ops_address.c 19.7 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/writeback.h>
20
#include <linux/gfs2_ondisk.h>
21
#include <linux/lm_interface.h>
D
David Teigland 已提交
22 23

#include "gfs2.h"
24
#include "incore.h"
D
David Teigland 已提交
25 26 27 28 29 30 31 32
#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"
33
#include "rgrp.h"
34
#include "ops_file.h"
35
#include "util.h"
S
Steven Whitehouse 已提交
36
#include "glops.h"
D
David Teigland 已提交
37

S
Steven Whitehouse 已提交
38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55

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
56
/**
57
 * gfs2_get_block - Fills in a buffer head with details about a block
D
David Teigland 已提交
58 59 60 61 62 63 64 65
 * @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
 */

66 67
int gfs2_get_block(struct inode *inode, sector_t lblock,
	           struct buffer_head *bh_result, int create)
D
David Teigland 已提交
68
{
69
	return gfs2_block_map(inode, lblock, create, bh_result);
D
David Teigland 已提交
70 71 72
}

/**
73
 * gfs2_get_block_noalloc - Fills in a buffer head with details about a block
D
David Teigland 已提交
74 75 76 77 78 79 80 81
 * @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
 */

82 83
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
84 85 86
{
	int error;

87
	error = gfs2_block_map(inode, lblock, 0, bh_result);
D
David Teigland 已提交
88 89
	if (error)
		return error;
90 91 92
	if (bh_result->b_blocknr == 0)
		return -EIO;
	return 0;
D
David Teigland 已提交
93 94
}

95 96
static int gfs2_get_block_direct(struct inode *inode, sector_t lblock,
				 struct buffer_head *bh_result, int create)
97
{
98
	return gfs2_block_map(inode, lblock, 0, bh_result);
99
}
100

D
David Teigland 已提交
101 102 103 104 105 106
/**
 * gfs2_writepage - Write complete page
 * @page: Page to write
 *
 * Returns: errno
 *
107 108
 * 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
109 110 111 112
 */

static int gfs2_writepage(struct page *page, struct writeback_control *wbc)
{
113
	struct inode *inode = page->mapping->host;
114 115
	struct gfs2_inode *ip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);
116 117 118
	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
119
	int error;
120
	int done_trans = 0;
D
David Teigland 已提交
121 122 123 124 125

	if (gfs2_assert_withdraw(sdp, gfs2_glock_is_held_excl(ip->i_gl))) {
		unlock_page(page);
		return -EIO;
	}
126
	if (current->journal_info)
127 128 129 130
		goto out_ignore;

	/* Is the page fully outside i_size? (truncate in progress) */
        offset = i_size & (PAGE_CACHE_SIZE-1);
131
	if (page->index > end_index || (page->index == end_index && !offset)) {
132
		page->mapping->a_ops->invalidatepage(page, 0);
D
David Teigland 已提交
133
		unlock_page(page);
134
		return 0; /* don't care */
D
David Teigland 已提交
135 136
	}

137 138 139 140
	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;
141 142 143 144
		if (!page_has_buffers(page)) {
			create_empty_buffers(page, inode->i_sb->s_blocksize,
					     (1 << BH_Dirty)|(1 << BH_Uptodate));
		}
145 146 147
		gfs2_page_add_databufs(ip, page, 0, sdp->sd_vfs->s_blocksize-1);
		done_trans = 1;
	}
148
	error = block_write_full_page(page, gfs2_get_block_noalloc, wbc);
149 150
	if (done_trans)
		gfs2_trans_end(sdp);
D
David Teigland 已提交
151 152
	gfs2_meta_cache_flush(ip);
	return error;
153 154 155 156 157

out_ignore:
	redirty_page_for_writepage(wbc, page);
	unlock_page(page);
	return 0;
D
David Teigland 已提交
158 159
}

