file.c 57.0 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4
/*
 *   fs/cifs/file.c
 *
 *   vfs operations that deal with files
S
Steve French 已提交
5 6
 *
 *   Copyright (C) International Business Machines  Corp., 2002,2007
L
Linus Torvalds 已提交
7
 *   Author(s): Steve French (sfrench@us.ibm.com)
J
[CIFS]  
Jeremy Allison 已提交
8
 *              Jeremy Allison (jra@samba.org)
L
Linus Torvalds 已提交
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
 *
 *   This library is free software; you can redistribute it and/or modify
 *   it under the terms of the GNU Lesser General Public License as published
 *   by the Free Software Foundation; either version 2.1 of the License, or
 *   (at your option) any later version.
 *
 *   This library is distributed in the hope that it will be useful,
 *   but WITHOUT ANY WARRANTY; without even the implied warranty of
 *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
 *   the GNU Lesser General Public License for more details.
 *
 *   You should have received a copy of the GNU Lesser General Public License
 *   along with this library; if not, write to the Free Software
 *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
 */
#include <linux/fs.h>
25
#include <linux/backing-dev.h>
L
Linus Torvalds 已提交
26 27 28 29
#include <linux/stat.h>
#include <linux/fcntl.h>
#include <linux/pagemap.h>
#include <linux/pagevec.h>
30
#include <linux/writeback.h>
31
#include <linux/task_io_accounting_ops.h>
32
#include <linux/delay.h>
L
Linus Torvalds 已提交
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
#include <asm/div64.h>
#include "cifsfs.h"
#include "cifspdu.h"
#include "cifsglob.h"
#include "cifsproto.h"
#include "cifs_unicode.h"
#include "cifs_debug.h"
#include "cifs_fs_sb.h"

static inline struct cifsFileInfo *cifs_init_private(
	struct cifsFileInfo *private_data, struct inode *inode,
	struct file *file, __u16 netfid)
{
	memset(private_data, 0, sizeof(struct cifsFileInfo));
	private_data->netfid = netfid;
S
Steve French 已提交
48
	private_data->pid = current->tgid;
L
Linus Torvalds 已提交
49
	init_MUTEX(&private_data->fh_sem);
50
	mutex_init(&private_data->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
51
	INIT_LIST_HEAD(&private_data->llist);
L
Linus Torvalds 已提交
52 53 54 55
	private_data->pfile = file; /* needed for writepage */
	private_data->pInode = inode;
	private_data->invalidHandle = FALSE;
	private_data->closePend = FALSE;
56 57 58 59
	/* we have to track num writers to the inode, since writepages
	does not tell us which handle the write is for so there can
	be a close (overlapping with write) of the filehandle that
	cifs_writepages chose to use */
S
Steve French 已提交
60
	atomic_set(&private_data->wrtPending, 0);
L
Linus Torvalds 已提交
61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88

	return private_data;
}

static inline int cifs_convert_flags(unsigned int flags)
{
	if ((flags & O_ACCMODE) == O_RDONLY)
		return GENERIC_READ;
	else if ((flags & O_ACCMODE) == O_WRONLY)
		return GENERIC_WRITE;
	else if ((flags & O_ACCMODE) == O_RDWR) {
		/* GENERIC_ALL is too much permission to request
		   can cause unnecessary access denied on create */
		/* return GENERIC_ALL; */
		return (GENERIC_READ | GENERIC_WRITE);
	}

	return 0x20197;
}

static inline int cifs_get_disposition(unsigned int flags)
{
	if ((flags & (O_CREAT | O_EXCL)) == (O_CREAT | O_EXCL))
		return FILE_CREATE;
	else if ((flags & (O_CREAT | O_TRUNC)) == (O_CREAT | O_TRUNC))
		return FILE_OVERWRITE_IF;
	else if ((flags & O_CREAT) == O_CREAT)
		return FILE_OPEN_IF;
89 90
	else if ((flags & O_TRUNC) == O_TRUNC)
		return FILE_OVERWRITE;
L
Linus Torvalds 已提交
91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107
	else
		return FILE_OPEN;
}

/* all arguments to this function must be checked for validity in caller */
static inline int cifs_open_inode_helper(struct inode *inode, struct file *file,
	struct cifsInodeInfo *pCifsInode, struct cifsFileInfo *pCifsFile,
	struct cifsTconInfo *pTcon, int *oplock, FILE_ALL_INFO *buf,
	char *full_path, int xid)
{
	struct timespec temp;
	int rc;

	/* want handles we can use to read with first
	   in the list so we do not have to walk the
	   list to search for one in prepare_write */
	if ((file->f_flags & O_ACCMODE) == O_WRONLY) {
S
Steve French 已提交
108
		list_add_tail(&pCifsFile->flist,
L
Linus Torvalds 已提交
109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124
			      &pCifsInode->openFileList);
	} else {
		list_add(&pCifsFile->flist,
			 &pCifsInode->openFileList);
	}
	write_unlock(&GlobalSMBSeslock);
	if (pCifsInode->clientCanCacheRead) {
		/* we have the inode open somewhere else
		   no need to discard cache data */
		goto client_can_cache;
	}

	/* BB need same check in cifs_create too? */
	/* if not oplocked, invalidate inode pages if mtime or file
	   size changed */
	temp = cifs_NTtimeToUnix(le64_to_cpu(buf->LastWriteTime));
125 126
	if (timespec_equal(&file->f_path.dentry->d_inode->i_mtime, &temp) &&
			   (file->f_path.dentry->d_inode->i_size ==
L
Linus Torvalds 已提交
127 128 129
			    (loff_t)le64_to_cpu(buf->EndOfFile))) {
		cFYI(1, ("inode unchanged on server"));
	} else {
130
		if (file->f_path.dentry->d_inode->i_mapping) {
L
Linus Torvalds 已提交
131 132
		/* BB no need to lock inode until after invalidate
		   since namei code should already have it locked? */
133 134 135
			rc = filemap_write_and_wait(file->f_path.dentry->d_inode->i_mapping);
			if (rc != 0)
				CIFS_I(file->f_path.dentry->d_inode)->write_behind_rc = rc;
L
Linus Torvalds 已提交
136 137 138
		}
		cFYI(1, ("invalidating remote inode since open detected it "
			 "changed"));
139
		invalidate_remote_inode(file->f_path.dentry->d_inode);
L
Linus Torvalds 已提交
140 141 142
	}

client_can_cache:
143
	if (pTcon->unix_ext)
144
		rc = cifs_get_inode_info_unix(&file->f_path.dentry->d_inode,
L
Linus Torvalds 已提交
145 146
			full_path, inode->i_sb, xid);
	else
147
		rc = cifs_get_inode_info(&file->f_path.dentry->d_inode,
L
Linus Torvalds 已提交
148 149 150 151 152 153
			full_path, buf, inode->i_sb, xid);

	if ((*oplock & 0xF) == OPLOCK_EXCLUSIVE) {
		pCifsInode->clientCanCacheAll = TRUE;
		pCifsInode->clientCanCacheRead = TRUE;
		cFYI(1, ("Exclusive Oplock granted on inode %p",
154
			 file->f_path.dentry->d_inode));
L
Linus Torvalds 已提交
155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182
	} else if ((*oplock & 0xF) == OPLOCK_READ)
		pCifsInode->clientCanCacheRead = TRUE;

	return rc;
}

int cifs_open(struct inode *inode, struct file *file)
{
	int rc = -EACCES;
	int xid, oplock;
	struct cifs_sb_info *cifs_sb;
	struct cifsTconInfo *pTcon;
	struct cifsFileInfo *pCifsFile;
	struct cifsInodeInfo *pCifsInode;
	struct list_head *tmp;
	char *full_path = NULL;
	int desiredAccess;
	int disposition;
	__u16 netfid;
	FILE_ALL_INFO *buf = NULL;

	xid = GetXid();

	cifs_sb = CIFS_SB(inode->i_sb);
	pTcon = cifs_sb->tcon;

	if (file->f_flags & O_CREAT) {
		/* search inode for this file and fill in file->private_data */
183
		pCifsInode = CIFS_I(file->f_path.dentry->d_inode);
L
Linus Torvalds 已提交
184 185 186 187 188 189 190 191 192 193
		read_lock(&GlobalSMBSeslock);
		list_for_each(tmp, &pCifsInode->openFileList) {
			pCifsFile = list_entry(tmp, struct cifsFileInfo,
					       flist);
			if ((pCifsFile->pfile == NULL) &&
			    (pCifsFile->pid == current->tgid)) {
				/* mode set in cifs_create */

				/* needed for writepage */
				pCifsFile->pfile = file;
194

L
Linus Torvalds 已提交
195 196 197 198 199 200 201 202 203 204 205 206
				file->private_data = pCifsFile;
				break;
			}
		}
		read_unlock(&GlobalSMBSeslock);
		if (file->private_data != NULL) {
			rc = 0;
			FreeXid(xid);
			return rc;
		} else {
			if (file->f_flags & O_EXCL)
				cERROR(1, ("could not find file instance for "
207
					   "new file %p", file));
L
Linus Torvalds 已提交
208 209 210
		}
	}

211
	full_path = build_path_from_dentry(file->f_path.dentry);
L
Linus Torvalds 已提交
212 213 214 215 216
	if (full_path == NULL) {
		FreeXid(xid);
		return -ENOMEM;
	}

217
	cFYI(1, ("inode = 0x%p file flags are 0x%x for %s",
L
Linus Torvalds 已提交
218 219 220 221 222
		 inode, file->f_flags, full_path));
	desiredAccess = cifs_convert_flags(file->f_flags);

/*********************************************************************
 *  open flag mapping table:
S
Steve French 已提交
223
 *
L
Linus Torvalds 已提交
224
 *	POSIX Flag            CIFS Disposition
S
Steve French 已提交
225
 *	----------            ----------------
L
Linus Torvalds 已提交
226 227 228 229 230 231 232
 *	O_CREAT               FILE_OPEN_IF
 *	O_CREAT | O_EXCL      FILE_CREATE
 *	O_CREAT | O_TRUNC     FILE_OVERWRITE_IF
 *	O_TRUNC               FILE_OVERWRITE
 *	none of the above     FILE_OPEN
 *
 *	Note that there is not a direct match between disposition
S
Steve French 已提交
233
 *	FILE_SUPERSEDE (ie create whether or not file exists although
L
Linus Torvalds 已提交
234 235 236 237
 *	O_CREAT | O_TRUNC is similar but truncates the existing
 *	file rather than creating a new file as FILE_SUPERSEDE does
 *	(which uses the attributes / metadata passed in on open call)
 *?
S
Steve French 已提交
238
 *?  O_SYNC is a reasonable match to CIFS writethrough flag
L
Linus Torvalds 已提交
239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
 *?  and the read write flags match reasonably.  O_LARGEFILE
 *?  is irrelevant because largefile support is always used
 *?  by this client. Flags O_APPEND, O_DIRECT, O_DIRECTORY,
 *	 O_FASYNC, O_NOFOLLOW, O_NONBLOCK need further investigation
 *********************************************************************/

	disposition = cifs_get_disposition(file->f_flags);

	if (oplockEnabled)
		oplock = REQ_OPLOCK;
	else
		oplock = FALSE;

