ops_address.c 19.4 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>
D
David Teigland 已提交
20 21

#include "gfs2.h"
22 23
#include "lm_interface.h"
#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
{
	int new = create;
69
	u64 dblock;
D
David Teigland 已提交
70
	int error;
S
Steven Whitehouse 已提交
71
	int boundary;
D
David Teigland 已提交
72

S
Steven Whitehouse 已提交
73
	error = gfs2_block_map(inode, lblock, &new, &dblock, &boundary);
D
David Teigland 已提交
74 75 76 77 78 79 80 81 82
	if (error)
		return error;

	if (!dblock)
		return 0;

	map_bh(bh_result, inode->i_sb, dblock);
	if (new)
		set_buffer_new(bh_result);
S
Steven Whitehouse 已提交
83 84
	if (boundary)
		set_buffer_boundary(bh_result);
D
David Teigland 已提交
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102

	return 0;
}

/**
 * get_block_noalloc - Fills in a buffer head with details about a block
 * @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
 */

static int get_block_noalloc(struct inode *inode, sector_t lblock,
			     struct buffer_head *bh_result, int create)
{
	int new = 0;
103
	u64 dblock;
D
David Teigland 已提交
104
	int error;
S
Steven Whitehouse 已提交
105
	int boundary;
D
David Teigland 已提交
106

S
Steven Whitehouse 已提交
107
	error = gfs2_block_map(inode, lblock, &new, &dblock, &boundary);
D
David Teigland 已提交
108 109 110 111 112
	if (error)
		return error;

	if (dblock)
		map_bh(bh_result, inode->i_sb, dblock);
113
	else if (gfs2_assert_withdraw(GFS2_SB(inode), !create))
D
David Teigland 已提交
114
		error = -EIO;
S
Steven Whitehouse 已提交
115 116
	if (boundary)
		set_buffer_boundary(bh_result);
D
David Teigland 已提交
117 118 119 120

	return error;
}

121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139
static int get_block_direct(struct inode *inode, sector_t lblock,
			    struct buffer_head *bh_result, int create)
{
	int new = 0;
	u64 dblock;
	int error, boundary;

	error = gfs2_block_map(inode, lblock, &new, &dblock, &boundary);
	if (error)
		return error;

	if (dblock) {
		map_bh(bh_result, inode->i_sb, dblock);
		if (boundary)
			set_buffer_boundary(bh_result);
	}

	return 0;
}
D
David Teigland 已提交
140 141 142 143 144 145
/**
 * gfs2_writepage - Write complete page
 * @page: Page to write
 *
 * Returns: errno
 *
146 147
 * 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
148 149 150 151
 */

static int gfs2_writepage(struct page *page, struct writeback_control *wbc)
{
152
	struct inode *inode = page->mapping->host;
153 154
	struct gfs2_inode *ip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);
155 156 157
	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
158
	int error;
159
	int done_trans = 0;
D
David Teigland 已提交
160 161 162 163 164

	if (gfs2_assert_withdraw(sdp, gfs2_glock_is_held_excl(ip->i_gl))) {
		unlock_page(page);
		return -EIO;
	}
165
	if (current->journal_info)
166 167 168 169
		goto out_ignore;

	/* Is the page fully outside i_size? (truncate in progress) */
        offset = i_size & (PAGE_CACHE_SIZE-1);
170
	if (page->index > end_index || (page->index == end_index && !offset)) {
171
		page->mapping->a_ops->invalidatepage(page, 0);
D
David Teigland 已提交
172
		unlock_page(page);
173
		return 0; /* don't care */
D
David Teigland 已提交
174 175
	}

