pagelist.c 29.1 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13
/*
 * linux/fs/nfs/pagelist.c
 *
 * A set of helper functions for managing NFS read and write requests.
 * The main purpose of these routines is to provide support for the
 * coalescing of several requests into a single RPC call.
 *
 * Copyright 2000, 2001 (c) Trond Myklebust <trond.myklebust@fys.uio.no>
 *
 */

#include <linux/slab.h>
#include <linux/file.h>
A
Alexey Dobriyan 已提交
14
#include <linux/sched.h>
L
Linus Torvalds 已提交
15
#include <linux/sunrpc/clnt.h>
16
#include <linux/nfs.h>
L
Linus Torvalds 已提交
17 18 19 20 21
#include <linux/nfs3.h>
#include <linux/nfs4.h>
#include <linux/nfs_page.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_mount.h>
22
#include <linux/export.h>
L
Linus Torvalds 已提交
23

24
#include "internal.h"
25
#include "pnfs.h"
26

27 28
#define NFSDBG_FACILITY		NFSDBG_PAGECACHE

29
static struct kmem_cache *nfs_page_cachep;
30
static const struct rpc_call_ops nfs_pgio_common_ops;
L
Linus Torvalds 已提交
31

32 33
static void nfs_free_request(struct nfs_page *);

34
static bool nfs_pgarray_set(struct nfs_page_array *p, unsigned int pagecount)
F
Fred Isaman 已提交
35 36 37 38 39 40 41 42 43 44 45 46
{
	p->npages = pagecount;
	if (pagecount <= ARRAY_SIZE(p->page_array))
		p->pagevec = p->page_array;
	else {
		p->pagevec = kcalloc(pagecount, sizeof(struct page *), GFP_KERNEL);
		if (!p->pagevec)
			p->npages = 0;
	}
	return p->pagevec != NULL;
}

47 48 49 50 51 52 53 54 55
void nfs_pgheader_init(struct nfs_pageio_descriptor *desc,
		       struct nfs_pgio_header *hdr,
		       void (*release)(struct nfs_pgio_header *hdr))
{
	hdr->req = nfs_list_entry(desc->pg_list.next);
	hdr->inode = desc->pg_inode;
	hdr->cred = hdr->req->wb_context->cred;
	hdr->io_start = req_offset(hdr->req);
	hdr->good_bytes = desc->pg_count;
56
	hdr->dreq = desc->pg_dreq;
57
	hdr->layout_private = desc->pg_layout_private;
58
	hdr->release = release;
59
	hdr->completion_ops = desc->pg_completion_ops;
60 61
	if (hdr->completion_ops->init_hdr)
		hdr->completion_ops->init_hdr(hdr);
62
}
B
Bryan Schumaker 已提交
63
EXPORT_SYMBOL_GPL(nfs_pgheader_init);
64 65 66 67 68 69 70 71 72 73 74 75 76

void nfs_set_pgio_error(struct nfs_pgio_header *hdr, int error, loff_t pos)
{
	spin_lock(&hdr->lock);
	if (pos < hdr->io_start + hdr->good_bytes) {
		set_bit(NFS_IOHDR_ERROR, &hdr->flags);
		clear_bit(NFS_IOHDR_EOF, &hdr->flags);
		hdr->good_bytes = pos - hdr->io_start;
		hdr->error = error;
	}
	spin_unlock(&hdr->lock);
}

L
Linus Torvalds 已提交
77 78 79
static inline struct nfs_page *
nfs_page_alloc(void)
{
80
	struct nfs_page	*p = kmem_cache_zalloc(nfs_page_cachep, GFP_NOIO);
81
	if (p)
L
Linus Torvalds 已提交
82 83 84 85 86 87 88 89 90 91
		INIT_LIST_HEAD(&p->wb_list);
	return p;
}

static inline void
nfs_page_free(struct nfs_page *p)
{
	kmem_cache_free(nfs_page_cachep, p);
}

92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140
static void
nfs_iocounter_inc(struct nfs_io_counter *c)
{
	atomic_inc(&c->io_count);
}

static void
nfs_iocounter_dec(struct nfs_io_counter *c)
{
	if (atomic_dec_and_test(&c->io_count)) {
		clear_bit(NFS_IO_INPROGRESS, &c->flags);
		smp_mb__after_clear_bit();
		wake_up_bit(&c->flags, NFS_IO_INPROGRESS);
	}
}

static int
__nfs_iocounter_wait(struct nfs_io_counter *c)
{
	wait_queue_head_t *wq = bit_waitqueue(&c->flags, NFS_IO_INPROGRESS);
	DEFINE_WAIT_BIT(q, &c->flags, NFS_IO_INPROGRESS);
	int ret = 0;

	do {
		prepare_to_wait(wq, &q.wait, TASK_KILLABLE);
		set_bit(NFS_IO_INPROGRESS, &c->flags);
		if (atomic_read(&c->io_count) == 0)
			break;
		ret = nfs_wait_bit_killable(&c->flags);
	} while (atomic_read(&c->io_count) != 0);
	finish_wait(wq, &q.wait);
	return ret;
}

/**
 * nfs_iocounter_wait - wait for i/o to complete
 * @c: nfs_io_counter to use
 *
 * returns -ERESTARTSYS if interrupted by a fatal signal.
 * Otherwise returns 0 once the io_count hits 0.
 */
int
nfs_iocounter_wait(struct nfs_io_counter *c)
{
	if (atomic_read(&c->io_count) == 0)
		return 0;
	return __nfs_iocounter_wait(c);
}

