ops_address.c 19.3 KB
Newer Older
D
David Teigland 已提交
1 2
/*
 * Copyright (C) Sistina Software, Inc.  1997-2003 All rights reserved.
3
 * Copyright (C) 2004-2006 Red Hat, Inc.  All rights reserved.
D
David Teigland 已提交
4 5 6
 *
 * This copyrighted material is made available to anyone wishing to use,
 * modify, copy, or redistribute it subject to the terms and conditions
7
 * of the GNU General Public License version 2.
D
David Teigland 已提交
8 9 10 11 12 13 14 15
 */

#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/completion.h>
#include <linux/buffer_head.h>
#include <linux/pagemap.h>
S
Steven Whitehouse 已提交
16
#include <linux/pagevec.h>
17
#include <linux/mpage.h>
18
#include <linux/fs.h>
19
#include <linux/gfs2_ondisk.h>
20
#include <linux/lm_interface.h>
D
David Teigland 已提交
21 22

#include "gfs2.h"
23
#include "incore.h"
D
David Teigland 已提交
24 25 26 27 28 29 30 31
#include "bmap.h"
#include "glock.h"
#include "inode.h"
#include "log.h"
#include "meta_io.h"
#include "ops_address.h"
#include "quota.h"
#include "trans.h"
32
#include "rgrp.h"
33
#include "ops_file.h"
34
#include "util.h"
S
Steven Whitehouse 已提交
35
#include "glops.h"
D
David Teigland 已提交
36

S
Steven Whitehouse 已提交
37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54

static void gfs2_page_add_databufs(struct gfs2_inode *ip, struct page *page,
				   unsigned int from, unsigned int to)
{
	struct buffer_head *head = page_buffers(page);
	unsigned int bsize = head->b_size;
	struct buffer_head *bh;
	unsigned int start, end;

	for (bh = head, start = 0; bh != head || !start;
	     bh = bh->b_this_page, start = end) {
		end = start + bsize;
		if (end <= from || start >= to)
			continue;
		gfs2_trans_add_bh(ip->i_gl, bh, 0);
	}
}

D
David Teigland 已提交
55
/**
56
 * gfs2_get_block - Fills in a buffer head with details about a block
D
David Teigland 已提交
57 58 59 60 61 62 63 64
 * @inode: The inode
 * @lblock: The block number to look up
 * @bh_result: The buffer head to return the result in
 * @create: Non-zero if we may add block to the file
 *
 * Returns: errno
 */

65 66
int gfs2_get_block(struct inode *inode, sector_t lblock,
	           struct buffer_head *bh_result, int create)
D
David Teigland 已提交
67
{
68
	return gfs2_block_map(inode, lblock, create, bh_result);
D
David Teigland 已提交
69 70 71
}

/**
72
 * gfs2_get_block_noalloc - Fills in a buffer head with details about a block
D
David Teigland 已提交
73 74 75 76 77 78 79 80
 * @inode: The inode
 * @lblock: The block number to look up
 * @bh_result: The buffer head to return the result in
 * @create: Non-zero if we may add block to the file
 *
 * Returns: errno
 */

81 82
static int gfs2_get_block_noalloc(struct inode *inode, sector_t lblock,
				  struct buffer_head *bh_result, int create)
D
David Teigland 已提交
83 84 85
{
	int error;

86
	error = gfs2_block_map(inode, lblock, 0, bh_result);
D
David Teigland 已提交
87 88
	if (error)
		return error;
89 90 91
	if (bh_result->b_blocknr == 0)
		return -EIO;
	return 0;
D
David Teigland 已提交
92 93
}

94 95
static int gfs2_get_block_direct(struct inode *inode, sector_t lblock,
				 struct buffer_head *bh_result, int create)
96
{
97
	return gfs2_block_map(inode, lblock, 0, bh_result);
98
}
99