160 161 162 163 164 165 166 167 168 169 170 171 172
/**
 * gfs2_writepages - Write a bunch of dirty pages back to disk
 * @mapping: The mapping to write
 * @wbc: Write-back control
 *
 * For journaled files and/or ordered writes this just falls back to the
 * kernel's default writepages path for now. We will probably want to change
 * that eventually (i.e. when we look at allocate on flush).
 *
 * For the data=writeback case though we can already ignore buffer heads
 * and write whole extents at once. This is a big reduction in the
 * number of I/O requests we send and the bmap calls we make in this case.
 */
173 174
static int gfs2_writepages(struct address_space *mapping,
			   struct writeback_control *wbc)
175 176 177 178 179 180 181 182 183 184 185
{
	struct inode *inode = mapping->host;
	struct gfs2_inode *ip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);

	if (sdp->sd_args.ar_data == GFS2_DATA_WRITEBACK && !gfs2_is_jdata(ip))
		return mpage_writepages(mapping, wbc, gfs2_get_block_noalloc);

	return generic_writepages(mapping, wbc);
}

D
David Teigland 已提交
186 187 188 189 190 191 192 193 194 195 196 197 198 199
/**
 * 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;

200
	BUG_ON(page->index);
S
Steven Whitehouse 已提交
201

D
David Teigland 已提交
202 203 204 205
	error = gfs2_meta_inode_buffer(ip, &dibh);
	if (error)
		return error;

206
	kaddr = kmap_atomic(page, KM_USER0);
S
Steven Whitehouse 已提交
207
	memcpy(kaddr, dibh->b_data + sizeof(struct gfs2_dinode),
D
David Teigland 已提交
208
	       ip->i_di.di_size);
S
Steven Whitehouse 已提交
209
	memset(kaddr + ip->i_di.di_size, 0, PAGE_CACHE_SIZE - ip->i_di.di_size);
210
	kunmap_atomic(kaddr, KM_USER0);
D
David Teigland 已提交
211 212 213 214 215 216 217 218 219 220 221

	brelse(dibh);

	SetPageUptodate(page);

	return 0;
}


/**
 * gfs2_readpage - readpage with locking
222 223
 * @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
224 225 226 227 228 229 230
 * @page: The page to read
 *
 * Returns: errno
 */

static int gfs2_readpage(struct file *file, struct page *page)
{
231 232
	struct gfs2_inode *ip = GFS2_I(page->mapping->host);
	struct gfs2_sbd *sdp = GFS2_SB(page->mapping->host);
233
	struct gfs2_file *gf = NULL;
234
	struct gfs2_holder gh;
D
David Teigland 已提交
235
	int error;
236
	int do_unlock = 0;
D
David Teigland 已提交
237

238
	if (likely(file != &gfs2_internal_file_sentinel)) {
239
		if (file) {
240
			gf = file->private_data;
241
			if (test_bit(GFF_EXLOCK, &gf->f_flags))
242
				/* gfs2_sharewrite_nopage has grabbed the ip->i_gl already */
243 244
				goto skip_lock;
		}
245
		gfs2_holder_init(ip->i_gl, LM_ST_SHARED, GL_ATIME|LM_FLAG_TRY_1CB, &gh);
246
		do_unlock = 1;
247
		error = gfs2_glock_nq_atime(&gh);
S
Steven Whitehouse 已提交
248
		if (unlikely(error))
249 250
			goto out_unlock;
	}
D
David Teigland 已提交
251

252
skip_lock:
253
	if (gfs2_is_stuffed(ip)) {
S
Steven Whitehouse 已提交
254 255
		error = stuffed_readpage(ip, page);
		unlock_page(page);
D
David Teigland 已提交
256
	} else
257
		error = mpage_readpage(page, gfs2_get_block);
D
David Teigland 已提交
258 259 260 261

