file.c 55.7 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
			filemap_write_and_wait(file->f_path.dentry->d_inode->i_mapping);
L
Linus Torvalds 已提交
134 135 136
		}
		cFYI(1, ("invalidating remote inode since open detected it "
			 "changed"));
137
		invalidate_remote_inode(file->f_path.dentry->d_inode);
L
Linus Torvalds 已提交
138 139 140
	}

client_can_cache:
141
	if (pTcon->unix_ext)
142
		rc = cifs_get_inode_info_unix(&file->f_path.dentry->d_inode,
L
Linus Torvalds 已提交
143 144
			full_path, inode->i_sb, xid);
	else
145
		rc = cifs_get_inode_info(&file->f_path.dentry->d_inode,
L
Linus Torvalds 已提交
146 147 148 149 150 151
			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",
152
			 file->f_path.dentry->d_inode));
L
Linus Torvalds 已提交
153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180
	} 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 */
181
		pCifsInode = CIFS_I(file->f_path.dentry->d_inode);
L
Linus Torvalds 已提交
182 183 184 185 186 187 188 189 190 191
		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;
192

L
Linus Torvalds 已提交
193 194 195 196 197 198 199 200 201 202 203 204
				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 "
205
					   "new file %p", file));
L
Linus Torvalds 已提交
206 207 208
		}
	}

209
	full_path = build_path_from_dentry(file->f_path.dentry);
L
Linus Torvalds 已提交
210 211 212 213 214
	if (full_path == NULL) {
		FreeXid(xid);
		return -ENOMEM;
	}

215
	cFYI(1, ("inode = 0x%p file flags are 0x%x for %s",
L
Linus Torvalds 已提交
216 217 218 219 220
		 inode, file->f_flags, full_path));
	desiredAccess = cifs_convert_flags(file->f_flags);

/*********************************************************************
 *  open flag mapping table:
S
Steve French 已提交
221
 *
L
Linus Torvalds 已提交
222
 *	POSIX Flag            CIFS Disposition
S
Steve French 已提交
223
 *	----------            ----------------
L
Linus Torvalds 已提交
224 225 226 227 228 229 230
 *	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
231
 *	FILE_SUPERSEDE (ie create whether or not file exists although
L
Linus Torvalds 已提交
232 233 234 235
 *	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
236
 *?  O_SYNC is a reasonable match to CIFS writethrough flag
L
Linus Torvalds 已提交
237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255
 *?  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
256 257
	/* 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
258 259 260 261 262 263
	   able to simply do a filemap_fdatawrite/filemap_fdatawait first */
	buf = kmalloc(sizeof(FILE_ALL_INFO), GFP_KERNEL);
	if (!buf) {
		rc = -ENOMEM;
		goto out;
	}
264 265

	if (cifs_sb->tcon->ses->capabilities & CAP_NT_SMBS)
S
Steve French 已提交
266
		rc = CIFSSMBOpen(xid, pTcon, full_path, disposition,
267
			 desiredAccess, CREATE_NOT_DIR, &netfid, &oplock, buf,
268 269
			 cifs_sb->local_nls, cifs_sb->mnt_cifs_flags
				 & CIFS_MOUNT_MAP_SPECIAL_CHR);
270 271 272
	else
		rc = -EIO; /* no NT SMB support fall into legacy open below */

273 274 275 276 277 278 279
	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
280
	if (rc) {
281
		cFYI(1, ("cifs_open returned 0x%x", rc));
L
Linus Torvalds 已提交
282 283 284 285 286 287 288 289 290 291 292 293
		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);

294
	pCifsInode = CIFS_I(file->f_path.dentry->d_inode);
L
Linus Torvalds 已提交
295 296 297 298 299 300 301 302
	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
303
	if (oplock & CIFS_CREATE_ACTION) {
L
Linus Torvalds 已提交
304 305
		/* time to set mode which we can not set earlier due to
		   problems creating new read-only files */
306
		if (pTcon->unix_ext) {
L
Linus Torvalds 已提交
307 308 309
			CIFSSMBUnixSetPerms(xid, pTcon, full_path,
					    inode->i_mode,
					    (__u64)-1, (__u64)-1, 0 /* dev */,
310
					    cifs_sb->local_nls,
S
Steve French 已提交
311
					    cifs_sb->mnt_cifs_flags &
312
						CIFS_MOUNT_MAP_SPECIAL_CHR);
L
Linus Torvalds 已提交
313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328
		} 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
329
/* Try to reacquire byte range locks that were released when session */
L
Linus Torvalds 已提交
330 331 332 333 334 335 336 337 338 339
/* 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;
}

340
static int cifs_reopen_file(struct file *file, int can_flush)
L
Linus Torvalds 已提交
341 342 343 344 345 346 347
{
	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
348
	struct inode *inode;
L
Linus Torvalds 已提交
349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366
	char *full_path = NULL;
	int desiredAccess;
	int disposition = FILE_OPEN;
	__u16 netfid;

	if (file->private_data) {
		pCifsFile = (struct cifsFileInfo *)file->private_data;
	} else
		return -EBADF;

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

367
	if (file->f_path.dentry == NULL) {
368 369 370 371 372 373 374
		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
375
	if (inode == NULL) {
376 377 378 379
		cERROR(1, ("inode not valid"));
		dump_stack();
		rc = -EBADF;
		goto reopen_error_exit;
L
Linus Torvalds 已提交
380
	}
381

L
Linus Torvalds 已提交
382 383
	cifs_sb = CIFS_SB(inode->i_sb);
	pTcon = cifs_sb->tcon;
384

L
Linus Torvalds 已提交
385 386 387 388
/* 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 */
389
	full_path = build_path_from_dentry(file->f_path.dentry);
L
Linus Torvalds 已提交
390
	if (full_path == NULL) {
391 392
		rc = -ENOMEM;
reopen_error_exit:
L
Linus Torvalds 已提交
393 394
		up(&pCifsFile->fh_sem);
		FreeXid(xid);
395
		return rc;
L
Linus Torvalds 已提交
396 397
	}