D
David Teigland 已提交
100 101 102 103 104 105
/**
 * gfs2_writepage - Write complete page
 * @page: Page to write
 *
 * Returns: errno
 *
106 107
 * Some of this is copied from block_write_full_page() although we still
 * call it to do most of the work.
D
David Teigland 已提交
108 109 110 111
 */

static int gfs2_writepage(struct page *page, struct writeback_control *wbc)
{
112
	struct inode *inode = page->mapping->host;
113 114
	struct gfs2_inode *ip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);
115 116 117
	loff_t i_size = i_size_read(inode);
	pgoff_t end_index = i_size >> PAGE_CACHE_SHIFT;
	unsigned offset;
D
David Teigland 已提交
118
	int error;
119
	int done_trans = 0;
D
David Teigland 已提交
120 121 122 123 124

	if (gfs2_assert_withdraw(sdp, gfs2_glock_is_held_excl(ip->i_gl))) {
		unlock_page(page);
		return -EIO;
	}
125
	if (current->journal_info)
126 127 128 129
		goto out_ignore;

	/* Is the page fully outside i_size? (truncate in progress) */
        offset = i_size & (PAGE_CACHE_SIZE-1);
130
	if (page->index > end_index || (page->index == end_index && !offset)) {
131
		page->mapping->a_ops->invalidatepage(page, 0);
D
David Teigland 已提交
132
		unlock_page(page);
133
		return 0; /* don't care */
D
David Teigland 已提交
134 135
	}

136 137 138 139
	if (sdp->sd_args.ar_data == GFS2_DATA_ORDERED || gfs2_is_jdata(ip)) {
		error = gfs2_trans_begin(sdp, RES_DINODE + 1, 0);
		if (error)
			goto out_ignore;
140 141 142 143
		if (!page_has_buffers(page)) {
			create_empty_buffers(page, inode->i_sb->s_blocksize,
					     (1 << BH_Dirty)|(1 << BH_Uptodate));
		}
144 145 146
		gfs2_page_add_databufs(ip, page, 0, sdp->sd_vfs->s_blocksize-1);
		done_trans = 1;
	}
147
	error = block_write_full_page(page, gfs2_get_block_noalloc, wbc);
148 149
	if (done_trans)
		gfs2_trans_end(sdp);
D
David Teigland 已提交
150 151
	gfs2_meta_cache_flush(ip);
	return error;
152 153 154 155 156

out_ignore:
	redirty_page_for_writepage(wbc, page);
	unlock_page(page);
	return 0;
D
David Teigland 已提交
157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172
}

/**
 * stuffed_readpage - Fill in a Linux page with stuffed file data
 * @ip: the inode
 * @page: the page
 *
 * Returns: errno
 */

static int stuffed_readpage(struct gfs2_inode *ip, struct page *page)
{
	struct buffer_head *dibh;
	void *kaddr;
	int error;

173
	BUG_ON(page->index);
S
Steven Whitehouse 已提交
174

D
David Teigland 已提交
175 176 177 178
	error = gfs2_meta_inode_buffer(ip, &dibh);
	if (error)
		return error;

179
	kaddr = kmap_atomic(page, KM_USER0);
S
Steven Whitehouse 已提交
180
	memcpy(kaddr, dibh->b_data + sizeof(struct gfs2_dinode),
D
David Teigland 已提交
181
	       ip->i_di.di_size);
S
Steven Whitehouse 已提交
182
	memset(kaddr + ip->i_di.di_size, 0, PAGE_CACHE_SIZE - ip->i_di.di_size);
183
	kunmap_atomic(kaddr, KM_USER0);
D
David Teigland 已提交
184 185 186 187 188 189 190 191 192 193 194

	brelse(dibh);

	SetPageUptodate(page);

	return 0;
}


/**
 * gfs2_readpage - readpage with locking
195 196
 * @file: The file to read a page for. N.B. This may be NULL if we are
 * reading an internal file.
D
David Teigland 已提交
197 198 199 200 201 202 203
 * @page: The page to read
 *
 * Returns: errno
 */

static int gfs2_readpage(struct file *file, struct page *page)
{
204 205
	struct gfs2_inode *ip = GFS2_I(page->mapping->host);
	struct gfs2_sbd *sdp = GFS2_SB(page->mapping->host);
206
	struct gfs2_file *gf = NULL;
207
	struct gfs2_holder gh;
D
David Teigland 已提交
208
	int error;
209
	int do_unlock = 0;
D
David Teigland 已提交
210

