in for us */
	req->r_num_pages = calc_pages_for(page_align, *plen);
481 482
	req->r_page_alignment = page_align;

A
Alex Elder 已提交
483 484
	ceph_osdc_build_request(req, off, *plen, num_op, ops,
				snapc, vino.snap, mtime);
485

S
Sage Weil 已提交
486 487
	return req;
}
488
EXPORT_SYMBOL(ceph_osdc_new_request);
S
Sage Weil 已提交
489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554

/*
 * 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;
}

555 556 557 558 559 560
/*
 * Resubmit requests pending on the given osd.
 */
static void __kick_osd_requests(struct ceph_osd_client *osdc,
				struct ceph_osd *osd)
{
561
	struct ceph_osd_request *req, *nreq;
562 563 564 565
	int err;

	dout("__kick_osd_requests osd%d\n", osd->o_osd);
	err = __reset_osd(osdc, osd);
566
	if (err)
567 568 569 570 571 572
		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);
573 574 575 576 577 578
		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) {
579 580 581 582 583
		/*
		 * reregister request prior to unregistering linger so
		 * that r_osd is preserved.
		 */
		BUG_ON(!list_empty(&req->r_req_lru_item));
584
		__register_request(osdc, req);
585 586 587
		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);
588 589
		dout("requeued lingering %p tid %llu osd%d\n", req, req->r_tid,
		     osd->o_osd);
590 591 592
	}
}

S
Sage Weil 已提交
593
/*
594
 * If the osd connection drops, we need to resubmit all requests.
S
Sage Weil 已提交
595 596 597 598 599 600 601 602 603 604 605
 */
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
606 607
	mutex_lock(&osdc->request_mutex);
	__kick_osd_requests(osdc, osd);
A
Alex Elder 已提交
608
	__send_queued(osdc);
S
Sage Weil 已提交
609
	mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
610 611 612 613 614 615
	up_read(&osdc->map_sem);
}

/*
 * Track open sessions with osds.
 */
616
static struct ceph_osd *create_osd(struct ceph_osd_client *osdc, int onum)
S
Sage Weil 已提交
617 618 619 620 621 622 623 624 625
{
	struct ceph_osd *osd;

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

	atomic_set(&osd->o_ref, 1);
	osd->o_osdc = osdc;
626
	osd->o_osd = onum;
627
	RB_CLEAR_NODE(&osd->o_node);
S
Sage Weil 已提交
628
	INIT_LIST_HEAD(&osd->o_requests);
629
	INIT_LIST_HEAD(&osd->o_linger_requests);
630
	INIT_LIST_HEAD(&osd->o_osd_lru);
S
Sage Weil 已提交
631 632
	osd->o_incarnation = 1;

633
	ceph_con_init(&osd->o_con, osd, &osd_con_ops, &osdc->client->msgr);
634

635
	INIT_LIST_HEAD(&osd->o_keepalive_item);
S
Sage Weil 已提交
636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654
	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);
655
	if (atomic_dec_and_test(&osd->o_ref) && osd->o_auth.authorizer) {
S
Sage Weil 已提交
656 657
		struct ceph_auth_client *ac = osd->o_osdc->client->monc.auth;

658
		if (ac->ops && ac->ops->destroy_authorizer)
659
			ac->ops->destroy_authorizer(ac, osd->o_auth.authorizer);
S
Sage Weil 已提交
660
		kfree(osd);
S
Sage Weil 已提交
661
	}
S
Sage Weil 已提交
662 663 664 665 666
}

/*
 * remove an osd from our map
 */
667
static void __remove_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
S
Sage Weil 已提交
668
{
669
	dout("__remove_osd %p\n", osd);
S
Sage Weil 已提交
670 671
	BUG_ON(!list_empty(&osd->o_requests));
	rb_erase(&osd->o_node, &osdc->osds);
672
	list_del_init(&osd->o_osd_lru);
S
Sage Weil 已提交
673 674 675 676
	ceph_con_close(&osd->o_con);
	put_osd(osd);
}

677 678
static void remove_all_osds(struct ceph_osd_client *osdc)
{
679
	dout("%s %p\n", __func__, osdc);
680 681 682 683 684 685 686 687 688
	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);
}

689 690 691 692 693 694
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);
695
	osd->lru_ttl = jiffies + osdc->client->options->osd_idle_ttl * HZ;
696 697 698 699 700 701 702 703 704
}

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);
}

705
static void remove_old_osds(struct ceph_osd_client *osdc)
706 707 708 709 710 711
{
	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) {
712
		if (time_before(jiffies, osd->lru_ttl))
713 714 715 716 717 718
			break;
		__remove_osd(osdc, osd);
	}
	mutex_unlock(&osdc->request_mutex);
}