141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280
/*
 * nfs_page_group_lock - lock the head of the page group
 * @req - request in group that is to be locked
 *
 * this lock must be held if modifying the page group list
 */
void
nfs_page_group_lock(struct nfs_page *req)
{
	struct nfs_page *head = req->wb_head;
	int err = -EAGAIN;

	WARN_ON_ONCE(head != head->wb_head);

	while (err)
		err = wait_on_bit_lock(&head->wb_flags, PG_HEADLOCK,
			nfs_wait_bit_killable, TASK_KILLABLE);
}

/*
 * nfs_page_group_unlock - unlock the head of the page group
 * @req - request in group that is to be unlocked
 */
void
nfs_page_group_unlock(struct nfs_page *req)
{
	struct nfs_page *head = req->wb_head;

	WARN_ON_ONCE(head != head->wb_head);

	smp_mb__before_clear_bit();
	clear_bit(PG_HEADLOCK, &head->wb_flags);
	smp_mb__after_clear_bit();
	wake_up_bit(&head->wb_flags, PG_HEADLOCK);
}

/*
 * nfs_page_group_sync_on_bit_locked
 *
 * must be called with page group lock held
 */
static bool
nfs_page_group_sync_on_bit_locked(struct nfs_page *req, unsigned int bit)
{
	struct nfs_page *head = req->wb_head;
	struct nfs_page *tmp;

	WARN_ON_ONCE(!test_bit(PG_HEADLOCK, &head->wb_flags));
	WARN_ON_ONCE(test_and_set_bit(bit, &req->wb_flags));

	tmp = req->wb_this_page;
	while (tmp != req) {
		if (!test_bit(bit, &tmp->wb_flags))
			return false;
		tmp = tmp->wb_this_page;
	}

	/* true! reset all bits */
	tmp = req;
	do {
		clear_bit(bit, &tmp->wb_flags);
		tmp = tmp->wb_this_page;
	} while (tmp != req);

	return true;
}

/*
 * nfs_page_group_sync_on_bit - set bit on current request, but only
 *   return true if the bit is set for all requests in page group
 * @req - request in page group
 * @bit - PG_* bit that is used to sync page group
 */
bool nfs_page_group_sync_on_bit(struct nfs_page *req, unsigned int bit)
{
	bool ret;

	nfs_page_group_lock(req);
	ret = nfs_page_group_sync_on_bit_locked(req, bit);
	nfs_page_group_unlock(req);

	return ret;
}

/*
 * nfs_page_group_init - Initialize the page group linkage for @req
 * @req - a new nfs request
 * @prev - the previous request in page group, or NULL if @req is the first
 *         or only request in the group (the head).
 */
static inline void
nfs_page_group_init(struct nfs_page *req, struct nfs_page *prev)
{
	WARN_ON_ONCE(prev == req);

	if (!prev) {
		req->wb_head = req;
		req->wb_this_page = req;
	} else {
		WARN_ON_ONCE(prev->wb_this_page != prev->wb_head);
		WARN_ON_ONCE(!test_bit(PG_HEADLOCK, &prev->wb_head->wb_flags));
		req->wb_head = prev->wb_head;
		req->wb_this_page = prev->wb_this_page;
		prev->wb_this_page = req;

		/* grab extra ref if head request has extra ref from
		 * the write/commit path to handle handoff between write
		 * and commit lists */
		if (test_bit(PG_INODE_REF, &prev->wb_head->wb_flags))
			kref_get(&req->wb_kref);
	}
}

/*
 * nfs_page_group_destroy - sync the destruction of page groups
 * @req - request that no longer needs the page group
 *
 * releases the page group reference from each member once all
 * members have called this function.
 */
static void
nfs_page_group_destroy(struct kref *kref)
{
	struct nfs_page *req = container_of(kref, struct nfs_page, wb_kref);
	struct nfs_page *tmp, *next;

	if (!nfs_page_group_sync_on_bit(req, PG_TEARDOWN))
		return;

	tmp = req;
	do {
		next = tmp->wb_this_page;
		/* unlink and free */
		tmp->wb_this_page = tmp;
		tmp->wb_head = tmp;
		nfs_free_request(tmp);
		tmp = next;
	} while (tmp != req);
}

L
Linus Torvalds 已提交
281 282
/**
 * nfs_create_request - Create an NFS read/write request.
283
 * @ctx: open context to use
L
Linus Torvalds 已提交
284
 * @page: page to write
285
 * @last: last nfs request created for this page group or NULL if head
L
Linus Torvalds 已提交
286 287 288 289
 * @offset: starting offset within the page for the write
 * @count: number of bytes to read/write
 *
 * The page must be locked by the caller. This makes sure we never
290
 * create two different requests for the same page.
L
Linus Torvalds 已提交
291 292 293
 * User should ensure it is safe to sleep in this function.
 */
struct nfs_page *
294
nfs_create_request(struct nfs_open_context *ctx, struct page *page,
295 296
		   struct nfs_page *last, unsigned int offset,
		   unsigned int count)
L
Linus Torvalds 已提交
297 298
{
	struct nfs_page		*req;
299
	struct nfs_lock_context *l_ctx;
L
Linus Torvalds 已提交
300

