linux/fs/cifs/file.c

3444 lines
88 KiB
C
Raw Normal View History

/*
* fs/cifs/file.c
*
* vfs operations that deal with files
*
* Copyright (C) International Business Machines Corp., 2002,2010
* Author(s): Steve French (sfrench@us.ibm.com)
* Jeremy Allison (jra@samba.org)
*
* 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>
#include <linux/backing-dev.h>
#include <linux/stat.h>
#include <linux/fcntl.h>
#include <linux/pagemap.h>
#include <linux/pagevec.h>
#include <linux/writeback.h>
#include <linux/task_io_accounting_ops.h>
#include <linux/delay.h>
#include <linux/mount.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include <linux/swap.h>
#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"
#include "fscache.h"
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 (READ_CONTROL | FILE_WRITE_ATTRIBUTES | FILE_READ_ATTRIBUTES |
FILE_WRITE_EA | FILE_APPEND_DATA | FILE_WRITE_DATA |
FILE_READ_DATA);
}
static u32 cifs_posix_convert_flags(unsigned int flags)
{
u32 posix_flags = 0;
if ((flags & O_ACCMODE) == O_RDONLY)
posix_flags = SMB_O_RDONLY;
else if ((flags & O_ACCMODE) == O_WRONLY)
posix_flags = SMB_O_WRONLY;
else if ((flags & O_ACCMODE) == O_RDWR)
posix_flags = SMB_O_RDWR;
if (flags & O_CREAT)
posix_flags |= SMB_O_CREAT;
if (flags & O_EXCL)
posix_flags |= SMB_O_EXCL;
if (flags & O_TRUNC)
posix_flags |= SMB_O_TRUNC;
/* be safe and imply O_SYNC for O_DSYNC */
vfs: Implement proper O_SYNC semantics While Linux provided an O_SYNC flag basically since day 1, it took until Linux 2.4.0-test12pre2 to actually get it implemented for filesystems, since that day we had generic_osync_around with only minor changes and the great "For now, when the user asks for O_SYNC, we'll actually give O_DSYNC" comment. This patch intends to actually give us real O_SYNC semantics in addition to the O_DSYNC semantics. After Jan's O_SYNC patches which are required before this patch it's actually surprisingly simple, we just need to figure out when to set the datasync flag to vfs_fsync_range and when not. This patch renames the existing O_SYNC flag to O_DSYNC while keeping it's numerical value to keep binary compatibility, and adds a new real O_SYNC flag. To guarantee backwards compatiblity it is defined as expanding to both the O_DSYNC and the new additional binary flag (__O_SYNC) to make sure we are backwards-compatible when compiled against the new headers. This also means that all places that don't care about the differences can just check O_DSYNC and get the right behaviour for O_SYNC, too - only places that actuall care need to check __O_SYNC in addition. Drivers and network filesystems have been updated in a fail safe way to always do the full sync magic if O_DSYNC is set. The few places setting O_SYNC for lower layers are kept that way for now to stay failsafe. We enforce that O_DSYNC is set when __O_SYNC is set early in the open path to make sure we always get these sane options. Note that parisc really screwed up their headers as they already define a O_DSYNC that has always been a no-op. We try to repair it by using it for the new O_DSYNC and redefinining O_SYNC to send both the traditional O_SYNC numerical value _and_ the O_DSYNC one. Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: Grant Grundler <grundler@parisc-linux.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: Ingo Molnar <mingo@elte.hu> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Andreas Dilger <adilger@sun.com> Acked-by: Trond Myklebust <Trond.Myklebust@netapp.com> Acked-by: Kyle McMartin <kyle@mcmartin.ca> Acked-by: Ulrich Drepper <drepper@redhat.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jan Kara <jack@suse.cz>
2009-10-27 10:05:28 +00:00
if (flags & O_DSYNC)
posix_flags |= SMB_O_SYNC;
if (flags & O_DIRECTORY)
posix_flags |= SMB_O_DIRECTORY;
if (flags & O_NOFOLLOW)
posix_flags |= SMB_O_NOFOLLOW;
if (flags & O_DIRECT)
posix_flags |= SMB_O_DIRECT;
return posix_flags;
}
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;
else if ((flags & O_TRUNC) == O_TRUNC)
return FILE_OVERWRITE;
else
return FILE_OPEN;
}
int cifs_posix_open(char *full_path, struct inode **pinode,
struct super_block *sb, int mode, unsigned int f_flags,
__u32 *poplock, __u16 *pnetfid, unsigned int xid)
{
int rc;
FILE_UNIX_BASIC_INFO *presp_data;
__u32 posix_flags = 0;
struct cifs_sb_info *cifs_sb = CIFS_SB(sb);
struct cifs_fattr fattr;
struct tcon_link *tlink;
struct cifs_tcon *tcon;
cFYI(1, "posix open %s", full_path);
presp_data = kzalloc(sizeof(FILE_UNIX_BASIC_INFO), GFP_KERNEL);
if (presp_data == NULL)
return -ENOMEM;
tlink = cifs_sb_tlink(cifs_sb);
if (IS_ERR(tlink)) {
rc = PTR_ERR(tlink);
goto posix_open_ret;
}
tcon = tlink_tcon(tlink);
mode &= ~current_umask();
posix_flags = cifs_posix_convert_flags(f_flags);
rc = CIFSPOSIXCreate(xid, tcon, posix_flags, mode, pnetfid, presp_data,
poplock, full_path, cifs_sb->local_nls,
cifs_sb->mnt_cifs_flags &
CIFS_MOUNT_MAP_SPECIAL_CHR);
cifs_put_tlink(tlink);
if (rc)
goto posix_open_ret;
if (presp_data->Type == cpu_to_le32(-1))
goto posix_open_ret; /* open ok, caller does qpathinfo */
if (!pinode)
goto posix_open_ret; /* caller does not need info */
cifs_unix_basic_to_fattr(&fattr, presp_data, cifs_sb);
/* get new inode and set it up */
if (*pinode == NULL) {
cifs_fill_uniqueid(sb, &fattr);
*pinode = cifs_iget(sb, &fattr);
if (!*pinode) {
rc = -ENOMEM;
goto posix_open_ret;
}
} else {
cifs_fattr_to_inode(*pinode, &fattr);
}
posix_open_ret:
kfree(presp_data);
return rc;
}
static int
cifs_nt_open(char *full_path, struct inode *inode, struct cifs_sb_info *cifs_sb,
struct cifs_tcon *tcon, unsigned int f_flags, __u32 *oplock,
struct cifs_fid *fid, unsigned int xid)
{
int rc;
int desired_access;
int disposition;
int create_options = CREATE_NOT_DIR;
FILE_ALL_INFO *buf;
if (!tcon->ses->server->ops->open)
return -ENOSYS;
desired_access = cifs_convert_flags(f_flags);
/*********************************************************************
* open flag mapping table:
*
* POSIX Flag CIFS Disposition
* ---------- ----------------
* 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
* FILE_SUPERSEDE (ie create whether or not file exists although
* 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)
*?
*? O_SYNC is a reasonable match to CIFS writethrough flag
*? 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(f_flags);
/* BB pass O_SYNC flag through on file attributes .. BB */
buf = kmalloc(sizeof(FILE_ALL_INFO), GFP_KERNEL);
if (!buf)
return -ENOMEM;
if (backup_cred(cifs_sb))
create_options |= CREATE_OPEN_BACKUP_INTENT;
rc = tcon->ses->server->ops->open(xid, tcon, full_path, disposition,
desired_access, create_options, fid,
oplock, buf, cifs_sb);
if (rc)
goto out;
if (tcon->unix_ext)
rc = cifs_get_inode_info_unix(&inode, full_path, inode->i_sb,
xid);
else
rc = cifs_get_inode_info(&inode, full_path, buf, inode->i_sb,
xid, &fid->netfid);
out:
kfree(buf);
return rc;
}
struct cifsFileInfo *
cifs_new_fileinfo(struct cifs_fid *fid, struct file *file,
struct tcon_link *tlink, __u32 oplock)
{
struct dentry *dentry = file->f_path.dentry;
struct inode *inode = dentry->d_inode;
struct cifsInodeInfo *cinode = CIFS_I(inode);
struct cifsFileInfo *cfile;
cfile = kzalloc(sizeof(struct cifsFileInfo), GFP_KERNEL);
if (cfile == NULL)
return cfile;
cfile->count = 1;
cfile->pid = current->tgid;
cfile->uid = current_fsuid();
cfile->dentry = dget(dentry);
cfile->f_flags = file->f_flags;
cfile->invalidHandle = false;
cfile->tlink = cifs_get_tlink(tlink);
mutex_init(&cfile->fh_mutex);
INIT_WORK(&cfile->oplock_break, cifs_oplock_break);
INIT_LIST_HEAD(&cfile->llist);
tlink_tcon(tlink)->ses->server->ops->set_fid(cfile, fid, oplock);
spin_lock(&cifs_file_list_lock);
list_add(&cfile->tlist, &(tlink_tcon(tlink)->openFileList));
/* if readable file instance put first in list*/
if (file->f_mode & FMODE_READ)
list_add(&cfile->flist, &cinode->openFileList);
else
list_add_tail(&cfile->flist, &cinode->openFileList);
spin_unlock(&cifs_file_list_lock);
file->private_data = cfile;
return cfile;
}
static void cifs_del_lock_waiters(struct cifsLockInfo *lock);
struct cifsFileInfo *
cifsFileInfo_get(struct cifsFileInfo *cifs_file)
{
spin_lock(&cifs_file_list_lock);
cifsFileInfo_get_locked(cifs_file);
spin_unlock(&cifs_file_list_lock);
return cifs_file;
}
/*
* Release a reference on the file private data. This may involve closing
* the filehandle out on the server. Must be called without holding
* cifs_file_list_lock.
*/
void cifsFileInfo_put(struct cifsFileInfo *cifs_file)
{
struct inode *inode = cifs_file->dentry->d_inode;
struct cifs_tcon *tcon = tlink_tcon(cifs_file->tlink);
struct cifsInodeInfo *cifsi = CIFS_I(inode);
struct cifs_sb_info *cifs_sb = CIFS_SB(inode->i_sb);
struct cifsLockInfo *li, *tmp;
spin_lock(&cifs_file_list_lock);
if (--cifs_file->count > 0) {
spin_unlock(&cifs_file_list_lock);
return;
}
/* remove it from the lists */
list_del(&cifs_file->flist);
list_del(&cifs_file->tlist);
if (list_empty(&cifsi->openFileList)) {
cFYI(1, "closing last open instance for inode %p",
cifs_file->dentry->d_inode);
/*
* In strict cache mode we need invalidate mapping on the last
* close because it may cause a error when we open this file
* again and get at least level II oplock.
*/
if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_STRICT_IO)
CIFS_I(inode)->invalid_mapping = true;
cifs_set_oplock_level(cifsi, 0);
}
spin_unlock(&cifs_file_list_lock);
cancel_work_sync(&cifs_file->oplock_break);
if (!tcon->need_reconnect && !cifs_file->invalidHandle) {
struct TCP_Server_Info *server = tcon->ses->server;
unsigned int xid;
int rc = -ENOSYS;
xid = get_xid();
if (server->ops->close)
rc = server->ops->close(xid, tcon, &cifs_file->fid);
free_xid(xid);
}
/* Delete any outstanding lock records. We'll lose them when the file
* is closed anyway.
*/
mutex_lock(&cifsi->lock_mutex);
list_for_each_entry_safe(li, tmp, &cifs_file->llist, llist) {
list_del(&li->llist);
cifs_del_lock_waiters(li);
kfree(li);
}
mutex_unlock(&cifsi->lock_mutex);
cifs_put_tlink(cifs_file->tlink);
dput(cifs_file->dentry);
kfree(cifs_file);
}
int cifs_open(struct inode *inode, struct file *file)
{
int rc = -EACCES;
unsigned int xid;
__u32 oplock;
struct cifs_sb_info *cifs_sb;
struct cifs_tcon *tcon;
struct tcon_link *tlink;
struct cifsFileInfo *cfile = NULL;
char *full_path = NULL;
bool posix_open_ok = false;
struct cifs_fid fid;
xid = get_xid();
cifs_sb = CIFS_SB(inode->i_sb);
tlink = cifs_sb_tlink(cifs_sb);
if (IS_ERR(tlink)) {
free_xid(xid);
return PTR_ERR(tlink);
}
tcon = tlink_tcon(tlink);
full_path = build_path_from_dentry(file->f_path.dentry);
if (full_path == NULL) {
cifs: Fix incorrect return code being printed in cFYI messages FreeXid() along with freeing Xid does add a cifsFYI debug message that prints rc (return code) as well. In some code paths where we set/return error code after calling FreeXid(), incorrect error code is being printed when cifsFYI is enabled. This could be misleading in few cases. For eg. In cifs_open() if cifs_fill_filedata() returns a valid pointer to cifsFileInfo, FreeXid() prints rc=-13 whereas 0 is actually being returned. Fix this by setting rc before calling FreeXid(). Basically convert FreeXid(xid); rc = -ERR; return -ERR; => FreeXid(xid); return rc; [Note that Christoph would like to replace the GetXid/FreeXid calls, which are primarily used for debugging. This seems like a good longer term goal, but although there is an alternative tracing facility, there are no examples yet available that I know of that we can use (yet) to convert this cifs function entry/exit logging, and for creating an identifier that we can use to correlate all dmesg log entries for a particular vfs operation (ie identify all log entries for a particular vfs request to cifs: e.g. a particular close or read or write or byte range lock call ... and just using the thread id is harder). Eventually when a replacement for this is available (e.g. when NFS switches over and various samples to look at in other file systems) we can remove the GetXid/FreeXid macro but in the meantime multiple people use this run time configurable logging all the time for debugging, and Suresh's patch fixes a problem which made it harder to notice some low memory problems in the log so it is worthwhile to fix this problem until a better logging approach is able to be used] Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: Suresh Jayaraman <sjayaraman@suse.de> Signed-off-by: Steve French <sfrench@us.ibm.com>
2009-06-25 12:42:34 +00:00
rc = -ENOMEM;
goto out;
}
cFYI(1, "inode = 0x%p file flags are 0x%x for %s",
inode, file->f_flags, full_path);
if (tcon->ses->server->oplocks)
oplock = REQ_OPLOCK;
else
oplock = 0;
if (!tcon->broken_posix_open && tcon->unix_ext &&
cap_unix(tcon->ses) && (CIFS_UNIX_POSIX_PATH_OPS_CAP &
le64_to_cpu(tcon->fsUnixInfo.Capability))) {
/* can not refresh inode info since size could be stale */
rc = cifs_posix_open(full_path, &inode, inode->i_sb,
cifs_sb->mnt_file_mode /* ignored */,
file->f_flags, &oplock, &fid.netfid, xid);
if (rc == 0) {
cFYI(1, "posix open succeeded");
posix_open_ok = true;
} else if ((rc == -EINVAL) || (rc == -EOPNOTSUPP)) {
if (tcon->ses->serverNOS)
cERROR(1, "server %s of type %s returned"
" unexpected error on SMB posix open"
", disabling posix open support."