S
Sage Weil 已提交
719 720 721
/*
 * reset osd connect
 */
722
static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
S
Sage Weil 已提交
723
{
A
Alex Elder 已提交
724
	struct ceph_entity_addr *peer_addr;
S
Sage Weil 已提交
725

726
	dout("__reset_osd %p osd%d\n", osd, osd->o_osd);
727 728
	if (list_empty(&osd->o_requests) &&
	    list_empty(&osd->o_linger_requests)) {
729
		__remove_osd(osdc, osd);
A
Alex Elder 已提交
730 731 732 733 734 735 736 737 738

		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;

739 740 741 742 743
		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
744 745

		return -EAGAIN;
S
Sage Weil 已提交
746
	}
A
Alex Elder 已提交
747 748 749 750 751 752

	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
753 754 755 756 757 758 759 760
}

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;

761
	dout("__insert_osd %p osd%d\n", new, new->o_osd);
S
Sage Weil 已提交
762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793
	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;
}

794 795 796
static void __schedule_osd_timeout(struct ceph_osd_client *osdc)
{
	schedule_delayed_work(&osdc->timeout_work,
797
			osdc->client->options->osd_keepalive_timeout * HZ);
798 799 800 801 802 803
}

static void __cancel_osd_timeout(struct ceph_osd_client *osdc)
{
	cancel_delayed_work(&osdc->timeout_work);
}
S
Sage Weil 已提交
804 805 806 807 808

/*
 * Register request, assign tid.  If this is the first request, set up
 * the timeout event.
 */
809 810
static void __register_request(struct ceph_osd_client *osdc,
			       struct ceph_osd_request *req)
S
Sage Weil 已提交
811 812
{
	req->r_tid = ++osdc->last_tid;
813
	req->r_request->hdr.tid = cpu_to_le64(req->r_tid);
814
	dout("__register_request %p tid %lld\n", req, req->r_tid);
S
Sage Weil 已提交
815 816 817 818
	__insert_request(osdc, req);
	ceph_osdc_get_request(req);
	osdc->num_requests++;
	if (osdc->num_requests == 1) {
819 820
		dout(" first request, scheduling timeout\n");
		__schedule_osd_timeout(osdc);
S
Sage Weil 已提交
821
	}
822 823 824 825 826 827 828
}

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
829 830 831 832 833 834 835 836 837
	mutex_unlock(&osdc->request_mutex);
}

/*
 * called under osdc->request_mutex
 */
static void __unregister_request(struct ceph_osd_client *osdc,
				 struct ceph_osd_request *req)
{
838 839 840 841 842 843
	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
844 845 846 847
	dout("__unregister_request %p tid %lld\n", req, req->r_tid);
	rb_erase(&req->r_node, &osdc->requests);
	osdc->num_requests--;

848 849
	if (req->r_osd) {
		/* make sure the original request isn't in flight. */
850
		ceph_msg_revoke(req->r_request);
851 852

		list_del_init(&req->r_osd_item);
853 854 855
		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);
856
			__move_osd_to_lru(osdc, req->r_osd);
857
		}
858
		if (list_empty(&req->r_linger_item))
859
			req->r_osd = NULL;
860
	}
S
Sage Weil 已提交
861

A
Alex Elder 已提交
862
	list_del_init(&req->r_req_lru_item);
S
Sage Weil 已提交
863 864
	ceph_osdc_put_request(req);

865 866 867
	if (osdc->num_requests == 0) {
		dout(" no requests, canceling timeout\n");
		__cancel_osd_timeout(osdc);
S
Sage Weil 已提交
868 869 870 871 872 873 874 875
	}
}

/*
 * Cancel a previously queued request message
 */
static void __cancel_request(struct ceph_osd_request *req)
{
876
	if (req->r_sent && req->r_osd) {
877
		ceph_msg_revoke(req->r_request);
S
Sage Weil 已提交
878 879 880 881
		req->r_sent = 0;
	}
}

882 883 884 885 886
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);
887 888 889
	if (req->r_osd)
		list_add_tail(&req->r_linger_osd,
			      &req->r_osd->o_linger_requests);
890 891 892 893 894 895
}

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
896
	list_del_init(&req->r_linger_item);
897 898 899 900 901 902 903 904
	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);
		}
905 906
		if (list_empty(&req->r_osd_item))
			req->r_osd = NULL;
907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936
	}
}

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
937 938 939
/*
 * 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
940
 * no up osd, set r_osd to NULL.  Move the request to the appropriate list
941
 * (unsent, homeless) or leave on in-flight lru.
S
Sage Weil 已提交
942 943 944 945 946
 *
 * Return 0 if unchanged, 1 if changed, or negative on error.
 *
 * Caller should hold map_sem for read and request_mutex.
 */
947
static int __map_request(struct ceph_osd_client *osdc,
948
			 struct ceph_osd_request *req, int force_resend)
S
Sage Weil 已提交
949
{
950
	struct ceph_pg pgid;
951 952
	int acting[CEPH_PG_MAX_SIZE];
	int o = -1, num = 0;
S
Sage Weil 已提交
953 954
	int err;