301 302
	if (test_bit(NFS_CONTEXT_BAD, &ctx->flags))
		return ERR_PTR(-EBADF);
303 304 305 306
	/* try to allocate the request struct */
	req = nfs_page_alloc();
	if (req == NULL)
		return ERR_PTR(-ENOMEM);
L
Linus Torvalds 已提交
307

308
	/* get lock context early so we can deal with alloc failures */
309 310
	l_ctx = nfs_get_lock_context(ctx);
	if (IS_ERR(l_ctx)) {
311
		nfs_page_free(req);
312
		return ERR_CAST(l_ctx);
313
	}
314
	req->wb_lock_context = l_ctx;
315
	nfs_iocounter_inc(&l_ctx->io_count);
316

L
Linus Torvalds 已提交
317 318 319 320
	/* Initialize the request struct. Initially, we assume a
	 * long write-back delay. This will be adjusted in
	 * update_nfs_request below if the region is not locked. */
	req->wb_page    = page;
321
	req->wb_index	= page_file_index(page);
L
Linus Torvalds 已提交
322 323 324 325 326
	page_cache_get(page);
	req->wb_offset  = offset;
	req->wb_pgbase	= offset;
	req->wb_bytes   = count;
	req->wb_context = get_nfs_open_context(ctx);
327
	kref_init(&req->wb_kref);
328
	nfs_page_group_init(req, last);
L
Linus Torvalds 已提交
329 330 331 332
	return req;
}

/**
333
 * nfs_unlock_request - Unlock request and wake up sleepers.
L
Linus Torvalds 已提交
334 335
 * @req:
 */
336
void nfs_unlock_request(struct nfs_page *req)
L
Linus Torvalds 已提交
337 338 339 340 341 342 343 344
{
	if (!NFS_WBACK_BUSY(req)) {
		printk(KERN_ERR "NFS: Invalid unlock attempted\n");
		BUG();
	}
	smp_mb__before_clear_bit();
	clear_bit(PG_BUSY, &req->wb_flags);
	smp_mb__after_clear_bit();
345
	wake_up_bit(&req->wb_flags, PG_BUSY);
346 347 348
}

/**
349 350
 * nfs_unlock_and_release_request - Unlock request and release the nfs_page
 * @req:
351
 */
352
void nfs_unlock_and_release_request(struct nfs_page *req)
353
{
354
	nfs_unlock_request(req);
L
Linus Torvalds 已提交
355 356 357
	nfs_release_request(req);
}

358
/*
L
Linus Torvalds 已提交
359 360 361
 * nfs_clear_request - Free up all resources allocated to the request
 * @req:
 *
362 363
 * Release page and open context resources associated with a read/write
 * request after it has completed.
L
Linus Torvalds 已提交
364
 */
365
static void nfs_clear_request(struct nfs_page *req)
L
Linus Torvalds 已提交
366
{
367
	struct page *page = req->wb_page;
368
	struct nfs_open_context *ctx = req->wb_context;
369
	struct nfs_lock_context *l_ctx = req->wb_lock_context;
370

371 372
	if (page != NULL) {
		page_cache_release(page);
L
Linus Torvalds 已提交
373 374
		req->wb_page = NULL;
	}
375
	if (l_ctx != NULL) {
376
		nfs_iocounter_dec(&l_ctx->io_count);
377 378 379
		nfs_put_lock_context(l_ctx);
		req->wb_lock_context = NULL;
	}
380 381 382 383
	if (ctx != NULL) {
		put_nfs_open_context(ctx);
		req->wb_context = NULL;
	}
L
Linus Torvalds 已提交
384 385 386 387 388 389 390 391
}

/**
 * nfs_release_request - Release the count on an NFS read/write request
 * @req: request to release
 *
 * Note: Should never be called with the spinlock held!
 */
392
static void nfs_free_request(struct nfs_page *req)
L
Linus Torvalds 已提交
393
{
394 395 396 397
	WARN_ON_ONCE(req->wb_this_page != req);

	/* extra debug: make sure no sync bits are still set */
	WARN_ON_ONCE(test_bit(PG_TEARDOWN, &req->wb_flags));
L
Linus Torvalds 已提交
398

399
	/* Release struct file and open context */
L
Linus Torvalds 已提交
400 401 402 403
	nfs_clear_request(req);
	nfs_page_free(req);
}

404 405
void nfs_release_request(struct nfs_page *req)
{
406
	kref_put(&req->wb_kref, nfs_page_group_destroy);
407 408
}

409 410 411 412 413 414
static int nfs_wait_bit_uninterruptible(void *word)
{
	io_schedule();
	return 0;
}

L
Linus Torvalds 已提交
415 416 417 418
/**
 * nfs_wait_on_request - Wait for a request to complete.
 * @req: request to wait upon.
 *
419
 * Interruptible by fatal signals only.
L
Linus Torvalds 已提交
420 421 422 423 424
 * The user is responsible for holding a count on the request.
 */
int
nfs_wait_on_request(struct nfs_page *req)
{
425 426 427
	return wait_on_bit(&req->wb_flags, PG_BUSY,
			nfs_wait_bit_uninterruptible,
			TASK_UNINTERRUPTIBLE);
L
Linus Torvalds 已提交
428 429
}