	/* BB pass O_SYNC flag through on file attributes .. BB */

	/* Also refresh inode by passing in file_info buf returned by SMBOpen
	   and calling get_inode_info with returned buf (at least helps
	   non-Unix server case) */

S
Steve French 已提交
258 259
	/* BB we can not do this if this is the second open of a file
	   and the first handle has writebehind data, we might be
L
Linus Torvalds 已提交
260 261 262 263 264 265
	   able to simply do a filemap_fdatawrite/filemap_fdatawait first */
	buf = kmalloc(sizeof(FILE_ALL_INFO), GFP_KERNEL);
	if (!buf) {
		rc = -ENOMEM;
		goto out;
	}
266 267

	if (cifs_sb->tcon->ses->capabilities & CAP_NT_SMBS)
S
Steve French 已提交
268
		rc = CIFSSMBOpen(xid, pTcon, full_path, disposition,
269
			 desiredAccess, CREATE_NOT_DIR, &netfid, &oplock, buf,
270 271
			 cifs_sb->local_nls, cifs_sb->mnt_cifs_flags
				 & CIFS_MOUNT_MAP_SPECIAL_CHR);
272 273 274
	else
		rc = -EIO; /* no NT SMB support fall into legacy open below */

275 276 277 278 279 280 281
	if (rc == -EIO) {
		/* Old server, try legacy style OpenX */
		rc = SMBLegacyOpen(xid, pTcon, full_path, disposition,
			desiredAccess, CREATE_NOT_DIR, &netfid, &oplock, buf,
			cifs_sb->local_nls, cifs_sb->mnt_cifs_flags
				& CIFS_MOUNT_MAP_SPECIAL_CHR);
	}
L
Linus Torvalds 已提交
282
	if (rc) {
283
		cFYI(1, ("cifs_open returned 0x%x", rc));
L
Linus Torvalds 已提交
284 285 286 287 288 289 290 291 292 293 294 295
		goto out;
	}
	file->private_data =
		kmalloc(sizeof(struct cifsFileInfo), GFP_KERNEL);
	if (file->private_data == NULL) {
		rc = -ENOMEM;
		goto out;
	}
	pCifsFile = cifs_init_private(file->private_data, inode, file, netfid);
	write_lock(&GlobalSMBSeslock);
	list_add(&pCifsFile->tlist, &pTcon->openFileList);

296
	pCifsInode = CIFS_I(file->f_path.dentry->d_inode);
L
Linus Torvalds 已提交
297 298 299 300 301 302 303 304
	if (pCifsInode) {
		rc = cifs_open_inode_helper(inode, file, pCifsInode,
					    pCifsFile, pTcon,
					    &oplock, buf, full_path, xid);
	} else {
		write_unlock(&GlobalSMBSeslock);
	}

S
Steve French 已提交
305
	if (oplock & CIFS_CREATE_ACTION) {
L
Linus Torvalds 已提交
306 307
		/* time to set mode which we can not set earlier due to
		   problems creating new read-only files */
308
		if (pTcon->unix_ext) {
L
Linus Torvalds 已提交
309 310 311
			CIFSSMBUnixSetPerms(xid, pTcon, full_path,
					    inode->i_mode,
					    (__u64)-1, (__u64)-1, 0 /* dev */,
312
					    cifs_sb->local_nls,
S
Steve French 已提交
313
					    cifs_sb->mnt_cifs_flags &
314
						CIFS_MOUNT_MAP_SPECIAL_CHR);
L
Linus Torvalds 已提交
315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
		} else {
			/* BB implement via Windows security descriptors eg
			   CIFSSMBWinSetPerms(xid, pTcon, full_path, mode,
					      -1, -1, local_nls);
			   in the meantime could set r/o dos attribute when
			   perms are eg: mode & 0222 == 0 */
		}
	}

out:
	kfree(buf);
	kfree(full_path);
	FreeXid(xid);
	return rc;
}

A
Adrian Bunk 已提交
331
/* Try to reacquire byte range locks that were released when session */
L
Linus Torvalds 已提交
332 333 334 335 336 337 338 339 340 341
/* to server was lost */
static int cifs_relock_file(struct cifsFileInfo *cifsFile)
{
	int rc = 0;

/* BB list all locks open on this file and relock */

	return rc;
}

342
static int cifs_reopen_file(struct file *file, int can_flush)
L
Linus Torvalds 已提交
343 344 345 346 347 348 349
{
	int rc = -EACCES;
	int xid, oplock;
	struct cifs_sb_info *cifs_sb;
	struct cifsTconInfo *pTcon;
	struct cifsFileInfo *pCifsFile;
	struct cifsInodeInfo *pCifsInode;
S
Steve French 已提交
350
	struct inode *inode;
L
Linus Torvalds 已提交
351 352 353 354 355
	char *full_path = NULL;
	int desiredAccess;
	int disposition = FILE_OPEN;
	__u16 netfid;

S
Steve French 已提交
356
	if (file->private_data)
L
Linus Torvalds 已提交
357
		pCifsFile = (struct cifsFileInfo *)file->private_data;
S
Steve French 已提交
358
	else
L
Linus Torvalds 已提交
359 360 361 362 363 364 365 366 367 368
		return -EBADF;

	xid = GetXid();
	down(&pCifsFile->fh_sem);
	if (pCifsFile->invalidHandle == FALSE) {
		up(&pCifsFile->fh_sem);
		FreeXid(xid);
		return 0;
	}

369
	if (file->f_path.dentry == NULL) {
370 371 372 373 374 375 376
		cERROR(1, ("no valid name if dentry freed"));
		dump_stack();
		rc = -EBADF;
		goto reopen_error_exit;
	}

	inode = file->f_path.dentry->d_inode;
S
Steve French 已提交
377
	if (inode == NULL) {
378 379 380 381
		cERROR(1, ("inode not valid"));
		dump_stack();
		rc = -EBADF;
		goto reopen_error_exit;
L
Linus Torvalds 已提交
382
	}
383

L
Linus Torvalds 已提交
384 385
	cifs_sb = CIFS_SB(inode->i_sb);
	pTcon = cifs_sb->tcon;
386

L
Linus Torvalds 已提交
387 388 389 390
/* can not grab rename sem here because various ops, including
   those that already have the rename sem can end up causing writepage
   to get called and if the server was down that means we end up here,
   and we can never tell if the caller already has the rename_sem */
391
	full_path = build_path_from_dentry(file->f_path.dentry);
L
Linus Torvalds 已提交
392
	if (full_path == NULL) {
393 394
		rc = -ENOMEM;
reopen_error_exit:
L
Linus Torvalds 已提交
395 396
		up(&pCifsFile->fh_sem);
		FreeXid(xid);
397
		return rc;
L
Linus Torvalds 已提交
398 399
	}

400
	cFYI(1, ("inode = 0x%p file flags 0x%x for %s",
S
Steve French 已提交
401
		 inode, file->f_flags, full_path));
L
Linus Torvalds 已提交
402 403 404 405 406 407 408 409
	desiredAccess = cifs_convert_flags(file->f_flags);

	if (oplockEnabled)
		oplock = REQ_OPLOCK;
	else
		oplock = FALSE;

	/* Can not refresh inode by passing in file_info buf to be returned
S
Steve French 已提交
410 411
	   by SMBOpen and then calling get_inode_info with returned buf
	   since file might have write behind data that needs to be flushed
L
Linus Torvalds 已提交
412 413 414 415 416
	   and server version of file size can be stale. If we knew for sure
	   that inode was not dirty locally we could do this */

	rc = CIFSSMBOpen(xid, pTcon, full_path, disposition, desiredAccess,
			 CREATE_NOT_DIR, &netfid, &oplock, NULL,
S
Steve French 已提交
417
			 cifs_sb->local_nls, cifs_sb->mnt_cifs_flags &
418
				CIFS_MOUNT_MAP_SPECIAL_CHR);
L
Linus Torvalds 已提交
419 420
	if (rc) {
		up(&pCifsFile->fh_sem);
421 422
		cFYI(1, ("cifs_open returned 0x%x", rc));
		cFYI(1, ("oplock: %d", oplock));
L
Linus Torvalds 已提交
423 424 425 426 427 428 429
	} else {
		pCifsFile->netfid = netfid;
		pCifsFile->invalidHandle = FALSE;
		up(&pCifsFile->fh_sem);
		pCifsInode = CIFS_I(inode);
		if (pCifsInode) {
			if (can_flush) {
430 431 432
				rc = filemap_write_and_wait(inode->i_mapping);
				if (rc != 0)
					CIFS_I(inode)->write_behind_rc = rc;
L
Linus Torvalds 已提交
433 434 435 436
			/* temporarily disable caching while we
			   go to server to get inode info */
				pCifsInode->clientCanCacheAll = FALSE;
				pCifsInode->clientCanCacheRead = FALSE;
437
				if (pTcon->unix_ext)
L
Linus Torvalds 已提交
438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453
					rc = cifs_get_inode_info_unix(&inode,
						full_path, inode->i_sb, xid);
				else
					rc = cifs_get_inode_info(&inode,
						full_path, NULL, inode->i_sb,
						xid);
			} /* else we are writing out data to server already
			     and could deadlock if we tried to flush data, and
			     since we do not know if we have data that would
			     invalidate the current end of file on the server
			     we can not go to the server to get the new inod
			     info */
			if ((oplock & 0xF) == OPLOCK_EXCLUSIVE) {
				pCifsInode->clientCanCacheAll = TRUE;
				pCifsInode->clientCanCacheRead = TRUE;
				cFYI(1, ("Exclusive Oplock granted on inode %p",
454
					 file->f_path.dentry->d_inode));
L
Linus Torvalds 已提交
455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473
			} else if ((oplock & 0xF) == OPLOCK_READ) {
				pCifsInode->clientCanCacheRead = TRUE;
				pCifsInode->clientCanCacheAll = FALSE;
			} else {
				pCifsInode->clientCanCacheRead = FALSE;
				pCifsInode->clientCanCacheAll = FALSE;
			}
			cifs_relock_file(pCifsFile);
		}
	}

	kfree(full_path);
	FreeXid(xid);
	return rc;
}

int cifs_close(struct inode *inode, struct file *file)
{
	int rc = 0;
474
	int xid, timeout;
L
Linus Torvalds 已提交
475 476 477 478 479 480 481 482 483 484
	struct cifs_sb_info *cifs_sb;
	struct cifsTconInfo *pTcon;
	struct cifsFileInfo *pSMBFile =
		(struct cifsFileInfo *)file->private_data;

	xid = GetXid();

	cifs_sb = CIFS_SB(inode->i_sb);
	pTcon = cifs_sb->tcon;
	if (pSMBFile) {
J
[CIFS]  
Jeremy Allison 已提交
485 486
		struct cifsLockInfo *li, *tmp;