176 177 178 179
	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;
180 181 182 183
		if (!page_has_buffers(page)) {
			create_empty_buffers(page, inode->i_sb->s_blocksize,
					     (1 << BH_Dirty)|(1 << BH_Uptodate));
		}
184 185 186 187 188 189
		gfs2_page_add_databufs(ip, page, 0, sdp->sd_vfs->s_blocksize-1);
		done_trans = 1;
	}
	error = block_write_full_page(page, get_block_noalloc, wbc);
	if (done_trans)
		gfs2_trans_end(sdp);
D
David Teigland 已提交
190 191
	gfs2_meta_cache_flush(ip);
	return error;
192 193 194 195 196

out_ignore:
	redirty_page_for_writepage(wbc, page);
	unlock_page(page);
	return 0;
D
David Teigland 已提交
197 198
}

S
Steven Whitehouse 已提交
199 200 201 202 203 204 205 206 207 208 209 210 211
static int zero_readpage(struct page *page)
{
	void *kaddr;

	kaddr = kmap_atomic(page, KM_USER0);
	memset(kaddr, 0, PAGE_CACHE_SIZE);
	kunmap_atomic(page, KM_USER0);

	SetPageUptodate(page);

	return 0;
}

D
David Teigland 已提交
212 213 214 215 216 217 218 219 220 221 222 223 224 225
/**
 * 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;

S
Steven Whitehouse 已提交
226 227 228 229
	/* Only the first page of a stuffed file might contain data */
	if (unlikely(page->index))
		return zero_readpage(page);

D
David Teigland 已提交
230 231 232 233
	error = gfs2_meta_inode_buffer(ip, &dibh);
	if (error)
		return error;

234
	kaddr = kmap_atomic(page, KM_USER0);
S
Steven Whitehouse 已提交
235
	memcpy(kaddr, dibh->b_data + sizeof(struct gfs2_dinode),
D
David Teigland 已提交
236
	       ip->i_di.di_size);
S
Steven Whitehouse 已提交
237
	memset(kaddr + ip->i_di.di_size, 0, PAGE_CACHE_SIZE - ip->i_di.di_size);
238
	kunmap_atomic(page, KM_USER0);
D
David Teigland 已提交
239 240 241 242 243 244 245 246 247 248 249

	brelse(dibh);

	SetPageUptodate(page);

	return 0;
}


/**
 * gfs2_readpage - readpage with locking
250 251
 * @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
252 253 254 255 256 257 258
 * @page: The page to read
 *
 * Returns: errno
 */

static int gfs2_readpage(struct file *file, struct page *page)
{
259 260
	struct gfs2_inode *ip = GFS2_I(page->mapping->host);
	struct gfs2_sbd *sdp = GFS2_SB(page->mapping->host);
261
	struct gfs2_holder gh;
D
David Teigland 已提交
262
	int error;
263
	int do_unlock = 0;
D
David Teigland 已提交
264

265
	if (likely(file != &gfs2_internal_file_sentinel)) {
266 267 268 269 270
		if (file) {
			struct gfs2_file *gf = file->private_data;
			if (test_bit(GFF_EXLOCK, &gf->f_flags))
				goto skip_lock;
		}
271
		gfs2_holder_init(ip->i_gl, LM_ST_SHARED, GL_ATIME|GL_AOP, &gh);
272
		do_unlock = 1;
273
		error = gfs2_glock_nq_m_atime(1, &gh);
S
Steven Whitehouse 已提交
274
		if (unlikely(error))
275 276
			goto out_unlock;
	}
D
David Teigland 已提交
277

278
skip_lock:
279
	if (gfs2_is_stuffed(ip)) {
S
Steven Whitehouse 已提交
280 281
		error = stuffed_readpage(ip, page);
		unlock_page(page);
D
David Teigland 已提交
282
	} else
283
		error = mpage_readpage(page, gfs2_get_block);
D
David Teigland 已提交
284 285 286 287

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

288
	if (file != &gfs2_internal_file_sentinel) {
289 290 291
		gfs2_glock_dq_m(1, &gh);
		gfs2_holder_uninit(&gh);
	}
292
out:
D
David Teigland 已提交
293
	return error;
294 295
out_unlock:
	unlock_page(page);
296
	if (do_unlock)
S
Steven Whitehouse 已提交
297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319
		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;
320 321
	struct gfs2_inode *ip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);
S
Steven Whitehouse 已提交
322 323 324
	struct gfs2_holder gh;
	unsigned page_idx;
	int ret;
325
	int do_unlock = 0;
S
Steven Whitehouse 已提交
326