430 431 432 433 434 435 436 437 438 439 440
/*
 * nfs_generic_pg_test - determine if requests can be coalesced
 * @desc: pointer to descriptor
 * @prev: previous request in desc, or NULL
 * @req: this request
 *
 * Returns zero if @req can be coalesced into @desc, otherwise it returns
 * the size of the request.
 */
size_t nfs_generic_pg_test(struct nfs_pageio_descriptor *desc,
			   struct nfs_page *prev, struct nfs_page *req)
441
{
442 443
	if (!prev)
		return req->wb_bytes;
444 445 446 447 448 449 450 451 452 453
	/*
	 * FIXME: ideally we should be able to coalesce all requests
	 * that are not block boundary aligned, but currently this
	 * is problematic for the case of bsize < PAGE_CACHE_SIZE,
	 * since nfs_flush_multi and nfs_pagein_multi assume you
	 * can have only one struct nfs_page.
	 */
	if (desc->pg_bsize < PAGE_SIZE)
		return 0;

454 455 456
	if (desc->pg_count + req->wb_bytes <= desc->pg_bsize)
		return req->wb_bytes;
	return 0;
457
}
458
EXPORT_SYMBOL_GPL(nfs_generic_pg_test);
459

460 461 462 463 464
static inline struct nfs_rw_header *NFS_RW_HEADER(struct nfs_pgio_header *hdr)
{
	return container_of(hdr, struct nfs_rw_header, header);
}

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
/**
 * nfs_rw_header_alloc - Allocate a header for a read or write
 * @ops: Read or write function vector
 */
struct nfs_rw_header *nfs_rw_header_alloc(const struct nfs_rw_ops *ops)
{
	struct nfs_rw_header *header = ops->rw_alloc_header();

	if (header) {
		struct nfs_pgio_header *hdr = &header->header;

		INIT_LIST_HEAD(&hdr->pages);
		INIT_LIST_HEAD(&hdr->rpc_list);
		spin_lock_init(&hdr->lock);
		atomic_set(&hdr->refcnt, 0);
		hdr->rw_ops = ops;
	}
	return header;
}
EXPORT_SYMBOL_GPL(nfs_rw_header_alloc);

/*
 * nfs_rw_header_free - Free a read or write header
 * @hdr: The header to free
 */
void nfs_rw_header_free(struct nfs_pgio_header *hdr)
{
	hdr->rw_ops->rw_free_header(NFS_RW_HEADER(hdr));
}
EXPORT_SYMBOL_GPL(nfs_rw_header_free);

496 497 498 499 500
/**
 * nfs_pgio_data_alloc - Allocate pageio data
 * @hdr: The header making a request
 * @pagecount: Number of pages to create
 */
501 502
static struct nfs_pgio_data *nfs_pgio_data_alloc(struct nfs_pgio_header *hdr,
						 unsigned int pagecount)
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
{
	struct nfs_pgio_data *data, *prealloc;

	prealloc = &NFS_RW_HEADER(hdr)->rpc_data;
	if (prealloc->header == NULL)
		data = prealloc;
	else
		data = kzalloc(sizeof(*data), GFP_KERNEL);
	if (!data)
		goto out;

	if (nfs_pgarray_set(&data->pages, pagecount)) {
		data->header = hdr;
		atomic_inc(&hdr->refcnt);
	} else {
		if (data != prealloc)
			kfree(data);
		data = NULL;
	}
out:
	return data;
}

/**
 * nfs_pgio_data_release - Properly free pageio data
 * @data: The data to release
 */
void nfs_pgio_data_release(struct nfs_pgio_data *data)
{
	struct nfs_pgio_header *hdr = data->header;
	struct nfs_rw_header *pageio_header = NFS_RW_HEADER(hdr);

	put_nfs_open_context(data->args.context);
	if (data->pages.pagevec != data->pages.page_array)
		kfree(data->pages.pagevec);
	if (data == &pageio_header->rpc_data) {
		data->header = NULL;
		data = NULL;
	}
	if (atomic_dec_and_test(&hdr->refcnt))
		hdr->completion_ops->completion(hdr);
	/* Note: we only free the rpc_task after callbacks are done.
	 * See the comment in rpc_free_task() for why
	 */
	kfree(data);
}
EXPORT_SYMBOL_GPL(nfs_pgio_data_release);

551 552 553 554 555 556 557 558
/**
 * nfs_pgio_rpcsetup - Set up arguments for a pageio call
 * @data: The pageio data
 * @count: Number of bytes to read
 * @offset: Initial offset
 * @how: How to commit data (writes only)
 * @cinfo: Commit information for the call (writes only)
 */
559
static void nfs_pgio_rpcsetup(struct nfs_pgio_data *data,
560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594
			      unsigned int count, unsigned int offset,
			      int how, struct nfs_commit_info *cinfo)
{
	struct nfs_page *req = data->header->req;