L
Linus Torvalds 已提交
487 488 489 490 491
		pSMBFile->closePend = TRUE;
		if (pTcon) {
			/* no sense reconnecting to close a file that is
			   already closed */
			if (pTcon->tidStatus != CifsNeedReconnect) {
492
				timeout = 2;
S
Steve French 已提交
493
				while ((atomic_read(&pSMBFile->wrtPending) != 0)
494
					&& (timeout <= 2048)) {
495 496 497 498 499
					/* Give write a better chance to get to
					server ahead of the close.  We do not
					want to add a wait_q here as it would
					increase the memory utilization as
					the struct would be in each open file,
S
Steve French 已提交
500
					but this should give enough time to
501
					clear the socket */
502 503
					cFYI(DBG2,
						("close delay, write pending"));
504 505
					msleep(timeout);
					timeout *= 4;
506
				}
S
Steve French 已提交
507
				if (atomic_read(&pSMBFile->wrtPending))
508 509
					cERROR(1,
						("close with pending writes"));
L
Linus Torvalds 已提交
510 511 512 513
				rc = CIFSSMBClose(xid, pTcon,
						  pSMBFile->netfid);
			}
		}
J
[CIFS]  
Jeremy Allison 已提交
514 515 516

		/* Delete any outstanding lock records.
		   We'll lose them when the file is closed anyway. */
517
		mutex_lock(&pSMBFile->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
518 519 520 521
		list_for_each_entry_safe(li, tmp, &pSMBFile->llist, llist) {
			list_del(&li->llist);
			kfree(li);
		}
522
		mutex_unlock(&pSMBFile->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
523

524
		write_lock(&GlobalSMBSeslock);
L
Linus Torvalds 已提交
525 526
		list_del(&pSMBFile->flist);
		list_del(&pSMBFile->tlist);
527
		write_unlock(&GlobalSMBSeslock);
528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544
		timeout = 10;
		/* We waited above to give the SMBWrite a chance to issue
		   on the wire (so we do not get SMBWrite returning EBADF
		   if writepages is racing with close.  Note that writepages
		   does not specify a file handle, so it is possible for a file
		   to be opened twice, and the application close the "wrong"
		   file handle - in these cases we delay long enough to allow
		   the SMBWrite to get on the wire before the SMB Close.
		   We allow total wait here over 45 seconds, more than
		   oplock break time, and more than enough to allow any write
		   to complete on the server, or to time out on the client */
		while ((atomic_read(&pSMBFile->wrtPending) != 0)
				&& (timeout <= 50000)) {
			cERROR(1, ("writes pending, delay free of handle"));
			msleep(timeout);
			timeout *= 8;
		}
L
Linus Torvalds 已提交
545 546 547 548 549 550
		kfree(pSMBFile->search_resume_name);
		kfree(file->private_data);
		file->private_data = NULL;
	} else
		rc = -EBADF;

551
	read_lock(&GlobalSMBSeslock);
L
Linus Torvalds 已提交
552 553 554 555 556 557 558
	if (list_empty(&(CIFS_I(inode)->openFileList))) {
		cFYI(1, ("closing last open instance for inode %p", inode));
		/* if the file is not open we do not know if we can cache info
		   on this inode, much less write behind and read ahead */
		CIFS_I(inode)->clientCanCacheRead = FALSE;
		CIFS_I(inode)->clientCanCacheAll  = FALSE;
	}
559
	read_unlock(&GlobalSMBSeslock);
S
Steve French 已提交
560
	if ((rc == 0) && CIFS_I(inode)->write_behind_rc)
L
Linus Torvalds 已提交
561 562 563 564 565 566 567 568 569 570 571 572 573
		rc = CIFS_I(inode)->write_behind_rc;
	FreeXid(xid);
	return rc;
}

int cifs_closedir(struct inode *inode, struct file *file)
{
	int rc = 0;
	int xid;
	struct cifsFileInfo *pCFileStruct =
	    (struct cifsFileInfo *)file->private_data;
	char *ptmp;

574
	cFYI(1, ("Closedir inode = 0x%p", inode));
L
Linus Torvalds 已提交
575 576 577 578 579

	xid = GetXid();

	if (pCFileStruct) {
		struct cifsTconInfo *pTcon;
S
Steve French 已提交
580 581
		struct cifs_sb_info *cifs_sb =
			CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
582 583 584 585

		pTcon = cifs_sb->tcon;

		cFYI(1, ("Freeing private data in close dir"));
586 587
		if ((pCFileStruct->srch_inf.endOfSearch == FALSE) &&
		   (pCFileStruct->invalidHandle == FALSE)) {
L
Linus Torvalds 已提交
588 589 590 591 592 593 594 595 596
			pCFileStruct->invalidHandle = TRUE;
			rc = CIFSFindClose(xid, pTcon, pCFileStruct->netfid);
			cFYI(1, ("Closing uncompleted readdir with rc %d",
				 rc));
			/* not much we can do if it fails anyway, ignore rc */
			rc = 0;
		}
		ptmp = pCFileStruct->srch_inf.ntwrk_buf_start;
		if (ptmp) {
597
			cFYI(1, ("closedir free smb buf in srch struct"));
L
Linus Torvalds 已提交
598
			pCFileStruct->srch_inf.ntwrk_buf_start = NULL;
S
Steve French 已提交
599
			if (pCFileStruct->srch_inf.smallBuf)
600 601 602
				cifs_small_buf_release(ptmp);
			else
				cifs_buf_release(ptmp);
L
Linus Torvalds 已提交
603 604 605
		}
		ptmp = pCFileStruct->search_resume_name;
		if (ptmp) {
606
			cFYI(1, ("closedir free resume name"));
L
Linus Torvalds 已提交
607 608 609 610 611 612 613 614 615 616 617
			pCFileStruct->search_resume_name = NULL;
			kfree(ptmp);
		}
		kfree(file->private_data);
		file->private_data = NULL;
	}
	/* BB can we lock the filestruct while this is going on? */
	FreeXid(xid);
	return rc;
}

J
[CIFS]  
Jeremy Allison 已提交
618 619 620
static int store_file_lock(struct cifsFileInfo *fid, __u64 len,
				__u64 offset, __u8 lockType)
{
S
Steve French 已提交
621 622
	struct cifsLockInfo *li =
		kmalloc(sizeof(struct cifsLockInfo), GFP_KERNEL);
J
[CIFS]  
Jeremy Allison 已提交
623 624 625 626 627
	if (li == NULL)
		return -ENOMEM;
	li->offset = offset;
	li->length = len;
	li->type = lockType;
628
	mutex_lock(&fid->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
629
	list_add(&li->llist, &fid->llist);
630
	mutex_unlock(&fid->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
631 632 633
	return 0;
}

L
Linus Torvalds 已提交
634 635 636 637 638 639 640 641 642
int cifs_lock(struct file *file, int cmd, struct file_lock *pfLock)
{
	int rc, xid;
	__u32 numLock = 0;
	__u32 numUnlock = 0;
	__u64 length;
	int wait_flag = FALSE;
	struct cifs_sb_info *cifs_sb;
	struct cifsTconInfo *pTcon;
643 644
	__u16 netfid;
	__u8 lockType = LOCKING_ANDX_LARGE_FILES;
J
[CIFS]  
Jeremy Allison 已提交
645
	int posix_locking;
L
Linus Torvalds 已提交
646 647 648 649 650 651 652

	length = 1 + pfLock->fl_end - pfLock->fl_start;
	rc = -EACCES;
	xid = GetXid();

	cFYI(1, ("Lock parm: 0x%x flockflags: "
		 "0x%x flocktype: 0x%x start: %lld end: %lld",
S
Steve French 已提交
653 654
		cmd, pfLock->fl_flags, pfLock->fl_type, pfLock->fl_start,
		pfLock->fl_end));
L
Linus Torvalds 已提交
655 656

	if (pfLock->fl_flags & FL_POSIX)
657
		cFYI(1, ("Posix"));
L
Linus Torvalds 已提交
658
	if (pfLock->fl_flags & FL_FLOCK)
659
		cFYI(1, ("Flock"));
L
Linus Torvalds 已提交
660
	if (pfLock->fl_flags & FL_SLEEP) {
661
		cFYI(1, ("Blocking lock"));
L
Linus Torvalds 已提交
662 663 664 665
		wait_flag = TRUE;
	}
	if (pfLock->fl_flags & FL_ACCESS)
		cFYI(1, ("Process suspended by mandatory locking - "
666
			 "not implemented yet"));
L
Linus Torvalds 已提交
667 668
	if (pfLock->fl_flags & FL_LEASE)
		cFYI(1, ("Lease on file - not implemented yet"));
S
Steve French 已提交
669
	if (pfLock->fl_flags &
L
Linus Torvalds 已提交
670 671 672 673 674 675 676
	    (~(FL_POSIX | FL_FLOCK | FL_SLEEP | FL_ACCESS | FL_LEASE)))
		cFYI(1, ("Unknown lock flags 0x%x", pfLock->fl_flags));

	if (pfLock->fl_type == F_WRLCK) {
		cFYI(1, ("F_WRLCK "));
		numLock = 1;
	} else if (pfLock->fl_type == F_UNLCK) {
677
		cFYI(1, ("F_UNLCK"));
L
Linus Torvalds 已提交
678
		numUnlock = 1;
679 680
		/* Check if unlock includes more than
		one lock range */
L
Linus Torvalds 已提交
681
	} else if (pfLock->fl_type == F_RDLCK) {
682
		cFYI(1, ("F_RDLCK"));
L
Linus Torvalds 已提交
683 684 685
		lockType |= LOCKING_ANDX_SHARED_LOCK;
		numLock = 1;
	} else if (pfLock->fl_type == F_EXLCK) {
686
		cFYI(1, ("F_EXLCK"));
L
Linus Torvalds 已提交
687 688
		numLock = 1;
	} else if (pfLock->fl_type == F_SHLCK) {
689
		cFYI(1, ("F_SHLCK"));
L
Linus Torvalds 已提交
690 691 692
		lockType |= LOCKING_ANDX_SHARED_LOCK;
		numLock = 1;
	} else
693
		cFYI(1, ("Unknown type of lock"));
L
Linus Torvalds 已提交
694

695
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
696 697 698 699 700 701
	pTcon = cifs_sb->tcon;

	if (file->private_data == NULL) {
		FreeXid(xid);
		return -EBADF;
	}
702 703
	netfid = ((struct cifsFileInfo *)file->private_data)->netfid;

J
[CIFS]  
Jeremy Allison 已提交
704 705
	posix_locking = (cifs_sb->tcon->ses->capabilities & CAP_UNIX) &&
			(CIFS_UNIX_FCNTL_CAP & le64_to_cpu(cifs_sb->tcon->fsUnixInfo.Capability));
L
Linus Torvalds 已提交
706

707 708 709
	/* BB add code here to normalize offset and length to
	account for negative length which we can not accept over the
	wire */
L
Linus Torvalds 已提交
710
	if (IS_GETLK(cmd)) {
S
Steve French 已提交
711
		if (posix_locking) {
712
			int posix_lock_type;
S
Steve French 已提交
713
			if (lockType & LOCKING_ANDX_SHARED_LOCK)
714 715 716 717
				posix_lock_type = CIFS_RDLCK;
			else
				posix_lock_type = CIFS_WRLCK;
			rc = CIFSSMBPosixLock(xid, pTcon, netfid, 1 /* get */,
718
					length,	pfLock,
719 720 721 722 723 724 725 726
					posix_lock_type, wait_flag);
			FreeXid(xid);
			return rc;
		}

