pagelist.c 23.2 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
static bool nfs_pgarray_set(struct nfs_page_array *p, unsigned int pagecount)
F
Fred Isaman 已提交
33 34 35 36 37 38 39 40 41 42 43 44
{
	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;
}

45 46 47 48 49 50 51 52 53
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;
54
	hdr->dreq = desc->pg_dreq;
55
	hdr->layout_private = desc->pg_layout_private;
56
	hdr->release = release;
57
	hdr->completion_ops = desc->pg_completion_ops;
58 59
	if (hdr->completion_ops->init_hdr)
		hdr->completion_ops->init_hdr(hdr);
60
}
B
Bryan Schumaker 已提交
61
EXPORT_SYMBOL_GPL(nfs_pgheader_init);
62 63 64 65 66 67 68 69 70 71 72 73 74

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
75 76 77
static inline struct nfs_page *
nfs_page_alloc(void)
{
78
	struct nfs_page	*p = kmem_cache_zalloc(nfs_page_cachep, GFP_NOIO);
79
	if (p)
L
Linus Torvalds 已提交
80 81 82 83 84 85 86 87 88 89
		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);
}

90 91 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
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);
}

L
Linus Torvalds 已提交
139 140
/**
 * nfs_create_request - Create an NFS read/write request.
141
 * @ctx: open context to use
L
Linus Torvalds 已提交
142 143 144 145 146 147
 * @inode: inode to which the request is attached
 * @page: page to write
 * @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
148
 * create two different requests for the same page.
L
Linus Torvalds 已提交
149 150 151 152 153 154 155 156
 * User should ensure it is safe to sleep in this function.
 */
struct nfs_page *
nfs_create_request(struct nfs_open_context *ctx, struct inode *inode,
		   struct page *page,
		   unsigned int offset, unsigned int count)
{
	struct nfs_page		*req;
157
	struct nfs_lock_context *l_ctx;
L
Linus Torvalds 已提交
158

159 160
	if (test_bit(NFS_CONTEXT_BAD, &ctx->flags))
		return ERR_PTR(-EBADF);
161 162 163 164
	/* try to allocate the request struct */
	req = nfs_page_alloc();
	if (req == NULL)
		return ERR_PTR(-ENOMEM);
L
Linus Torvalds 已提交
165

166
	/* get lock context early so we can deal with alloc failures */
167 168
	l_ctx = nfs_get_lock_context(ctx);
	if (IS_ERR(l_ctx)) {
169
		nfs_page_free(req);
170
		return ERR_CAST(l_ctx);
171
	}
172
	req->wb_lock_context = l_ctx;
173
	nfs_iocounter_inc(&l_ctx->io_count);
174

L
Linus Torvalds 已提交
175 176 177 178
	/* 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;
179
	req->wb_index	= page_file_index(page);
L
Linus Torvalds 已提交
180 181 182 183 184
	page_cache_get(page);
	req->wb_offset  = offset;
	req->wb_pgbase	= offset;
	req->wb_bytes   = count;
	req->wb_context = get_nfs_open_context(ctx);
185
	kref_init(&req->wb_kref);
L
Linus Torvalds 已提交
186 187 188 189
	return req;
}

/**
190
 * nfs_unlock_request - Unlock request and wake up sleepers.
L
Linus Torvalds 已提交
191 192
 * @req:
 */
193
void nfs_unlock_request(struct nfs_page *req)
L
Linus Torvalds 已提交
194 195 196 197 198 199 200 201
{
	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();
202
	wake_up_bit(&req->wb_flags, PG_BUSY);
203 204 205
}

/**
206 207
 * nfs_unlock_and_release_request - Unlock request and release the nfs_page
 * @req:
208
 */
209
void nfs_unlock_and_release_request(struct nfs_page *req)
210
{
211
	nfs_unlock_request(req);
L
Linus Torvalds 已提交
212 213 214
	nfs_release_request(req);
}

215
/*
L
Linus Torvalds 已提交
216 217 218
 * nfs_clear_request - Free up all resources allocated to the request
 * @req:
 *
219 220
 * Release page and open context resources associated with a read/write
 * request after it has completed.
L
Linus Torvalds 已提交
221
 */
222
static void nfs_clear_request(struct nfs_page *req)
L
Linus Torvalds 已提交
223
{
224
	struct page *page = req->wb_page;
225
	struct nfs_open_context *ctx = req->wb_context;
226
	struct nfs_lock_context *l_ctx = req->wb_lock_context;
227

228 229
	if (page != NULL) {
		page_cache_release(page);
L
Linus Torvalds 已提交
230 231
		req->wb_page = NULL;
	}
232
	if (l_ctx != NULL) {
233
		nfs_iocounter_dec(&l_ctx->io_count);
234 235 236
		nfs_put_lock_context(l_ctx);
		req->wb_lock_context = NULL;
	}
237 238 239 240
	if (ctx != NULL) {
		put_nfs_open_context(ctx);
		req->wb_context = NULL;
	}
L
Linus Torvalds 已提交
241 242 243 244 245 246 247 248 249
}