" Check if server update available.",
tcon->ses->serverName,
tcon->ses->serverNOS);
tcon->broken_posix_open = true;
} else if ((rc != -EIO) && (rc != -EREMOTE) &&
(rc != -EOPNOTSUPP)) /* path not found or net err */
goto out;
/*
* Else fallthrough to retry open the old way on network i/o
* or DFS errors.
*/
}
if (!posix_open_ok) {
rc = cifs_nt_open(full_path, inode, cifs_sb, tcon,
file->f_flags, &oplock, &fid, xid);
if (rc)
goto out;
}
cfile = cifs_new_fileinfo(&fid, file, tlink, oplock);
if (cfile == NULL) {
if (tcon->ses->server->ops->close)
tcon->ses->server->ops->close(xid, tcon, &fid);
rc = -ENOMEM;
goto out;
}
cifs_fscache_set_inode_cookie(inode, file);
if ((oplock & CIFS_CREATE_ACTION) && !posix_open_ok && tcon->unix_ext) {
/*
* Time to set mode which we can not set earlier due to
* problems creating new read-only files.
*/
struct cifs_unix_set_info_args args = {
.mode = inode->i_mode,
.uid = NO_CHANGE_64,
.gid = NO_CHANGE_64,
.ctime = NO_CHANGE_64,
.atime = NO_CHANGE_64,
.mtime = NO_CHANGE_64,
.device = 0,
};
CIFSSMBUnixSetFileInfo(xid, tcon, &args, fid.netfid,
cfile->pid);
}
out:
kfree(full_path);
free_xid(xid);
cifs_put_tlink(tlink);
return rc;
}
/*
* Try to reacquire byte range locks that were released when session
* 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;
}
static int
cifs_reopen_file(struct cifsFileInfo *cfile, bool can_flush)
{
int rc = -EACCES;
unsigned int xid;
__u32 oplock;
struct cifs_sb_info *cifs_sb;
struct cifs_tcon *tcon;
struct TCP_Server_Info *server;
struct cifsInodeInfo *cinode;
struct inode *inode;
char *full_path = NULL;
int desired_access;
int disposition = FILE_OPEN;
int create_options = CREATE_NOT_DIR;
struct cifs_fid fid;
xid = get_xid();
mutex_lock(&cfile->fh_mutex);
if (!cfile->invalidHandle) {
mutex_unlock(&cfile->fh_mutex);
cifs: Fix incorrect return code being printed in cFYI messages FreeXid() along with freeing Xid does add a cifsFYI debug message that prints rc (return code) as well. In some code paths where we set/return error code after calling FreeXid(), incorrect error code is being printed when cifsFYI is enabled. This could be misleading in few cases. For eg. In cifs_open() if cifs_fill_filedata() returns a valid pointer to cifsFileInfo, FreeXid() prints rc=-13 whereas 0 is actually being returned. Fix this by setting rc before calling FreeXid(). Basically convert FreeXid(xid); rc = -ERR; return -ERR; => FreeXid(xid); return rc; [Note that Christoph would like to replace the GetXid/FreeXid calls, which are primarily used for debugging. This seems like a good longer term goal, but although there is an alternative tracing facility, there are no examples yet available that I know of that we can use (yet) to convert this cifs function entry/exit logging, and for creating an identifier that we can use to correlate all dmesg log entries for a particular vfs operation (ie identify all log entries for a particular vfs request to cifs: e.g. a particular close or read or write or byte range lock call ... and just using the thread id is harder). Eventually when a replacement for this is available (e.g. when NFS switches over and various samples to look at in other file systems) we can remove the GetXid/FreeXid macro but in the meantime multiple people use this run time configurable logging all the time for debugging, and Suresh's patch fixes a problem which made it harder to notice some low memory problems in the log so it is worthwhile to fix this problem until a better logging approach is able to be used] Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: Suresh Jayaraman <sjayaraman@suse.de> Signed-off-by: Steve French <sfrench@us.ibm.com>
2009-06-25 12:42:34 +00:00
rc = 0;
free_xid(xid);
cifs: Fix incorrect return code being printed in cFYI messages FreeXid() along with freeing Xid does add a cifsFYI debug message that prints rc (return code) as well. In some code paths where we set/return error code after calling FreeXid(), incorrect error code is being printed when cifsFYI is enabled. This could be misleading in few cases. For eg. In cifs_open() if cifs_fill_filedata() returns a valid pointer to cifsFileInfo, FreeXid() prints rc=-13 whereas 0 is actually being returned. Fix this by setting rc before calling FreeXid(). Basically convert FreeXid(xid); rc = -ERR; return -ERR; => FreeXid(xid); return rc; [Note that Christoph would like to replace the GetXid/FreeXid calls, which are primarily used for debugging. This seems like a good longer term goal, but although there is an alternative tracing facility, there are no examples yet available that I know of that we can use (yet) to convert this cifs function entry/exit logging, and for creating an identifier that we can use to correlate all dmesg log entries for a particular vfs operation (ie identify all log entries for a particular vfs request to cifs: e.g. a particular close or read or write or byte range lock call ... and just using the thread id is harder). Eventually when a replacement for this is available (e.g. when NFS switches over and various samples to look at in other file systems) we can remove the GetXid/FreeXid macro but in the meantime multiple people use this run time configurable logging all the time for debugging, and Suresh's patch fixes a problem which made it harder to notice some low memory problems in the log so it is worthwhile to fix this problem until a better logging approach is able to be used] Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: Suresh Jayaraman <sjayaraman@suse.de> Signed-off-by: Steve French <sfrench@us.ibm.com>
2009-06-25 12:42:34 +00:00
return rc;
}
inode = cfile->dentry->d_inode;
cifs_sb = CIFS_SB(inode->i_sb);
tcon = tlink_tcon(cfile->tlink);
server = tcon->ses->server;
/*
* 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.
*/
full_path = build_path_from_dentry(cfile->dentry);
if (full_path == NULL) {
rc = -ENOMEM;
mutex_unlock(&cfile->fh_mutex);
free_xid(xid);
return rc;
}
cFYI(1, "inode = 0x%p file flags 0x%x for %s", inode, cfile->f_flags,
full_path);
if (tcon->ses->server->oplocks)
oplock = REQ_OPLOCK;
else
oplock = 0;
if (tcon->unix_ext && cap_unix(tcon->ses) &&
(CIFS_UNIX_POSIX_PATH_OPS_CAP &
le64_to_cpu(tcon->fsUnixInfo.Capability))) {
/*
* O_CREAT, O_EXCL and O_TRUNC already had their effect on the
* original open. Must mask them off for a reopen.
*/
unsigned int oflags = cfile->f_flags &
~(O_CREAT | O_EXCL | O_TRUNC);
rc = cifs_posix_open(full_path, NULL, inode->i_sb,
cifs_sb->mnt_file_mode /* ignored */,
oflags, &oplock, &fid.netfid, xid);
if (rc == 0) {
cFYI(1, "posix reopen succeeded");
goto reopen_success;
}
/*
* fallthrough to retry open the old way on errors, especially
* in the reconnect path it is important to retry hard
*/
}
desired_access = cifs_convert_flags(cfile->f_flags);
if (backup_cred(cifs_sb))
create_options |= CREATE_OPEN_BACKUP_INTENT;
/*
* Can not refresh inode by passing in file_info buf to be returned by
* CIFSSMBOpen and then calling get_inode_info with returned buf since
* file might have write behind data that needs to be flushed 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 = server->ops->open(xid, tcon, full_path, disposition,
desired_access, create_options, &fid, &oplock,
NULL, cifs_sb);
if (rc) {
mutex_unlock(&cfile->fh_mutex);
cFYI(1, "cifs_reopen returned 0x%x", rc);
cFYI(1, "oplock: %d", oplock);
goto reopen_error_exit;
}
reopen_success:
cfile->invalidHandle = false;
mutex_unlock(&cfile->fh_mutex);
cinode = CIFS_I(inode);
if (can_flush) {
rc = filemap_write_and_wait(inode->i_mapping);
mapping_set_error(inode->i_mapping, rc);
if (tcon->unix_ext)
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, NULL);
}
/*
* 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 inode info.
*/
server->ops->set_fid(cfile, &fid, oplock);
cifs_relock_file(cfile);
reopen_error_exit:
kfree(full_path);
free_xid(xid);
return rc;
}
int cifs_close(struct inode *inode, struct file *file)
{
if (file->private_data != NULL) {
cifsFileInfo_put(file->private_data);
file->private_data = NULL;
}
/* return code from the ->release op is always ignored */
return 0;
}
int cifs_closedir(struct inode *inode, struct file *file)
{
int rc = 0;
unsigned int xid;
struct cifsFileInfo *cfile = file->private_data;
char *tmp;
cFYI(1, "Closedir inode = 0x%p", inode);
xid = get_xid();
if (cfile) {
struct cifs_tcon *tcon = tlink_tcon(cfile->tlink);
cFYI(1, "Freeing private data in close dir");
spin_lock(&cifs_file_list_lock);
if (!cfile->srch_inf.endOfSearch && !cfile->invalidHandle) {
cfile->invalidHandle = true;
spin_unlock(&cifs_file_list_lock);
rc = CIFSFindClose(xid, tcon, cfile->fid.netfid);
cFYI(1, "Closing uncompleted readdir with rc %d", rc);
/* not much we can do if it fails anyway, ignore rc */
rc = 0;
} else
spin_unlock(&cifs_file_list_lock);
tmp = cfile->srch_inf.ntwrk_buf_start;
if (tmp) {
cFYI(1, "closedir free smb buf in srch struct");
cfile->srch_inf.ntwrk_buf_start = NULL;
if (cfile->srch_inf.smallBuf)
cifs_small_buf_release(tmp);
else
cifs_buf_release(tmp);
}
cifs_put_tlink(cfile->tlink);
kfree(file->private_data);
file->private_data = NULL;
}
/* BB can we lock the filestruct while this is going on? */
free_xid(xid);
return rc;
}
static struct cifsLockInfo *
cifs_lock_init(__u64 offset, __u64 length, __u8 type)
{
struct cifsLockInfo *lock =
kmalloc(sizeof(struct cifsLockInfo), GFP_KERNEL);
if (!lock)
return lock;
lock->offset = offset;
lock->length = length;
lock->type = type;
lock->pid = current->tgid;
INIT_LIST_HEAD(&lock->blist);
init_waitqueue_head(&lock->block_q);
return lock;
}
static void
cifs_del_lock_waiters(struct cifsLockInfo *lock)
{
struct cifsLockInfo *li, *tmp;
list_for_each_entry_safe(li, tmp, &lock->blist, blist) {
list_del_init(&li->blist);
wake_up(&li->block_q);
}
}
static bool
cifs_find_fid_lock_conflict(struct cifsFileInfo *cfile, __u64 offset,
__u64 length, __u8 type, struct cifsFileInfo *cur,
struct cifsLockInfo **conf_lock)
{
struct cifsLockInfo *li;
struct TCP_Server_Info *server = tlink_tcon(cfile->tlink)->ses->server;
list_for_each_entry(li, &cfile->llist, llist) {
if (offset + length <= li->offset ||
offset >= li->offset + li->length)
continue;
else if ((type & server->vals->shared_lock_type) &&
((server->ops->compare_fids(cur, cfile) &&
current->tgid == li->pid) || type == li->type))
continue;
else {
*conf_lock = li;
return true;
}
}
return false;
}
static bool
cifs_find_lock_conflict(struct cifsFileInfo *cfile, __u64 offset, __u64 length,
__u8 type, struct cifsLockInfo **conf_lock)
{
bool rc = false;
struct cifsFileInfo *fid, *tmp;
struct cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode);
spin_lock(&cifs_file_list_lock);
list_for_each_entry_safe(fid, tmp, &cinode->openFileList, flist) {
rc = cifs_find_fid_lock_conflict(fid, offset, length, type,
cfile, conf_lock);
if (rc)
break;
}
spin_unlock(&cifs_file_list_lock);
return rc;
}
/*
* Check if there is another lock that prevents us to set the lock (mandatory
* style). If such a lock exists, update the flock structure with its
* properties. Otherwise, set the flock type to F_UNLCK if we can cache brlocks
* or leave it the same if we can't. Returns 0 if we don't need to request to
* the server or 1 otherwise.
*/
static int
cifs_lock_test(struct cifsFileInfo *cfile, __u64 offset, __u64 length,
__u8 type, struct file_lock *flock)
{
int rc = 0;
struct cifsLockInfo *conf_lock;
struct cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode);
struct TCP_Server_Info *server = tlink_tcon(cfile->tlink)->ses->server;
bool exist;
mutex_lock(&cinode->lock_mutex);
exist = cifs_find_lock_conflict(cfile, offset, length, type,
&conf_lock);
if (exist) {
flock->fl_start = conf_lock->offset;
flock->fl_end = conf_lock->offset + conf_lock->length - 1;
flock->fl_pid = conf_lock->pid;
if (conf_lock->type & server->vals->shared_lock_type)
flock->fl_type = F_RDLCK;
else
flock->fl_type = F_WRLCK;
} else if (!cinode->can_cache_brlcks)
rc = 1;
else
flock->fl_type = F_UNLCK;
mutex_unlock(&cinode->lock_mutex);
return rc;
}
static void
cifs_lock_add(struct cifsFileInfo *cfile, struct cifsLockInfo *lock)
{
struct cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode);
mutex_lock(&cinode->lock_mutex);
list_add_tail(&lock->llist, &cfile->llist);
mutex_unlock(&cinode->lock_mutex);
}
/*
* Set the byte-range lock (mandatory style). Returns:
* 1) 0, if we set the lock and don't need to request to the server;
* 2) 1, if no locks prevent us but we need to request to the server;
* 3) -EACCESS, if there is a lock that prevents us and wait is false.