	/* Set up the RPC argument and reply structs
	 * NB: take care not to mess about with data->commit et al. */

	data->args.fh     = NFS_FH(data->header->inode);
	data->args.offset = req_offset(req) + offset;
	/* pnfs_set_layoutcommit needs this */
	data->mds_offset = data->args.offset;
	data->args.pgbase = req->wb_pgbase + offset;
	data->args.pages  = data->pages.pagevec;
	data->args.count  = count;
	data->args.context = get_nfs_open_context(req->wb_context);
	data->args.lock_context = req->wb_lock_context;
	data->args.stable  = NFS_UNSTABLE;
	switch (how & (FLUSH_STABLE | FLUSH_COND_STABLE)) {
	case 0:
		break;
	case FLUSH_COND_STABLE:
		if (nfs_reqs_to_commit(cinfo))
			break;
	default:
		data->args.stable = NFS_FILE_SYNC;
	}

	data->res.fattr   = &data->fattr;
	data->res.count   = count;
	data->res.eof     = 0;
	data->res.verf    = &data->verf;
	nfs_fattr_init(&data->fattr);
}

595 596 597 598 599
/**
 * nfs_pgio_prepare - Prepare pageio data to go over the wire
 * @task: The current task
 * @calldata: pageio data to prepare
 */
600
static void nfs_pgio_prepare(struct rpc_task *task, void *calldata)
601 602 603 604 605 606 607 608
{
	struct nfs_pgio_data *data = calldata;
	int err;
	err = NFS_PROTO(data->header->inode)->pgio_rpc_prepare(task, data);
	if (err)
		rpc_exit(task, err);
}

609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654
int nfs_initiate_pgio(struct rpc_clnt *clnt, struct nfs_pgio_data *data,
		      const struct rpc_call_ops *call_ops, int how, int flags)
{
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_argp = &data->args,
		.rpc_resp = &data->res,
		.rpc_cred = data->header->cred,
	};
	struct rpc_task_setup task_setup_data = {
		.rpc_client = clnt,
		.task = &data->task,
		.rpc_message = &msg,
		.callback_ops = call_ops,
		.callback_data = data,
		.workqueue = nfsiod_workqueue,
		.flags = RPC_TASK_ASYNC | flags,
	};
	int ret = 0;

	data->header->rw_ops->rw_initiate(data, &msg, &task_setup_data, how);

	dprintk("NFS: %5u initiated pgio call "
		"(req %s/%llu, %u bytes @ offset %llu)\n",
		data->task.tk_pid,
		data->header->inode->i_sb->s_id,
		(unsigned long long)NFS_FILEID(data->header->inode),
		data->args.count,
		(unsigned long long)data->args.offset);

	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task)) {
		ret = PTR_ERR(task);
		goto out;
	}
	if (how & FLUSH_SYNC) {
		ret = rpc_wait_for_completion_task(task);
		if (ret == 0)
			ret = task->tk_status;
	}
	rpc_put_task(task);
out:
	return ret;
}
EXPORT_SYMBOL_GPL(nfs_initiate_pgio);

655
static int nfs_do_multiple_pgios(struct list_head *head,
656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675
				 const struct rpc_call_ops *call_ops,
				 int how)
{
	struct nfs_pgio_data *data;
	int ret = 0;

	while (!list_empty(head)) {
		int ret2;

		data = list_first_entry(head, struct nfs_pgio_data, list);
		list_del_init(&data->list);

		ret2 = nfs_initiate_pgio(NFS_CLIENT(data->header->inode),
					 data, call_ops, how, 0);
		if (ret == 0)
			 ret = ret2;
	}
	return ret;
}

676 677 678 679 680
/**
 * nfs_pgio_error - Clean up from a pageio error
 * @desc: IO descriptor
 * @hdr: pageio header
 */
681
static int nfs_pgio_error(struct nfs_pageio_descriptor *desc,
682 683 684 685 686 687 688 689 690 691 692 693 694 695
			  struct nfs_pgio_header *hdr)
{
	struct nfs_pgio_data *data;

	set_bit(NFS_IOHDR_REDO, &hdr->flags);
	while (!list_empty(&hdr->rpc_list)) {
		data = list_first_entry(&hdr->rpc_list, struct nfs_pgio_data, list);
		list_del(&data->list);
		nfs_pgio_data_release(data);
	}
	desc->pg_completion_ops->error_cleanup(&desc->pg_list);
	return -ENOMEM;
}

696 697 698 699
/**
 * nfs_pgio_release - Release pageio data
 * @calldata: The pageio data to release
 */
700
static void nfs_pgio_release(void *calldata)
701 702 703 704 705 706 707
{
	struct nfs_pgio_data *data = calldata;
	if (data->header->rw_ops->rw_release)
		data->header->rw_ops->rw_release(data);
	nfs_pgio_data_release(data);
}

L
Linus Torvalds 已提交
708
/**
709 710
 * nfs_pageio_init - initialise a page io descriptor
 * @desc: pointer to descriptor
711 712 713 714
 * @inode: pointer to inode
 * @doio: pointer to io function
 * @bsize: io block size
 * @io_flags: extra parameters for the io function
715
 */
716 717
void nfs_pageio_init(struct nfs_pageio_descriptor *desc,
		     struct inode *inode,
718
		     const struct nfs_pageio_ops *pg_ops,
719
		     const struct nfs_pgio_completion_ops *compl_ops,
720
		     const struct nfs_rw_ops *rw_ops,
721
		     size_t bsize,
722
		     int io_flags)
723 724
{
	INIT_LIST_HEAD(&desc->pg_list);
725
	desc->pg_bytes_written = 0;
726 727 728
	desc->pg_count = 0;
	desc->pg_bsize = bsize;
	desc->pg_base = 0;
729
	desc->pg_moreio = 0;
730
	desc->pg_recoalesce = 0;
731
	desc->pg_inode = inode;
732
	desc->pg_ops = pg_ops;
733
	desc->pg_completion_ops = compl_ops;
734
	desc->pg_rw_ops = rw_ops;
735 736
	desc->pg_ioflags = io_flags;
	desc->pg_error = 0;
737
	desc->pg_lseg = NULL;
738
	desc->pg_dreq = NULL;
739
	desc->pg_layout_private = NULL;
740
}
B
Bryan Schumaker 已提交
741
EXPORT_SYMBOL_GPL(nfs_pageio_init);
742