398
	cFYI(1, ("inode = 0x%p file flags 0x%x for %s",
S
Steve French 已提交
399
		 inode, file->f_flags, full_path));
L
Linus Torvalds 已提交
400 401 402 403 404 405 406 407
	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
408 409
	   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
410 411 412 413 414
	   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
415
			 cifs_sb->local_nls, cifs_sb->mnt_cifs_flags &
416
				CIFS_MOUNT_MAP_SPECIAL_CHR);
L
Linus Torvalds 已提交
417 418
	if (rc) {
		up(&pCifsFile->fh_sem);
419 420
		cFYI(1, ("cifs_open returned 0x%x", rc));
		cFYI(1, ("oplock: %d", oplock));
L
Linus Torvalds 已提交
421 422 423 424 425 426 427
	} else {
		pCifsFile->netfid = netfid;
		pCifsFile->invalidHandle = FALSE;
		up(&pCifsFile->fh_sem);
		pCifsInode = CIFS_I(inode);
		if (pCifsInode) {
			if (can_flush) {
428
				filemap_write_and_wait(inode->i_mapping);
L
Linus Torvalds 已提交
429 430 431 432
			/* temporarily disable caching while we
			   go to server to get inode info */
				pCifsInode->clientCanCacheAll = FALSE;
				pCifsInode->clientCanCacheRead = FALSE;
433
				if (pTcon->unix_ext)
L
Linus Torvalds 已提交
434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449
					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",
450
					 file->f_path.dentry->d_inode));
L
Linus Torvalds 已提交
451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469
			} 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;
470
	int xid, timeout;
L
Linus Torvalds 已提交
471 472 473 474 475 476 477 478 479 480
	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
481 482
		struct cifsLockInfo *li, *tmp;

L
Linus Torvalds 已提交
483 484 485 486 487
		pSMBFile->closePend = TRUE;
		if (pTcon) {
			/* no sense reconnecting to close a file that is
			   already closed */
			if (pTcon->tidStatus != CifsNeedReconnect) {
488
				timeout = 2;
S
Steve French 已提交
489
				while ((atomic_read(&pSMBFile->wrtPending) != 0)
490
					&& (timeout <= 2048)) {
491 492 493 494 495
					/* 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
496
					but this should give enough time to
497
					clear the socket */
498
#ifdef CONFIG_CIFS_DEBUG2
S
Steve French 已提交
499
					cFYI(1, ("close delay, write pending"));
500
#endif /* DEBUG2 */
501 502
					msleep(timeout);
					timeout *= 4;
503
				}
S
Steve French 已提交
504
				if (atomic_read(&pSMBFile->wrtPending))
505 506
					cERROR(1,
						("close with pending writes"));
L
Linus Torvalds 已提交
507 508 509 510
				rc = CIFSSMBClose(xid, pTcon,
						  pSMBFile->netfid);
			}
		}
J
[CIFS]  
Jeremy Allison 已提交
511 512 513

		/* Delete any outstanding lock records.
		   We'll lose them when the file is closed anyway. */
514
		mutex_lock(&pSMBFile->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
515 516 517 518
		list_for_each_entry_safe(li, tmp, &pSMBFile->llist, llist) {
			list_del(&li->llist);
			kfree(li);
		}
519
		mutex_unlock(&pSMBFile->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
520

521
		write_lock(&GlobalSMBSeslock);
L
Linus Torvalds 已提交
522 523
		list_del(&pSMBFile->flist);
		list_del(&pSMBFile->tlist);
524
		write_unlock(&GlobalSMBSeslock);
525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541
		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
542 543 544 545 546 547
		kfree(pSMBFile->search_resume_name);
		kfree(file->private_data);
		file->private_data = NULL;
	} else
		rc = -EBADF;

548
	read_lock(&GlobalSMBSeslock);
L
Linus Torvalds 已提交
549 550 551 552 553 554 555
	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;
	}
556
	read_unlock(&GlobalSMBSeslock);
S
Steve French 已提交
557
	if ((rc == 0) && CIFS_I(inode)->write_behind_rc)
L
Linus Torvalds 已提交
558 559 560 561 562 563 564 565 566 567 568 569 570
		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;

571
	cFYI(1, ("Closedir inode = 0x%p", inode));
L
Linus Torvalds 已提交
572 573 574 575 576

	xid = GetXid();

	if (pCFileStruct) {
		struct cifsTconInfo *pTcon;
S
Steve French 已提交
577 578
		struct cifs_sb_info *cifs_sb =
			CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
579 580 581 582

		pTcon = cifs_sb->tcon;

		cFYI(1, ("Freeing private data in close dir"));
583 584
		if ((pCFileStruct->srch_inf.endOfSearch == FALSE) &&
		   (pCFileStruct->invalidHandle == FALSE)) {
L
Linus Torvalds 已提交
585 586 587 588 589 590 591 592 593
			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) {
594
			cFYI(1, ("closedir free smb buf in srch struct"));
L
Linus Torvalds 已提交
595
			pCFileStruct->srch_inf.ntwrk_buf_start = NULL;
S
Steve French 已提交
596
			if (pCFileStruct->srch_inf.smallBuf)
597 598 599
				cifs_small_buf_release(ptmp);
			else
				cifs_buf_release(ptmp);
L
Linus Torvalds 已提交
600 601 602
		}
		ptmp = pCFileStruct->search_resume_name;
		if (ptmp) {
603
			cFYI(1, ("closedir free resume name"));
L
Linus Torvalds 已提交
604 605 606 607 608 609 610 611 612 613 614
			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
615 616 617
static int store_file_lock(struct cifsFileInfo *fid, __u64 len,
				__u64 offset, __u8 lockType)
{
S
Steve French 已提交
618 619
	struct cifsLockInfo *li =
		kmalloc(sizeof(struct cifsLockInfo), GFP_KERNEL);
J
[CIFS]  
Jeremy Allison 已提交
620 621 622 623 624
	if (li == NULL)
		return -ENOMEM;
	li->offset = offset;
	li->length = len;
	li->type = lockType;
625
	mutex_lock(&fid->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
626
	list_add(&li->llist, &fid->llist);
627
	mutex_unlock(&fid->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
628 629 630
	return 0;
}