*/
static int
cifs_lock_add_if(struct cifsFileInfo *cfile, struct cifsLockInfo *lock,
bool wait)
{
struct cifsLockInfo *conf_lock;
struct cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode);
bool exist;
int rc = 0;
try_again:
exist = false;
mutex_lock(&cinode->lock_mutex);
exist = cifs_find_lock_conflict(cfile, lock->offset, lock->length,
lock->type, &conf_lock);
if (!exist && cinode->can_cache_brlcks) {
list_add_tail(&lock->llist, &cfile->llist);
mutex_unlock(&cinode->lock_mutex);
return rc;
}
if (!exist)
rc = 1;
else if (!wait)
rc = -EACCES;
else {
list_add_tail(&lock->blist, &conf_lock->blist);
mutex_unlock(&cinode->lock_mutex);
rc = wait_event_interruptible(lock->block_q,
(lock->blist.prev == &lock->blist) &&
(lock->blist.next == &lock->blist));
if (!rc)
goto try_again;
mutex_lock(&cinode->lock_mutex);
list_del_init(&lock->blist);
}
mutex_unlock(&cinode->lock_mutex);
return rc;
}
/*
* Check if there is another lock that prevents us to set the lock (posix
* style). If such a lock exists, update the flock structure with its
* properties. Otherwise, set the flock type to F_UNLCK if we can cache brlocks
* or leave it the same if we can't. Returns 0 if we don't need to request to
* the server or 1 otherwise.
*/
static int
cifs_posix_lock_test(struct file *file, struct file_lock *flock)
{
int rc = 0;
struct cifsInodeInfo *cinode = CIFS_I(file->f_path.dentry->d_inode);
unsigned char saved_type = flock->fl_type;
if ((flock->fl_flags & FL_POSIX) == 0)
return 1;
mutex_lock(&cinode->lock_mutex);
posix_test_lock(file, flock);
if (flock->fl_type == F_UNLCK && !cinode->can_cache_brlcks) {
flock->fl_type = saved_type;
rc = 1;
}
mutex_unlock(&cinode->lock_mutex);
return rc;
}
/*
* Set the byte-range lock (posix style). Returns:
* 1) 0, if we set the lock and don't need to request to the server;
* 2) 1, if we need to request to the server;
* 3) <0, if the error occurs while setting the lock.
*/
static int
cifs_posix_lock_set(struct file *file, struct file_lock *flock)
{
struct cifsInodeInfo *cinode = CIFS_I(file->f_path.dentry->d_inode);
int rc = 1;
if ((flock->fl_flags & FL_POSIX) == 0)
return rc;
try_again:
mutex_lock(&cinode->lock_mutex);
if (!cinode->can_cache_brlcks) {
mutex_unlock(&cinode->lock_mutex);
return rc;
}
rc = posix_lock_file(file, flock, NULL);
mutex_unlock(&cinode->lock_mutex);
if (rc == FILE_LOCK_DEFERRED) {
rc = wait_event_interruptible(flock->fl_wait, !flock->fl_next);
if (!rc)
goto try_again;
locks_delete_block(flock);
}
return rc;
}
static int
cifs_push_mandatory_locks(struct cifsFileInfo *cfile)
{
unsigned int xid;
int rc = 0, stored_rc;
struct cifsLockInfo *li, *tmp;
struct cifs_tcon *tcon;
struct cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode);
unsigned int num, max_num, max_buf;
LOCKING_ANDX_RANGE *buf, *cur;
int types[] = {LOCKING_ANDX_LARGE_FILES,
LOCKING_ANDX_SHARED_LOCK | LOCKING_ANDX_LARGE_FILES};
int i;
xid = get_xid();
tcon = tlink_tcon(cfile->tlink);
mutex_lock(&cinode->lock_mutex);
if (!cinode->can_cache_brlcks) {
mutex_unlock(&cinode->lock_mutex);
free_xid(xid);
return rc;
}
/*
* Accessing maxBuf is racy with cifs_reconnect - need to store value
* and check it for zero before using.
*/
max_buf = tcon->ses->server->maxBuf;
if (!max_buf) {
mutex_unlock(&cinode->lock_mutex);
free_xid(xid);
return -EINVAL;
}
max_num = (max_buf - sizeof(struct smb_hdr)) /
sizeof(LOCKING_ANDX_RANGE);
buf = kzalloc(max_num * sizeof(LOCKING_ANDX_RANGE), GFP_KERNEL);
if (!buf) {
mutex_unlock(&cinode->lock_mutex);
free_xid(xid);
return -ENOMEM;
}
for (i = 0; i < 2; i++) {
cur = buf;
num = 0;
list_for_each_entry_safe(li, tmp, &cfile->llist, llist) {
if (li->type != types[i])
continue;
cur->Pid = cpu_to_le16(li->pid);
cur->LengthLow = cpu_to_le32((u32)li->length);
cur->LengthHigh = cpu_to_le32((u32)(li->length>>32));
cur->OffsetLow = cpu_to_le32((u32)li->offset);
cur->OffsetHigh = cpu_to_le32((u32)(li->offset>>32));
if (++num == max_num) {
stored_rc = cifs_lockv(xid, tcon,
cfile->fid.netfid,
(__u8)li->type, 0, num,
buf);
if (stored_rc)
rc = stored_rc;
cur = buf;
num = 0;
} else
cur++;
}
if (num) {
stored_rc = cifs_lockv(xid, tcon, cfile->fid.netfid,
(__u8)types[i], 0, num, buf);
if (stored_rc)
rc = stored_rc;
}
}
cinode->can_cache_brlcks = false;
mutex_unlock(&cinode->lock_mutex);
kfree(buf);
free_xid(xid);
return rc;
}
/* copied from fs/locks.c with a name change */
#define cifs_for_each_lock(inode, lockp) \
for (lockp = &inode->i_flock; *lockp != NULL; \
lockp = &(*lockp)->fl_next)
struct lock_to_push {
struct list_head llist;
__u64 offset;
__u64 length;
__u32 pid;
__u16 netfid;
__u8 type;
};
static int
cifs_push_posix_locks(struct cifsFileInfo *cfile)
{
struct cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode);
struct cifs_tcon *tcon = tlink_tcon(cfile->tlink);
struct file_lock *flock, **before;
unsigned int count = 0, i = 0;
int rc = 0, xid, type;
struct list_head locks_to_send, *el;
struct lock_to_push *lck, *tmp;
__u64 length;
xid = get_xid();
mutex_lock(&cinode->lock_mutex);
if (!cinode->can_cache_brlcks) {
mutex_unlock(&cinode->lock_mutex);
free_xid(xid);
return rc;
}
lock_flocks();
cifs_for_each_lock(cfile->dentry->d_inode, before) {
if ((*before)->fl_flags & FL_POSIX)
count++;
}
unlock_flocks();
INIT_LIST_HEAD(&locks_to_send);
/*
* Allocating count locks is enough because no FL_POSIX locks can be
* added to the list while we are holding cinode->lock_mutex that
* protects locking operations of this inode.
*/
for (; i < count; i++) {
lck = kmalloc(sizeof(struct lock_to_push), GFP_KERNEL);
if (!lck) {
rc = -ENOMEM;
goto err_out;
}
list_add_tail(&lck->llist, &locks_to_send);
}
el = locks_to_send.next;
lock_flocks();
cifs_for_each_lock(cfile->dentry->d_inode, before) {
flock = *before;
if ((flock->fl_flags & FL_POSIX) == 0)
continue;
if (el == &locks_to_send) {
/*
* The list ended. We don't have enough allocated
* structures - something is really wrong.
*/
cERROR(1, "Can't push all brlocks!");
break;
}
length = 1 + flock->fl_end - flock->fl_start;
if (flock->fl_type == F_RDLCK || flock->fl_type == F_SHLCK)
type = CIFS_RDLCK;
else
type = CIFS_WRLCK;
lck = list_entry(el, struct lock_to_push, llist);
lck->pid = flock->fl_pid;
lck->netfid = cfile->fid.netfid;
lck->length = length;
lck->type = type;
lck->offset = flock->fl_start;
el = el->next;
}
unlock_flocks();
list_for_each_entry_safe(lck, tmp, &locks_to_send, llist) {
int stored_rc;
stored_rc = CIFSSMBPosixLock(xid, tcon, lck->netfid, lck->pid,
lck->offset, lck->length, NULL,
lck->type, 0);
if (stored_rc)
rc = stored_rc;
list_del(&lck->llist);
kfree(lck);
}
out:
cinode->can_cache_brlcks = false;
mutex_unlock(&cinode->lock_mutex);
free_xid(xid);
return rc;
err_out:
list_for_each_entry_safe(lck, tmp, &locks_to_send, llist) {
list_del(&lck->llist);
kfree(lck);
}
goto out;
}
static int
cifs_push_locks(struct cifsFileInfo *cfile)
{
struct cifs_sb_info *cifs_sb = CIFS_SB(cfile->dentry->d_sb);
struct cifs_tcon *tcon = tlink_tcon(cfile->tlink);
if (cap_unix(tcon->ses) &&
(CIFS_UNIX_FCNTL_CAP & le64_to_cpu(tcon->fsUnixInfo.Capability)) &&
((cifs_sb->mnt_cifs_flags & CIFS_MOUNT_NOPOSIXBRL) == 0))
return cifs_push_posix_locks(cfile);
return cifs_push_mandatory_locks(cfile);
}
static void
cifs_read_flock(struct file_lock *flock, __u32 *type, int *lock, int *unlock,
bool *wait_flag, struct TCP_Server_Info *server)
{
if (flock->fl_flags & FL_POSIX)
cFYI(1, "Posix");
if (flock->fl_flags & FL_FLOCK)
cFYI(1, "Flock");
if (flock->fl_flags & FL_SLEEP) {
cFYI(1, "Blocking lock");
*wait_flag = true;
}
if (flock->fl_flags & FL_ACCESS)
cFYI(1, "Process suspended by mandatory locking - "
"not implemented yet");
if (flock->fl_flags & FL_LEASE)
cFYI(1, "Lease on file - not implemented yet");
if (flock->fl_flags &
(~(FL_POSIX | FL_FLOCK | FL_SLEEP | FL_ACCESS | FL_LEASE)))
cFYI(1, "Unknown lock flags 0x%x", flock->fl_flags);
*type = server->vals->large_lock_type;
if (flock->fl_type == F_WRLCK) {
cFYI(1, "F_WRLCK ");
*type |= server->vals->exclusive_lock_type;
*lock = 1;
} else if (flock->fl_type == F_UNLCK) {
cFYI(1, "F_UNLCK");
*type |= server->vals->unlock_lock_type;
*unlock = 1;
/* Check if unlock includes more than one lock range */
} else if (flock->fl_type == F_RDLCK) {
cFYI(1, "F_RDLCK");
*type |= server->vals->shared_lock_type;
*lock = 1;
} else if (flock->fl_type == F_EXLCK) {
cFYI(1, "F_EXLCK");
*type |= server->vals->exclusive_lock_type;
*lock = 1;
} else if (flock->fl_type == F_SHLCK) {
cFYI(1, "F_SHLCK");
*type |= server->vals->shared_lock_type;
*lock = 1;
} else
cFYI(1, "Unknown type of lock");
}
static int
cifs_mandatory_lock(unsigned int xid, struct cifsFileInfo *cfile, __u64 offset,
__u64 length, __u32 type, int lock, int unlock, bool wait)
{
return CIFSSMBLock(xid, tlink_tcon(cfile->tlink), cfile->fid.netfid,
current->tgid, length, offset, unlock, lock,
(__u8)type, wait, 0);
}
static int
cifs_getlk(struct file *file, struct file_lock *flock, __u32 type,
bool wait_flag, bool posix_lck, unsigned int xid)
{
int rc = 0;
__u64 length = 1 + flock->fl_end - flock->fl_start;
struct cifsFileInfo *cfile = (struct cifsFileInfo *)file->private_data;
struct cifs_tcon *tcon = tlink_tcon(cfile->tlink);
struct TCP_Server_Info *server = tcon->ses->server;
__u16 netfid = cfile->fid.netfid;
if (posix_lck) {
int posix_lock_type;
rc = cifs_posix_lock_test(file, flock);
if (!rc)
return rc;
if (type & server->vals->shared_lock_type)
posix_lock_type = CIFS_RDLCK;
else
posix_lock_type = CIFS_WRLCK;
rc = CIFSSMBPosixLock(xid, tcon, netfid, current->tgid,
flock->fl_start, length, flock,
posix_lock_type, wait_flag);
return rc;
}
rc = cifs_lock_test(cfile, flock->fl_start, length, type, flock);
if (!rc)
return rc;
/* BB we could chain these into one lock request BB */
rc = cifs_mandatory_lock(xid, cfile, flock->fl_start, length, type,
1, 0, false);
if (rc == 0) {
rc = cifs_mandatory_lock(xid, cfile, flock->fl_start, length,
type, 0, 1, false);
flock->fl_type = F_UNLCK;
if (rc != 0)
cERROR(1, "Error unlocking previously locked "
"range %d during test of lock", rc);
return 0;
}
if (type & server->vals->shared_lock_type) {
flock->fl_type = F_WRLCK;
return 0;
}
rc = cifs_mandatory_lock(xid, cfile, flock->fl_start, length,
type | server->vals->shared_lock_type, 1, 0,
false);
if (rc == 0) {
rc = cifs_mandatory_lock(xid, cfile, flock->fl_start, length,
type | server->vals->shared_lock_type,
0, 1, false);
flock->fl_type = F_RDLCK;
if (rc != 0)
cERROR(1, "Error unlocking previously locked "
"range %d during test of lock", rc);
} else
flock->fl_type = F_WRLCK;
return 0;
}
static void
cifs_move_llist(struct list_head *source, struct list_head *dest)
{
struct list_head *li, *tmp;
list_for_each_safe(li, tmp, source)
list_move(li, dest);
}
static void
cifs_free_llist(struct list_head *llist)
{
struct cifsLockInfo *li, *tmp;
list_for_each_entry_safe(li, tmp, llist, llist) {
cifs_del_lock_waiters(li);
list_del(&li->llist);
kfree(li);
}
}
static int
cifs_unlock_range(struct cifsFileInfo *cfile, struct file_lock *flock,
unsigned int xid)
{
int rc = 0, stored_rc;
int types[] = {LOCKING_ANDX_LARGE_FILES,
LOCKING_ANDX_SHARED_LOCK | LOCKING_ANDX_LARGE_FILES};
unsigned int i;
unsigned int max_num, num, max_buf;
LOCKING_ANDX_RANGE *buf, *cur;
struct cifs_tcon *tcon = tlink_tcon(cfile->tlink);
struct cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode);
struct cifsLockInfo *li, *tmp;
__u64 length = 1 + flock->fl_end - flock->fl_start;
struct list_head tmp_llist;
INIT_LIST_HEAD(&tmp_llist);
/*
* Accessing maxBuf is racy with cifs_reconnect - need to store value
* and check it for zero before using.
