diff options
Diffstat (limited to 'fs/cifs/file.c')
| -rw-r--r-- | fs/cifs/file.c | 2412 |
1 files changed, 1614 insertions, 798 deletions
diff --git a/fs/cifs/file.c b/fs/cifs/file.c index 4dd9283885e..e90a1e9aa62 100644 --- a/fs/cifs/file.c +++ b/fs/cifs/file.c @@ -43,6 +43,7 @@ #include "cifs_fs_sb.h" #include "fscache.h" + static inline int cifs_convert_flags(unsigned int flags) { if ((flags & O_ACCMODE) == O_RDONLY) @@ -72,10 +73,14 @@ static u32 cifs_posix_convert_flags(unsigned int flags) else if ((flags & O_ACCMODE) == O_RDWR) posix_flags = SMB_O_RDWR; - if (flags & O_CREAT) + if (flags & O_CREAT) { posix_flags |= SMB_O_CREAT; - if (flags & O_EXCL) - posix_flags |= SMB_O_EXCL; + if (flags & O_EXCL) + posix_flags |= SMB_O_EXCL; + } else if (flags & O_EXCL) + cifs_dbg(FYI, "Application %s pid %d has incorrectly set O_EXCL flag but not O_CREAT on file open. Ignoring O_EXCL\n", + current->comm, current->tgid); + if (flags & O_TRUNC) posix_flags |= SMB_O_TRUNC; /* be safe and imply O_SYNC for O_DSYNC */ @@ -107,7 +112,7 @@ static inline int cifs_get_disposition(unsigned int flags) int cifs_posix_open(char *full_path, struct inode **pinode, struct super_block *sb, int mode, unsigned int f_flags, - __u32 *poplock, __u16 *pnetfid, int xid) + __u32 *poplock, __u16 *pnetfid, unsigned int xid) { int rc; FILE_UNIX_BASIC_INFO *presp_data; @@ -117,7 +122,7 @@ int cifs_posix_open(char *full_path, struct inode **pinode, struct tcon_link *tlink; struct cifs_tcon *tcon; - cFYI(1, "posix open %s", full_path); + cifs_dbg(FYI, "posix open %s\n", full_path); presp_data = kzalloc(sizeof(FILE_UNIX_BASIC_INFO), GFP_KERNEL); if (presp_data == NULL) @@ -169,16 +174,21 @@ posix_open_ret: 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 *poplock, - __u16 *pnetfid, int xid) + struct cifs_tcon *tcon, unsigned int f_flags, __u32 *oplock, + struct cifs_fid *fid, unsigned int xid) { int rc; - int desiredAccess; + int desired_access; int disposition; int create_options = CREATE_NOT_DIR; FILE_ALL_INFO *buf; + struct TCP_Server_Info *server = tcon->ses->server; + struct cifs_open_parms oparms; + + if (!server->ops->open) + return -ENOSYS; - desiredAccess = cifs_convert_flags(f_flags); + desired_access = cifs_convert_flags(f_flags); /********************************************************************* * open flag mapping table: @@ -215,16 +225,16 @@ cifs_nt_open(char *full_path, struct inode *inode, struct cifs_sb_info *cifs_sb, if (backup_cred(cifs_sb)) create_options |= CREATE_OPEN_BACKUP_INTENT; - if (tcon->ses->capabilities & CAP_NT_SMBS) - rc = CIFSSMBOpen(xid, tcon, full_path, disposition, - desiredAccess, create_options, pnetfid, poplock, buf, - cifs_sb->local_nls, cifs_sb->mnt_cifs_flags - & CIFS_MOUNT_MAP_SPECIAL_CHR); - else - rc = SMBLegacyOpen(xid, tcon, full_path, disposition, - desiredAccess, CREATE_NOT_DIR, pnetfid, poplock, buf, - cifs_sb->local_nls, cifs_sb->mnt_cifs_flags - & CIFS_MOUNT_MAP_SPECIAL_CHR); + oparms.tcon = tcon; + oparms.cifs_sb = cifs_sb; + oparms.desired_access = desired_access; + oparms.create_options = create_options; + oparms.disposition = disposition; + oparms.path = full_path; + oparms.fid = fid; + oparms.reconnect = false; + + rc = server->ops->open(xid, &oparms, oplock, buf); if (rc) goto out; @@ -234,54 +244,111 @@ cifs_nt_open(char *full_path, struct inode *inode, struct cifs_sb_info *cifs_sb, xid); else rc = cifs_get_inode_info(&inode, full_path, buf, inode->i_sb, - xid, pnetfid); + xid, fid); out: kfree(buf); return rc; } +static bool +cifs_has_mand_locks(struct cifsInodeInfo *cinode) +{ + struct cifs_fid_locks *cur; + bool has_locks = false; + + down_read(&cinode->lock_sem); + list_for_each_entry(cur, &cinode->llist, llist) { + if (!list_empty(&cur->locks)) { + has_locks = true; + break; + } + } + up_read(&cinode->lock_sem); + return has_locks; +} + struct cifsFileInfo * -cifs_new_fileinfo(__u16 fileHandle, struct file *file, +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 *pCifsInode = CIFS_I(inode); - struct cifsFileInfo *pCifsFile; - - pCifsFile = kzalloc(sizeof(struct cifsFileInfo), GFP_KERNEL); - if (pCifsFile == NULL) - return pCifsFile; - - pCifsFile->count = 1; - pCifsFile->netfid = fileHandle; - pCifsFile->pid = current->tgid; - pCifsFile->uid = current_fsuid(); - pCifsFile->dentry = dget(dentry); - pCifsFile->f_flags = file->f_flags; - pCifsFile->invalidHandle = false; - pCifsFile->tlink = cifs_get_tlink(tlink); - mutex_init(&pCifsFile->fh_mutex); - INIT_WORK(&pCifsFile->oplock_break, cifs_oplock_break); + struct cifsInodeInfo *cinode = CIFS_I(inode); + struct cifsFileInfo *cfile; + struct cifs_fid_locks *fdlocks; + struct cifs_tcon *tcon = tlink_tcon(tlink); + struct TCP_Server_Info *server = tcon->ses->server; + + cfile = kzalloc(sizeof(struct cifsFileInfo), GFP_KERNEL); + if (cfile == NULL) + return cfile; + + fdlocks = kzalloc(sizeof(struct cifs_fid_locks), GFP_KERNEL); + if (!fdlocks) { + kfree(cfile); + return NULL; + } + + INIT_LIST_HEAD(&fdlocks->locks); + fdlocks->cfile = cfile; + cfile->llist = fdlocks; + down_write(&cinode->lock_sem); + list_add(&fdlocks->llist, &cinode->llist); + up_write(&cinode->lock_sem); + + 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); + INIT_WORK(&cfile->oplock_break, cifs_oplock_break); + mutex_init(&cfile->fh_mutex); + + cifs_sb_active(inode->i_sb); + + /* + * If the server returned a read oplock and we have mandatory brlocks, + * set oplock level to None. + */ + if (server->ops->is_read_op(oplock) && cifs_has_mand_locks(cinode)) { + cifs_dbg(FYI, "Reset oplock val from read to None due to mand locks\n"); + oplock = 0; + } spin_lock(&cifs_file_list_lock); - list_add(&pCifsFile->tlist, &(tlink_tcon(tlink)->openFileList)); + if (fid->pending_open->oplock != CIFS_OPLOCK_NO_CHANGE && oplock) + oplock = fid->pending_open->oplock; + list_del(&fid->pending_open->olist); + + fid->purge_cache = false; + server->ops->set_fid(cfile, fid, oplock); + + list_add(&cfile->tlist, &tcon->openFileList); /* if readable file instance put first in list*/ if (file->f_mode & FMODE_READ) - list_add(&pCifsFile->flist, &pCifsInode->openFileList); + list_add(&cfile->flist, &cinode->openFileList); else - list_add_tail(&pCifsFile->flist, &pCifsInode->openFileList); + list_add_tail(&cfile->flist, &cinode->openFileList); spin_unlock(&cifs_file_list_lock); - cifs_set_oplock_level(pCifsInode, oplock); - pCifsInode->can_cache_brlcks = pCifsInode->clientCanCacheAll; + if (fid->purge_cache) + cifs_zap_mapping(inode); - file->private_data = pCifsFile; - return pCifsFile; + 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 @@ -292,9 +359,13 @@ 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 TCP_Server_Info *server = tcon->ses->server; struct cifsInodeInfo *cifsi = CIFS_I(inode); - struct cifs_sb_info *cifs_sb = CIFS_SB(inode->i_sb); + struct super_block *sb = inode->i_sb; + struct cifs_sb_info *cifs_sb = CIFS_SB(sb); struct cifsLockInfo *li, *tmp; + struct cifs_fid fid; + struct cifs_pending_open open; spin_lock(&cifs_file_list_lock); if (--cifs_file->count > 0) { @@ -302,20 +373,26 @@ void cifsFileInfo_put(struct cifsFileInfo *cifs_file) return; } + if (server->ops->get_lease_key) + server->ops->get_lease_key(inode, &fid); + + /* store open in pending opens to make sure we don't miss lease break */ + cifs_add_pending_open_locked(&fid, cifs_file->tlink, &open); + /* 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 */ + cifs_dbg(FYI, "closing last open instance for inode %p\n", + 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; - + set_bit(CIFS_INO_INVALID_MAPPING, &cifsi->flags); cifs_set_oplock_level(cifsi, 0); } spin_unlock(&cifs_file_list_lock); @@ -323,53 +400,63 @@ void cifsFileInfo_put(struct cifsFileInfo *cifs_file) cancel_work_sync(&cifs_file->oplock_break); if (!tcon->need_reconnect && !cifs_file->invalidHandle) { - int xid, rc; + struct TCP_Server_Info *server = tcon->ses->server; + unsigned int xid; - xid = GetXid(); - rc = CIFSSMBClose(xid, tcon, cifs_file->netfid); - FreeXid(xid); + xid = get_xid(); + if (server->ops->close) + server->ops->close(xid, tcon, &cifs_file->fid); + _free_xid(xid); } - /* Delete any outstanding lock records. We'll lose them when the file + cifs_del_pending_open(&open); + + /* + * 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, &cifsi->llist, llist) { - if (li->netfid != cifs_file->netfid) - continue; + down_write(&cifsi->lock_sem); + list_for_each_entry_safe(li, tmp, &cifs_file->llist->locks, llist) { list_del(&li->llist); cifs_del_lock_waiters(li); kfree(li); } - mutex_unlock(&cifsi->lock_mutex); + list_del(&cifs_file->llist->llist); + kfree(cifs_file->llist); + up_write(&cifsi->lock_sem); cifs_put_tlink(cifs_file->tlink); dput(cifs_file->dentry); + cifs_sb_deactive(sb); kfree(cifs_file); } int cifs_open(struct inode *inode, struct file *file) + { int rc = -EACCES; - int xid; + unsigned int xid; __u32 oplock; struct cifs_sb_info *cifs_sb; + struct TCP_Server_Info *server; struct cifs_tcon *tcon; struct tcon_link *tlink; - struct cifsFileInfo *pCifsFile = NULL; + struct cifsFileInfo *cfile = NULL; char *full_path = NULL; bool posix_open_ok = false; - __u16 netfid; + struct cifs_fid fid; + struct cifs_pending_open open; - xid = GetXid(); + xid = get_xid(); cifs_sb = CIFS_SB(inode->i_sb); tlink = cifs_sb_tlink(cifs_sb); if (IS_ERR(tlink)) { - FreeXid(xid); + free_xid(xid); return PTR_ERR(tlink); } tcon = tlink_tcon(tlink); + server = tcon->ses->server; full_path = build_path_from_dentry(file->f_path.dentry); if (full_path == NULL) { @@ -377,51 +464,61 @@ int cifs_open(struct inode *inode, struct file *file) goto out; } - cFYI(1, "inode = 0x%p file flags are 0x%x for %s", + cifs_dbg(FYI, "inode = 0x%p file flags are 0x%x for %s\n", inode, file->f_flags, full_path); - if (enable_oplocks) + if (server->oplocks) oplock = REQ_OPLOCK; else oplock = 0; if (!tcon->broken_posix_open && tcon->unix_ext && - (tcon->ses->capabilities & CAP_UNIX) && - (CIFS_UNIX_POSIX_PATH_OPS_CAP & - le64_to_cpu(tcon->fsUnixInfo.Capability))) { + 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, &netfid, xid); + file->f_flags, &oplock, &fid.netfid, xid); if (rc == 0) { - cFYI(1, "posix open succeeded"); + cifs_dbg(FYI, "posix open succeeded\n"); 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); + cifs_dbg(VFS, "server %s of type %s returned unexpected error on SMB posix open, disabling posix open support. Check if server update available.