743 744 745 746 747
/**
 * nfs_pgio_result - Basic pageio error handling
 * @task: The task that ran
 * @calldata: Pageio data to check
 */
748
static void nfs_pgio_result(struct rpc_task *task, void *calldata)
749 750 751 752 753 754 755 756 757 758 759 760 761 762 763
{
	struct nfs_pgio_data *data = calldata;
	struct inode *inode = data->header->inode;

	dprintk("NFS: %s: %5u, (status %d)\n", __func__,
		task->tk_pid, task->tk_status);

	if (data->header->rw_ops->rw_done(task, data, inode) != 0)
		return;
	if (task->tk_status < 0)
		nfs_set_pgio_error(data->header, task->tk_status, data->args.offset);
	else
		data->header->rw_ops->rw_result(task, data);
}

764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849
/*
 * Generate multiple small requests to read or write a single
 * contiguous dirty on one page.
 */
static int nfs_pgio_multi(struct nfs_pageio_descriptor *desc,
			  struct nfs_pgio_header *hdr)
{
	struct nfs_page *req = hdr->req;
	struct page *page = req->wb_page;
	struct nfs_pgio_data *data;
	size_t wsize = desc->pg_bsize, nbytes;
	unsigned int offset;
	int requests = 0;
	struct nfs_commit_info cinfo;

	nfs_init_cinfo(&cinfo, desc->pg_inode, desc->pg_dreq);

	if ((desc->pg_ioflags & FLUSH_COND_STABLE) &&
	    (desc->pg_moreio || nfs_reqs_to_commit(&cinfo) ||
	     desc->pg_count > wsize))
		desc->pg_ioflags &= ~FLUSH_COND_STABLE;

	offset = 0;
	nbytes = desc->pg_count;
	do {
		size_t len = min(nbytes, wsize);

		data = nfs_pgio_data_alloc(hdr, 1);
		if (!data)
			return nfs_pgio_error(desc, hdr);
		data->pages.pagevec[0] = page;
		nfs_pgio_rpcsetup(data, len, offset, desc->pg_ioflags, &cinfo);
		list_add(&data->list, &hdr->rpc_list);
		requests++;
		nbytes -= len;
		offset += len;
	} while (nbytes != 0);

	nfs_list_remove_request(req);
	nfs_list_add_request(req, &hdr->pages);
	desc->pg_rpc_callops = &nfs_pgio_common_ops;
	return 0;
}

/*
 * Create an RPC task for the given read or write request and kick it.
 * The page must have been locked by the caller.
 *
 * It may happen that the page we're passed is not marked dirty.
 * This is the case if nfs_updatepage detects a conflicting request
 * that has been written but not committed.
 */
static int nfs_pgio_one(struct nfs_pageio_descriptor *desc,
			struct nfs_pgio_header *hdr)
{
	struct nfs_page		*req;
	struct page		**pages;
	struct nfs_pgio_data	*data;
	struct list_head *head = &desc->pg_list;
	struct nfs_commit_info cinfo;

	data = nfs_pgio_data_alloc(hdr, nfs_page_array_len(desc->pg_base,
							   desc->pg_count));
	if (!data)
		return nfs_pgio_error(desc, hdr);

	nfs_init_cinfo(&cinfo, desc->pg_inode, desc->pg_dreq);
	pages = data->pages.pagevec;
	while (!list_empty(head)) {
		req = nfs_list_entry(head->next);
		nfs_list_remove_request(req);
		nfs_list_add_request(req, &hdr->pages);
		*pages++ = req->wb_page;
	}

	if ((desc->pg_ioflags & FLUSH_COND_STABLE) &&
	    (desc->pg_moreio || nfs_reqs_to_commit(&cinfo)))
		desc->pg_ioflags &= ~FLUSH_COND_STABLE;

	/* Set up the argument struct */
	nfs_pgio_rpcsetup(data, desc->pg_count, 0, desc->pg_ioflags, &cinfo);
	list_add(&data->list, &hdr->rpc_list);
	desc->pg_rpc_callops = &nfs_pgio_common_ops;
	return 0;
}

850
static int nfs_generic_pg_pgios(struct nfs_pageio_descriptor *desc)
851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873
{
	struct nfs_rw_header *rw_hdr;
	struct nfs_pgio_header *hdr;
	int ret;

	rw_hdr = nfs_rw_header_alloc(desc->pg_rw_ops);
	if (!rw_hdr) {
		desc->pg_completion_ops->error_cleanup(&desc->pg_list);
		return -ENOMEM;
	}
	hdr = &rw_hdr->header;
	nfs_pgheader_init(desc, hdr, nfs_rw_header_free);
	atomic_inc(&hdr->refcnt);
	ret = nfs_generic_pgio(desc, hdr);
	if (ret == 0)
		ret = nfs_do_multiple_pgios(&hdr->rpc_list,
					    desc->pg_rpc_callops,
					    desc->pg_ioflags);
	if (atomic_dec_and_test(&hdr->refcnt))
		hdr->completion_ops->completion(hdr);
	return ret;
}