		/* BB we could chain these into one lock request BB */
		rc = CIFSSMBLock(xid, pTcon, netfid, length, pfLock->fl_start,
				 0, 1, lockType, 0 /* wait flag */ );
L
Linus Torvalds 已提交
727
		if (rc == 0) {
S
Steve French 已提交
728
			rc = CIFSSMBLock(xid, pTcon, netfid, length,
L
Linus Torvalds 已提交
729 730 731 732 733 734
					 pfLock->fl_start, 1 /* numUnlock */ ,
					 0 /* numLock */ , lockType,
					 0 /* wait flag */ );
			pfLock->fl_type = F_UNLCK;
			if (rc != 0)
				cERROR(1, ("Error unlocking previously locked "
735
					   "range %d during test of lock", rc));
L
Linus Torvalds 已提交
736 737 738 739 740 741 742 743 744 745 746
			rc = 0;

		} else {
			/* if rc == ERR_SHARING_VIOLATION ? */
			rc = 0;	/* do not change lock type to unlock
				   since range in use */
		}

		FreeXid(xid);
		return rc;
	}
J
[CIFS]  
Jeremy Allison 已提交
747 748 749 750 751 752 753 754 755

	if (!numLock && !numUnlock) {
		/* if no lock or unlock then nothing
		to do since we do not know what it is */
		FreeXid(xid);
		return -EOPNOTSUPP;
	}

	if (posix_locking) {
756
		int posix_lock_type;
S
Steve French 已提交
757
		if (lockType & LOCKING_ANDX_SHARED_LOCK)
758 759 760
			posix_lock_type = CIFS_RDLCK;
		else
			posix_lock_type = CIFS_WRLCK;
761

S
Steve French 已提交
762
		if (numUnlock == 1)
763
			posix_lock_type = CIFS_UNLCK;
J
[CIFS]  
Jeremy Allison 已提交
764

765
		rc = CIFSSMBPosixLock(xid, pTcon, netfid, 0 /* set */,
766
				      length, pfLock,
767
				      posix_lock_type, wait_flag);
J
[CIFS]  
Jeremy Allison 已提交
768
	} else {
S
Steve French 已提交
769 770
		struct cifsFileInfo *fid =
			(struct cifsFileInfo *)file->private_data;
J
[CIFS]  
Jeremy Allison 已提交
771 772

		if (numLock) {
S
Steve French 已提交
773 774
			rc = CIFSSMBLock(xid, pTcon, netfid, length,
					pfLock->fl_start,
J
[CIFS]  
Jeremy Allison 已提交
775 776 777 778 779 780 781 782 783 784 785 786 787
					0, numLock, lockType, wait_flag);

			if (rc == 0) {
				/* For Windows locks we must store them. */
				rc = store_file_lock(fid, length,
						pfLock->fl_start, lockType);
			}
		} else if (numUnlock) {
			/* For each stored lock that this unlock overlaps
			   completely, unlock it. */
			int stored_rc = 0;
			struct cifsLockInfo *li, *tmp;

788
			rc = 0;
789
			mutex_lock(&fid->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
790 791
			list_for_each_entry_safe(li, tmp, &fid->llist, llist) {
				if (pfLock->fl_start <= li->offset &&
S
Steve French 已提交
792
						(pfLock->fl_start + length) >=
793
						(li->offset + li->length)) {
S
Steve French 已提交
794 795
					stored_rc = CIFSSMBLock(xid, pTcon,
							netfid,
J
[CIFS]  
Jeremy Allison 已提交
796 797 798 799 800 801 802 803 804
							li->length, li->offset,
							1, 0, li->type, FALSE);
					if (stored_rc)
						rc = stored_rc;

					list_del(&li->llist);
					kfree(li);
				}
			}
805
			mutex_unlock(&fid->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
806 807 808
		}
	}

809
	if (pfLock->fl_flags & FL_POSIX)
L
Linus Torvalds 已提交
810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825
		posix_lock_file_wait(file, pfLock);
	FreeXid(xid);
	return rc;
}

ssize_t cifs_user_write(struct file *file, const char __user *write_data,
	size_t write_size, loff_t *poffset)
{
	int rc = 0;
	unsigned int bytes_written = 0;
	unsigned int total_written;
	struct cifs_sb_info *cifs_sb;
	struct cifsTconInfo *pTcon;
	int xid, long_op;
	struct cifsFileInfo *open_file;

826
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
827 828 829 830 831

	pTcon = cifs_sb->tcon;

	/* cFYI(1,
	   (" write %d bytes to offset %lld of %s", write_size,
832
	   *poffset, file->f_path.dentry->d_name.name)); */
L
Linus Torvalds 已提交
833 834 835

	if (file->private_data == NULL)
		return -EBADF;
836
	open_file = (struct cifsFileInfo *) file->private_data;
837

L
Linus Torvalds 已提交
838 839
	xid = GetXid();

840
	if (*poffset > file->f_path.dentry->d_inode->i_size)
841
		long_op = CIFS_VLONG_OP; /* writes past EOF take long time */
L
Linus Torvalds 已提交
842
	else
843
		long_op = CIFS_LONG_OP;
L
Linus Torvalds 已提交
844 845 846 847 848 849 850 851 852 853 854 855

	for (total_written = 0; write_size > total_written;
	     total_written += bytes_written) {
		rc = -EAGAIN;
		while (rc == -EAGAIN) {
			if (file->private_data == NULL) {
				/* file has been closed on us */
				FreeXid(xid);
			/* if we have gotten here we have written some data
			   and blocked, and the file has been freed on us while
			   we blocked so return what we managed to write */
				return total_written;
S
Steve French 已提交
856
			}
L
Linus Torvalds 已提交
857 858 859 860 861 862 863 864 865 866 867 868
			if (open_file->closePend) {
				FreeXid(xid);
				if (total_written)
					return total_written;
				else
					return -EBADF;
			}
			if (open_file->invalidHandle) {
				/* we could deadlock if we called
				   filemap_fdatawait from here so tell
				   reopen_file not to flush data to server
				   now */
869
				rc = cifs_reopen_file(file, FALSE);
L
Linus Torvalds 已提交
870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889
				if (rc != 0)
					break;
			}

			rc = CIFSSMBWrite(xid, pTcon,
				open_file->netfid,
				min_t(const int, cifs_sb->wsize,
				      write_size - total_written),
				*poffset, &bytes_written,
				NULL, write_data + total_written, long_op);
		}
		if (rc || (bytes_written == 0)) {
			if (total_written)
				break;
			else {
				FreeXid(xid);
				return rc;
			}
		} else
			*poffset += bytes_written;
890
		long_op = CIFS_STD_OP; /* subsequent writes fast -
L
Linus Torvalds 已提交
891 892 893
				    15 seconds is plenty */
	}

894
	cifs_stats_bytes_written(pTcon, total_written);
L
Linus Torvalds 已提交
895 896

	/* since the write may have blocked check these pointers again */
897 898
	if ((file->f_path.dentry) && (file->f_path.dentry->d_inode)) {
		struct inode *inode = file->f_path.dentry->d_inode;
S
Steve French 已提交
899 900
/* Do not update local mtime - server will set its actual value on write
 *		inode->i_ctime = inode->i_mtime =
901 902 903 904 905
 * 			current_fs_time(inode->i_sb);*/
		if (total_written > 0) {
			spin_lock(&inode->i_lock);
			if (*poffset > file->f_path.dentry->d_inode->i_size)
				i_size_write(file->f_path.dentry->d_inode,
L
Linus Torvalds 已提交
906
					*poffset);
907
			spin_unlock(&inode->i_lock);
L
Linus Torvalds 已提交
908
		}
S
Steve French 已提交
909
		mark_inode_dirty_sync(file->f_path.dentry->d_inode);
L
Linus Torvalds 已提交
910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925
	}
	FreeXid(xid);
	return total_written;
}

static ssize_t cifs_write(struct file *file, const char *write_data,
	size_t write_size, loff_t *poffset)
{
	int rc = 0;
	unsigned int bytes_written = 0;
	unsigned int total_written;
	struct cifs_sb_info *cifs_sb;
	struct cifsTconInfo *pTcon;
	int xid, long_op;
	struct cifsFileInfo *open_file;

926
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
927 928 929

	pTcon = cifs_sb->tcon;

S
Steve French 已提交
930
	cFYI(1, ("write %zd bytes to offset %lld of %s", write_size,
931
	   *poffset, file->f_path.dentry->d_name.name));
L
Linus Torvalds 已提交
932 933 934

	if (file->private_data == NULL)
		return -EBADF;
935
	open_file = (struct cifsFileInfo *)file->private_data;
936

L
Linus Torvalds 已提交
937 938
	xid = GetXid();

939
	if (*poffset > file->f_path.dentry->d_inode->i_size)
940
		long_op = CIFS_VLONG_OP; /* writes past EOF can be slow */
L
Linus Torvalds 已提交
941
	else
942
		long_op = CIFS_LONG_OP;
L
Linus Torvalds 已提交
943 944 945 946 947 948 949 950 951 952

	for (total_written = 0; write_size > total_written;
	     total_written += bytes_written) {
		rc = -EAGAIN;
		while (rc == -EAGAIN) {
			if (file->private_data == NULL) {
				/* file has been closed on us */
				FreeXid(xid);
			/* if we have gotten here we have written some data
			   and blocked, and the file has been freed on us
S
Steve French 已提交
953
			   while we blocked so return what we managed to
L
Linus Torvalds 已提交
954 955
			   write */
				return total_written;
S
Steve French 已提交
956
			}
L
Linus Torvalds 已提交
957 958 959 960 961 962 963 964 965 966
			if (open_file->closePend) {
				FreeXid(xid);
				if (total_written)
					return total_written;
				else
					return -EBADF;
			}
			if (open_file->invalidHandle) {
				/* we could deadlock if we called
				   filemap_fdatawait from here so tell
S
Steve French 已提交
967
				   reopen_file not to flush data to
L
Linus Torvalds 已提交
968
				   server now */
969
				rc = cifs_reopen_file(file, FALSE);
L
Linus Torvalds 已提交
970 971 972
				if (rc != 0)
					break;
			}
S
Steve French 已提交
973 974
			if (experimEnabled || (pTcon->ses->server &&
				((pTcon->ses->server->secMode &
S
Steve French 已提交
975
				(SECMODE_SIGN_REQUIRED | SECMODE_SIGN_ENABLED))
S
Steve French 已提交
976
				== 0))) {
977 978 979
				struct kvec iov[2];
				unsigned int len;