327
	if (likely(file != &gfs2_internal_file_sentinel)) {
328 329 330 331 332
		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
333 334
		gfs2_holder_init(ip->i_gl, LM_ST_SHARED,
				 LM_FLAG_TRY_1CB|GL_ATIME|GL_AOP, &gh);
335
		do_unlock = 1;
S
Steven Whitehouse 已提交
336 337 338 339 340 341
		ret = gfs2_glock_nq_m_atime(1, &gh);
		if (ret == GLR_TRYFAILED) 
			goto out_noerror;
		if (unlikely(ret))
			goto out_unlock;
	}
342
skip_lock:
S
Steven Whitehouse 已提交
343 344 345 346
	if (gfs2_is_stuffed(ip)) {
		struct pagevec lru_pvec;
		pagevec_init(&lru_pvec, 0);
		for (page_idx = 0; page_idx < nr_pages; page_idx++) {
347 348
			struct page *page = list_entry(pages->prev, struct page, lru);
			prefetchw(&page->flags);
S
Steven Whitehouse 已提交
349 350 351 352 353 354 355
			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);
356 357
			} else {
				page_cache_release(page);
S
Steven Whitehouse 已提交
358 359 360 361 362 363 364 365 366
			}
		}
		pagevec_lru_add(&lru_pvec);
		ret = 0;
	} else {
		/* What we really want to do .... */
		ret = mpage_readpages(mapping, pages, nr_pages, gfs2_get_block);
	}

367
	if (do_unlock) {
S
Steven Whitehouse 已提交
368 369 370 371 372 373 374 375 376 377 378 379
		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:
	/* unlock all pages, we can't do any I/O right now */
	for (page_idx = 0; page_idx < nr_pages; page_idx++) {
380
		struct page *page = list_entry(pages->prev, struct page, lru);
S
Steven Whitehouse 已提交
381 382 383 384
		list_del(&page->lru);
		unlock_page(page);
		page_cache_release(page);
	}
385
	if (do_unlock)
S
Steven Whitehouse 已提交
386
		gfs2_holder_uninit(&gh);
387
	goto out;
D
David Teigland 已提交
388 389 390 391 392 393 394 395 396 397 398 399 400 401 402
}

/**
 * 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)
{
403 404
	struct gfs2_inode *ip = GFS2_I(page->mapping->host);
	struct gfs2_sbd *sdp = GFS2_SB(page->mapping->host);
405 406
	unsigned int data_blocks, ind_blocks, rblocks;
	int alloc_required;
D
David Teigland 已提交
407
	int error = 0;
408 409 410
	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;
D
David Teigland 已提交
411

412
	gfs2_holder_init(ip->i_gl, LM_ST_EXCLUSIVE, GL_ATIME|GL_AOP, &ip->i_gh);
413 414 415
	error = gfs2_glock_nq_m_atime(1, &ip->i_gh);
	if (error)
		goto out_uninit;
D
David Teigland 已提交
416

417 418 419 420 421
	gfs2_write_calc_reserv(ip, to - from, &data_blocks, &ind_blocks);

	error = gfs2_write_alloc_required(ip, pos, from - to, &alloc_required);
	if (error)
		goto out_unlock;
D
David Teigland 已提交
422

423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452

	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;

		error = gfs2_quota_check(ip, ip->i_di.di_uid, ip->i_di.di_gid);
		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)) {
453
			error = gfs2_unstuff_dinode(ip, page);
454 455 456
			if (error == 0)
				goto prepare_write;
		} else if (!PageUptodate(page))
D
David Teigland 已提交
457
			error = stuffed_readpage(ip, page);
458
		goto out;
459 460
	}

461
prepare_write:
462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478
	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
479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496