\n", + 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 */ + /* + * Else fallthrough to retry open the old way on network i/o + * or DFS errors. + */ } + if (server->ops->get_lease_key) + server->ops->get_lease_key(inode, &fid); + + cifs_add_pending_open(&fid, tlink, &open); + if (!posix_open_ok) { + if (server->ops->get_lease_key) + server->ops->get_lease_key(inode, &fid); + rc = cifs_nt_open(full_path, inode, cifs_sb, tcon, - file->f_flags, &oplock, &netfid, xid); - if (rc) + file->f_flags, &oplock, &fid, xid); + if (rc) { + cifs_del_pending_open(&open); goto out; + } } - pCifsFile = cifs_new_fileinfo(netfid, file, tlink, oplock); - if (pCifsFile == NULL) { - CIFSSMBClose(xid, tcon, netfid); + cfile = cifs_new_fileinfo(&fid, file, tlink, oplock); + if (cfile == NULL) { + if (server->ops->close) + server->ops->close(xid, tcon, &fid); + cifs_del_pending_open(&open); rc = -ENOMEM; goto out; } @@ -429,162 +526,208 @@ int cifs_open(struct inode *inode, struct file *file) 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 */ + /* + * 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, + .uid = INVALID_UID, /* no change */ + .gid = INVALID_GID, /* no change */ .ctime = NO_CHANGE_64, .atime = NO_CHANGE_64, .mtime = NO_CHANGE_64, .device = 0, }; - CIFSSMBUnixSetFileInfo(xid, tcon, &args, netfid, - pCifsFile->pid); + CIFSSMBUnixSetFileInfo(xid, tcon, &args, fid.netfid, + cfile->pid); } out: kfree(full_path); - FreeXid(xid); + 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) +static int cifs_push_posix_locks(struct cifsFileInfo *cfile); + +/* + * Try to reacquire byte range locks that were released when session + * to server was lost. + */ +static int +cifs_relock_file(struct cifsFileInfo *cfile) { + struct cifs_sb_info *cifs_sb = CIFS_SB(cfile->dentry->d_sb); + struct cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode); + struct cifs_tcon *tcon = tlink_tcon(cfile->tlink); int rc = 0; -/* BB list all locks open on this file and relock */ + down_read(&cinode->lock_sem); + if (cinode->can_cache_brlcks) { + /* can cache locks - no need to relock */ + up_read(&cinode->lock_sem); + return rc; + } + if (cap_unix(tcon->ses) && + (CIFS_UNIX_FCNTL_CAP & le64_to_cpu(tcon->fsUnixInfo.Capability)) && + ((cifs_sb->mnt_cifs_flags & CIFS_MOUNT_NOPOSIXBRL) == 0)) + rc = cifs_push_posix_locks(cfile); + else + rc = tcon->ses->server->ops->push_mand_locks(cfile); + + up_read(&cinode->lock_sem); return rc; } -static int cifs_reopen_file(struct cifsFileInfo *pCifsFile, bool can_flush) +static int +cifs_reopen_file(struct cifsFileInfo *cfile, bool can_flush) { int rc = -EACCES; - int xid; + unsigned int xid; __u32 oplock; struct cifs_sb_info *cifs_sb; struct cifs_tcon *tcon; - struct cifsInodeInfo *pCifsInode; + struct TCP_Server_Info *server; + struct cifsInodeInfo *cinode; struct inode *inode; char *full_path = NULL; - int desiredAccess; + int desired_access; int disposition = FILE_OPEN; int create_options = CREATE_NOT_DIR; - __u16 netfid; + struct cifs_open_parms oparms; - xid = GetXid(); - mutex_lock(&pCifsFile->fh_mutex); - if (!pCifsFile->invalidHandle) { - mutex_unlock(&pCifsFile->fh_mutex); + xid = get_xid(); + mutex_lock(&cfile->fh_mutex); + if (!cfile->invalidHandle) { + mutex_unlock(&cfile->fh_mutex); rc = 0; - FreeXid(xid); + free_xid(xid); return rc; } - inode = pCifsFile->dentry->d_inode; + inode = cfile->dentry->d_inode; cifs_sb = CIFS_SB(inode->i_sb); - tcon = tlink_tcon(pCifsFile->tlink); + 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(pCifsFile->dentry); + /* + * 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(&pCifsFile->fh_mutex); - FreeXid(xid); + mutex_unlock(&cfile->fh_mutex); + free_xid(xid); return rc; } - cFYI(1, "inode = 0x%p file flags 0x%x for %s", - inode, pCifsFile->f_flags, full_path); + cifs_dbg(FYI, "inode = 0x%p file flags 0x%x for %s\n", + inode, cfile->f_flags, full_path); - if (enable_oplocks) + if (tcon->ses->server->oplocks) oplock = REQ_OPLOCK; else oplock = 0; - if (tcon->unix_ext && (tcon->ses->capabilities & CAP_UNIX) && + if (tcon->unix_ext && cap_unix(tcon->ses) && (CIFS_UNIX_POSIX_PATH_OPS_CAP & - le64_to_cpu(tcon->fsUnixInfo.Capability))) { - + 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 = pCifsFile->f_flags & + 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, &netfid, xid); + cifs_sb->mnt_file_mode /* ignored */, + oflags, &oplock, &cfile->fid.netfid, xid); if (rc == 0) { - cFYI(1, "posix reopen succeeded"); + cifs_dbg(FYI, "posix reopen succeeded\n"); + oparms.reconnect = true; goto reopen_success; } - /* fallthrough to retry open the old way on errors, especially - in the reconnect path it is important to retry hard */ + /* + * fallthrough to retry open the old way on errors, especially + * in the reconnect path it is important to retry hard + */ } - desiredAccess = cifs_convert_flags(pCifsFile->f_flags); + 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 SMBOpen 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 */ + if (server->ops->get_lease_key) + server->ops->get_lease_key(inode, &cfile->fid); + + oparms.tcon = tcon; + oparms.cifs_sb = cifs_sb; + oparms.desired_access = desired_access; + oparms.create_options = create_options; + oparms.disposition = disposition; + oparms.path = full_path; + oparms.fid = &cfile->fid; + oparms.reconnect = true; + + /* + * Can not refresh inode by passing in file_info buf to be returned by + * ops->open 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, &oparms, &oplock, NULL); + if (rc == -ENOENT && oparms.reconnect == false) { + /* durable handle timeout is expired - open the file again */ + rc = server->ops->open(xid, &oparms, &oplock, NULL); + /* indicate that we need to relock the file */ + oparms.reconnect = true; + } - rc = CIFSSMBOpen(xid, tcon, full_path, disposition, desiredAccess, - create_options, &netfid, &oplock, NULL, - cifs_sb->local_nls, cifs_sb->mnt_cifs_flags & - CIFS_MOUNT_MAP_SPECIAL_CHR); if (rc) { - mutex_unlock(&pCifsFile->fh_mutex); - cFYI(1, "cifs_open returned 0x%x", rc); - cFYI(1, "oplock: %d", oplock); + mutex_unlock(&cfile->fh_mutex); + cifs_dbg(FYI, "cifs_reopen returned 0x%x\n", rc); + cifs_dbg(FYI, "oplock: %d\n", oplock); goto reopen_error_exit; } reopen_success: - pCifsFile->netfid = netfid; - pCifsFile->invalidHandle = false; - mutex_unlock(&pCifsFile->fh_mutex); - pCifsInode = CIFS_I(inode); + 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); + 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 inod - info */ - - cifs_set_oplock_level(pCifsInode, oplock); + 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. + */ - cifs_relock_file(pCifsFile); + server->ops->set_fid(cfile, &cfile->fid, oplock); + if (oparms.reconnect) + cifs_relock_file(cfile); reopen_error_exit: kfree(full_path); - FreeXid(xid); + free_xid(xid); return rc; } @@ -602,50 +745,56 @@ int cifs_close(struct inode *inode, struct file *file) int cifs_closedir(struct inode *inode, struct file *file) { int rc = 0; - int xid; - struct cifsFileInfo *pCFileStruct = file->private_data; - char *ptmp; + unsigned int xid; + struct cifsFileInfo *cfile = file->private_data; + struct cifs_tcon *tcon; + struct TCP_Server_Info *server; + char *buf; - cFYI(1, "Closedir inode = 0x%p", inode); + cifs_dbg(FYI, "Closedir inode = 0x%p\n", inode); - xid = GetXid(); + if (cfile == NULL) + return rc; - if (pCFileStruct) { - struct cifs_tcon *pTcon = tlink_tcon(pCFileStruct->tlink); + xid = get_xid(); + tcon = tlink_tcon(cfile->tlink); + server = tcon->ses->server; - cFYI(1, "Freeing private data in close dir"); - spin_lock(&cifs_file_list_lock); - if (!pCFileStruct->srch_inf.endOfSearch && - !pCFileStruct->invalidHandle) { - pCFileStruct->invalidHandle = true; - spin_unlock(&cifs_file_list_lock); - rc = CIFSFindClose(xid, pTcon, pCFileStruct->netfid); - cFYI(1, "Closing uncompleted readdir with rc %d", - rc); - /* not much we can do if it fails anyway, ignore rc */ - rc = 0; - } else - spin_unlock(&cifs_file_list_lock); - ptmp = pCFileStruct->srch_inf.ntwrk_buf_start; - if (ptmp) { - cFYI(1, "closedir free smb buf in srch struct"); - pCFileStruct->srch_inf.ntwrk_buf_start = NULL; - if (pCFileStruct->srch_inf.smallBuf) - cifs_small_buf_release(ptmp); - else - cifs_buf_release(ptmp); - } - cifs_put_tlink(pCFileStruct->tlink); - kfree(file->private_data); - file->private_data = NULL; + cifs_dbg(FYI, "Freeing private data in close dir\n"); + spin_lock(&cifs_file_list_lock); + if (!cfile->srch_inf.endOfSearch && !cfile->invalidHandle) { + cfile->invalidHandle = true; + spin_unlock(&cifs_file_list_lock); + if (server->ops->close_dir) + rc = server->ops->close_dir(xid, tcon, &cfile->fid); + else + rc = -ENOSYS; + cifs_dbg(FYI, "Closing uncompleted readdir with rc %d\n", rc); + /* not much we can do if it fails anyway, ignore rc */ + rc = 0; + } else + spin_unlock(&cifs_file_list_lock); + + buf = cfile->srch_inf.ntwrk_buf_start; + if (buf) { + cifs_dbg(FYI, "closedir free smb buf in srch struct\n"); + cfile->srch_inf.ntwrk_buf_start = NULL; + if (cfile->srch_inf.smallBuf) + cifs_small_buf_release(buf); + else + cifs_buf_release(buf); } + + cifs_put_tlink(cfile->tlink); + kfree(file->private_data); + file->private_data = NULL; /* BB can we lock the filestruct while this is going on? */ - FreeXid(xid); + free_xid(xid); return rc; } static struct cifsLockInfo * -cifs_lock_init(__u64 offset, __u64 length, __u8 type, __u16 netfid) +cifs_lock_init(__u64 offset, __u64 length, __u8 type) { struct cifsLockInfo *lock = kmalloc(sizeof(struct cifsLockInfo), GFP_KERNEL); @@ -654,14 +803,13 @@ cifs_lock_init(__u64 offset, __u64 length, __u8 type, __u16 netfid) lock->offset = offset; lock->length = length; lock->type = type; - lock->netfid = netfid; lock->pid = current->tgid; INIT_LIST_HEAD(&lock->blist); init_waitqueue_head(&lock->block_q); return lock; } -static void +void cifs_del_lock_waiters(struct cifsLockInfo *lock) { struct cifsLockInfo *li, *tmp; @@ -671,35 +819,59 @@ cifs_del_lock_waiters(struct cifsLockInfo *lock) } } +#define CIFS_LOCK_OP 0 +#define CIFS_READ_OP 1 +#define CIFS_WRITE_OP 2 + +/* @rw_check : 0 - no op, 1 - read, 2 - write */ static bool -__cifs_find_lock_conflict(struct cifsInodeInfo *cinode, __u64 offset, - __u64 length, __u8 type, __u16 netfid, - struct cifsLockInfo **conf_lock) +cifs_find_fid_lock_conflict(struct cifs_fid_locks *fdlocks, __u64 offset, + __u64 length, __u8 type, struct cifsFileInfo *cfile, + struct cifsLockInfo **conf_lock, int rw_check) { - struct cifsLockInfo *li, *tmp; + struct cifsLockInfo *li; + struct cifsFileInfo *cur_cfile = fdlocks->cfile; + struct TCP_Server_Info *server = tlink_tcon(cfile->tlink)->ses->server; - list_for_each_entry_safe(li, tmp, &cinode->llist, llist) { + list_for_each_entry(li, &fdlocks->locks, llist) { if (offset + length <= li->offset || offset >= li->offset + li->length) continue; - else if ((type & LOCKING_ANDX_SHARED_LOCK) && - ((netfid == li->netfid && current->tgid == li->pid) || - type == li->type)) + if (rw_check != CIFS_LOCK_OP && current->tgid == li->pid && + server->ops->compare_fids(cfile, cur_cfile)) { + /* shared lock prevents write op through the same fid */ + if (!