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

262
	if (do_unlock) {
263 264 265
		gfs2_glock_dq_m(1, &gh);
		gfs2_holder_uninit(&gh);
	}
266
out:
D
David Teigland 已提交
267
	return error;
268
out_unlock:
269 270
	if (error == GLR_TRYFAILED)
		error = AOP_TRUNCATED_PAGE;
271
	unlock_page(page);
272
	if (do_unlock)
S
Steven Whitehouse 已提交
273 274 275 276 277 278 279 280 281 282 283 284 285
		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.
286
 * 2. We don't handle stuffed files here we let readpage do the honours.
S
Steven Whitehouse 已提交
287 288 289 290 291 292 293 294 295
 * 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;
296 297
	struct gfs2_inode *ip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);
S
Steven Whitehouse 已提交
298
	struct gfs2_holder gh;
299
	int ret = 0;
300
	int do_unlock = 0;
S
Steven Whitehouse 已提交
301

302
	if (likely(file != &gfs2_internal_file_sentinel)) {
303 304 305 306 307
		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
308
		gfs2_holder_init(ip->i_gl, LM_ST_SHARED,
309
				 LM_FLAG_TRY_1CB|GL_ATIME, &gh);
310
		do_unlock = 1;
311
		ret = gfs2_glock_nq_atime(&gh);
312
		if (ret == GLR_TRYFAILED)
S
Steven Whitehouse 已提交
313 314 315 316
			goto out_noerror;
		if (unlikely(ret))
			goto out_unlock;
	}
317
skip_lock:
318
	if (!gfs2_is_stuffed(ip))
S
Steven Whitehouse 已提交
319 320
		ret = mpage_readpages(mapping, pages, nr_pages, gfs2_get_block);

321
	if (do_unlock) {
S
Steven Whitehouse 已提交
322 323 324 325 326 327 328 329 330 331
		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:
332
	if (do_unlock)
S
Steven Whitehouse 已提交
333
		gfs2_holder_uninit(&gh);
334
	goto out;
D
David Teigland 已提交
335 336 337 338 339 340 341 342 343 344 345 346 347 348 349
}

/**
 * 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)
{
350 351
	struct gfs2_inode *ip = GFS2_I(page->mapping->host);
	struct gfs2_sbd *sdp = GFS2_SB(page->mapping->host);
352 353
	unsigned int data_blocks, ind_blocks, rblocks;
	int alloc_required;
D
David Teigland 已提交
354
	int error = 0;
355 356 357
	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;
358 359
	unsigned int write_len = to - from;

D
David Teigland 已提交
360

361
	gfs2_holder_init(ip->i_gl, LM_ST_EXCLUSIVE, GL_ATIME|LM_FLAG_TRY_1CB, &ip->i_gh);
362
	error = gfs2_glock_nq_atime(&ip->i_gh);
363
	if (unlikely(error)) {
364 365
		if (error == GLR_TRYFAILED) {
			unlock_page(page);
366
			error = AOP_TRUNCATED_PAGE;
367
		}
368
		goto out_uninit;
369
	}
D
David Teigland 已提交
370

371
	gfs2_write_calc_reserv(ip, write_len, &data_blocks, &ind_blocks);
372

373
	error = gfs2_write_alloc_required(ip, pos, write_len, &alloc_required);
374 375
	if (error)
		goto out_unlock;
D
David Teigland 已提交
376

377

378
	ip->i_alloc.al_requested = 0;
379 380 381 382 383 384 385
	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;

386
		error = gfs2_quota_check(ip, ip->i_inode.i_uid, ip->i_inode.i_gid);
387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407
		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)) {
408
			error = gfs2_unstuff_dinode(ip, page);
409 410 411
			if (error == 0)
				goto prepare_write;
		} else if (!PageUptodate(page))
D
David Teigland 已提交
412
			error = stuffed_readpage(ip, page);
413
		goto out;
414 415
	}

416
prepare_write:
417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433
	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
434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451