	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;
497 498
	struct gfs2_inode *ip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);
499 500 501
	int error = -EOPNOTSUPP;
	struct buffer_head *dibh;
	struct gfs2_alloc *al = &ip->i_alloc;;
D
David Teigland 已提交
502

503 504 505 506 507 508 509 510 511
	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);

D
David Teigland 已提交
512
	if (gfs2_is_stuffed(ip)) {
513
		u64 file_size;
D
David Teigland 已提交
514 515
		void *kaddr;

516
		file_size = ((u64)page->index << PAGE_CACHE_SHIFT) + to;
D
David Teigland 已提交
517

518
		kaddr = kmap_atomic(page, KM_USER0);
D
David Teigland 已提交
519
		memcpy(dibh->b_data + sizeof(struct gfs2_dinode) + from,
520
		       kaddr + from, to - from);
521
		kunmap_atomic(page, KM_USER0);
D
David Teigland 已提交
522 523 524 525 526 527

		SetPageUptodate(page);

		if (inode->i_size < file_size)
			i_size_write(inode, file_size);
	} else {
528 529
		if (sdp->sd_args.ar_data == GFS2_DATA_ORDERED ||
		    gfs2_is_jdata(ip))
530
			gfs2_page_add_databufs(ip, page, from, to);
D
David Teigland 已提交
531 532 533 534 535
		error = generic_commit_write(file, page, from, to);
		if (error)
			goto fail;
	}

536 537 538 539 540 541 542 543 544 545 546 547 548
	if (ip->i_di.di_size < inode->i_size)
		ip->i_di.di_size = inode->i_size;

	gfs2_dinode_out(&ip->i_di, dibh->b_data);
	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
549 550
	return 0;

551 552 553 554 555 556 557 558 559 560 561 562
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
563 564 565 566 567 568 569 570 571 572 573 574 575 576
	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)
{
577
	struct gfs2_inode *ip = GFS2_I(mapping->host);
D
David Teigland 已提交
578 579 580 581 582 583 584 585 586
	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))
587
		dblock = generic_block_bmap(mapping, lblock, gfs2_get_block);
D
David Teigland 已提交
588 589 590 591 592 593 594 595

	gfs2_glock_dq_uninit(&i_gh);

	return dblock;
}

static void discard_buffer(struct gfs2_sbd *sdp, struct buffer_head *bh)
{
596
	struct gfs2_bufdata *bd;
D
David Teigland 已提交
597 598

	gfs2_log_lock(sdp);
599
	bd = bh->b_private;
600 601
	if (bd) {
		bd->bd_bh = NULL;
602
		bh->b_private = NULL;
603 604
	}
	gfs2_log_unlock(sdp);
D
David Teigland 已提交
605 606 607 608 609 610 611 612 613 614 615

	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);
}

616
static void gfs2_invalidatepage(struct page *page, unsigned long offset)
D
David Teigland 已提交
617
{
618
	struct gfs2_sbd *sdp = GFS2_SB(page->mapping->host);
D
David Teigland 已提交
619 620 621 622 623
	struct buffer_head *head, *bh, *next;
	unsigned int curr_off = 0;

	BUG_ON(!PageLocked(page));
	if (!page_has_buffers(page))
624
		return;
D
David Teigland 已提交
625 626 627 628 629 630 631 632 633 634 635 636 637 638

	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)
639
		try_to_release_page(page, 0);
D
David Teigland 已提交
640

641
	return;
D
David Teigland 已提交
642 643
}

S
Steven Whitehouse 已提交
644 645 646
static ssize_t gfs2_direct_IO(int rw, struct kiocb *iocb,
			      const struct iovec *iov, loff_t offset,
			      unsigned long nr_segs)
647 648 649
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;
650
	struct gfs2_inode *ip = GFS2_I(inode);
651 652 653
	struct gfs2_holder gh;
	int rv;