874 875 876 877 878 879 880 881 882
int nfs_generic_pgio(struct nfs_pageio_descriptor *desc,
		     struct nfs_pgio_header *hdr)
{
	if (desc->pg_bsize < PAGE_CACHE_SIZE)
		return nfs_pgio_multi(desc, hdr);
	return nfs_pgio_one(desc, hdr);
}
EXPORT_SYMBOL_GPL(nfs_generic_pgio);

883 884 885 886 887 888 889 890 891 892 893 894 895
static bool nfs_match_open_context(const struct nfs_open_context *ctx1,
		const struct nfs_open_context *ctx2)
{
	return ctx1->cred == ctx2->cred && ctx1->state == ctx2->state;
}

static bool nfs_match_lock_context(const struct nfs_lock_context *l1,
		const struct nfs_lock_context *l2)
{
	return l1->lockowner.l_owner == l2->lockowner.l_owner
		&& l1->lockowner.l_pid == l2->lockowner.l_pid;
}

896 897 898 899 900 901 902 903 904 905 906
/**
 * nfs_can_coalesce_requests - test two requests for compatibility
 * @prev: pointer to nfs_page
 * @req: pointer to nfs_page
 *
 * The nfs_page structures 'prev' and 'req' are compared to ensure that the
 * page data area they describe is contiguous, and that their RPC
 * credentials, NFSv4 open state, and lockowners are the same.
 *
 * Return 'true' if this is the case, else return 'false'.
 */
907 908 909
static bool nfs_can_coalesce_requests(struct nfs_page *prev,
				      struct nfs_page *req,
				      struct nfs_pageio_descriptor *pgio)
910
{
911 912
	size_t size;

913 914 915 916 917 918 919 920 921 922 923 924 925 926
	if (prev) {
		if (!nfs_match_open_context(req->wb_context, prev->wb_context))
			return false;
		if (req->wb_context->dentry->d_inode->i_flock != NULL &&
		    !nfs_match_lock_context(req->wb_lock_context,
					    prev->wb_lock_context))
			return false;
		if (req->wb_pgbase != 0)
			return false;
		if (prev->wb_pgbase + prev->wb_bytes != PAGE_CACHE_SIZE)
			return false;
		if (req_offset(req) != req_offset(prev) + prev->wb_bytes)
			return false;
	}
927 928 929
	size = pgio->pg_ops->pg_test(pgio, prev, req);
	WARN_ON_ONCE(size && size != req->wb_bytes);
	return size > 0;
930 931 932
}

/**
933
 * nfs_pageio_do_add_request - Attempt to coalesce a request into a page list.
934 935 936 937 938 939
 * @desc: destination io descriptor
 * @req: request
 *
 * Returns true if the request 'req' was successfully coalesced into the
 * existing list of pages 'desc'.
 */
940 941
static int nfs_pageio_do_add_request(struct nfs_pageio_descriptor *desc,
				     struct nfs_page *req)
942
{
943
	struct nfs_page *prev = NULL;
944 945
	if (desc->pg_count != 0) {
		prev = nfs_list_entry(desc->pg_list.prev);
946
	} else {
947 948
		if (desc->pg_ops->pg_init)
			desc->pg_ops->pg_init(desc, req);
949
		desc->pg_base = req->wb_pgbase;
950
	}
951 952
	if (!nfs_can_coalesce_requests(prev, req, desc))
		return 0;
953 954
	nfs_list_remove_request(req);
	nfs_list_add_request(req, &desc->pg_list);
955
	desc->pg_count += req->wb_bytes;
956 957 958
	return 1;
}

959 960 961 962 963 964
/*
 * Helper for nfs_pageio_add_request and nfs_pageio_complete
 */
static void nfs_pageio_doio(struct nfs_pageio_descriptor *desc)
{
	if (!list_empty(&desc->pg_list)) {
965
		int error = desc->pg_ops->pg_doio(desc);
966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981
		if (error < 0)
			desc->pg_error = error;
		else
			desc->pg_bytes_written += desc->pg_count;
	}
	if (list_empty(&desc->pg_list)) {
		desc->pg_count = 0;
		desc->pg_base = 0;
	}
}

/**
 * nfs_pageio_add_request - Attempt to coalesce a request into a page list.
 * @desc: destination io descriptor
 * @req: request
 *
982 983 984
 * This may split a request into subrequests which are all part of the
 * same page group.
 *
985 986 987
 * Returns true if the request 'req' was successfully coalesced into the
 * existing list of pages 'desc'.
 */
988
static int __nfs_pageio_add_request(struct nfs_pageio_descriptor *desc,
989
			   struct nfs_page *req)
990
{
991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041
	struct nfs_page *subreq;
	unsigned int bytes_left = 0;
	unsigned int offset, pgbase;

	nfs_page_group_lock(req);

	subreq = req;
	bytes_left = subreq->wb_bytes;
	offset = subreq->wb_offset;
	pgbase = subreq->wb_pgbase;

	do {
		if (!nfs_pageio_do_add_request(desc, subreq)) {
			/* make sure pg_test call(s) did nothing */
			WARN_ON_ONCE(subreq->wb_bytes != bytes_left);
			WARN_ON_ONCE(subreq->wb_offset != offset);
			WARN_ON_ONCE(subreq->wb_pgbase != pgbase);

			nfs_page_group_unlock(req);
			desc->pg_moreio = 1;
			nfs_pageio_doio(desc);
			if (desc->pg_error < 0)
				return 0;
			desc->pg_moreio = 0;
			if (desc->pg_recoalesce)
				return 0;
			/* retry add_request for this subreq */
			nfs_page_group_lock(req);
			continue;
		}