955
	dout("map_request %p tid %lld\n", req, req->r_tid);
956
	err = ceph_calc_object_layout(&pgid, req->r_oid,
S
Sage Weil 已提交
957
				      &req->r_file_layout, osdc->osdmap);
958 959
	if (err) {
		list_move(&req->r_req_lru_item, &osdc->req_notarget);
S
Sage Weil 已提交
960
		return err;
961
	}
962 963
	req->r_pgid = pgid;

964 965 966 967 968
	err = ceph_calc_pg_acting(osdc->osdmap, pgid, acting);
	if (err > 0) {
		o = acting[0];
		num = err;
	}
S
Sage Weil 已提交
969

970 971
	if ((!force_resend &&
	     req->r_osd && req->r_osd->o_osd == o &&
972 973 974
	     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
975 976 977
	    (req->r_osd == NULL && o == -1))
		return 0;  /* no change */

978 979
	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
980 981
	     req->r_osd ? req->r_osd->o_osd : -1);

982 983 984 985
	/* 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
986 987 988 989 990 991 992 993
	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) {
994
		err = -ENOMEM;
995
		req->r_osd = create_osd(osdc, o);
996 997
		if (!req->r_osd) {
			list_move(&req->r_req_lru_item, &osdc->req_notarget);
998
			goto out;
999
		}
S
Sage Weil 已提交
1000

1001
		dout("map_request osd %p is osd%d\n", req->r_osd, o);
S
Sage Weil 已提交
1002 1003
		__insert_osd(osdc, req->r_osd);

1004 1005 1006
		ceph_con_open(&req->r_osd->o_con,
			      CEPH_ENTITY_TYPE_OSD, o,
			      &osdc->osdmap->osd_addr[o]);
S
Sage Weil 已提交
1007 1008
	}

1009 1010
	if (req->r_osd) {
		__remove_osd_from_lru(req->r_osd);
S
Sage Weil 已提交
1011
		list_add(&req->r_osd_item, &req->r_osd->o_requests);
1012 1013 1014
		list_move(&req->r_req_lru_item, &osdc->req_unsent);
	} else {
		list_move(&req->r_req_lru_item, &osdc->req_notarget);
1015
	}
1016
	err = 1;   /* osd or pg changed */
S
Sage Weil 已提交
1017 1018 1019 1020 1021 1022 1023 1024

out:
	return err;
}

/*
 * caller should hold map_sem (for read) and request_mutex
 */
1025 1026
static void __send_request(struct ceph_osd_client *osdc,
			   struct ceph_osd_request *req)
S
Sage Weil 已提交
1027
{
1028
	void *p;
S
Sage Weil 已提交
1029

1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043
	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));
1044

S
Sage Weil 已提交
1045
	req->r_stamp = jiffies;
1046
	list_move_tail(&req->r_req_lru_item, &osdc->req_lru);
S
Sage Weil 已提交
1047 1048 1049 1050 1051 1052

	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;
}

1053 1054 1055
/*
 * Send any requests in the queue (req_unsent).
 */
A
Alex Elder 已提交
1056
static void __send_queued(struct ceph_osd_client *osdc)
1057 1058 1059
{
	struct ceph_osd_request *req, *tmp;

A
Alex Elder 已提交
1060 1061
	dout("__send_queued\n");
	list_for_each_entry_safe(req, tmp, &osdc->req_unsent, r_req_lru_item)
1062 1063 1064
		__send_request(osdc, req);
}

S
Sage Weil 已提交
1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
/*
 * 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
1078
	struct ceph_osd_request *req;
S
Sage Weil 已提交
1079
	struct ceph_osd *osd;
1080
	unsigned long keepalive =
1081
		osdc->client->options->osd_keepalive_timeout * HZ;
1082
	struct list_head slow_osds;
S
Sage Weil 已提交
1083 1084 1085 1086 1087 1088 1089
	dout("timeout\n");
	down_read(&osdc->map_sem);

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

	mutex_lock(&osdc->request_mutex);

1090 1091 1092 1093 1094 1095 1096
	/*
	 * 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
1097
		if (time_before(jiffies, req->r_stamp + keepalive))
1098 1099 1100 1101 1102
			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
1103
		     req->r_tid, osd->o_osd);
1104 1105 1106 1107 1108 1109
		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
1110 1111 1112
		ceph_con_keepalive(&osd->o_con);
	}

1113
	__schedule_osd_timeout(osdc);
A
Alex Elder 已提交
1114
	__send_queued(osdc);
S
Sage Weil 已提交
1115 1116 1117 1118
	mutex_unlock(&osdc->request_mutex);
	up_read(&osdc->map_sem);
}