(li->type & server->vals->shared_lock_type) || + rw_check != CIFS_WRITE_OP) + continue; + } + if ((type & server->vals->shared_lock_type) && + ((server->ops->compare_fids(cfile, cur_cfile) && + current->tgid == li->pid) || type == li->type)) continue; - else { + if (conf_lock) *conf_lock = li; - return true; - } + return true; } return false; } -static bool -cifs_find_lock_conflict(struct cifsInodeInfo *cinode, struct cifsLockInfo *lock, - struct cifsLockInfo **conf_lock) +bool +cifs_find_lock_conflict(struct cifsFileInfo *cfile, __u64 offset, __u64 length, + __u8 type, struct cifsLockInfo **conf_lock, + int rw_check) { - return __cifs_find_lock_conflict(cinode, lock->offset, lock->length, - lock->type, lock->netfid, conf_lock); + bool rc = false; + struct cifs_fid_locks *cur; + struct cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode); + + list_for_each_entry(cur, &cinode->llist, llist) { + rc = cifs_find_fid_lock_conflict(cur, offset, length, type, + cfile, conf_lock, rw_check); + if (rc) + break; + } + + return rc; } /* @@ -710,22 +882,24 @@ cifs_find_lock_conflict(struct cifsInodeInfo *cinode, struct cifsLockInfo *lock, * the server or 1 otherwise. */ static int -cifs_lock_test(struct cifsInodeInfo *cinode, __u64 offset, __u64 length, - __u8 type, __u16 netfid, struct file_lock *flock) +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); + down_read(&cinode->lock_sem); - exist = __cifs_find_lock_conflict(cinode, offset, length, type, netfid, - &conf_lock); + exist = cifs_find_lock_conflict(cfile, offset, length, type, + &conf_lock, CIFS_LOCK_OP); 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 & LOCKING_ANDX_SHARED_LOCK) + if (conf_lock->type & server->vals->shared_lock_type) flock->fl_type = F_RDLCK; else flock->fl_type = F_WRLCK; @@ -734,16 +908,17 @@ cifs_lock_test(struct cifsInodeInfo *cinode, __u64 offset, __u64 length, else flock->fl_type = F_UNLCK; - mutex_unlock(&cinode->lock_mutex); + up_read(&cinode->lock_sem); return rc; } static void -cifs_lock_add(struct cifsInodeInfo *cinode, struct cifsLockInfo *lock) +cifs_lock_add(struct cifsFileInfo *cfile, struct cifsLockInfo *lock) { - mutex_lock(&cinode->lock_mutex); - list_add_tail(&lock->llist, &cinode->llist); - mutex_unlock(&cinode->lock_mutex); + struct cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode); + down_write(&cinode->lock_sem); + list_add_tail(&lock->llist, &cfile->llist->locks); + up_write(&cinode->lock_sem); } /* @@ -753,21 +928,23 @@ cifs_lock_add(struct cifsInodeInfo *cinode, struct cifsLockInfo *lock) * 3) -EACCESS, if there is a lock that prevents us and wait is false. */ static int -cifs_lock_add_if(struct cifsInodeInfo *cinode, struct cifsLockInfo *lock, +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); + down_write(&cinode->lock_sem); - exist = cifs_find_lock_conflict(cinode, lock, &conf_lock); + exist = cifs_find_lock_conflict(cfile, lock->offset, lock->length, + lock->type, &conf_lock, CIFS_LOCK_OP); if (!exist && cinode->can_cache_brlcks) { - list_add_tail(&lock->llist, &cinode->llist); - mutex_unlock(&cinode->lock_mutex); + list_add_tail(&lock->llist, &cfile->llist->locks); + up_write(&cinode->lock_sem); return rc; } @@ -777,17 +954,17 @@ try_again: rc = -EACCES; else { list_add_tail(&lock->blist, &conf_lock->blist); - mutex_unlock(&cinode->lock_mutex); + up_write(&cinode->lock_sem); 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); + down_write(&cinode->lock_sem); list_del_init(&lock->blist); } - mutex_unlock(&cinode->lock_mutex); + up_write(&cinode->lock_sem); return rc; } @@ -802,13 +979,13 @@ 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); + struct cifsInodeInfo *cinode = CIFS_I(file_inode(file)); unsigned char saved_type = flock->fl_type; if ((flock->fl_flags & FL_POSIX) == 0) return 1; - mutex_lock(&cinode->lock_mutex); + down_read(&cinode->lock_sem); posix_test_lock(file, flock); if (flock->fl_type == F_UNLCK && !cinode->can_cache_brlcks) { @@ -816,7 +993,7 @@ cifs_posix_lock_test(struct file *file, struct file_lock *flock) rc = 1; } - mutex_unlock(&cinode->lock_mutex); + up_read(&cinode->lock_sem); return rc; } @@ -829,58 +1006,68 @@ cifs_posix_lock_test(struct file *file, struct file_lock *flock) static int cifs_posix_lock_set(struct file *file, struct file_lock *flock) { - struct cifsInodeInfo *cinode = CIFS_I(file->f_path.dentry->d_inode); + struct cifsInodeInfo *cinode = CIFS_I(file_inode(file)); int rc = 1; if ((flock->fl_flags & FL_POSIX) == 0) return rc; - mutex_lock(&cinode->lock_mutex); +try_again: + down_write(&cinode->lock_sem); if (!cinode->can_cache_brlcks) { - mutex_unlock(&cinode->lock_mutex); + up_write(&cinode->lock_sem); return rc; } - rc = posix_lock_file_wait(file, flock); - mutex_unlock(&cinode->lock_mutex); + + rc = posix_lock_file(file, flock, NULL); + up_write(&cinode->lock_sem); + if (rc == FILE_LOCK_DEFERRED) { + rc = wait_event_interruptible(flock->fl_wait, !flock->fl_next); + if (!rc) + goto try_again; + posix_unblock_lock(flock); + } return rc; } -static int +int cifs_push_mandatory_locks(struct cifsFileInfo *cfile) { - int xid, rc = 0, stored_rc; + 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; + 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 = GetXid(); + xid = get_xid(); tcon = tlink_tcon(cfile->tlink); - mutex_lock(&cinode->lock_mutex); - if (!cinode->can_cache_brlcks) { - mutex_unlock(&cinode->lock_mutex); - FreeXid(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) { + free_xid(xid); + return -EINVAL; } - max_num = (tcon->ses->server->maxBuf - sizeof(struct smb_hdr)) / - sizeof(LOCKING_ANDX_RANGE); + 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); - FreeXid(xid); - return rc; + free_xid(xid); + return -ENOMEM; } for (i = 0; i < 2; i++) { cur = buf; num = 0; - list_for_each_entry_safe(li, tmp, &cinode->llist, llist) { + list_for_each_entry_safe(li, tmp, &cfile->llist->locks, llist) { if (li->type != types[i]) continue; cur->Pid = cpu_to_le16(li->pid); @@ -889,8 +1076,10 @@ cifs_push_mandatory_locks(struct cifsFileInfo *cfile) 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->netfid, - li->type, 0, num, buf); + stored_rc = cifs_lockv(xid, tcon, + cfile->fid.netfid, + (__u8)li->type, 0, num, + buf); if (stored_rc) rc = stored_rc; cur = buf; @@ -900,18 +1089,15 @@ cifs_push_mandatory_locks(struct cifsFileInfo *cfile) } if (num) { - stored_rc = cifs_lockv(xid, tcon, cfile->netfid, - types[i], 0, num, buf); + 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); - FreeXid(xid); + free_xid(xid); return rc; } @@ -920,58 +1106,86 @@ cifs_push_mandatory_locks(struct cifsFileInfo *cfile) 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 inode *inode = cfile->dentry->d_inode; struct cifs_tcon *tcon = tlink_tcon(cfile->tlink); struct file_lock *flock, **before; - struct cifsLockInfo *lck, *tmp; + 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; - struct list_head locks_to_send; - xid = GetXid(); + xid = get_xid(); - mutex_lock(&cinode->lock_mutex); - if (!cinode->can_cache_brlcks) { - mutex_unlock(&cinode->lock_mutex); - FreeXid(xid); - return rc; + spin_lock(&inode->i_lock); + cifs_for_each_lock(inode, before) { + if ((*before)->fl_flags & FL_POSIX) + count++; } + spin_unlock(&inode->i_lock); INIT_LIST_HEAD(&locks_to_send); - lock_flocks(); - cifs_for_each_lock(cfile->dentry->d_inode, before) { + /* + * Allocating count locks is enough because no FL_POSIX locks can be + * added to the list while we are holding cinode->lock_sem 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; + spin_lock(&inode->i_lock); + cifs_for_each_lock(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. + */ + cifs_dbg(VFS, "Can't push all brlocks!\n"); + 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 = cifs_lock_init(flock->fl_start, length, type, - cfile->netfid); - if (!lck) { - rc = -ENOMEM; - goto send_locks; - } + lck = list_entry(el, struct lock_to_push, llist); lck->pid = flock->fl_pid; - - list_add_tail(&lck->llist, &locks_to_send); + lck->netfid = cfile->fid.netfid; + lck->length = length; + lck->type = type; + lck->offset = flock->fl_start; + el = el->next; } - -send_locks: - unlock_flocks(); + spin_unlock(&inode->i_lock); list_for_each_entry_safe(lck, tmp, &locks_to_send, llist) { - struct file_lock tmp_lock; int stored_rc; - tmp_lock.fl_start = lck->offset; stored_rc = CIFSSMBPosixLock(xid, tcon, lck->netfid, lck->pid, - 0, lck->length, &tmp_lock, + lck->offset, lck->length, NULL, lck->type, 0); if (stored_rc) rc = stored_rc; @@ -979,81 +1193,101 @@ send_locks: kfree(lck); } - cinode->can_cache_brlcks = false; - mutex_unlock(&cinode->lock_mutex); - - FreeXid(xid); +out: + 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 cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode); struct cifs_tcon *tcon = tlink_tcon(cfile->tlink); + int rc = 0; - if ((tcon->ses->capabilities & CAP_UNIX) && + /* we are going to update can_cache_brlcks here - need a write access */ + down_write(&cinode->lock_sem); + if (!cinode->can_cache_brlcks) { + up_write(&cinode->lock_sem); + return rc; + } + + 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); + rc = cifs_push_posix_locks(cfile); + else + rc = tcon->ses->server->ops->push_mand_locks(cfile); - return cifs_push_mandatory_locks(cfile); + cinode->can_cache_brlcks = false; + up_write(&cinode->lock_sem); + return rc; } static void -cifs_read_flock(struct file_lock *flock, __u8 *type, int *lock, int *unlock, - bool *wait_flag) +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"); + cifs_dbg(FYI, "Posix\n"); if (flock->fl_flags & FL_FLOCK) - cFYI(1, "Flock"); + cifs_dbg(FYI, "Flock\n"); if (flock->fl_flags & FL_SLEEP) { - cFYI(1, "Blocking lock"); + cifs_dbg(FYI, "Blocking lock\n"); *wait_flag = true; } if (flock->fl_flags & FL_ACCESS) - cFYI(1, "Process suspended by mandatory locking - " - "not implemented yet"); + cifs_dbg(FYI, "Process suspended by mandatory locking - not implemented yet\n"); if (flock->fl_flags & FL_LEASE) - cFYI(1, "Lease on file - not implemented yet"); + cifs_dbg(FYI, "Lease on file - not implemented yet\n"); if (flock->fl_flags & - (~(FL_POSIX | FL_FLOCK | FL_SLEEP | FL_ACCESS | FL_LEASE))) - cFYI(1, "Unknown lock flags 0x%x", flock->fl_flags); + (~(FL_POSIX | FL_FLOCK | FL_SLEEP | + FL_ACCESS | FL_LEASE | FL_CLOSE))) + cifs_dbg(FYI, "Unknown lock flags 0x%x\n", flock->fl_flags); - *type = LOCKING_ANDX_LARGE_FILES; + *type = server->vals->large_lock_type; if (flock->fl_type == F_WRLCK) { - cFYI(1, "F_WRLCK "); + cifs_dbg(FYI, "F_WRLCK\n"); + *type |= server->vals->exclusive_lock_type; *lock = 1; } else if (flock->fl_type == F_UNLCK) { - cFYI(1, "F_UNLCK"); + cifs_dbg(FYI, "F_UNLCK\n"); + *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 |= LOCKING_ANDX_SHARED_LOCK; + cifs_dbg(FYI, "F_RDLCK\n"); + *type |= server->vals->shared_lock_type; *lock = 