L
Linus Torvalds 已提交
631 632 633 634 635 636 637 638 639
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;
640 641
	__u16 netfid;
	__u8 lockType = LOCKING_ANDX_LARGE_FILES;
J
[CIFS]  
Jeremy Allison 已提交
642
	int posix_locking;
L
Linus Torvalds 已提交
643 644 645 646 647 648 649

	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
650 651
		cmd, pfLock->fl_flags, pfLock->fl_type, pfLock->fl_start,
		pfLock->fl_end));
L
Linus Torvalds 已提交
652 653

	if (pfLock->fl_flags & FL_POSIX)
654
		cFYI(1, ("Posix"));
L
Linus Torvalds 已提交
655
	if (pfLock->fl_flags & FL_FLOCK)
656
		cFYI(1, ("Flock"));
L
Linus Torvalds 已提交
657
	if (pfLock->fl_flags & FL_SLEEP) {
658
		cFYI(1, ("Blocking lock"));
L
Linus Torvalds 已提交
659 660 661 662
		wait_flag = TRUE;
	}
	if (pfLock->fl_flags & FL_ACCESS)
		cFYI(1, ("Process suspended by mandatory locking - "
663
			 "not implemented yet"));
L
Linus Torvalds 已提交
664 665
	if (pfLock->fl_flags & FL_LEASE)
		cFYI(1, ("Lease on file - not implemented yet"));
S
Steve French 已提交
666
	if (pfLock->fl_flags &
L
Linus Torvalds 已提交
667 668 669 670 671 672 673
	    (~(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) {
674
		cFYI(1, ("F_UNLCK"));
L
Linus Torvalds 已提交
675
		numUnlock = 1;
676 677
		/* Check if unlock includes more than
		one lock range */
L
Linus Torvalds 已提交
678
	} else if (pfLock->fl_type == F_RDLCK) {
679
		cFYI(1, ("F_RDLCK"));
L
Linus Torvalds 已提交
680 681 682
		lockType |= LOCKING_ANDX_SHARED_LOCK;
		numLock = 1;
	} else if (pfLock->fl_type == F_EXLCK) {
683
		cFYI(1, ("F_EXLCK"));
L
Linus Torvalds 已提交
684 685
		numLock = 1;
	} else if (pfLock->fl_type == F_SHLCK) {
686
		cFYI(1, ("F_SHLCK"));
L
Linus Torvalds 已提交
687 688 689
		lockType |= LOCKING_ANDX_SHARED_LOCK;
		numLock = 1;
	} else
690
		cFYI(1, ("Unknown type of lock"));
L
Linus Torvalds 已提交
691

692
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
693 694 695 696 697 698
	pTcon = cifs_sb->tcon;

	if (file->private_data == NULL) {
		FreeXid(xid);
		return -EBADF;
	}
699 700
	netfid = ((struct cifsFileInfo *)file->private_data)->netfid;

J
[CIFS]  
Jeremy Allison 已提交
701 702
	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
703

704 705 706
	/* 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
707
	if (IS_GETLK(cmd)) {
S
Steve French 已提交
708
		if (posix_locking) {
709
			int posix_lock_type;
S
Steve French 已提交
710
			if (lockType & LOCKING_ANDX_SHARED_LOCK)
711 712 713 714
				posix_lock_type = CIFS_RDLCK;
			else
				posix_lock_type = CIFS_WRLCK;
			rc = CIFSSMBPosixLock(xid, pTcon, netfid, 1 /* get */,
715
					length,	pfLock,
716 717 718 719 720 721 722 723
					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
724
		if (rc == 0) {
S
Steve French 已提交
725
			rc = CIFSSMBLock(xid, pTcon, netfid, length,
L
Linus Torvalds 已提交
726 727 728 729 730 731
					 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 "
732
					   "range %d during test of lock", rc));
L
Linus Torvalds 已提交
733 734 735 736 737 738 739 740 741 742 743
			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
744 745 746 747 748 749 750 751 752

	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) {
753
		int posix_lock_type;
S
Steve French 已提交
754
		if (lockType & LOCKING_ANDX_SHARED_LOCK)
755 756 757
			posix_lock_type = CIFS_RDLCK;
		else
			posix_lock_type = CIFS_WRLCK;
758

S
Steve French 已提交
759
		if (numUnlock == 1)
760
			posix_lock_type = CIFS_UNLCK;
J
[CIFS]  
Jeremy Allison 已提交
761

762
		rc = CIFSSMBPosixLock(xid, pTcon, netfid, 0 /* set */,
763
				      length, pfLock,
764
				      posix_lock_type, wait_flag);
J
[CIFS]  
Jeremy Allison 已提交
765
	} else {
S
Steve French 已提交
766 767
		struct cifsFileInfo *fid =
			(struct cifsFileInfo *)file->private_data;
J
[CIFS]  
Jeremy Allison 已提交
768 769

		if (numLock) {
S
Steve French 已提交
770 771
			rc = CIFSSMBLock(xid, pTcon, netfid, length,
					pfLock->fl_start,
J
[CIFS]  
Jeremy Allison 已提交
772 773 774 775 776 777 778 779 780 781 782 783 784
					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;