211
	if (likely(file != &gfs2_internal_file_sentinel)) {
212
		if (file) {
213
			gf = file->private_data;
214
			if (test_bit(GFF_EXLOCK, &gf->f_flags))
215
				/* gfs2_sharewrite_nopage has grabbed the ip->i_gl already */
216 217
				goto skip_lock;
		}
218
		gfs2_holder_init(ip->i_gl, LM_ST_SHARED, GL_ATIME|LM_FLAG_TRY_1CB, &gh);
219
		do_unlock = 1;
220
		error = gfs2_glock_nq_atime(&gh);
S
Steven Whitehouse 已提交
221
		if (unlikely(error))
222 223
			goto out_unlock;
	}
D
David Teigland 已提交
224

225
skip_lock:
226
	if (gfs2_is_stuffed(ip)) {
S
Steven Whitehouse 已提交
227 228
		error = stuffed_readpage(ip, page);
		unlock_page(page);
D
David Teigland 已提交
229
	} else
230
		error = mpage_readpage(page, gfs2_get_block);
D
David Teigland 已提交
231 232 233 234

	if (unlikely(test_bit(SDF_SHUTDOWN, &sdp->sd_flags)))
		error = -EIO;

235
	if (do_unlock) {
236 237 238
		gfs2_glock_dq_m(1, &gh);
		gfs2_holder_uninit(&gh);
	}
239
out:
D
David Teigland 已提交
240
	return error;
241
out_unlock:
242 243
	if (error == GLR_TRYFAILED)
		error = AOP_TRUNCATED_PAGE;
244
	unlock_page(page);
245
	if (do_unlock)
S
Steven Whitehouse 已提交
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268
		gfs2_holder_uninit(&gh);
	goto out;
}

/**
 * gfs2_readpages - Read a bunch of pages at once
 *
 * Some notes:
 * 1. This is only for readahead, so we can simply ignore any things
 *    which are slightly inconvenient (such as locking conflicts between
 *    the page lock and the glock) and return having done no I/O. Its
 *    obviously not something we'd want to do on too regular a basis.
 *    Any I/O we ignore at this time will be done via readpage later.
 * 2. We have to handle stuffed files here too.
 * 3. mpage_readpages() does most of the heavy lifting in the common case.
 * 4. gfs2_get_block() is relied upon to set BH_Boundary in the right places.
 * 5. We use LM_FLAG_TRY_1CB here, effectively we then have lock-ahead as
 *    well as read-ahead.
 */
static int gfs2_readpages(struct file *file, struct address_space *mapping,
			  struct list_head *pages, unsigned nr_pages)
{
	struct inode *inode = mapping->host;
269 270
	struct gfs2_inode *ip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);
S
Steven Whitehouse 已提交
271 272 273
	struct gfs2_holder gh;
	unsigned page_idx;
	int ret;
274
	int do_unlock = 0;
S
Steven Whitehouse 已提交
275