1; } else if (flock->fl_type == F_EXLCK) { - cFYI(1, "F_EXLCK"); + cifs_dbg(FYI, "F_EXLCK\n"); + *type |= server->vals->exclusive_lock_type; *lock = 1; } else if (flock->fl_type == F_SHLCK) { - cFYI(1, "F_SHLCK"); - *type |= LOCKING_ANDX_SHARED_LOCK; + cifs_dbg(FYI, "F_SHLCK\n"); + *type |= server->vals->shared_lock_type; *lock = 1; } else - cFYI(1, "Unknown type of lock"); + cifs_dbg(FYI, "Unknown type of lock\n"); } static int -cifs_getlk(struct file *file, struct file_lock *flock, __u8 type, - bool wait_flag, bool posix_lck, int xid) +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 cifsInodeInfo *cinode = CIFS_I(cfile->dentry->d_inode); - __u16 netfid = cfile->netfid; + struct TCP_Server_Info *server = tcon->ses->server; + __u16 netfid = cfile->fid.netfid; if (posix_lck) { int posix_lock_type; @@ -1062,59 +1296,57 @@ cifs_getlk(struct file *file, struct file_lock *flock, __u8 type, if (!rc) return rc; - if (type & LOCKING_ANDX_SHARED_LOCK) + 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, - 1 /* get */, length, flock, + flock->fl_start, length, flock, posix_lock_type, wait_flag); return rc; } - rc = cifs_lock_test(cinode, flock->fl_start, length, type, netfid, - flock); + 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 = CIFSSMBLock(xid, tcon, netfid, current->tgid, length, - flock->fl_start, 0, 1, type, 0, 0); + rc = server->ops->mand_lock(xid, cfile, flock->fl_start, length, type, + 1, 0, false); if (rc == 0) { - rc = CIFSSMBLock(xid, tcon, netfid, current->tgid, - length, flock->fl_start, 1, 0, - type, 0, 0); + rc = server->ops->mand_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); + cifs_dbg(VFS, "Error unlocking previously locked range %d during test of lock\n", + rc); return 0; } - if (type & LOCKING_ANDX_SHARED_LOCK) { + if (type & server->vals->shared_lock_type) { flock->fl_type = F_WRLCK; return 0; } - rc = CIFSSMBLock(xid, tcon, netfid, current->tgid, length, - flock->fl_start, 0, 1, - type | LOCKING_ANDX_SHARED_LOCK, 0, 0); + type &= ~server->vals->exclusive_lock_type; + + rc = server->ops->mand_lock(xid, cfile, flock->fl_start, length, + type | server->vals->shared_lock_type, + 1, 0, false); if (rc == 0) { - rc = CIFSSMBLock(xid, tcon, netfid, current->tgid, - length, flock->fl_start, 1, 0, - type | LOCKING_ANDX_SHARED_LOCK, - 0, 0); + rc = server->ops->mand_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); + cifs_dbg(VFS, "Error unlocking previously locked range %d during test of lock\n", + rc); } else flock->fl_type = F_WRLCK; return 0; } -static void +void cifs_move_llist(struct list_head *source, struct list_head *dest) { struct list_head *li, *tmp; @@ -1122,7 +1354,7 @@ cifs_move_llist(struct list_head *source, struct list_head *dest) list_move(li, dest); } -static void +void cifs_free_llist(struct list_head *llist) { struct cifsLockInfo *li, *tmp; @@ -1133,14 +1365,15 @@ cifs_free_llist(struct list_head *llist) } } -static int -cifs_unlock_range(struct cifsFileInfo *cfile, struct file_lock *flock, int xid) +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; + 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); @@ -1150,102 +1383,106 @@ cifs_unlock_range(struct cifsFileInfo *cfile, struct file_lock *flock, int xid) INIT_LIST_HEAD(&tmp_llist); - max_num = (tcon->ses->server->maxBuf - sizeof(struct smb_hdr)) / - sizeof(LOCKING_ANDX_RANGE); + /* + * 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); + down_write(&cinode->lock_sem); for (i = 0; i < 2; i++) { cur = buf; num = 0; - list_for_each_entry_safe(li, tmp, &cinode->llist, llist) { + list_for_each_entry_safe(li, tmp, &cfile->llist->locks, llist) { if (flock->fl_start > li->offset || (flock->fl_start + length) < (li->offset + li->length)) continue; if (current->tgid != li->pid) continue; - if (cfile->netfid != li->netfid) - continue; if (types[i] != li->type) continue; - if (!cinode->can_cache_brlcks) { - 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 inode 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->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 inode list. - */ - cifs_move_llist(&tmp_llist, - &cinode->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++; - } else { + if (cinode->can_cache_brlcks) { /* * We can cache brlock requests - simply remove - * a lock from the inode list. + * 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->locks); + 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->netfid, + stored_rc = cifs_lockv(xid, tcon, cfile->fid.netfid, types[i], num, 0, buf); if (stored_rc) { - cifs_move_llist(&tmp_llist, &cinode->llist); + cifs_move_llist(&tmp_llist, + &cfile->llist->locks); rc = stored_rc; } else cifs_free_llist(&tmp_llist); } } - mutex_unlock(&cinode->lock_mutex); + up_write(&cinode->lock_sem); kfree(buf); return rc; } static int -cifs_setlk(struct file *file, struct file_lock *flock, __u8 type, - bool wait_flag, bool posix_lck, int lock, int unlock, int xid) +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 cifsInodeInfo *cinode = CIFS_I(file->f_path.dentry->d_inode); - __u16 netfid = cfile->netfid; + struct TCP_Server_Info *server = tcon->ses->server; + struct inode *inode = cfile->dentry->d_inode; if (posix_lck) { int posix_lock_type; @@ -1254,7 +1491,7 @@ cifs_setlk(struct file *file, struct file_lock *flock, __u8 type, if (!rc || rc < 0) return rc; - if (type & LOCKING_ANDX_SHARED_LOCK) + if (type & server->vals->shared_lock_type) posix_lock_type = CIFS_RDLCK; else posix_lock_type = CIFS_WRLCK; @@ -1262,35 +1499,52 @@ cifs_setlk(struct file *file, struct file_lock *flock, __u8 type, if (unlock == 1) posix_lock_type = CIFS_UNLCK; - rc = CIFSSMBPosixLock(xid, tcon, netfid, current->tgid, - 0 /* set */, length, flock, - posix_lock_type, wait_flag); + rc = CIFSSMBPosixLock(xid, tcon, cfile->fid.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, netfid); + lock = cifs_lock_init(flock->fl_start, length, type); if (!lock) return -ENOMEM; - rc = cifs_lock_add_if(cinode, lock, wait_flag); - if (rc < 0) + rc = cifs_lock_add_if(cfile, lock, wait_flag); + if (rc < 0) { kfree(lock); - if (rc <= 0) + return rc; + } + if (!rc) goto out; - rc = CIFSSMBLock(xid, tcon, netfid, current->tgid, length, - flock->fl_start, 0, 1, type, wait_flag, 0); + /* + * Windows 7 server can delay breaking lease from read to None + * if we set a byte-range lock on a file - break it explicitly + * before sending the lock to the server to be sure the next + * read won't conflict with non-overlapted locks due to + * pagereading. + */ + if (!CIFS_CACHE_WRITE(CIFS_I(inode)) && + CIFS_CACHE_READ(CIFS_I(inode))) { + cifs_zap_mapping(inode); + cifs_dbg(FYI, "Set no oplock for inode=%p due to mand locks\n", + inode); + CIFS_I(inode)->oplock = 0; + } + + rc = server->ops->mand_lock(xid, cfile, flock->fl_start, length, + type, 1, 0, wait_flag); if (rc) { kfree(lock); - goto out; + return rc; } - cifs_lock_add(cinode, lock); + cifs_lock_add(cfile, lock); } else if (unlock) - rc = cifs_unlock_range(cfile, flock, xid); + rc = server->ops->mand_unlock_range(cfile, flock, xid); out: if (flock->fl_flags & FL_POSIX) @@ -1309,24 +1563,26 @@ int cifs_lock(struct file *file, int cmd, struct file_lock *flock) struct cifsInodeInfo *cinode; struct cifsFileInfo *cfile; __u16 netfid; - __u8 type; + __u32 type; rc = -EACCES; - xid = GetXid(); + 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); + cifs_dbg(FYI, "Lock parm: 0x%x flockflags: 0x%x flocktype: 0x%x start: %lld end: %lld\n", + cmd, flock->fl_flags, flock->fl_type, + flock->fl_start, flock->fl_end); - cifs_read_flock(flock, &type, &lock, &unlock, &wait_flag); - - cifs_sb = CIFS_SB(file->f_path.dentry->d_sb); cfile = (struct cifsFileInfo *)file->private_data; tcon = tlink_tcon(cfile->tlink); - netfid = cfile->netfid; - cinode = CIFS_I(file->f_path.dentry->d_inode); - if ((tcon->ses->capabilities & CAP_UNIX) && + 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_inode(file)); + + 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; @@ -1336,7 +1592,7 @@ int cifs_lock(struct file *file, int cmd, struct file_lock *flock) */ if (IS_GETLK(cmd)) { rc = cifs_getlk(file, flock, type, wait_flag, posix_lck, xid); - FreeXid(xid); + free_xid(xid); return rc; } @@ -1345,17 +1601,20 @@ int cifs_lock(struct file *file, int cmd, struct file_lock *flock) * if no lock or unlock then nothing to do since we do not * know what it is */ - FreeXid(xid); + free_xid(xid); return -EOPNOTSUPP; } rc = cifs_setlk(file, flock, type, wait_flag, posix_lck, lock, unlock, xid); - FreeXid(xid); + free_xid(xid); return rc; } -/* update the file size (if needed) after a write */ +/* + * 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) @@ -1366,28 +1625,33 @@ cifs_update_eof(struct cifsInodeInfo *cifsi, loff_t offset, 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) +static ssize_t +cifs_write(struct cifsFileInfo *open_file, __u32 pid, const char *write_data, + size_t write_size, loff_t *offset) { int rc = 0; unsigned int bytes_written = 0; unsigned int total_written; struct cifs_sb_info *cifs_sb; - struct cifs_tcon *pTcon; - int xid; + struct cifs_tcon *tcon; + struct TCP_Server_Info *server; + 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); + cifs_dbg(FYI, "write %zd bytes to offset %lld of %s\n", + write_size, *offset, dentry->d_name.name); + + tcon = tlink_tcon(open_file->tlink); + server = tcon->ses->server; - pTcon = tlink_tcon(open_file->tlink); + if (!server->ops->sync_write) + return -ENOSYS; - xid = GetXid(); + xid = get_xid(); for (total_written = 0; write_size > total_written; total_written += bytes_written) { @@ -1411,37 +1675,38 @@ static ssize_t cifs_write(struct cifsFileInfo *open_file, __u32 pid, /* 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->netfid; io_parms.pid = pid; - io_parms.tcon = pTcon; - io_parms.offset = *poffset; + io_parms.tcon = tcon; + io_parms.offset = *offset; io_parms.length = len; - rc = CIFSSMBWrite2(xid, &io_parms, &bytes_written, iov, - 1, 0); + rc = server->ops->sync_write(xid, open_file, &io_parms, + &bytes_written, iov, 1); } if (rc || (bytes_written == 0)) { if (total_written) break; else { - FreeXid(xid); + free_xid(xid); return rc; } } else { - cifs_update_eof(cifsi, *poffset, bytes_written); - *poffset += bytes_written; + spin_lock(&dentry->d_inode->i_lock); + cifs_update_eof(cifsi, *offset, bytes_written); + spin_unlock(&dentry->d_inode->i_lock); + *offset += bytes_written; } } - cifs_stats_bytes_written(pTcon, total_written); + cifs_stats_bytes_written(tcon, 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); + if (*offset > dentry->d_inode->i_size) + i_size_write(dentry->d_inode, *offset); spin_unlock(&dentry->d_inode->i_lock); } mark_inode_dirty_sync(dentry->d_inode); - FreeXid(xid); + free_xid(xid); return total_written; } @@ -1460,13 +1725,13 @@ struct