785
			rc = 0;
786
			mutex_lock(&fid->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
787 788
			list_for_each_entry_safe(li, tmp, &fid->llist, llist) {
				if (pfLock->fl_start <= li->offset &&
S
Steve French 已提交
789
						(pfLock->fl_start + length) >=
790
						(li->offset + li->length)) {
S
Steve French 已提交
791 792
					stored_rc = CIFSSMBLock(xid, pTcon,
							netfid,
J
[CIFS]  
Jeremy Allison 已提交
793 794 795 796 797 798 799 800 801
							li->length, li->offset,
							1, 0, li->type, FALSE);
					if (stored_rc)
						rc = stored_rc;

					list_del(&li->llist);
					kfree(li);
				}
			}
802
			mutex_unlock(&fid->lock_mutex);
J
[CIFS]  
Jeremy Allison 已提交
803 804 805
		}
	}

806
	if (pfLock->fl_flags & FL_POSIX)
L
Linus Torvalds 已提交
807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822
		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;

823
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
824 825 826 827 828

	pTcon = cifs_sb->tcon;

	/* cFYI(1,
	   (" write %d bytes to offset %lld of %s", write_size,
829
	   *poffset, file->f_path.dentry->d_name.name)); */
L
Linus Torvalds 已提交
830 831 832

	if (file->private_data == NULL)
		return -EBADF;
833
	open_file = (struct cifsFileInfo *) file->private_data;
834

L
Linus Torvalds 已提交
835 836
	xid = GetXid();

837
	if (*poffset > file->f_path.dentry->d_inode->i_size)
L
Linus Torvalds 已提交
838 839 840 841 842 843 844 845 846 847 848 849 850 851 852
		long_op = 2; /* writes past end of file can take a long time */
	else
		long_op = 1;

	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
853
			}
L
Linus Torvalds 已提交
854 855 856 857 858 859 860 861 862 863 864 865
			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 */
866
				rc = cifs_reopen_file(file, FALSE);
L
Linus Torvalds 已提交
867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890
				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;
		long_op = FALSE; /* subsequent writes fast -
				    15 seconds is plenty */
	}

891
	cifs_stats_bytes_written(pTcon, total_written);
L
Linus Torvalds 已提交
892 893

	/* since the write may have blocked check these pointers again */
894 895
	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
896 897
/* Do not update local mtime - server will set its actual value on write
 *		inode->i_ctime = inode->i_mtime =
898 899 900 901 902
 * 			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
903
					*poffset);
904
			spin_unlock(&inode->i_lock);
L
Linus Torvalds 已提交
905
		}
S
Steve French 已提交
906
		mark_inode_dirty_sync(file->f_path.dentry->d_inode);
L
Linus Torvalds 已提交
907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922
	}
	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;

923
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
924 925 926

	pTcon = cifs_sb->tcon;

S
Steve French 已提交
927
	cFYI(1, ("write %zd bytes to offset %lld of %s", write_size,
928
	   *poffset, file->f_path.dentry->d_name.name));
L
Linus Torvalds 已提交
929 930 931

	if (file->private_data == NULL)
		return -EBADF;
932
	open_file = (struct cifsFileInfo *)file->private_data;
933

L
Linus Torvalds 已提交
934 935
	xid = GetXid();

936
	if (*poffset > file->f_path.dentry->d_inode->i_size)
L
Linus Torvalds 已提交
937 938 939 940 941 942 943 944 945 946 947 948 949
		long_op = 2; /* writes past end of file can take a long time */
	else
		long_op = 1;

	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
950
			   while we blocked so return what we managed to
L
Linus Torvalds 已提交
951 952
			   write */
				return total_written;
S
Steve French 已提交
953
			}
L
Linus Torvalds 已提交
954 955 956 957 958 959 960 961 962 963
			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
964
				   reopen_file not to flush data to
L
Linus Torvalds 已提交
965
				   server now */
966
				rc = cifs_reopen_file(file, FALSE);
L
Linus Torvalds 已提交
967 968 969
				if (rc != 0)
					break;
			}
S
Steve French 已提交
970 971
			if (experimEnabled || (pTcon->ses->server &&
				((pTcon->ses->server->secMode &
S
Steve French 已提交
972
				(SECMODE_SIGN_REQUIRED | SECMODE_SIGN_ENABLED))
S
Steve French 已提交
973
				== 0))) {
974 975 976
				struct kvec iov[2];
				unsigned int len;

977
				len = min((size_t)cifs_sb->wsize,
978 979 980 981 982
					  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;
983
				rc = CIFSSMBWrite2(xid, pTcon,
984
						open_file->netfid, len,
985
						*poffset, &bytes_written,
986
						iov, 1, long_op);
987
			} else
988 989 990 991 992 993 994
				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
995 996 997 998 999 1000 1001 1002 1003 1004
		}
		if (rc || (bytes_written == 0)) {
			if (total_written)
				break;
			else {
				FreeXid(xid);
				return rc;
			}
		} else
			*poffset += bytes_written;
S
Steve French 已提交
1005
		long_op = FALSE; /* subsequent writes fast -
L
Linus Torvalds 已提交
1006 1007 1008
				    15 seconds is plenty */
	}

1009
	cifs_stats_bytes_written(pTcon, total_written);
L
Linus Torvalds 已提交
1010 1011

	/* since the write may have blocked check these pointers again */
1012
	if ((file->f_path.dentry) && (file->f_path.dentry->d_inode)) {
1013
/*BB We could make this contingent on superblock ATIME flag too */
1014 1015 1016 1017 1018 1019 1020 1021
/*		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
1022
		}
1023
		mark_inode_dirty_sync(file->f_path.dentry->d_inode);
L
Linus Torvalds 已提交
1024 1025 1026 1027 1028
	}
	FreeXid(xid);
	return total_written;
}

1029
struct cifsFileInfo *find_writable_file(struct cifsInodeInfo *cifs_inode)
1030 1031
{
	struct cifsFileInfo *open_file;
1032
	int rc;
1033

1034 1035 1036 1037
	/* 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
1038 1039
	if (cifs_inode == NULL) {
		cERROR(1, ("Null inode passed to cifs_writeable_file"));
1040 1041 1042 1043
		dump_stack();
		return NULL;
	}