276
	if (likely(file != &gfs2_internal_file_sentinel)) {
277 278 279 280 281
		if (file) {
			struct gfs2_file *gf = file->private_data;
			if (test_bit(GFF_EXLOCK, &gf->f_flags))
				goto skip_lock;
		}
S
Steven Whitehouse 已提交
282
		gfs2_holder_init(ip->i_gl, LM_ST_SHARED,
283
				 LM_FLAG_TRY_1CB|GL_ATIME, &gh);
284
		do_unlock = 1;
285
		ret = gfs2_glock_nq_atime(&gh);
286
		if (ret == GLR_TRYFAILED)
S
Steven Whitehouse 已提交
287 288 289 290
			goto out_noerror;
		if (unlikely(ret))
			goto out_unlock;
	}
291
skip_lock:
S
Steven Whitehouse 已提交
292 293 294 295
	if (gfs2_is_stuffed(ip)) {
		struct pagevec lru_pvec;
		pagevec_init(&lru_pvec, 0);
		for (page_idx = 0; page_idx < nr_pages; page_idx++) {
296 297
			struct page *page = list_entry(pages->prev, struct page, lru);
			prefetchw(&page->flags);
S
Steven Whitehouse 已提交
298 299 300 301 302 303 304
			list_del(&page->lru);
			if (!add_to_page_cache(page, mapping,
					       page->index, GFP_KERNEL)) {
				ret = stuffed_readpage(ip, page);
				unlock_page(page);
				if (!pagevec_add(&lru_pvec, page))
					 __pagevec_lru_add(&lru_pvec);
305 306
			} else {
				page_cache_release(page);
S
Steven Whitehouse 已提交
307 308 309 310 311 312 313 314 315
			}
		}
		pagevec_lru_add(&lru_pvec);
		ret = 0;
	} else {
		/* What we really want to do .... */
		ret = mpage_readpages(mapping, pages, nr_pages, gfs2_get_block);
	}

316
	if (do_unlock) {
S
Steven Whitehouse 已提交
317 318 319 320 321 322 323 324 325 326
		gfs2_glock_dq_m(1, &gh);
		gfs2_holder_uninit(&gh);
	}
out:
	if (unlikely(test_bit(SDF_SHUTDOWN, &sdp->sd_flags)))
		ret = -EIO;
	return ret;
out_noerror:
	ret = 0;
out_unlock:
327
	if (do_unlock)
S
Steven Whitehouse 已提交
328
		gfs2_holder_uninit(&gh);
329
	goto out;
D
David Teigland 已提交
330 331 332 333 334 335 336 337 338 339 340 341 342 343 344
}

/**
 * gfs2_prepare_write - Prepare to write a page to a file
 * @file: The file to write to
 * @page: The page which is to be prepared for writing
 * @from: From (byte range within page)
 * @to: To (byte range within page)
 *
 * Returns: errno
 */

static int gfs2_prepare_write(struct file *file, struct page *page,
			      unsigned from, unsigned to)
{
345 346
	struct gfs2_inode *ip = GFS2_I(page->mapping->host);
	struct gfs2_sbd *sdp = GFS2_SB(page->mapping->host);
347 348
	unsigned int data_blocks, ind_blocks, rblocks;
	int alloc_required;
D
David Teigland 已提交
349
	int error = 0;
350 351 352
	loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + from;
	loff_t end = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to;
	struct gfs2_alloc *al;
353 354
	unsigned int write_len = to - from;

D
David Teigland 已提交
355

356
	gfs2_holder_init(ip->i_gl, LM_ST_EXCLUSIVE, GL_ATIME|LM_FLAG_TRY_1CB, &ip->i_gh);
357
	error = gfs2_glock_nq_atime(&ip->i_gh);
358 359 360
	if (unlikely(error)) {
		if (error == GLR_TRYFAILED)
			error = AOP_TRUNCATED_PAGE;
361
		goto out_uninit;
362
	}
D
David Teigland 已提交
363

364
	gfs2_write_calc_reserv(ip, write_len, &data_blocks, &ind_blocks);
365

366
	error = gfs2_write_alloc_required(ip, pos, write_len, &alloc_required);
367 368
	if (error)
		goto out_unlock;
D
David Teigland 已提交
369

370

371
	ip->i_alloc.al_requested = 0;
372 373 374 375 376 377 378
	if (alloc_required) {
		al = gfs2_alloc_get(ip);

		error = gfs2_quota_lock(ip, NO_QUOTA_CHANGE, NO_QUOTA_CHANGE);
		if (error)
			goto out_alloc_put;

