osd_client.c 54.6 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_vino vino,
S
Sage Weil 已提交
67 68 69 70
		       struct ceph_file_layout *layout,
		       u64 off, u64 *plen,
		       struct ceph_osd_request *req,
		       struct ceph_osd_req_op *op)
S
Sage Weil 已提交
71
{
72 73 74 75
	u64 orig_len = *plen;
	u64 bno = 0;
	u64 objoff = 0;
	u64 objlen = 0;
S
Sage Weil 已提交
76
	int r;
S
Sage Weil 已提交
77

78 79 80
	/* object extent? */
	r = ceph_calc_file_object_mapping(layout, off, orig_len, &bno,
					  &objoff, &objlen);
S
Sage Weil 已提交
81 82
	if (r < 0)
		return r;
83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107
	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",
	     bno, objoff, objlen, req->r_num_pages);
S
Sage Weil 已提交
108

109
	snprintf(req->r_oid, sizeof(req->r_oid), "%llx.%08llx", vino.ino, bno);
S
Sage Weil 已提交
110
	req->r_oid_len = strlen(req->r_oid);
S
Sage Weil 已提交
111 112

	return r;
S
Sage Weil 已提交
113 114 115 116 117
}

/*
 * requests
 */
S
Sage Weil 已提交
118
void ceph_osdc_release_request(struct kref *kref)
S
Sage Weil 已提交
119
{
S
Sage Weil 已提交
120 121 122 123 124 125
	struct ceph_osd_request *req = container_of(kref,
						    struct ceph_osd_request,
						    r_kref);

	if (req->r_request)
		ceph_msg_put(req->r_request);
126
	if (req->r_con_filling_msg) {
127 128
		dout("%s revoking msg %p from con %p\n", __func__,
		     req->r_reply, req->r_con_filling_msg);
129
		ceph_msg_revoke_incoming(req->r_reply);
130
		req->r_con_filling_msg->ops->put(req->r_con_filling_msg);
131
		req->r_con_filling_msg = NULL;
132
	}
133 134
	if (req->r_reply)
		ceph_msg_put(req->r_reply);
S
Sage Weil 已提交
135 136 137 138
	if (req->r_own_pages)
		ceph_release_page_vector(req->r_pages,
					 req->r_num_pages);
	ceph_put_snap_context(req->r_snapc);
139
	ceph_pagelist_release(&req->r_trail);
S
Sage Weil 已提交
140 141 142 143
	if (req->r_mempool)
		mempool_free(req, req->r_osdc->req_mempool);
	else
		kfree(req);
S
Sage Weil 已提交
144
}
145
EXPORT_SYMBOL(ceph_osdc_release_request);
146

147
struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc,
S
Sage Weil 已提交
148
					       struct ceph_snap_context *snapc,
A
Alex Elder 已提交
149
					       unsigned int num_op,
150
					       bool use_mempool,
151
					       gfp_t gfp_flags)
S
Sage Weil 已提交
152 153 154
{
	struct ceph_osd_request *req;
	struct ceph_msg *msg;
155
	size_t msg_size = sizeof(struct ceph_osd_request_head);
156

157
	msg_size += num_op*sizeof(struct ceph_osd_op);
S
Sage Weil 已提交
158 159

	if (use_mempool) {
160
		req = mempool_alloc(osdc->req_mempool, gfp_flags);
S
Sage Weil 已提交
161 162
		memset(req, 0, sizeof(*req));
	} else {
163
		req = kzalloc(sizeof(*req), gfp_flags);
S
Sage Weil 已提交
164 165
	}
	if (req == NULL)
166
		return NULL;
S
Sage Weil 已提交
167 168 169

	req->r_osdc = osdc;
	req->r_mempool = use_mempool;
170

S
Sage Weil 已提交
171
	kref_init(&req->r_kref);
S
Sage Weil 已提交
172 173
	init_completion(&req->r_completion);
	init_completion(&req->r_safe_completion);
174
	RB_CLEAR_NODE(&req->r_node);
S
Sage Weil 已提交
175
	INIT_LIST_HEAD(&req->r_unsafe_item);
176 177
	INIT_LIST_HEAD(&req->r_linger_item);
	INIT_LIST_HEAD(&req->r_linger_osd);
178
	INIT_LIST_HEAD(&req->r_req_lru_item);
179 180
	INIT_LIST_HEAD(&req->r_osd_item);

S
Sage Weil 已提交
181 182 183 184 185
	/* create reply message */
	if (use_mempool)
		msg = ceph_msgpool_get(&osdc->msgpool_op_reply, 0);
	else
		msg = ceph_msg_new(CEPH_MSG_OSD_OPREPLY,
186
				   OSD_OPREPLY_FRONT_LEN, gfp_flags, true);
187
	if (!msg) {
S
Sage Weil 已提交
188
		ceph_osdc_put_request(req);
189
		return NULL;
S
Sage Weil 已提交
190 191 192
	}
	req->r_reply = msg;

193
	ceph_pagelist_init(&req->r_trail);
194

S
Sage Weil 已提交
195
	/* create request message; allow space for oid */
196
	msg_size += MAX_OBJ_NAME_SIZE;
S
Sage Weil 已提交
197 198 199
	if (snapc)
		msg_size += sizeof(u64) * snapc->num_snaps;
	if (use_mempool)
200
		msg = ceph_msgpool_get(&osdc->msgpool_op, 0);
S
Sage Weil 已提交
201
	else
202
		msg = ceph_msg_new(CEPH_MSG_OSD_OP, msg_size, gfp_flags, true);
203
	if (!msg) {
S
Sage Weil 已提交
204
		ceph_osdc_put_request(req);
205
		return NULL;
S
Sage Weil 已提交
206
	}
207

S
Sage Weil 已提交
208
	memset(msg->front.iov_base, 0, msg->front.iov_len);
209 210 211 212 213

	req->r_request = msg;

	return req;
}
214
EXPORT_SYMBOL(ceph_osdc_alloc_request);
215

216 217 218 219 220 221
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);

222
	switch (src->op) {
A
Alex Elder 已提交
223 224
	case CEPH_OSD_OP_STAT:
		break;
225 226 227 228 229 230 231 232 233 234 235
	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;
236 237 238 239 240
	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);