1119 1120 1121 1122 1123 1124
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 =
1125
		osdc->client->options->osd_idle_ttl * HZ >> 2;
1126 1127 1128

	dout("osds timeout\n");
	down_read(&osdc->map_sem);
1129
	remove_old_osds(osdc);
1130 1131 1132 1133 1134 1135
	up_read(&osdc->map_sem);

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

S
Sage Weil 已提交
1136 1137 1138 1139 1140 1141 1142
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 */
}

1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162
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
1163 1164 1165 1166
/*
 * handle osd op reply.  either call the callback if it is specified,
 * or do the completion to wake up the waiting thread.
 */
1167 1168
static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
			 struct ceph_connection *con)
S
Sage Weil 已提交
1169
{
1170
	void *p, *end;
S
Sage Weil 已提交
1171 1172
	struct ceph_osd_request *req;
	u64 tid;
1173 1174
	int object_len;
	int numops, payload_len, flags;
1175
	s32 result;
1176 1177 1178 1179 1180 1181 1182
	s32 retry_attempt;
	struct ceph_pg pg;
	int err;
	u32 reassert_epoch;
	u64 reassert_version;
	u32 osdmap_epoch;
	int i;
S
Sage Weil 已提交
1183

1184
	tid = le64_to_cpu(msg->hdr.tid);
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196
	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
1197
		goto bad;
1198 1199 1200 1201 1202 1203 1204 1205

	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
1206 1207 1208 1209 1210 1211 1212 1213 1214
	/* 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);
1215 1216 1217 1218 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

	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
1247

1248
	/*
1249
	 * if this connection filled our message, drop our reference now, to
1250 1251
	 * avoid a (safe but slower) revoke later.
	 */
1252
	if (req->r_con_filling_msg == con && req->r_reply == msg) {
S
Sage Weil 已提交
1253
		dout(" dropping con_filling_msg ref %p\n", con);
1254
		req->r_con_filling_msg = NULL;
1255
		con->ops->put(con);
1256 1257
	}

S
Sage Weil 已提交
1258
	if (!req->r_got_reply) {
1259
		unsigned int bytes;
S
Sage Weil 已提交
1260

1261
		req->r_result = result;
S
Sage Weil 已提交
1262 1263 1264 1265 1266 1267 1268
		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, */
1269 1270
		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
1271 1272 1273 1274

		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
1275
		mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1276 1277 1278 1279 1280
		goto done;
	}

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

1281 1282 1283
	if (req->r_linger && (flags & CEPH_OSD_FLAG_ONDISK))
		__register_linger_request(osdc, req);

S
Sage Weil 已提交
1284
	/* either this is a read, or we got the safe response */
1285 1286
	if (result < 0 ||
	    (flags & CEPH_OSD_FLAG_ONDISK) ||
S
Sage Weil 已提交
1287 1288 1289 1290 1291 1292 1293 1294
	    ((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
1295
		complete_all(&req->r_completion);
S
Sage Weil 已提交
1296

S
Sage Weil 已提交
1297 1298
	if (flags & CEPH_OSD_FLAG_ONDISK)
		complete_request(req);
S
Sage Weil 已提交
1299 1300

done:
1301
	dout("req=%p req->r_linger=%d\n", req, req->r_linger);
S
Sage Weil 已提交
1302 1303 1304
	ceph_osdc_put_request(req);
	return;

1305 1306
bad_put:
	ceph_osdc_put_request(req);
S
Sage Weil 已提交
1307
bad:
1308 1309
	pr_err("corrupt osd_op_reply got %d %d\n",
	       (int)msg->front.iov_len, le32_to_cpu(msg->hdr.front_len));
1310
	ceph_msg_dump(msg);
S
Sage Weil 已提交
1311 1312
}

1313
static void reset_changed_osds(struct ceph_osd_client *osdc)
S
Sage Weil 已提交
1314 1315 1316
{
	struct rb_node *p, *n;

1317 1318
	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
1319

1320 1321 1322 1323 1324 1325 1326
		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
1327
	}
1328 1329 1330
}

/*
1331 1332
 * Requeue requests whose mapping to an OSD has changed.  If requests map to
 * no osd, request a new map.
1333
 *
1334
 * Caller should hold map_sem for read.
1335
 */
1336
static void kick_requests(struct ceph_osd_client *osdc, int force_resend)
1337
{
1338
	struct ceph_osd_request *req, *nreq;
1339 1340 1341
	struct rb_node *p;
	int needmap = 0;
	int err;
1342

1343
	dout("kick_requests %s\n", force_resend ? " (force resend)" : "");
1344
	mutex_lock(&osdc->request_mutex);
1345
	for (p = rb_first(&osdc->requests); p; ) {
1346
		req = rb_entry(p, struct ceph_osd_request, r_node);
1347
		p = rb_next(p);
1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365