*/
max_buf = tcon->ses->server->maxBuf;
if (!max_buf)
return -EINVAL;
max_num = (max_buf - sizeof(struct smb_hdr)) /
sizeof(LOCKING_ANDX_RANGE);
buf = kzalloc(max_num * sizeof(LOCKING_ANDX_RANGE), GFP_KERNEL);
if (!buf)
return -ENOMEM;
mutex_lock(&cinode->lock_mutex);
for (i = 0; i < 2; i++) {
cur = buf;
num = 0;
list_for_each_entry_safe(li, tmp, &cfile->llist, llist) {
if (flock->fl_start > li->offset ||
(flock->fl_start + length) <
(li->offset + li->length))
continue;
if (current->tgid != li->pid)
continue;
if (types[i] != li->type)
continue;
if (cinode->can_cache_brlcks) {
/*
* We can cache brlock requests - simply remove
* a lock from the file's list.
*/
list_del(&li->llist);
cifs_del_lock_waiters(li);
kfree(li);
continue;
}
cur->Pid = cpu_to_le16(li->pid);
cur->LengthLow = cpu_to_le32((u32)li->length);
cur->LengthHigh = cpu_to_le32((u32)(li->length>>32));
cur->OffsetLow = cpu_to_le32((u32)li->offset);
cur->OffsetHigh = cpu_to_le32((u32)(li->offset>>32));
/*
* We need to save a lock here to let us add it again to
* the file's list if the unlock range request fails on
* the server.
*/
list_move(&li->llist, &tmp_llist);
if (++num == max_num) {
stored_rc = cifs_lockv(xid, tcon,
cfile->fid.netfid,
li->type, num, 0, buf);
if (stored_rc) {
/*
* We failed on the unlock range
* request - add all locks from the tmp
* list to the head of the file's list.
*/
cifs_move_llist(&tmp_llist,
&cfile->llist);
rc = stored_rc;
} else
/*
* The unlock range request succeed -
* free the tmp list.
*/
cifs_free_llist(&tmp_llist);
cur = buf;
num = 0;
} else
cur++;
}
if (num) {
stored_rc = cifs_lockv(xid, tcon, cfile->fid.netfid,
types[i], num, 0, buf);
if (stored_rc) {
cifs_move_llist(&tmp_llist, &cfile->llist);
rc = stored_rc;
} else
cifs_free_llist(&tmp_llist);
}
}
mutex_unlock(&cinode->lock_mutex);
kfree(buf);
return rc;
}
static int
cifs_setlk(struct file *file, struct file_lock *flock, __u32 type,
bool wait_flag, bool posix_lck, int lock, int unlock,
unsigned int xid)
{
int rc = 0;
__u64 length = 1 + flock->fl_end - flock->fl_start;
struct cifsFileInfo *cfile = (struct cifsFileInfo *)file->private_data;
struct cifs_tcon *tcon = tlink_tcon(cfile->tlink);
struct TCP_Server_Info *server = tcon->ses->server;
__u16 netfid = cfile->fid.netfid;
if (posix_lck) {
int posix_lock_type;
rc = cifs_posix_lock_set(file, flock);
if (!rc || rc < 0)
return rc;
if (type & server->vals->shared_lock_type)
posix_lock_type = CIFS_RDLCK;
else
posix_lock_type = CIFS_WRLCK;
if (unlock == 1)
posix_lock_type = CIFS_UNLCK;
rc = CIFSSMBPosixLock(xid, tcon, netfid, current->tgid,
flock->fl_start, length, NULL,
posix_lock_type, wait_flag);
goto out;
}
if (lock) {
struct cifsLockInfo *lock;
lock = cifs_lock_init(flock->fl_start, length, type);
if (!lock)
return -ENOMEM;
rc = cifs_lock_add_if(cfile, lock, wait_flag);
if (rc < 0)
kfree(lock);
if (rc <= 0)
goto out;
rc = cifs_mandatory_lock(xid, cfile, flock->fl_start, length,
type, 1, 0, wait_flag);
if (rc) {
kfree(lock);
goto out;
}
cifs_lock_add(cfile, lock);
} else if (unlock)
rc = cifs_unlock_range(cfile, flock, xid);
out:
if (flock->fl_flags & FL_POSIX)
posix_lock_file_wait(file, flock);
return rc;
}
int cifs_lock(struct file *file, int cmd, struct file_lock *flock)
{
int rc, xid;
int lock = 0, unlock = 0;
bool wait_flag = false;
bool posix_lck = false;
struct cifs_sb_info *cifs_sb;
struct cifs_tcon *tcon;
struct cifsInodeInfo *cinode;
struct cifsFileInfo *cfile;
__u16 netfid;
__u32 type;
rc = -EACCES;
xid = get_xid();
cFYI(1, "Lock parm: 0x%x flockflags: 0x%x flocktype: 0x%x start: %lld "
"end: %lld", cmd, flock->fl_flags, flock->fl_type,
flock->fl_start, flock->fl_end);
cfile = (struct cifsFileInfo *)file->private_data;
tcon = tlink_tcon(cfile->tlink);
cifs_read_flock(flock, &type, &lock, &unlock, &wait_flag,
tcon->ses->server);
cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
netfid = cfile->fid.netfid;
cinode = CIFS_I(file->f_path.dentry->d_inode);
if (cap_unix(tcon->ses) &&
(CIFS_UNIX_FCNTL_CAP & le64_to_cpu(tcon->fsUnixInfo.Capability)) &&
((cifs_sb->mnt_cifs_flags & CIFS_MOUNT_NOPOSIXBRL) == 0))
posix_lck = true;
/*
* BB add code here to normalize offset and length to account for
* negative length which we can not accept over the wire.
*/
if (IS_GETLK(cmd)) {
rc = cifs_getlk(file, flock, type, wait_flag, posix_lck, xid);
free_xid(xid);
return rc;
}
if (!lock && !unlock) {
/*
* if no lock or unlock then nothing to do since we do not
* know what it is
*/
free_xid(xid);
return -EOPNOTSUPP;
}
rc = cifs_setlk(file, flock, type, wait_flag, posix_lck, lock, unlock,
xid);
free_xid(xid);
return rc;
}
/*
* update the file size (if needed) after a write. Should be called with
* the inode->i_lock held
*/
void
cifs_update_eof(struct cifsInodeInfo *cifsi, loff_t offset,
unsigned int bytes_written)
{
loff_t end_of_write = offset + bytes_written;
if (end_of_write > cifsi->server_eof)
cifsi->server_eof = end_of_write;
}
static ssize_t cifs_write(struct cifsFileInfo *open_file, __u32 pid,
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 cifs_tcon *pTcon;
unsigned int xid;
struct dentry *dentry = open_file->dentry;
struct cifsInodeInfo *cifsi = CIFS_I(dentry->d_inode);
struct cifs_io_parms io_parms;
cifs_sb = CIFS_SB(dentry->d_sb);
cFYI(1, "write %zd bytes to offset %lld of %s", write_size,
*poffset, dentry->d_name.name);
pTcon = tlink_tcon(open_file->tlink);
xid = get_xid();
for (total_written = 0; write_size > total_written;
total_written += bytes_written) {
rc = -EAGAIN;
while (rc == -EAGAIN) {
struct kvec iov[2];
unsigned int len;
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 */
rc = cifs_reopen_file(open_file, false);
if (rc != 0)
break;
}
len = min((size_t)cifs_sb->wsize,
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;
io_parms.netfid = open_file->fid.netfid;
io_parms.pid = pid;
io_parms.tcon = pTcon;
io_parms.offset = *poffset;
io_parms.length = len;
rc = CIFSSMBWrite2(xid, &io_parms, &bytes_written, iov,
1, 0);
}
if (rc || (bytes_written == 0)) {
if (total_written)
break;
else {
free_xid(xid);
return rc;
}
} else {
spin_lock(&dentry->d_inode->i_lock);
cifs_update_eof(cifsi, *poffset, bytes_written);
spin_unlock(&dentry->d_inode->i_lock);
*poffset += bytes_written;
}
}
cifs_stats_bytes_written(pTcon, total_written);
if (total_written > 0) {
spin_lock(&dentry->d_inode->i_lock);
if (*poffset > dentry->d_inode->i_size)
i_size_write(dentry->d_inode, *poffset);
spin_unlock(&dentry->d_inode->i_lock);
}
mark_inode_dirty_sync(dentry->d_inode);
free_xid(xid);
return total_written;
}
struct cifsFileInfo *find_readable_file(struct cifsInodeInfo *cifs_inode,
bool fsuid_only)
{
struct cifsFileInfo *open_file = NULL;
struct cifs_sb_info *cifs_sb = CIFS_SB(cifs_inode->vfs_inode.i_sb);
/* only filter by fsuid on multiuser mounts */
if (!(cifs_sb->mnt_cifs_flags & CIFS_MOUNT_MULTIUSER))
fsuid_only = false;
spin_lock(&cifs_file_list_lock);
/* 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 (fsuid_only && open_file->uid != current_fsuid())
continue;
if (OPEN_FMODE(open_file->f_flags) & FMODE_READ) {
if (!open_file->invalidHandle) {
/* found a good file */
/* lock it so it will not be closed on us */
cifsFileInfo_get_locked(open_file);
spin_unlock(&cifs_file_list_lock);
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 */
}
spin_unlock(&cifs_file_list_lock);
return NULL;
}
struct cifsFileInfo *find_writable_file(struct cifsInodeInfo *cifs_inode,
bool fsuid_only)
{
struct cifsFileInfo *open_file, *inv_file = NULL;
struct cifs_sb_info *cifs_sb;
bool any_available = false;
int rc;
unsigned int refind = 0;
/* 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 */
if (cifs_inode == NULL) {
cERROR(1, "Null inode passed to cifs_writeable_file");
dump_stack();
return NULL;
}
cifs_sb = CIFS_SB(cifs_inode->vfs_inode.i_sb);
/* only filter by fsuid on multiuser mounts */
if (!(cifs_sb->mnt_cifs_flags & CIFS_MOUNT_MULTIUSER))
fsuid_only = false;
spin_lock(&cifs_file_list_lock);
refind_writable:
if (refind > MAX_REOPEN_ATT) {
spin_unlock(&cifs_file_list_lock);
return NULL;
}
list_for_each_entry(open_file, &cifs_inode->openFileList, flist) {
if (!any_available && open_file->pid != current->tgid)
continue;
if (fsuid_only && open_file->uid != current_fsuid())
continue;
if (OPEN_FMODE(open_file->f_flags) & FMODE_WRITE) {
if (!open_file->invalidHandle) {
/* found a good writable file */
cifsFileInfo_get_locked(open_file);
spin_unlock(&cifs_file_list_lock);
return open_file;
} else {
if (!inv_file)
inv_file = open_file;
}
}
}
/* couldn't find useable FH with same pid, try any available */
if (!any_available) {
any_available = true;
goto refind_writable;
}
if (inv_file) {
any_available = false;
cifsFileInfo_get_locked(inv_file);
}
spin_unlock(&cifs_file_list_lock);
if (inv_file) {
rc = cifs_reopen_file(inv_file, false);
if (!rc)
return inv_file;
else {
spin_lock(&cifs_file_list_lock);
list_move_tail(&inv_file->flist,
&cifs_inode->openFileList);
spin_unlock(&cifs_file_list_lock);
cifsFileInfo_put(inv_file);
spin_lock(&cifs_file_list_lock);
++refind;
goto refind_writable;
}
}
return NULL;
}
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 inode *inode;
struct cifsFileInfo *open_file;
if (!mapping || !mapping->host)
return -EFAULT;
inode = page->mapping->host;
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)
to = (unsigned)(mapping->host->i_size - offset);
open_file = find_writable_file(CIFS_I(mapping->host), false);
if (open_file) {
bytes_written = cifs_write(open_file, open_file->pid,
write_data, to - from, &offset);
cifsFileInfo_put(open_file);
/* Does mm or vfs already set times? */
inode->i_atime = inode->i_mtime = current_fs_time(inode->i_sb);
if ((bytes_written > 0) && (offset))
rc = 0;
else if (bytes_written < 0)
rc = bytes_written;
} else {
cFYI(1, "No writeable filehandles for inode");
rc = -EIO;
}
kunmap(page);
return rc;
}
/*
* Marshal up the iov array, reserving the first one for the header. Also,
* set wdata->bytes.