1044
	read_lock(&GlobalSMBSeslock);
1045
refind_writable:
1046 1047 1048 1049 1050 1051
	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))) {
1052
			atomic_inc(&open_file->wrtPending);
1053 1054 1055 1056 1057 1058 1059

			if (!open_file->invalidHandle) {
				/* found a good writable file */
				read_unlock(&GlobalSMBSeslock);
				return open_file;
			}
	
1060
			read_unlock(&GlobalSMBSeslock);
1061 1062 1063 1064 1065 1066 1067
			/* Had to unlock since following call can block */
			rc = cifs_reopen_file(open_file->pfile, FALSE);
			if (!rc) { 
				if (!open_file->closePend)
					return open_file;
				else { /* start over in case this was deleted */
				       /* since the list could be modified */
1068
					read_lock(&GlobalSMBSeslock);
1069
					atomic_dec(&open_file->wrtPending);
1070
					goto refind_writable;
1071 1072
				}
			}
1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095

			/* 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);
			
			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. */
1096 1097 1098 1099 1100 1101
		}
	}
	read_unlock(&GlobalSMBSeslock);
	return NULL;
}

L
Linus Torvalds 已提交
1102 1103 1104 1105 1106 1107 1108 1109 1110 1111
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;
1112
	struct cifsFileInfo *open_file;
L
Linus Torvalds 已提交
1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137

	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
1138
		to = (unsigned)(mapping->host->i_size - offset);
L
Linus Torvalds 已提交
1139

1140 1141 1142 1143
	open_file = find_writable_file(CIFS_I(mapping->host));
	if (open_file) {
		bytes_written = cifs_write(open_file->pfile, write_data,
					   to-from, &offset);
1144
		atomic_dec(&open_file->wrtPending);
L
Linus Torvalds 已提交
1145
		/* Does mm or vfs already set times? */
1146 1147 1148 1149 1150 1151
		inode->i_atime = inode->i_mtime = current_fs_time(inode->i_sb);
		if ((bytes_written > 0) && (offset)) {
			rc = 0;
		} else if (bytes_written < 0) {
			if (rc != -EBADF)
				rc = bytes_written;
L
Linus Torvalds 已提交
1152
		}
1153
	} else {
L
Linus Torvalds 已提交
1154 1155 1156 1157 1158 1159 1160 1161 1162
		cFYI(1, ("No writeable filehandles for inode"));
		rc = -EIO;
	}

	kunmap(page);
	return rc;
}

static int cifs_writepages(struct address_space *mapping,
1163
			   struct writeback_control *wbc)
L
Linus Torvalds 已提交
1164
{
1165 1166 1167 1168 1169
	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;
1170
	pgoff_t end;
1171
	pgoff_t index;
S
Steve French 已提交
1172 1173
	int range_whole = 0;
	struct kvec *iov;
1174
	int len;
1175 1176 1177 1178
	int n_iov = 0;
	pgoff_t next;
	int nr_pages;
	__u64 offset = 0;
1179
	struct cifsFileInfo *open_file;
1180 1181 1182 1183
	struct page *page;
	struct pagevec pvec;
	int rc = 0;
	int scanned = 0;
L
Linus Torvalds 已提交
1184 1185
	int xid;

1186
	cifs_sb = CIFS_SB(mapping->host->i_sb);
1187

1188 1189 1190 1191 1192 1193 1194
	/*
	 * 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
1195 1196 1197 1198
	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)
1199
				return generic_writepages(mapping, wbc);
1200

1201
	iov = kmalloc(32 * sizeof(struct kvec), GFP_KERNEL);
S
Steve French 已提交
1202
	if (iov == NULL)
1203 1204 1205
		return generic_writepages(mapping, wbc);


1206 1207 1208 1209 1210 1211
	/*
	 * 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;
1212
		kfree(iov);
1213 1214 1215
		return 0;
	}

L
Linus Torvalds 已提交
1216 1217
	xid = GetXid();

1218
	pagevec_init(&pvec, 0);
1219
	if (wbc->range_cyclic) {
1220
		index = mapping->writeback_index; /* Start from prev offset */
1221 1222 1223 1224 1225 1226
		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;
1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261
		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;
			}

1262
			if (!wbc->range_cyclic && page->index > end) {
1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277
				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) ||
1278
					!clear_page_dirty_for_io(page)) {
1279 1280 1281
				unlock_page(page);
				break;
			}
1282

1283 1284 1285 1286 1287 1288
			/*
			 * This actually clears the dirty bit in the radix tree.
			 * See cifs_writepage() for more commentary.
			 */
			set_page_writeback(page);

1289 1290 1291
			if (page_offset(page) >= mapping->host->i_size) {
				done = 1;
				unlock_page(page);
1292
				end_page_writeback(page);
1293 1294 1295
				break;
			}

1296 1297 1298 1299 1300
			/*
			 * BB can we get rid of this?  pages are held by pvec
			 */
			page_cache_get(page);

1301 1302 1303
			len = min(mapping->host->i_size - page_offset(page),
				  (loff_t)PAGE_CACHE_SIZE);

1304 1305 1306
			/* reserve iov[0] for the smb header */
			n_iov++;
			iov[n_iov].iov_base = kmap(page);
1307 1308
			iov[n_iov].iov_len = len;
			bytes_to_write += len;
1309 1310 1311 1312 1313 1314 1315 1316 1317 1318

			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) {
1319 1320 1321 1322 1323 1324 1325 1326
			/* 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
1327
			} else {
1328 1329 1330 1331 1332 1333 1334
				rc = CIFSSMBWrite2(xid, cifs_sb->tcon,
						   open_file->netfid,
						   bytes_to_write, offset,
						   &bytes_written, iov, n_iov,
						   1);
				atomic_dec(&open_file->wrtPending);
				if (rc || bytes_written < bytes_to_write) {
1335
					cERROR(1, ("Write2 ret %d, wrote %d",
1336 1337 1338 1339 1340 1341 1342 1343
						  rc, bytes_written));
					/* BB what if continued retry is
					   requested via mount flags? */
					set_bit(AS_EIO, &mapping->flags);
				} else {
					cifs_stats_bytes_written(cifs_sb->tcon,
								 bytes_written);
				}
1344 1345 1346
			}
			for (i = 0; i < n_iov; i++) {
				page = pvec.pages[first + i];
1347 1348 1349 1350
				/* 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
1351 1352
				when page marked as error */
				if (rc)
1353
					SetPageError(page);
1354 1355
				kunmap(page);
				unlock_page(page);
1356
				end_page_writeback(page);
1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
				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;
	}
1374
	if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0))
1375 1376
		mapping->writeback_index = index;