241
		ceph_pagelist_append(&req->r_trail, src->cls.class_name,
242
				     src->cls.class_len);
243
		ceph_pagelist_append(&req->r_trail, src->cls.method_name,
244
				     src->cls.method_len);
245
		ceph_pagelist_append(&req->r_trail, src->cls.indata,
246 247
				     src->cls.indata_len);
		break;
248 249
	case CEPH_OSD_OP_STARTSYNC:
		break;
250 251 252 253 254 255
	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;
256 257 258 259
	default:
		pr_err("unrecognized osd opcode %d\n", dst->op);
		WARN_ON(1);
		break;
260 261 262
	case CEPH_OSD_OP_MAPEXT:
	case CEPH_OSD_OP_MASKTRUNC:
	case CEPH_OSD_OP_SPARSE_READ:
263
	case CEPH_OSD_OP_NOTIFY:
264 265 266 267 268 269 270 271 272 273 274 275
	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:
276
	case CEPH_OSD_OP_ROLLBACK:
277 278 279 280 281 282 283 284 285 286 287 288 289
	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:
290
	case CEPH_OSD_OP_GETXATTR:
291
	case CEPH_OSD_OP_GETXATTRS:
292 293
	case CEPH_OSD_OP_CMPXATTR:
	case CEPH_OSD_OP_SETXATTR:
294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317
	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;
318 319 320 321
	}
	dst->payload_len = cpu_to_le32(src->payload_len);
}

322 323 324 325 326
/*
 * build new request AND message
 *
 */
void ceph_osdc_build_request(struct ceph_osd_request *req,
A
Alex Elder 已提交
327
			     u64 off, u64 len, unsigned int num_op,
328
			     struct ceph_osd_req_op *src_ops,
329
			     struct ceph_snap_context *snapc, u64 snap_id,
330
			     struct timespec *mtime)
331 332 333
{
	struct ceph_msg *msg = req->r_request;
	struct ceph_osd_request_head *head;
334
	struct ceph_osd_req_op *src_op;
335 336 337 338
	struct ceph_osd_op *op;
	void *p;
	size_t msg_size = sizeof(*head) + num_op*sizeof(*op);
	int flags = req->r_flags;
339
	u64 data_len;
340
	int i;
341

342 343
	WARN_ON((flags & (CEPH_OSD_FLAG_READ|CEPH_OSD_FLAG_WRITE)) == 0);

S
Sage Weil 已提交
344
	head = msg->front.iov_base;
345
	head->snapid = cpu_to_le64(snap_id);
S
Sage Weil 已提交
346 347 348 349 350 351 352 353 354
	op = (void *)(head + 1);
	p = (void *)(op + num_op);

	req->r_snapc = ceph_get_snap_context(snapc);

	head->client_inc = cpu_to_le32(1); /* always, for now. */
	head->flags = cpu_to_le32(flags);
	if (flags & CEPH_OSD_FLAG_WRITE)
		ceph_encode_timespec(&head->mtime, mtime);
A
Alex Elder 已提交
355
	BUG_ON(num_op > (unsigned int) ((u16) -1));
S
Sage Weil 已提交
356 357 358
	head->num_ops = cpu_to_le16(num_op);

	/* fill in oid */
359 360 361
	head->object_len = cpu_to_le32(req->r_oid_len);
	memcpy(p, req->r_oid, req->r_oid_len);
	p += req->r_oid_len;
S
Sage Weil 已提交
362

363
	src_op = src_ops;
A
Alex Elder 已提交
364 365
	while (num_op--)
		osd_req_encode_op(req, op++, src_op++);
366

S
Sage Weil 已提交
367 368 369 370 371 372 373 374 375
	if (snapc) {
		head->snap_seq = cpu_to_le64(snapc->seq);
		head->num_snaps = cpu_to_le32(snapc->num_snaps);
		for (i = 0; i < snapc->num_snaps; i++) {
			put_unaligned_le64(snapc->snaps[i], p);
			p += sizeof(u64);
		}
	}

376
	data_len = req->r_trail.length;
377 378
	if (flags & CEPH_OSD_FLAG_WRITE) {
		req->r_request->hdr.data_off = cpu_to_le16(off);
379
		data_len += len;
380
	}
381
	req->r_request->hdr.data_len = cpu_to_le32(data_len);
382 383
	req->r_request->page_alignment = req->r_page_alignment;

S
Sage Weil 已提交
384
	BUG_ON(p > msg->front.iov_base + msg->front.iov_len);
385 386 387
	msg_size = p - msg->front.iov_base;
	msg->front.iov_len = msg_size;
	msg->hdr.front_len = cpu_to_le32(msg_size);
388 389
	return;
}
390
EXPORT_SYMBOL(ceph_osdc_build_request);
391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412

/*
 * 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,
413
					       bool use_mempool,
414
					       int page_align)
415
{
A
Alex Elder 已提交
416
	struct ceph_osd_req_op ops[2];
417
	struct ceph_osd_request *req;
A
Alex Elder 已提交
418
	unsigned int num_op = 1;
419
	int r;
420

A
Alex Elder 已提交
421 422
	memset(&ops, 0, sizeof ops);

423 424 425 426 427 428
	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
429 430
		num_op++;
	}
431

A
Alex Elder 已提交
432 433
	req = ceph_osdc_alloc_request(osdc, snapc, num_op, use_mempool,
					GFP_NOFS);
434
	if (!req)
435
		return ERR_PTR(-ENOMEM);
436
	req->r_flags = flags;
437 438

	/* calculate max write size */
439
	r = calc_layout(vino, layout, off, plen, req, ops);
440 441
	if (r < 0)
		return ERR_PTR(r);
442 443
	req->r_file_layout = *layout;  /* keep a copy */

444 445 446
	/* 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);
447 448
	req->r_page_alignment = page_align;

A
Alex Elder 已提交
449 450
	ceph_osdc_build_request(req, off, *plen, num_op, ops,
				snapc, vino.snap, mtime);
451

S
Sage Weil 已提交
452 453
	return req;
}
454
EXPORT_SYMBOL(ceph_osdc_new_request);
S
Sage Weil 已提交
455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520

/*
 * 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;
}

521 522 523 524 525 526
/*
 * Resubmit requests pending on the given osd.
 */
static void __kick_osd_requests(struct ceph_osd_client *osdc,
				struct ceph_osd *osd)
{
527
	struct ceph_osd_request *req, *nreq;
528 529 530 531
	int err;

	dout("__kick_osd_requests osd%d\n", osd->o_osd);
	err = __reset_osd(osdc, osd);
532
	if (err)
533 534 535 536 537 538
		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);
539 540 541 542 543 544
		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) {
545 546 547 548 549
		/*
		 * reregister request prior to unregistering linger so
		 * that r_osd is preserved.
		 */
		BUG_ON(!list_empty(&req->r_req_lru_item));
550
		__register_request(osdc, req);
551 552 553
		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);
554 555
		dout("requeued lingering %p tid %llu osd%d\n", req, req->r_tid,
		     osd->o_osd);
556 557 558
	}
}