379
		error = gfs2_quota_check(ip, ip->i_inode.i_uid, ip->i_inode.i_gid);
380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
		if (error)
			goto out_qunlock;

		al->al_requested = data_blocks + ind_blocks;
		error = gfs2_inplace_reserve(ip);
		if (error)
			goto out_qunlock;
	}

	rblocks = RES_DINODE + ind_blocks;
	if (gfs2_is_jdata(ip))
		rblocks += data_blocks ? data_blocks : 1;
	if (ind_blocks || data_blocks)
		rblocks += RES_STATFS + RES_QUOTA;

	error = gfs2_trans_begin(sdp, rblocks, 0);
	if (error)
		goto out;

	if (gfs2_is_stuffed(ip)) {
		if (end > sdp->sd_sb.sb_bsize - sizeof(struct gfs2_dinode)) {
401
			error = gfs2_unstuff_dinode(ip, page);
402 403 404
			if (error == 0)
				goto prepare_write;
		} else if (!PageUptodate(page))
D
David Teigland 已提交
405
			error = stuffed_readpage(ip, page);
406
		goto out;
407 408
	}

409
prepare_write:
410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426
	error = block_prepare_write(page, from, to, gfs2_get_block);

out:
	if (error) {
		gfs2_trans_end(sdp);
		if (alloc_required) {
			gfs2_inplace_release(ip);
out_qunlock:
			gfs2_quota_unlock(ip);
out_alloc_put:
			gfs2_alloc_put(ip);
		}
out_unlock:
		gfs2_glock_dq_m(1, &ip->i_gh);
out_uninit:
		gfs2_holder_uninit(&ip->i_gh);
	}
D
David Teigland 已提交
427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444

	return error;
}

/**
 * gfs2_commit_write - Commit write to a file
 * @file: The file to write to
 * @page: The page containing the data
 * @from: From (byte range within page)
 * @to: To (byte range within page)
 *
 * Returns: errno
 */

static int gfs2_commit_write(struct file *file, struct page *page,
			     unsigned from, unsigned to)
{
	struct inode *inode = page->mapping->host;
445 446
	struct gfs2_inode *ip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);
447 448
	int error = -EOPNOTSUPP;
	struct buffer_head *dibh;
449 450
	struct gfs2_alloc *al = &ip->i_alloc;
	struct gfs2_dinode *di;
D
David Teigland 已提交
451

452 453 454 455 456 457 458 459
	if (gfs2_assert_withdraw(sdp, gfs2_glock_is_locked_by_me(ip->i_gl)))
                goto fail_nounlock;

	error = gfs2_meta_inode_buffer(ip, &dibh);
	if (error)
		goto fail_endtrans;

	gfs2_trans_add_bh(ip->i_gl, dibh, 1);
460
	di = (struct gfs2_dinode *)dibh->b_data;
461

D
David Teigland 已提交
462
	if (gfs2_is_stuffed(ip)) {
463
		u64 file_size;
D
David Teigland 已提交
464 465
		void *kaddr;

466
		file_size = ((u64)page->index << PAGE_CACHE_SHIFT) + to;
D
David Teigland 已提交
467

468
		kaddr = kmap_atomic(page, KM_USER0);
D
David Teigland 已提交
469
		memcpy(dibh->b_data + sizeof(struct gfs2_dinode) + from,
470
		       kaddr + from, to - from);
471
		kunmap_atomic(kaddr, KM_USER0);
D
David Teigland 已提交
472 473 474

		SetPageUptodate(page);

475
		if (inode->i_size < file_size) {
D
David Teigland 已提交
476
			i_size_write(inode, file_size);
477 478
			mark_inode_dirty(inode);
		}
D
David Teigland 已提交
479
	} else {
480 481
		if (sdp->sd_args.ar_data == GFS2_DATA_ORDERED ||
		    gfs2_is_jdata(ip))
482
			gfs2_page_add_databufs(ip, page, from, to);
D
David Teigland 已提交
483 484 485 486 487
		error = generic_commit_write(file, page, from, to);
		if (error)
			goto fail;
	}

488
	if (ip->i_di.di_size < inode->i_size) {
489
		ip->i_di.di_size = inode->i_size;
490 491 492
		di->di_size = cpu_to_be64(inode->i_size);
	}