980
				len = min((size_t)cifs_sb->wsize,
981 982 983 984 985
					  write_size - total_written);
				/* iov[0] is reserved for smb header */
				iov[1].iov_base = (char *)write_data +
						  total_written;
				iov[1].iov_len = len;
986
				rc = CIFSSMBWrite2(xid, pTcon,
987
						open_file->netfid, len,
988
						*poffset, &bytes_written,
989
						iov, 1, long_op);
990
			} else
991 992 993 994 995 996 997
				rc = CIFSSMBWrite(xid, pTcon,
					 open_file->netfid,
					 min_t(const int, cifs_sb->wsize,
					       write_size - total_written),
					 *poffset, &bytes_written,
					 write_data + total_written,
					 NULL, long_op);
L
Linus Torvalds 已提交
998 999 1000 1001 1002 1003 1004 1005 1006 1007
		}
		if (rc || (bytes_written == 0)) {
			if (total_written)
				break;
			else {
				FreeXid(xid);
				return rc;
			}
		} else
			*poffset += bytes_written;
1008
		long_op = CIFS_STD_OP; /* subsequent writes fast -
L
Linus Torvalds 已提交
1009 1010 1011
				    15 seconds is plenty */
	}

1012
	cifs_stats_bytes_written(pTcon, total_written);
L
Linus Torvalds 已提交
1013 1014

	/* since the write may have blocked check these pointers again */
1015
	if ((file->f_path.dentry) && (file->f_path.dentry->d_inode)) {
1016
/*BB We could make this contingent on superblock ATIME flag too */
1017 1018 1019 1020 1021 1022 1023 1024
/*		file->f_path.dentry->d_inode->i_ctime =
		file->f_path.dentry->d_inode->i_mtime = CURRENT_TIME;*/
		if (total_written > 0) {
			spin_lock(&file->f_path.dentry->d_inode->i_lock);
			if (*poffset > file->f_path.dentry->d_inode->i_size)
				i_size_write(file->f_path.dentry->d_inode,
					     *poffset);
			spin_unlock(&file->f_path.dentry->d_inode->i_lock);
L
Linus Torvalds 已提交
1025
		}
1026
		mark_inode_dirty_sync(file->f_path.dentry->d_inode);
L
Linus Torvalds 已提交
1027 1028 1029 1030 1031
	}
	FreeXid(xid);
	return total_written;
}

S
Steve French 已提交
1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
#ifdef CONFIG_CIFS_EXPERIMENTAL
struct cifsFileInfo *find_readable_file(struct cifsInodeInfo *cifs_inode)
{
	struct cifsFileInfo *open_file = NULL;

	read_lock(&GlobalSMBSeslock);
	/* we could simply get the first_list_entry since write-only entries
	   are always at the end of the list but since the first entry might
	   have a close pending, we go through the whole list */
	list_for_each_entry(open_file, &cifs_inode->openFileList, flist) {
		if (open_file->closePend)
			continue;
		if (open_file->pfile && ((open_file->pfile->f_flags & O_RDWR) ||
		    (open_file->pfile->f_flags & O_RDONLY))) {
			if (!open_file->invalidHandle) {
				/* found a good file */
				/* lock it so it will not be closed on us */
				atomic_inc(&open_file->wrtPending);
				read_unlock(&GlobalSMBSeslock);
				return open_file;
			} /* else might as well continue, and look for
			     another, or simply have the caller reopen it
			     again rather than trying to fix this handle */
		} else /* write only file */
			break; /* write only files are last so must be done */
	}
	read_unlock(&GlobalSMBSeslock);
	return NULL;
}
#endif

1063
struct cifsFileInfo *find_writable_file(struct cifsInodeInfo *cifs_inode)
1064 1065
{
	struct cifsFileInfo *open_file;
1066
	int rc;
1067

1068 1069 1070 1071
	/* Having a null inode here (because mapping->host was set to zero by
	the VFS or MM) should not happen but we had reports of on oops (due to
	it being zero) during stress testcases so we need to check for it */

S
Steve French 已提交
1072 1073
	if (cifs_inode == NULL) {
		cERROR(1, ("Null inode passed to cifs_writeable_file"));
1074 1075 1076 1077
		dump_stack();
		return NULL;
	}

1078
	read_lock(&GlobalSMBSeslock);
1079
refind_writable:
1080 1081 1082 1083 1084 1085
	list_for_each_entry(open_file, &cifs_inode->openFileList, flist) {
		if (open_file->closePend)
			continue;
		if (open_file->pfile &&
		    ((open_file->pfile->f_flags & O_RDWR) ||
		     (open_file->pfile->f_flags & O_WRONLY))) {
1086
			atomic_inc(&open_file->wrtPending);
1087 1088 1089 1090 1091 1092

			if (!open_file->invalidHandle) {
				/* found a good writable file */
				read_unlock(&GlobalSMBSeslock);
				return open_file;
			}
S
Steve French 已提交
1093

1094
			read_unlock(&GlobalSMBSeslock);
1095 1096
			/* Had to unlock since following call can block */
			rc = cifs_reopen_file(open_file->pfile, FALSE);
S
Steve French 已提交
1097
			if (!rc) {
1098 1099 1100 1101
				if (!open_file->closePend)
					return open_file;
				else { /* start over in case this was deleted */
				       /* since the list could be modified */
1102
					read_lock(&GlobalSMBSeslock);
1103
					atomic_dec(&open_file->wrtPending);
1104
					goto refind_writable;
1105 1106
				}
			}
1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119

			/* if it fails, try another handle if possible -
			(we can not do this if closePending since
			loop could be modified - in which case we
			have to start at the beginning of the list
			again. Note that it would be bad
			to hold up writepages here (rather than
			in caller) with continuous retries */
			cFYI(1, ("wp failed on reopen file"));
			read_lock(&GlobalSMBSeslock);
			/* can not use this handle, no write
			   pending on this one after all */
			atomic_dec(&open_file->wrtPending);
S
Steve French 已提交
1120

1121 1122 1123 1124 1125 1126 1127 1128 1129
			if (open_file->closePend) /* list could have changed */
				goto refind_writable;
			/* else we simply continue to the next entry. Thus
			   we do not loop on reopen errors.  If we
			   can not reopen the file, for example if we
			   reconnected to a server with another client
			   racing to delete or lock the file we would not
			   make progress if we restarted before the beginning
			   of the loop here. */
1130 1131 1132 1133 1134 1135
		}
	}
	read_unlock(&GlobalSMBSeslock);
	return NULL;
}

L
Linus Torvalds 已提交
1136 1137 1138 1139 1140 1141 1142 1143 1144 1145
static int cifs_partialpagewrite(struct page *page, unsigned from, unsigned to)
{
	struct address_space *mapping = page->mapping;
	loff_t offset = (loff_t)page->index << PAGE_CACHE_SHIFT;
	char *write_data;
	int rc = -EFAULT;
	int bytes_written = 0;
	struct cifs_sb_info *cifs_sb;
	struct cifsTconInfo *pTcon;
	struct inode *inode;
1146
	struct cifsFileInfo *open_file;
L
Linus Torvalds 已提交
1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171

	if (!mapping || !mapping->host)
		return -EFAULT;

	inode = page->mapping->host;
	cifs_sb = CIFS_SB(inode->i_sb);
	pTcon = cifs_sb->tcon;

	offset += (loff_t)from;
	write_data = kmap(page);
	write_data += from;

	if ((to > PAGE_CACHE_SIZE) || (from > to)) {
		kunmap(page);
		return -EIO;
	}

	/* racing with truncate? */
	if (offset > mapping->host->i_size) {
		kunmap(page);
		return 0; /* don't care */
	}

	/* check to make sure that we are not extending the file */
	if (mapping->host->i_size - offset < (loff_t)to)
S
Steve French 已提交
1172
		to = (unsigned)(mapping->host->i_size - offset);
L
Linus Torvalds 已提交
1173

1174 1175 1176 1177
	open_file = find_writable_file(CIFS_I(mapping->host));
	if (open_file) {
		bytes_written = cifs_write(open_file->pfile, write_data,
					   to-from, &offset);
1178
		atomic_dec(&open_file->wrtPending);
L
Linus Torvalds 已提交
1179
		/* Does mm or vfs already set times? */
1180
		inode->i_atime = inode->i_mtime = current_fs_time(inode->i_sb);
1181
		if ((bytes_written > 0) && (offset))
1182
			rc = 0;
1183 1184
		else if (bytes_written < 0)
			rc = bytes_written;
1185
	} else {
L
Linus Torvalds 已提交
1186 1187 1188 1189 1190 1191 1192 1193 1194
		cFYI(1, ("No writeable filehandles for inode"));
		rc = -EIO;
	}

	kunmap(page);
	return rc;
}

static int cifs_writepages(struct address_space *mapping,
1195
			   struct writeback_control *wbc)
L
Linus Torvalds 已提交
1196
{
1197 1198 1199 1200 1201
	struct backing_dev_info *bdi = mapping->backing_dev_info;
	unsigned int bytes_to_write;
	unsigned int bytes_written;
	struct cifs_sb_info *cifs_sb;
	int done = 0;
1202
	pgoff_t end;
1203
	pgoff_t index;
S
Steve French 已提交
1204 1205
	int range_whole = 0;
	struct kvec *iov;
1206
	int len;
1207 1208 1209 1210
	int n_iov = 0;
	pgoff_t next;
	int nr_pages;
	__u64 offset = 0;
1211
	struct cifsFileInfo *open_file;
1212 1213 1214 1215
	struct page *page;
	struct pagevec pvec;
	int rc = 0;
	int scanned = 0;
L
Linus Torvalds 已提交
1216 1217
	int xid;

1218
	cifs_sb = CIFS_SB(mapping->host->i_sb);
1219

1220 1221 1222 1223 1224 1225 1226
	/*
	 * If wsize is smaller that the page cache size, default to writing
	 * one page at a time via cifs_writepage
	 */
	if (cifs_sb->wsize < PAGE_CACHE_SIZE)
		return generic_writepages(mapping, wbc);

S
Steve French 已提交
1227 1228 1229 1230
	if ((cifs_sb->tcon->ses) && (cifs_sb->tcon->ses->server))
		if (cifs_sb->tcon->ses->server->secMode &
				(SECMODE_SIGN_REQUIRED | SECMODE_SIGN_ENABLED))
			if (!experimEnabled)
1231
				return generic_writepages(mapping, wbc);
1232

1233
	iov = kmalloc(32 * sizeof(struct kvec), GFP_KERNEL);
S
Steve French 已提交
1234
	if (iov == NULL)
1235 1236 1237
		return generic_writepages(mapping, wbc);


1238 1239 1240 1241 1242 1243
	/*
	 * BB: Is this meaningful for a non-block-device file system?
	 * If it is, we should test it again after we do I/O
	 */
	if (wbc->nonblocking && bdi_write_congested(bdi)) {
		wbc->encountered_congestion = 1;
1244
		kfree(iov);
1245 1246 1247
		return 0;
	}

L
Linus Torvalds 已提交
1248 1249
	xid = GetXid();