L
Linus Torvalds 已提交
1377
	FreeXid(xid);
1378
	kfree(iov);
L
Linus Torvalds 已提交
1379 1380 1381
	return rc;
}

S
Steve French 已提交
1382
static int cifs_writepage(struct page *page, struct writeback_control *wbc)
L
Linus Torvalds 已提交
1383 1384 1385 1386 1387 1388 1389
{
	int rc = -EFAULT;
	int xid;

	xid = GetXid();
/* BB add check for wbc flags */
	page_cache_get(page);
S
Steve French 已提交
1390
	if (!PageUptodate(page)) {
L
Linus Torvalds 已提交
1391 1392
		cFYI(1, ("ppw - page not up to date"));
	}
1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403

	/*
	 * 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
1404
	set_page_writeback(page);
L
Linus Torvalds 已提交
1405 1406 1407
	rc = cifs_partialpagewrite(page, 0, PAGE_CACHE_SIZE);
	SetPageUptodate(page); /* BB add check for error and Clearuptodate? */
	unlock_page(page);
1408 1409
	end_page_writeback(page);
	page_cache_release(page);
L
Linus Torvalds 已提交
1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423
	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
1424
	cFYI(1, ("commit write for page %p up to position %lld for %d",
L
Linus Torvalds 已提交
1425
		 page, position, to));
1426
	spin_lock(&inode->i_lock);
L
Linus Torvalds 已提交
1427 1428 1429
	if (position > inode->i_size) {
		i_size_write(inode, position);
	}
1430
	spin_unlock(&inode->i_lock);
L
Linus Torvalds 已提交
1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451
	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
1452
	} else {
L
Linus Torvalds 已提交
1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463
		set_page_dirty(page);
	}

	FreeXid(xid);
	return rc;
}

int cifs_fsync(struct file *file, struct dentry *dentry, int datasync)
{
	int xid;
	int rc = 0;
1464
	struct inode *inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
1465 1466 1467

	xid = GetXid();

S
Steve French 已提交
1468
	cFYI(1, ("Sync file - name: %s datasync: 0x%x",
L
Linus Torvalds 已提交
1469
		dentry->d_name.name, datasync));
1470

L
Linus Torvalds 已提交
1471 1472 1473 1474 1475 1476 1477
	rc = filemap_fdatawrite(inode->i_mapping);
	if (rc == 0)
		CIFS_I(inode)->write_behind_rc = 0;
	FreeXid(xid);
	return rc;
}

1478
/* static void cifs_sync_page(struct page *page)
L
Linus Torvalds 已提交
1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491
{
	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)
1492
		return; */
L
Linus Torvalds 已提交
1493

S
Steve French 已提交
1494
/*	fill in rpages then
L
Linus Torvalds 已提交
1495 1496
	result = cifs_pagein_inode(inode, index, rpages); */ /* BB finish */

1497
/*	cFYI(1, ("rpages is %d for sync page of Index %ld", rpages, index));
L
Linus Torvalds 已提交
1498

1499
#if 0
L
Linus Torvalds 已提交
1500 1501 1502
	if (rc < 0)
		return rc;
	return 0;
1503
#endif
L
Linus Torvalds 已提交
1504 1505 1506 1507 1508 1509
} */

/*
 * As file closes, flush all cached write data for this inode checking
 * for write behind errors.
 */
1510
int cifs_flush(struct file *file, fl_owner_t id)
L
Linus Torvalds 已提交
1511
{
S
Steve French 已提交
1512
	struct inode *inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526
	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);
	if (!rc) /* reset wb rc if we were able to write out dirty pages */
		CIFS_I(inode)->write_behind_rc = 0;
1527

S
Steve French 已提交
1528
	cFYI(1, ("Flush inode %p file %p rc %d", inode, file, rc));
L
Linus Torvalds 已提交
1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548

	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();
1549
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563
	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"));
	}
	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
1564
		current_read_size = min_t(const int, read_size - total_read,
L
Linus Torvalds 已提交
1565 1566 1567 1568
					  cifs_sb->rsize);
		rc = -EAGAIN;
		smb_read_data = NULL;
		while (rc == -EAGAIN) {
1569
			int buf_type = CIFS_NO_BUFFER;
S
Steve French 已提交
1570
			if ((open_file->invalidHandle) &&
L
Linus Torvalds 已提交
1571
			    (!open_file->closePend)) {
1572
				rc = cifs_reopen_file(file, TRUE);
L
Linus Torvalds 已提交
1573 1574 1575
				if (rc != 0)
					break;
			}
1576
			rc = CIFSSMBRead(xid, pTcon,
1577 1578 1579 1580
					 open_file->netfid,
					 current_read_size, *poffset,
					 &bytes_read, &smb_read_data,
					 &buf_type);
L
Linus Torvalds 已提交
1581 1582
			pSMBr = (struct smb_com_read_rsp *)smb_read_data;
			if (smb_read_data) {
1583 1584 1585 1586 1587 1588 1589 1590
				if (copy_to_user(current_offset,
						smb_read_data +
						4 /* RFC1001 length field */ +
						le16_to_cpu(pSMBr->DataOffset),
						bytes_read)) {
					rc = -EFAULT;
				}