S
Sage Weil 已提交
559
/*
560
 * If the osd connection drops, we need to resubmit all requests.
S
Sage Weil 已提交
561 562 563 564 565 566 567 568 569 570 571
 */
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
572 573
	mutex_lock(&osdc->request_mutex);
	__kick_osd_requests(osdc, osd);
A
Alex Elder 已提交
574
	__send_queued(osdc);
S
Sage Weil 已提交
575
	mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
576 577 578 579 580 581
	up_read(&osdc->map_sem);
}

/*
 * Track open sessions with osds.
 */
582
static struct ceph_osd *create_osd(struct ceph_osd_client *osdc, int onum)
S
Sage Weil 已提交
583 584 585 586 587 588 589 590 591
{
	struct ceph_osd *osd;

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

	atomic_set(&osd->o_ref, 1);
	osd->o_osdc = osdc;
592
	osd->o_osd = onum;
593
	RB_CLEAR_NODE(&osd->o_node);
S
Sage Weil 已提交
594
	INIT_LIST_HEAD(&osd->o_requests);
595
	INIT_LIST_HEAD(&osd->o_linger_requests);
596
	INIT_LIST_HEAD(&osd->o_osd_lru);
S
Sage Weil 已提交
597 598
	osd->o_incarnation = 1;

599
	ceph_con_init(&osd->o_con, osd, &osd_con_ops, &osdc->client->msgr);
600

601
	INIT_LIST_HEAD(&osd->o_keepalive_item);
S
Sage Weil 已提交
602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620
	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);
621
	if (atomic_dec_and_test(&osd->o_ref) && osd->o_auth.authorizer) {
S
Sage Weil 已提交
622 623
		struct ceph_auth_client *ac = osd->o_osdc->client->monc.auth;

624
		if (ac->ops && ac->ops->destroy_authorizer)
625
			ac->ops->destroy_authorizer(ac, osd->o_auth.authorizer);
S
Sage Weil 已提交
626
		kfree(osd);
S
Sage Weil 已提交
627
	}
S
Sage Weil 已提交
628 629 630 631 632
}

/*
 * remove an osd from our map
 */
633
static void __remove_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
S
Sage Weil 已提交
634
{
635
	dout("__remove_osd %p\n", osd);
S
Sage Weil 已提交
636 637
	BUG_ON(!list_empty(&osd->o_requests));
	rb_erase(&osd->o_node, &osdc->osds);
638
	list_del_init(&osd->o_osd_lru);
S
Sage Weil 已提交
639 640 641 642
	ceph_con_close(&osd->o_con);
	put_osd(osd);
}

643 644
static void remove_all_osds(struct ceph_osd_client *osdc)
{
645
	dout("%s %p\n", __func__, osdc);
646 647 648 649 650 651 652 653 654
	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);
}

655 656 657 658 659 660
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);
661
	osd->lru_ttl = jiffies + osdc->client->options->osd_idle_ttl * HZ;
662 663 664 665 666 667 668 669 670
}

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);
}

671
static void remove_old_osds(struct ceph_osd_client *osdc)
672 673 674 675 676 677
{
	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) {
678
		if (time_before(jiffies, osd->lru_ttl))
679 680 681 682 683 684
			break;
		__remove_osd(osdc, osd);
	}
	mutex_unlock(&osdc->request_mutex);
}

S
Sage Weil 已提交
685 686 687
/*
 * reset osd connect
 */
688
static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd)
S
Sage Weil 已提交
689
{
A
Alex Elder 已提交
690
	struct ceph_entity_addr *peer_addr;
S
Sage Weil 已提交
691

692
	dout("__reset_osd %p osd%d\n", osd, osd->o_osd);
693 694
	if (list_empty(&osd->o_requests) &&
	    list_empty(&osd->o_linger_requests)) {
695
		__remove_osd(osdc, osd);
A
Alex Elder 已提交
696 697 698 699 700 701 702 703 704

		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;

705 706 707 708 709
		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
710 711

		return -EAGAIN;
S
Sage Weil 已提交
712
	}
A
Alex Elder 已提交
713 714 715 716 717 718

	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
719 720 721 722 723 724 725 726
}

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;

727
	dout("__insert_osd %p osd%d\n", new, new->o_osd);
S
Sage Weil 已提交
728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759
	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;
}

760 761 762
static void __schedule_osd_timeout(struct ceph_osd_client *osdc)
{
	schedule_delayed_work(&osdc->timeout_work,
763
			osdc->client->options->osd_keepalive_timeout * HZ);
764 765 766 767 768 769
}

static void __cancel_osd_timeout(struct ceph_osd_client *osdc)
{
	cancel_delayed_work(&osdc->timeout_work);
}
S
Sage Weil 已提交
770 771 772 773 774

/*
 * Register request, assign tid.  If this is the first request, set up
 * the timeout event.
 */
775 776
static void __register_request(struct ceph_osd_client *osdc,
			       struct ceph_osd_request *req)
S
Sage Weil 已提交
777 778
{
	req->r_tid = ++osdc->last_tid;
779
	req->r_request->hdr.tid = cpu_to_le64(req->r_tid);
780
	dout("__register_request %p tid %lld\n", req, req->r_tid);
S
Sage Weil 已提交
781 782 783 784
	__insert_request(osdc, req);
	ceph_osdc_get_request(req);
	osdc->num_requests++;
	if (osdc->num_requests == 1) {
785 786
		dout(" first request, scheduling timeout\n");
		__schedule_osd_timeout(osdc);
S
Sage Weil 已提交
787
	}
788 789 790 791 792 793 794
}

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
795 796 797 798 799 800 801 802 803
	mutex_unlock(&osdc->request_mutex);
}

/*
 * called under osdc->request_mutex
 */
static void __unregister_request(struct ceph_osd_client *osdc,
				 struct ceph_osd_request *req)
{
804 805 806 807 808 809
	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
810 811 812 813
	dout("__unregister_request %p tid %lld\n", req, req->r_tid);
	rb_erase(&req->r_node, &osdc->requests);
	osdc->num_requests--;