493 494 495 496 497 498 499 500 501
	brelse(dibh);
	gfs2_trans_end(sdp);
	if (al->al_requested) {
		gfs2_inplace_release(ip);
		gfs2_quota_unlock(ip);
		gfs2_alloc_put(ip);
	}
	gfs2_glock_dq_m(1, &ip->i_gh);
	gfs2_holder_uninit(&ip->i_gh);
D
David Teigland 已提交
502 503
	return 0;

504 505 506 507 508 509 510 511 512 513 514 515
fail:
	brelse(dibh);
fail_endtrans:
	gfs2_trans_end(sdp);
	if (al->al_requested) {
		gfs2_inplace_release(ip);
		gfs2_quota_unlock(ip);
		gfs2_alloc_put(ip);
	}
	gfs2_glock_dq_m(1, &ip->i_gh);
	gfs2_holder_uninit(&ip->i_gh);
fail_nounlock:
D
David Teigland 已提交
516 517 518 519 520 521 522 523 524 525 526 527 528 529
	ClearPageUptodate(page);
	return error;
}

/**
 * gfs2_bmap - Block map function
 * @mapping: Address space info
 * @lblock: The block to map
 *
 * Returns: The disk address for the block or 0 on hole or error
 */

static sector_t gfs2_bmap(struct address_space *mapping, sector_t lblock)
{
530
	struct gfs2_inode *ip = GFS2_I(mapping->host);
D
David Teigland 已提交
531 532 533 534 535 536 537 538 539
	struct gfs2_holder i_gh;
	sector_t dblock = 0;
	int error;

	error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, LM_FLAG_ANY, &i_gh);
	if (error)
		return 0;

	if (!gfs2_is_stuffed(ip))
540
		dblock = generic_block_bmap(mapping, lblock, gfs2_get_block);
D
David Teigland 已提交
541 542 543 544 545 546 547 548

	gfs2_glock_dq_uninit(&i_gh);

	return dblock;
}

static void discard_buffer(struct gfs2_sbd *sdp, struct buffer_head *bh)
{
549
	struct gfs2_bufdata *bd;
D
David Teigland 已提交
550 551

	gfs2_log_lock(sdp);
552
	bd = bh->b_private;
553 554
	if (bd) {
		bd->bd_bh = NULL;
555
		bh->b_private = NULL;
556 557
	}
	gfs2_log_unlock(sdp);
D
David Teigland 已提交
558 559 560 561 562 563 564 565 566 567 568

	lock_buffer(bh);
	clear_buffer_dirty(bh);
	bh->b_bdev = NULL;
	clear_buffer_mapped(bh);
	clear_buffer_req(bh);
	clear_buffer_new(bh);
	clear_buffer_delay(bh);
	unlock_buffer(bh);
}

569
static void gfs2_invalidatepage(struct page *page, unsigned long offset)
D
David Teigland 已提交
570
{
571
	struct gfs2_sbd *sdp = GFS2_SB(page->mapping->host);
D
David Teigland 已提交
572 573 574 575 576
	struct buffer_head *head, *bh, *next;
	unsigned int curr_off = 0;

	BUG_ON(!PageLocked(page));
	if (!page_has_buffers(page))
577
		return;
D
David Teigland 已提交
578 579 580 581 582 583 584 585 586 587 588 589 590 591

	bh = head = page_buffers(page);
	do {
		unsigned int next_off = curr_off + bh->b_size;
		next = bh->b_this_page;

		if (offset <= curr_off)
			discard_buffer(sdp, bh);

		curr_off = next_off;
		bh = next;
	} while (bh != head);

	if (!offset)
592
		try_to_release_page(page, 0);
D
David Teigland 已提交
593

594
	return;
D
David Teigland 已提交
595 596
}

S
Steven Whitehouse 已提交
597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626
/**
 * gfs2_ok_for_dio - check that dio is valid on this file
 * @ip: The inode
 * @rw: READ or WRITE
 * @offset: The offset at which we are reading or writing
 *
 * Returns: 0 (to ignore the i/o request and thus fall back to buffered i/o)
 *          1 (to accept the i/o request)
 */
static int gfs2_ok_for_dio(struct gfs2_inode *ip, int rw, loff_t offset)
{
	/*
	 * Should we return an error here? I can't see that O_DIRECT for
	 * a journaled file makes any sense. For now we'll silently fall
	 * back to buffered I/O, likewise we do the same for stuffed
	 * files since they are (a) small and (b) unaligned.
	 */
	if (gfs2_is_jdata(ip))
		return 0;