/**
 * 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!
 */
250
static void nfs_free_request(struct kref *kref)
L
Linus Torvalds 已提交
251
{
252
	struct nfs_page *req = container_of(kref, struct nfs_page, wb_kref);
L
Linus Torvalds 已提交
253

254
	/* Release struct file and open context */
L
Linus Torvalds 已提交
255 256 257 258
	nfs_clear_request(req);
	nfs_page_free(req);
}

259 260 261 262 263
void nfs_release_request(struct nfs_page *req)
{
	kref_put(&req->wb_kref, nfs_free_request);
}

264 265 266 267 268 269
static int nfs_wait_bit_uninterruptible(void *word)
{
	io_schedule();
	return 0;
}

L
Linus Torvalds 已提交
270 271 272 273
/**
 * nfs_wait_on_request - Wait for a request to complete.
 * @req: request to wait upon.
 *
274
 * Interruptible by fatal signals only.
L
Linus Torvalds 已提交
275 276 277 278 279
 * The user is responsible for holding a count on the request.
 */
int
nfs_wait_on_request(struct nfs_page *req)
{
280 281 282
	return wait_on_bit(&req->wb_flags, PG_BUSY,
			nfs_wait_bit_uninterruptible,
			TASK_UNINTERRUPTIBLE);
L
Linus Torvalds 已提交
283 284
}

285
bool nfs_generic_pg_test(struct nfs_pageio_descriptor *desc, struct nfs_page *prev, struct nfs_page *req)
286 287 288 289 290 291 292 293 294 295 296 297 298
{
	/*
	 * 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;

	return desc->pg_count + req->wb_bytes <= desc->pg_bsize;
}
299
EXPORT_SYMBOL_GPL(nfs_generic_pg_test);
300

301 302 303 304 305
static inline struct nfs_rw_header *NFS_RW_HEADER(struct nfs_pgio_header *hdr)
{
	return container_of(hdr, struct nfs_rw_header, header);
}

306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336
/**
 * 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);

337 338 339 340 341
/**
 * nfs_pgio_data_alloc - Allocate pageio data
 * @hdr: The header making a request
 * @pagecount: Number of pages to create
 */
342 343
static struct nfs_pgio_data *nfs_pgio_data_alloc(struct nfs_pgio_header *hdr,
						 unsigned int pagecount)
344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391
{
	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);

392 393 394 395 396 397 398 399
/**
 * 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)
 */
400
static void nfs_pgio_rpcsetup(struct nfs_pgio_data *data,
401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435
			      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);
}

436 437 438 439 440
/**
 * nfs_pgio_prepare - Prepare pageio data to go over the wire
 * @task: The current task
 * @calldata: pageio data to prepare
 */
441
static void nfs_pgio_prepare(struct rpc_task *task, void *calldata)
442 443 444 445 446 447 448 449
{
	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);
}

450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495
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);

496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516
int nfs_do_multiple_pgios(struct list_head *head,
				 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;
}

517 518 519 520 521
/**
 * nfs_pgio_error - Clean up from a pageio error
 * @desc: IO descriptor
 * @hdr: pageio header
 */
522
static int nfs_pgio_error(struct nfs_pageio_descriptor *desc,
523 524 525 526 527 528 529 530 531 532 533 534 535 536
			  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;
}

537 538 539 540
/**
 * nfs_pgio_release - Release pageio data
 * @calldata: The pageio data to release
 */
541
static void nfs_pgio_release(void *calldata)
542 543 544 545 546 547 548
{
	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
549
/**
550 551
 * nfs_pageio_init - initialise a page io descriptor
 * @desc: pointer to descriptor
552 553 554 555
 * @inode: pointer to inode
 * @doio: pointer to io function
 * @bsize: io block size
 * @io_flags: extra parameters for the io function
556
 */
557 558
void nfs_pageio_init(struct nfs_pageio_descriptor *desc,
		     struct inode *inode,
559
		     const struct nfs_pageio_ops *pg_ops,
560
		     const struct nfs_pgio_completion_ops *compl_ops,
561
		     const struct nfs_rw_ops *rw_ops,
562
		     size_t bsize,
563
		     int io_flags)
564 565
{
	INIT_LIST_HEAD(&desc->pg_list);
566
	desc->pg_bytes_written = 0;
567 568 569
	desc->pg_count = 0;
	desc->pg_bsize = bsize;
	desc->pg_base = 0;
570
	desc->pg_moreio = 0;
571
	desc->pg_recoalesce = 0;
572
	desc->pg_inode = inode;
573
	desc->pg_ops = pg_ops;
574
	desc->pg_completion_ops = compl_ops;
575
	desc->pg_rw_ops = rw_ops;
576 577
	desc->pg_ioflags = io_flags;
	desc->pg_error = 0;
578
	desc->pg_lseg = NULL;
579
	desc->pg_dreq = NULL;
580
	desc->pg_layout_private = NULL;
581
}
B
Bryan Schumaker 已提交
582
EXPORT_SYMBOL_GPL(nfs_pageio_init);
583