814 815
	if (req->r_osd) {
		/* make sure the original request isn't in flight. */
816
		ceph_msg_revoke(req->r_request);
817 818

		list_del_init(&req->r_osd_item);
819 820 821
		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);
822
			__move_osd_to_lru(osdc, req->r_osd);
823
		}
824
		if (list_empty(&req->r_linger_item))
825
			req->r_osd = NULL;
826
	}
S
Sage Weil 已提交
827

A
Alex Elder 已提交
828
	list_del_init(&req->r_req_lru_item);
S
Sage Weil 已提交
829 830
	ceph_osdc_put_request(req);

831 832 833
	if (osdc->num_requests == 0) {
		dout(" no requests, canceling timeout\n");
		__cancel_osd_timeout(osdc);
S
Sage Weil 已提交
834 835 836 837 838 839 840 841
	}
}

/*
 * Cancel a previously queued request message
 */
static void __cancel_request(struct ceph_osd_request *req)
{
842
	if (req->r_sent && req->r_osd) {
843
		ceph_msg_revoke(req->r_request);
S
Sage Weil 已提交
844 845 846 847
		req->r_sent = 0;
	}
}

848 849 850 851 852
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);
853 854 855
	if (req->r_osd)
		list_add_tail(&req->r_linger_osd,
			      &req->r_osd->o_linger_requests);
856 857 858 859 860 861
}

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
862
	list_del_init(&req->r_linger_item);
863 864 865 866 867 868 869 870
	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);
		}
871 872
		if (list_empty(&req->r_osd_item))
			req->r_osd = NULL;
873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902
	}
}

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
903 904 905
/*
 * 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
906
 * no up osd, set r_osd to NULL.  Move the request to the appropriate list
907
 * (unsent, homeless) or leave on in-flight lru.
S
Sage Weil 已提交
908 909 910 911 912
 *
 * Return 0 if unchanged, 1 if changed, or negative on error.
 *
 * Caller should hold map_sem for read and request_mutex.
 */
913
static int __map_request(struct ceph_osd_client *osdc,
914
			 struct ceph_osd_request *req, int force_resend)
S
Sage Weil 已提交
915
{
916
	struct ceph_pg pgid;
917 918
	int acting[CEPH_PG_MAX_SIZE];
	int o = -1, num = 0;
S
Sage Weil 已提交
919 920
	int err;

921
	dout("map_request %p tid %lld\n", req, req->r_tid);
922
	err = ceph_calc_object_layout(&pgid, req->r_oid,
S
Sage Weil 已提交
923
				      &req->r_file_layout, osdc->osdmap);
924 925
	if (err) {
		list_move(&req->r_req_lru_item, &osdc->req_notarget);
S
Sage Weil 已提交
926
		return err;
927
	}
928 929
	req->r_pgid = pgid;

930 931 932 933 934
	err = ceph_calc_pg_acting(osdc->osdmap, pgid, acting);
	if (err > 0) {
		o = acting[0];
		num = err;
	}
S
Sage Weil 已提交
935

936 937
	if ((!force_resend &&
	     req->r_osd && req->r_osd->o_osd == o &&
938 939 940
	     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
941 942 943
	    (req->r_osd == NULL && o == -1))
		return 0;  /* no change */

944 945
	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
946 947
	     req->r_osd ? req->r_osd->o_osd : -1);

948 949 950 951
	/* 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
952 953 954 955 956 957 958 959
	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) {
960
		err = -ENOMEM;
961
		req->r_osd = create_osd(osdc, o);
962 963
		if (!req->r_osd) {
			list_move(&req->r_req_lru_item, &osdc->req_notarget);
964
			goto out;
965
		}
S
Sage Weil 已提交
966

967
		dout("map_request osd %p is osd%d\n", req->r_osd, o);
S
Sage Weil 已提交
968 969
		__insert_osd(osdc, req->r_osd);

970 971 972
		ceph_con_open(&req->r_osd->o_con,
			      CEPH_ENTITY_TYPE_OSD, o,
			      &osdc->osdmap->osd_addr[o]);
S
Sage Weil 已提交
973 974
	}

975 976
	if (req->r_osd) {
		__remove_osd_from_lru(req->r_osd);
S
Sage Weil 已提交
977
		list_add(&req->r_osd_item, &req->r_osd->o_requests);
978 979 980
		list_move(&req->r_req_lru_item, &osdc->req_unsent);
	} else {
		list_move(&req->r_req_lru_item, &osdc->req_notarget);
981
	}
982
	err = 1;   /* osd or pg changed */
S
Sage Weil 已提交
983 984 985 986 987 988 989 990

out:
	return err;
}

/*
 * caller should hold map_sem (for read) and request_mutex
 */
991 992
static void __send_request(struct ceph_osd_client *osdc,
			   struct ceph_osd_request *req)
S
Sage Weil 已提交
993 994 995 996 997 998 999
{
	struct ceph_osd_request_head *reqhead;

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

	reqhead = req->r_request->front.iov_base;
1000
	reqhead->snapid = cpu_to_le64(req->r_snapid);
S
Sage Weil 已提交
1001 1002 1003 1004
	reqhead->osdmap_epoch = cpu_to_le32(osdc->osdmap->epoch);
	reqhead->flags |= cpu_to_le32(req->r_flags);  /* e.g., RETRY */
	reqhead->reassert_version = req->r_reassert_version;

1005 1006 1007 1008 1009
	reqhead->layout.ol_pgid.ps = cpu_to_le16(req->r_pgid.seed);
	reqhead->layout.ol_pgid.pool = cpu_to_le32(req->r_pgid.pool);
	reqhead->layout.ol_pgid.preferred = cpu_to_le16(-1);
	reqhead->layout.ol_stripe_unit = 0;

S
Sage Weil 已提交
1010
	req->r_stamp = jiffies;
1011
	list_move_tail(&req->r_req_lru_item, &osdc->req_lru);
S
Sage Weil 已提交
1012 1013 1014 1015 1016 1017

	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;
}

1018 1019 1020
/*
 * Send any requests in the queue (req_unsent).
 */
A
Alex Elder 已提交
1021
static void __send_queued(struct ceph_osd_client *osdc)
1022 1023 1024
{
	struct ceph_osd_request *req, *tmp;

A
Alex Elder 已提交
1025 1026
	dout("__send_queued\n");
	list_for_each_entry_safe(req, tmp, &osdc->req_unsent, r_req_lru_item)
1027 1028 1029
		__send_request(osdc, req);
}