1250
	pagevec_init(&pvec, 0);
1251
	if (wbc->range_cyclic) {
1252
		index = mapping->writeback_index; /* Start from prev offset */
1253 1254 1255 1256 1257 1258
		end = -1;
	} else {
		index = wbc->range_start >> PAGE_CACHE_SHIFT;
		end = wbc->range_end >> PAGE_CACHE_SHIFT;
		if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
			range_whole = 1;
1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293
		scanned = 1;
	}
retry:
	while (!done && (index <= end) &&
	       (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
			PAGECACHE_TAG_DIRTY,
			min(end - index, (pgoff_t)PAGEVEC_SIZE - 1) + 1))) {
		int first;
		unsigned int i;

		first = -1;
		next = 0;
		n_iov = 0;
		bytes_to_write = 0;

		for (i = 0; i < nr_pages; i++) {
			page = pvec.pages[i];
			/*
			 * At this point we hold neither mapping->tree_lock nor
			 * lock on the page itself: the page may be truncated or
			 * invalidated (changing page->mapping to NULL), or even
			 * swizzled back from swapper_space to tmpfs file
			 * mapping
			 */

			if (first < 0)
				lock_page(page);
			else if (TestSetPageLocked(page))
				break;

			if (unlikely(page->mapping != mapping)) {
				unlock_page(page);
				break;
			}

1294
			if (!wbc->range_cyclic && page->index > end) {
1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309
				done = 1;
				unlock_page(page);
				break;
			}

			if (next && (page->index != next)) {
				/* Not next consecutive page */
				unlock_page(page);
				break;
			}

			if (wbc->sync_mode != WB_SYNC_NONE)
				wait_on_page_writeback(page);

			if (PageWriteback(page) ||
1310
					!clear_page_dirty_for_io(page)) {
1311 1312 1313
				unlock_page(page);
				break;
			}
1314

1315 1316 1317 1318 1319 1320
			/*
			 * This actually clears the dirty bit in the radix tree.
			 * See cifs_writepage() for more commentary.
			 */
			set_page_writeback(page);

1321 1322 1323
			if (page_offset(page) >= mapping->host->i_size) {
				done = 1;
				unlock_page(page);
1324
				end_page_writeback(page);
1325 1326 1327
				break;
			}

1328 1329 1330 1331 1332
			/*
			 * BB can we get rid of this?  pages are held by pvec
			 */
			page_cache_get(page);

1333 1334 1335
			len = min(mapping->host->i_size - page_offset(page),
				  (loff_t)PAGE_CACHE_SIZE);

1336 1337 1338
			/* reserve iov[0] for the smb header */
			n_iov++;
			iov[n_iov].iov_base = kmap(page);
1339 1340
			iov[n_iov].iov_len = len;
			bytes_to_write += len;
1341 1342 1343 1344 1345 1346 1347 1348 1349 1350

			if (first < 0) {
				first = i;
				offset = page_offset(page);
			}
			next = page->index + 1;
			if (bytes_to_write + PAGE_CACHE_SIZE > cifs_sb->wsize)
				break;
		}
		if (n_iov) {
1351 1352 1353 1354 1355 1356 1357 1358
			/* Search for a writable handle every time we call
			 * CIFSSMBWrite2.  We can't rely on the last handle
			 * we used to still be valid
			 */
			open_file = find_writable_file(CIFS_I(mapping->host));
			if (!open_file) {
				cERROR(1, ("No writable handles for inode"));
				rc = -EBADF;
S
Steve French 已提交
1359
			} else {
1360 1361 1362 1363
				rc = CIFSSMBWrite2(xid, cifs_sb->tcon,
						   open_file->netfid,
						   bytes_to_write, offset,
						   &bytes_written, iov, n_iov,
1364
						   CIFS_LONG_OP);
1365 1366
				atomic_dec(&open_file->wrtPending);
				if (rc || bytes_written < bytes_to_write) {
1367
					cERROR(1, ("Write2 ret %d, wrote %d",
1368 1369 1370
						  rc, bytes_written));
					/* BB what if continued retry is
					   requested via mount flags? */
1371 1372 1373 1374
					if (rc == -ENOSPC)
						set_bit(AS_ENOSPC, &mapping->flags);
					else
						set_bit(AS_EIO, &mapping->flags);
1375 1376 1377 1378
				} else {
					cifs_stats_bytes_written(cifs_sb->tcon,
								 bytes_written);
				}
1379 1380 1381
			}
			for (i = 0; i < n_iov; i++) {
				page = pvec.pages[first + i];
1382 1383 1384 1385
				/* Should we also set page error on
				success rc but too little data written? */
				/* BB investigate retry logic on temporary
				server crash cases and how recovery works
S
Steve French 已提交
1386 1387
				when page marked as error */
				if (rc)
1388
					SetPageError(page);
1389 1390
				kunmap(page);
				unlock_page(page);
1391
				end_page_writeback(page);
1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408
				page_cache_release(page);
			}
			if ((wbc->nr_to_write -= n_iov) <= 0)
				done = 1;
			index = next;
		}
		pagevec_release(&pvec);
	}
	if (!scanned && !done) {
		/*
		 * We hit the last page and there is more work to be done: wrap
		 * back to the start of the file
		 */
		scanned = 1;
		index = 0;
		goto retry;
	}
1409
	if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0))
1410 1411
		mapping->writeback_index = index;

L
Linus Torvalds 已提交
1412
	FreeXid(xid);
1413
	kfree(iov);
L
Linus Torvalds 已提交
1414 1415 1416
	return rc;
}

S
Steve French 已提交
1417
static int cifs_writepage(struct page *page, struct writeback_control *wbc)
L
Linus Torvalds 已提交
1418 1419 1420 1421 1422 1423 1424
{
	int rc = -EFAULT;
	int xid;

	xid = GetXid();
/* BB add check for wbc flags */
	page_cache_get(page);
S
Steve French 已提交
1425
	if (!PageUptodate(page))
L
Linus Torvalds 已提交
1426
		cFYI(1, ("ppw - page not up to date"));
1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437

	/*
	 * Set the "writeback" flag, and clear "dirty" in the radix tree.
	 *
	 * A writepage() implementation always needs to do either this,
	 * or re-dirty the page with "redirty_page_for_writepage()" in
	 * the case of a failure.
	 *
	 * Just unlocking the page will cause the radix tree tag-bits
	 * to fail to update with the state of the page correctly.
	 */
S
Steve French 已提交
1438
	set_page_writeback(page);
L
Linus Torvalds 已提交
1439 1440 1441
	rc = cifs_partialpagewrite(page, 0, PAGE_CACHE_SIZE);
	SetPageUptodate(page); /* BB add check for error and Clearuptodate? */
	unlock_page(page);
1442 1443
	end_page_writeback(page);
	page_cache_release(page);
L
Linus Torvalds 已提交
1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457
	FreeXid(xid);
	return rc;
}

static int cifs_commit_write(struct file *file, struct page *page,
	unsigned offset, unsigned to)
{
	int xid;
	int rc = 0;
	struct inode *inode = page->mapping->host;
	loff_t position = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to;
	char *page_data;

	xid = GetXid();
S
Steve French 已提交
1458
	cFYI(1, ("commit write for page %p up to position %lld for %d",
L
Linus Torvalds 已提交
1459
		 page, position, to));
1460
	spin_lock(&inode->i_lock);
S
Steve French 已提交
1461
	if (position > inode->i_size)
L
Linus Torvalds 已提交
1462
		i_size_write(inode, position);
S
Steve French 已提交
1463

1464
	spin_unlock(&inode->i_lock);
L
Linus Torvalds 已提交
1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485
	if (!PageUptodate(page)) {
		position =  ((loff_t)page->index << PAGE_CACHE_SHIFT) + offset;
		/* can not rely on (or let) writepage write this data */
		if (to < offset) {
			cFYI(1, ("Illegal offsets, can not copy from %d to %d",
				offset, to));
			FreeXid(xid);
			return rc;
		}
		/* this is probably better than directly calling
		   partialpage_write since in this function the file handle is
		   known which we might as well	leverage */
		/* BB check if anything else missing out of ppw
		   such as updating last write time */
		page_data = kmap(page);
		rc = cifs_write(file, page_data + offset, to-offset,
				&position);
		if (rc > 0)
			rc = 0;
		/* else if (rc < 0) should we set writebehind rc? */
		kunmap(page);
S
Steve French 已提交
1486
	} else {
L
Linus Torvalds 已提交
1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497
		set_page_dirty(page);
	}

	FreeXid(xid);
	return rc;
}

int cifs_fsync(struct file *file, struct dentry *dentry, int datasync)
{
	int xid;
	int rc = 0;
1498
	struct inode *inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
1499 1500 1501

	xid = GetXid();

S
Steve French 已提交
1502
	cFYI(1, ("Sync file - name: %s datasync: 0x%x",
L
Linus Torvalds 已提交
1503
		dentry->d_name.name, datasync));
1504

1505 1506 1507
	rc = filemap_write_and_wait(inode->i_mapping);
	if (rc == 0) {
		rc = CIFS_I(inode)->write_behind_rc;
L
Linus Torvalds 已提交
1508
		CIFS_I(inode)->write_behind_rc = 0;
1509
	}
L
Linus Torvalds 已提交
1510 1511 1512 1513
	FreeXid(xid);
	return rc;
}

1514
/* static void cifs_sync_page(struct page *page)
L
Linus Torvalds 已提交
1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527
{
	struct address_space *mapping;
	struct inode *inode;
	unsigned long index = page->index;
	unsigned int rpages = 0;
	int rc = 0;

	cFYI(1, ("sync page %p",page));
	mapping = page->mapping;
	if (!mapping)
		return 0;
	inode = mapping->host;
	if (!inode)
1528
		return; */
L
Linus Torvalds 已提交
1529

S
Steve French 已提交
1530
/*	fill in rpages then
L
Linus Torvalds 已提交
1531 1532
	result = cifs_pagein_inode(inode, index, rpages); */ /* BB finish */

1533
/*	cFYI(1, ("rpages is %d for sync page of Index %ld", rpages, index));
L
Linus Torvalds 已提交
1534

1535
#if 0
L
Linus Torvalds 已提交
1536 1537 1538
	if (rc < 0)
		return rc;
	return 0;
1539
#endif
L
Linus Torvalds 已提交
1540 1541 1542 1543 1544 1545
} */

/*
 * As file closes, flush all cached write data for this inode checking
 * for write behind errors.
 */
1546
int cifs_flush(struct file *file, fl_owner_t id)
L
Linus Torvalds 已提交
1547
{
S
Steve French 已提交
1548
	struct inode *inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560
	int rc = 0;

	/* Rather than do the steps manually:
	   lock the inode for writing
	   loop through pages looking for write behind data (dirty pages)
	   coalesce into contiguous 16K (or smaller) chunks to write to server
	   send to server (prefer in parallel)
	   deal with writebehind errors
	   unlock inode for writing
	   filemapfdatawrite appears easier for the time being */

	rc = filemap_fdatawrite(inode->i_mapping);
1561 1562 1563
	/* reset wb rc if we were able to write out dirty pages */
	if (!rc) {
		rc = CIFS_I(inode)->write_behind_rc;
L
Linus Torvalds 已提交
1564
		CIFS_I(inode)->write_behind_rc = 0;
1565
	}
1566