S
Steven Whitehouse 已提交
654 655
	if (rw == READ)
		mutex_lock(&inode->i_mutex);
656
	/*
S
Steven Whitehouse 已提交
657
	 * Shared lock, even if its a write, since we do no allocation
658 659 660 661 662 663 664
	 * on this path. All we need change is atime.
	 */
	gfs2_holder_init(ip->i_gl, LM_ST_SHARED, GL_ATIME, &gh);
	rv = gfs2_glock_nq_m_atime(1, &gh);
	if (rv)
		goto out;

S
Steven Whitehouse 已提交
665 666 667
	if (offset > i_size_read(inode))
		goto out;

668 669 670 671 672 673 674 675 676 677 678 679
	/*
	 * 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))
		goto out;

	if (gfs2_is_stuffed(ip))
		goto out;

S
Steven Whitehouse 已提交
680 681 682
	rv = blockdev_direct_IO_own_locking(rw, iocb, inode,
					    inode->i_sb->s_bdev,
					    iov, offset, nr_segs,
683
					    get_block_direct, NULL);
684 685 686
out:
	gfs2_glock_dq_m(1, &gh);
	gfs2_holder_uninit(&gh);
S
Steven Whitehouse 已提交
687 688
	if (rw == READ)
		mutex_unlock(&inode->i_mutex);
689 690 691 692

	return rv;
}

S
Steven Whitehouse 已提交
693 694 695 696 697 698 699 700 701 702 703 704
/**
 * 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;
705 706
static unsigned limit = 0;

707
	if (limit > 3)
708
		return;
709
	limit++;
S
Steven Whitehouse 已提交
710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746

	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;

	fs_warn(sdp, "gl = (%u, %llu)\n", 
		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");
	}
}

/**
747
 * gfs2_releasepage - free the metadata associated with a page
S
Steven Whitehouse 已提交
748 749 750 751 752 753 754 755 756 757 758 759 760 761 762
 * @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;
763
	unsigned long t = jiffies + gfs2_tune_get(sdp, gt_stall_secs) * HZ;
S
Steven Whitehouse 已提交
764 765 766 767 768 769 770

	if (!page_has_buffers(page))
		goto out;

	head = bh = page_buffers(page);
	do {
		while (atomic_read(&bh->b_count)) {
771 772 773 774 775 776 777
			if (!atomic_read(&aspace->i_writecount))
				return 0;

			if (time_after_eq(jiffies, t)) {
				stuck_releasepage(bh);
				/* should we withdraw here? */
				return 0;
S
Steven Whitehouse 已提交
778 779
			}

780
			yield();
S
Steven Whitehouse 已提交
781 782 783
		}

		gfs2_assert_warn(sdp, !buffer_pinned(bh));
784
		gfs2_assert_warn(sdp, !buffer_dirty(bh));
S
Steven Whitehouse 已提交
785

786
		gfs2_log_lock(sdp);
S
Steven Whitehouse 已提交
787 788 789 790 791
		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);
792 793 794
			bd->bd_bh = NULL;
			if (!list_empty(&bd->bd_le.le_list))
				bd = NULL;
S
Steven Whitehouse 已提交
795 796
			bh->b_private = NULL;
		}
797 798 799
		gfs2_log_unlock(sdp);
		if (bd)
			kmem_cache_free(gfs2_bufdata_cachep, bd);
S
Steven Whitehouse 已提交
800 801

		bh = bh->b_this_page;
802
	} while (bh != head);
S
Steven Whitehouse 已提交
803

S
Steven Whitehouse 已提交
804
out:
S
Steven Whitehouse 已提交
805 806 807
	return try_to_free_buffers(page);
}

808
const struct address_space_operations gfs2_file_aops = {
D
David Teigland 已提交
809 810
	.writepage = gfs2_writepage,
	.readpage = gfs2_readpage,
S
Steven Whitehouse 已提交
811
	.readpages = gfs2_readpages,
D
David Teigland 已提交
812 813 814 815 816
	.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
817
	.releasepage = gfs2_releasepage,
D
David Teigland 已提交
818 819 820
	.direct_IO = gfs2_direct_IO,
};