*/
static void
cifs_writepages_marshal_iov(struct kvec *iov, struct cifs_writedata *wdata)
{
int i;
struct inode *inode = wdata->cfile->dentry->d_inode;
loff_t size = i_size_read(inode);
/* marshal up the pages into iov array */
wdata->bytes = 0;
for (i = 0; i < wdata->nr_pages; i++) {
iov[i + 1].iov_len = min(size - page_offset(wdata->pages[i]),
(loff_t)PAGE_CACHE_SIZE);
iov[i + 1].iov_base = kmap(wdata->pages[i]);
wdata->bytes += iov[i + 1].iov_len;
}
}
static int cifs_writepages(struct address_space *mapping,
struct writeback_control *wbc)
{
struct cifs_sb_info *cifs_sb = CIFS_SB(mapping->host->i_sb);
bool done = false, scanned = false, range_whole = false;
pgoff_t end, index;
struct cifs_writedata *wdata;
struct page *page;
int rc = 0;
/*
* If wsize is smaller than 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);
[PATCH] writeback: fix range handling When a writeback_control's `start' and `end' fields are used to indicate a one-byte-range starting at file offset zero, the required values of .start=0,.end=0 mean that the ->writepages() implementation has no way of telling that it is being asked to perform a range request. Because we're currently overloading (start == 0 && end == 0) to mean "this is not a write-a-range request". To make all this sane, the patch changes range of writeback_control. So caller does: If it is calling ->writepages() to write pages, it sets range (range_start/end or range_cyclic) always. And if range_cyclic is true, ->writepages() thinks the range is cyclic, otherwise it just uses range_start and range_end. This patch does, - Add LLONG_MAX, LLONG_MIN, ULLONG_MAX to include/linux/kernel.h -1 is usually ok for range_end (type is long long). But, if someone did, range_end += val; range_end is "val - 1" u64val = range_end >> bits; u64val is "~(0ULL)" or something, they are wrong. So, this adds LLONG_MAX to avoid nasty things, and uses LLONG_MAX for range_end. - All callers of ->writepages() sets range_start/end or range_cyclic. - Fix updates of ->writeback_index. It seems already bit strange. If it starts at 0 and ended by check of nr_to_write, this last index may reduce chance to scan end of file. So, this updates ->writeback_index only if range_cyclic is true or whole-file is scanned. Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> Cc: Nathan Scott <nathans@sgi.com> Cc: Anton Altaparmakov <aia21@cantab.net> Cc: Steven French <sfrench@us.ibm.com> Cc: "Vladimir V. Saveliev" <vs@namesys.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 09:03:26 +00:00
if (wbc->range_cyclic) {
index = mapping->writeback_index; /* Start from prev offset */
[PATCH] writeback: fix range handling When a writeback_control's `start' and `end' fields are used to indicate a one-byte-range starting at file offset zero, the required values of .start=0,.end=0 mean that the ->writepages() implementation has no way of telling that it is being asked to perform a range request. Because we're currently overloading (start == 0 && end == 0) to mean "this is not a write-a-range request". To make all this sane, the patch changes range of writeback_control. So caller does: If it is calling ->writepages() to write pages, it sets range (range_start/end or range_cyclic) always. And if range_cyclic is true, ->writepages() thinks the range is cyclic, otherwise it just uses range_start and range_end. This patch does, - Add LLONG_MAX, LLONG_MIN, ULLONG_MAX to include/linux/kernel.h -1 is usually ok for range_end (type is long long). But, if someone did, range_end += val; range_end is "val - 1" u64val = range_end >> bits; u64val is "~(0ULL)" or something, they are wrong. So, this adds LLONG_MAX to avoid nasty things, and uses LLONG_MAX for range_end. - All callers of ->writepages() sets range_start/end or range_cyclic. - Fix updates of ->writeback_index. It seems already bit strange. If it starts at 0 and ended by check of nr_to_write, this last index may reduce chance to scan end of file. So, this updates ->writeback_index only if range_cyclic is true or whole-file is scanned. Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> Cc: Nathan Scott <nathans@sgi.com> Cc: Anton Altaparmakov <aia21@cantab.net> Cc: Steven French <sfrench@us.ibm.com> Cc: "Vladimir V. Saveliev" <vs@namesys.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 09:03:26 +00:00
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 = true;
scanned = true;
}
retry:
while (!done && index <= end) {
unsigned int i, nr_pages, found_pages;
pgoff_t next = 0, tofind;
struct page **pages;
tofind = min((cifs_sb->wsize / PAGE_CACHE_SIZE) - 1,
end - index) + 1;
wdata = cifs_writedata_alloc((unsigned int)tofind,
cifs_writev_complete);
if (!wdata) {
rc = -ENOMEM;
break;
}
/*
* find_get_pages_tag seems to return a max of 256 on each
* iteration, so we must call it several times in order to
* fill the array or the wsize is effectively limited to
* 256 * PAGE_CACHE_SIZE.
*/
found_pages = 0;
pages = wdata->pages;
do {
nr_pages = find_get_pages_tag(mapping, &index,
PAGECACHE_TAG_DIRTY,
tofind, pages);
found_pages += nr_pages;
tofind -= nr_pages;
pages += nr_pages;
} while (nr_pages && tofind && index <= end);
if (found_pages == 0) {
kref_put(&wdata->refcount, cifs_writedata_release);
break;
}
nr_pages = 0;
for (i = 0; i < found_pages; i++) {
page = wdata->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 (nr_pages == 0)
lock_page(page);
else if (!trylock_page(page))
break;
if (unlikely(page->mapping != mapping)) {
unlock_page(page);
break;
}
[PATCH] writeback: fix range handling When a writeback_control's `start' and `end' fields are used to indicate a one-byte-range starting at file offset zero, the required values of .start=0,.end=0 mean that the ->writepages() implementation has no way of telling that it is being asked to perform a range request. Because we're currently overloading (start == 0 && end == 0) to mean "this is not a write-a-range request". To make all this sane, the patch changes range of writeback_control. So caller does: If it is calling ->writepages() to write pages, it sets range (range_start/end or range_cyclic) always. And if range_cyclic is true, ->writepages() thinks the range is cyclic, otherwise it just uses range_start and range_end. This patch does, - Add LLONG_MAX, LLONG_MIN, ULLONG_MAX to include/linux/kernel.h -1 is usually ok for range_end (type is long long). But, if someone did, range_end += val; range_end is "val - 1" u64val = range_end >> bits; u64val is "~(0ULL)" or something, they are wrong. So, this adds LLONG_MAX to avoid nasty things, and uses LLONG_MAX for range_end. - All callers of ->writepages() sets range_start/end or range_cyclic. - Fix updates of ->writeback_index. It seems already bit strange. If it starts at 0 and ended by check of nr_to_write, this last index may reduce chance to scan end of file. So, this updates ->writeback_index only if range_cyclic is true or whole-file is scanned. Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> Cc: Nathan Scott <nathans@sgi.com> Cc: Anton Altaparmakov <aia21@cantab.net> Cc: Steven French <sfrench@us.ibm.com> Cc: "Vladimir V. Saveliev" <vs@namesys.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 09:03:26 +00:00
if (!wbc->range_cyclic && page->index > end) {
done = true;
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) ||
!clear_page_dirty_for_io(page)) {
unlock_page(page);
break;
}
/*
* This actually clears the dirty bit in the radix tree.
* See cifs_writepage() for more commentary.
*/
set_page_writeback(page);
if (page_offset(page) >= mapping->host->i_size) {
done = true;
unlock_page(page);
end_page_writeback(page);
break;
}
wdata->pages[i] = page;
next = page->index + 1;
++nr_pages;
}
/* reset index to refind any pages skipped */
if (nr_pages == 0)
index = wdata->pages[0]->index + 1;
/* put any pages we aren't going to use */
for (i = nr_pages; i < found_pages; i++) {
page_cache_release(wdata->pages[i]);
wdata->pages[i] = NULL;
}
/* nothing to write? */
if (nr_pages == 0) {
kref_put(&wdata->refcount, cifs_writedata_release);
continue;
}
wdata->sync_mode = wbc->sync_mode;
wdata->nr_pages = nr_pages;
wdata->offset = page_offset(wdata->pages[0]);
wdata->marshal_iov = cifs_writepages_marshal_iov;
do {
if (wdata->cfile != NULL)
cifsFileInfo_put(wdata->cfile);
wdata->cfile = find_writable_file(CIFS_I(mapping->host),
false);
if (!wdata->cfile) {
cERROR(1, "No writable handles for inode");
rc = -EBADF;
break;
}
wdata->pid = wdata->cfile->pid;
rc = cifs_async_writev(wdata);
} while (wbc->sync_mode == WB_SYNC_ALL && rc == -EAGAIN);
for (i = 0; i < nr_pages; ++i)
unlock_page(wdata->pages[i]);
/* send failure -- clean up the mess */
if (rc != 0) {
for (i = 0; i < nr_pages; ++i) {
if (rc == -EAGAIN)
redirty_page_for_writepage(wbc,
wdata->pages[i]);
else
SetPageError(wdata->pages[i]);
end_page_writeback(wdata->pages[i]);
page_cache_release(wdata->pages[i]);
}
if (rc != -EAGAIN)
mapping_set_error(mapping, rc);
}
kref_put(&wdata->refcount, cifs_writedata_release);
wbc->nr_to_write -= nr_pages;
if (wbc->nr_to_write <= 0)
done = true;
index = next;
}
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 = true;
index = 0;
goto retry;
}
[PATCH] writeback: fix range handling When a writeback_control's `start' and `end' fields are used to indicate a one-byte-range starting at file offset zero, the required values of .start=0,.end=0 mean that the ->writepages() implementation has no way of telling that it is being asked to perform a range request. Because we're currently overloading (start == 0 && end == 0) to mean "this is not a write-a-range request". To make all this sane, the patch changes range of writeback_control. So caller does: If it is calling ->writepages() to write pages, it sets range (range_start/end or range_cyclic) always. And if range_cyclic is true, ->writepages() thinks the range is cyclic, otherwise it just uses range_start and range_end. This patch does, - Add LLONG_MAX, LLONG_MIN, ULLONG_MAX to include/linux/kernel.h -1 is usually ok for range_end (type is long long). But, if someone did, range_end += val; range_end is "val - 1" u64val = range_end >> bits; u64val is "~(0ULL)" or something, they are wrong. So, this adds LLONG_MAX to avoid nasty things, and uses LLONG_MAX for range_end. - All callers of ->writepages() sets range_start/end or range_cyclic. - Fix updates of ->writeback_index. It seems already bit strange. If it starts at 0 and ended by check of nr_to_write, this last index may reduce chance to scan end of file. So, this updates ->writeback_index only if range_cyclic is true or whole-file is scanned. Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> Cc: Nathan Scott <nathans@sgi.com> Cc: Anton Altaparmakov <aia21@cantab.net> Cc: Steven French <sfrench@us.ibm.com> Cc: "Vladimir V. Saveliev" <vs@namesys.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 09:03:26 +00:00
if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0))
mapping->writeback_index = index;
return rc;
}
static int
cifs_writepage_locked(struct page *page, struct writeback_control *wbc)
{
int rc;
unsigned int xid;
xid = get_xid();
/* BB add check for wbc flags */
page_cache_get(page);
if (!PageUptodate(page))
cFYI(1, "ppw - page not up to date");
/*
* 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.
*/
set_page_writeback(page);
retry_write:
rc = cifs_partialpagewrite(page, 0, PAGE_CACHE_SIZE);
if (rc == -EAGAIN && wbc->sync_mode == WB_SYNC_ALL)
goto retry_write;
else if (rc == -EAGAIN)
redirty_page_for_writepage(wbc, page);
else if (rc != 0)
SetPageError(page);
else
SetPageUptodate(page);
end_page_writeback(page);
page_cache_release(page);
free_xid(xid);
return rc;
}
static int cifs_writepage(struct page *page, struct writeback_control *wbc)
{
int rc = cifs_writepage_locked(page, wbc);
unlock_page(page);
return rc;
}
static int cifs_write_end(struct file *file, struct address_space *mapping,
loff_t pos, unsigned len, unsigned copied,
struct page *page, void *fsdata)
{
int rc;
struct inode *inode = mapping->host;
struct cifsFileInfo *cfile = file->private_data;
struct cifs_sb_info *cifs_sb = CIFS_SB(cfile->dentry->d_sb);
__u32 pid;
if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_RWPIDFORWARD)
pid = cfile->pid;
else
pid = current->tgid;
cFYI(1, "write_end for page %p from pos %lld with %d bytes",
page, pos, copied);
if (PageChecked(page)) {
if (copied == len)
SetPageUptodate(page);
ClearPageChecked(page);
} else if (!PageUptodate(page) && copied == PAGE_CACHE_SIZE)
SetPageUptodate(page);
if (!PageUptodate(page)) {
char *page_data;
unsigned offset = pos & (PAGE_CACHE_SIZE - 1);
unsigned int xid;
xid = get_xid();
/* 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(cfile, pid, page_data + offset, copied, &pos);
/* if (rc < 0) should we set writebehind rc? */
kunmap(page);
free_xid(xid);
} else {
rc = copied;
pos += copied;
set_page_dirty(page);
}
if (rc > 0) {
spin_lock(&inode->i_lock);
if (pos > inode->i_size)
i_size_write(inode, pos);
spin_unlock(&inode->i_lock);
}
unlock_page(page);
page_cache_release(page);
return rc;
}
int cifs_strict_fsync(struct file *file, loff_t start, loff_t end,
int datasync)
{
unsigned int xid;
int rc = 0;
struct cifs_tcon *tcon;
struct TCP_Server_Info *server;
struct cifsFileInfo *smbfile = file->private_data;
struct inode *inode = file->f_path.dentry->d_inode;
struct cifs_sb_info *cifs_sb = CIFS_SB(inode->i_sb);
rc = filemap_write_and_wait_range(inode->i_mapping, start, end);
if (rc)
return rc;
mutex_lock(&inode->i_mutex);
xid = get_xid();
cFYI(1, "Sync file - name: %s datasync: 0x%x",
file->f_path.dentry->d_name.name, datasync);
if (!CIFS_I(inode)->clientCanCacheRead) {
rc = cifs_invalidate_mapping(inode);
if (rc) {
cFYI(1, "rc: %d during invalidate phase", rc);
rc = 0; /* don't care about it in fsync */
}
}
tcon = tlink_tcon(smbfile->tlink);
if (!(cifs_sb->mnt_cifs_flags & CIFS_MOUNT_NOSSYNC)) {
server = tcon->ses->server;
if (server->ops->flush)
rc = server->ops->flush(xid, tcon, &smbfile->fid);
else
rc = -ENOSYS;
}
free_xid(xid);
mutex_unlock(&inode->i_mutex);
return rc;
}
int cifs_fsync(struct file *file, loff_t start, loff_t end, int datasync)
{
unsigned int xid;
int rc = 0;
struct cifs_tcon *tcon;
struct TCP_Server_Info *server;
struct cifsFileInfo *smbfile = file->private_data;
struct cifs_sb_info *cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
struct inode *inode = file->f_mapping->host;
rc = filemap_write_and_wait_range(inode->i_mapping, start, end);
if (rc)
return rc;
mutex_lock(&inode->i_mutex);
xid = get_xid();
cFYI(1, "Sync file - name: %s datasync: 0x%x",
file->f_path.dentry->d_name.name, datasync);
tcon = tlink_tcon(smbfile->tlink);
if (!(cifs_sb->mnt_cifs_flags & CIFS_MOUNT_NOSSYNC)) {
server = tcon->ses->server;
if (server->ops->flush)
rc = server->ops->flush(xid, tcon, &smbfile->fid);
else
rc = -ENOSYS;
}
free_xid(xid);
mutex_unlock(&inode->i_mutex);
return rc;
}
/*
* As file closes, flush all cached write data for this inode checking
* for write behind errors.