584 585 586 587 588
/**
 * nfs_pgio_result - Basic pageio error handling
 * @task: The task that ran
 * @calldata: Pageio data to check
 */
589
static void nfs_pgio_result(struct rpc_task *task, void *calldata)
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604
{
	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);
}

605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699
/*
 * 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;
}

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);

700 701 702 703 704 705 706 707 708 709 710 711 712
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;
}

713 714 715 716 717 718 719 720 721 722 723
/**
 * 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'.
 */
724 725 726
static bool nfs_can_coalesce_requests(struct nfs_page *prev,
				      struct nfs_page *req,
				      struct nfs_pageio_descriptor *pgio)
727
{
728
	if (!nfs_match_open_context(req->wb_context, prev->wb_context))
729
		return false;
730 731
	if (req->wb_context->dentry->d_inode->i_flock != NULL &&
	    !nfs_match_lock_context(req->wb_lock_context, prev->wb_lock_context))
732
		return false;
733
	if (req->wb_pgbase != 0)
734
		return false;
735
	if (prev->wb_pgbase + prev->wb_bytes != PAGE_CACHE_SIZE)
736
		return false;
737 738
	if (req_offset(req) != req_offset(prev) + prev->wb_bytes)
		return false;
739
	return pgio->pg_ops->pg_test(pgio, prev, req);
740 741 742
}

/**
743
 * nfs_pageio_do_add_request - Attempt to coalesce a request into a page list.
744 745 746 747 748 749
 * @desc: destination io descriptor
 * @req: request
 *
 * Returns true if the request 'req' was successfully coalesced into the
 * existing list of pages 'desc'.
 */
750 751
static int nfs_pageio_do_add_request(struct nfs_pageio_descriptor *desc,
				     struct nfs_page *req)
752 753 754 755 756
{
	if (desc->pg_count != 0) {
		struct nfs_page *prev;

		prev = nfs_list_entry(desc->pg_list.prev);
757
		if (!nfs_can_coalesce_requests(prev, req, desc))
758
			return 0;
759
	} else {
760 761
		if (desc->pg_ops->pg_init)
			desc->pg_ops->pg_init(desc, req);
762
		desc->pg_base = req->wb_pgbase;
763
	}
764 765
	nfs_list_remove_request(req);
	nfs_list_add_request(req, &desc->pg_list);
766
	desc->pg_count += req->wb_bytes;
767 768 769
	return 1;
}

770 771 772 773 774 775
/*
 * 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)) {
776
		int error = desc->pg_ops->pg_doio(desc);
777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795
		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
 *
 * Returns true if the request 'req' was successfully coalesced into the
 * existing list of pages 'desc'.
 */
796
static int __nfs_pageio_add_request(struct nfs_pageio_descriptor *desc,
797
			   struct nfs_page *req)
798 799
{
	while (!nfs_pageio_do_add_request(desc, req)) {
800
		desc->pg_moreio = 1;
801 802 803
		nfs_pageio_doio(desc);
		if (desc->pg_error < 0)
			return 0;
804
		desc->pg_moreio = 0;
805 806
		if (desc->pg_recoalesce)
			return 0;
807 808 809 810
	}
	return 1;
}

811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851
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
852
EXPORT_SYMBOL_GPL(nfs_pageio_add_request);
853

854 855 856 857 858 859
/**
 * 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)
{
860 861 862 863 864 865 866
	for (;;) {
		nfs_pageio_doio(desc);
		if (!desc->pg_recoalesce)
			break;
		if (!nfs_do_recoalesce(desc))
			break;
	}
867
}
B
Bryan Schumaker 已提交
868
EXPORT_SYMBOL_GPL(nfs_pageio_complete);
869

870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885
/**
 * 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)
886
			nfs_pageio_complete(desc);
887 888 889
	}
}

D
David Howells 已提交
890
int __init nfs_init_nfspagecache(void)
L
Linus Torvalds 已提交
891 892 893 894
{
	nfs_page_cachep = kmem_cache_create("nfs_page",
					    sizeof(struct nfs_page),
					    0, SLAB_HWCACHE_ALIGN,
895
					    NULL);
L
Linus Torvalds 已提交
896 897 898 899 900 901
	if (nfs_page_cachep == NULL)
		return -ENOMEM;

	return 0;
}

902
void nfs_destroy_nfspagecache(void)
L
Linus Torvalds 已提交
903
{
904
	kmem_cache_destroy(nfs_page_cachep);
L
Linus Torvalds 已提交
905 906
}

907
static const struct rpc_call_ops nfs_pgio_common_ops = {
908 909 910 911
	.rpc_call_prepare = nfs_pgio_prepare,
	.rpc_call_done = nfs_pgio_result,
	.rpc_release = nfs_pgio_release,
};