		/*
		 * 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;
		}

1366
		err = __map_request(osdc, req, force_resend);
1367 1368 1369 1370 1371 1372
		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) {
1373 1374 1375 1376
			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);
1377
				req->r_flags |= CEPH_OSD_FLAG_RETRY;
1378 1379
			}
		}
1380 1381 1382 1383 1384 1385
	}

	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);

1386
		err = __map_request(osdc, req, force_resend);
1387
		dout("__map_request returned %d\n", err);
1388 1389 1390 1391 1392 1393 1394 1395
		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;
1396
		}
1397 1398 1399 1400

		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);
1401
		__unregister_linger_request(osdc, req);
1402
	}
S
Sage Weil 已提交
1403 1404 1405 1406 1407 1408
	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);
	}
1409
	reset_changed_osds(osdc);
1410
}
1411 1412


S
Sage Weil 已提交
1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435
/*
 * 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));
1436 1437
	if (ceph_check_fsid(osdc->client, &fsid) < 0)
		return;
S
Sage Weil 已提交
1438 1439 1440 1441 1442 1443 1444 1445

	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);
1446 1447
		epoch = ceph_decode_32(&p);
		maplen = ceph_decode_32(&p);
S
Sage Weil 已提交
1448 1449 1450 1451 1452 1453 1454
		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,
1455
							  &osdc->client->msgr);
S
Sage Weil 已提交
1456 1457 1458 1459
			if (IS_ERR(newmap)) {
				err = PTR_ERR(newmap);
				goto bad;
			}
1460
			BUG_ON(!newmap);
S
Sage Weil 已提交
1461 1462 1463 1464
			if (newmap != osdc->osdmap) {
				ceph_osdmap_destroy(osdc->osdmap);
				osdc->osdmap = newmap;
			}
1465
			kick_requests(osdc, 0);
S
Sage Weil 已提交
1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480
		} 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);
1481 1482
		epoch = ceph_decode_32(&p);
		maplen = ceph_decode_32(&p);
S
Sage Weil 已提交
1483 1484 1485 1486 1487 1488 1489 1490 1491
		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 {
1492 1493
			int skipped_map = 0;

S
Sage Weil 已提交
1494 1495 1496 1497 1498 1499
			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;
			}
1500
			BUG_ON(!newmap);
S
Sage Weil 已提交
1501 1502
			oldmap = osdc->osdmap;
			osdc->osdmap = newmap;
1503 1504 1505
			if (oldmap) {
				if (oldmap->epoch + 1 < newmap->epoch)
					skipped_map = 1;
S
Sage Weil 已提交
1506
				ceph_osdmap_destroy(oldmap);
1507 1508
			}
			kick_requests(osdc, skipped_map);
S
Sage Weil 已提交
1509 1510 1511 1512 1513 1514 1515 1516
		}
		p += maplen;
		nr_maps--;
	}

done:
	downgrade_write(&osdc->map_sem);
	ceph_monc_got_osdmap(&osdc->client->monc, osdc->osdmap->epoch);
1517 1518 1519 1520 1521 1522 1523 1524 1525

	/*
	 * 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
1526 1527 1528
	mutex_lock(&osdc->request_mutex);
	__send_queued(osdc);
	mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1529
	up_read(&osdc->map_sem);
1530
	wake_up_all(&osdc->client->auth_wq);
S
Sage Weil 已提交
1531 1532 1533 1534
	return;

bad:
	pr_err("osdc handle_map corrupt msg\n");
1535
	ceph_msg_dump(msg);
S
Sage Weil 已提交
1536 1537 1538 1539
	up_write(&osdc->map_sem);
	return;
}

1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621
/*
 * 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 *),
1622
			   void *data, struct ceph_osd_event **pevent)
1623 1624 1625 1626 1627 1628 1629 1630 1631
{
	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;
1632
	event->one_shot = 0;
1633 1634 1635
	event->data = data;
	event->osdc = osdc;
	INIT_LIST_HEAD(&event->osd_node);
1636
	RB_CLEAR_NODE(&event->node);
1637 1638 1639 1640 1641 1642 1643 1644 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
	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
 */
1683 1684
static void handle_watch_notify(struct ceph_osd_client *osdc,
				struct ceph_msg *msg)
1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704
{
	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) {
1705
		BUG_ON(event->one_shot);
1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716
		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;
		}
1717
		INIT_WORK(&event_work->work, do_event_work);
1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738
		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;
}