S
Sage Weil 已提交
1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042
/*
 * 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
1043
	struct ceph_osd_request *req;
S
Sage Weil 已提交
1044
	struct ceph_osd *osd;
1045
	unsigned long keepalive =
1046
		osdc->client->options->osd_keepalive_timeout * HZ;
1047
	struct list_head slow_osds;
S
Sage Weil 已提交
1048 1049 1050 1051 1052 1053 1054
	dout("timeout\n");
	down_read(&osdc->map_sem);

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

	mutex_lock(&osdc->request_mutex);

1055 1056 1057 1058 1059 1060 1061
	/*
	 * 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
1062
		if (time_before(jiffies, req->r_stamp + keepalive))
1063 1064 1065 1066 1067
			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
1068
		     req->r_tid, osd->o_osd);
1069 1070 1071 1072 1073 1074
		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
1075 1076 1077
		ceph_con_keepalive(&osd->o_con);
	}

1078
	__schedule_osd_timeout(osdc);
A
Alex Elder 已提交
1079
	__send_queued(osdc);
S
Sage Weil 已提交
1080 1081 1082 1083
	mutex_unlock(&osdc->request_mutex);
	up_read(&osdc->map_sem);
}

1084 1085 1086 1087 1088 1089
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 =
1090
		osdc->client->options->osd_idle_ttl * HZ >> 2;
1091 1092 1093

	dout("osds timeout\n");
	down_read(&osdc->map_sem);
1094
	remove_old_osds(osdc);
1095 1096 1097 1098 1099 1100
	up_read(&osdc->map_sem);

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

S
Sage Weil 已提交
1101 1102 1103 1104 1105 1106 1107
static void complete_request(struct ceph_osd_request *req)
{
	if (req->r_safe_callback)
		req->r_safe_callback(req, NULL);
	complete_all(&req->r_safe_completion);  /* fsync waiter */
}

S
Sage Weil 已提交
1108 1109 1110 1111
/*
 * handle osd op reply.  either call the callback if it is specified,
 * or do the completion to wake up the waiting thread.
 */
1112 1113
static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
			 struct ceph_connection *con)
S
Sage Weil 已提交
1114 1115 1116 1117 1118
{
	struct ceph_osd_reply_head *rhead = msg->front.iov_base;
	struct ceph_osd_request *req;
	u64 tid;
	int numops, object_len, flags;
1119
	s32 result;
S
Sage Weil 已提交
1120

1121
	tid = le64_to_cpu(msg->hdr.tid);
S
Sage Weil 已提交
1122 1123 1124 1125
	if (msg->front.iov_len < sizeof(*rhead))
		goto bad;
	numops = le32_to_cpu(rhead->num_ops);
	object_len = le32_to_cpu(rhead->object_len);
1126
	result = le32_to_cpu(rhead->result);
S
Sage Weil 已提交
1127 1128 1129
	if (msg->front.iov_len != sizeof(*rhead) + object_len +
	    numops * sizeof(struct ceph_osd_op))
		goto bad;
1130
	dout("handle_reply %p tid %llu result %d\n", msg, tid, (int)result);
S
Sage Weil 已提交
1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141
	/* lookup */
	mutex_lock(&osdc->request_mutex);
	req = __lookup_request(osdc, tid);
	if (req == NULL) {
		dout("handle_reply tid %llu dne\n", tid);
		mutex_unlock(&osdc->request_mutex);
		return;
	}
	ceph_osdc_get_request(req);
	flags = le32_to_cpu(rhead->flags);

1142
	/*
1143
	 * if this connection filled our message, drop our reference now, to
1144 1145
	 * avoid a (safe but slower) revoke later.
	 */
1146
	if (req->r_con_filling_msg == con && req->r_reply == msg) {
S
Sage Weil 已提交
1147
		dout(" dropping con_filling_msg ref %p\n", con);
1148
		req->r_con_filling_msg = NULL;
1149
		con->ops->put(con);
1150 1151
	}

S
Sage Weil 已提交
1152
	if (!req->r_got_reply) {
1153
		unsigned int bytes;
S
Sage Weil 已提交
1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167

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

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

		req->r_got_reply = 1;
	} else if ((flags & CEPH_OSD_FLAG_ONDISK) == 0) {
		dout("handle_reply tid %llu dup ack\n", tid);
S
Sage Weil 已提交
1168
		mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1169 1170 1171 1172 1173
		goto done;
	}

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

1174 1175 1176
	if (req->r_linger && (flags & CEPH_OSD_FLAG_ONDISK))
		__register_linger_request(osdc, req);

S
Sage Weil 已提交
1177
	/* either this is a read, or we got the safe response */
1178 1179
	if (result < 0 ||
	    (flags & CEPH_OSD_FLAG_ONDISK) ||
S
Sage Weil 已提交
1180 1181 1182 1183 1184 1185 1186 1187
	    ((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
1188
		complete_all(&req->r_completion);
S
Sage Weil 已提交
1189

S
Sage Weil 已提交
1190 1191
	if (flags & CEPH_OSD_FLAG_ONDISK)
		complete_request(req);
S
Sage Weil 已提交
1192 1193

done:
1194
	dout("req=%p req->r_linger=%d\n", req, req->r_linger);
S
Sage Weil 已提交
1195 1196 1197 1198 1199 1200 1201
	ceph_osdc_put_request(req);
	return;

bad:
	pr_err("corrupt osd_op_reply got %d %d expected %d\n",
	       (int)msg->front.iov_len, le32_to_cpu(msg->hdr.front_len),
	       (int)sizeof(*rhead));
1202
	ceph_msg_dump(msg);
S
Sage Weil 已提交
1203 1204
}

1205
static void reset_changed_osds(struct ceph_osd_client *osdc)
S
Sage Weil 已提交
1206 1207 1208
{
	struct rb_node *p, *n;

1209 1210
	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
1211

1212 1213 1214 1215 1216 1217 1218
		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
1219
	}
1220 1221 1222
}

/*
1223 1224
 * Requeue requests whose mapping to an OSD has changed.  If requests map to
 * no osd, request a new map.
1225
 *
1226
 * Caller should hold map_sem for read.
1227
 */
1228
static void kick_requests(struct ceph_osd_client *osdc, int force_resend)
1229
{
1230
	struct ceph_osd_request *req, *nreq;
1231 1232 1233
	struct rb_node *p;
	int needmap = 0;
	int err;
1234