cifsFileInfo *find_readable_file(struct cifsInodeInfo *cifs_inode, 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()) + if (fsuid_only && !uid_eq(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(open_file); + cifsFileInfo_get_locked(open_file); spin_unlock(&cifs_file_list_lock); return open_file; } /* else might as well continue, and look for @@ -1482,17 +1747,18 @@ struct cifsFileInfo *find_readable_file(struct cifsInodeInfo *cifs_inode, struct cifsFileInfo *find_writable_file(struct cifsInodeInfo *cifs_inode, bool fsuid_only) { - struct cifsFileInfo *open_file; + 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"); + cifs_dbg(VFS, "Null inode passed to cifs_writeable_file\n"); dump_stack(); return NULL; } @@ -1505,40 +1771,25 @@ struct cifsFileInfo *find_writable_file(struct cifsInodeInfo *cifs_inode, 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()) + if (fsuid_only && !uid_eq(open_file->uid, current_fsuid())) continue; if (OPEN_FMODE(open_file->f_flags) & FMODE_WRITE) { - cifsFileInfo_get(open_file); - 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; } - - spin_unlock(&cifs_file_list_lock); - - /* Had to unlock since following call can block */ - rc = cifs_reopen_file(open_file, false); - if (!rc) - return open_file; - - /* if it fails, try another handle if possible */ - cFYI(1, "wp failed on reopen file"); - cifsFileInfo_put(open_file); - - spin_lock(&cifs_file_list_lock); - - /* else we simply continue to the next entry. Thus - we do not loop on reopen errors. If we - can not reopen the file, for example if we - reconnected to a server with another client - racing to delete or lock the file we would not - make progress if we restarted before the beginning - of the loop here. */ } } /* couldn't find useable FH with same pid, try any available */ @@ -1546,7 +1797,30 @@ refind_writable: 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; } @@ -1596,7 +1870,7 @@ static int cifs_partialpagewrite(struct page *page, unsigned from, unsigned to) else if (bytes_written < 0) rc = bytes_written; } else { - cFYI(1, "No writeable filehandles for inode"); + cifs_dbg(FYI, "No writeable filehandles for inode\n"); rc = -EIO; } @@ -1611,6 +1885,7 @@ static int cifs_writepages(struct address_space *mapping, bool done = false, scanned = false, range_whole = false; pgoff_t end, index; struct cifs_writedata *wdata; + struct TCP_Server_Info *server; struct page *page; int rc = 0; @@ -1640,7 +1915,8 @@ retry: tofind = min((cifs_sb->wsize / PAGE_CACHE_SIZE) - 1, end - index) + 1; - wdata = cifs_writedata_alloc((unsigned int)tofind); + wdata = cifs_writedata_alloc((unsigned int)tofind, + cifs_writev_complete); if (!wdata) { rc = -ENOMEM; break; @@ -1716,7 +1992,7 @@ retry: */ set_page_writeback(page); - if (page_offset(page) >= mapping->host->i_size) { + if (page_offset(page) >= i_size_read(mapping->host)) { done = true; unlock_page(page); end_page_writeback(page); @@ -1747,6 +2023,13 @@ retry: wdata->sync_mode = wbc->sync_mode; wdata->nr_pages = nr_pages; wdata->offset = page_offset(wdata->pages[0]); + wdata->pagesz = PAGE_CACHE_SIZE; + wdata->tailsz = + min(i_size_read(mapping->host) - + page_offset(wdata->pages[nr_pages - 1]), + (loff_t)PAGE_CACHE_SIZE); + wdata->bytes = ((nr_pages - 1) * PAGE_CACHE_SIZE) + + wdata->tailsz; do { if (wdata->cfile != NULL) @@ -1754,11 +2037,14 @@ retry: wdata->cfile = find_writable_file(CIFS_I(mapping->host), false); if (!wdata->cfile) { - cERROR(1, "No writable handles for inode"); + cifs_dbg(VFS, "No writable handles for inode\n"); rc = -EBADF; break; } - rc = cifs_async_writev(wdata); + wdata->pid = wdata->cfile->pid; + server = tlink_tcon(wdata->cfile->tlink)->ses->server; + rc = server->ops->async_writev(wdata, + cifs_writedata_release); } while (wbc->sync_mode == WB_SYNC_ALL && rc == -EAGAIN); for (i = 0; i < nr_pages; ++i) @@ -1807,13 +2093,13 @@ static int cifs_writepage_locked(struct page *page, struct writeback_control *wbc) { int rc; - int xid; + unsigned int xid; - xid = GetXid(); + xid = get_xid(); /* BB add check for wbc flags */ page_cache_get(page); if (!PageUptodate(page)) - cFYI(1, "ppw - page not up to date"); + cifs_dbg(FYI, "ppw - page not up to date\n"); /* * Set the "writeback" flag, and clear "dirty" in the radix tree. @@ -1838,7 +2124,7 @@ retry_write: SetPageUptodate(page); end_page_writeback(page); page_cache_release(page); - FreeXid(xid); + free_xid(xid); return rc; } @@ -1864,7 +2150,7 @@ static int cifs_write_end(struct file *file, struct address_space *mapping, else pid = current->tgid; - cFYI(1, "write_end for page %p from pos %lld with %d bytes", + cifs_dbg(FYI, "write_end for page %p from pos %lld with %d bytes\n", page, pos, copied); if (PageChecked(page)) { @@ -1877,9 +2163,9 @@ static int cifs_write_end(struct file *file, struct address_space *mapping, if (!PageUptodate(page)) { char *page_data; unsigned offset = pos & (PAGE_CACHE_SIZE - 1); - int xid; + unsigned int xid; - xid = GetXid(); + 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 */ @@ -1890,7 +2176,7 @@ static int cifs_write_end(struct file *file, struct address_space *mapping, /* if (rc < 0) should we set writebehind rc? */ kunmap(page); - FreeXid(xid); + free_xid(xid); } else { rc = copied; pos += copied; @@ -1913,11 +2199,12 @@ static int cifs_write_end(struct file *file, struct address_space *mapping, int cifs_strict_fsync(struct file *file, loff_t start, loff_t end, int datasync) { - int xid; + 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 inode *inode = file_inode(file); struct cifs_sb_info *cifs_sb = CIFS_SB(inode->i_sb); rc = filemap_write_and_wait_range(inode->i_mapping, start, end); @@ -1925,33 +2212,39 @@ int cifs_strict_fsync(struct file *file, loff_t start, loff_t end, return rc; mutex_lock(&inode->i_mutex); - xid = GetXid(); + xid = get_xid(); - cFYI(1, "Sync file - name: %s datasync: 0x%x", - file->f_path.dentry->d_name.name, datasync); + cifs_dbg(FYI, "Sync file - name: %s datasync: 0x%x\n", + file->f_path.dentry->d_name.name, datasync); - if (!CIFS_I(inode)->clientCanCacheRead) { - rc = cifs_invalidate_mapping(inode); + if (!CIFS_CACHE_READ(CIFS_I(inode))) { + rc = cifs_zap_mapping(inode); if (rc) { - cFYI(1, "rc: %d during invalidate phase", rc); + cifs_dbg(FYI, "rc: %d during invalidate phase\n", rc); rc = 0; /* don't care about it in fsync */ } } tcon = tlink_tcon(smbfile->tlink); - if (!(cifs_sb->mnt_cifs_flags & CIFS_MOUNT_NOSSYNC)) - rc = CIFSSMBFlush(xid, tcon, smbfile->netfid); + 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; + } - FreeXid(xid); + free_xid(xid); mutex_unlock(&inode->i_mutex); return rc; } int cifs_fsync(struct file *file, loff_t start, loff_t end, int datasync) { - int xid; + 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; @@ -1961,16 +2254,21 @@ int cifs_fsync(struct file *file, loff_t start, loff_t end, int datasync) return rc; mutex_lock(&inode->i_mutex); - xid = GetXid(); + xid = get_xid(); - cFYI(1, "Sync file - name: %s datasync: 0x%x", - file->f_path.dentry->d_name.name, datasync); + cifs_dbg(FYI, "Sync file - name: %s datasync: 0x%x\n", + file->f_path.dentry->d_name.name, datasync); tcon = tlink_tcon(smbfile->tlink); - if (!(cifs_sb->mnt_cifs_flags & CIFS_MOUNT_NOSSYNC)) - rc = CIFSSMBFlush(xid, tcon, smbfile->netfid); + 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; + } - FreeXid(xid); + free_xid(xid); mutex_unlock(&inode->i_mutex); return rc; } @@ -1981,13 +2279,13 @@ int cifs_fsync(struct file *file, loff_t start, loff_t end, int datasync) */ int cifs_flush(struct file *file, fl_owner_t id) { - struct inode *inode = file->f_path.dentry->d_inode; + struct inode *inode = file_inode(file); 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); + cifs_dbg(FYI, "Flush inode %p file %p rc %d\n", inode, file, rc); return rc; } @@ -1999,7 +2297,7 @@ cifs_write_allocate_pages(struct page **pages, unsigned long num_pages) unsigned long i; for (i = 0; i < num_pages; i++) { - pages[i] = alloc_page(__GFP_HIGHMEM); + pages[i] = alloc_page(GFP_KERNEL|__GFP_HIGHMEM); if (!pages[i]) { /* * save number of pages we have already allocated and @@ -2007,15 +2305,14 @@ cifs_write_allocate_pages(struct page **pages, unsigned long num_pages) */ num_pages = i; rc = -ENOMEM; - goto error; + break; } } - return rc; - -error: - for (i = 0; i < num_pages; i++) - put_page(pages[i]); + if (rc) { + for (i = 0; i < num_pages; i++) + put_page(pages[i]); + } return rc; } @@ -2026,9 +2323,7 @@ size_t get_numpages(const size_t wsize, const size_t len, size_t *cur_len) size_t clen; clen = min_t(const size_t, len, wsize); - num_pages = clen / PAGE_CACHE_SIZE; - if (clen % PAGE_CACHE_SIZE) - num_pages++; + num_pages = DIV_ROUND_UP(clen, PAGE_SIZE); if (cur_len) *cur_len = clen; @@ -2036,139 +2331,226 @@ size_t get_numpages(const size_t wsize, const size_t len, size_t *cur_len) return num_pages; } +static void +cifs_uncached_writedata_release(struct kref *refcount) +{ + int i; + struct cifs_writedata *wdata = container_of(refcount, + struct cifs_writedata, refcount); + + for (i = 0; i < wdata->nr_pages; i++) + put_page(wdata->pages[i]); + cifs_writedata_release(refcount); +} + +static void +cifs_uncached_writev_complete(struct work_struct *work) +{ + 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); + + kref_put(&wdata->refcount, cifs_uncached_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; + struct TCP_Server_Info *server; + + server = tlink_tcon(wdata->cfile->tlink)->ses->server; + + do { + if (wdata->cfile->invalidHandle) { + rc = cifs_reopen_file(wdata->cfile, false); + if (rc != 0) + continue; + } + rc = server->ops->async_writev(wdata, + cifs_uncached_writedata_release); + } 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) +cifs_iovec_write(struct file *file, struct iov_iter *from, loff_t *poffset) { - unsigned int written; - unsigned long num_pages, npages, i; - size_t copied, len, cur_len; + unsigned long nr_pages, i; + size_t bytes, copied, len, cur_len; ssize_t total_written = 0; - struct kvec *to_send; - struct page **pages; - struct iov_iter it; - struct inode *inode; + loff_t offset; struct cifsFileInfo *open_file; - struct cifs_tcon *pTcon; + struct cifs_tcon *tcon; struct cifs_sb_info *cifs_sb; - struct cifs_io_parms io_parms; - int xid, rc; - __u32 pid; - - len = iov_length(iov, nr_segs); - if (!len) - return 0; + struct cifs_writedata *wdata, *tmp; + struct list_head wdata_list; + int rc; + pid_t pid; + len = iov_iter_count(from); rc = generic_write_checks(file, poffset, &len, 0); if (rc) return rc; - cifs_sb = CIFS_SB(file->f_path.dentry->d_sb); - num_pages = get_numpages(cifs_sb->wsize, len, &cur_len); + if (!len) + return 0; - pages = kmalloc(sizeof(struct pages *)*num_pages, GFP_KERNEL); - if (!pages) - return -ENOMEM; + iov_iter_truncate(from, len); - to_send = kmalloc(sizeof(struct kvec)*(num_pages + 1), GFP_KERNEL); - if (!to_send) { - kfree(pages); - return -ENOMEM; - } + 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); - rc = cifs_write_allocate_pages(pages, num_pages); - if (rc) { - kfree(pages); - kfree(to_send); - return rc; - } + if (!tcon->ses->server->ops->async_writev) + return -ENOSYS; - xid = GetXid(); - open_file = file->private_data; + offset = *poffset; if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_RWPIDFORWARD) pid = open_file->pid; else pid = current->tgid; - pTcon = tlink_tcon(open_file->tlink); - inode = file->f_path.dentry->d_inode; + do { + size_t save_len; - iov_iter_init(&it, iov, nr_segs, len, 0); - npages = num_pages; + 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; + } - do { - size_t save_len = cur_len; - for (i = 0; i < npages; i++) { - copied = min_t(const size_t, cur_len, PAGE_CACHE_SIZE); - copied = iov_iter_copy_from_user(pages[i], &it, 0, - copied); - cur_len -= copied; - iov_iter_advance(&it, copied); - to_send[i+1].iov_base = kmap(pages[i]); - to_send[i+1].iov_len = copied; + 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++) { + bytes = min_t(size_t, cur_len, PAGE_SIZE); + copied = copy_page_from_iter(wdata->pages[i], 0, bytes, + from); + cur_len -= copied; + /* + * If we didn't copy as much as we expected, then that + * may mean we trod into an unmapped area. Stop copying + * at that point. On the next pass through the big + * loop, we'll likely end up getting a zero-length + * write and bailing out of it. + */ + if (copied < bytes) + break; + } cur_len = save_len - cur_len; - do { - if (open_file->invalidHandle) { - rc = cifs_reopen_file(open_file, false); - if (rc != 0) - break; - } - io_parms.netfid = open_file->netfid; - io_parms.pid = pid; - io_parms.tcon = pTcon; - io_parms.offset = *poffset; - io_parms.length = cur_len; - rc = CIFSSMBWrite2(xid, &io_parms, &written, to_send, - npages, 0); - } while (rc == -EAGAIN); - - for (i = 0; i < npages; i++) - kunmap(pages[i]); - - if (written) { - len -= written; - total_written += written; - cifs_update_eof(CIFS_I(inode), *poffset, written); - *poffset += written; - } else if (rc < 0) { - if (!total_written) - total_written = rc; + /* + * If we have no data to send, then that probably means that + * the copy above failed altogether. That's most likely because + * the address in the iovec was bogus. Set the rc to -EFAULT, + * free anything we allocated and bail out. + */ + if (!cur_len) { + for (i = 0; i < nr_pages; i++) + put_page(wdata->pages[i]); + kfree(wdata); + rc = -EFAULT; + break; + } + + /* + * i + 1 now represents the number of pages we actually used in + * the copy phase above. Bring nr_pages down to that, and free + * any pages that we didn't use. + */ + for ( ; nr_pages > i + 1; nr_pages--) + put_page(wdata->pages[nr_pages - 1]); + + 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->pagesz = PAGE_SIZE; + wdata->tailsz = cur_len - ((nr_pages - 1) * PAGE_SIZE); + rc = cifs_uncached_retry_writev(wdata); + if (rc) { + kref_put(&wdata->refcount, + cifs_uncached_writedata_release); break; } - /* get length and number of kvecs of the next write */ - npages = get_numpages(cifs_sb->wsize, len, &cur_len); + list_add_tail(&wdata->list, &wdata_list); + offset += cur_len; + len -= cur_len; } while (len > 0); - if (total_written > 0) { - spin_lock(&inode->i_lock); - if (*poffset > inode->i_size) - i_size_write(inode, *poffset); - spin_unlock(&inode->i_lock); + /* + * 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_uncached_writedata_release); } - cifs_stats_bytes_written(pTcon, total_written); - mark_inode_dirty_sync(inode); + if (total_written > 0) + *poffset += total_written; - for (i = 0; i < num_pages; i++) - put_page(pages[i]); - kfree(to_send); - kfree(pages); - FreeXid(xid); - return 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 cifs_user_writev(struct kiocb *iocb, struct iov_iter *from) { ssize_t written; struct inode *inode; + loff_t pos = iocb->ki_pos; - inode = iocb->ki_filp->f_path.dentry->d_inode; + inode = file_inode(iocb->ki_filp); /* * BB - optimize the way when signing is disabled. We can drop this @@ -2176,69 +2558,300 @@ ssize_t cifs_user_writev(struct kiocb *iocb, const struct iovec *iov, * write request. */ - written = cifs_iovec_write(iocb->ki_filp, iov, nr_segs, &pos); + written = cifs_iovec_write(iocb->ki_filp, from, &pos); if (written > 0) { - CIFS_I(inode)->invalid_mapping = true; + set_bit(CIFS_INO_INVALID_MAPPING, &CIFS_I(inode)->flags); 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) +static ssize_t +cifs_writev(struct kiocb *iocb, struct iov_iter *from) { - struct inode *inode; + struct file *file = iocb->ki_filp; + struct cifsFileInfo *cfile = (struct cifsFileInfo *)file->private_data; + struct inode *inode = file->f_mapping->host; + struct cifsInodeInfo *cinode = CIFS_I(inode); + struct TCP_Server_Info *server = tlink_tcon(cfile->tlink)->ses->server; + ssize_t rc = -EACCES; + loff_t lock_pos = iocb->ki_pos; + + /* + * We need to hold the sem to be sure nobody modifies lock list + * with a brlock that prevents writing. + */ + down_read(&cinode->lock_sem); + mutex_lock(&inode->i_mutex); + if (file->f_flags & O_APPEND) + lock_pos = i_size_read(inode); + if (!cifs_find_lock_conflict(cfile, lock_pos, iov_iter_count(from), + server->vals->exclusive_lock_type, NULL, + CIFS_WRITE_OP)) { + rc = __generic_file_write_iter(iocb, from); + mutex_unlock(&inode->i_mutex); + + if (rc > 0) { + ssize_t err; + + err = generic_write_sync(file, iocb->ki_pos - rc, rc); + if (err < 0) + rc = err; + } + } else { + mutex_unlock(&inode->i_mutex); + } + up_read(&cinode->lock_sem); + return rc; +} - inode = iocb->ki_filp->f_path.dentry->d_inode; +ssize_t +cifs_strict_writev(struct kiocb *iocb, struct iov_iter *from) +{ + struct inode *inode = file_inode(iocb->ki_filp); + struct cifsInodeInfo *cinode = CIFS_I(inode); + struct cifs_sb_info *cifs_sb = CIFS_SB(inode->i_sb); + struct cifsFileInfo *cfile = (struct cifsFileInfo *) + iocb->ki_filp->private_data; + struct cifs_tcon *tcon = tlink_tcon(cfile->tlink); + ssize_t written; - if (CIFS_I(inode)->clientCanCacheAll) - return generic_file_aio_write(iocb, iov, nr_segs, pos); + written = cifs_get_writer(cinode); + if (written) + return written; + if (CIFS_CACHE_WRITE(cinode)) { + if (cap_unix(tcon->ses) && + (CIFS_UNIX_FCNTL_CAP & le64_to_cpu(tcon->fsUnixInfo.Capability)) + && ((cifs_sb->mnt_cifs_flags & CIFS_MOUNT_NOPOSIXBRL) == 0)) { + written = generic_file_write_iter(iocb, from); + goto out; + } + written = cifs_writev(iocb, from); + goto out; + } /* - * 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. + * For non-oplocked files 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. */ + written = cifs_user_writev(iocb, from); + if (written > 0 && CIFS_CACHE_READ(cinode)) { + /* + * Windows 7 server can delay breaking level2 oplock if a write + * request comes - break it on the client to prevent reading + * an old data. + */ + cifs_zap_mapping(inode); + cifs_dbg(FYI, "Set no oplock for inode=%p after a write operation\n", + inode); + cinode->oplock = 0; + } +out: + cifs_put_writer(cinode); + return written; +} - return cifs_user_writev(iocb, iov, nr_segs, pos); +static struct cifs_readdata * +cifs_readdata_alloc(unsigned int nr_pages, work_func_t complete) +{ + struct cifs_readdata *rdata; + + rdata = kzalloc(sizeof(*rdata) + (sizeof(struct page *) * nr_pages), + GFP_KERNEL); + if (rdata != NULL) { + kref_init(&rdata->refcount); + INIT_LIST_HEAD(&rdata->list); + init_completion(&rdata->done); + INIT_WORK(&rdata->work, complete); + } + + return rdata; } -static ssize_t -cifs_iovec_read(struct file *file, const struct iovec *iov, - unsigned long nr_segs, loff_t *poffset) +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 cifs_readdata *rdata, unsigned int nr_pages) +{ + int rc = 0; + struct page *page; + unsigned int i; + + for (i = 0; i < nr_pages; i++) { + page = alloc_page(GFP_KERNEL|__GFP_HIGHMEM); + if (!page) { + rc = -ENOMEM; + break; + } + rdata->pages[i] = page; + } + + if (rc) { + for (i = 0; i < nr_pages; i++) { + put_page(rdata->pages[i]); + rdata->pages[i] = NULL; + } + } + return rc; +} + +static void +cifs_uncached_readdata_release(struct kref *refcount) +{ + struct cifs_readdata *rdata = container_of(refcount, + struct cifs_readdata, refcount); + unsigned int i; + + for (i = 0; i < rdata->nr_pages; i++) { + put_page(rdata->pages[i]); + rdata->pages[i] = NULL; + } + cifs_readdata_release(refcount); +} + +static int +cifs_retry_async_readv(struct cifs_readdata *rdata) { int rc; - int xid; - ssize_t total_read; - unsigned int bytes_read = 0; + 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 + * @iter: destination for our data + * + * 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 int +cifs_readdata_to_iov(struct cifs_readdata *rdata, struct iov_iter *iter) +{ + size_t remaining = rdata->bytes; + unsigned int i; + + for (i = 0; i < rdata->nr_pages; i++) { + struct page *page = rdata->pages[i]; + size_t copy = min_t(size_t, remaining, PAGE_SIZE); + size_t written = copy_page_to_iter(page, 0, copy, iter); + remaining -= written; + if (written < copy && iov_iter_count(iter) > 0) + break; + } + return remaining ? -EFAULT : 0; +} + +static void +cifs_uncached_readv_complete(struct work_struct *work) +{ + struct cifs_readdata *rdata = container_of(work, + struct cifs_readdata, work); + + complete(&rdata->done); + kref_put(&rdata->refcount, cifs_uncached_readdata_release); +} + +static int +cifs_uncached_read_into_pages(struct TCP_Server_Info *server, + struct cifs_readdata *rdata, unsigned int len) +{ + int total_read = 0, result = 0; + unsigned int i; + unsigned int nr_pages = rdata->nr_pages; + struct kvec iov; + + rdata->tailsz = PAGE_SIZE; + for (i = 0; i < nr_pages; i++) { + struct page *page = rdata->pages[i]; + + if (len >= PAGE_SIZE) { + /* enough data to fill the page */ + iov.iov_base = kmap(page); + iov.iov_len = PAGE_SIZE; + cifs_dbg(FYI, "%u: iov_base=%p iov_len=%zu\n", + i, iov.iov_base, iov.iov_len); + len -= PAGE_SIZE; + } else if (len > 0) { + /* enough for partial page, fill and zero the rest */ + iov.iov_base = kmap(page); + iov.iov_len = len; + cifs_dbg(FYI, "%u: iov_base=%p iov_len=%zu\n", + i, iov.iov_base, iov.iov_len); + memset(iov.iov_base + len, '\0', PAGE_SIZE - len); + rdata->tailsz = len; + len = 0; + } else { + /* no need to hold page hostage */ + rdata->pages[i] = NULL; + rdata->nr_pages--; + put_page(page); + continue; + } + + result = cifs_readv_from_socket(server, &iov, 1, iov.iov_len); + kunmap(page); + if (result < 0) + break; + + total_read += result; + } + + return total_read > 0 ? total_read : result; +} + +ssize_t cifs_user_readv(struct kiocb *iocb, struct iov_iter *to) +{ + struct file *file = iocb->ki_filp; + ssize_t rc; size_t len, cur_len; - int iov_offset = 0; + ssize_t total_read = 0; + loff_t offset = iocb->ki_pos; + unsigned int npages; struct cifs_sb_info *cifs_sb; - struct cifs_tcon *pTcon; + struct cifs_tcon *tcon; struct cifsFileInfo *open_file; - struct smb_com_read_rsp *pSMBr; - struct cifs_io_parms io_parms; - char *read_data; - unsigned int rsize; - __u32 pid; - - if (!nr_segs) - return 0; + struct cifs_readdata *rdata, *tmp; + struct list_head rdata_list; + pid_t pid; - len = iov_length(iov, nr_segs); + len = iov_iter_count(to); if (!len) return 0; - xid = GetXid(); + INIT_LIST_HEAD(&rdata_list); 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); - open_file = file->private_data; - pTcon = tlink_tcon(open_file->tlink); + 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; @@ -2246,81 +2859,99 @@ cifs_iovec_read(struct file *file, const struct iovec *iov, pid = current->tgid; if ((file->f_flags & O_ACCMODE) == O_WRONLY) - cFYI(1, "attempting read on write only file instance"); + cifs_dbg(FYI, "attempting read on write only file instance\n"); - for (total_read = 0; total_read < len; total_read += bytes_read) { - cur_len = min_t(const size_t, len - total_read, rsize); - rc = -EAGAIN; - read_data = NULL; + do { + cur_len = min_t(const size_t, len - total_read, cifs_sb->rsize); + npages = DIV_ROUND_UP(cur_len, PAGE_SIZE); - while (rc == -EAGAIN) { - int buf_type = CIFS_NO_BUFFER; - if (open_file->invalidHandle) { - rc = cifs_reopen_file(open_file, true); - if (rc != 0) - break; - } - io_parms.netfid = open_file->netfid; - io_parms.pid = pid; - io_parms.tcon = pTcon; - io_parms.offset = *poffset; - io_parms.length = cur_len; - rc = CIFSSMBRead(xid, &io_parms, &bytes_read, - &read_data, &buf_type); - pSMBr = (struct smb_com_read_rsp *)read_data; - if (read_data) { - char *data_offset = read_data + 4 + - le16_to_cpu(pSMBr->DataOffset); - if (memcpy_toiovecend(iov, data_offset, - iov_offset, bytes_read)) - rc = -EFAULT; - if (buf_type == CIFS_SMALL_BUFFER) - cifs_small_buf_release(read_data); - else if (buf_type == CIFS_LARGE_BUFFER) - cifs_buf_release(read_data); - read_data = NULL; - iov_offset += bytes_read; - } + /* allocate a readdata struct */ + rdata = cifs_readdata_alloc(npages, + cifs_uncached_readv_complete); + if (!rdata) { + rc = -ENOMEM; + break; } - if (rc || (bytes_read == 0)) { - if (total_read) { - break; + rc = cifs_read_allocate_pages(rdata, npages); + if (rc) + goto error; + + rdata->cfile = cifsFileInfo_get(open_file); + rdata->nr_pages = npages; + rdata->offset = offset; + rdata->bytes = cur_len; + rdata->pid = pid; + rdata->pagesz = PAGE_SIZE; + rdata->read_into_pages = cifs_uncached_read_into_pages; + + 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; + + len = iov_iter_count(to); + /* the loop below should proceed in the order of increasing offsets */ + list_for_each_entry_safe(rdata, tmp, &rdata_list, list) { + again: + if (!rc) { + /* FIXME: freezable sleep too? */ + rc = wait_for_completion_killable(&rdata->done); + if (rc) + rc = -EINTR; + else if (rdata->result) { + rc = rdata->result; + /* resend call if it's a retryable error */ + if (rc == -EAGAIN) { + rc = cifs_retry_async_readv(rdata); + goto again; + } } else { - FreeXid(xid); - return rc; + rc = cifs_readdata_to_iov(rdata, to); } - } else { - cifs_stats_bytes_read(pTcon, bytes_read); - *poffset += bytes_read; + } + list_del_init(&rdata->list); + kref_put(&rdata->refcount, cifs_uncached_readdata_release); } - FreeXid(xid); - return total_read; -} + total_read = len - iov_iter_count(to); -ssize_t cifs_user_readv(struct kiocb *iocb, const struct iovec *iov, - unsigned long nr_segs, loff_t pos) -{ - ssize_t read; + cifs_stats_bytes_read(tcon, total_read); - read = cifs_iovec_read(iocb->ki_filp, iov, nr_segs, &pos); - if (read > 0) - iocb->ki_pos = pos; + /* mask nodata case */ + if (rc == -ENODATA) + rc = 0; - return read; + if (total_read) { + iocb->ki_pos += total_read; + return total_read; + } + return rc; } -ssize_t cifs_strict_readv(struct kiocb *iocb, const struct iovec *iov, - unsigned long nr_segs, loff_t pos) +ssize_t +cifs_strict_readv(struct kiocb *iocb, struct iov_iter *to) { - 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); + struct inode *inode = file_inode(iocb->ki_filp); + struct cifsInodeInfo *cinode = CIFS_I(inode); + struct cifs_sb_info *cifs_sb = CIFS_SB(inode->i_sb); + struct cifsFileInfo *cfile = (struct cifsFileInfo *) + iocb->ki_filp->private_data; + struct cifs_tcon *tcon = tlink_tcon(cfile->tlink); + int rc = -EACCES; /* * In strict cache mode we need to read from the server all the time @@ -2330,12 +2961,29 @@ ssize_t cifs_strict_readv(struct kiocb *iocb, const struct iovec *iov, * on pages affected by this read but not on the region from pos to * pos+len-1. */ + if (!CIFS_CACHE_READ(cinode)) + return cifs_user_readv(iocb, to); - return cifs_user_readv(iocb, iov, nr_segs, pos); + 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 generic_file_read_iter(iocb, to); + + /* + * We need to hold the sem to be sure nobody modifies lock list + * with a brlock that prevents reading. + */ + down_read(&cinode->lock_sem); + if (!cifs_find_lock_conflict(cfile, iocb->ki_pos, iov_iter_count(to), + tcon->ses->server->vals->shared_lock_type, + NULL, CIFS_READ_OP)) + rc = generic_file_read_iter(iocb, to); + up_read(&cinode->lock_sem); + return rc; } -static ssize_t cifs_read(struct file *file, char *read_data, size_t read_size, - loff_t *poffset) +static ssize_t +cifs_read(struct file *file, char *read_data, size_t read_size, loff_t *offset) { int rc = -EACCES; unsigned int bytes_read = 0; @@ -2343,15 +2991,16 @@ static ssize_t cifs_read(struct file *file, char *read_data, size_t read_size, unsigned int current_read_size; unsigned int rsize; struct cifs_sb_info *cifs_sb; - struct cifs_tcon *pTcon; - int xid; - char *current_offset; + struct cifs_tcon *tcon; + struct TCP_Server_Info *server; + unsigned int xid; + char *cur_offset; struct cifsFileInfo *open_file; struct cifs_io_parms io_parms; int buf_type = CIFS_NO_BUFFER; __u32 pid; - xid = GetXid(); + 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 */ @@ -2359,11 +3008,17 @@ static ssize_t cifs_read(struct file *file, char *read_data, size_t read_size, if (file->private_data == NULL) { rc = -EBADF; - FreeXid(xid); + free_xid(xid); return rc; } open_file = file->private_data; - pTcon = tlink_tcon(open_file->tlink); + tcon = tlink_tcon(open_file->tlink); + server = tcon->ses->server; + + if (!server->ops->sync_read) { + free_xid(xid); + return -ENOSYS; + } if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_RWPIDFORWARD) pid = open_file->pid; @@ -2371,17 +3026,17 @@ static ssize_t cifs_read(struct file *file, char *read_data, size_t read_size, pid = current->tgid; if ((file->f_flags & O_ACCMODE) == O_WRONLY) - cFYI(1, "attempting read on write only file instance"); + cifs_dbg(FYI, "attempting read on write only file instance\n"); - for (total_read = 0, current_offset = read_data; - read_size > total_read; - total_read += bytes_read, current_offset += bytes_read) { + for (total_read = 0, cur_offset = read_data; read_size > total_read; + total_read += bytes_read, cur_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 ((pTcon->ses) && - !(pTcon->ses->capabilities & CAP_LARGE_FILES)) { + /* + * 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); } @@ -2392,27 +3047,27 @@ static ssize_t cifs_read(struct file *file, char *read_data, size_t read_size, if (rc != 0) break; } - io_parms.netfid = open_file->netfid; io_parms.pid = pid; - io_parms.tcon = pTcon; - io_parms.offset = *poffset; + io_parms.tcon = tcon; + io_parms.offset = *offset; io_parms.length = current_read_size; - rc = CIFSSMBRead(xid, &io_parms, &bytes_read, - ¤t_offset, &buf_type); + rc = server->ops->sync_read(xid, open_file, &io_parms, + &bytes_read, &cur_offset, + &buf_type); } if (rc || (bytes_read == 0)) { if (total_read) { break; } else { - FreeXid(xid); + free_xid(xid); return rc; } } else { - cifs_stats_bytes_read(pTcon, total_read); - *poffset += bytes_read; + cifs_stats_bytes_read(tcon, total_read); + *offset += bytes_read; } } - FreeXid(xid); + free_xid(xid); return total_read; } @@ -2431,18 +3086,20 @@ cifs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf) static struct vm_operations_struct cifs_file_vm_ops = { .fault = filemap_fault, + .map_pages = filemap_map_pages, .page_mkwrite = cifs_page_mkwrite, + .remap_pages = generic_file_remap_pages, }; 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; + struct inode *inode = file_inode(file); - xid = GetXid(); + xid = get_xid(); - if (!CIFS_I(inode)->clientCanCacheRead) { - rc = cifs_invalidate_mapping(inode); + if (!CIFS_CACHE_READ(CIFS_I(inode))) { + rc = cifs_zap_mapping(inode); if (rc) return rc; } @@ -2450,7 +3107,7 @@ int cifs_file_strict_mmap(struct file *file, struct vm_area_struct *vma) rc = generic_file_mmap(file, vma); if (rc == 0) vma->vm_ops = &cifs_file_vm_ops; - FreeXid(xid); + free_xid(xid); return rc; } @@ -2458,20 +3115,125 @@ int cifs_file_mmap(struct file *file, struct vm_area_struct *vma) { int rc, xid; - xid = GetXid(); + xid = get_xid(); rc = cifs_revalidate_file(file); if (rc) { - cFYI(1, "Validation prior to mmap failed, error=%d", rc); - FreeXid(xid); + cifs_dbg(FYI, "Validation prior to mmap failed, error=%d\n", + rc); + free_xid(xid); return rc; } rc = generic_file_mmap(file, vma); if (rc == 0) vma->vm_ops = &cifs_file_vm_ops; - FreeXid(xid); + free_xid(xid); return rc; } +static void +cifs_readv_complete(struct work_struct *work) +{ + unsigned int i; + struct cifs_readdata *rdata = container_of(work, + struct cifs_readdata, work); + + for (i = 0; i < rdata->nr_pages; i++) { + struct page *page = rdata->pages[i]; + + lru_cache_add_file(page); + + if (rdata->result == 0) { + 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); + rdata->pages[i] = NULL; + } + kref_put(&rdata->refcount, cifs_readdata_release); +} + +static int +cifs_readpages_read_into_pages(struct TCP_Server_Info *server, + struct cifs_readdata *rdata, unsigned int len) +{ + int total_read = 0, result = 0; + unsigned int i; + u64 eof; + pgoff_t eof_index; + unsigned int nr_pages = rdata->nr_pages; + struct kvec iov; + + /* 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; + cifs_dbg(FYI, "eof=%llu eof_index=%lu\n", eof, eof_index); + + rdata->tailsz = PAGE_CACHE_SIZE; + for (i = 0; i < nr_pages; i++) { + struct page *page = rdata->pages[i]; + + if (len >= PAGE_CACHE_SIZE) { + /* enough data to fill the page */ + iov.iov_base = kmap(page); + iov.iov_len = PAGE_CACHE_SIZE; + cifs_dbg(FYI, "%u: idx=%lu iov_base=%p iov_len=%zu\n", + i, page->index, iov.iov_base, iov.iov_len); + len -= PAGE_CACHE_SIZE; + } else if (len > 0) { + /* enough for partial page, fill and zero the rest */ + iov.iov_base = kmap(page); + iov.iov_len = len; + cifs_dbg(FYI, "%u: idx=%lu iov_base=%p iov_len=%zu\n", + i, page->index, iov.iov_base, iov.iov_len); + memset(iov.iov_base + len, + '\0', PAGE_CACHE_SIZE - len); + rdata->tailsz = len; + len = 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); + lru_cache_add_file(page); + flush_dcache_page(page); + SetPageUptodate(page); + unlock_page(page); + page_cache_release(page); + rdata->pages[i] = NULL; + rdata->nr_pages--; + continue; + } else { + /* no need to hold page hostage */ + lru_cache_add_file(page); + unlock_page(page); + page_cache_release(page); + rdata->pages[i] = NULL; + rdata->nr_pages--; + continue; + } + + result = cifs_readv_from_socket(server, &iov, 1, iov.iov_len); + kunmap(page); + if (result < 0) + break; + + total_read += result; + } + + return total_read > 0 ? total_read : result; +} + static int cifs_readpages(struct file *file, struct address_space *mapping, struct list_head *page_list, unsigned num_pages) { @@ -2494,6 +3256,9 @@ static int cifs_readpages(struct file *file, struct address_space *mapping, /* * Reads as many pages as possible from fscache. Returns -ENOBUFS * immediately if the cookie is negative + * + * After this point, every page in the list might have PG_fscache set, + * so we will need to clean that up off of every page we don't use. */ rc = cifs_readpages_from_fscache(mapping->host, mapping, page_list, &num_pages); @@ -2508,8 +3273,8 @@ static int cifs_readpages(struct file *file, struct address_space *mapping, rc = 0; INIT_LIST_HEAD(&tmplist); - cFYI(1, "%s: file=%p mapping=%p num_pages=%u", __func__, file, - mapping, num_pages); + cifs_dbg(FYI, "%s: file=%p mapping=%p num_pages=%u\n", + __func__, file, mapping, num_pages); /* * Start with the page at end of list and move it to private @@ -2523,6 +3288,7 @@ static int cifs_readpages(struct file *file, struct address_space *mapping, * the rdata->pages, then we want them in increasing order. */ while (!list_empty(page_list)) { + unsigned int i; unsigned int bytes = PAGE_CACHE_SIZE; unsigned int expected_index; unsigned int nr_pages = 1; @@ -2574,7 +3340,7 @@ static int cifs_readpages(struct file *file, struct address_space *mapping, nr_pages++; } - rdata = cifs_readdata_alloc(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) { @@ -2587,41 +3353,45 @@ static int cifs_readpages(struct file *file, struct address_space *mapping, break; } - spin_lock(&cifs_file_list_lock); - cifsFileInfo_get(open_file); - spin_unlock(&cifs_file_list_lock); - rdata->cfile = open_file; + rdata->cfile = cifsFileInfo_get(open_file); rdata->mapping = mapping; rdata->offset = offset; rdata->bytes = bytes; rdata->pid = pid; - list_splice_init(&tmplist, &rdata->pages); + rdata->pagesz = PAGE_CACHE_SIZE; + rdata->read_into_pages = cifs_readpages_read_into_pages; - do { - if (open_file->invalidHandle) { - rc = cifs_reopen_file(open_file, true); - if (rc != 0) - continue; - } - rc = cifs_async_readv(rdata); - } while (rc == -EAGAIN); + list_for_each_entry_safe(page, tpage, &tmplist, lru) { + list_del(&page->lru); + rdata->pages[rdata->nr_pages++] = page; + } + rc = cifs_retry_async_readv(rdata); if (rc != 0) { - list_for_each_entry_safe(page, tpage, &rdata->pages, - lru) { - list_del(&page->lru); + for (i = 0; i < rdata->nr_pages; i++) { + page = rdata->pages[i]; lru_cache_add_file(page); unlock_page(page); page_cache_release(page); } - cifs_readdata_free(rdata); + kref_put(&rdata->refcount, cifs_readdata_release); break; } + + kref_put(&rdata->refcount, cifs_readdata_release); } + /* Any pages that have been shown to fscache but didn't get added to + * the pagecache must be uncached before they get returned to the + * allocator. + */ + cifs_fscache_readpages_cancel(mapping->host, page_list); return rc; } +/* + * cifs_readpage_worker must be called with the page pinned + */ static int cifs_readpage_worker(struct file *file, struct page *page, loff_t *poffset) { @@ -2629,11 +3399,10 @@ static int cifs_readpage_worker(struct file *file, struct page *page, int rc; /* Is the page cached? */ - rc = cifs_readpage_from_fscache(file->f_path.dentry->d_inode, page); + rc = cifs_readpage_from_fscache(file_inode(file), 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 */ @@ -2642,10 +3411,10 @@ static int cifs_readpage_worker(struct file *file, struct page *page, if (rc < 0) goto io_error; else - cFYI(1, "Bytes read %d", rc); + cifs_dbg(FYI, "Bytes read %d\n", rc); - file->f_path.dentry->d_inode->i_atime = - current_fs_time(file->f_path.dentry->d_inode->i_sb); + file_inode(file)->i_atime = + current_fs_time(file_inode(file)->i_sb); if (PAGE_CACHE_SIZE > rc) memset(read_data + rc, 0, PAGE_CACHE_SIZE - rc); @@ -2654,13 +3423,13 @@ static int cifs_readpage_worker(struct file *file, struct page *page, SetPageUptodate(page); /* send this page to the cache */ - cifs_readpage_to_fscache(file->f_path.dentry->d_inode, page); + cifs_readpage_to_fscache(file_inode(file), page); rc = 0; io_error: kunmap(page); - page_cache_release(page); + unlock_page(page); read_complete: return rc; @@ -2670,24 +3439,22 @@ static int cifs_readpage(struct file *file, struct page *page) { loff_t offset = (loff_t)page->index << PAGE_CACHE_SHIFT; int rc = -EACCES; - int xid; + unsigned int xid; - xid = GetXid(); + xid = get_xid(); if (file->private_data == NULL) { rc = -EBADF; - FreeXid(xid); + free_xid(xid); return rc; } - cFYI(1, "readpage %p at offset %d 0x%x\n", + cifs_dbg(FYI, "readpage %p at offset %d 0x%x\n", page, (int)offset, (int)offset); rc = cifs_readpage_worker(file, page, &offset); - unlock_page(page); - - FreeXid(xid); + free_xid(xid); return rc; } @@ -2740,6 +3507,7 @@ static int cifs_write_begin(struct file *file, struct address_space *mapping, loff_t pos, unsigned len, unsigned flags, struct page **pagep, void **fsdata) { + int oncethru = 0; pgoff_t index = pos >> PAGE_CACHE_SHIFT; loff_t offset = pos & (PAGE_CACHE_SIZE - 1); loff_t page_start = pos & PAGE_MASK; @@ -2747,8 +3515,9 @@ static int cifs_write_begin(struct file *file, struct address_space *mapping, struct page *page; int rc = 0; - cFYI(1, "write_begin from %lld len %d", (long long)pos, len); + cifs_dbg(FYI, "write_begin from %lld len %d\n", (long long)pos, len); +start: page = grab_cache_page_write_begin(mapping, index, flags); if (!page) { rc = -ENOMEM; @@ -2772,7 +3541,7 @@ static int cifs_write_begin(struct file *file, struct address_space *mapping, * 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) { + if (CIFS_CACHE_READ(CIFS_I(mapping->host))) { i_size = i_size_read(mapping->host); if (page_start >= i_size || (offset == 0 && (pos + len) >= i_size)) { @@ -2790,13 +3559,16 @@ static int cifs_write_begin(struct file *file, struct address_space *mapping, } } - if ((file->f_flags & O_ACCMODE) != O_WRONLY) { + if ((file->f_flags & O_ACCMODE) != O_WRONLY && !oncethru) { /* * 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); + page_cache_release(page); + oncethru = 1; + goto 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 @@ -2816,11 +3588,12 @@ static int cifs_release_page(struct page *page, gfp_t gfp) return cifs_fscache_release_page(page, gfp); } -static void cifs_invalidate_page(struct page *page, unsigned long offset) +static void cifs_invalidate_page(struct page *page, unsigned int offset, + unsigned int length) { struct cifsInodeInfo *cifsi = CIFS_I(page->mapping->host); - if (offset == 0) + if (offset == 0 && length == PAGE_CACHE_SIZE) cifs_fscache_invalidate_page(page, &cifsi->vfs_inode); } @@ -2836,7 +3609,7 @@ static int cifs_launder_page(struct page *page) .range_end = range_end, }; - cFYI(1, "Launder page: %p", page); + cifs_dbg(FYI, "Launder page: %p\n", page); if (clear_page_dirty_for_io(page)) rc = cifs_writepage_locked(page, &wbc); @@ -2845,31 +3618,53 @@ static int cifs_launder_page(struct page *page) return rc; } +static int +cifs_pending_writers_wait(void *unused) +{ + schedule(); + return 0; +} + 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); + struct cifs_tcon *tcon = tlink_tcon(cfile->tlink); + struct TCP_Server_Info *server = tcon->ses->server; int rc = 0; + wait_on_bit(&cinode->flags, CIFS_INODE_PENDING_WRITERS, + cifs_pending_writers_wait, TASK_UNINTERRUPTIBLE); + + server->ops->downgrade_oplock(server, cinode, + test_bit(CIFS_INODE_DOWNGRADE_OPLOCK_TO_L2, &cinode->flags)); + + if (!CIFS_CACHE_WRITE(cinode) && CIFS_CACHE_READ(cinode) && + cifs_has_mand_locks(cinode)) { + cifs_dbg(FYI, "Reset oplock to None for inode=%p due to mand locks\n", + inode); + cinode->oplock = 0; + } + if (inode && S_ISREG(inode->i_mode)) { - if (cinode->clientCanCacheRead) + if (CIFS_CACHE_READ(cinode)) break_lease(inode, O_RDONLY); else break_lease(inode, O_WRONLY); rc = filemap_fdatawrite(inode->i_mapping); - if (cinode->clientCanCacheRead == 0) { + if (!CIFS_CACHE_READ(cinode)) { rc = filemap_fdatawait(inode->i_mapping); mapping_set_error(inode->i_mapping, rc); - invalidate_remote_inode(inode); + cifs_zap_mapping(inode); } - cFYI(1, "Oplock flush inode %p rc %d", inode, rc); + cifs_dbg(FYI, "Oplock flush inode %p rc %d\n", inode, rc); } rc = cifs_push_locks(cfile); if (rc) - cERROR(1, "Push locks rc = %d", rc); + cifs_dbg(VFS, "Push locks rc = %d\n", rc); /* * releasing stale oplock after recent reconnect of smb session using @@ -2878,14 +3673,34 @@ void cifs_oplock_break(struct work_struct *work) * disconnected since oplock already released by the server */ if (!cfile->oplock_break_cancelled) { - rc = CIFSSMBLock(0, tlink_tcon(cfile->tlink), cfile->netfid, - current->tgid, 0, 0, 0, 0, - LOCKING_ANDX_OPLOCK_RELEASE, false, - cinode->clientCanCacheRead ? 1 : 0); - cFYI(1, "Oplock release rc = %d", rc); + rc = tcon->ses->server->ops->oplock_response(tcon, &cfile->fid, + cinode); + cifs_dbg(FYI, "Oplock release rc = %d\n", rc); } + cifs_done_oplock_break(cinode); } +/* + * The presence of cifs_direct_io() in the address space ops vector + * allowes open() O_DIRECT flags which would have failed otherwise. + * + * In the non-cached mode (mount with cache=none), we shunt off direct read and write requests + * so this method should never be called. + * + * Direct IO is not yet supported in the cached mode. + */ +static ssize_t +cifs_direct_io(int rw, struct kiocb *iocb, struct iov_iter *iter, + loff_t pos) +{ + /* + * FIXME + * Eventually need to support direct IO for non forcedirectio mounts + */ + return -EINVAL; +} + + const struct address_space_operations cifs_addr_ops = { .readpage = cifs_readpage, .readpages = cifs_readpages, @@ -2895,6 +3710,7 @@ const struct address_space_operations cifs_addr_ops = { .write_end = cifs_write_end, .set_page_dirty = __set_page_dirty_nobuffers, .releasepage = cifs_release_page, + .direct_IO = cifs_direct_io, .invalidatepage = cifs_invalidate_page, .launder_page = cifs_launder_page, }; |