S
Sage Weil 已提交
1739 1740 1741 1742 1743 1744 1745
/*
 * 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
1746
	int rc = 0;
S
Sage Weil 已提交
1747 1748 1749

	req->r_request->pages = req->r_pages;
	req->r_request->nr_pages = req->r_num_pages;
1750 1751 1752
#ifdef CONFIG_BLOCK
	req->r_request->bio = req->r_bio;
#endif
1753
	req->r_request->trail = &req->r_trail;
S
Sage Weil 已提交
1754 1755 1756 1757 1758

	register_request(osdc, req);

	down_read(&osdc->map_sem);
	mutex_lock(&osdc->request_mutex);
S
Sage Weil 已提交
1759 1760 1761 1762 1763 1764
	/*
	 * 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) {
1765
		rc = __map_request(osdc, req, 0);
1766 1767 1768 1769 1770 1771
		if (rc < 0) {
			if (nofail) {
				dout("osdc_start_request failed map, "
				     " will retry %lld\n", req->r_tid);
				rc = 0;
			}
1772
			goto out_unlock;
1773
		}
1774 1775 1776 1777
		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 {
1778
			__send_request(osdc, req);
S
Sage Weil 已提交
1779
		}
1780
		rc = 0;
S
Sage Weil 已提交
1781
	}
1782 1783

out_unlock:
S
Sage Weil 已提交
1784 1785 1786 1787
	mutex_unlock(&osdc->request_mutex);
	up_read(&osdc->map_sem);
	return rc;
}
1788
EXPORT_SYMBOL(ceph_osdc_start_request);
S
Sage Weil 已提交
1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801

/*
 * 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);
1802
		__unregister_request(osdc, req);
S
Sage Weil 已提交
1803
		mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1804
		complete_request(req);
1805
		dout("wait_request tid %llu canceled/timed out\n", req->r_tid);
S
Sage Weil 已提交
1806 1807 1808 1809 1810 1811
		return rc;
	}

	dout("wait_request tid %llu result %d\n", req->r_tid, req->r_result);
	return req->r_result;
}
1812
EXPORT_SYMBOL(ceph_osdc_wait_request);
S
Sage Weil 已提交
1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845

/*
 * 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);
}
1846
EXPORT_SYMBOL(ceph_osdc_sync);
S
Sage Weil 已提交
1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863

/*
 * 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;
1864
	INIT_LIST_HEAD(&osdc->osd_lru);
S
Sage Weil 已提交
1865
	osdc->requests = RB_ROOT;
1866
	INIT_LIST_HEAD(&osdc->req_lru);
1867 1868
	INIT_LIST_HEAD(&osdc->req_unsent);
	INIT_LIST_HEAD(&osdc->req_notarget);
1869
	INIT_LIST_HEAD(&osdc->req_linger);
S
Sage Weil 已提交
1870 1871
	osdc->num_requests = 0;
	INIT_DELAYED_WORK(&osdc->timeout_work, handle_timeout);
1872
	INIT_DELAYED_WORK(&osdc->osds_timeout_work, handle_osds_timeout);
1873 1874 1875
	spin_lock_init(&osdc->event_lock);
	osdc->event_tree = RB_ROOT;
	osdc->event_count = 0;
1876 1877

	schedule_delayed_work(&osdc->osds_timeout_work,
1878
	   round_jiffies_relative(osdc->client->options->osd_idle_ttl * HZ));
S
Sage Weil 已提交
1879

1880
	err = -ENOMEM;
S
Sage Weil 已提交
1881 1882 1883
	osdc->req_mempool = mempool_create_kmalloc_pool(10,
					sizeof(struct ceph_osd_request));
	if (!osdc->req_mempool)
1884
		goto out;
S
Sage Weil 已提交
1885

1886 1887
	err = ceph_msgpool_init(&osdc->msgpool_op, CEPH_MSG_OSD_OP,
				OSD_OP_FRONT_LEN, 10, true,
1888
				"osd_op");
S
Sage Weil 已提交
1889
	if (err < 0)
1890
		goto out_mempool;
1891
	err = ceph_msgpool_init(&osdc->msgpool_op_reply, CEPH_MSG_OSD_OPREPLY,
1892 1893
				OSD_OPREPLY_FRONT_LEN, 10, true,
				"osd_op_reply");
S
Sage Weil 已提交
1894 1895
	if (err < 0)
		goto out_msgpool;
1896 1897 1898 1899 1900 1901 1902