S
Steve French 已提交
1567
	cFYI(1, ("Flush inode %p file %p rc %d", inode, file, rc));
L
Linus Torvalds 已提交
1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587

	return rc;
}

ssize_t cifs_user_read(struct file *file, char __user *read_data,
	size_t read_size, loff_t *poffset)
{
	int rc = -EACCES;
	unsigned int bytes_read = 0;
	unsigned int total_read = 0;
	unsigned int current_read_size;
	struct cifs_sb_info *cifs_sb;
	struct cifsTconInfo *pTcon;
	int xid;
	struct cifsFileInfo *open_file;
	char *smb_read_data;
	char __user *current_offset;
	struct smb_com_read_rsp *pSMBr;

	xid = GetXid();
1588
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
1589 1590 1591 1592 1593 1594 1595 1596
	pTcon = cifs_sb->tcon;

	if (file->private_data == NULL) {
		FreeXid(xid);
		return -EBADF;
	}
	open_file = (struct cifsFileInfo *)file->private_data;

S
Steve French 已提交
1597
	if ((file->f_flags & O_ACCMODE) == O_WRONLY)
L
Linus Torvalds 已提交
1598
		cFYI(1, ("attempting read on write only file instance"));
S
Steve French 已提交
1599

L
Linus Torvalds 已提交
1600 1601 1602
	for (total_read = 0, current_offset = read_data;
	     read_size > total_read;
	     total_read += bytes_read, current_offset += bytes_read) {
S
Steve French 已提交
1603
		current_read_size = min_t(const int, read_size - total_read,
L
Linus Torvalds 已提交
1604 1605 1606 1607
					  cifs_sb->rsize);
		rc = -EAGAIN;
		smb_read_data = NULL;
		while (rc == -EAGAIN) {
1608
			int buf_type = CIFS_NO_BUFFER;
S
Steve French 已提交
1609
			if ((open_file->invalidHandle) &&
L
Linus Torvalds 已提交
1610
			    (!open_file->closePend)) {
1611
				rc = cifs_reopen_file(file, TRUE);
L
Linus Torvalds 已提交
1612 1613 1614
				if (rc != 0)
					break;
			}
1615
			rc = CIFSSMBRead(xid, pTcon,
1616 1617 1618 1619
					 open_file->netfid,
					 current_read_size, *poffset,
					 &bytes_read, &smb_read_data,
					 &buf_type);
L
Linus Torvalds 已提交
1620 1621
			pSMBr = (struct smb_com_read_rsp *)smb_read_data;
			if (smb_read_data) {
1622 1623 1624 1625
				if (copy_to_user(current_offset,
						smb_read_data +
						4 /* RFC1001 length field */ +
						le16_to_cpu(pSMBr->DataOffset),
S
Steve French 已提交
1626
						bytes_read))
1627 1628
					rc = -EFAULT;

S
Steve French 已提交
1629
				if (buf_type == CIFS_SMALL_BUFFER)
1630
					cifs_small_buf_release(smb_read_data);
S
Steve French 已提交
1631
				else if (buf_type == CIFS_LARGE_BUFFER)
1632
					cifs_buf_release(smb_read_data);
L
Linus Torvalds 已提交
1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643
				smb_read_data = NULL;
			}
		}
		if (rc || (bytes_read == 0)) {
			if (total_read) {
				break;
			} else {
				FreeXid(xid);
				return rc;
			}
		} else {
1644
			cifs_stats_bytes_read(pTcon, bytes_read);
L
Linus Torvalds 已提交
1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664
			*poffset += bytes_read;
		}
	}
	FreeXid(xid);
	return total_read;
}


static ssize_t cifs_read(struct file *file, char *read_data, size_t read_size,
	loff_t *poffset)
{
	int rc = -EACCES;
	unsigned int bytes_read = 0;
	unsigned int total_read;
	unsigned int current_read_size;
	struct cifs_sb_info *cifs_sb;
	struct cifsTconInfo *pTcon;
	int xid;
	char *current_offset;
	struct cifsFileInfo *open_file;
1665
	int buf_type = CIFS_NO_BUFFER;
L
Linus Torvalds 已提交
1666 1667

	xid = GetXid();
1668
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679
	pTcon = cifs_sb->tcon;

	if (file->private_data == NULL) {
		FreeXid(xid);
		return -EBADF;
	}
	open_file = (struct cifsFileInfo *)file->private_data;

	if ((file->f_flags & O_ACCMODE) == O_WRONLY)
		cFYI(1, ("attempting read on write only file instance"));

S
Steve French 已提交
1680
	for (total_read = 0, current_offset = read_data;
L
Linus Torvalds 已提交
1681 1682 1683 1684
	     read_size > total_read;
	     total_read += bytes_read, current_offset += bytes_read) {
		current_read_size = min_t(const int, read_size - total_read,
					  cifs_sb->rsize);
1685 1686
		/* For windows me and 9x we do not want to request more
		than it negotiated since it will refuse the read then */
S
Steve French 已提交
1687
		if ((pTcon->ses) &&
1688 1689 1690 1691
			!(pTcon->ses->capabilities & CAP_LARGE_FILES)) {
			current_read_size = min_t(const int, current_read_size,
					pTcon->ses->server->maxBuf - 128);
		}
L
Linus Torvalds 已提交
1692 1693
		rc = -EAGAIN;
		while (rc == -EAGAIN) {
S
Steve French 已提交
1694
			if ((open_file->invalidHandle) &&
L
Linus Torvalds 已提交
1695
			    (!open_file->closePend)) {
1696
				rc = cifs_reopen_file(file, TRUE);
L
Linus Torvalds 已提交
1697 1698 1699
				if (rc != 0)
					break;
			}
1700
			rc = CIFSSMBRead(xid, pTcon,
1701 1702 1703 1704
					 open_file->netfid,
					 current_read_size, *poffset,
					 &bytes_read, &current_offset,
					 &buf_type);
L
Linus Torvalds 已提交
1705 1706 1707 1708 1709 1710 1711 1712 1713
		}
		if (rc || (bytes_read == 0)) {
			if (total_read) {
				break;
			} else {
				FreeXid(xid);
				return rc;
			}
		} else {
1714
			cifs_stats_bytes_read(pTcon, total_read);
L
Linus Torvalds 已提交
1715 1716 1717 1718 1719 1720 1721 1722 1723
			*poffset += bytes_read;
		}
	}
	FreeXid(xid);
	return total_read;
}

int cifs_file_mmap(struct file *file, struct vm_area_struct *vma)
{
1724
	struct dentry *dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739
	int rc, xid;

	xid = GetXid();
	rc = cifs_revalidate(dentry);
	if (rc) {
		cFYI(1, ("Validation prior to mmap failed, error=%d", rc));
		FreeXid(xid);
		return rc;
	}
	rc = generic_file_mmap(file, vma);
	FreeXid(xid);
	return rc;
}


S
Steve French 已提交
1740
static void cifs_copy_cache_pages(struct address_space *mapping,
L
Linus Torvalds 已提交
1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757
	struct list_head *pages, int bytes_read, char *data,
	struct pagevec *plru_pvec)
{
	struct page *page;
	char *target;

	while (bytes_read > 0) {
		if (list_empty(pages))
			break;

		page = list_entry(pages->prev, struct page, lru);
		list_del(&page->lru);

		if (add_to_page_cache(page, mapping, page->index,
				      GFP_KERNEL)) {
			page_cache_release(page);
			cFYI(1, ("Add page cache failed"));
1758 1759
			data += PAGE_CACHE_SIZE;
			bytes_read -= PAGE_CACHE_SIZE;
L
Linus Torvalds 已提交
1760 1761 1762
			continue;
		}

S
Steve French 已提交
1763
		target = kmap_atomic(page, KM_USER0);
L
Linus Torvalds 已提交
1764 1765 1766 1767

		if (PAGE_CACHE_SIZE > bytes_read) {
			memcpy(target, data, bytes_read);
			/* zero the tail end of this partial page */
S
Steve French 已提交
1768
			memset(target + bytes_read, 0,
L
Linus Torvalds 已提交
1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795
			       PAGE_CACHE_SIZE - bytes_read);
			bytes_read = 0;
		} else {
			memcpy(target, data, PAGE_CACHE_SIZE);
			bytes_read -= PAGE_CACHE_SIZE;
		}
		kunmap_atomic(target, KM_USER0);

		flush_dcache_page(page);
		SetPageUptodate(page);
		unlock_page(page);
		if (!pagevec_add(plru_pvec, page))
			__pagevec_lru_add(plru_pvec);
		data += PAGE_CACHE_SIZE;
	}
	return;
}

static int cifs_readpages(struct file *file, struct address_space *mapping,
	struct list_head *page_list, unsigned num_pages)
{
	int rc = -EACCES;
	int xid;
	loff_t offset;
	struct page *page;
	struct cifs_sb_info *cifs_sb;
	struct cifsTconInfo *pTcon;
S
Steve French 已提交
1796
	unsigned int bytes_read = 0;
S
Steve French 已提交
1797
	unsigned int read_size, i;
L
Linus Torvalds 已提交
1798 1799 1800 1801
	char *smb_read_data = NULL;
	struct smb_com_read_rsp *pSMBr;
	struct pagevec lru_pvec;
	struct cifsFileInfo *open_file;
1802
	int buf_type = CIFS_NO_BUFFER;
L
Linus Torvalds 已提交
1803 1804 1805 1806 1807 1808 1809

	xid = GetXid();
	if (file->private_data == NULL) {
		FreeXid(xid);
		return -EBADF;
	}
	open_file = (struct cifsFileInfo *)file->private_data;
1810
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
1811
	pTcon = cifs_sb->tcon;
1812

L
Linus Torvalds 已提交
1813
	pagevec_init(&lru_pvec, 0);
1814
		cFYI(DBG2, ("rpages: num pages %d", num_pages));
L
Linus Torvalds 已提交
1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827
	for (i = 0; i < num_pages; ) {
		unsigned contig_pages;
		struct page *tmp_page;
		unsigned long expected_index;

		if (list_empty(page_list))
			break;

		page = list_entry(page_list->prev, struct page, lru);
		offset = (loff_t)page->index << PAGE_CACHE_SHIFT;

		/* count adjacent pages that we will read into */
		contig_pages = 0;
S
Steve French 已提交
1828
		expected_index =
L
Linus Torvalds 已提交
1829
			list_entry(page_list->prev, struct page, lru)->index;
S
Steve French 已提交
1830
		list_for_each_entry_reverse(tmp_page, page_list, lru) {
L
Linus Torvalds 已提交
1831 1832 1833 1834
			if (tmp_page->index == expected_index) {
				contig_pages++;
				expected_index++;
			} else
S
Steve French 已提交
1835
				break;
L
Linus Torvalds 已提交
1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846
		}
		if (contig_pages + i >  num_pages)
			contig_pages = num_pages - i;

		/* for reads over a certain size could initiate async
		   read ahead */

		read_size = contig_pages * PAGE_CACHE_SIZE;
		/* Read size needs to be in multiples of one page */
		read_size = min_t(const unsigned int, read_size,
				  cifs_sb->rsize & PAGE_CACHE_MASK);
1847
		cFYI(DBG2, ("rpages: read size 0x%x  contiguous pages %d",
1848
				read_size, contig_pages));
L
Linus Torvalds 已提交
1849 1850
		rc = -EAGAIN;
		while (rc == -EAGAIN) {
S
Steve French 已提交
1851
			if ((open_file->invalidHandle) &&
L
Linus Torvalds 已提交
1852
			    (!open_file->closePend)) {
1853
				rc = cifs_reopen_file(file, TRUE);
L
Linus Torvalds 已提交
1854 1855 1856 1857
				if (rc != 0)
					break;
			}