1235
	dout("kick_requests %s\n", force_resend ? " (force resend)" : "");
1236
	mutex_lock(&osdc->request_mutex);
1237
	for (p = rb_first(&osdc->requests); p; ) {
1238
		req = rb_entry(p, struct ceph_osd_request, r_node);
1239
		p = rb_next(p);
1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257

		/*
		 * 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;
		}

1258
		err = __map_request(osdc, req, force_resend);
1259 1260 1261 1262 1263 1264
		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) {
1265 1266 1267 1268
			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);
1269
				req->r_flags |= CEPH_OSD_FLAG_RETRY;
1270 1271
			}
		}
1272 1273 1274 1275 1276 1277
	}

	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);

1278
		err = __map_request(osdc, req, force_resend);
1279
		dout("__map_request returned %d\n", err);
1280 1281 1282 1283 1284 1285 1286 1287
		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;
1288
		}
1289 1290 1291 1292

		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);
1293
		__unregister_linger_request(osdc, req);
1294
	}
S
Sage Weil 已提交
1295 1296 1297 1298 1299 1300
	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);
	}
1301
	reset_changed_osds(osdc);
1302
}
1303 1304


S
Sage Weil 已提交
1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327
/*
 * 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));
1328 1329
	if (ceph_check_fsid(osdc->client, &fsid) < 0)
		return;
S
Sage Weil 已提交
1330 1331 1332 1333 1334 1335 1336 1337

	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);
1338 1339
		epoch = ceph_decode_32(&p);
		maplen = ceph_decode_32(&p);
S
Sage Weil 已提交
1340 1341 1342 1343 1344 1345 1346
		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,
1347
							  &osdc->client->msgr);
S
Sage Weil 已提交
1348 1349 1350 1351
			if (IS_ERR(newmap)) {
				err = PTR_ERR(newmap);
				goto bad;
			}
1352
			BUG_ON(!newmap);
S
Sage Weil 已提交
1353 1354 1355 1356
			if (newmap != osdc->osdmap) {
				ceph_osdmap_destroy(osdc->osdmap);
				osdc->osdmap = newmap;
			}
1357
			kick_requests(osdc, 0);
S
Sage Weil 已提交
1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372
		} 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);
1373 1374
		epoch = ceph_decode_32(&p);
		maplen = ceph_decode_32(&p);
S
Sage Weil 已提交
1375 1376 1377 1378 1379 1380 1381 1382 1383
		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 {
1384 1385
			int skipped_map = 0;

S
Sage Weil 已提交
1386 1387 1388 1389 1390 1391
			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;
			}
1392
			BUG_ON(!newmap);
S
Sage Weil 已提交
1393 1394
			oldmap = osdc->osdmap;
			osdc->osdmap = newmap;
1395 1396 1397
			if (oldmap) {
				if (oldmap->epoch + 1 < newmap->epoch)
					skipped_map = 1;
S
Sage Weil 已提交
1398
				ceph_osdmap_destroy(oldmap);
1399 1400
			}
			kick_requests(osdc, skipped_map);
S
Sage Weil 已提交
1401 1402 1403 1404 1405 1406 1407 1408
		}
		p += maplen;
		nr_maps--;
	}

done:
	downgrade_write(&osdc->map_sem);
	ceph_monc_got_osdmap(&osdc->client->monc, osdc->osdmap->epoch);
1409 1410 1411 1412 1413 1414 1415 1416 1417

	/*
	 * 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
1418 1419 1420
	mutex_lock(&osdc->request_mutex);
	__send_queued(osdc);
	mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1421
	up_read(&osdc->map_sem);
1422
	wake_up_all(&osdc->client->auth_wq);
S
Sage Weil 已提交
1423 1424 1425 1426
	return;

bad:
	pr_err("osdc handle_map corrupt msg\n");
1427
	ceph_msg_dump(msg);
S
Sage Weil 已提交
1428 1429 1430 1431
	up_write(&osdc->map_sem);
	return;
}

1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513
/*
 * 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 *),
1514
			   void *data, struct ceph_osd_event **pevent)
1515 1516 1517 1518 1519 1520 1521 1522 1523
{
	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;
1524
	event->one_shot = 0;
1525 1526 1527
	event->data = data;
	event->osdc = osdc;
	INIT_LIST_HEAD(&event->osd_node);
1528
	RB_CLEAR_NODE(&event->node);
1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574
	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
 */
1575 1576
static void handle_watch_notify(struct ceph_osd_client *osdc,
				struct ceph_msg *msg)
1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596
{
	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) {
1597
		BUG_ON(event->one_shot);
1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608
		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;
		}
1609
		INIT_WORK(&event_work->work, do_event_work);
1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630
		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
1631 1632 1633 1634 1635 1636 1637
/*
 * 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
1638
	int rc = 0;
S
Sage Weil 已提交
1639 1640 1641

	req->r_request->pages = req->r_pages;
	req->r_request->nr_pages = req->r_num_pages;
1642 1643 1644
#ifdef CONFIG_BLOCK
	req->r_request->bio = req->r_bio;
#endif
1645
	req->r_request->trail = &req->r_trail;
S
Sage Weil 已提交
1646 1647 1648 1649 1650

	register_request(osdc, req);

	down_read(&osdc->map_sem);
	mutex_lock(&osdc->request_mutex);
S
Sage Weil 已提交
1651 1652 1653 1654 1655 1656
	/*
	 * 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) {
1657
		rc = __map_request(osdc, req, 0);
1658 1659 1660 1661 1662 1663
		if (rc < 0) {
			if (nofail) {
				dout("osdc_start_request failed map, "
				     " will retry %lld\n", req->r_tid);
				rc = 0;
			}
1664
			goto out_unlock;
1665
		}
1666 1667 1668 1669
		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 {
1670
			__send_request(osdc, req);
S
Sage Weil 已提交
1671
		}
1672
		rc = 0;
S
Sage Weil 已提交
1673
	}
1674 1675

out_unlock:
S
Sage Weil 已提交
1676 1677 1678 1679
	mutex_unlock(&osdc->request_mutex);
	up_read(&osdc->map_sem);
	return rc;
}
1680
EXPORT_SYMBOL(ceph_osdc_start_request);
S
Sage Weil 已提交
1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693