	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
1903
	return 0;
1904

S
Sage Weil 已提交
1905 1906
out_msgpool:
	ceph_msgpool_destroy(&osdc->msgpool_op);
1907 1908 1909 1910
out_mempool:
	mempool_destroy(osdc->req_mempool);
out:
	return err;
S
Sage Weil 已提交
1911 1912 1913 1914
}

void ceph_osdc_stop(struct ceph_osd_client *osdc)
{
1915 1916
	flush_workqueue(osdc->notify_wq);
	destroy_workqueue(osdc->notify_wq);
S
Sage Weil 已提交
1917
	cancel_delayed_work_sync(&osdc->timeout_work);
1918
	cancel_delayed_work_sync(&osdc->osds_timeout_work);
S
Sage Weil 已提交
1919 1920 1921 1922
	if (osdc->osdmap) {
		ceph_osdmap_destroy(osdc->osdmap);
		osdc->osdmap = NULL;
	}
1923
	remove_all_osds(osdc);
S
Sage Weil 已提交
1924 1925
	mempool_destroy(osdc->req_mempool);
	ceph_msgpool_destroy(&osdc->msgpool_op);
S
Sage Weil 已提交
1926
	ceph_msgpool_destroy(&osdc->msgpool_op_reply);
S
Sage Weil 已提交
1927 1928 1929 1930 1931 1932 1933 1934 1935 1936
}

/*
 * 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,
1937
			struct page **pages, int num_pages, int page_align)
S
Sage Weil 已提交
1938 1939 1940 1941 1942 1943 1944 1945 1946
{
	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,
1947
				    false, page_align);
1948 1949
	if (IS_ERR(req))
		return PTR_ERR(req);
S
Sage Weil 已提交
1950 1951 1952 1953

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

1954 1955
	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
1956 1957 1958 1959 1960 1961 1962 1963 1964

	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;
}
1965
EXPORT_SYMBOL(ceph_osdc_readpages);
S
Sage Weil 已提交
1966 1967 1968 1969 1970 1971 1972 1973 1974 1975

/*
 * 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,
1976
			 struct page **pages, int num_pages)
S
Sage Weil 已提交
1977 1978 1979
{
	struct ceph_osd_request *req;
	int rc = 0;
1980
	int page_align = off & ~PAGE_MASK;
S
Sage Weil 已提交
1981 1982 1983 1984

	BUG_ON(vino.snap != CEPH_NOSNAP);
	req = ceph_osdc_new_request(osdc, layout, vino, off, &len,
				    CEPH_OSD_OP_WRITE,
1985
				    CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE,
1986
				    snapc, 0,
S
Sage Weil 已提交
1987
				    truncate_seq, truncate_size, mtime,
1988
				    true, page_align);
1989 1990
	if (IS_ERR(req))
		return PTR_ERR(req);
S
Sage Weil 已提交
1991 1992 1993 1994 1995 1996

	/* 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);

1997
	rc = ceph_osdc_start_request(osdc, req, true);
S
Sage Weil 已提交
1998 1999 2000 2001 2002 2003 2004 2005 2006
	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;
}
2007
EXPORT_SYMBOL(ceph_osdc_writepages);
S
Sage Weil 已提交
2008 2009 2010 2011 2012 2013 2014

/*
 * handle incoming message
 */
static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
{
	struct ceph_osd *osd = con->private;
2015
	struct ceph_osd_client *osdc;
S
Sage Weil 已提交
2016 2017 2018
	int type = le16_to_cpu(msg->hdr.type);

	if (!osd)
S
Sage Weil 已提交
2019
		goto out;
2020
	osdc = osd->o_osdc;
S
Sage Weil 已提交
2021 2022 2023 2024 2025 2026

	switch (type) {
	case CEPH_MSG_OSD_MAP:
		ceph_osdc_handle_map(osdc, msg);
		break;
	case CEPH_MSG_OSD_OPREPLY:
2027
		handle_reply(osdc, msg, con);
S
Sage Weil 已提交
2028
		break;
2029 2030 2031
	case CEPH_MSG_WATCH_NOTIFY:
		handle_watch_notify(osdc, msg);
		break;
S
Sage Weil 已提交
2032 2033 2034 2035 2036

	default:
		pr_err("received unknown message type %d %s\n", type,
		       ceph_msg_type_name(type));
	}
S
Sage Weil 已提交
2037
out:
S
Sage Weil 已提交
2038 2039 2040
	ceph_msg_put(msg);
}

2041
/*
2042 2043
 * lookup and return message for incoming reply.  set up reply message
 * pages.
2044 2045
 */
static struct ceph_msg *get_reply(struct ceph_connection *con,
2046 2047
				  struct ceph_msg_header *hdr,
				  int *skip)
S
Sage Weil 已提交
2048 2049 2050
{
	struct ceph_osd *osd = con->private;
	struct ceph_osd_client *osdc = osd->o_osdc;
2051
	struct ceph_msg *m;
2052
	struct ceph_osd_request *req;
2053 2054
	int front = le32_to_cpu(hdr->front_len);
	int data_len = le32_to_cpu(hdr->data_len);
2055
	u64 tid;
S
Sage Weil 已提交
2056