*/
int cifs_flush(struct file *file, fl_owner_t id)
{
struct inode *inode = file->f_path.dentry->d_inode;
int rc = 0;
if (file->f_mode & FMODE_WRITE)
rc = filemap_write_and_wait(inode->i_mapping);
cFYI(1, "Flush inode %p file %p rc %d", inode, file, rc);
return rc;
}
static int
cifs_write_allocate_pages(struct page **pages, unsigned long num_pages)
{
int rc = 0;
unsigned long i;
for (i = 0; i < num_pages; i++) {
pages[i] = alloc_page(GFP_KERNEL|__GFP_HIGHMEM);
if (!pages[i]) {
/*
* save number of pages we have already allocated and
* return with ENOMEM error
*/
num_pages = i;
rc = -ENOMEM;
break;
}
}
if (rc) {
for (i = 0; i < num_pages; i++)
put_page(pages[i]);
}
return rc;
}
static inline
size_t get_numpages(const size_t wsize, const size_t len, size_t *cur_len)
{
size_t num_pages;
size_t clen;
clen = min_t(const size_t, len, wsize);
num_pages = DIV_ROUND_UP(clen, PAGE_SIZE);
if (cur_len)
*cur_len = clen;
return num_pages;
}
static void
cifs_uncached_marshal_iov(struct kvec *iov, struct cifs_writedata *wdata)
{
int i;
size_t bytes = wdata->bytes;
/* marshal up the pages into iov array */
for (i = 0; i < wdata->nr_pages; i++) {
iov[i + 1].iov_len = min_t(size_t, bytes, PAGE_SIZE);
iov[i + 1].iov_base = kmap(wdata->pages[i]);
bytes -= iov[i + 1].iov_len;
}
}
static void
cifs_uncached_writev_complete(struct work_struct *work)
{
int i;
struct cifs_writedata *wdata = container_of(work,
struct cifs_writedata, work);
struct inode *inode = wdata->cfile->dentry->d_inode;
struct cifsInodeInfo *cifsi = CIFS_I(inode);
spin_lock(&inode->i_lock);
cifs_update_eof(cifsi, wdata->offset, wdata->bytes);
if (cifsi->server_eof > inode->i_size)
i_size_write(inode, cifsi->server_eof);
spin_unlock(&inode->i_lock);
complete(&wdata->done);
if (wdata->result != -EAGAIN) {
for (i = 0; i < wdata->nr_pages; i++)
put_page(wdata->pages[i]);
}
kref_put(&wdata->refcount, cifs_writedata_release);
}
/* attempt to send write to server, retry on any -EAGAIN errors */
static int
cifs_uncached_retry_writev(struct cifs_writedata *wdata)
{
int rc;
do {
if (wdata->cfile->invalidHandle) {
rc = cifs_reopen_file(wdata->cfile, false);
if (rc != 0)
continue;
}
rc = cifs_async_writev(wdata);
} while (rc == -EAGAIN);
return rc;
}
static ssize_t
cifs_iovec_write(struct file *file, const struct iovec *iov,
unsigned long nr_segs, loff_t *poffset)
{
unsigned long nr_pages, i;
size_t copied, len, cur_len;
ssize_t total_written = 0;
loff_t offset;
struct iov_iter it;
struct cifsFileInfo *open_file;
struct cifs_tcon *tcon;
struct cifs_sb_info *cifs_sb;
struct cifs_writedata *wdata, *tmp;
struct list_head wdata_list;
int rc;
pid_t pid;
len = iov_length(iov, nr_segs);
if (!len)
return 0;
rc = generic_write_checks(file, poffset, &len, 0);
if (rc)
return rc;
INIT_LIST_HEAD(&wdata_list);
cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
open_file = file->private_data;
tcon = tlink_tcon(open_file->tlink);
offset = *poffset;
if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_RWPIDFORWARD)
pid = open_file->pid;
else
pid = current->tgid;
iov_iter_init(&it, iov, nr_segs, len, 0);
do {
size_t save_len;
nr_pages = get_numpages(cifs_sb->wsize, len, &cur_len);
wdata = cifs_writedata_alloc(nr_pages,
cifs_uncached_writev_complete);
if (!wdata) {
rc = -ENOMEM;
break;
}
rc = cifs_write_allocate_pages(wdata->pages, nr_pages);
if (rc) {
kfree(wdata);
break;
}
save_len = cur_len;
for (i = 0; i < nr_pages; i++) {
copied = min_t(const size_t, cur_len, PAGE_SIZE);
copied = iov_iter_copy_from_user(wdata->pages[i], &it,
0, copied);
cur_len -= copied;
iov_iter_advance(&it, copied);
}
cur_len = save_len - cur_len;
wdata->sync_mode = WB_SYNC_ALL;
wdata->nr_pages = nr_pages;
wdata->offset = (__u64)offset;
wdata->cfile = cifsFileInfo_get(open_file);
wdata->pid = pid;
wdata->bytes = cur_len;
wdata->marshal_iov = cifs_uncached_marshal_iov;
rc = cifs_uncached_retry_writev(wdata);
if (rc) {
kref_put(&wdata->refcount, cifs_writedata_release);
break;
}
list_add_tail(&wdata->list, &wdata_list);
offset += cur_len;
len -= cur_len;
} while (len > 0);
/*
* If at least one write was successfully sent, then discard any rc
* value from the later writes. If the other write succeeds, then
* we'll end up returning whatever was written. If it fails, then
* we'll get a new rc value from that.
*/
if (!list_empty(&wdata_list))
rc = 0;
/*
* Wait for and collect replies for any successful sends in order of
* increasing offset. Once an error is hit or we get a fatal signal
* while waiting, then return without waiting for any more replies.
*/
restart_loop:
list_for_each_entry_safe(wdata, tmp, &wdata_list, list) {
if (!rc) {
/* FIXME: freezable too? */
rc = wait_for_completion_killable(&wdata->done);
if (rc)
rc = -EINTR;
else if (wdata->result)
rc = wdata->result;
else
total_written += wdata->bytes;
/* resend call if it's a retryable error */
if (rc == -EAGAIN) {
rc = cifs_uncached_retry_writev(wdata);
goto restart_loop;
}
}
list_del_init(&wdata->list);
kref_put(&wdata->refcount, cifs_writedata_release);
}
if (total_written > 0)
*poffset += total_written;
cifs_stats_bytes_written(tcon, total_written);
return total_written ? total_written : (ssize_t)rc;
}
ssize_t cifs_user_writev(struct kiocb *iocb, const struct iovec *iov,
unsigned long nr_segs, loff_t pos)
{
ssize_t written;
struct inode *inode;
inode = iocb->ki_filp->f_path.dentry->d_inode;
/*
* BB - optimize the way when signing is disabled. We can drop this
* extra memory-to-memory copying and use iovec buffers for constructing
* write request.
*/
written = cifs_iovec_write(iocb->ki_filp, iov, nr_segs, &pos);
if (written > 0) {
CIFS_I(inode)->invalid_mapping = true;
iocb->ki_pos = pos;
}
return written;
}
ssize_t cifs_strict_writev(struct kiocb *iocb, const struct iovec *iov,
unsigned long nr_segs, loff_t pos)
{
struct inode *inode;
inode = iocb->ki_filp->f_path.dentry->d_inode;
if (CIFS_I(inode)->clientCanCacheAll)
return generic_file_aio_write(iocb, iov, nr_segs, pos);
/*
* In strict cache mode we need to write the data to the server exactly
* from the pos to pos+len-1 rather than flush all affected pages
* because it may cause a error with mandatory locks on these pages but
* not on the region from pos to ppos+len-1.
*/
return cifs_user_writev(iocb, iov, nr_segs, pos);
}
static struct cifs_readdata *
cifs_readdata_alloc(unsigned int nr_vecs, work_func_t complete)
{
struct cifs_readdata *rdata;
rdata = kzalloc(sizeof(*rdata) +
sizeof(struct kvec) * nr_vecs, GFP_KERNEL);
if (rdata != NULL) {
kref_init(&rdata->refcount);
INIT_LIST_HEAD(&rdata->list);
init_completion(&rdata->done);
INIT_WORK(&rdata->work, complete);
INIT_LIST_HEAD(&rdata->pages);
}
return rdata;
}
void
cifs_readdata_release(struct kref *refcount)
{
struct cifs_readdata *rdata = container_of(refcount,
struct cifs_readdata, refcount);
if (rdata->cfile)
cifsFileInfo_put(rdata->cfile);
kfree(rdata);
}
static int
cifs_read_allocate_pages(struct list_head *list, unsigned int npages)
{
int rc = 0;
struct page *page, *tpage;
unsigned int i;
for (i = 0; i < npages; i++) {
page = alloc_page(GFP_KERNEL|__GFP_HIGHMEM);
if (!page) {
rc = -ENOMEM;
break;
}
list_add(&page->lru, list);
}
if (rc) {
list_for_each_entry_safe(page, tpage, list, lru) {
list_del(&page->lru);
put_page(page);
}
}
return rc;
}
static void
cifs_uncached_readdata_release(struct kref *refcount)
{
struct page *page, *tpage;
struct cifs_readdata *rdata = container_of(refcount,
struct cifs_readdata, refcount);
list_for_each_entry_safe(page, tpage, &rdata->pages, lru) {
list_del(&page->lru);
put_page(page);
}
cifs_readdata_release(refcount);
}
static int
cifs_retry_async_readv(struct cifs_readdata *rdata)
{
int rc;
struct TCP_Server_Info *server;
server = tlink_tcon(rdata->cfile->tlink)->ses->server;
do {
if (rdata->cfile->invalidHandle) {
rc = cifs_reopen_file(rdata->cfile, true);
if (rc != 0)
continue;
}
rc = server->ops->async_readv(rdata);
} while (rc == -EAGAIN);
return rc;
}
/**
* cifs_readdata_to_iov - copy data from pages in response to an iovec
* @rdata: the readdata response with list of pages holding data
* @iov: vector in which we should copy the data
* @nr_segs: number of segments in vector
* @offset: offset into file of the first iovec
* @copied: used to return the amount of data copied to the iov
*
* This function copies data from a list of pages in a readdata response into
* an array of iovecs. It will first calculate where the data should go
* based on the info in the readdata and then copy the data into that spot.
*/
static ssize_t
cifs_readdata_to_iov(struct cifs_readdata *rdata, const struct iovec *iov,
unsigned long nr_segs, loff_t offset, ssize_t *copied)
{
int rc = 0;
struct iov_iter ii;
size_t pos = rdata->offset - offset;
struct page *page, *tpage;
ssize_t remaining = rdata->bytes;
unsigned char *pdata;
/* set up iov_iter and advance to the correct offset */
iov_iter_init(&ii, iov, nr_segs, iov_length(iov, nr_segs), 0);
iov_iter_advance(&ii, pos);
*copied = 0;
list_for_each_entry_safe(page, tpage, &rdata->pages, lru) {
ssize_t copy;
/* copy a whole page or whatever's left */
copy = min_t(ssize_t, remaining, PAGE_SIZE);
/* ...but limit it to whatever space is left in the iov */
copy = min_t(ssize_t, copy, iov_iter_count(&ii));
/* go while there's data to be copied and no errors */
if (copy && !rc) {
pdata = kmap(page);
rc = memcpy_toiovecend(ii.iov, pdata, ii.iov_offset,
(int)copy);
kunmap(page);
if (!rc) {
*copied += copy;
remaining -= copy;
iov_iter_advance(&ii, copy);
}
}
list_del(&page->lru);
put_page(page);
}
return rc;
}
static void
cifs_uncached_readv_complete(struct work_struct *work)
{
struct cifs_readdata *rdata = container_of(work,
struct cifs_readdata, work);
/* if the result is non-zero then the pages weren't kmapped */
if (rdata->result == 0) {
struct page *page;
list_for_each_entry(page, &rdata->pages, lru)
kunmap(page);
}
complete(&rdata->done);
kref_put(&rdata->refcount, cifs_uncached_readdata_release);
}
static int
cifs_uncached_read_marshal_iov(struct cifs_readdata *rdata,
unsigned int remaining)
{
int len = 0;
struct page *page, *tpage;
rdata->nr_iov = 1;
list_for_each_entry_safe(page, tpage, &rdata->pages, lru) {
if (remaining >= PAGE_SIZE) {
/* enough data to fill the page */
rdata->iov[rdata->nr_iov].iov_base = kmap(page);
rdata->iov[rdata->nr_iov].iov_len = PAGE_SIZE;
cFYI(1, "%u: idx=%lu iov_base=%p iov_len=%zu",
rdata->nr_iov, page->index,
rdata->iov[rdata->nr_iov].iov_base,
rdata->iov[rdata->nr_iov].iov_len);
++rdata->nr_iov;
len += PAGE_SIZE;
remaining -= PAGE_SIZE;
} else if (remaining > 0) {
/* enough for partial page, fill and zero the rest */
rdata->iov[rdata->nr_iov].iov_base = kmap(page);
rdata->iov[rdata->nr_iov].iov_len = remaining;
cFYI(1, "%u: idx=%lu iov_base=%p iov_len=%zu",
rdata->nr_iov, page->index,
rdata->iov[rdata->nr_iov].iov_base,
rdata->iov[rdata->nr_iov].iov_len);
memset(rdata->iov[rdata->nr_iov].iov_base + remaining,
'\0', PAGE_SIZE - remaining);
++rdata->nr_iov;
len += remaining;
remaining = 0;
} else {
/* no need to hold page hostage */
list_del(&page->lru);
put_page(page);
}
}
return len;
}
static ssize_t
cifs_iovec_read(struct file *file, const struct iovec *iov,
unsigned long nr_segs, loff_t *poffset)
{
ssize_t rc;
size_t len, cur_len;
ssize_t total_read = 0;
loff_t offset = *poffset;
unsigned int npages;
struct cifs_sb_info *cifs_sb;
struct cifs_tcon *tcon;
struct cifsFileInfo *open_file;
struct cifs_readdata *rdata, *tmp;
struct list_head rdata_list;
pid_t pid;
if (!nr_segs)
return 0;
len = iov_length(iov, nr_segs);
if (!len)
return 0;
INIT_LIST_HEAD(&rdata_list);
cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
open_file = file->private_data;
tcon = tlink_tcon(open_file->tlink);
if (!tcon->ses->server->ops->async_readv)
return -ENOSYS;
if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_RWPIDFORWARD)
pid = open_file->pid;
else
pid = current->tgid;
if ((file->f_flags & O_ACCMODE) == O_WRONLY)
cFYI(1, "attempting read on write only file instance");
do {
cur_len = min_t(const size_t, len - total_read, cifs_sb->rsize);
npages = DIV_ROUND_UP(cur_len, PAGE_SIZE);
/* allocate a readdata struct */
rdata = cifs_readdata_alloc(npages,
cifs_uncached_readv_complete);
if (!rdata) {
rc = -ENOMEM;
goto error;
}
rc = cifs_read_allocate_pages(&rdata->pages, npages);
if (rc)
goto error;
rdata->cfile = cifsFileInfo_get(open_file);
rdata->offset = offset;
rdata->bytes = cur_len;
rdata->pid = pid;
rdata->marshal_iov = cifs_uncached_read_marshal_iov;
rc = cifs_retry_async_readv(rdata);
error:
if (rc) {
kref_put(&rdata->refcount,
cifs_uncached_readdata_release);
break;
}
list_add_tail(&rdata->list, &rdata_list);
offset += cur_len;
len -= cur_len;
} while (len > 0);
/* if at least one read request send succeeded, then reset rc */
if (!list_empty(&rdata_list))
rc = 0;
/* the loop below should proceed in the order of increasing offsets */
restart_loop:
list_for_each_entry_safe(rdata, tmp, &rdata_list, list) {
if (!rc) {
ssize_t copied;
/* FIXME: freezable sleep too? */
rc = wait_for_completion_killable(&rdata->done);
if (rc)
rc = -EINTR;
else if (rdata->result)
rc = rdata->result;
else {
rc = cifs_readdata_to_iov(rdata, iov,
nr_segs, *poffset,
&copied);
total_read += copied;
}
/* resend call if it's a retryable error */
if (rc == -EAGAIN) {
rc = cifs_retry_async_readv(rdata);
goto restart_loop;
}
}
list_del_init(&rdata->list);
kref_put(&rdata->refcount, cifs_uncached_readdata_release);
}
cifs_stats_bytes_read(tcon, total_read);
*poffset += total_read;
/* mask nodata case */
if (rc == -ENODATA)
rc = 0;
return total_read ? total_read : rc;
}
ssize_t cifs_user_readv(struct kiocb *iocb, const struct iovec *iov,
unsigned long nr_segs, loff_t pos)
{
ssize_t read;
read = cifs_iovec_read(iocb->ki_filp, iov, nr_segs, &pos);
if (read > 0)
iocb->ki_pos = pos;
return read;
}
ssize_t cifs_strict_readv(struct kiocb *iocb, const struct iovec *iov,
unsigned long nr_segs, loff_t pos)
{
struct inode *inode;
inode = iocb->ki_filp->f_path.dentry->d_inode;
if (CIFS_I(inode)->clientCanCacheRead)
return generic_file_aio_read(iocb, iov, nr_segs, pos);
/*
* In strict cache mode we need to read from the server all the time
* if we don't have level II oplock because the server can delay mtime
* change - so we can't make a decision about inode invalidating.