	if (gfs2_is_stuffed(ip))
		return 0;

	if (offset > i_size_read(&ip->i_inode))
		return 0;
	return 1;
}



S
Steven Whitehouse 已提交
627 628 629
static ssize_t gfs2_direct_IO(int rw, struct kiocb *iocb,
			      const struct iovec *iov, loff_t offset,
			      unsigned long nr_segs)
630 631 632
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;
633
	struct gfs2_inode *ip = GFS2_I(inode);
634 635 636 637
	struct gfs2_holder gh;
	int rv;

	/*
S
Steven Whitehouse 已提交
638 639 640 641 642 643
	 * Deferred lock, even if its a write, since we do no allocation
	 * on this path. All we need change is atime, and this lock mode
	 * ensures that other nodes have flushed their buffered read caches
	 * (i.e. their page cache entries for this inode). We do not,
	 * unfortunately have the option of only flushing a range like
	 * the VFS does.
644
	 */
S
Steven Whitehouse 已提交
645
	gfs2_holder_init(ip->i_gl, LM_ST_DEFERRED, GL_ATIME, &gh);
646
	rv = gfs2_glock_nq_atime(&gh);
647
	if (rv)
S
Steven Whitehouse 已提交
648 649 650 651 652 653 654 655
		return rv;
	rv = gfs2_ok_for_dio(ip, rw, offset);
	if (rv != 1)
		goto out; /* dio not valid, fall back to buffered i/o */

	rv = blockdev_direct_IO_no_locking(rw, iocb, inode, inode->i_sb->s_bdev,
					   iov, offset, nr_segs,
					   gfs2_get_block_direct, NULL);
656 657 658 659 660 661
out:
	gfs2_glock_dq_m(1, &gh);
	gfs2_holder_uninit(&gh);
	return rv;
}

S
Steven Whitehouse 已提交
662 663 664 665 666 667 668 669 670 671 672 673
/**
 * stuck_releasepage - We're stuck in gfs2_releasepage().  Print stuff out.
 * @bh: the buffer we're stuck on
 *
 */

static void stuck_releasepage(struct buffer_head *bh)
{
	struct inode *inode = bh->b_page->mapping->host;
	struct gfs2_sbd *sdp = inode->i_sb->s_fs_info;
	struct gfs2_bufdata *bd = bh->b_private;
	struct gfs2_glock *gl;
674 675
static unsigned limit = 0;

676
	if (limit > 3)
677
		return;
678
	limit++;
S
Steven Whitehouse 已提交
679 680 681 682 683 684 685 686 687 688 689 690

	fs_warn(sdp, "stuck in gfs2_releasepage() %p\n", inode);
	fs_warn(sdp, "blkno = %llu, bh->b_count = %d\n",
		(unsigned long long)bh->b_blocknr, atomic_read(&bh->b_count));
	fs_warn(sdp, "pinned = %u\n", buffer_pinned(bh));
	fs_warn(sdp, "bh->b_private = %s\n", (bd) ? "!NULL" : "NULL");

	if (!bd)
		return;

	gl = bd->bd_gl;

691
	fs_warn(sdp, "gl = (%u, %llu)\n",
S
Steven Whitehouse 已提交
692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715
		gl->gl_name.ln_type, (unsigned long long)gl->gl_name.ln_number);

	fs_warn(sdp, "bd_list_tr = %s, bd_le.le_list = %s\n",
		(list_empty(&bd->bd_list_tr)) ? "no" : "yes",
		(list_empty(&bd->bd_le.le_list)) ? "no" : "yes");

	if (gl->gl_ops == &gfs2_inode_glops) {
		struct gfs2_inode *ip = gl->gl_object;
		unsigned int x;

		if (!ip)
			return;

		fs_warn(sdp, "ip = %llu %llu\n",
			(unsigned long long)ip->i_num.no_formal_ino,
			(unsigned long long)ip->i_num.no_addr);

		for (x = 0; x < GFS2_MAX_META_HEIGHT; x++)
			fs_warn(sdp, "ip->i_cache[%u] = %s\n",
				x, (ip->i_cache[x]) ? "!NULL" : "NULL");
	}
}