2057 2058 2059 2060 2061 2062
	tid = le64_to_cpu(hdr->tid);
	mutex_lock(&osdc->request_mutex);
	req = __lookup_request(osdc, tid);
	if (!req) {
		*skip = 1;
		m = NULL;
2063 2064
		dout("get_reply unknown tid %llu from osd%d\n", tid,
		     osd->o_osd);
2065 2066
		goto out;
	}
S
Sage Weil 已提交
2067 2068

	if (req->r_con_filling_msg) {
2069
		dout("%s revoking msg %p from old con %p\n", __func__,
S
Sage Weil 已提交
2070
		     req->r_reply, req->r_con_filling_msg);
2071
		ceph_msg_revoke_incoming(req->r_reply);
2072
		req->r_con_filling_msg->ops->put(req->r_con_filling_msg);
2073
		req->r_con_filling_msg = NULL;
2074 2075
	}

S
Sage Weil 已提交
2076 2077 2078
	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);
2079
		m = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, front, GFP_NOFS, false);
2080
		if (!m)
S
Sage Weil 已提交
2081 2082 2083 2084 2085 2086
			goto out;
		ceph_msg_put(req->r_reply);
		req->r_reply = m;
	}
	m = ceph_msg_get(req->r_reply);

2087
	if (data_len > 0) {
2088
		int want = calc_pages_for(req->r_page_alignment, data_len);
2089

2090
		if (req->r_pages && unlikely(req->r_num_pages < want)) {
2091 2092 2093
			pr_warning("tid %lld reply has %d bytes %d pages, we"
				   " had only %d pages ready\n", tid, data_len,
				   want, req->r_num_pages);
2094 2095
			*skip = 1;
			ceph_msg_put(m);
2096
			m = NULL;
2097
			goto out;
2098
		}
2099 2100
		m->pages = req->r_pages;
		m->nr_pages = req->r_num_pages;
2101
		m->page_alignment = req->r_page_alignment;
2102 2103 2104
#ifdef CONFIG_BLOCK
		m->bio = req->r_bio;
#endif
2105
	}
2106
	*skip = 0;
2107
	req->r_con_filling_msg = con->ops->get(con);
S
Sage Weil 已提交
2108
	dout("get_reply tid %lld %p\n", tid, m);
2109 2110 2111

out:
	mutex_unlock(&osdc->request_mutex);
2112
	return m;
2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123

}

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
2124
	*skip = 0;
2125 2126
	switch (type) {
	case CEPH_MSG_OSD_MAP:
2127
	case CEPH_MSG_WATCH_NOTIFY:
2128
		return ceph_msg_new(type, front, GFP_NOFS, false);
2129 2130 2131 2132 2133 2134 2135 2136
	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
2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155
}

/*
 * 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);
}

2156 2157 2158
/*
 * authentication
 */
2159 2160 2161 2162 2163
/*
 * 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,
2164
					int *proto, int force_new)
2165 2166 2167 2168
{
	struct ceph_osd *o = con->private;
	struct ceph_osd_client *osdc = o->o_osdc;
	struct ceph_auth_client *ac = osdc->client->monc.auth;
2169
	struct ceph_auth_handshake *auth = &o->o_auth;
2170

2171
	if (force_new && auth->authorizer) {
2172 2173
		if (ac->ops && ac->ops->destroy_authorizer)
			ac->ops->destroy_authorizer(ac, auth->authorizer);
2174 2175
		auth->authorizer = NULL;
	}
2176
	if (!auth->authorizer && ac->ops && ac->ops->create_authorizer) {
2177 2178
		int ret = ac->ops->create_authorizer(ac, CEPH_ENTITY_TYPE_OSD,
							auth);
2179
		if (ret)
2180
			return ERR_PTR(ret);
2181 2182
	}
	*proto = ac->protocol;
2183

2184
	return auth;
2185 2186 2187 2188 2189 2190 2191 2192 2193
}


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;

2194 2195 2196 2197
	/*
	 * XXX If ac->ops or ac->ops->verify_authorizer_reply is null,
	 * XXX which do we do:  succeed or fail?
	 */
2198
	return ac->ops->verify_authorizer_reply(ac, o->o_auth.authorizer, len);
2199 2200
}

2201 2202 2203 2204 2205 2206
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;

2207
	if (ac->ops && ac->ops->invalidate_authorizer)
2208 2209 2210 2211
		ac->ops->invalidate_authorizer(ac, CEPH_ENTITY_TYPE_OSD);

	return ceph_monc_validate_auth(&osdc->client->monc);
}
2212

2213
static const struct ceph_connection_operations osd_con_ops = {
S
Sage Weil 已提交
2214 2215 2216
	.get = get_osd_con,
	.put = put_osd_con,
	.dispatch = dispatch,
2217 2218
	.get_authorizer = get_authorizer,
	.verify_authorizer_reply = verify_authorizer_reply,
2219
	.invalidate_authorizer = invalidate_authorizer,
S
Sage Weil 已提交
2220
	.alloc_msg = alloc_msg,
2221
	.fault = osd_reset,
S
Sage Weil 已提交
2222
};