* And we can also fail with pagereading if there are mandatory locks
* on pages affected by this read but not on the region from pos to
* pos+len-1.
*/
return cifs_user_readv(iocb, iov, nr_segs, pos);
}
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;
unsigned int rsize;
struct cifs_sb_info *cifs_sb;
struct cifs_tcon *tcon;
unsigned int xid;
char *current_offset;
struct cifsFileInfo *open_file;
struct cifs_io_parms io_parms;
int buf_type = CIFS_NO_BUFFER;
__u32 pid;
xid = get_xid();
cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
/* FIXME: set up handlers for larger reads and/or convert to async */
rsize = min_t(unsigned int, cifs_sb->rsize, CIFSMaxBufSize);
if (file->private_data == NULL) {
cifs: Fix incorrect return code being printed in cFYI messages FreeXid() along with freeing Xid does add a cifsFYI debug message that prints rc (return code) as well. In some code paths where we set/return error code after calling FreeXid(), incorrect error code is being printed when cifsFYI is enabled. This could be misleading in few cases. For eg. In cifs_open() if cifs_fill_filedata() returns a valid pointer to cifsFileInfo, FreeXid() prints rc=-13 whereas 0 is actually being returned. Fix this by setting rc before calling FreeXid(). Basically convert FreeXid(xid); rc = -ERR; return -ERR; => FreeXid(xid); return rc; [Note that Christoph would like to replace the GetXid/FreeXid calls, which are primarily used for debugging. This seems like a good longer term goal, but although there is an alternative tracing facility, there are no examples yet available that I know of that we can use (yet) to convert this cifs function entry/exit logging, and for creating an identifier that we can use to correlate all dmesg log entries for a particular vfs operation (ie identify all log entries for a particular vfs request to cifs: e.g. a particular close or read or write or byte range lock call ... and just using the thread id is harder). Eventually when a replacement for this is available (e.g. when NFS switches over and various samples to look at in other file systems) we can remove the GetXid/FreeXid macro but in the meantime multiple people use this run time configurable logging all the time for debugging, and Suresh's patch fixes a problem which made it harder to notice some low memory problems in the log so it is worthwhile to fix this problem until a better logging approach is able to be used] Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: Suresh Jayaraman <sjayaraman@suse.de> Signed-off-by: Steve French <sfrench@us.ibm.com>
2009-06-25 12:42:34 +00:00
rc = -EBADF;
free_xid(xid);
cifs: Fix incorrect return code being printed in cFYI messages FreeXid() along with freeing Xid does add a cifsFYI debug message that prints rc (return code) as well. In some code paths where we set/return error code after calling FreeXid(), incorrect error code is being printed when cifsFYI is enabled. This could be misleading in few cases. For eg. In cifs_open() if cifs_fill_filedata() returns a valid pointer to cifsFileInfo, FreeXid() prints rc=-13 whereas 0 is actually being returned. Fix this by setting rc before calling FreeXid(). Basically convert FreeXid(xid); rc = -ERR; return -ERR; => FreeXid(xid); return rc; [Note that Christoph would like to replace the GetXid/FreeXid calls, which are primarily used for debugging. This seems like a good longer term goal, but although there is an alternative tracing facility, there are no examples yet available that I know of that we can use (yet) to convert this cifs function entry/exit logging, and for creating an identifier that we can use to correlate all dmesg log entries for a particular vfs operation (ie identify all log entries for a particular vfs request to cifs: e.g. a particular close or read or write or byte range lock call ... and just using the thread id is harder). Eventually when a replacement for this is available (e.g. when NFS switches over and various samples to look at in other file systems) we can remove the GetXid/FreeXid macro but in the meantime multiple people use this run time configurable logging all the time for debugging, and Suresh's patch fixes a problem which made it harder to notice some low memory problems in the log so it is worthwhile to fix this problem until a better logging approach is able to be used] Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: Suresh Jayaraman <sjayaraman@suse.de> Signed-off-by: Steve French <sfrench@us.ibm.com>
2009-06-25 12:42:34 +00:00
return rc;
}
open_file = file->private_data;
tcon = tlink_tcon(open_file->tlink);
if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_RWPIDFORWARD)
pid = open_file->pid;
else
pid = current->tgid;
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) {
current_read_size = min_t(uint, read_size - total_read, rsize);
/*
* For windows me and 9x we do not want to request more than it
* negotiated since it will refuse the read then.
*/
if ((tcon->ses) && !(tcon->ses->capabilities &
tcon->ses->server->vals->cap_large_files)) {
current_read_size = min_t(uint, current_read_size,
CIFSMaxBufSize);
}
rc = -EAGAIN;
while (rc == -EAGAIN) {
if (open_file->invalidHandle) {
rc = cifs_reopen_file(open_file, true);
if (rc != 0)
break;
}
io_parms.netfid = open_file->fid.netfid;
io_parms.pid = pid;
io_parms.tcon = tcon;
io_parms.offset = *poffset;
io_parms.length = current_read_size;
rc = CIFSSMBRead(xid, &io_parms, &bytes_read,
&current_offset, &buf_type);
}
if (rc || (bytes_read == 0)) {
if (total_read) {
break;
} else {
free_xid(xid);
return rc;
}
} else {
cifs_stats_bytes_read(tcon, total_read);
*poffset += bytes_read;
}
}
free_xid(xid);
return total_read;
}
/*
* If the page is mmap'ed into a process' page tables, then we need to make
* sure that it doesn't change while being written back.
*/
static int
cifs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
{
struct page *page = vmf->page;
lock_page(page);
return VM_FAULT_LOCKED;
}
static struct vm_operations_struct cifs_file_vm_ops = {
.fault = filemap_fault,
.page_mkwrite = cifs_page_mkwrite,
};
int cifs_file_strict_mmap(struct file *file, struct vm_area_struct *vma)
{
int rc, xid;
struct inode *inode = file->f_path.dentry->d_inode;
xid = get_xid();
if (!CIFS_I(inode)->clientCanCacheRead) {
rc = cifs_invalidate_mapping(inode);
if (rc)
return rc;
}
rc = generic_file_mmap(file, vma);
if (rc == 0)
vma->vm_ops = &cifs_file_vm_ops;
free_xid(xid);
return rc;
}
int cifs_file_mmap(struct file *file, struct vm_area_struct *vma)
{
int rc, xid;
xid = get_xid();
rc = cifs_revalidate_file(file);
if (rc) {
cFYI(1, "Validation prior to mmap failed, error=%d", rc);
free_xid(xid);
return rc;
}
rc = generic_file_mmap(file, vma);
if (rc == 0)
vma->vm_ops = &cifs_file_vm_ops;
free_xid(xid);
return rc;
}
static void
cifs_readv_complete(struct work_struct *work)
{
struct cifs_readdata *rdata = container_of(work,
struct cifs_readdata, work);
struct page *page, *tpage;
list_for_each_entry_safe(page, tpage, &rdata->pages, lru) {
list_del(&page->lru);
lru_cache_add_file(page);
if (rdata->result == 0) {
kunmap(page);
flush_dcache_page(page);
SetPageUptodate(page);
}
unlock_page(page);
if (rdata->result == 0)
cifs_readpage_to_fscache(rdata->mapping->host, page);
page_cache_release(page);
}
kref_put(&rdata->refcount, cifs_readdata_release);
}
static int
cifs_readpages_marshal_iov(struct cifs_readdata *rdata, unsigned int remaining)
{
int len = 0;
struct page *page, *tpage;
u64 eof;
pgoff_t eof_index;
/* determine the eof that the server (probably) has */
eof = CIFS_I(rdata->mapping->host)->server_eof;
eof_index = eof ? (eof - 1) >> PAGE_CACHE_SHIFT : 0;
cFYI(1, "eof=%llu eof_index=%lu", eof, eof_index);
rdata->nr_iov = 1;
list_for_each_entry_safe(page, tpage, &rdata->pages, lru) {
if (remaining >= PAGE_CACHE_SIZE) {
/* enough data to fill the page */
rdata->iov[rdata->nr_iov].iov_base = kmap(page);
rdata->iov[rdata->nr_iov].iov_len = PAGE_CACHE_SIZE;
cFYI(1, "%u: idx=%lu iov_base=%p iov_len=%zu",
rdata->nr_iov, page->index,
rdata->iov[rdata->nr_iov].iov_base,
rdata->iov[rdata->nr_iov].iov_len);
++rdata->nr_iov;
len += PAGE_CACHE_SIZE;
remaining -= PAGE_CACHE_SIZE;
} else if (remaining > 0) {
/* enough for partial page, fill and zero the rest */
rdata->iov[rdata->nr_iov].iov_base = kmap(page);
rdata->iov[rdata->nr_iov].iov_len = remaining;
cFYI(1, "%u: idx=%lu iov_base=%p iov_len=%zu",
rdata->nr_iov, page->index,
rdata->iov[rdata->nr_iov].iov_base,
rdata->iov[rdata->nr_iov].iov_len);
memset(rdata->iov[rdata->nr_iov].iov_base + remaining,
'\0', PAGE_CACHE_SIZE - remaining);
++rdata->nr_iov;
len += remaining;
remaining = 0;
} else if (page->index > eof_index) {
/*
* The VFS will not try to do readahead past the
* i_size, but it's possible that we have outstanding
* writes with gaps in the middle and the i_size hasn't
* caught up yet. Populate those with zeroed out pages
* to prevent the VFS from repeatedly attempting to
* fill them until the writes are flushed.
*/
zero_user(page, 0, PAGE_CACHE_SIZE);
list_del(&page->lru);
lru_cache_add_file(page);
flush_dcache_page(page);
SetPageUptodate(page);
unlock_page(page);
page_cache_release(page);
} else {
/* no need to hold page hostage */
list_del(&page->lru);
lru_cache_add_file(page);
unlock_page(page);
page_cache_release(page);
}
}
return len;
}
static int cifs_readpages(struct file *file, struct address_space *mapping,
struct list_head *page_list, unsigned num_pages)
{
int rc;
struct list_head tmplist;
struct cifsFileInfo *open_file = file->private_data;
struct cifs_sb_info *cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
unsigned int rsize = cifs_sb->rsize;
pid_t pid;
/*
* Give up immediately if rsize is too small to read an entire page.
* The VFS will fall back to readpage. We should never reach this
* point however since we set ra_pages to 0 when the rsize is smaller
* than a cache page.
*/
if (unlikely(rsize < PAGE_CACHE_SIZE))
return 0;
/*
* Reads as many pages as possible from fscache. Returns -ENOBUFS
* immediately if the cookie is negative
*/
rc = cifs_readpages_from_fscache(mapping->host, mapping, page_list,
&num_pages);
if (rc == 0)
return rc;
if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_RWPIDFORWARD)
pid = open_file->pid;
else
pid = current->tgid;
rc = 0;
INIT_LIST_HEAD(&tmplist);
cFYI(1, "%s: file=%p mapping=%p num_pages=%u", __func__, file,
mapping, num_pages);
/*
* Start with the page at end of list and move it to private
* list. Do the same with any following pages until we hit
* the rsize limit, hit an index discontinuity, or run out of
* pages. Issue the async read and then start the loop again
* until the list is empty.
*
* Note that list order is important. The page_list is in
* the order of declining indexes. When we put the pages in
* the rdata->pages, then we want them in increasing order.
*/
while (!list_empty(page_list)) {
unsigned int bytes = PAGE_CACHE_SIZE;
unsigned int expected_index;
unsigned int nr_pages = 1;
loff_t offset;
struct page *page, *tpage;
struct cifs_readdata *rdata;
page = list_entry(page_list->prev, struct page, lru);
/*
* Lock the page and put it in the cache. Since no one else
* should have access to this page, we're safe to simply set
* PG_locked without checking it first.