1858
			rc = CIFSSMBRead(xid, pTcon,
1859 1860 1861 1862
					 open_file->netfid,
					 read_size, offset,
					 &bytes_read, &smb_read_data,
					 &buf_type);
1863
			/* BB more RC checks ? */
S
Steve French 已提交
1864
			if (rc == -EAGAIN) {
L
Linus Torvalds 已提交
1865
				if (smb_read_data) {
S
Steve French 已提交
1866
					if (buf_type == CIFS_SMALL_BUFFER)
1867
						cifs_small_buf_release(smb_read_data);
S
Steve French 已提交
1868
					else if (buf_type == CIFS_LARGE_BUFFER)
1869
						cifs_buf_release(smb_read_data);
L
Linus Torvalds 已提交
1870 1871 1872 1873 1874 1875 1876 1877
					smb_read_data = NULL;
				}
			}
		}
		if ((rc < 0) || (smb_read_data == NULL)) {
			cFYI(1, ("Read error in readpages: %d", rc));
			break;
		} else if (bytes_read > 0) {
1878
			task_io_account_read(bytes_read);
L
Linus Torvalds 已提交
1879 1880 1881 1882 1883 1884
			pSMBr = (struct smb_com_read_rsp *)smb_read_data;
			cifs_copy_cache_pages(mapping, page_list, bytes_read,
				smb_read_data + 4 /* RFC1001 hdr */ +
				le16_to_cpu(pSMBr->DataOffset), &lru_pvec);

			i +=  bytes_read >> PAGE_CACHE_SHIFT;
1885
			cifs_stats_bytes_read(pTcon, bytes_read);
S
Steve French 已提交
1886
			if ((bytes_read & PAGE_CACHE_MASK) != bytes_read) {
L
Linus Torvalds 已提交
1887 1888
				i++; /* account for partial page */

S
Steve French 已提交
1889
				/* server copy of file can have smaller size
L
Linus Torvalds 已提交
1890
				   than client */
S
Steve French 已提交
1891 1892
				/* BB do we need to verify this common case ?
				   this case is ok - if we are at server EOF
L
Linus Torvalds 已提交
1893 1894
				   we will hit it on next read */

O
OGAWA Hirofumi 已提交
1895
				/* break; */
L
Linus Torvalds 已提交
1896 1897 1898 1899 1900
			}
		} else {
			cFYI(1, ("No bytes read (%d) at offset %lld . "
				 "Cleaning remaining pages from readahead list",
				 bytes_read, offset));
S
Steve French 已提交
1901
			/* BB turn off caching and do new lookup on
L
Linus Torvalds 已提交
1902 1903 1904 1905
			   file size at server? */
			break;
		}
		if (smb_read_data) {
S
Steve French 已提交
1906
			if (buf_type == CIFS_SMALL_BUFFER)
1907
				cifs_small_buf_release(smb_read_data);
S
Steve French 已提交
1908
			else if (buf_type == CIFS_LARGE_BUFFER)
1909
				cifs_buf_release(smb_read_data);
L
Linus Torvalds 已提交
1910 1911 1912 1913 1914 1915 1916 1917 1918
			smb_read_data = NULL;
		}
		bytes_read = 0;
	}

	pagevec_lru_add(&lru_pvec);

/* need to free smb_read_data buf before exit */
	if (smb_read_data) {
S
Steve French 已提交
1919
		if (buf_type == CIFS_SMALL_BUFFER)
1920
			cifs_small_buf_release(smb_read_data);
S
Steve French 已提交
1921
		else if (buf_type == CIFS_LARGE_BUFFER)
1922
			cifs_buf_release(smb_read_data);
L
Linus Torvalds 已提交
1923
		smb_read_data = NULL;
S
Steve French 已提交
1924
	}
L
Linus Torvalds 已提交
1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938

	FreeXid(xid);
	return rc;
}

static int cifs_readpage_worker(struct file *file, struct page *page,
	loff_t *poffset)
{
	char *read_data;
	int rc;

	page_cache_get(page);
	read_data = kmap(page);
	/* for reads over a certain size could initiate async read ahead */
S
Steve French 已提交
1939

L
Linus Torvalds 已提交
1940
	rc = cifs_read(file, read_data, PAGE_CACHE_SIZE, poffset);
S
Steve French 已提交
1941

L
Linus Torvalds 已提交
1942 1943 1944
	if (rc < 0)
		goto io_error;
	else
S
Steve French 已提交
1945 1946
		cFYI(1, ("Bytes read %d", rc));

1947 1948
	file->f_path.dentry->d_inode->i_atime =
		current_fs_time(file->f_path.dentry->d_inode->i_sb);
S
Steve French 已提交
1949

L
Linus Torvalds 已提交
1950 1951 1952 1953 1954 1955
	if (PAGE_CACHE_SIZE > rc)
		memset(read_data + rc, 0, PAGE_CACHE_SIZE - rc);

	flush_dcache_page(page);
	SetPageUptodate(page);
	rc = 0;
S
Steve French 已提交
1956

L
Linus Torvalds 已提交
1957
io_error:
S
Steve French 已提交
1958
	kunmap(page);
L
Linus Torvalds 已提交
1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975
	page_cache_release(page);
	return rc;
}

static int cifs_readpage(struct file *file, struct page *page)
{
	loff_t offset = (loff_t)page->index << PAGE_CACHE_SHIFT;
	int rc = -EACCES;
	int xid;

	xid = GetXid();

	if (file->private_data == NULL) {
		FreeXid(xid);
		return -EBADF;
	}

S
Steve French 已提交
1976
	cFYI(1, ("readpage %p at offset %d 0x%x\n",
L
Linus Torvalds 已提交
1977 1978 1979 1980 1981 1982 1983 1984 1985 1986
		 page, (int)offset, (int)offset));

	rc = cifs_readpage_worker(file, page, &offset);

	unlock_page(page);

	FreeXid(xid);
	return rc;
}

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
static int is_inode_writable(struct cifsInodeInfo *cifs_inode)
{
	struct cifsFileInfo *open_file;

	read_lock(&GlobalSMBSeslock);
	list_for_each_entry(open_file, &cifs_inode->openFileList, flist) {
		if (open_file->closePend)
			continue;
		if (open_file->pfile &&
		    ((open_file->pfile->f_flags & O_RDWR) ||
		     (open_file->pfile->f_flags & O_WRONLY))) {
			read_unlock(&GlobalSMBSeslock);
			return 1;
		}
	}
	read_unlock(&GlobalSMBSeslock);
	return 0;
}

L
Linus Torvalds 已提交
2006 2007 2008
/* We do not want to update the file size from server for inodes
   open for write - to avoid races with writepage extending
   the file - in the future we could consider allowing
S
Steve French 已提交
2009
   refreshing the inode only on increases in the file size
L
Linus Torvalds 已提交
2010 2011
   but this is tricky to do without racing with writebehind
   page caching in the current Linux kernel design */
2012
int is_size_safe_to_change(struct cifsInodeInfo *cifsInode, __u64 end_of_file)
L
Linus Torvalds 已提交
2013
{
2014 2015
	if (!cifsInode)
		return 1;
2016

2017 2018
	if (is_inode_writable(cifsInode)) {
		/* This inode is open for write at least once */
2019 2020 2021
		struct cifs_sb_info *cifs_sb;

		cifs_sb = CIFS_SB(cifsInode->vfs_inode.i_sb);
S
Steve French 已提交
2022
		if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_DIRECT_IO) {
S
Steve French 已提交
2023
			/* since no page cache to corrupt on directio
2024 2025 2026 2027
			we can change size safely */
			return 1;
		}

S
Steve French 已提交
2028
		if (i_size_read(&cifsInode->vfs_inode) < end_of_file)
2029 2030
			return 1;

2031
		return 0;
2032
	} else
2033
		return 1;
L
Linus Torvalds 已提交
2034 2035 2036 2037 2038 2039
}

static int cifs_prepare_write(struct file *file, struct page *page,
	unsigned from, unsigned to)
{
	int rc = 0;
2040 2041 2042
	loff_t i_size;
	loff_t offset;

S
Steve French 已提交
2043
	cFYI(1, ("prepare write for page %p from %d to %d", page, from, to));
2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062
	if (PageUptodate(page))
		return 0;

	/* If we are writing a full page it will be up to date,
	   no need to read from the server */
	if ((to == PAGE_CACHE_SIZE) && (from == 0)) {
		SetPageUptodate(page);
		return 0;
	}

	offset = (loff_t)page->index << PAGE_CACHE_SHIFT;
	i_size = i_size_read(page->mapping->host);

	if ((offset >= i_size) ||
	    ((from == 0) && (offset + to) >= i_size)) {
		/*
		 * We don't need to read data beyond the end of the file.
		 * zero it, and set the page uptodate
		 */
2063
		simple_prepare_write(file, page, from, to);
2064 2065
		SetPageUptodate(page);
	} else if ((file->f_flags & O_ACCMODE) != O_WRONLY) {
L
Linus Torvalds 已提交
2066
		/* might as well read a page, it is fast enough */
2067 2068 2069 2070 2071 2072
		rc = cifs_readpage_worker(file, page, &offset);
	} else {
		/* we could try using another file handle if there is one -
		   but how would we lock it to prevent close of that handle
		   racing with this read? In any case
		   this will be written out by commit_write so is fine */
L
Linus Torvalds 已提交
2073 2074
	}

S
Steve French 已提交
2075 2076
	/* we do not need to pass errors back
	   e.g. if we do not have read access to the file
2077 2078
	   because cifs_commit_write will do the right thing.  -- shaggy */

L
Linus Torvalds 已提交
2079 2080 2081
	return 0;
}

2082
const struct address_space_operations cifs_addr_ops = {
L
Linus Torvalds 已提交
2083 2084 2085
	.readpage = cifs_readpage,
	.readpages = cifs_readpages,
	.writepage = cifs_writepage,
2086
	.writepages = cifs_writepages,
L
Linus Torvalds 已提交
2087 2088 2089 2090 2091 2092
	.prepare_write = cifs_prepare_write,
	.commit_write = cifs_commit_write,
	.set_page_dirty = __set_page_dirty_nobuffers,
	/* .sync_page = cifs_sync_page, */
	/* .direct_IO = */
};
D
Dave Kleikamp 已提交
2093 2094 2095 2096 2097 2098

/*
 * cifs_readpages requires the server to support a buffer large enough to
 * contain the header plus one complete page of data.  Otherwise, we need
 * to leave cifs_readpages out of the address space operations.
 */
2099
const struct address_space_operations cifs_addr_ops_smallbuf = {
D
Dave Kleikamp 已提交
2100 2101 2102 2103 2104 2105 2106 2107 2108
	.readpage = cifs_readpage,
	.writepage = cifs_writepage,
	.writepages = cifs_writepages,
	.prepare_write = cifs_prepare_write,
	.commit_write = cifs_commit_write,
	.set_page_dirty = __set_page_dirty_nobuffers,
	/* .sync_page = cifs_sync_page, */
	/* .direct_IO = */
};