		/* check for buggy pg_test call(s) */
		WARN_ON_ONCE(subreq->wb_bytes + subreq->wb_pgbase > PAGE_SIZE);
		WARN_ON_ONCE(subreq->wb_bytes > bytes_left);
		WARN_ON_ONCE(subreq->wb_bytes == 0);

		bytes_left -= subreq->wb_bytes;
		offset += subreq->wb_bytes;
		pgbase += subreq->wb_bytes;

		if (bytes_left) {
			subreq = nfs_create_request(req->wb_context,
					req->wb_page,
					subreq, pgbase, bytes_left);
			nfs_lock_request(subreq);
			subreq->wb_offset  = offset;
			subreq->wb_index = req->wb_index;
		}
	} while (bytes_left > 0);

	nfs_page_group_unlock(req);
1042 1043 1044
	return 1;
}

1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
static int nfs_do_recoalesce(struct nfs_pageio_descriptor *desc)
{
	LIST_HEAD(head);

	do {
		list_splice_init(&desc->pg_list, &head);
		desc->pg_bytes_written -= desc->pg_count;
		desc->pg_count = 0;
		desc->pg_base = 0;
		desc->pg_recoalesce = 0;

		while (!list_empty(&head)) {
			struct nfs_page *req;

			req = list_first_entry(&head, struct nfs_page, wb_list);
			nfs_list_remove_request(req);
			if (__nfs_pageio_add_request(desc, req))
				continue;
			if (desc->pg_error < 0)
				return 0;
			break;
		}
	} while (desc->pg_recoalesce);
	return 1;
}

int nfs_pageio_add_request(struct nfs_pageio_descriptor *desc,
		struct nfs_page *req)
{
	int ret;

	do {
		ret = __nfs_pageio_add_request(desc, req);
		if (ret)
			break;
		if (desc->pg_error < 0)
			break;
		ret = nfs_do_recoalesce(desc);
	} while (ret);
	return ret;
}
B
Bryan Schumaker 已提交
1086
EXPORT_SYMBOL_GPL(nfs_pageio_add_request);
1087

1088 1089 1090 1091 1092 1093
/**
 * nfs_pageio_complete - Complete I/O on an nfs_pageio_descriptor
 * @desc: pointer to io descriptor
 */
void nfs_pageio_complete(struct nfs_pageio_descriptor *desc)
{
1094 1095 1096 1097 1098 1099 1100
	for (;;) {
		nfs_pageio_doio(desc);
		if (!desc->pg_recoalesce)
			break;
		if (!nfs_do_recoalesce(desc))
			break;
	}
1101
}
B
Bryan Schumaker 已提交
1102
EXPORT_SYMBOL_GPL(nfs_pageio_complete);
1103

1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119
/**
 * nfs_pageio_cond_complete - Conditional I/O completion
 * @desc: pointer to io descriptor
 * @index: page index
 *
 * It is important to ensure that processes don't try to take locks
 * on non-contiguous ranges of pages as that might deadlock. This
 * function should be called before attempting to wait on a locked
 * nfs_page. It will complete the I/O if the page index 'index'
 * is not contiguous with the existing list of pages in 'desc'.
 */
void nfs_pageio_cond_complete(struct nfs_pageio_descriptor *desc, pgoff_t index)
{
	if (!list_empty(&desc->pg_list)) {
		struct nfs_page *prev = nfs_list_entry(desc->pg_list.prev);
		if (index != prev->wb_index + 1)
1120
			nfs_pageio_complete(desc);
1121 1122 1123
	}
}

D
David Howells 已提交
1124
int __init nfs_init_nfspagecache(void)
L
Linus Torvalds 已提交
1125 1126 1127 1128
{
	nfs_page_cachep = kmem_cache_create("nfs_page",
					    sizeof(struct nfs_page),
					    0, SLAB_HWCACHE_ALIGN,
1129
					    NULL);
L
Linus Torvalds 已提交
1130 1131 1132 1133 1134 1135
	if (nfs_page_cachep == NULL)
		return -ENOMEM;

	return 0;
}

1136
void nfs_destroy_nfspagecache(void)
L
Linus Torvalds 已提交
1137
{
1138
	kmem_cache_destroy(nfs_page_cachep);
L
Linus Torvalds 已提交
1139 1140
}

1141
static const struct rpc_call_ops nfs_pgio_common_ops = {
1142 1143 1144 1145
	.rpc_call_prepare = nfs_pgio_prepare,
	.rpc_call_done = nfs_pgio_result,
	.rpc_release = nfs_pgio_release,
};
1146 1147 1148 1149 1150

const struct nfs_pageio_ops nfs_pgio_rw_ops = {
	.pg_test = nfs_generic_pg_test,
	.pg_doio = nfs_generic_pg_pgios,
};