/*
 * 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);
1694
		__unregister_request(osdc, req);
S
Sage Weil 已提交
1695
		mutex_unlock(&osdc->request_mutex);
S
Sage Weil 已提交
1696
		complete_request(req);
1697
		dout("wait_request tid %llu canceled/timed out\n", req->r_tid);
S
Sage Weil 已提交
1698 1699 1700 1701 1702 1703
		return rc;
	}

	dout("wait_request tid %llu result %d\n", req->r_tid, req->r_result);
	return req->r_result;
}
1704
EXPORT_SYMBOL(ceph_osdc_wait_request);
S
Sage Weil 已提交
1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737

/*
 * 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);
}
1738
EXPORT_SYMBOL(ceph_osdc_sync);
S
Sage Weil 已提交
1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755

/*
 * 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;
1756
	INIT_LIST_HEAD(&osdc->osd_lru);
S
Sage Weil 已提交
1757
	osdc->requests = RB_ROOT;
1758
	INIT_LIST_HEAD(&osdc->req_lru);
1759 1760
	INIT_LIST_HEAD(&osdc->req_unsent);
	INIT_LIST_HEAD(&osdc->req_notarget);
1761
	INIT_LIST_HEAD(&osdc->req_linger);
S
Sage Weil 已提交
1762 1763
	osdc->num_requests = 0;
	INIT_DELAYED_WORK(&osdc->timeout_work, handle_timeout);
1764
	INIT_DELAYED_WORK(&osdc->osds_timeout_work, handle_osds_timeout);
1765 1766 1767
	spin_lock_init(&osdc->event_lock);
	osdc->event_tree = RB_ROOT;
	osdc->event_count = 0;
1768 1769

	schedule_delayed_work(&osdc->osds_timeout_work,
1770
	   round_jiffies_relative(osdc->client->options->osd_idle_ttl * HZ));
S
Sage Weil 已提交
1771

1772
	err = -ENOMEM;
S
Sage Weil 已提交
1773 1774 1775
	osdc->req_mempool = mempool_create_kmalloc_pool(10,
					sizeof(struct ceph_osd_request));
	if (!osdc->req_mempool)
1776
		goto out;
S
Sage Weil 已提交
1777

1778 1779
	err = ceph_msgpool_init(&osdc->msgpool_op, CEPH_MSG_OSD_OP,
				OSD_OP_FRONT_LEN, 10, true,
1780
				"osd_op");
S
Sage Weil 已提交
1781
	if (err < 0)
1782
		goto out_mempool;
1783
	err = ceph_msgpool_init(&osdc->msgpool_op_reply, CEPH_MSG_OSD_OPREPLY,
1784 1785
				OSD_OPREPLY_FRONT_LEN, 10, true,
				"osd_op_reply");
S
Sage Weil 已提交
1786 1787
	if (err < 0)
		goto out_msgpool;
1788 1789 1790 1791 1792 1793 1794

	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
1795
	return 0;
1796

S
Sage Weil 已提交
1797 1798
out_msgpool:
	ceph_msgpool_destroy(&osdc->msgpool_op);
1799 1800 1801 1802
out_mempool:
	mempool_destroy(osdc->req_mempool);
out:
	return err;
S
Sage Weil 已提交
1803 1804 1805 1806
}

void ceph_osdc_stop(struct ceph_osd_client *osdc)
{
1807 1808
	flush_workqueue(osdc->notify_wq);
	destroy_workqueue(osdc->notify_wq);
S
Sage Weil 已提交
1809
	cancel_delayed_work_sync(&osdc->timeout_work);
1810
	cancel_delayed_work_sync(&osdc->osds_timeout_work);
S
Sage Weil 已提交
1811 1812 1813 1814
	if (osdc->osdmap) {
		ceph_osdmap_destroy(osdc->osdmap);
		osdc->osdmap = NULL;
	}
1815
	remove_all_osds(osdc);
S
Sage Weil 已提交
1816 1817
	mempool_destroy(osdc->req_mempool);
	ceph_msgpool_destroy(&osdc->msgpool_op);
S
Sage Weil 已提交
1818
	ceph_msgpool_destroy(&osdc->msgpool_op_reply);
S
Sage Weil 已提交
1819 1820 1821 1822 1823 1824 1825 1826 1827 1828
}

/*
 * 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,
1829
			struct page **pages, int num_pages, int page_align)
S
Sage Weil 已提交
1830 1831 1832 1833 1834 1835 1836 1837 1838
{
	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,
1839
				    false, page_align);
1840 1841
	if (IS_ERR(req))
		return PTR_ERR(req);
S
Sage Weil 已提交
1842 1843 1844 1845

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

1846 1847
	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
1848 1849 1850 1851 1852 1853 1854 1855 1856

	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;
}
1857
EXPORT_SYMBOL(ceph_osdc_readpages);
S
Sage Weil 已提交
1858 1859 1860 1861 1862 1863 1864 1865 1866 1867

/*
 * 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,
1868
			 struct page **pages, int num_pages)
S
Sage Weil 已提交
1869 1870 1871
{
	struct ceph_osd_request *req;
	int rc = 0;
1872
	int page_align = off & ~PAGE_MASK;
S
Sage Weil 已提交
1873 1874 1875 1876

	BUG_ON(vino.snap != CEPH_NOSNAP);
	req = ceph_osdc_new_request(osdc, layout, vino, off, &len,
				    CEPH_OSD_OP_WRITE,
1877
				    CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE,
1878
				    snapc, 0,
S
Sage Weil 已提交
1879
				    truncate_seq, truncate_size, mtime,
1880
				    true, page_align);
1881 1882
	if (IS_ERR(req))
		return PTR_ERR(req);
S
Sage Weil 已提交
1883 1884 1885 1886 1887 1888

	/* 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);

1889
	rc = ceph_osdc_start_request(osdc, req, true);
S
Sage Weil 已提交
1890 1891 1892 1893 1894 1895 1896 1897 1898
	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;
}
1899
EXPORT_SYMBOL(ceph_osdc_writepages);
S
Sage Weil 已提交
1900 1901 1902 1903 1904 1905 1906