S
Steve French 已提交
1591
				if (buf_type == CIFS_SMALL_BUFFER)
1592
					cifs_small_buf_release(smb_read_data);
S
Steve French 已提交
1593
				else if (buf_type == CIFS_LARGE_BUFFER)
1594
					cifs_buf_release(smb_read_data);
L
Linus Torvalds 已提交
1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605
				smb_read_data = NULL;
			}
		}
		if (rc || (bytes_read == 0)) {
			if (total_read) {
				break;
			} else {
				FreeXid(xid);
				return rc;
			}
		} else {
1606
			cifs_stats_bytes_read(pTcon, bytes_read);
L
Linus Torvalds 已提交
1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626
			*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;
1627
	int buf_type = CIFS_NO_BUFFER;
L
Linus Torvalds 已提交
1628 1629

	xid = GetXid();
1630
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641
	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
1642
	for (total_read = 0, current_offset = read_data;
L
Linus Torvalds 已提交
1643 1644 1645 1646
	     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);
1647 1648
		/* 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
1649
		if ((pTcon->ses) &&
1650 1651 1652 1653
			!(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
1654 1655
		rc = -EAGAIN;
		while (rc == -EAGAIN) {
S
Steve French 已提交
1656
			if ((open_file->invalidHandle) &&
L
Linus Torvalds 已提交
1657
			    (!open_file->closePend)) {
1658
				rc = cifs_reopen_file(file, TRUE);
L
Linus Torvalds 已提交
1659 1660 1661
				if (rc != 0)
					break;
			}
1662
			rc = CIFSSMBRead(xid, pTcon,
1663 1664 1665 1666
					 open_file->netfid,
					 current_read_size, *poffset,
					 &bytes_read, &current_offset,
					 &buf_type);
L
Linus Torvalds 已提交
1667 1668 1669 1670 1671 1672 1673 1674 1675
		}
		if (rc || (bytes_read == 0)) {
			if (total_read) {
				break;
			} else {
				FreeXid(xid);
				return rc;
			}
		} else {
1676
			cifs_stats_bytes_read(pTcon, total_read);
L
Linus Torvalds 已提交
1677 1678 1679 1680 1681 1682 1683 1684 1685
			*poffset += bytes_read;
		}
	}
	FreeXid(xid);
	return total_read;
}

int cifs_file_mmap(struct file *file, struct vm_area_struct *vma)
{
1686
	struct dentry *dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701
	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
1702
static void cifs_copy_cache_pages(struct address_space *mapping,
L
Linus Torvalds 已提交
1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719
	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"));
1720 1721
			data += PAGE_CACHE_SIZE;
			bytes_read -= PAGE_CACHE_SIZE;
L
Linus Torvalds 已提交
1722 1723 1724
			continue;
		}

S
Steve French 已提交
1725
		target = kmap_atomic(page, KM_USER0);
L
Linus Torvalds 已提交
1726 1727 1728 1729

		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
1730
			memset(target + bytes_read, 0,
L
Linus Torvalds 已提交
1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757
			       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
1758
	unsigned int bytes_read = 0;
S
Steve French 已提交
1759
	unsigned int read_size, i;
L
Linus Torvalds 已提交
1760 1761 1762 1763
	char *smb_read_data = NULL;
	struct smb_com_read_rsp *pSMBr;
	struct pagevec lru_pvec;
	struct cifsFileInfo *open_file;
1764
	int buf_type = CIFS_NO_BUFFER;
L
Linus Torvalds 已提交
1765 1766 1767 1768 1769 1770 1771

	xid = GetXid();
	if (file->private_data == NULL) {
		FreeXid(xid);
		return -EBADF;
	}
	open_file = (struct cifsFileInfo *)file->private_data;
1772
	cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
L
Linus Torvalds 已提交
1773
	pTcon = cifs_sb->tcon;
1774

L
Linus Torvalds 已提交
1775
	pagevec_init(&lru_pvec, 0);
1776
#ifdef CONFIG_CIFS_DEBUG2
S
Steve French 已提交
1777 1778
		cFYI(1, ("rpages: num pages %d", num_pages));
#endif
L
Linus Torvalds 已提交
1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791
	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
1792
		expected_index =
L
Linus Torvalds 已提交
1793
			list_entry(page_list->prev, struct page, lru)->index;
S
Steve French 已提交
1794
		list_for_each_entry_reverse(tmp_page, page_list, lru) {
L
Linus Torvalds 已提交
1795 1796 1797 1798
			if (tmp_page->index == expected_index) {
				contig_pages++;
				expected_index++;
			} else
S
Steve French 已提交
1799
				break;
L
Linus Torvalds 已提交
1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810
		}
		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);
1811
#ifdef CONFIG_CIFS_DEBUG2
S
Steve French 已提交
1812
		cFYI(1, ("rpages: read size 0x%x  contiguous pages %d",
1813
				read_size, contig_pages));
S
Steve French 已提交
1814
#endif
L
Linus Torvalds 已提交
1815 1816
		rc = -EAGAIN;
		while (rc == -EAGAIN) {
S
Steve French 已提交
1817
			if ((open_file->invalidHandle) &&
L
Linus Torvalds 已提交
1818
			    (!open_file->closePend)) {
1819
				rc = cifs_reopen_file(file, TRUE);
L
Linus Torvalds 已提交
1820 1821 1822 1823
				if (rc != 0)
					break;
			}

1824
			rc = CIFSSMBRead(xid, pTcon,
1825 1826 1827 1828
					 open_file->netfid,
					 read_size, offset,
					 &bytes_read, &smb_read_data,
					 &buf_type);
1829
			/* BB more RC checks ? */
S
Steve French 已提交
1830
			if (rc == -EAGAIN) {
L
Linus Torvalds 已提交
1831
				if (smb_read_data) {
S
Steve French 已提交
1832
					if (buf_type == CIFS_SMALL_BUFFER)
1833
						cifs_small_buf_release(smb_read_data);
S
Steve French 已提交
1834
					else if (buf_type == CIFS_LARGE_BUFFER)
1835
						cifs_buf_release(smb_read_data);
L
Linus Torvalds 已提交
1836 1837 1838 1839 1840 1841 1842 1843
					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) {
1844
			task_io_account_read(bytes_read);
L
Linus Torvalds 已提交
1845 1846 1847 1848 1849 1850
			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;
1851
			cifs_stats_bytes_read(pTcon, bytes_read);
S
Steve French 已提交
1852
			if ((bytes_read & PAGE_CACHE_MASK) != bytes_read) {
L
Linus Torvalds 已提交
1853 1854
				i++; /* account for partial page */