	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;
452 453
	struct gfs2_inode *ip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);
454 455
	int error = -EOPNOTSUPP;
	struct buffer_head *dibh;
456 457
	struct gfs2_alloc *al = &ip->i_alloc;
	struct gfs2_dinode *di;
D
David Teigland 已提交
458

459 460 461 462 463 464 465 466
	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);
467
	di = (struct gfs2_dinode *)dibh->b_data;
468

D
David Teigland 已提交
469
	if (gfs2_is_stuffed(ip)) {
470
		u64 file_size;
D
David Teigland 已提交
471 472
		void *kaddr;

473
		file_size = ((u64)page->index << PAGE_CACHE_SHIFT) + to;
D
David Teigland 已提交
474

475
		kaddr = kmap_atomic(page, KM_USER0);
D
David Teigland 已提交
476
		memcpy(dibh->b_data + sizeof(struct gfs2_dinode) + from,
477
		       kaddr + from, to - from);
478
		kunmap_atomic(kaddr, KM_USER0);
D
David Teigland 已提交
479 480 481

		SetPageUptodate(page);

482
		if (inode->i_size < file_size) {
D
David Teigland 已提交
483
			i_size_write(inode, file_size);
484 485
			mark_inode_dirty(inode);
		}
D
David Teigland 已提交
486
	} else {
487 488
		if (sdp->sd_args.ar_data == GFS2_DATA_ORDERED ||
		    gfs2_is_jdata(ip))
489
			gfs2_page_add_databufs(ip, page, from, to);
D
David Teigland 已提交
490 491 492 493 494
		error = generic_commit_write(file, page, from, to);
		if (error)
			goto fail;
	}

495
	if (ip->i_di.di_size < inode->i_size) {
496
		ip->i_di.di_size = inode->i_size;
497 498 499
		di->di_size = cpu_to_be64(inode->i_size);
	}

500 501 502 503 504 505 506 507 508
	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
509 510
	return 0;

511 512 513 514 515 516 517 518 519 520 521 522
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
523 524 525 526 527 528 529 530 531 532 533 534 535 536
	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)
{
537
	struct gfs2_inode *ip = GFS2_I(mapping->host);
D
David Teigland 已提交
538 539 540 541 542 543 544 545 546
	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))
547
		dblock = generic_block_bmap(mapping, lblock, gfs2_get_block);
D
David Teigland 已提交
548 549 550 551 552 553 554 555

	gfs2_glock_dq_uninit(&i_gh);

	return dblock;
}

static void discard_buffer(struct gfs2_sbd *sdp, struct buffer_head *bh)
{
556
	struct gfs2_bufdata *bd;
D
David Teigland 已提交
557 558

	gfs2_log_lock(sdp);
559
	bd = bh->b_private;
560 561
	if (bd) {
		bd->bd_bh = NULL;
562
		bh->b_private = NULL;
563 564
	}
	gfs2_log_unlock(sdp);
D
David Teigland 已提交
565 566 567 568 569 570 571 572 573 574 575

	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);
}

576
static void gfs2_invalidatepage(struct page *page, unsigned long offset)
D
David Teigland 已提交
577
{
578
	struct gfs2_sbd *sdp = GFS2_SB(page->mapping->host);
D
David Teigland 已提交
579 580 581 582 583
	struct buffer_head *head, *bh, *next;
	unsigned int curr_off = 0;

	BUG_ON(!PageLocked(page));
	if (!page_has_buffers(page))
584
		return;
D
David Teigland 已提交
585 586 587 588 589 590 591 592 593 594 595 596 597 598

	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)
599
		try_to_release_page(page, 0);
D
David Teigland 已提交
600

601
	return;
D
David Teigland 已提交
602 603
}

S
Steven Whitehouse 已提交
604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633
/**
 * 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
634 635 636
static ssize_t gfs2_direct_IO(int rw, struct kiocb *iocb,
			      const struct iovec *iov, loff_t offset,
			      unsigned long nr_segs)
637 638 639
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;
640
	struct gfs2_inode *ip = GFS2_I(inode);
641 642 643 644
	struct gfs2_holder gh;
	int rv;