/**
716
 * gfs2_releasepage - free the metadata associated with a page
S
Steven Whitehouse 已提交
717 718 719 720 721 722 723 724 725 726 727 728 729 730 731
 * @page: the page that's being released
 * @gfp_mask: passed from Linux VFS, ignored by us
 *
 * Call try_to_free_buffers() if the buffers in this page can be
 * released.
 *
 * Returns: 0
 */

int gfs2_releasepage(struct page *page, gfp_t gfp_mask)
{
	struct inode *aspace = page->mapping->host;
	struct gfs2_sbd *sdp = aspace->i_sb->s_fs_info;
	struct buffer_head *bh, *head;
	struct gfs2_bufdata *bd;
732
	unsigned long t = jiffies + gfs2_tune_get(sdp, gt_stall_secs) * HZ;
S
Steven Whitehouse 已提交
733 734 735 736 737 738 739

	if (!page_has_buffers(page))
		goto out;

	head = bh = page_buffers(page);
	do {
		while (atomic_read(&bh->b_count)) {
740 741 742
			if (!atomic_read(&aspace->i_writecount))
				return 0;

743 744 745
			if (!(gfp_mask & __GFP_WAIT))
				return 0;

746 747 748 749
			if (time_after_eq(jiffies, t)) {
				stuck_releasepage(bh);
				/* should we withdraw here? */
				return 0;
S
Steven Whitehouse 已提交
750 751
			}

752
			yield();
S
Steven Whitehouse 已提交
753 754 755
		}

		gfs2_assert_warn(sdp, !buffer_pinned(bh));
756
		gfs2_assert_warn(sdp, !buffer_dirty(bh));
S
Steven Whitehouse 已提交
757

758
		gfs2_log_lock(sdp);
S
Steven Whitehouse 已提交
759 760 761 762 763
		bd = bh->b_private;
		if (bd) {
			gfs2_assert_warn(sdp, bd->bd_bh == bh);
			gfs2_assert_warn(sdp, list_empty(&bd->bd_list_tr));
			gfs2_assert_warn(sdp, !bd->bd_ail);
764 765 766
			bd->bd_bh = NULL;
			if (!list_empty(&bd->bd_le.le_list))
				bd = NULL;
S
Steven Whitehouse 已提交
767 768
			bh->b_private = NULL;
		}
769 770 771
		gfs2_log_unlock(sdp);
		if (bd)
			kmem_cache_free(gfs2_bufdata_cachep, bd);
S
Steven Whitehouse 已提交
772 773

		bh = bh->b_this_page;
774
	} while (bh != head);
S
Steven Whitehouse 已提交
775

S
Steven Whitehouse 已提交
776
out:
S
Steven Whitehouse 已提交
777 778 779
	return try_to_free_buffers(page);
}

780
const struct address_space_operations gfs2_file_aops = {
D
David Teigland 已提交
781 782
	.writepage = gfs2_writepage,
	.readpage = gfs2_readpage,
S
Steven Whitehouse 已提交
783
	.readpages = gfs2_readpages,
D
David Teigland 已提交
784 785 786 787 788
	.sync_page = block_sync_page,
	.prepare_write = gfs2_prepare_write,
	.commit_write = gfs2_commit_write,
	.bmap = gfs2_bmap,
	.invalidatepage = gfs2_invalidatepage,
S
Steven Whitehouse 已提交
789
	.releasepage = gfs2_releasepage,
D
David Teigland 已提交
790 791 792
	.direct_IO = gfs2_direct_IO,
};