*/
__set_page_locked(page);
rc = add_to_page_cache_locked(page, mapping,
page->index, GFP_KERNEL);
/* give up if we can't stick it in the cache */
if (rc) {
__clear_page_locked(page);
break;
}
/* move first page to the tmplist */
offset = (loff_t)page->index << PAGE_CACHE_SHIFT;
list_move_tail(&page->lru, &tmplist);
/* now try and add more pages onto the request */
expected_index = page->index + 1;
list_for_each_entry_safe_reverse(page, tpage, page_list, lru) {
/* discontinuity ? */
if (page->index != expected_index)
break;
/* would this page push the read over the rsize? */
if (bytes + PAGE_CACHE_SIZE > rsize)
break;
__set_page_locked(page);
if (add_to_page_cache_locked(page, mapping,
page->index, GFP_KERNEL)) {
__clear_page_locked(page);
break;
}
list_move_tail(&page->lru, &tmplist);
bytes += PAGE_CACHE_SIZE;
expected_index++;
nr_pages++;
}
rdata = cifs_readdata_alloc(nr_pages, cifs_readv_complete);
if (!rdata) {
/* best to give up if we're out of mem */
list_for_each_entry_safe(page, tpage, &tmplist, lru) {
list_del(&page->lru);
lru_cache_add_file(page);
unlock_page(page);
page_cache_release(page);
}
rc = -ENOMEM;
break;
}
rdata->cfile = cifsFileInfo_get(open_file);
rdata->mapping = mapping;
rdata->offset = offset;
rdata->bytes = bytes;
rdata->pid = pid;
rdata->marshal_iov = cifs_readpages_marshal_iov;
list_splice_init(&tmplist, &rdata->pages);
rc = cifs_retry_async_readv(rdata);
if (rc != 0) {
list_for_each_entry_safe(page, tpage, &rdata->pages,
lru) {
list_del(&page->lru);
lru_cache_add_file(page);
unlock_page(page);
page_cache_release(page);
}
kref_put(&rdata->refcount, cifs_readdata_release);
break;
}
kref_put(&rdata->refcount, cifs_readdata_release);
}
return rc;
}
static int cifs_readpage_worker(struct file *file, struct page *page,
loff_t *poffset)
{
char *read_data;
int rc;
/* Is the page cached? */
rc = cifs_readpage_from_fscache(file->f_path.dentry->d_inode, page);
if (rc == 0)
goto read_complete;
page_cache_get(page);
read_data = kmap(page);
/* for reads over a certain size could initiate async read ahead */
rc = cifs_read(file, read_data, PAGE_CACHE_SIZE, poffset);
if (rc < 0)
goto io_error;
else
cFYI(1, "Bytes read %d", rc);
file->f_path.dentry->d_inode->i_atime =
current_fs_time(file->f_path.dentry->d_inode->i_sb);
if (PAGE_CACHE_SIZE > rc)
memset(read_data + rc, 0, PAGE_CACHE_SIZE - rc);
flush_dcache_page(page);
SetPageUptodate(page);
/* send this page to the cache */
cifs_readpage_to_fscache(file->f_path.dentry->d_inode, page);
rc = 0;
io_error:
kunmap(page);
page_cache_release(page);
read_complete:
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;
unsigned int xid;
xid = get_xid();
if (file->private_data == NULL) {
cifs: Fix incorrect return code being printed in cFYI messages FreeXid() along with freeing Xid does add a cifsFYI debug message that prints rc (return code) as well. In some code paths where we set/return error code after calling FreeXid(), incorrect error code is being printed when cifsFYI is enabled. This could be misleading in few cases. For eg. In cifs_open() if cifs_fill_filedata() returns a valid pointer to cifsFileInfo, FreeXid() prints rc=-13 whereas 0 is actually being returned. Fix this by setting rc before calling FreeXid(). Basically convert FreeXid(xid); rc = -ERR; return -ERR; => FreeXid(xid); return rc; [Note that Christoph would like to replace the GetXid/FreeXid calls, which are primarily used for debugging. This seems like a good longer term goal, but although there is an alternative tracing facility, there are no examples yet available that I know of that we can use (yet) to convert this cifs function entry/exit logging, and for creating an identifier that we can use to correlate all dmesg log entries for a particular vfs operation (ie identify all log entries for a particular vfs request to cifs: e.g. a particular close or read or write or byte range lock call ... and just using the thread id is harder). Eventually when a replacement for this is available (e.g. when NFS switches over and various samples to look at in other file systems) we can remove the GetXid/FreeXid macro but in the meantime multiple people use this run time configurable logging all the time for debugging, and Suresh's patch fixes a problem which made it harder to notice some low memory problems in the log so it is worthwhile to fix this problem until a better logging approach is able to be used] Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: Suresh Jayaraman <sjayaraman@suse.de> Signed-off-by: Steve French <sfrench@us.ibm.com>
2009-06-25 12:42:34 +00:00
rc = -EBADF;
free_xid(xid);
cifs: Fix incorrect return code being printed in cFYI messages FreeXid() along with freeing Xid does add a cifsFYI debug message that prints rc (return code) as well. In some code paths where we set/return error code after calling FreeXid(), incorrect error code is being printed when cifsFYI is enabled. This could be misleading in few cases. For eg. In cifs_open() if cifs_fill_filedata() returns a valid pointer to cifsFileInfo, FreeXid() prints rc=-13 whereas 0 is actually being returned. Fix this by setting rc before calling FreeXid(). Basically convert FreeXid(xid); rc = -ERR; return -ERR; => FreeXid(xid); return rc; [Note that Christoph would like to replace the GetXid/FreeXid calls, which are primarily used for debugging. This seems like a good longer term goal, but although there is an alternative tracing facility, there are no examples yet available that I know of that we can use (yet) to convert this cifs function entry/exit logging, and for creating an identifier that we can use to correlate all dmesg log entries for a particular vfs operation (ie identify all log entries for a particular vfs request to cifs: e.g. a particular close or read or write or byte range lock call ... and just using the thread id is harder). Eventually when a replacement for this is available (e.g. when NFS switches over and various samples to look at in other file systems) we can remove the GetXid/FreeXid macro but in the meantime multiple people use this run time configurable logging all the time for debugging, and Suresh's patch fixes a problem which made it harder to notice some low memory problems in the log so it is worthwhile to fix this problem until a better logging approach is able to be used] Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: Suresh Jayaraman <sjayaraman@suse.de> Signed-off-by: Steve French <sfrench@us.ibm.com>
2009-06-25 12:42:34 +00:00
return rc;
}
cFYI(1, "readpage %p at offset %d 0x%x",
page, (int)offset, (int)offset);
rc = cifs_readpage_worker(file, page, &offset);
unlock_page(page);
free_xid(xid);
return rc;
}
static int is_inode_writable(struct cifsInodeInfo *cifs_inode)
{
struct cifsFileInfo *open_file;
spin_lock(&cifs_file_list_lock);
list_for_each_entry(open_file, &cifs_inode->openFileList, flist) {
if (OPEN_FMODE(open_file->f_flags) & FMODE_WRITE) {
spin_unlock(&cifs_file_list_lock);
return 1;
}
}
spin_unlock(&cifs_file_list_lock);
return 0;
}
/* 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
refreshing the inode only on increases in the file size
but this is tricky to do without racing with writebehind
page caching in the current Linux kernel design */
bool is_size_safe_to_change(struct cifsInodeInfo *cifsInode, __u64 end_of_file)
{
if (!cifsInode)
return true;
if (is_inode_writable(cifsInode)) {
/* This inode is open for write at least once */
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) {
/* since no page cache to corrupt on directio
we can change size safely */
return true;
}
if (i_size_read(&cifsInode->vfs_inode) < end_of_file)
return true;
return false;
} else
return true;
}
static int cifs_write_begin(struct file *file, struct address_space *mapping,
loff_t pos, unsigned len, unsigned flags,
struct page **pagep, void **fsdata)
{
pgoff_t index = pos >> PAGE_CACHE_SHIFT;
loff_t offset = pos & (PAGE_CACHE_SIZE - 1);
loff_t page_start = pos & PAGE_MASK;
loff_t i_size;
struct page *page;
int rc = 0;
cFYI(1, "write_begin from %lld len %d", (long long)pos, len);
fs: symlink write_begin allocation context fix With the write_begin/write_end aops, page_symlink was broken because it could no longer pass a GFP_NOFS type mask into the point where the allocations happened. They are done in write_begin, which would always assume that the filesystem can be entered from reclaim. This bug could cause filesystem deadlocks. The funny thing with having a gfp_t mask there is that it doesn't really allow the caller to arbitrarily tinker with the context in which it can be called. It couldn't ever be GFP_ATOMIC, for example, because it needs to take the page lock. The only thing any callers care about is __GFP_FS anyway, so turn that into a single flag. Add a new flag for write_begin, AOP_FLAG_NOFS. Filesystems can now act on this flag in their write_begin function. Change __grab_cache_page to accept a nofs argument as well, to honour that flag (while we're there, change the name to grab_cache_page_write_begin which is more instructive and does away with random leading underscores). This is really a more flexible way to go in the end anyway -- if a filesystem happens to want any extra allocations aside from the pagecache ones in ints write_begin function, it may now use GFP_KERNEL (rather than GFP_NOFS) for common case allocations (eg. ocfs2_alloc_write_ctxt, for a random example). [kosaki.motohiro@jp.fujitsu.com: fix ubifs] [kosaki.motohiro@jp.fujitsu.com: fix fuse] Signed-off-by: Nick Piggin <npiggin@suse.de> Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: <stable@kernel.org> [2.6.28.x] Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> [ Cleaned up the calling convention: just pass in the AOP flags untouched to the grab_cache_page_write_begin() function. That just simplifies everybody, and may even allow future expansion of the logic. - Linus ] Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-04 20:00:53 +00:00
page = grab_cache_page_write_begin(mapping, index, flags);
if (!page) {
rc = -ENOMEM;
goto out;
}
if (PageUptodate(page))
goto out;
/*
* If we write a full page it will be up to date, no need to read from
* the server. If the write is short, we'll end up doing a sync write
* instead.
*/
if (len == PAGE_CACHE_SIZE)
goto out;
/*
* optimize away the read when we have an oplock, and we're not
* expecting to use any of the data we'd be reading in. That
* is, when the page lies beyond the EOF, or straddles the EOF
* and the write will cover all of the existing data.
*/
if (CIFS_I(mapping->host)->clientCanCacheRead) {
i_size = i_size_read(mapping->host);
if (page_start >= i_size ||
(offset == 0 && (pos + len) >= i_size)) {
zero_user_segments(page, 0, offset,
offset + len,
PAGE_CACHE_SIZE);
/*
* PageChecked means that the parts of the page
* to which we're not writing are considered up
* to date. Once the data is copied to the
* page, it can be set uptodate.
*/
SetPageChecked(page);
goto out;
}
}
if ((file->f_flags & O_ACCMODE) != O_WRONLY) {
/*
* might as well read a page, it is fast enough. If we get
* an error, we don't need to return it. cifs_write_end will
* do a sync write instead since PG_uptodate isn't set.
*/
cifs_readpage_worker(file, page, &page_start);
} 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 write_end so is fine */
}
out:
*pagep = page;
return rc;
}
static int cifs_release_page(struct page *page, gfp_t gfp)
{
if (PagePrivate(page))
return 0;
return cifs_fscache_release_page(page, gfp);
}
static void cifs_invalidate_page(struct page *page, unsigned long offset)
{
struct cifsInodeInfo *cifsi = CIFS_I(page->mapping->host);
if (offset == 0)
cifs_fscache_invalidate_page(page, &cifsi->vfs_inode);
}
static int cifs_launder_page(struct page *page)
{
int rc = 0;
loff_t range_start = page_offset(page);
loff_t range_end = range_start + (loff_t)(PAGE_CACHE_SIZE - 1);
struct writeback_control wbc = {
.sync_mode = WB_SYNC_ALL,
.nr_to_write = 0,
.range_start = range_start,
.range_end = range_end,
};
cFYI(1, "Launder page: %p", page);
if (clear_page_dirty_for_io(page))
rc = cifs_writepage_locked(page, &wbc);
cifs_fscache_invalidate_page(page, page->mapping->host);
return rc;
}
void cifs_oplock_break(struct work_struct *work)
{
struct cifsFileInfo *cfile = container_of(work, struct cifsFileInfo,
oplock_break);
struct inode *inode = cfile->dentry->d_inode;
struct cifsInodeInfo *cinode = CIFS_I(inode);
int rc = 0;
if (inode && S_ISREG(inode->i_mode)) {
if (cinode->clientCanCacheRead)
break_lease(inode, O_RDONLY);
else
break_lease(inode, O_WRONLY);
rc = filemap_fdatawrite(inode->i_mapping);
if (cinode->clientCanCacheRead == 0) {
rc = filemap_fdatawait(inode->i_mapping);
mapping_set_error(inode->i_mapping, rc);
invalidate_remote_inode(inode);
}
cFYI(1, "Oplock flush inode %p rc %d", inode, rc);
}
rc = cifs_push_locks(cfile);
if (rc)
cERROR(1, "Push locks rc = %d", rc);
/*
* releasing stale oplock after recent reconnect of smb session using
* a now incorrect file handle is not a data integrity issue but do
* not bother sending an oplock release if session to server still is
* disconnected since oplock already released by the server
*/
if (!cfile->oplock_break_cancelled) {
rc = CIFSSMBLock(0, tlink_tcon(cfile->tlink), cfile->fid.netfid,
current->tgid, 0, 0, 0, 0,
LOCKING_ANDX_OPLOCK_RELEASE, false,
cinode->clientCanCacheRead ? 1 : 0);
cFYI(1, "Oplock release rc = %d", rc);
}
}
const struct address_space_operations cifs_addr_ops = {
.readpage = cifs_readpage,
.readpages = cifs_readpages,
.writepage = cifs_writepage,
.writepages = cifs_writepages,
.write_begin = cifs_write_begin,
.write_end = cifs_write_end,
.set_page_dirty = __set_page_dirty_nobuffers,
.releasepage = cifs_release_page,
.invalidatepage = cifs_invalidate_page,
.launder_page = cifs_launder_page,
};
/*
* 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.
*/
const struct address_space_operations cifs_addr_ops_smallbuf = {
.readpage = cifs_readpage,
.writepage = cifs_writepage,
.writepages = cifs_writepages,
.write_begin = cifs_write_begin,
.write_end = cifs_write_end,
.set_page_dirty = __set_page_dirty_nobuffers,
.releasepage = cifs_release_page,
.invalidatepage = cifs_invalidate_page,
.launder_page = cifs_launder_page,
};