	/*
S
Steven Whitehouse 已提交
645 646 647 648 649 650
	 * 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.
651
	 */
S
Steven Whitehouse 已提交
652
	gfs2_holder_init(ip->i_gl, LM_ST_DEFERRED, GL_ATIME, &gh);
653
	rv = gfs2_glock_nq_atime(&gh);
654
	if (rv)
S
Steven Whitehouse 已提交
655 656 657 658 659 660 661 662
		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);
663 664 665 666 667 668
out:
	gfs2_glock_dq_m(1, &gh);
	gfs2_holder_uninit(&gh);
	return rv;
}

S
Steven Whitehouse 已提交
669 670 671 672 673 674 675 676 677 678 679 680
/**
 * 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;
681 682
static unsigned limit = 0;

683
	if (limit > 3)
684
		return;
685
	limit++;
S
Steven Whitehouse 已提交
686 687 688 689 690 691 692 693 694 695 696 697

	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;

698
	fs_warn(sdp, "gl = (%u, %llu)\n",
S
Steven Whitehouse 已提交
699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722
		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");
	}
}

/**
723
 * gfs2_releasepage - free the metadata associated with a page
S
Steven Whitehouse 已提交
724 725 726 727 728 729 730 731 732 733 734 735 736 737 738
 * @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;
739
	unsigned long t = jiffies + gfs2_tune_get(sdp, gt_stall_secs) * HZ;
S
Steven Whitehouse 已提交
740 741 742 743 744 745 746

	if (!page_has_buffers(page))
		goto out;

	head = bh = page_buffers(page);
	do {
		while (atomic_read(&bh->b_count)) {
747 748 749
			if (!atomic_read(&aspace->i_writecount))
				return 0;

750 751 752
			if (!(gfp_mask & __GFP_WAIT))
				return 0;

753 754 755 756
			if (time_after_eq(jiffies, t)) {
				stuck_releasepage(bh);
				/* should we withdraw here? */
				return 0;
S
Steven Whitehouse 已提交
757 758
			}

759
			yield();
S
Steven Whitehouse 已提交
760 761 762
		}

		gfs2_assert_warn(sdp, !buffer_pinned(bh));
763
		gfs2_assert_warn(sdp, !buffer_dirty(bh));
S
Steven Whitehouse 已提交
764

765
		gfs2_log_lock(sdp);
S
Steven Whitehouse 已提交
766 767 768 769 770
		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);
771 772 773
			bd->bd_bh = NULL;
			if (!list_empty(&bd->bd_le.le_list))
				bd = NULL;
S
Steven Whitehouse 已提交
774 775
			bh->b_private = NULL;
		}
776 777 778
		gfs2_log_unlock(sdp);
		if (bd)
			kmem_cache_free(gfs2_bufdata_cachep, bd);
S
Steven Whitehouse 已提交
779 780

		bh = bh->b_this_page;
781
	} while (bh != head);
S
Steven Whitehouse 已提交
782

S
Steven Whitehouse 已提交
783
out:
S
Steven Whitehouse 已提交
784 785 786
	return try_to_free_buffers(page);
}

787
const struct address_space_operations gfs2_file_aops = {
D
David Teigland 已提交
788
	.writepage = gfs2_writepage,
789
	.writepages = gfs2_writepages,
D
David Teigland 已提交
790
	.readpage = gfs2_readpage,
S
Steven Whitehouse 已提交
791
	.readpages = gfs2_readpages,
D
David Teigland 已提交
792 793 794 795 796
	.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
797
	.releasepage = gfs2_releasepage,
D
David Teigland 已提交
798 799 800
	.direct_IO = gfs2_direct_IO,
};