S
Steve French 已提交
1855
				/* server copy of file can have smaller size
L
Linus Torvalds 已提交
1856
				   than client */
S
Steve French 已提交
1857 1858
				/* 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
1859 1860
				   we will hit it on next read */

O
OGAWA Hirofumi 已提交
1861
				/* break; */
L
Linus Torvalds 已提交
1862 1863 1864 1865 1866
			}
		} 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
1867
			/* BB turn off caching and do new lookup on
L
Linus Torvalds 已提交
1868 1869 1870 1871
			   file size at server? */
			break;
		}
		if (smb_read_data) {
S
Steve French 已提交
1872
			if (buf_type == CIFS_SMALL_BUFFER)
1873
				cifs_small_buf_release(smb_read_data);
S
Steve French 已提交
1874
			else if (buf_type == CIFS_LARGE_BUFFER)
1875
				cifs_buf_release(smb_read_data);
L
Linus Torvalds 已提交
1876 1877 1878 1879 1880 1881 1882 1883 1884
			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
1885
		if (buf_type == CIFS_SMALL_BUFFER)
1886
			cifs_small_buf_release(smb_read_data);
S
Steve French 已提交
1887
		else if (buf_type == CIFS_LARGE_BUFFER)
1888
			cifs_buf_release(smb_read_data);
L
Linus Torvalds 已提交
1889
		smb_read_data = NULL;
S
Steve French 已提交
1890
	}
L
Linus Torvalds 已提交
1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904

	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
1905

L
Linus Torvalds 已提交
1906
	rc = cifs_read(file, read_data, PAGE_CACHE_SIZE, poffset);
S
Steve French 已提交
1907

L
Linus Torvalds 已提交
1908 1909 1910
	if (rc < 0)
		goto io_error;
	else
S
Steve French 已提交
1911 1912
		cFYI(1, ("Bytes read %d", rc));

1913 1914
	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
1915

L
Linus Torvalds 已提交
1916 1917 1918 1919 1920 1921
	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
1922

L
Linus Torvalds 已提交
1923
io_error:
S
Steve French 已提交
1924
	kunmap(page);
L
Linus Torvalds 已提交
1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941
	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
1942
	cFYI(1, ("readpage %p at offset %d 0x%x\n",
L
Linus Torvalds 已提交
1943 1944 1945 1946 1947 1948 1949 1950 1951 1952
		 page, (int)offset, (int)offset));

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

	unlock_page(page);

	FreeXid(xid);
	return rc;
}

1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971
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
1972 1973 1974
/* 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
1975
   refreshing the inode only on increases in the file size
L
Linus Torvalds 已提交
1976 1977
   but this is tricky to do without racing with writebehind
   page caching in the current Linux kernel design */
1978
int is_size_safe_to_change(struct cifsInodeInfo *cifsInode, __u64 end_of_file)
L
Linus Torvalds 已提交
1979
{
1980 1981
	if (!cifsInode)
		return 1;
1982

1983 1984
	if (is_inode_writable(cifsInode)) {
		/* This inode is open for write at least once */
1985 1986 1987 1988
		struct cifs_sb_info *cifs_sb;

		cifs_sb = CIFS_SB(cifsInode->vfs_inode.i_sb);
		if ( cifs_sb->mnt_cifs_flags & CIFS_MOUNT_DIRECT_IO ) {
S
Steve French 已提交
1989
			/* since no page cache to corrupt on directio
1990 1991 1992 1993
			we can change size safely */
			return 1;
		}

S
Steve French 已提交
1994
		if (i_size_read(&cifsInode->vfs_inode) < end_of_file)
1995 1996
			return 1;

1997
		return 0;
1998
	} else
1999
		return 1;
L
Linus Torvalds 已提交
2000 2001 2002 2003 2004 2005
}

static int cifs_prepare_write(struct file *file, struct page *page,
	unsigned from, unsigned to)
{
	int rc = 0;
2006 2007 2008
	loff_t i_size;
	loff_t offset;

S
Steve French 已提交
2009
	cFYI(1, ("prepare write for page %p from %d to %d", page, from, to));
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
	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
		 */
2029
		simple_prepare_write(file, page, from, to);
2030 2031
		SetPageUptodate(page);
	} else if ((file->f_flags & O_ACCMODE) != O_WRONLY) {
L
Linus Torvalds 已提交
2032
		/* might as well read a page, it is fast enough */
2033 2034 2035 2036 2037 2038
		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
2039 2040
	}

S
Steve French 已提交
2041 2042
	/* we do not need to pass errors back
	   e.g. if we do not have read access to the file
2043 2044
	   because cifs_commit_write will do the right thing.  -- shaggy */

L
Linus Torvalds 已提交
2045 2046 2047
	return 0;
}

2048
const struct address_space_operations cifs_addr_ops = {
L
Linus Torvalds 已提交
2049 2050 2051
	.readpage = cifs_readpage,
	.readpages = cifs_readpages,
	.writepage = cifs_writepage,
2052
	.writepages = cifs_writepages,
L
Linus Torvalds 已提交
2053 2054 2055 2056 2057 2058
	.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
2059 2060 2061 2062 2063 2064

/*
 * 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.
 */
2065
const struct address_space_operations cifs_addr_ops_smallbuf = {
D
Dave Kleikamp 已提交
2066 2067 2068 2069 2070 2071 2072 2073 2074
	.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 = */
};