/*
 * handle incoming message
 */
static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
{
	struct ceph_osd *osd = con->private;
1907
	struct ceph_osd_client *osdc;
S
Sage Weil 已提交
1908 1909 1910
	int type = le16_to_cpu(msg->hdr.type);

	if (!osd)
S
Sage Weil 已提交
1911
		goto out;
1912
	osdc = osd->o_osdc;
S
Sage Weil 已提交
1913 1914 1915 1916 1917 1918

	switch (type) {
	case CEPH_MSG_OSD_MAP:
		ceph_osdc_handle_map(osdc, msg);
		break;
	case CEPH_MSG_OSD_OPREPLY:
1919
		handle_reply(osdc, msg, con);
S
Sage Weil 已提交
1920
		break;
1921 1922 1923
	case CEPH_MSG_WATCH_NOTIFY:
		handle_watch_notify(osdc, msg);
		break;
S
Sage Weil 已提交
1924 1925 1926 1927 1928

	default:
		pr_err("received unknown message type %d %s\n", type,
		       ceph_msg_type_name(type));
	}
S
Sage Weil 已提交
1929
out:
S
Sage Weil 已提交
1930 1931 1932
	ceph_msg_put(msg);
}

1933
/*
1934 1935
 * lookup and return message for incoming reply.  set up reply message
 * pages.
1936 1937
 */
static struct ceph_msg *get_reply(struct ceph_connection *con,
1938 1939
				  struct ceph_msg_header *hdr,
				  int *skip)
S
Sage Weil 已提交
1940 1941 1942
{
	struct ceph_osd *osd = con->private;
	struct ceph_osd_client *osdc = osd->o_osdc;
1943
	struct ceph_msg *m;
1944
	struct ceph_osd_request *req;
1945 1946
	int front = le32_to_cpu(hdr->front_len);
	int data_len = le32_to_cpu(hdr->data_len);
1947
	u64 tid;
S
Sage Weil 已提交
1948

1949 1950 1951 1952 1953 1954
	tid = le64_to_cpu(hdr->tid);
	mutex_lock(&osdc->request_mutex);
	req = __lookup_request(osdc, tid);
	if (!req) {
		*skip = 1;
		m = NULL;
1955 1956
		dout("get_reply unknown tid %llu from osd%d\n", tid,
		     osd->o_osd);
1957 1958
		goto out;
	}
S
Sage Weil 已提交
1959 1960

	if (req->r_con_filling_msg) {
1961
		dout("%s revoking msg %p from old con %p\n", __func__,
S
Sage Weil 已提交
1962
		     req->r_reply, req->r_con_filling_msg);
1963
		ceph_msg_revoke_incoming(req->r_reply);
1964
		req->r_con_filling_msg->ops->put(req->r_con_filling_msg);
1965
		req->r_con_filling_msg = NULL;
1966 1967
	}

S
Sage Weil 已提交
1968 1969 1970
	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);
1971
		m = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, front, GFP_NOFS, false);
1972
		if (!m)
S
Sage Weil 已提交
1973 1974 1975 1976 1977 1978
			goto out;
		ceph_msg_put(req->r_reply);
		req->r_reply = m;
	}
	m = ceph_msg_get(req->r_reply);

1979
	if (data_len > 0) {
1980
		int want = calc_pages_for(req->r_page_alignment, data_len);
1981

1982
		if (req->r_pages && unlikely(req->r_num_pages < want)) {
1983 1984 1985
			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);
1986 1987
			*skip = 1;
			ceph_msg_put(m);
1988
			m = NULL;
1989
			goto out;
1990
		}
1991 1992
		m->pages = req->r_pages;
		m->nr_pages = req->r_num_pages;
1993
		m->page_alignment = req->r_page_alignment;
1994 1995 1996
#ifdef CONFIG_BLOCK
		m->bio = req->r_bio;
#endif
1997
	}
1998
	*skip = 0;
1999
	req->r_con_filling_msg = con->ops->get(con);
S
Sage Weil 已提交
2000
	dout("get_reply tid %lld %p\n", tid, m);
2001 2002 2003

out:
	mutex_unlock(&osdc->request_mutex);
2004
	return m;
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

}

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
2016
	*skip = 0;
2017 2018
	switch (type) {
	case CEPH_MSG_OSD_MAP:
2019
	case CEPH_MSG_WATCH_NOTIFY:
2020
		return ceph_msg_new(type, front, GFP_NOFS, false);
2021 2022 2023 2024 2025 2026 2027 2028
	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
2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047
}

/*
 * 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);
}

2048 2049 2050
/*
 * authentication
 */
2051 2052 2053 2054 2055
/*
 * 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,
2056
					int *proto, int force_new)
2057 2058 2059 2060
{
	struct ceph_osd *o = con->private;
	struct ceph_osd_client *osdc = o->o_osdc;
	struct ceph_auth_client *ac = osdc->client->monc.auth;
2061
	struct ceph_auth_handshake *auth = &o->o_auth;
2062

2063
	if (force_new && auth->authorizer) {
2064 2065
		if (ac->ops && ac->ops->destroy_authorizer)
			ac->ops->destroy_authorizer(ac, auth->authorizer);
2066 2067
		auth->authorizer = NULL;
	}
2068
	if (!auth->authorizer && ac->ops && ac->ops->create_authorizer) {
2069 2070
		int ret = ac->ops->create_authorizer(ac, CEPH_ENTITY_TYPE_OSD,
							auth);
2071
		if (ret)
2072
			return ERR_PTR(ret);
2073 2074
	}
	*proto = ac->protocol;
2075

2076
	return auth;
2077 2078 2079 2080 2081 2082 2083 2084 2085
}


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;

2086 2087 2088 2089
	/*
	 * XXX If ac->ops or ac->ops->verify_authorizer_reply is null,
	 * XXX which do we do:  succeed or fail?
	 */
2090
	return ac->ops->verify_authorizer_reply(ac, o->o_auth.authorizer, len);
2091 2092
}

2093 2094 2095 2096 2097 2098
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;

2099
	if (ac->ops && ac->ops->invalidate_authorizer)
2100 2101 2102 2103
		ac->ops->invalidate_authorizer(ac, CEPH_ENTITY_TYPE_OSD);

	return ceph_monc_validate_auth(&osdc->client->monc);
}
2104

2105
static const struct ceph_connection_operations osd_con_ops = {
S
Sage Weil 已提交
2106 2107 2108
	.get = get_osd_con,
	.put = put_osd_con,
	.dispatch = dispatch,
2109 2110
	.get_authorizer = get_authorizer,
	.verify_authorizer_reply = verify_authorizer_reply,
2111
	.invalidate_authorizer = invalidate_authorizer,
S
Sage Weil 已提交
2112
	.alloc_msg = alloc_msg,
2113
	.fault = osd_reset,
S
Sage Weil 已提交
2114
};